Organization of FliN Subunits in the Flagellar Motor of Escherichia coli

FliN is a major constituent of the C ring in the flagellar basalbody of many bacteria. It is present in >100 copies per flagellumand together with FliM and FliG forms the switch complex thatfunctions in flagellar assembly, rotation, and clockwise-counterclockwiseswitching. FliN is essential for flagellar assembly and switching,but its precise functions are unknown. The C-terminal part ofthe protein is best conserved and most important for function;a crystal structure of this C-terminal domain of FliN from Thermotogamaritima revealed a saddle-shaped dimer formed mainly from ßstrands (P. N. Brown, M. A. A. Mathews, L. A. Joss, C. P. Hill,and D. F. Blair, J. Bacteriol. 187:2890-2902, 2005). Equilibriumsedimentation studies showed that FliN can form stable tetramersand that a FliM₁FliN₄ complex is also stable. Here, we haveexamined the organization of FliN subunits by using targetedcross-linking. Cys residues were introduced at various positionsin FliN, singly or in pairs, and disulfide cross-linking wasinduced by oxidation. Efficient cross-linking was observed forcertain positions near the ends of the dimer and for some positionsin the structurally uncharacterized N-terminal domain. Certaincombinations of two Cys replacements gave a high yield of cross-linkedtetramer. The results support a model in which FliN is organizedin doughnut-shaped tetramers, stabilized in part by contactsinvolving the N-terminal domain. Electron microscopic reconstructionsshow a bulge at the bottom of the C-ring whose size and shapeare a close match for the hypothesized FliN tetramer.

INTRODUCTION

Top

Introduction

Many species of bacteria are propelled by rotating flagella(2) that obtain energy from the membrane ion gradient (16, 23;for reviews, see references 1 and 21) The flagella are builtfrom about 25 different proteins, most of which have roles informing the flagellar basal body, hook, and filament (22, 26).Only a few proteins function in rotation and switching. Themembrane proteins MotA and MotB form ion-conducting complexesthat function as the stator (3, 8-11, 33). The proteins FliN,FliM, and FliG form the rotor-mounted switch complex that functionsin flagellar assembly, rotation, and clockwise-counterclockwise(CW/CCW) direction control (44, 45). The switch complex contains>100 copies of FliN (47), about 34 copies of FliM (47), and25 copies of FliG (13, 18, 35, 36, 47). Together, these formthe C ring, a 45-nm-diameter, drum-shaped structure at the bottom(the cytoplasm-proximal end) of the basal body (Fig. 1) (14,19, 40).

View larger version (26K):
[in this window]
[in a new window]
FIG. 1. Cartoon of the flagellar basal body indicating the approximate locations of the proteins that form the stator (MotA and MotB) and the switch complex (FliG, FliM, and FliN). OM, outer membrane; PG, peptidoglycan; IM, inner membrane.

FliG functions directly in rotation of the flagellum (17, 24).It interacts with the FliF protein that forms the MS ring (13,15, 20, 30) and with the membrane-associated stator proteinMotA (48) and must presumably be located in the upper (membrane-proximal)part of the C ring (Fig. 1). FliM and FliN have not been locatedprecisely but are probably more distant from the membrane becausethey do not function as directly in rotation (24) and are notbelieved to interact directly with FliF or MotA. FliM is closelyinvolved in direction switching (34) and contains a bindingsite for the CW-signaling protein CheY (28, 41, 43). FliN hasbeen implicated in CW/CCW switching (17) and in the export processthat occurs during flagellar assembly (42), but its exact functionsare unknown.

The FliN protein of Escherichia coli or Salmonella contains137 residues (27). The C-terminal part of the protein, fromabout residue 58 to the end, is best conserved and most importantfor function (37). A FliN protein lacking 57 N-terminal residuesis sufficient for flagellar assembly and rotation, but the cellsswarm at about one-third of the wild-type rate in soft agar(37) and show relatively slow, erratic motility in liquid (K.Paul and D. Blair, unpublished data). In the flagellum, FliNinteracts with the C-terminal domain of FliM (FliM_C) (28, 41).Portions of FliM_C show sequence homology to FliN (28) and maytherefore show some similarity in folding. FliN might also bindto FliG, but the evidence for this is less strong (39).

Two crystal structures were recently solved for most of theFliN protein from Thermotoga maritima (Protein Data Bank [PDB]entries 1o6a and 1yab) (5). Both structures show a saddle-shapeddimer formed mainly from ß strands, with short ${alpha}$ helicesat the ends (Fig. 2). The two structures differ somewhat inthe position of ${alpha}$ -helix 1, which packs near the body of the dimerin one crystal (1o6a) but extends to contact an adjacent dimerin the other (1yab). The FliN dimer exhibits a sizable surfacehydrophobic patch, formed from the side chains of 5 residues(10 residues in the dimer) with a strongly conserved nonpolarcharacter. A mutation in the hydropobic patch (Val113 $->$ Asp) affectsboth flagellar assembly and switching (5).

View larger version (44K):
[in this window]
[in a new window]
FIG. 2. (A) Crystal structure of FliN (PDB accession no. 1yab), with the hydrophobic patch indicated by gray coloring. (B) Differing positions of helix 1 (colored gray) in the two FliN crystal structures. In the 1o6a structure, the two helices of the dimer have different apparent lengths because the structural model begins at residue 68 of one subunit and residue 74 of the other. (C) Model for the FliN tetramer based upon the subunit organization seen in the crystal structure of HrcQB_C (12). (D) Example of a dimer-dimer contact occurring in the FliN crystal structure 1o6a. (E) Ring-shaped tetramer of FliN subunits seen in the crystal structure 1yab. Panels A and B were prepared using Pymol, and panels C and D were prepared using Rasmol.

Equilibrium sedimentation experiments showed that the FliN proteinof E. coli forms a stable tetramer in solution and that theFliN and FliM proteins of T. maritima form a stable complexwith the composition FliM₁FliN₄ (5). (Sedimentation experimentswith FliM used the T. maritima proteins because E. coli FliMis prone to aggregation.) Accordingly, we proposed that FliM₁FliN₄complexes might be structural units of the C ring (5). The subunitcomposition of the C ring is not yet firmly established, however;an estimate of subunit stoichiometry in purified basal bodiesof Salmonella enterica serovar Typhimurium gave a FliM:FliNratio closer to 1:3 (47). Because FliM_C shows sequence similarityto FliN, it might be capable of replacing a FliN subunit toform a FliM-FliN heterodimer, and such heterodimers in combinationwith FliN dimers could account for a 1:3 subunit composition.The location and arrangement of the FliN subunits in the C ringare also unknown. The related protein HrcQB_C crystallized asan elongated tetramer in which the dimers are joined end toend (12). Patterns of residue conservation indicate that FliNdimers might associate in a similar end-to-end fashion (5),but the subunit associations seen in FliN crystals indicatethat other arrangements are also possible, including some withthe shape of a ring (Fig. 2C to E).

Here, we have used targeted cross-linking to test various modelsfor the organization of FliN. Cys residues were introduced invarious positions on the surface of FliN, singly or in pairs;disulfide cross-linking was induced; and the products were analyzedon immunoblots. The subunit replacement model was tested bymaking Cys replacements at predicted interfacial positions inFliN and FliM_C. The results indicate that FliN is organizedin tetramers and does not form subunit-exchanged heterodimerswith FliM_C. The FliN tetramer is not organized like the tetramerseen in the HrcQB_C crystal structure but has a doughnut shapesimilar to the subunit organization seen in one of the FliNcrystal structures. The FliN doughnut seen in the crystal structureis puckered and as a result is predicted to be bistable (toexist in two equally stable conformations). In the flagellum,the two conformations would be nonequivalent and might be relevantfor CW/CCW switching. Some Cys replacements were also made inthe structurally uncharacterized N-terminal domain of FliN.Occurrences of efficient cross-linking show that the N-terminalsegments of two FliN subunits are in proximity to each otherand are near the hydrophobic patch of another dimer. These interactionsappear to stabilize the tetramer, because a protein lackingthe N-terminal domain formed mainly dimers. Electron microscopicreconstructions (40, 46) show a bulge at the bottom of the Cring of the appropriate size and shape to fit the FliN tetramers.

MATERIALS AND METHODS

Top

Materials and Methods

Strains and media. The E. coli strains and plasmids used are listed in Table 1.Procedures for transformation and plasmid isolation were describedpreviously (38). TB contained 10 g tryptone and 5 g NaCl perliter. SB contained, per liter, 12-g tryptone, 24-g yeast extract,5-ml glycerol, 3.8-g KH₂PO₄, and 12.5-g K₂HPO₄. Ampicillin (Amp)was used at 125 µg/ml in liquid medium and at 50 µg/mlin swarm plates. Chloramphenicol was used at 50 µg/mlin liquid and 12.5 µg/ml in swarm plates. IPTG (isopropyl-ß-D-thiogalactopyranoside)was prepared as a 0.1 M stock in water.

View this table:
[in this window]
[in a new window]
TABLE 1. Strains and plasmids

Site-directed mutagenesis and assays of swarming. Single-stranded DNA preparation and site-directed mutagenesiswere carried out according to the Altered Sites procedure (Promega),with fliN cloned in plasmid pLS4 (24) or with fliM cloned inpHT41 (38). The mutations were confirmed by DNA sequencing.The mutated fliN and fliM genes were transferred into plasmids(pHT39 and pDFB97, respectively) that allow IPTG-regulated expressionfrom the tac promoter. Swarming in soft agar, swimming in liquid,and flagellation were assayed as described previously (38) byusing the fliN deletion strain DFB223 transformed with wild-typeor mutant plasmids. Swarm plates contained TB, 0.27% agar, andan appropriate antibiotic(s).

Cross-linking. Initial cross-linking experiments used the catalyst Cu[1,10-phenanthroline]₃(Cu-phenanthroline) and were performed as described previously(4), with minor modifications. fliN deletion strain DFB223 orflhDC deletion strain RP3098 was transformed with wild-typeor mutant plasmids. RP3098 expresses no flagellar genes, exceptthose expressed from plasmids, and thus does not assemble flagella.Cells were cultured overnight in TB and antibiotics at 32°C,then diluted 100 fold into TB containing appropriate antibiotics,and grown for 1 h at 32°C. The culture was then inducedwith 25 µM IPTG, and growth was continued for an additional3.5 h. The absorbance of each culture was measured at 600 nm.The least-dense culture was transferred to a centrifuge tubein its entirety, and the other cultures were transferred involumes adjusted (by using A₆₀₀ readings) to give equal numbersof cells. The cells were pelleted (3,000 x g; 10 min) and resuspendedin 200 µl lysis buffer (50 mM Tris [pH 8], 0.5 M sucrose,10 mM EDTA, 0.2-mg/ml lysozyme). The cells were incubated onice for 1 h and then rapidly diluted by the addition of 1.8ml of ice-cold water. Samples were divided into two 1-ml fractions,and 110 µl of cross-linking reagent was added to one ofthe tubes and 110 µl of 50% ethanol was added to the other.The cross-linking reagent contained 4 mM CuSO₄ and 16 mM 1,10-phenanthrolinein 50% ethanol and was freshly prepared from a 1 M stock of1,10-phenanthroline in 95% ethanol and a 400 mM stock of CuSO₄in water. Samples were incubated at room temperature with gentlerotation for 5 min, sonicated for 15 s (50% duty cycle, powerlevel 2; Branson model 450 sonifier), and incubated an additional5 min. Reactions were quenched by the addition of N-ethylmaleimide(22 µl of a 1 M stock in 95% ethanol) and EDTA (126 µlof a 0.5 M stock). A total of 100 µl of each sample wasmixed with 100 µl of nonreducing gel-loading buffer, boiled,and used for electrophoresis.

For some cross-linking experiments using iodine, cells werecultured, collected as described above, and then treated with0.2 mM I₂ added from a 2 mM stock in 95% ethanol. N-Ethylmaleimidewas added after 5 min, and then cells were mixed with gel loadingbuffer (lacking reducing agent), boiled, loaded on gels, andanalyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis(SDS-PAGE) and immunoblotting, as described below.

SDS-PAGE and immunoblotting. Protein samples were separated on 10% or 12% SDS-PAGE minigels(Bio-Rad MiniProtean system). Proteins were transferred to nitrocelluloseusing a semidry transfer apparatus (Bio-Rad). Rabbit polyclonalantibody against FliN was prepared as described previously (37)and was used at an $~$ 1,000-fold dilution. Bands were visualizedusing the Super Signal West Picoluminol system (Pierce) andX-ray film. Bands were quantified by video densitometry usingthe public domain image-processing program NIH Image (developedat the U.S. National Institutes of Health and available onlineat http://rsb.info.nih.gov/nih-image/).

Gel filtration chromatography. The fliN deletion strain DFB223 was transformed with plasmidsencoding either full-length FliN or FliN lacking 57 N-terminalresidues. Cells were grown at 37°C in 500 ml SB-Amp brothto an absorbance of 0.7 at 600 nm, then IPTG was added to 400µM, and growth was continued for 3 h. Cells were pelletedand resuspended in 40 ml of buffer containing 50 mM Tris-HCl(pH 8.0), 500 mM NaCl, 1 mM EDTA, and 1 mM phenylmethylsulfonylfluoride. Cells were lysed by passage through a French pressurecell, and the lysate was centrifuged (23,000 x g; 45 min) topellet the membranes. The supernatant was collected in a freshtube. A 2-ml sample of the supernatant was filtered and loadedonto a Superdex-300 size exclusion column in a fast-performanceliquid chromatography apparatus (ÄKTA; Amersham). The columnwas run in 50 mM Tris-HCl (pH 8.0)-200 mM NaCl, and fractionswere collected and analyzed by SDS-PAGE and immunoblotting.

RESULTS

Top

Results

Shape of the FliN tetramer. The crystal structure is known for residues 68 to 154 of T.maritima FliN (5) (PDB accession numbers 1yab and 1o6a), correspondingto residues 51 to 137 of the E. coli protein. The cross-linkingexperiments used the protein from E. coli. The wild-type proteincontains no Cys residues. In the part of FliN of known structure,10 Cys replacements were made on the surface and 1 was madeat a more interior position near the dimer twofold axis. SeveralCys replacements were also made in the N-terminal domain, whichhas not yet been crystallized. The positions of the mutationsare shown in Fig. 3, either on the crystal structure (for positions54 and higher) or on a sequence alignment of the N-terminalsegment (for the replacements in the N-terminal domain). A secondary-structureprediction for the N-terminal part of the protein is also shown.The Cys replacements in the N-terminal domain were in a segmentthat is relatively well conserved and predicted to form an ${alpha}$ helix.

View larger version (63K):
[in this window]
[in a new window]
FIG. 3. (A) Positions of Cys replacements on the FliN dimer. The structure shown is 1yab (5). Labels in roman type are for subunit B and those in italic type are for subunit A. (B) Positions of the Cys replacements in the N-terminal domain (arrows). The structure of this domain is not known; secondary structures predicted by a neural-net algorithm (32) are shown on a sequence alignment of the N-terminal segments from some closely related species, with springs indicating ${alpha}$ helices and lines indicating coil. S.t., Salmonella typhimurium; Y.p., Yersinia pestis.

First, we examined cross-linking of proteins with single Cysresidues on the surface of the domain of known structure. Onedouble mutant (77C/83C) was also studied, because these positionsare predicted to be near each other in the HrcQB_C-based modelfor the FliN tetramer (5, 12). The Cys-substituted FliN proteinswere expressed in the fliN deletion strain DFB223, cells werelysed, and copper phenanthroline was added to catalyze oxidativecross-linking. The deletion in strain DFB223 is fully complementedby plasmid-encoded fliN (37), and so the cells in these experimentswere flagellate. Parallel experiments were also conducted withan flhDC strain (RP3098) that does not express any chromosomalflagellar genes and that does not assemble flagella to determinewhether similar subunit contacts occur when the protein is notassembled into motors. Proteins were resolved on SDS-PAGE gels,and the products of cross-linking were characterized on FliNimmunoblots.

Wild-type FliN and most of the Cys-replacement proteins showedno cross-linking when treated with Cu-phenanthroline. Just fourof the mutant proteins, with Cys replacements at positions 54,57, 58, and 136, showed cross-linking upon oxidation (Fig. 4).(Numbers are for the E. coli protein; T. maritima numbers arehigher by 17.) Each showed a single major cross-linked productwith an apparent molecular weight of 42 kDa. Although this waslarger than expected for a normally migrating FliN dimer (predictedmass, 31 kDa), cross-linked species often exhibit aberrant mobility,and other evidence described below supports its assignment asa FliN dimer. These four positions are near the ends of theFliN dimer (Fig. 3), in a region that is the major site of contactbetween dimers in the 1yab FliN crystal structure (Fig. 2E)(5). This region did not contribute to the dimer-dimer interfacein the HrcQB_C-based model for the FliN tetramer (see Fig. S1Ain the supplemental material). Cross-linking results were similarfor the flagellum-forming strain and the flhDC strain (Fig.4). Thus, the cross-linked product did not contain any flagellarproteins besides FliN, and the contacts through positions 54,57, 58, and 136 did not require the protein to be assembledinto flagella. Given the location of these residues at the endsof the (crystallographic) FliN dimer, the cross-linking wasnot between the two subunits of this dimer but must occur withina larger assembly—presumably, the FliN tetramer knownfrom sedimentation equilibrium experiments (5).

View larger version (31K):
[in this window]
[in a new window]
FIG. 4. Efficient cross-linking by four Cys replacements on the FliN domain of known structure. (A) Cross-linking by Cu[1,10-phenanthroline]₃. The experiments used either the fliN-null strain DFB223 or the flhDC strain RP3098, as indicated. DFB223 is flagellate when it expresses the FliN protein from the plasmid, whereas strain RP3098 is nonflagellate, and so the FliN is not present in flagellar motors. Experiments used either a 10% gel (57C) or 12% gels (58C, 54C, and 136C). Products of cross-linking, here and in panel B, are visualized on anti-FliN immunoblots. (B) Cross-linking of the same proteins by iodine. The experiment used the fliN-null strain DFB223 and 10% polyacrylamide gels. Iodine was used at 0.4 mM for the 54C and 57C proteins and 0.2 mM for the 58C and 136C proteins. (These were chosen as those showing the highest yield, in a series of experiments using 0.2, 0.4, or 0.6 mM.)

Fluctuations in protein position or orientation might allowcross-linking between sites that are somewhat separated in thelowest-energy configuration (7). Previous studies found thatsuch motional effects are minimized when iodine rather thancopper phenanthroline is used for oxidation (4, 6); so to placetighter constraints on residue proximity, we examined cross-linkingby iodine. Although yields were lower than those with copperphenanthroline, the proteins with Cys replacements at position54, 57, or 136 showed a readily detectable cross-linked productupon treatment with iodine. Position 58, which showed a loweryield than the others with copper phenanthroline, did not showany cross-linking using iodine.

FliN and FliM_C do not form a subunit-exchanged heterodimer. As discussed above, sequence similarities between FliN and FliM_Craise the possibility that FliM-FliN heterodimers form by replacementof a FliN subunit by FliM_C. To test this subunit-replacementmodel, we introduced a Cys replacement at a position that isnear the twofold axis of the FliN dimer seen in the crystalstructures (Leu 68 in E. coli, corresponding to Leu 85 in T.maritima). The L68C protein gave cross-linked FliN dimer invery high yield (Fig. 5), again with an apparent mass of approximately42 kDa. We conclude that position 68 of each FliN subunit isclose to position 68 of the other subunit in the dimer, as isseen in the crystal structures. Although the yield of cross-linkingshowed some variability between experiments, it was typically>80% and appeared to be quantitative (i.e., with no detectablemonomer remaining) in some experiments. Such complete conversioninto dimer would not be possible if a significant fraction ofthe FliN were present in FliM-FliN heterodimers formed by asubunit replacement. As a further test, we looked for cross-linkingbetween FliN with Cys at position 68 and FliM with Cys at position265, the residue that corresponds with FliN-Leu68 in a FliM-FliNsequence alignment. No FliM-FliN cross-linking was detectedusing either FliN or FliM immunoblots, and the major productwas again FliN dimer (Fig. 5; the FliN blot is shown). We concludethat subunit-exchanged FliM-FliN heterodimers do not occur orare present as only a minor species. These data support a tetramericFliN complex and argue against a complex with composition FliN₃FliM₁.

View larger version (32K):
[in this window]
[in a new window]
FIG. 5. (Left) Efficient cross-linking of the FliN protein with Cys at position 68, near the twofold axis of the dimer. Cross-linking was induced by Cu[1,10-phenanthroline]₃. The experiment used either the nonflagellate flhDC strain RP3098 (flhDC), or the fliN-null strain DFB223 (fla+), as indicated. Two representative experiments in strain DFB223 are shown. Proteins were resolved on 12% polyacrylamide gels, and the blot used anti-FliN antibody. (Right) Absence of cross-linking between FliN (L68C) and FliM (F265C). FliM has a molecular weight of approximately 38 kDa, and a FliM-FliN heterodimer would have a molecular weight of 52.7 kDa. Phe265 is the residue of FliM that aligns with Leu68 in a FliM:FliN sequence comparison (28). The cross-linked dimer shows closer-to-normal mobility in this experiment, which used a lower acrylamide concentration (10%). A FliN immunoblot is shown; FliM blots (not shown) exhibited only the FliM monomer.

To probe FliN tetramer organization further, we examined cross-linkingof a FliN protein with one Cys replacement at position 68 nearthe twofold axis of the dimer and another at position 54 nearthe end(s) of the dimer. Cross-linking was induced with copperphenanthroline, and products were examined using immunoblots,as before. Representative results are shown in Fig. 6. In theflagellate strain, cross-linking of the 54C/68C double mutantgave a species with an apparent mass of about 70 kDa as theonly major product. In the nonflagellate strain, the resultswere similar, except that a small amount of dimer was also observed.Because the 54C and 68C replacements individually showed veryefficient cross-linking (Fig. 4 and 6), we assigned the 70-kDaband as a FliN tetramer cross-linked through both positions.The apparent mass was close to that expected (predicted tetramermass, 62 kDa). Although the tetramer formed in very high yield,larger products were not observed in significant amounts.

View larger version (41K):
[in this window]
[in a new window]
FIG. 6. Cross-linking of the FliN protein with Cys residues at both positions 54 and 68 (and corresponding single-Cys controls). Cross-linking was induced by Cu[1,10-phenanthroline]₃. Proteins were resolved on a 10% polyacrylamide gel and visualized with anti-FliN immunoblots. The experiment used either the flagellate strain (fliN-null strain complemented with fliN on a plasmid) or the nonflagellate flhDC strain, as indicated.

Organization of a segment in the N-terminal domain. As discussed above, the N-terminal domain of FliN contributesto optimal swarming motility but is not essential for flagellarassembly. We hypothesized that the N-terminal segment mightcontribute to the stability of the FliN tetramer. To test this,we used gel filtration to examine the state of association ofa FliN protein deleted for residues 1 to 57. The FliN( ${Delta}$ 1-57)protein ran with an apparent mass of about 20 kDa in the sizingcolumn (Fig. 7), close to that expected for a dimer (predicteddimer mass, 16 kDa) and too small to be a tetramer. The full-lengthFliN protein, by contrast, exhibited a broad peak centered ataround 100 kDa, consistent with previous sedimentation experimentsindicating a tetramer with nonglobular shape (5). Thus, theN-terminal domain appears necessary for the formation of a stableFliN tetramer.

View larger version (27K):
[in this window]
[in a new window]
FIG. 7. Size exclusion chromatography of full-length FliN and FliN lacking residues 1 to 57. Cell lysates were chromatographed on a Superdex-300 column, and eluted fractions were examined on FliN immunoblots, as described in Materials and Methods. Positions of the elution peaks for three molecular mass standards are indicated at the top.

To examine further the organization of the N-terminal domain,we studied cross-linking of proteins with single Cys replacementsin a relatively well-conserved, predicted ${alpha}$ -helical segment (Fig.3). An initial experiment used Cu-phenanthroline with Cys replacementspositions 16, 19, 22, and 25. In each case, treatment with Cu-phenanthrolinegave cross-linked dimer in high yield (Fig. 8A), indicatingthat the N-terminal segments of different FliN subunits arein proximity to each other. The protein cross-linked throughthese positions migrated with an apparent mass of 35 kDa ingels, close to the mass expected for a FliN dimer.

View larger version (51K):
[in this window]
[in a new window]
FIG. 8. (A) Cross-linking of FliN proteins with Cys replacements at four positions in the N-terminal domain, using Cu[1,10-phenanthroline]₃. Experiments used either the nonflagellate flhDC strain or the flagellate strain, as indicated. (B) Cross-linking of FliN proteins with Cys replacements in 10 positions in the N-terminal segment, using 0.4 mM iodine. This experiment used strain DB223. Experiments for both panels used 10% polyacrylamide gels and anti-FliN immunoblots.

To place tighter constraints on the arrangement of the N-terminalsegment, we examined iodine-induced cross-linking of FliN proteinswith Cys in 10 positions in the predicted helical segment (Fig.3). Although yields were relatively low, the proteins with Cysat position 16, 19, or 20 produced cross-linked dimer at readilydetectable levels (Fig. 8B). If this segment of FliN were ${alpha}$ helical,then these positions would lie on one face, and the patternof cross-linking would suggest that the N-terminal segmentsof two FliN subunits approach most closely through this face.

Next, we examined cross-linking of FliN proteins with a Cysreplacement at position 19 in the N-terminal domain and a secondreplacement at position 68 near the dimer twofold axis. Oxidationusing copper phenanthroline gave two major products, both inthe range expected for a tetramer. Two products in the rangeexpected for a dimer and one in the range expected for a trimerwere also seen, but in smaller amounts (Fig. 9). Species largerthan a tetramer did not occur in significant amounts. Resultswere similar in the flagellate strain and the nonflagellateflhDC strain, and so the cross-linked products did not containany flagellar proteins besides FliN. Because FliN in solutionis known to form a stable tetramer and patterns of cross-linkingwere similar whether or not flagella were assembled, we concludedthat the cross-linking occurred within rather than between FliNtetramers. Given the results with the individual Cys replacementsat positions 19 and 68, one interpretation might be that FliNdimers joined by a 68-68 disulfide can cross-link into tetramersthrough position 19. Such tetramers could contain either threeor four disulfide bonds (i.e., be either linear or closed),which could account for the occurrence of two major products.An alternative possibility is that the N-terminal segment ispositioned near the hydrophobic patch, and cross-linking betweenpositions 19 and 68 gives rise to tetramers and other products.

View larger version (26K):
[in this window]
[in a new window]
FIG. 9. (A) Cross-linking of the FliN protein with Cys residues at both positions 19 and 68 and single-Cys controls. Cross-linking used Cu[1,10-phenanthroline]₃. (B) Cross-linking of the 19-68 double mutant and other double mutants, using iodine (0.2 mM). In all experiments, proteins were resolved on 10% polyacrylamide gels and visualized on FliN immunoblots. (C) Testing alternative models for the disposition of the N-terminal segment. Two alternative topologies are pictured. The four lines (two thick and two thin) represent the four subunits in the FliN tetramer. Cross-linking of the 19C-54C double mutant with Cu-phenanthroline is predicted to give only dimer in one case but tetramer in the other. The experiment, shown on the right, gave a high yield of tetramer.

To distinguish between these possibilities, we first examinedcross-linking of the 19C-68C double mutant by using iodine.The 19C replacement by itself showed only very weak cross-linkingwith iodine, and so iodine treatment should give little or notetramer if tetramer formation requires the 19-19 disulfide.Iodine treatment of the 19C-68C double mutant gave both dimerand tetramer in high yield, as well as species at a higher molecularweight (Fig. 9B). Because iodine was not expected to induceefficient 19-19 cross-linking, this suggested that the tetramerand larger species were formed by cross-linking between positions19 and 68. Some tetramer was also formed upon iodine treatmentof the 16C-68C, 22C-68C, and 25C-68C double mutants (Fig. 9B).Positions 22 and 25, like 19, showed no iodine-induced cross-linkingby themselves, and so the production of tetramer in the 22C-68Cand 25C-68C mutants reinforces the conclusion that position68 can cross-link to positions in the N-terminal segment.

Tetramer formation might still be due to 68-68 cross-linkingin combination with 19-19 cross-linking, if we postulate thatthe Cys replacement at position 68 affects the position or mobilityof the N-terminal segment to increase the efficiency of 19-19cross-linking. Thus, as a further means of distinguishing themodels, we examined cross-linking of FliN with one Cys replacementsat position 19 and another at position 54. Cu-phenanthrolinewas used for this experiment, because unlike iodine it allowsefficient cross-linking through positions in the N-terminalsegment, including position 19. One model predicts formationof tetramer upon treatment with copper phenanthroline, whilethe other predicts formation of only dimer (Fig. 9C). Some cross-linkeddimer was present prior to treatment with the oxidation catalyst,as was also observed with the 54C single replacement. Treatmentwith Cu-phenanthroline gave an additional dimer form and a highyield of tetramer. This supports the model in which the N-terminalsegments are in proximity to the hydrophobic patch of anotherdimer (Fig. 9C).

Function of the Cys mutant proteins. The FliN proteins that displayed efficient cross-linking weretested by swarming assays to determine whether their functionwas altered by the Cys replacements. The proteins with Cys replacementsat positions 16, 19, 22, 25, 54, 57, and 136 allowed cells toswarm at approximately wild-type rates. The replacement at position58 caused a roughly twofold reduction in swarming rate. Thereplacement at position 68, either alone or in combination withother Cys replacements, allowed flagellar assembly and vigorousswimming, but the cells swarmed at <1/10 of the wild-typerate in soft agar and exhibited aberrantly smooth motility inliquid, indicating a CCW motor bias.

DISCUSSION

Top

Discussion

Based on patterns of sequence conservation, we suggested previously(5) that FliN might be organized similarly to the tetramer ofHrcQB_C (a FliN paralog) seen in a crystal structure (12). However,a subunit organization like that in the HrcQB_C crystal cannotaccount for the pattern of disulfide cross-linking observedhere with FliN (see Fig. S1 in the supplemental material). Thepositions of high-yield cross-linking instead indicate a doughnut-likearrangement, substantially similar to the organization seenin one of the FliN crystal structures (1yab) (Fig. 10A) (5).The subunits contacts in this arrangement are provided largelyby ${alpha}$ -helix 1, particularly positions 54 and 57. An alignmentof FliN sequences (not shown) shows that hydrophobic characteris very well conserved at these positions. Although helix 1participates in close dimer-dimer contacts in one of the crystals(1yab) (5), it does not preserve these contacts in the othercrystal form (1o6a), which might account for its different positioningin the two structures (Fig. 2B).

View larger version (47K):
[in this window]
[in a new window]
FIG. 10. Model for FliN subunit arrangement and location. (A) Doughnut-like arrangement seen in the FliN crystal structure 1yab (5). Positions that gave a high yield of cross-linked dimer are indicated. At each dimer-dimer interface, the structure shows interpretable density for residue 136 of only one subunit, extending only to residue 135 in the other subunit. Residue 136 occurs in an ${alpha}$ helix; to obtain a rough estimate of its position, this helix from the first subunit was overlaid on the (shorter) helix of the second. (B) Proposed location of the FliN tetramers in the flagellar basal body. The location and orientation of one FliN tetramer are indicated; the flagellum would contain approximately 34 FliN tetramers in all. The side view of the basal body is from reference 40 (with permission of the publisher), and the end view is from reference 46 (with permission of the publisher). Given this location for FliN and evidence that FliG must lie near the top (15, 20, 25, 30, 48), it follows that FliM is located in the middle of the C-ring wall, as indicated. (C, left) Arrangement of N-terminal segments that can account for the cross-linking results and for the contribution of the segment to stability of the tetramer. The segment containing residues 15 to 25 is hypothesized to form an ${alpha}$ helix. Two such helices are in proximity to each other and to the hydrophobic patch of the other FliN dimer, bringing residues 19 and 68 into proximity. Although the arrangement pictured is stable enough to allow high-yield cross-linking between positions 19 and 68, the results also indicate that the segments are dynamic (see Discussion). (Right) Detailed view of the hypothesized interface between the N-terminal segments and the body of the dimer. The view is along the segments, one pointing toward the viewer and the other pointing away. Positions that cross-link most efficiently to position 68 are shaded, and positions that allow some cross-linking between the two N-terminal segments are circled.

Our results show that little if any of the FliN in the celloccurs in the form of subunit-exchanged FliM_C/FliN heterodimers.A Cys residue near the twofold axis of the FliN dimer gave disulfide-linkeddimer in very high yield, and no FliM-FliN cross-linking wasobserved when a second Cys residue was introduced at the correspondingposition in FliM_C (Fig. 5). We therefore favor a model in whichFliM₁FliN₄ units, rather than FliM₁FliN₃ units, are constituentsof the C ring. Sedimentation equilibrium studies have establishedthe stability of a FliM₁FliN₄ complex of the T. maritima proteins(5). The FliM protein of E. coli is less amenable to such studies,owing to its tendency to aggregate, but we are attempting toidentify better-behaved fragments that might allow further testsof this model for the complex.

The shape and dimensions of the hypothesized FliN tetramer providestrong constraints on its location in the C ring. Electron microscopicreconstructions of the basal body at approximately 20-Åresolution (14, 40) show that a ring with these dimensions wouldnot fit in the part of the C ring just under FliG but is a closematch for the bulge at the bottom edge of the ring. The FliNtetramer appears a good fit for this location in both side andend views of the basal body (Fig. 10B). Given this locationfor FliN at the bottom and the previously inferred locationfor FliG at the top of the C ring (13, 15, 20, 25, 30, 48),we conclude that FliM must be situated in the middle of theC-ring wall (Fig. 10B). Although the crystal structure of FliMhas not been reported, a crystal structure is known for therelated chemotaxis protein CheC (31). The structure shows anellipsoidal protein with dimensions (approximately 2.5 nm by3.5 nm by 5 nm) that are a good fit for this location.

Although we still lack detailed structural information on theN-terminal part of FliN, the present results cast some lighton its function and organization. This domain contributes tothe stability of the tetramer, because a truncated FliN lackingresidues 1 to 57 formed primarily dimers. The cross-linkingresults establish that the N-terminal segments of two FliN subunitsare in fairly close contact (Fig. 8) and that this assemblyof two N-terminal segments, most likely two ${alpha}$ helices, is positionednear the hydrophobic patch of another FliN dimer (Fig. 9). Ifresidues 15 to 25 form an ${alpha}$ helix, as predicted, then the observedpattern of cross-linking can be accounted for in a simple modelfor the segment organization (Fig. 10C).

In E. coli, the N-terminal segment is joined to the rest ofFliN by a Gly-rich segment (GGG_43-45) that should allow considerableconformational flexibility. The cross-linking experiments givetwo indications that the N-terminal segments are not rigidlyfixed but can undergo some movement relative to each other andto the patch: yields of intersegment cross-linking were muchhigher when Cu-phenanthroline was used instead of iodine (Fig.8A), and several different positions in the N-terminal segmentshowed some ability to cross-link to the same position (residue68) in the main body of the protein (Fig. 9B). Given this dynamiccharacter, we predict that the N-terminal segments could bedisplaced from the hydrophobic patch readily, to allow interactionswith other proteins, such as FliH (K. Paul and D. F. Blair,unpublished data). Flexibility of the N-terminal segment mightalso allow it to reach between different tetramers; such bridgingcould account for the occurrence of products larger than tetramerin some of the cross-linking experiments with double Cys proteins(Fig. 9B).

Studies of spontaneous mutants showed that FliN has some rolein CW/CCW switching (17). In this context, we note that theFliN tetramer in the 1yab crystal structure is puckered, ascan be seen in side views (Fig. 10A and B; see Fig. S2 in thesupplemental material). If such puckering occurs in the FliNtetramer in the flagellum, then the tetramer would be capableof existing in two conformations (puckered in one directionor the other). The two conformations would be equivalent inan isolated tetramer and thus equally stable, but in the contextof the C-ring they would be nonequivalent and might correspondto CW and CCW states of the switch (see Fig. S2 in the supplementalmaterial).

Finally, we note that the subunit arrangement we propose islikely to be relevant for other species that contain FliN butmight be different in species that use the much larger proteinFliY. These include most gram-positive species and spirochaetes.The organization of FliY and the shape of the C ring in thosespecies remain to be examined.

ACKNOWLEDGMENTS

We thank Sandy Parkinson for strains; Jonathan McMurry, BerthaGonzáles-Pedrajo, Dennis Thomas, Perry Brown, Bryan Lowder,and Jacob Harmon for discussions and advice; Moises Terrazasfor assistance with swarming-rate measurements; Stan Williamsand Marian Price-Carter for assistance with FPLC experiments;and Marian Price-Carter for critical reading of the manuscript.

This work was supported by grants R01-GM64664 and R01-GM64663from the National Institute of General Medical Sciences andgrant R01-EB2041 from the National Institute of Biomedical Imagingand Bioengineering. The Protein-DNA Core Facility at the Universityof Utah receives support from the National Cancer Institute(5P30 CA42014).

FOOTNOTES

* Corresponding author. Mailing address: Department of Biology, University of Utah, Salt Lake City, UT 84112. Phone: (801) 585-3709. Fax: (801) 581-4668. E-mail: blair{at}bioscience.utah.edu.

Supplemental material for this article may be found at http://jb.asm.org/.

REFERENCES

Top