The Flat-Ribbon Configuration of the Periplasmic Flagella of Borrelia burgdorferi and Its Relationship to Motility and Morphology

Electron cryotomography was used to analyze the structure ofthe Lyme disease spirochete, Borrelia burgdorferi. This methodologyoffers a new means for studying the native architecture of bacteriaby eliminating the chemical fixing, dehydration, and stainingsteps of conventional electron microscopy. Using electron cryotomography,we noted that membrane blebs formed at the ends of the cells.These blebs may be precursors to vesicles that are releasedfrom cells grown in vivo and in vitro. We found that the periplasmicspace of B. burgdorferi was quite narrow (16.0 nm) comparedto those of Escherichia coli and Pseudomonas aeruginosa. However,in the vicinity of the periplasmic flagella, this space wasconsiderably wider (42.3 nm). In contrast to previous results,the periplasmic flagella did not form a bundle but rather formeda tight-fitting ribbon that wraps around the protoplasmic cellcylinder in a right-handed sense. We show how the ribbon configurationof the assembled periplasmic flagella is more advantageous thana bundle for both swimming and forming the flat-wave morphology.Previous results indicate that B. burgdorferi motility is dependenton the rotation of the periplasmic flagella in generating backward-movingwaves along the length of the cell. This swimming requires thatthe rotation of the flagella exerts force on the cell cylinder.Accordingly, a ribbon is more beneficial than a bundle, as thisconfiguration allows each periplasmic flagellum to have directcontact with the cell cylinder in order to exert that force,and it minimizes interference between the rotating filaments.

INTRODUCTION

Top

INTRODUCTION

Spirochetes are a monophyltic phylum with a unique morphology(8, 35) These bacteria have a protoplasmic cell cylinder, whichincludes the plasma membrane and peptidoglycan layer, and anouter membrane. The region between the plasma membrane and theouter membrane constitutes the periplasmic space. The periplasmicflagella, which are subterminally attached to the ends of theprotoplasmic cell cylinder, reside in this space. A given periplasmicflagellum is attached at only one end, extends toward the centerof the cell, and is rotated by a basal motor anchored to theprotoplasmic cell cylinder. The periplasmic flagella at eachend form a group of filaments, and depending on the species,each group contains from one to hundreds of periplasmic flagella.

The motility of Borrelia burgdorferi, the Lyme disease spirochete,is quite complex (see references 10, 25, and 26 for recent reviewson spirochete motility). This species is capable of swimmingboth in low-viscosity media and also in viscous gel-like mediathat inhibit the motility of most other bacteria (19, 23). Atypical B. burgdorferi cell runs, stops, flexes (pauses andforms a distorted shape), and reverses direction. Several linesof evidence indicate that during a run, the two groups of 7to 11 periplasmic flagella rotate asymmetrically; i.e., onegroup rotates clockwise (CW), and the other rotates counter-clockwise(CCW) (10, 24, 32). (As a frame of reference, a given flagellumis viewed from its end along the filament toward its insertioninto the protoplasmic cell cylinder.) A cell in the flexingmode is thought to have its groups of periplasmic flagella rotatingin the same direction; i.e., either both rotate CW or both rotateCCW (32). During a running interval, the cell has a flat-waveappearance, with waves of constant amplitude being propagatedfrom its anterior to its posterior end (19). Rotation of thetwo groups of periplasmic flagella in opposite directions generatesbackward-moving waves along the cell body that propel the cellforward (10, 18, 19, 19, 25; see http://stock.cabm.rutgers.edu/blast/video1.mov).B. burgdorferi has many motility and chemotaxis genes in commonwith those of rod-shaped bacteria (16, 25); it is chemotacticto many compounds, including glucosamine, N-acetylglucosamine,and glutamate (1). However, the paradigm for spirochete chemotaxisis notably different from that of other bacteria such as Escherichiacoli and Salmonella enterica serovar Typhimurium (10). For example,B. burgdorferi rotates its groups of periplasmic flagella asymmetricallyduring a run (10, 24), whereas for E. coli and S. enterica serovarTyphimurium, all the flagella rotate CCW during the run (44).In addition, although CheY and CheA homologs are involved inB. burgdorferi chemotaxis (1, 24, 32), the nature of the signalthat coordinates rotation of the two groups of periplasmic flagellais unknown (10).

The complex geometry of B. burgdorferi is beginning to be understood.B. burgdorferi cells are approximately 10 to 20 µm longand 0.33 µm in diameter (18, 19). The periplasmic flagellaattached to one end of the cell are long enough to overlap withthose of the other end (20). Because the cell cylinder is rodshaped in mutants that lack the periplasmic flagella and theflat-wave morphology is regained in genetically complementedstrains that regain the periplasmic flagella, these organellesare concluded to have a skeletal function (31, 41). Purifiedperiplasmic flagella are tightly coiled left-handed helices,with most having a helix pitch of 1.48 µm and a helixdiameter of 0.28 µm (11). In addition, the periplasmicflagella undergo a helical transformation as a function of pH,as found with many flagella of other bacteria (S. Satoshi, S.I. Aizawa, M. Motaleb, and N. W. Charon, unpublished). In high-voltageelectron micrographs of intact cells, the periplasmic flagellaappear as a left-handed helical bundle with a helix pitch equalto the cell's wavelength (18). Although the bundle wraps aroundthe body axis (i.e., the center of the cell cylinder, as ifit were a sausage) in a right-handed sense, along the cell axis(i.e., since the cell resembles a sine wave, the cell axis isdefined as an abscissa) it is left handed (18). Because theshapes of the isolated periplasmic flagella and cell cylindersare so markedly different from those seen in the intact cells,they evidently exert force on one another to influence eachother's shape. Recent experiments and calculations using elasticitytheory, coupled with measurements of the mechanical propertiesof purified periplasmic flagella and protoplasmic cell cylindersemploying laser tweezers, indicate that the flat-wave cell morphologyis a natural consequence of the interaction of helical periplasmicflagella and the rod-shaped cell cylinder (C. Dombrowski, W.Kan, M. A. Motaleb, N. W. Charon, R. E. Goldstein, and C. W.Wolgemuth, submitted for publication).

Cryoelectron microscopy and electron cryotomography, also referredto as cryoelectron tomography, offer a new methodology for studyingthe architecture of bacteria (22, 30, 45, 47). Previous electronmicroscopic analysis used specimens of cells that were chemicallyfixed and stained; this methodology has been shown to introduceartifacts. For example, electron microscopy of hydrated sectionsof E. coli and Pseudomonas aeruginosa indicated that the periplasmicspace is markedly thinner than indicated by results previouslyobtained using fixed cells (28). Here we analyzed the structureof B. burgdorferi using electron cryotomography and comparedour results to those recently reported for Treponema primitiaand Treponema denticola (21, 34). We found not only that isits periplasmic space quite thin but that the periplasmic flagellado not form a bundle as previously thought (18, 31). Instead,the periplasmic flagella assemble into a very tightly packedflat ribbon that also widens the periplasmic space in the domainwhere they reside. Furthermore, we show how the ribbon configurationof the assembled periplasmic flagella is optimal for both swimmingand forming the flat-wave morphology.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Strains, culture conditions, and sample preparation. The high-passage B. burgdorferi strain B31A was used for allanalyses (6). Cells were grown in BSK complete medium (Sigma-Aldrich)at 34°C in an atmosphere of 3.0% CO₂ (31). To prepare cellsfor electron cryotomography, approximately 1.5 ml of late-logarithmic-phasecells was centrifuged at 1,200 x g for 5 minutes in a microcentrifugeat room temperature. Approximately 1.4 ml of the supernatantfluid was discarded, and the cell pellet was gently resuspendedby pipetting up and down in the remaining 100 µl and thencooled in ice. To prepare grids, first a thin carbon coat wasevaporated onto Quantifoil electron microscopy specimen grids(R3.5/1; Quantifoil Microtools, Jena, Germany). The grids wereimmediately treated with a 10-nm colloidal gold solution toprovide fiducial markers for alignment of the tomographic tiltseries (36). Approximately 5 µl of the cell suspensionwas applied to the grid without dilution or washing. The excessmedium was blotted with filter paper. The grid was immediatelyplunge-frozen into liquid ethane (13) and stored under liquidnitrogen for future examination.

Electron microscopy and tomographic reconstruction. Images were recorded at –178°C, using a JEOL JEM4000FXinstrument equipped with a Gatan GIF2002 energy filter. Themicroscope was operated at a 400-kV acceleration voltage inzero-loss energy-filtered mode. Single-axis tilt series werecollected with a 1° increment and a 120° angular range.The thickness of the ice layer was 300 to 400 nm, as measuredby electron energy-loss spectroscopy (14). The total electrondose for a tilt series was 70 to 90 e^–/Å², withthe higher dose used with thicker specimens. The calculatedresolution in the x-y plane was 8 nm. The calculated z (depth)resolution was 12 nm, due to the elongation factor from the"missing wedge" caused by the limited tilt range (37). The underfocusvalue, 15 µm, was chosen to maximize the transfer of informationat the expected resolution limit in order to optimize the signal-to-noiseratio (29). All image processing was done using SPIDER (15),and the reconstructions were computed by weighted back-projection.Isosurface models were traced on a WACOM Cintiq 20WSX tablet(Wacom Co., Saitama, Japan) using AMIRA software (Mercury ComputerSystems, Chelmsford, MA). To confirm chirality determinations,previously embedded Leptonema illini (formerly Leptospira illini),which is a known right-handed helical spirochete, was analyzedin the same manner as frozen B. burgdorferi cells (9, 18). Theconstructed images were originally found to be reversed fromleft to right (mirror images), and thus the final images aspresented here were corrected for this reversal. A total ofeight cells were analyzed, and measurements of specific parameterswere made only from clearly defined images.

RESULTS

Top

RESULTS

Overall cell morphology. Electron cryotomography was used to analyze B. burgdorferi cells.Cells were propagated in growth medium, concentrated by centrifugation,dropped onto electron microscopy grids with a support film perforatedwith small holes, and immediately plunged into liquid ethane.Because the spirochetes are flexible, distortion of a cell occurredat the edges of the holes of the support film. The analyseswe report are exclusively on those regions of the cells whichfell across the holes. B. burgdorferi cells are considerablylonger than the diameter of the grid holes. Consequently, onlya relatively small region of the cell (2 to 3 µm, whichis less than 25% of the cell length) was analyzed in a giventomogram.

The cells in general had a flat-wave appearance as seen in livingor fixed cells. The large size of the spirochetes (peak-to-peakwave amplitude of approximately 780 nm [18]) and the shallowwater layer (approximately 300 to 400 nm) resulted in the flatwave lying parallel to the grid. We compared the shape and diameter(d_ic) of the inner cell, as measured between the center of theplasma membrane at each side of the cell, in regions where thecell was curved (bend region) to those in regions that weremore linear (interbend region). Our rationale is that bendingof the cell could cause compression and distortion. Becausethe total cell diameter varied in the regions where the periplasmicflagella reside (see below), d_ic was used to test for compression.We found that there was no obvious distortion of the inner cell;all were circular. In addition, the d_ic was 285 ± 49nm within the bends (n = 9 bends on seven cells), and 294 ±35 nm in the interbend regions (n = 7 regions on seven cells).These results suggest that bending of the cells did not dramaticallyalter the inner diameter of the cell.

The general shape of the cells was similar in many respectsto previous determinations using light microscopy, standardtransmission electron microscopy, and high-voltage electronmicroscopy (18, 19). All cells appeared intact, with an attachedouter membrane, plasma membrane, and periplasmic space, andthe periplasmic flagella were clearly evident (Fig. 1 and 2a;see Movie S1 in the supplemental material). Cells were circularin cross section, with a diameter of approximately 310 nm inregions without periplasmic flagella, which compares reasonablywell with results determined using high-voltage electron microscopy(330 nm) (18). At the ends of the cells, we often saw a bulb-likeformation referred to as a bleb (Fig. 3). These blebs did notcontain periplasmic flagella and had contents that appearedslightly denser than the cytoplasm. In addition, these structureswere separated from the rest of the cell body and were continuouswith the outer membrane.

View larger version (98K):
[in this window]
[in a new window]

FIG. 1. Electron cryotomography cross-section, 1.8 nm thick. The outer membrane (OM), plasma membrane (PM), periplasmic flagella (PFs), and periplasmic space (PS) are identified. Note the circular shape of the plasma membrane, the ribbon formation of the periplasmic flagella, and that the periplasmic space is wider in the domain where the periplasmic flagella reside. Bar, 50 nm.

View larger version (78K):
[in this window]
[in a new window]

FIG. 2. (a) A 1.8-nm-thick longitudinal slice of a cell body. The outer membrane (OM), plasma membrane (PM), periplasmic flagella (PFs), and periplasmic space (PS) are identified. Bar, 200 nm. (b) Three-dimensional view of sections of a cell, showing that the periplasmic space (arrows) widens in the domain where the periplasmic flagella (red) reside. Blue represents plasma membrane, and brown represents outer membrane.

View larger version (177K):
[in this window]
[in a new window]

FIG. 3. A 1.8-nm-thick longitudinal slice of a cell end. The ends of the cell were often bulb shaped. Bar, 100 nm.

Outer membrane and periplasmic space. The outer membrane was noticeably different than that seen inchemically fixed and stained cells. While the outer membranesometimes appears as having an irregular ruffled or uneven appearancein standard sections of fixed cells (see Fig. 3 in reference18 and Fig. 4 in reference 31), it was quite smooth in appearancein the tomograms (Fig. 1 and 2a). The width of the periplasmicspace was significantly greater in regions containing periplasmicflagella than in regions where there were no periplasmic flagella(Fig. 1 and 2a and b; see Movies S1 and S2 in the supplementalmaterial). As measured between the centers of the outer membraneand plasma membrane, the width of the periplasmic space was22.7 ± 3.9 nm in regions without periplasmic flagella(n = 17 spaces on eight cells) and 49.0 ± 6.9 nm in regionscontaining periplasmic flagella (n = 11 spaces on eight cells).The periplasmic space as measured between the inner and outermembranes was 16.0 ± 3.7 nm (n = 17 spaces on eight cells)in regions without the periplasmic flagella and 42.3 ±6.8 nm (11 spaces on eight cells) in regions with the flagella.

View larger version (147K):
[in this window]
[in a new window]

FIG. 4. A 9-nm-thick longitudinal slice of a cell end, showing four basal bodies in an approximately linear arrangement (arrowheads). The dark centers of the basal bodies are the regions where the filaments are seen to emanate by tomography. The putative peptidoglycan layer is indicated by the arrow. Bars, 200 nm in main image and 100 nm in inset.

Arrangement of basal bodies. The basal bodies of the periplasmic flagella were evident onlyat the ends of the cells in a subterminal region (Fig. 4). Theouter surfaces of the basal bodies (i.e., those parts of thebasal bodies that directly face the periplasmic space), withthe hook region exposed, were approximately 38 nm in diameter,which is similar to those reported for T. primitia (35 nm) butsomewhat greater than those of T. denticola (29 nm) (21, 33).Each basal body appeared as a ring, with a dense region in thecenter. Flagellar filaments could be seen emanating from thedark center of the basal bodies and were seen to curve towardthe center of the cell by tomography (not shown). The arrangementof basal bodies near the ends of the cell body was similar tothat seen in negatively stained preparations as forming a somewhatlinear arrangement parallel to the long axis of the cell (2,7, 20), and they were found to lie along a line in the terminalbending region of the cell. The spacing between the four adjacentbasal bodies as seen in Fig. 4 was approximately 90.8 ±4.2 nm. In most tomograms, the peptidoglycan layer was not evident,but occasionally this putative wall layer was seen within theperiplasmic space (Fig. 4).

Arrangement of the periplasmic flagella. The periplasmic flagella were clearly visible in the tomograms,and their appearance was quite different than that seen in fixedcells. These organelles formed an elegantly constructed ribbon(Fig. 1, 2b, and 5a and b; see Movies S1, S2, and S3 in thesupplemental material). The width of the ribbon was directlydependent on the number of periplasmic flagella and ranged from92.9 nm to 160.9 nm, with a mean of 120.0 ± 26.7 nm (Fig.6). The contribution of each periplasmic flagellum and surroundingregion to the width of the ribbon was a mean of 19.2 ±1.7 nm (range of 16.7 to 20.6 nm). Because previous resultsfor unsheathed B. burgdorferi periplasmic flagella indicateda diameter of approximately 16 nm (20), evidently there is littlespace (3.0 nm or less) between neighboring periplasmic flagellawithin the ribbon. Significantly, the flat-ribbon formationby the periplasmic flagella is markedly different from thatof stained and fixed cells. Previous results indicated thatthese structures formed a round bundle of 67 nm in diameter,and in thin sections they appeared as irregularly arranged filaments(18, 31). We found that an occasional filament was separatedfrom the ribbon and was located on the opposite side of thecell cylinder (Fig. 2). In one ribbon of 158.27 nm that wascomposed of nine periplasmic flagella, one of the flagellarfilaments in the middle region appeared to terminate. The spacethat it would occupy is evident in the tomogram; the remainingflagellar filaments moved to fill in the vacated space (Fig.7). Specifically, the adjoining filaments are 19.0 nm apartat the point where the filament first disappears and 7.4 nmapart at a distance of 141 nm. These results suggest that thereare forces acting on the filaments to bring the adjoining filamentstogether. In one case, we observed a wide ribbon close to theplasma membrane and another narrower ribbon with fewer filamentsin one region close to the outer membrane (Fig. 2b). In allpreparations, the ribbon wrapped around the body axis in a right-handedsense (Fig. 5a and b; see Movies S1, S2, and S3 in the supplementalmaterial), which is in agreement with the high-voltage electronmicroscopy analysis (18). In addition, the angle subtended bythe ribbon was approximately 1.13 rad in the interbend regions(Fig. 5a), which is also in agreement with the high-voltageelectron microscopy analyses (18).

View larger version (37K):
[in this window]
[in a new window]

FIG. 5. (a) A 1.8-nm-thick longitudinal slice of a ribbon of nine periplasmic flagella. The ribbon wraps around the cytoplasmic cylinder in a right-handed sense (bottom of cell as viewed from the top). Bar, 200 nm. (b) Three-dimensional view of sections of a cell, showing that the periplasmic flagella (red) form a ribbon and wrap around the cell cylinder (blue) in a right-handed sense. The outer membrane is not shown.

View larger version (10K):
[in this window]
[in a new window]

FIG. 6. Ribbon formation. The width of the ribbons as a function of the number of periplasmic flagella (PFs) is shown. The coefficient of correlation was 0.90.

View larger version (122K):
[in this window]
[in a new window]

FIG. 7. A 1.8-nm-thick longitudinal slice of a ribbon of nine periplasmic flagella. In this case, one of the flagella terminates (arrow) within the ribbon. Bar, 200 nm.

DISCUSSION

Top

DISCUSSION

Electron cryotomography revealed that B. burgdorferi producedouter membrane blebs, and these structures were located at theends of the cells. These blebs contained electron-dense material.B. burgdorferi is known to spontaneously generate membrane vesicleswhen grown in vitro and in vivo (12, 17, 37a). These vesicles,which are purified by passing culture supernatants though a2-µm filter, are less than 0.25 µm in diameter andhave been shown to contain DNA and several outer membrane-associatedproteins, such as OspA and OspB, and an 83-kDa protein (12,17, 43). In many species of bacteria, vesicles are postulatedto play important roles in protein secretion, virulence, andpromotion of cell-cell communication in biofilms (5, 27, 42).In B. burgdorferi, vesicles are believed to be able to readilypenetrate tissues and initiate inflammation (43). Little isknown about the precise details of how these structures areformed in other bacteria, but they are associated with the formationof blebs on the bacterial surface (27). In B. burgdorferi, wefound that the blebs containing dense material are an extensionof the outer membrane, which is in agreement with previous resultsof others using transmission electron microscopy (17). We find,however, that these blebs are located primarily at the endsof cells and not along the length of the cell as previouslydescribed (17).

The width of the periplasmic space in the region where no periplasmicflagella reside is consistent with what is found for T. denticolabut less than in other gram negative bacteria. To compare ourresults with published values, we made two measurements. Ifwe measured this space between the region of highest densityof both the outer membrane and inner membrane, it was 22.7 nm.This is the easier measurement to make, because the region ofthe membrane of highest density is quite clear. On the otherhand, if we just measured the space between the inner and outermembranes, it was considerably less, i.e., 16.0 nm. However,in this case, it is more difficult to determine the boundariesof the membrane, as some clarity is lost. When measuring betweenthe outer and inner membranes, electron microscopy of ultramicrotomesections of frozen-hydrated cells revealed a periplasmic spaceof 24 nm for P. aeruginosa and 22 nm for E. coli (28), but aswith our results, there was considerable variability. In T.denticola, the periplasmic space between the membranes was reportedto be 15 to 22 nm (21), which is similar to our results withB. burgdorferi. For T. primitia, the periplasmic space was foundto be 28 nm, as measured by the line of highest density betweenthe membranes, which is somewhat wider than for B. burgdorferi.In T. primitia, two layers of wall are believed to be presentin the periplasmic space (34), which is different from whatis reported for B. burgdorferi.

The periplasmic space in B. burgdorferi contains the peptidoglycanlayer, but we were not able to consistently visualize this structure,apparently due to its generally low density in this organism.Nevertheless, we know that peptidoglycan is present in thesebacteria from chemical analyses, susceptibility to lysozymeafter outer membrane disruption, and sensitivity to specificpeptidoglycan wall-acting antibiotics (3, 39, 40, 46). Recentexperiments suggest that the peptidoglycan of B. burgdorferiis quite flexible compared to sacculi from E. coli (Dombrowskiet al., submitted), which may be related to a decrease in peptidoglycancross-linking compared to that in other bacteria. In contrastto B. burgdorferi, both T. primitia and T. denticola have relativelyrigid helical cell cylinders, and in both cases the peptidoglycancan be more readily visualized (21, 34, 38).

The periplasmic space in the vicinity where the periplasmicflagella reside was notably wider (42.3 nm) than in the regionwithout these organelles (16.0 nm). Similar results of an increasewidth in the periplasmic space where the periplasmic flagellareside have been noted for T. primitia and T. denticola (21,34). In T. denticola, the increase in width in this region wasless than the diameter of the periplasmic flagella, and thesuggestion was made that the peptidoglycan layer was thinnerin that region. (21, 34). In contrast, in B. burgdorferi, thisspace is greater than the diameter of the periplasmic flagella(16 nm [20]). Therefore, our results suggest that the outermembrane fits tightly about the cell cylinder and that the periplasmicspace widens due to the obstruction produced by the periplasmicflagella. In addition, in the bend regions of the cell, theperiplasmic flagella may pull in toward the cell axis, causingthe periplasmic space to widen more than the diameter of thefilaments (see, for example, Fig. 3 in reference 18).

In contrast to the observations with plastic-embedded cells,which indicated that the periplasmic flagella form a bundlein the periplasmic space (31), electron cryotomography revealedthat the periplasmic flagella form a tight ribbon in this spaceas they wrap around the cell cylinder. The periplasmic flagellaare reported to have diameters of 16 nm (unsheathed) and 21nm (sheathed) (20). The sheath is associated with FlaA, andit is localized proximal to the hook region (S. Shabata, S.Aizawa, M. Motaleb, and N. Charon, unpublished data). Givena diameter of 16 nm per filament, we estimate that there isapproximately 3 nm between each filament in the ribbon. In addition,in agreement with the high-voltage electron microscopy of embeddedcells (18), the ribbon was found to wrap around the body axisin a right-handed sense. These results indicate that each periplasmicflagellum is tightly associated with the cell cylinder, andthus each can equally participate in dictating the flat-wavemorphology.

One obvious question is the following: how does the tight ribbonof periplasmic flagella form? Perhaps the spacing of the insertionof the periplasmic flagella is critical for ribbon formation,as these organelles are inserted on a line along the subterminalends of the cell (2, 7, 20), which we find are approximately90 nm apart. In addition, formation of the ribbon may dependon having active periplasmic flagella that rotate in a mannersimilar to bundle formation in E. coli and S. enterica (44).Previous results indicate that B. burgdorferi cells incubatedin the cold or cells treated with the iononophore carbonyl cyanide-m-chlorophenylhydrazone,which dissipates the proton gradient, become immotile but stillretain the flat-wave morphology (19, 31). One possible explanationfor these results is that cells treated in this manner couldhave already formed the ribbon structure before treatment. Thus,flagellar rotation may be necessary to form but not maintainthe ribbon structure. Although we precooled the cells in icebefore preparing the grids, motile cells prepared without precoolingstill had the flat-ribbon configuration of the periplasmic flagella(J. Benach, T. LaRocca, M. Marko, and C. Hsieh, unpublisheddata). Thus, the ribbon configuration is not the result of coolingthe cells in ice. We observed that if a filament is terminatedbefore the others in a ribbon, the adjoining periplasmic flagellamove to occupy the space. These results suggest that a forcebrings the periplasmic flagella together in a ribbon, and itmay be that flagellar rotation helps produce this force. Weexpect that future experiments targeting genes that result inmutants that retain their periplasmic flagella but result inparalysis will allow us to determine the role of flagellar rotationand ribbon formation. Several questions remain unanswered. InFig. 2, one of the periplasmic flagella is separated from theribbon. Is this separation a consequence of that filament rotatingin the opposite direction relative to the others? In addition,we observed that in one ribbon there was a region where therewere two layers of ribbons. It is not known if the periplasmicflagella from one end form the double layer or if one layeris derived from the filaments from the other end. Finally, ifmost of the periplasmic flagella interact with the peptidoglycanlayer as in a ribbon, how does an individual filament affectcell shape? Perhaps mutant analysis would allow the determinationof how cell shape is influenced by the number of periplasmicflagella.

A final interesting question is the following: what role doesthis ribbon structure play in the motility of B. burgdorferi?We assume that the shape of the cell is explicitly tied to motility;i.e., optimal swimming is dependent on the flat-wave shape.Theoretical work on the morphology of B. burgdorferi suggeststhat the flat-wave shape depends strongly on the stiffness ofthe flagella compared to that of the cell cylinder (Dombrowskiet al., submitted). If this model is correct, then there isa minimum number of flagella required to produce a flat-waveshape with a reasonable amplitude. If motility requires thatthe rotation of the flagella exerts force directly on the cellcylinder to generate the backward-moving flat waves, then aribbon is clearly beneficial. As can be seen in Fig. 8a, andas is consistent with the results presented in this study, sevenperiplasmic flagella lie in a ribbon along the top side of thecell cylinder and rotate CCW. Each flagellum pushes againstthe cell cylinder to generate a backward-moving wave. This wavemay be the result of the periplasmic flagella pushing againstthe cell cylinder with some slippage as they rotate. In addition,the torque generated may also cause a counter rotation of thecell cylinder (4, 10, 18, 19). Since each flagellum is in contactwith the cell cylinder, each imparts an equal force and torqueto the cell cylinder. Furthermore, since there is a small spacebetween each flagellum, there is little interaction betweenthe rotating flagella. Figure 8b represents what is observedin cells fixed in plastic (18, 31). The bundle configurationof flagella requires that some of the flagella are not in contactwith the cell cylinder. For these flagella to exert an equalforce and torque on the cell cylinder, their rotation must beconveyed to the cell cylinder through interaction with otherflagella. However, the rotation of a flagellum that is in contactwith another flagellum and rotating in the same direction impedesand interferes with the motion of the neighboring flagellum.Therefore, a bundle configuration would likely decrease therotation of the flagella and lead to a smaller force and torquebeing applied to the cell cylinder. We expect that a detailedunderstanding of both ribbon formation and its physical interactionwith the cell cylinder will lead to a more complete mechanicalmodel of how flagellar rotation and wave deformations drivemotility.

View larger version (23K):
[in this window]
[in a new window]

FIG. 8. Cartoon comparison of the effectiveness of a periplasmic flagellar ribbon compared to that of a bundle. (a) Cross-sectional view of the ribbon configuration observed in electron cryotomography images. (b) Schematic of a bundle configuration of periplasmic flagella, which represents a configuration similar to that seen in plastic-embedded cells (18, 31). The cell body is gray, and its surface is dashed. The periplasmic flagella are rotating CCW. See text for details.

ACKNOWLEDGMENTS

This research was supported by Public Health Service grant AI-29743awarded to N.W.C., Public Health Service grant GM 0072004 awardedto C.W.W., Public Health Service grant AR054582 awarded to M.A.M.,National Science Foundation grant DMS 0201063 to S.F.G., andPublic Health Service grant P41 RR01219, which supports theWadsworth Center's Resource for Biological Complexity as a NationalBiotechnological Resource.

We are grateful for resources from the University of MinnesotaSupercomputing Institute. We also appreciate the unpublishedinformation from S. I. Aizawa, S. Shibata. J. Benach, and T.LaRocca. We thank D. Dorward for helpful information.

FOOTNOTES

* Corresponding author. Mailing address: Department of Microbiology, Immunology and Cell Biology, Robert C. Byrd Health Sciences Center, West Virginia University, Morgantown, WV 26506-9177. Phone: (304) 293-4170. Fax: (304) 293-7823. E-mail: ncharon{at}hsc.wvu.edu

Published ahead of print on 14 November 2008.

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

REFERENCES

Top