Enhanced Binding of Altered H-NS Protein to Flagellar Rotor Protein FliG Causes Increased Flagellar Rotational Speed and Hypermotility in Escherichia coli*

H-NS is an Escherichia coli nucleoid protein known only to function as a modulator of gene expression. In this study, we found that specific single amino acid substitutions in H-NS caused an approximately 50% increase in flagellum rotational speed. In fluorescence anisotropy and chemical cross-linking assays, H-NS interacted with the flagellar torque-generating rotor protein FliG to form a complex with a K d of 2.15 μm. Furthermore, one of the altered H-NS proteins that exhibited high speed flagellum rotation bound FliG 50% tighter than wild-type H-NS. These results demonstrate the first non-regulatory role for H-NS and provide a direct correlation between H-NS-FliG binding affinities, flagellar rotation, and motor torque generation.

It has been shown that hns insertion mutations render bacteria non-motile, suggesting that H-NS is a positive regulator of flagellar gene expression (19,20). H-NS is a 15.4-kDa nucleoid-associated DNA-binding protein (21)(22)(23) that modulates the expression of many unrelated genes in E. coli (24 -26). In most instances, such as fimB (27) and proU (28) expression, H-NS acts as a direct transcriptional repressor.
Here we characterized two independent hns point mutations, hnsT108I and hnsA18E, in relation to E. coli motility. Rather than displaying a non-motile behavior, strains carrying these mutations exhibited an unprecedented hypermotility. This novel hypermotile phenotype was a result of enhanced binding of mutant protein H-NST108I to the flagellar rotor, causing increased flagellar rotational speed.
Swarming and Motility-AL127 was transformed with various plasmids. Fresh colonies were picked, inoculated onto low agar (0.3%) tryptone plates, and incubated at 30°C. Swarm diameters were measured every 2-3 h, beginning 9 h post-inoculation. Differences in motility rates were calculated by comparing the swarm diameter of AL127 expressing an hns mutant plasmid relative to the same strain expressing a wild-type hns gene, on the same plate for three individual experiments.
Flagellation-Strains were grown at 30°C overnight in T broth with the appropriate antibiotics and then diluted 25-fold into fresh media and incubated until the A 600 reached 0.3-0.5. Cells from these cultures were incubated on polylysine-coated coverslips for 15 min at room temperature, rinsed twice with phosphate-buffered saline, and fixed in 3% glutaraldehyde, 0.1 M sodium phosphate buffer (pH 7.4). Bacteriacoated coverslips were dehydrated, critical point dried, attached to aluminum stubs, and sputter coated with gold:palladium. Bacterial flagella were visualized by scanning electron microscopy (Cambridge Stereo Scan S200, LEO Electron Microscopy, Inc.). Flagella were counted on at least 10 fields per strain.
Tethering Experiments-Tethering for the flagellar rotational bias and speed experiments was performed as described previously (30), with the following modifications: diluted cultures were grown without inducer until the A 600 reached 0.4 -0.8; flagella were sheared with a Tissue Tearor model 985-370 (Biospec Products, Inc.); and a 1:100 dilution of anti-flagella antibody (36) was used. Bacteria were observed by dark field microscopy, recorded to videotape, played back at 1/5 the speed, and scored using The Observer 3.1 videotape analysis system software (Noldus Information Technology). Typically, 30 individual cells were quantitated for 30 s each for CCW bias or 1 min each for flagellum rotational speed.
Protein Purifications-E. coli wild-type H-NS and H-NST108I were each purified as described before (27) except the mutant protein lysate was washed from the double stranded DNA-cellulose column at a lower salt concentration (125 versus 250 mM NaCl) than the wild-type cell lysate. His-tagged FliG was nickel-affinity purified basically by the method of Toker and Macnab (31) from BL21(DE3) cultures harboring pET28NdefliG. Protein solutions were concentrated with Centricon-10 columns as instructed by the manufacturer (Amicon). Protein concentrations were measured with the Bio-Rad Dc protein assay kit (Bio-Rad).
Fluorescence Anisotropy-H-NS was labeled with the thiol-reactive probe, fluorescein 5-maleimide, after reduction with tris-(2-carboxyethyl)phosphine hydrochloride, according to the manufacturer's instructions (Molecular Probes). Labeled protein was separated from excess probe by G-25 Sephadex protein spin column purification (Boehringer Mannheim). Anisotropy was measured at excitation and emission wavelengths of 490 and 515 nm, respectively. Data was collected on a LS 50B Luminescence Spectrometer with FL Data Manager software (Perkin-Elmer). The K d was calculated as the FliG concentration corresponding to one-half the maximum anisotropy value on the H-NS-FliG binding curve.
Chemical Cross-linking-FliG was first attached to the heterobifunctional cross-linker, sulfosuccinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate (Pierce) by incubating for 1 h at room temperature. Reactions were quenched by the addition of 1 M Tris (pH 7.5), H-NS (wild-type or T108I) was added, and the entire reaction was incubated another hour. The cross-linking reactions were terminated by the addition of 5ϫ Laemmli denaturing sample buffer. Reactions containing equal amounts of each protein (5 M) were electrophoresed on a 4 -20% denaturing gradient gel (Jule Biotechnologies, Inc.) and visualized by Coomassie or Sypro-Orange (Molecular Probes) staining. Complexes were quantitated by volumetric integration on a Gel-Doc 1000 with Molecular Analyst 2.1 software (Bio-Rad). To ensure that we added equal amounts of wild-type H-NS and H-NST108I to FliG, we also quantitated free, uncomplexed H-NS protein.
Antibody Production and Western Blotting-A C-terminal 6xHis-H-NS fusion protein was expressed from pDMG1 and purified under denaturing conditions according to the manufacturer's instructions (Qiagen). An emulsion of TiterMax (Vaxcel) and 300 g of 6xHis-H-NS was used to inoculate and boost New Zealand White rabbits. Anti-H-NS antiserum obtained was used at a 1:5000 dilution. Anti-FliG antiserum was kindly provided by David Blair (University of Utah, Salt Lake City, UT) and used at a 1:7000 dilution. H-NS-FliG cross-linked complex reactions were run on 12% SDS-polyacrylamide gel electrophoresis, transferred, probed with antiserum, and detected as described previously (27).

Effect of hns Mutations on E. coli Motility-H-NS is a global
regulator of E. coli gene expression (reviewed in Refs. 32 and 33). Mutations in hns have pleiotropic effects on the cell altering synthesis of a variety of gene products involved in numerous, diverse biological pathways (24 -26). In particular, trans-poson insertions in hns result in the loss of bacterial motility due to the lack of master operon flhCD expression (20). We isolated a set of random hns mutations 2 and determined their effect on motility. Plasmids carrying either no insertion (pA-CYC vector alone), a wild-type hns gene (pTHK116), or hns point mutations T108I (pMASS46 -1), or A18E (pMASS73-4) were separately introduced into AL127, an E. coli hns2-tetR (34) insertion mutant background strain (Table I). Transformants were inoculated onto semi-solid agar plates and motility was determined by measuring the diameter of the bacterial swarm over time. Motility was classified into three distinct categories based on swarm size (Fig. 1). Strains lacking H-NS were non-motile. Motility was restored in these strains upon the addition of a wild-type hns clone in trans. However, the hnsT108I and hnsA18E mutations each conferred a greater than 2-fold increase (2.1 Ϯ .21 and 2.5 Ϯ .17, respectively) in swarm rate as compared with the wild-type strain. These data represent the first instance in which any E. coli mutation, hns or otherwise, resulted in a hypermotile phenotype.
Flagellation-H-NS is a positive transcriptional regulator of the flhCD operon, the master operon which controls expression of all other flagellar components (1). In the absence of H-NS, flagellar genes are not expressed resulting in the loss of intact flagella and motility (20). One explanation for the hypermotile phenotype that we observed with specific hns point mutations was that these alleles affected flagella biosynthesis and/or assembly. To test this hypothesis, we examined strains expressing different hns alleles by scanning electron microscopy (SEM). Visually, there was no discernible difference in cell size or shape or flagellum length or distribution between strains carrying a wild-type hns allele ( Fig. 2A) or either hns mutation (Fig. 2, B and C). There was also no statistical difference in the average number of flagella/cell between the control and hnsA18E strains and only a very slight increase expressed from the hnsT108I allele (Table II). We concluded that the increased motility rates exhibited by the hnsT108I and hnsA18E alleles were not due to an alteration in the number or construction of functional flagella.
Flagellar Rotational Behavior-E. coli swim by rotating their flagella filaments (35,36) either CCW or CW (7). Motion is an alternating series of smooth runs (CCW rotation) and abrupt directional changes called tumbles (CW rotation) (6). It is possible that the hnsT108I and hnsA18E mutations caused a shift in flagellar rotational bias resulting in hypermotility.  Strains bearing hns mutant alleles may "run" (versus tumble) a higher percentage of the time than their wild-type counterparts. To address this issue, we performed tethering experiments (30) to compare the rotational bias of strains containing wild-type or mutant hns genes. In this assay, individual cells harboring plasmid-based hns alleles were tethered to a microscope slide via an anti-flagella antibody. The spinning direction of cell bodies was observed and bias was calculated to be the proportion of the time cells spent rotating CCW (Table III). Wild-type cells generally spin CCW with occasional pauses and direction reversals (36). Both of the wild-type controls (Table III, lines 1 and 4) adopted this spinning mode, rotating CCW 85% of the time. Additional controls included a "gutted" strain (RBB1041) which is deleted for the chemotaxis genes and spins only in the default CCW direction (37) (line 2), and a CW-biased cheZ mutant (RP1616) (38) which only tumbles (line 3). Strains with hns mutant alleles T108I or A18E (lines 5 and 6) did not favor a higher running (CCW) bias over wild-type strains. In fact, these strains along with two other hns mutations which did not exhibit a hypermotile phenotype (data not shown), tumbled or paused slightly more frequently than wildtype cells (lines 7 and 8). Thus, the hns mutations did not affect the function of the chemotaxis apparatus or the mechanism of flagellar switching.
Flagellar Speed-Flagella propel bacteria by rotating motordriven helical filaments (35,36) whereby swimming speed is directly related to flagellar rotational speed (39). Free-swimming bacteria can transverse 20 -60 m/s in liquid media (40), whereas tethered bacteria can rotate their cell bodies 2-9 revolutions/s (36). We postulated that the hnsT108I and hnsA18E mutations caused hypermotile bacterial swimming behavior by directly increasing the speed at which individual flagella rotated. Once again, we employed bacterial tethering to measure rotational speed using a host strain (AL128) containing a cheA-cheZ deletion as well as an hns2-tetR mutation. This strain enabled us to survey cells that were all rotating in one direction, CCW, without the complications of switching. Rotational speed data (Table IV) were accumulated from cells of the same size and tethered at the same point to maintain similar drag coefficients, overall cell geometry, and load. The controls (Table  IV, lines 1 and 2) displayed similar rotational speeds regardless of the location of hns (chromosome-or plasmid-based). In contrast, the addition of either hns point mutation in trans (lines 3 and 4) resulted in significant (p Ͻ .005 and p Ͻ .025, respectively) increases in the speed of the tethered cell body. We concluded that hnsA18E and hnsT108I accelerated flagellar speeds 44 -62% (Table IV) over wild-type levels without affecting flagellar assembly ( Fig. 2 and Table II) or switching (Table  III). This direct effect on flagellum rotational speed likely caused the original swarm plate hypermotile phenotype displayed by these hns mutant alleles.  with each other (11,12) to form a switch complex peripherally attached to the inner membrane, facing the cytoplasm (41)(42)(43). All three proteins are involved in flagellar assembly, switching, and rotation (13,14). Since FliG is the motor protein most involved in speed and torque generation (12,14,15), we examined the binding between H-NS and FliG in fluorescence anisotropy assays. This technique quantitates protein-protein interactions by measuring the change in anisotropy of a fluorescently labeled protein upon the addition of a second unlabeled protein (44).
Increasing amounts of purified N terminus His-tagged FliG was added to fluorescein-labeled H-NS and emission anisotropy of the fluorophore was assessed (Fig. 3). As the concentration of FliG increased the anisotropy values also increased, reflecting the formation of a slower rotating complex and protein binding. The leveling off of the curve at higher FliG concentrations demonstrated protein binding saturation (45). We calculated the dissociation constant (K d ) for the H-NS-FliG complex to be 2.15 Ϯ .25 M, well within the range of biologically relevant protein associations. These data represent the first direct biochemical evidence that H-NS binds to FliG and forms a complex with the flagellar motor machinery.
H-NS-FliG Binding via Chemical Cross-linking-Since wildtype H-NS bound FliG in the anisotropy studies (Fig. 3), we wanted to determine if the mutant H-NST108I also bound FliG. A difference in binding affinities between FliG and wildtype or mutant H-NS protein might account for the hypermotile swarm phenotype seen with the mutant. We compared binding capabilities of each of these proteins to FliG by chemical cross-linking.
Equal amounts of either wild-type or T108I H-NS were crosslinked to FliG. H-NS-FliG complexes were trapped, run on SDS-polyacrylamide gels and stained (Fig. 4A). Tandem Western blots were also performed on identical reactions with anti-H-NS (Fig. 4B) and anti-FliG antiserums (Fig. 4C) to confirm that the complexes contained both proteins. Several conclusions can be drawn from these results. First, in agreement with previous work (11,12), FliG self-associated (Fig. 4C) in the absence (reaction 3) and presence (reaction 6) of cross-linker. Second, both wild-type and T108I H-NS dimerized (Fig. 4A,  reactions 4 and 5). Also, wild-type H-NS (Fig. 4, A, B, and C, reaction 7) and H-NST108I (Fig. 4, A, B, and C, reaction 8) each bound to FliG forming a complex of approximately 55 kDa. Furthermore, H-NST108I bound FliG 50% tighter than wildtype H-NS (compare Fig. 4, A-C, reactions 7 and 8) at the same protein molar ratios. Although H-NS and FliG obviously bound to each other in vitro, this complex did not seem to be the major FliG interaction since most of the protein remained in the monomer form (Fig. 4, A and C, reactions 7 and 8). These results demonstrated that the hnsT108I mutation conferred an enhanced binding affinity of H-NS for FliG. This increased attachment of H-NST108I to the torque-generating flagellar rotor protein FliG probably accounted for the increased motility rates and flagellum rotational speeds in E. coli strains carrying the hns mutation. DISCUSSION Hypermotility-We studied the effect of two hns point mutations, hnsT108I and hnsA18E, on E. coli motility. In swarm plate assays, we observed that each of these hns mutations bestowed a hypermotile phenotype (Fig. 1). Swarm rates for strains carrying the hns mutant alleles were approximately twice as fast as strains with the wild-type hns allele in trans. Previously characterized mutations in genes involved in flagella functions have led to three basic mutant phenotype classifications: non-flagellated (Fla Ϫ ) (15), non-rotating flagella ("paralyzed", Mot Ϫ ) (14), and skewed chemotaxis, either CCWbiased (14) or CW-biased (46). In terms of swarming, mutations typically cause a decrease in motility (47). To the best of our    knowledge, hnsT108I and hnsA18E are the first documented mutations that cause a 2-fold increase in swarm rates.
The mutations studied here represent single H-NS amino acid substitutions of a C-terminal threonine to isoleucine (T108I) and an N-terminal alanine to glutamic acid (A18E). The N-terminal domain of H-NS is ill defined but the C-terminal third of the protein encompasses the DNA-binding domain (48,49). If H-NSA18E acts by the same mechanism as H-NST108I to increase flagella speed then these mutants may define a FliG-specific binding domain on the H-NS surface. Residues at position 18 and 108 may form a pocket which is near or embodies the actual FliG-binding site. Thus, alterations in these amino acids may enhance FliG binding directly or cause rearrangements in adjacent residues. This possibility suggests that the N and C termini of H-NS are close to each other in the three-dimensional protein structure.
Three ways in which altered H-NS protein could affect motility are by changing the (i) number of flagella assembled, (ii) flagellar directional bias, or (iii) spinning speed of flagella. H-NS is a positive transcriptional activator of the flagellar master operon (20,50). Thus, it is possible that altered H-NS could be a stronger activator and up-regulate the expression of flagellar structural components. Upon inspection, it was evident that all strains demonstrated the same flagella profile (Fig. 2) and only mutation hnsT108I exhibited a minor increase of 0.7 flagella/cell (Table II). We addressed the issue of whether cells bearing the mutant hns alleles were running their flagella CCW a longer proportion of time. Regardless of their motility phenotype, multiple hns mutant alleles all displayed similar CCW biases that were not greater than wild-type cells (Table  III). Finally, we thought that faster rotating flagella could perhaps account for the increase in swarm motility. In modified tethering assays, we showed that hnsT108I-and hnsA18Econtaining strains did indeed rotate their flagella an average of 53% faster than wild-type cells (Table IV).
H-NS-FliG Interactions-Since our hns mutations seemed to be specifically affecting flagellar speed, we investigated the possible interaction between H-NS and the protein complex that controls rotation. Of the three proteins that make up the rotor portion of the motor, we chose to test FliG for the following reasons. First, FliG is situated at the cytoplasmic face of the inner membrane attached to the embedded MS ring through interactions with FliF (41)(42)(43). This location allows free access to cytoplasmic H-NS. Second, FliG is presumably part of the active flagellar rotating unit along with the MS ring, rod, hook, and filament (3,12,51). Thus, H-NS binding directly to a portion of the moving apparatus could potentially alter speed. Furthermore, FliG functions primarily in flagella rotation and torque generation (12,14,15) rather than assembly or switching. Finally, there is previous genetic data indicating that fliGand hns-encoded fusion proteins interacted in a yeast twohybrid system (11).
In our anisotropy studies, it was evident that wild-type H-NS bound FliG, yielding a typical binding curve (Fig. 3). We estimated the K d to be in the micromolar range, confirming that the H-NS-FliG interaction was significant and biologically relevant. Our data represent the first biochemical evidence to support Marykwas' (11) genetic-based conclusion that H-NS binds to FliG. It also represents the first time H-NS has directly been shown to bind an E. coli protein other than itself or homologs. As a modulator of gene expression, H-NS usually functions by binding DNA and exerting an effect on transcription (reviewed in Refs. 32 and 33). It has been suggested by suppressor mutation isolation (52) and co-purification (53) that H-NS may interact with other E. coli proteins. However, the only other protein proven to bind H-NS is the bacteriophage T7 gene 5.5 protein product (54).
We carried out cross-linking reactions (Fig. 4) in order to compare the relative binding affinities of FliG for either wildtype H-NS or H-NST108I. We were careful to equalize and monitor H-NS quantities such that any differences in complex intensity were due to binding tightness rather than unequivalent protein concentrations. Each protein dimerized and FliG bound each H-NS species forming 55-kDa heterodimeric complexes, with mutant H-NST108I binding FliG 50% tighter than wild-type H-NS. We are confident that the complexes observed represent an H-NS-FliG interaction since the same overlaid bands were recognized by both anti-H-NS and anti-FliG antiserum (Fig. 4, B and C).
There have been many studies examining the flagellar motor proteins yet H-NS has not been previously found in these complexes. This missing observation may be because nonflagellar components were not assayed for, or the H-NS-FliG binding may be transient, such that, a complex would have to be trapped by a covalent cross-linker rather than pulled down via co-precipitations. The caveat of such a cross-linking experiment is that we are not truly mimicking conditions in vivo. In the cell, FliG exists in multiple copies per motor (43) and is known to interact with other proteins such as FliM, FliN, and MotA (12). Therefore, the addition of any of these proteins in our cross-linking assay could alter H-NS-FliG binding mechanisms. Regardless, it seems likely that the hypermotility exhibited by the hns mutation is a direct consequence of enhanced binding of H-NST108I to the flagellar rotor protein, FliG. Models-No existing model sufficiently accounts for the unexpected effects of H-NS on flagella rotation. Thus, we present a speculative hypothesis of H-NS action (Fig. 5) based on the data presented here as well as previous investigators results (10 -12, 31, 41). We postulate that H-NS is involved in torque generation through its interactions with FliG. Given the fact that (i) hns mutations resulting in the hypermotile phenotype specifically affected flagellar rotational speed, (ii) speed, torque, and presumably proton-motive force are all directly related (39,55), and (iii) the C-terminal domain of FliG functions specifically in torque generation (15), we position H-NS at the interface between the rotor and stator, directly linked to the C terminus of FliG (Fig. 5A). Tighter binding of mutant H-NST108I to FliG (Fig. 5B) may cause increases in flagellar speed by altering the conformation of FliG relative to the other rotor proteins and/or the MotA⅐B complex, thus, compacting the motor complex and allowing faster rotation by creating less friction within the surrounding stationary MotA⅐B ring complex (56). Alternatively, the movement or rate of proton flux through this channel could also be affected, or the H-NS-FliG complex may play a regulatory role by altering expression of genes downstream of fliG in the flagellar assembly cascade.
In totality, this study provides several unique observations about H-NS and E. coli motility. We have defined a new swarming mutant class illustrated by a hypermotile phenotype. We have shown that H-NS plays a rare, non-regulatory role in motility. We provided the first biochemical evidence that H-NS binds to the flagellar rotor protein FliG and demonstrated that the tightness of this interaction determines flagellar rotational speed. Future directions will include testing our model in order to provide a suitable mechanistic view of H-NS activity on bacterial motility.