A Chaperone for the Stator Units of a Bacterial Flagellum

The bacterial flagellum is a reversible rotating motor powered by ion transport through stator units, which also exert torque on the rotor component to turn the flagellum for motility. Species-specific adaptations to flagellar motors impact stator function to meet the demands of each species to sufficiently power flagellar rotation. We identified another evolutionary adaptation by discovering that FlgX of Campylobacter jejuni preserves the integrity of stator units by functioning as a chaperone to protect stator proteins from degradation by the FtsH protease complex due to the physiology of the bacterium. FlgX is required to maintain a level of stator units sufficient to power the naturally high-torque flagellar motor of C. jejuni for motility in intestinal mucosal layers to colonize hosts. Our work continues to identify an increasing number of adaptations to flagellar motors across bacterial species that provide the mechanics necessary for producing an effective rotating nanomachine for motility.

units is hypothesized to cause a conformational change resulting in a pushing force that exerts torque upon the rotor to turn the extracellular filament to propel bacteria through different environmental milieus (8).
Components forming the stator units are fairly extensively conserved across motile bacterial species. As studied in Escherichia coli and Salmonella species, each stator unit is a heterohexamer of MotA and MotB (MotA 4 MotB 2 ) (9,10). Both proteins are localized to the inner membrane, with MotA possessing four transmembrane domains and two large cytoplasmic domains whereas MotB has one transmembrane domain linked to a large periplasmic domain (11)(12)(13)(14)(15). MotB transmembrane domains complete formation of an ion channel upon complexing with MotA proteins to form a stator unit. MotB also contains the conserved aspartate residue for ion binding and translocation (16). Stator unit incorporation into the motor results in conformational changes that remove a "plug" from the channel, allowing ion passage, which in turn results in a major restructuring of the MotB periplasmic domain to interact with peptidoglycan for the tethering that is essential for stator and motor function (17)(18)(19)(20)(21). Ion passage also causes conformational changes in MotA within a stator unit such that the cytoplasmic region contacting FliG of the rotor creates a pushing force that generates torque for flagellar rotation (8,22,23).
The mechanics of stator function are thought to be largely conserved across flagellar motors of diverse bacterial species. However, some significant differences have been noted. In many Vibrio species, motility is powered by sodium ions through the PomAB stator, which is similar to the MotAB counterpart that is powered by hydrogen ions (24)(25)(26). Some motile bacteria such as Shewanella oneidensis and Bacillus subtilis possess both hydrogen-powered and sodium-powered stator units, which increases the fuel options for powering motility (27,28). In pseudomonads, two different stator units, MotAB and MotCD, are produced and exchanged in the motors (29)(30)(31). Whereas MotAB stators are used only for swimming motility, MotCD stators are employed for both swimming motility and swarming motility.
E. coli and Salmonella stator units are dynamic components of the flagellar motor that can differ in number per motor and can also be exchanged (32)(33)(34)(35)(36). Although 1 stator unit can power rotation, up to 11 can be incorporated per motor in a loaddependent manner that correlates with increased torque (32,34). However, other bacterial species, notably, Campylobacter jejuni, Helicobacter pylori, and Vibrio species, produce motors with evolutionary adaptations that appear as disk or ring appendages that impact stator integration and, consequently, the supply of power and generation of torque (3,(37)(38)(39)(40)(41). These structures form around the periplasmic rod and rings and radiate outward to serve as scaffolds, resulting in integration of stator units at a greater radial distance around the flagellar rotor than is seen with Salmonella (3,(37)(38)(39). Due to expanded rotor diameter compared to the Salmonella flagellar motor, C. jejuni and Vibrio motors accommodate 17 and 13 stator units into the respective flagellar motors (3). The increased number of stator units and the placement of the stator units at a greater radial distance from the axle of the flagellar motor are thought to contribute to increased torque and to the observed higher velocity of motility of the C. jejuni and Vibrio motors (3,37,42,43). A current hypothesis is that the number of stator units in flagellar motors with these scaffolds (such as in C. jejuni) is fixed and may not vary depending on the load imparted on the flagellum as seen in E. coli or Salmonella (36,37).
In this work, we expanded the adaptations in flagellar motors that impact stator function and motor output. We determined that C. jejuni FlgX, which was previously identified to be required for motility (44), is a chaperone that functions to maintain stator unit integrity for host colonization and full motility. The use of a chaperone for the stator units is an unusual feature in motile bacteria, as a chaperone specific for stator units has not been recognized in other flagellar systems. We describe additional findings that suggest that the requirement for FlgX-dependent stability of stator units may be due to a specific characteristic of the physiology of C. jejuni, with the FtsH inner membrane protease as a major contributor to stator protein degradation. This work  (Fig. S2). Complementation of the ΔflgX mutant with a plasmid that constitutively expressed WT FlgX, FLAG-FlgX, or FlgX-FLAG restored WT levels of both stator proteins (Fig. 1B). Note that all FLAG-tagged FlgX proteins used in this experiment and those described were the full-length 165-amino-acid-long sequence. We were unable to compare the relative levels of FlgX produced from the native chromosomal locus with FlgX expressed from complementing plasmids as we were unable to generate specific antiserum against FlgX after multiple attempts. We eliminated the possibility of a role for FlgX in transcription of motAB, as semiquantitative real-time PCR (qRT-PCR) analysis revealed similar levels of motA and motB expression in WT C. jejuni and the ΔflgX mutant with or without a complementing plasmid for expression of FlgX (Fig. S3) (44). We confirmed that MotA and MotB coimmunoprecipitated with both FLAG-FlgX or FlgX-FLAG when expressed in C. jejuni ΔflgX after formaldehyde cross-linking of cells ( Fig. 2A). MotA and MotB did not bind the immunoprecipitation resin nonspecifically as these proteins were not detected in samples in which WT FlgX without a FLAG tag was produced or in a sample lacking FlgX ( Fig. 2A). This interaction between FlgX and the stator proteins was specific as FlgX did not coimmunoprecipitate the FliF MS ring protein, which is an inner membrane flagellar protein like the stator proteins, or the cytoplasmic ␣ subunit of RNA polymerase (RpoA; Fig. 2A).
We next conducted experiments to discern whether FlgX interacted with MotA or with MotB or with both proteins to maintain stator protein stability and function. For this analysis, we constructed mutants lacking either motA or motB in C. jejuni ΔflgX and then expressed FLAG-FlgX in trans from a plasmid for coimmunoprecipitation assays. Native MotA immunoprecipitated with FLAG-FlgX at similar levels in the ΔflgX and ΔflgX ΔmotB mutants, even though the MotA levels were slightly lower in the ΔflgX ΔmotB mutant (Fig. 2B). These data indicated that FlgX-MotA interactions do not require MotB. Due to significantly reduced levels of MotB in the ΔmotA mutant (Fig. 2B), we were unable to determine whether FlgX also directly interacts with MotB.
FlgX is predicted to be a cytoplasmic protein, lacking both a signal sequence for secretion and transmembrane domains for integration into the inner membrane. We isolated proteins from the cytoplasmic, membrane, and periplasmic fractions of C. jejuni ΔflgX and ΔflgX ΔmotAB mutants producing FLAG-FlgX to determine the compartmental location of FlgX in C. jejuni. As predicted, we observed FlgX to localize exclusively to the C. jejuni cytoplasm with RpoA and not with any other compartment (Fig. 3). Although we anticipated that FlgX might show some localization to the inner membrane due to interactions with stator units, FlgX was not found to be associated with the inner membrane (Fig. 3). Thus, the cytoplasmic location did not change regardless of whether the stator proteins were produced. We propose that FlgX interacts with a cytoplasmic domain of MotA to maintain stability of MotAB stator units in C. jejuni on the basis of the following observations: (i) FlgX is localized to the cytoplasm; (ii) MotA has prominent cytoplasmic domains, in contrast to MotB, which is predicted to have a small N-terminal cytoplasmic domain of only ϳ18 residues; and (iii) FlgX efficiently coimmunoprecipitated MotA in the absence of MotB.
Although stator units are dynamic components of the Salmonella and E. coli flagellar motors, a current hypothesis suggests that the stator units of C. jejuni and other bacteria that integrate stators into the flagellar motor via scaffolding proteins are fixed and may not fluctuate in number with the load exerted upon the flagellum (37). Previous structural analysis of the C. jejuni flagellar motor by electron cryotomography revealed that 17 stator complexes are integrated into the flagellar motor, with MotA contacting FliG and MotB (likely interacting with the PflB proximal disk scaffolding protein) (3). In line with these observations, C. jejuni ΔpflB lacks MotAB stator units in jejuni ΔmotAB. C. jejuni cultures were standardized to similar cellular densities and then fractionated to recover proteins from the cytoplasmic (C), inner membrane (IM), periplasmic (P), and outer membrane (OM) fractions. Samples from whole-cell lysates (WCL) were also recovered. Equal amounts of proteins were loaded for all strains. FLAG-FlgX was detected by a monoclonal anti-FLAG antibody. Specific antisera were used to detect proteins as specific markers for different fractions, including RpoA (cytoplasm), PflB (inner membrane), Cjj81176_0382 (periplasm), and FlgP (outer membrane).
The FlgX Chaperone for Flagellar Stator Units ® the flagellar motor (3). We determined whether FlgX is required for stability of stator proteins in a nonfunctional motor without a rotor (ΔfliG mutant) or when stator units are unable to integrate into the flagellar motor (ΔpflB mutant). We observed that FlgX was required to maintain WT levels of the stator proteins in the flagellar motors of both the ΔfliG mutant and the ΔpflB mutant (Fig. 4). Due to uncertainty with respect to analyzing C. jejuni under conditions in which all stator units would be integrated around the rotor without any unincorporated stator units produced elsewhere in the cell, we were unable to determine whether FlgX remains associated with stator units after their incorporation into the motor to maintain their integrity.
Suppressor analysis of C. jejuni ⌬flgX. Taking into consideration the results presented above, we propose that FlgX likely functions as a chaperone to preserve the integrity of C. jejuni stator units to ensure power and generation of torque for flagellar motility. We performed a suppressor analysis with C. jejuni ΔflgX to attempt to isolate mutants with augmented motility and to provide insight into how FlgX ensures stator integrity as a chaperone. For this analysis, we inoculated C. jejuni ΔflgX mutants (with an in-frame deletion of residues 6 to 153 or residues 101 to 153) into MH motility agar, followed by incubation at 37°C for up to 7 days. A total of 18 independent suppressor mutants were isolated from motile flares originating from the point of inoculation. All ΔflgX suppressor mutants demonstrated a greater level of motility and higher levels of MotA and MotB in lysates than the parental ΔflgX strains. Panels A and B of Fig. 5 show the motility phenotypes and stator protein levels for all ΔflgX suppressor mutants isolated from the mutant with a deletion of codons 101 to 153 of flgX (suppressors S1 to S10). Similar elevated levels of motility and MotA and MotB stator proteins were also observed for the eight suppressor mutants isolated from the C. jejuni ΔflgX lacking codons 6 to 153 (suppressors S11 to S18; data not shown).
Genomic sequencing or targeted sequencing of genes of interest in the ΔflgX suppressor mutants revealed three classes of suppressor mutations ( Table 1). The first class of suppressor mutations involved five alterations to the motAB locus. Four suppressor mutations were identified as point mutations in motA to produce MotA H138Y (3 independent isolates) and MotA A201V (a single isolate). The H138 residue is located in the larger of the two cytoplasmic domains of MotA ( Fig. S4A and B). This domain interacts with FliG for torque generation and rotor rotation, but H138Y is not hypothesized to be involved in these processes. The A201 residue is predicted to be the last residue of the most C-terminal transmembrane domain of MotA ( Fig. S4A and B). The suppressor mutation in suppressor S7 resulted in a duplication of a 13-kb region of the chromosome encompassing motAB, effectively creating a motAB merodiploid that increased MotA and MotB production in the absence of FlgX (Fig. 5B). The members of the second class of suppressor mutants all contained mutations within ftsH, which encodes an essential inner membrane AAA protease that largely assists in quality control of inner membrane proteins, along with degradation of some cytoplasmic proteins (Table 1) (45). Four ΔflgX suppressor mutants contained independent transversions occurring at the same nucleotide to result in production of FtsH M250I , and another suppressor mutation was due to a transversion to produce FtsH S243T . In structurally analyzed FtsH proteins, these residues are near the entry pore Suppressor mutant S10 with the motA H138Y missense mutation served as a control. The level of expression of ftsH in C. jejuni ΔflgX as measured by qRT-PCR is set to 1. Expression of ftsH suppressor mutants is shown relative to that seen with C. jejuni ΔflgX. Error bars indicate standard deviations. Statistically significant differences in ftsH expression levels between C. jejuni ΔflgX and suppressor mutants (*, P Ͻ 0.05) as performed by Student's t test are indicated.
The FlgX Chaperone for Flagellar Stator Units ® of the FtsH ATPase complex and are possibly important for pore substrate recognition of misfolded substrate proteins (46). Two other suppressor mutations identified were in-frame duplications of part of the coding region for ftsH, including an 18-nucleotide duplication and a 210-nucleotide duplication ( Table 1).
The members of the final class of suppressor mutants all contained alterations within Cjj81176_1135, encoding a predicted PrmA homolog that functions as the L11 ribosomal protein methyltransferase (Table 1). Cjj81176_1135 is immediately upstream of ftsH, and the two genes are likely cotranscribed. All six mutations in Cjj81176_1135 involved insertions or deletions of nucleotides that caused frameshift mutations and truncation of the Cjj81176_1135 reading frame. We hypothesized that these frameshift mutations in Cjj81176_1135 may have caused polar effects on transcription or the on stability of the ftsH mRNA. qRT-PCR analysis revealed that the ftsH mRNA levels in these Cjj81176_1135 mutants were 3-fold to 5-fold lower than in the parental ΔflgX mutant or in a ΔflgX motA H138Y suppressor (suppressor S10) that was also identified in this suppressor screen (Fig. 5C). Thus, a reduction in the level of ftsH transcription in these Cjj81176_1135 mutants likely resulted in the increased levels of MotA and MotB in the absence of FlgX.
We examined the sufficiency of the H138Y alteration of MotA for restoring stator protein levels and motility to the ΔflgX mutant by replacing WT motA gene with motA H138Y on the chromosome of WT C. jejuni 81-176 and the ΔflgX mutant. MotA topology predictions suggest that H138Y resides in the first of two large cytoplasmic domains of the protein. The levels of motility, in addition to those of both MotA and MotB proteins, were comparable between WT C. jejuni and C. jejuni motA H138Y (Fig. 6A and B). Importantly, MotA and MotB levels were increased in the ΔflgX motA H138Y mutant compared to the ΔflgX mutant, and these levels were similar to those of the original ΔflgX motA H138Y suppressor mutant (S10) that had been isolated. Furthermore, we found that the ΔflgX motA H138Y mutant supported a greater level of motility than the ΔflgX mutant containing WT motA (Fig. 6A). Measurements of the motile rings in motility agar indicated that the level of migration from the point of inoculation of the ΔflgX motA H138Y mutant was 83.5% (Ϯ19.5%, P Ͻ 0.05) of the level seen with WT C. jejuni, indicating only a modest but a significant reduction in motility. These observations demonstrate that the H138Y alteration in MotA was nearly sufficient to restore the levels of integrity and function of the stator units seen in the absence of FlgX to WT levels. We also observed that H138 was not essential for binding by FlgX, as FLAG-FlgX coimmunoprecipitated WT MotA and MotA H138Y comparably well (Fig. 6C). Cjj81176_1135 Insertion of T at base 231 Frameshift S13 Cjj81176_1135 Deletion of A at base 12 Insertion of A at base 12 Frameshift a Suppressor mutants S1 to S10 were isolated from DAR4803 (containing a deletion of codons 101 to 153 of flgX), and mutants S11 to S18 were isolated from DAR2340 (containing a deletion of codons 6 to 153 of flgX). Spontaneous motility mutants were isolated from motility agar after incubation at 37°C for up to 8 days. b The location of the mutation was identified by genomic sequencing or by sequencing of a specific gene of interest.
We attempted to determine whether the FtsH M250I mutation was sufficient for restoring stator protein levels and motility to the ΔflgX mutant. However, we were unable to replace WT FtsH with FtsH M250I because construction of such a mutant requires an intermediate step to disrupt wild-type ftsH (an essential gene) with an antibiotic resistance cassette prior to replacement with the ftsH M250I allele. We also attempted to construct plasmids to express WT ftsH and ftsH M250I in trans and then to eliminate ftsH on the chromosome. However, the level of in trans expression of either allele from multicopy plasmids was likely too high relative to native ftsH expression from the chromosome to accurately interpret whether FtsH M250I had reduced protease activity with respect to the stator proteins (data not shown).
Analysis of stator proteins upon heterologous expression in C. jejuni and E. coli. As we are unaware of the existence of any previous report identifying a chaperone specific for stator units in a bacterial flagellar system, we investigated whether the requirement of FlgX as a chaperone for stator units is linked to an aspect of C. jejuni physiology and/or to a specific feature of the C. jejuni stator proteins. For this approach, we compared the levels of production of C. jejuni and Salmonella enterica serovar Typhimurium MotAB stator units upon heterologous expression in C. jejuni ΔmotAB in the presence or absence of FlgX and in E. coli. We complemented the C. jejuni and E. coli strains with plasmids to express Salmonella motAB or C. jejuni motAB from identical promoters and plasmid backbones to ensure levels of protein production that were as similar as possible. FLAG-tagged epitopes were fused to the C terminus of MotB for each motAB locus so that the same antibody could be used to detect MotB-FLAG to eliminate differences in methods for protein detection. We attempted to fuse a FLAG tag to the N terminus of MotA for similar analyses, but FLAG-MotA does not efficiently complement C. jejuni ΔmotA and the protein was not stable for detection.
Expression of C. jejuni motAB-FLAG and Salmonella motAB-FLAG in C. jejuni ΔmotAB containing FlgX revealed that both MotB-FLAG proteins (and presumably MotA partner proteins) were stable, with an apparent increased level of Salmonella MotB present compared to C. jejuni MotB (Fig. 7A). The differences in the sizes of the MotB proteins was due to Salmonella motB naturally encoding a larger protein. Whereas no C. jejuni MotB-FLAG was detected in C. jejuni ΔmotAB without FlgX, Salmonella MotB-FLAG was detected in both the lysates and the total membrane fraction in this strain. These data suggest that the Salmonella stator unit is not as sensitive to degradation under the physiological conditions present in C. jejuni as the native C. jejuni stator proteins. We also noticed higher levels of Salmonella MotB-FLAG in C. jejuni with FlgX than without, indicating that FlgX has an ability to enhance levels of heterologous stator proteins. In contrast to expression in C. jejuni, the C. jejuni stator proteins and the Salmonella stator proteins were produced at comparable levels and were stable in E. coli, which does not encode any discernible FlgX ortholog (Fig. 7B). Therefore, C. jejuni stators are stable without FlgX in other bacterial species. These observations point to the physiology of C. jejuni, presumably due to its altered or augmented FtsH activity, targeting C. jejuni stator proteins for degradation and thus requiring FlgX as a chaperone for maintaining stator unit integrity for flagellar motor rotation and motility.
In vivo requirement of FlgX for host colonization. C. jejuni requires flagellar motility for infection of humans to promote diarrheal disease and optimal colonization of the intestinal tract of avian species for commensalism (47)(48)(49)(50). As stator units are needed to supply power and generate torque for flagellar rotation and motility, we assessed the requirement for FlgX for cecal colonization of the chick intestinal tract. We also examined whether MotA H138Y , which promotes increased stator unit stability and motility in the absence of FlgX, could suppress any potential requirement for FlgX for host colonization. We subjected day-of-hatch chicks to orally gavage with ϳ10 4 CFU of WT C. jejuni, the ΔflgX mutant, and the reconstructed ΔflgX motA H138Y mutant and then determined the level of colonization in the ceca of chicks 14 days postinfection. WT C. jejuni reached an average of 1.24 ϫ 10 9 CFU per gram of cecal content, but the ΔflgX mutant showed a 10,000-fold reduction in colonization (average of 1.04 ϫ 10 5 CFU per gram of cecal content; Fig. 8). We did not identify any ΔflgX suppressor mutants that developed in vivo to enhance motility. Replacement of wild-type motA with motA H138Y in the ΔflgX mutant caused the colonization capacity of the ΔflgX mutant to improve by 3 orders of magnitude to 1.2 ϫ 10 8 CFU per gram cecal content on average, which was only 10-fold lower than the level seen with WT C. jejuni. Thus, the colonization defect of the ΔflgX mutant was almost completely restored by a single-point mutation in MotA. These data suggest that the primary in vivo role of FlgX is as a chaperone to maintain stator stability to supply power and generate torque for flagellar rotation for the motility that is essential for host interactions.

DISCUSSION
Stators are essential components of the flagellar motor that power rotation by transporting hydrogen or sodium ions and exert torque on the flagellar rotor for rotation. Efficient production and integration of stator units into the flagellar motor are essential for motility. C. jejuni FlgX was initially found to be required for motility and to interact with MotA and MotB (44). In this work, we provide a detailed analysis impli- cating FlgX as a chaperone that ensures the integrity of the MotAB stator unit for supplying power and torque for flagellar motility. We base this conclusion on the following observations: (i) production of WT levels of MotA and MotB in C. jejuni requires FlgX; (ii) FlgX is not a regulatory factor required for transcription of motAB; (iii) FlgX can be isolated in a complex with MotA and MotB, likely through a direct interaction between FlgX and MotA; (iv) FlgX is required for MotA and MotB stability in flagellar mutants unable to integrate stator units into the motor, suggesting that FlgX is required for the stability of stator units prior to incorporation into the motor; and (iv) MotA and MotB levels and motility were partially to fully restored in the ΔflgX mutant with suppressor mutations that reduced ftsH expression or (presumably) activity of the FtsH inner membrane protease, indicating that FlgX protects stator units from proteolysis. The one issue concerning a function often attributed to some other chaperone proteins that we currently cannot resolve is whether FlgX is released from the stator unit upon integration of the stators into the motor. Unfortunately, we cannot design an experiment that would give strong conclusive data for assessment. To date, a chaperone for stator complexes has not been identified in any other flagellar system. Along with the scaffolding structures composed of FlgP, PflA, and PflB in C. jejuni (3), FlgX represents another unique evolutionary adaptation that has occurred in Campylobacter species to influence the mechanics of a high-torque motor for motility.
Our data suggest that FlgX may interact with only MotA to maintain stability of the stator complexes. MotA is predicted to contain two large cytoplasmic domains, whereas MotB is predicted to contain a small N-terminal domain of about 18 amino acids. Because  Although FlgX was able to increase Salmonella MotB (and presumably MotA) stability in C. jejuni, proteins with significant homology to FlgX are present only in Campylobacter species and closely related epsilonproteobacterial species. Even in Helicobacter pylori, another epsilonproteobacterium, the idea of the presence of a functional FlgX ortholog is dubious, with only one protein identified with low homology (i.e., HPG27_1307; 28% identical and 36% similar to FlgX in 47 of 183 residues). Considering the number of flagellar systems that have been investigated across bacterial species, C. jejuni and closely related species may represent an exclusive example in producing flagellar systems requiring a protein such as FlgX to function as a chaperone for stator units. This proposition prompts an intriguing question. Why does C. jejuni need FlgX as a chaperone to maintain stator stability when other motile bacterial species apparently do not? On the basis of the results from our suppressor screen, suppressor mutations in motA or those in the ftsH locus that either reduce the levels of ftsH expression or presumably alter FtsH activity allow higher levels of stator proteins and motility in the absence of FlgX. These findings suggest that the physiology of C. jejuni may include an altered or augmented form of FtsH activity relative to other motile bacteria that consequently targets stator proteins for destruction. Currently, we do not know whether the C. jejuni stator proteins are initially produced in C. jejuni in a relatively more unstructured form that might make them more susceptible to degradation by FtsH. However, when C. jejuni stator proteins were produced in E. coli, they were comparable to Salmonella stator proteins in stability. This finding suggests the C. jejuni proteins are not generally less structurally stable than other bacterial stator proteins, at least when produced in other systems.
It is unclear why the physiology of C. jejuni might have evolved to include FtsH activity targeting stator units, and perhaps other inner membrane proteins in general, more readily than that in other bacterial species. C. jejuni does have an optimal growth temperature of 42°C, which is the normal body temperature of its natural avian host. It is possible that this optimal growth temperature may have necessitated an elevated or altered form of FtsH activity as a quality control mechanism for increased protein unfolding and turnover. However, altered FtsH activity might be expected to target many proteins nonspecifically such that numerous proteins, including the MotA and MotB stator proteins, would require their own specific chaperones, which has not been a common theme revealed in C. jejuni biology so far.
Another intriguing possibility is that targeted destruction of the stator units by FtsH in C. jejuni might be a mechanism to remove incorporated stators from the flagellar motor to reduce power and rotation. Currently, it is unknown whether stator units in the flagellum are dynamic or whether the stator number is directly correlated to the load exerted on the flagellum in C. jejuni. In E. coli, one stator complex is sufficient to power rotation, but stator numbers increase to up to 11 per motor to augment rotational speed and power when the external load exerted on the flagellum rises (33-36, 51, 52). A current hypothesis suggests that the stator number in the C. jejuni flagellar motor may be fixed at 17 stator units per motor (3,37). If FlgX is released from stator units after incorporation into the motor, targeting these incorporated stator units for proteolysis by FtsH may be a mechanism to reduce power for rotation and to disassemble stator units from the flagellar motor. Currently, we are unable to discern whether FlgX is associated with stators after incorporation into the motor. However, it is difficult to envision the existence of a natural niche or condition in which C. jejuni would need to reduce stator function and torque since the bacterium is found primarily associated with intestinal mucus layers in avian, animal, or human hosts. Production of a high-torque flagellar motor seems optimal for motility through viscous milieus such as these host mucus layers. Thus, evolving FlgX as an adaptation to the flagellar motor of C. jejuni likely ensures optimal production of stator units to fully and consistently supply power and torque for the high-torque flagellar motor.
We also isolated suppressor mutations within motA that partially restored motility, with motA H138Y almost completely restoring motility to the ΔflgX mutant. The H138 residue of MotA is not required for interactions with FlgX as the mutant protein appeared to coimmunoprecipitate with FlgX as well as WT MotA. MotA A201, the target of the other MotA suppressor mutation, is predicted to occur at a transmembranecytoplasm interface (see Fig. S4A in the supplemental material). The MotA A201V suppressor mutant restored motility to the ΔflgX mutant but at a reduced level compared to MotA H138Y . FtsH usually recognizes target proteins for degradation via nonpolar regions of approximately 20 amino acids in length at the N or C termini of proteins without a strictly defined sequence (53)(54)(55)(56). H138 and A201 of C. jejuni MotA are positioned at more central positions of the protein. Thus, we do not anticipate that either H138Y or A201V would have disrupted a proteolytic site for FtsH. Instead, these point mutations may have altered the structure of MotA or the MotA 4 MotB 2 heterohexameric complex such that MotA is more resistant to unfolding and recognition by FtsH. H138 is not conserved in Salmonella MotA, but this residue is naturally a tyrosine in H. pylori MotA (Fig. S4B). This natural alteration of MotA in H. pylori may negate the need of this epsilonproteobacterium to possess a FlgX ortholog to preserve stator integrity. While C. jejuni ΔflgX displayed a 10,000-fold defect in commensal colonization of the chick ceca, the simple alteration of H138 of MotA to a tyrosine that occurred in one of the suppressor mutants restored colonization levels 1,000-fold for the ΔflgX mutant. Furthermore, the colonization capacity of C. jejuni ΔflgX motA H138Y was only 10-fold lower than that WT C. jejuni. Due to this nearly complete restoration of stator protein levels, motility, and colonization, we think FlgX likely does not have other significant in vivo roles for C. jejuni outside maintaining stability of stator complexes. The modest reduction in in vitro motility conferred by MotA H138Y compared to WT MotA may have contributed to the 10-fold colonization defect observed in the ΔflgX motA H138Y mutant relative to WT C. jejuni.
Our work has identified a unique role for FlgX in C. jejuni as a chaperone for the stator units of a bacterial flagellum. The employment of FlgX to maintain a sufficient level of stator units to power flagellar rotation represents another evolutionary adaptation of the C. jejuni flagellar motor and of bacterial flagellar motors in general. The maintenance of the integrity of stator units via FlgX, along with the evolutionary adaptations in the flagellar motor structure due to the incorporation of disk scaffolds that integrate more stator units and position them at a greater radial distance to increase torque, demonstrates the exquisite ability of C. jejuni to adapt and construct a high-torque motor to facilitate flagellar motility.

MATERIALS AND METHODS
Bacterial strains and plasmids. The C. jejuni 81-176 strains used in this study are described in Table 2. All plasmids used or constructed for studies in this work are described in Table 3. Unless otherwise indicated, C. jejuni was grown from freezer stocks on Mueller-Hinton (MH) agar under microaerobic conditions (85% N 2 , 10% CO 2 , 5% O 2 ) at 37°C for 48 h and then restreaked on MH agar and grown for 16 h under identical conditions prior to each experiment. Antibiotics were added to MH media as required at the following concentrations: 10 g/ml trimethoprim; 15 g/ml chloramphenicol; 30 g/ml cefoperazone; or 0.5, 1, 2, or 5 mg/ml streptomycin. All C. jejuni strains were stored at Ϫ80°C in a 85% MH broth-15% glycerol solution. Escherichia coli DH5␣ and DH5/pRK212.1 were grown on Luria-Bertani (LB) agar or in LB broth containing 100 g/ml ampicillin, 15 g/ml chloramphenicol, or 12.5 g/ml tetracycline as appropriate. All E. coli strains were stored at Ϫ80°C in a 80% LB broth-20% glycerol solution.
Construction of C. jejuni mutants. C. jejuni mutants were constructed by electroporation of plasmid DNA or natural transformation of in vitro-methylated plasmid DNA following previously described methods (57,58). All plasmids were constructed by ligation of DNA fragments into plasmids by the use of T4 DNA ligase or Gibson assembly mastermix (New England Biolabs).
Construction of plasmids for complementation. Plasmids to complement ΔflgX strains were constructed by amplifying flgX from codon 2 to the stop codon (with and without a FLAG tag attached to the 3= end prior to the stop codon) from the 81-176 genome, followed by assembly into BamHIdigested pECO102 or pDAR964 to yield pDAR5009, pDAR5011, and pDAR5010. Plasmids to complement ΔmotA and ΔmotB mutants were created by amplifying the coding sequence from codon 2 to the stop codon of motA or codon 2 to the penultimate codon of motB with DNA for a FLAG tag, and a stop codon was added to the 3= end. These fragments were then cloned into the BamHI site of pDAR964 and the BamHI and PstI sites of pECO102, resulting in plasmids pDAR4319 and pDAR2053, respectively. Plasmids to express the motAB loci from C. jejuni 81-176 or S. enterica serovar Typhimurium IR715 were created by amplifying the coding region of motA and motB, which are adjacent to each other on the respective chromosomes. The motAB region from codon 2 of motA through the penultimate codon of motB was amplified with primers that added DNA for a FLAG tag to the 3= end of motB followed by a stop codon. These primers also contained 5= BamHI sites for introduction into BamHI-digested pECO102, resulting in The FlgX Chaperone for Flagellar Stator Units ® plasmids pDAR5266 and pDAR5220. All plasmids were transformed into E. coli DH5␣ and pRK212.1, which served as donor strains for conjugation into C. jejuni. Plasmids were then conjugated into C. jejuni strains as previously described (59). Motility assays. After standard growth for 16 h on MH agar, C. jejuni strains were resuspended from plates in MH broth and diluted to an optical density at 600 nm (OD 600 ) of 0.8. Each bacterial strain was stabbed into semisolid MH motility media containing 0.4% agar using an inoculation needle and then incubated for 24 to 48 h at 37°C under microaerobic conditions. When appropriate, motility agar that contained chloramphenicol was used to maintain plasmids for in trans complementation of mutants.
Generation of antisera. For generation of anti-FlgP rabbit antiserum, recombinant 6ϫHis-tagged FlgP was purified as previously described (60). Purified recombinant protein was used to immunize rabbits for generation of antiserum by a commercial vendor (Cocalico Biologicals).
Protein preparation and immunoblotting analyses. Whole-cell lysates (WCL) were prepared as previously described (61). Fractionation of C. jejuni into subcellular compartments for analysis of protein localization was performed as previously described (62), with slight modifications. Briefly, after growth, C. jejuni strains were resuspended in 1ϫ phosphate-buffered saline (PBS) and diluted to an OD 600 of 0.8 to obtain equal densities of bacteria for all cultures before any fractionation procedures were performed. For WCL, 1 ml of bacterial culture was pelleted, washed once in PBS, and resuspended in 50 l 1ϫ SDS-PAGE loading buffer. To generate periplasmic and cytoplasmic fractions, 20 ml of bacterial culture was washed twice with 2 ml of PBS containing 0.1% gelatin (PBSG) and then resuspended in 2 ml of PBSG containing 20 mg/ml polymyxin B sulfate (Sigma) to compromise the outer membrane and release the periplasmic contents. After centrifugation, the supernatant was saved as the periplasmic fraction, and the recovered pellets consisted of whole spheroplasts. The spheroplast preparation was resuspended in 1 ml 10 mM HEPES and sonicated. Following centrifugation, the supernatants representing soluble cytoplasmic proteins were recovered. For the inner and outer membrane protein fractions, 5-ml aliquots of bacterial culture were pelleted and washed once with 1 ml of 10 mM HEPES. Bacteria were then resuspended in 1 ml of 10 mM HEPES and sonicated. Insoluble material representing total membrane proteins were recovered after centrifugation for 30 min at 16,000 ϫ g. Membranes were resuspended in 10 mM HEPES containing 1% N-lauroylsarcosine sodium salt to solubilize inner membrane proteins. The soluble inner membrane proteins were separated from the insoluble outer membrane proteins by centrifugation for 30 min at 16,000 ϫ g.
For analysis of proteins from E. coli whole-cell lysates, total membrane, and soluble fractions, overnight cultures of E. coli were inoculated at a 1:20 dilution and grown at 37°C with shaking to an OD 600 of 0.8. For whole-cell lysates, 1 ml of culture was collected by centrifugation and resuspended in 50 ml of 1 ϫ SDS-PAGE loading buffer. For separation of soluble and total membrane fractions, 5 ml of culture was collected by centrifugation, washed once in 1 ml of 10 mM HEPES, and then sonicated. After sonication, samples were centrifuged for 30 min at full speed to pellet total membrane fractions, with the supernatant representing the soluble proteins from the cytoplasm and periplasm.
Semiquantitative real-time PCR analysis. WT C. jejuni and isogenic mutant strains were suspended from MH agar plates after growth, total RNA was extracted with RiboZol (Amresco), and RNA was treated with DNase I (Invitrogen). RNA was diluted to a concentration of 5 ng/l before analysis. Semiquantitative real-time PCR (qRT-PCR) was performed using a 7500 real-time PCR system (Applied Biosystems) with gyrA mRNA detection as an endogenous control. For measurements of motA and motB mRNA transcript levels, strain CRG479 (strain 81-176 rpsL Sm /pDAR964) served as the control to determine relative gene expression levels in isogenic mutants. For measurements of ftsH mRNA transcript levels, DAR4803 (strain 81-176 rpsL Sm ΔflgX with deletion of codons 101 to 153) served as the control to determine relative gene expression levels in isogenic suppressor mutants.
In vivo immunoprecipitation of C. jejuni proteins. Coimmunoprecipitation of proteins from C. jejuni strains expressing FLAG-tagged FlgX proteins was performed as previously described (63), with a few modifications. Briefly, after growth on MH agar with appropriate antibiotics for 16 h at 37°C under microaerobic conditions, bacteria from two plates of growth were suspended in PBS and collected by centrifugation. Bacteria were resuspended in 2 ml PBS and then cross-linked by the addition of formaldehyde (0.1% final concentration) for 30 min at room temperature with shaking, followed by quenching with 0.4 ml of 1 M glycine for 10 min. Bacteria were then collected by centrifugation and then disrupted by osmotic lysis with sequential addition of 0.5 ml 200 mM Tris (pH 8), 1 ml 200 mM Tris (pH 8), 1 M sucrose, 0.1 ml 10 mM EDTA, 0.1 ml 10 mg/ml lysozyme, 3 ml double-distilled water (dH 2 O), and 0.3 ml 100 mM phenylmethylsulfonyl fluoride (PMSF) (64). After incubation on ice for 15 min, 5 ml of lysis solution (50 mM Tris [pH 8.0], 10 mM MgCl 2 , 2% Triton X-100) was added. Samples were incubated on ice for 45 min and then centrifuged at 16,000 ϫ g for 20 min. A 30-l volume of anti-FLAG M2 affinity gel resin was added to the supernatant, and the reaction mixture was then incubated at 4°C overnight with agitation. The resin was pelleted by centrifugation at 4°C for 10 min at 6,000 ϫ g followed by 3 washes with radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris [pH 8.0], 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100). For immunoblotting, the resin was resuspended in 70 l SDS-PAGE loading buffer, boiled for 5 min, and analyzed by 10% SDS-PAGE and immunoblotting with specific antisera.
C. jejuni ⌬flgX suppressor analysis. To isolate and identify suppressor mutants that partially restored the motility phenotype of C. jejuni ΔflgX, DAR2340 (strain 81-176 rpsL Sm ΔflgX, which contains a large deletion of flgX from codons 6 to 153) and DARH4803 (strain 81-176 rpsL Sm ΔflgX, which contains a smaller deletion of flgX from codons 101 to 153) were resuspended in MH broth from plates after growth to an OD 600 of 0.8 and stabbed into MH semisolid motility agar. Bacteria were then incubated for up to 10 days at 37°C under microaerobic conditions. Potential suppressor mutants were identified as motile flares that emanated from the point of inoculation of different motility stabs. A small agar plug from the leading edge of motile flares was recovered and subjected to vortex mixing in MH broth. Suppressor mutants from each plug were isolated by plating serial dilutions on MH agar. Three isolates from each agar plug were saved.
Genomic DNAs from a parental C. jejuni ΔflgX strain (DRH4803) and from corresponding suppressor mutant strains were prepared as previously described (65). Briefly, isolates were grown on Mueller-Hinton agar plates supplemented with 10 g/ml trimethoprim at 37°C under microaerobic conditions. Bacteria were harvested, and genomic DNA was isolated using a Qiagen DNeasy blood and tissue kit. The resulting DNA was treated with RNase If (New England BioLabs) and cleaned using a Zymo genomic DNA clean and concentrator kit. Prior to submission for sequencing, the DNA samples were run on a 1.0% agarose gel to check DNA integrity. Genomic DNA from the experiments described above was used to generate bar-coded Bioo NEXTflex DNA libraries at the Indiana University Center for Genomics and Bioinformatics. These libraries were cleaned and verified using an Agilent 2200 TapeStation before pooling and sequencing on the Illumina NextSeq platform were performed. Paired-end reads were demultiplexed before the analysis described below was performed.
Single nucleotide polymorphisms (SNPs) with a minimum variant frequency of 20% were identified in regions of the parent and suppressor mutant genomes with at least 5ϫ coverage. Individual SNPs identified in suppressor mutant genomes were compared to those identified within the parent genome. SNPs that were unique to the suppressor mutant genomes were presumed to be associated with the suppressor mutant phenotypes and investigated further. Identification of other suppressor mutants for which genomic sequencing was not applied involved PCR amplification of one or more suspected genes and then sequencing of PCR products.
To reconstitute the motA H138Y suppressor allele identified in DAR4803 S10 on the chromosome of WT C. jejuni, DNA fragments containing the motAB locus that encode the H138Y mutation with approximately 0.5 kb of upstream or downstream sequence were amplified. The fragments were then assembled into EcoRI-digested pUC19 to yield pDAR5002. pDRH3330 was introduced into DAR2340 (strain 81-176 rpsL Sm ΔflgX), and transformants were obtained on MH agar containing chloramphenicol. Colony PCR verified creation of DAR5007 (strain 81-176 rpsL Sm ΔflgX motA::cat-rpsL). motA::cat-rpsL was replaced with motA H138Y in MB1225 (strain 81-176 rpsL Sm motA::cat-rpsL) and DAR5007 by introduction of pDAR5002 and selection of transformants on MH agar with streptomycin. Potential transformants were screened for chloramphenicol sensitivity and were then verified by colony PCR and sequencing to obtain DAR5148 (strain 81-176 rpsL Sm motA H138Y ) and DAR5112 (strain 81-176 rpsL Sm ΔflgX motA H138Y ).
Chick colonization assays. All uses of animals in this work were approved by the IACUC at the University of Texas Southwestern Medical Center. The ability of wild-type C. jejuni or mutant strains to colonize the ceca of chicks after oral inoculation was determined as previously described (49). Briefly, fertilized chicken eggs (SPAFAS) were incubated for 21 days at 37.8°C with appropriate humidity and rotation in a Sportsman II model 1502 incubator (Georgia Quail Farms Manufacturing Company). Approximately 12 to 36 h after hatching, chicks were orally infected with 100 l of PBS containing approximately 10 4 CFU of WT C. jejuni or an isogenic mutant strain. Strains were prepared for oral gavage by resuspension from MH agar after growth and dilution in PBS to an OD 600 of 0.4 followed by serial dilution to obtain the appropriate inoculum for oral gavage of chicks. The CFU count of the inoculum was determined by serial dilution on MH agar. At day 14 postinfection, chicks were sacrificed and the cecal contents were recovered, weighed, and suspended in PBS to 0.1 g cecal content/ml PBS. Serial dilutions were spread on MH agar containing TMP and cefoperazone. Bacteria were grown for 72 h at 37°C under microaerobic conditions and then counted to determine the CFU per gram of cecal content for each chick. Approximately 50 recovered colonies per chick were stabbed into MH motility agar to determine whether suppressors had developed during in vivo growth. Motility was assessed after 24 to 30 h of incubation at 37°C under microaerobic conditions. Statistical analyses were performed by the use of the Mann-Whitney U test, with statistically significant differences between wild-type and mutant strains indicated with P values of Ͻ0.05.