INTRODUCTION

Top Previous Next References

Motility is arguably the most impressive feature of a microbe's physiology. Active movement ultimately uncovered microbes as living organisms to the first microscopic inspection by Antonie van Leeuwenhoek more than 300 years ago (31) and since this time has attracted the curiosity of innumerable scientists. In general, motile prokaryotic microorganisms move in aqueous environments by swimming or by control of buoyancy or along surfaces by using distinct modes of surface translocation. Most research has focused on understanding swimming motility in prokaryotes. Fundamental insights have been gained from thorough studies of the molecular architecture of the flagellum, from the energy transduction mechanism and mode of force generation, and from the control of motility, all of which serve now as paradigms for motility in biology (for a review, see reference 14). Notably, as shown by Waterbury et al. for a cyanobacterium (128), not all swimming bacteria are flagellated. Swimming motility is advantageous only for microorganisms living in aqueous habitats. Many microbes, however, live in environments with a low water content or changing humidity. These environments include biofilms, microbial mats, and soil, where the exploration of a new food source by means of swimming motility is unfeasible. These organisms are faced with the challenge of how to move on surfaces that are covered with only a few layers of water molecules. One solution of some swimming bacteria is to produce excessive lateral flagella that enable them to swarm in a thin fluid layer on a solid surface (for a review, see reference 46). However, many other prokaryotes employ one of the two modes of active cell translocation on a solid surface: gliding or twitching.

Historically, gliding is defined as the movement of a nonflagellated cell in the direction of its long axis on a surface (51). This rather broad definition, which does not specify a molecular apparatus or a mode of force generation, has therefore been used to describe movements by many phylogenetically unrelated bacteria (Fig. 1). As will be shown in this review, gliding movement can be generated by fundamentally different molecular mechanisms that can operate simultaneously even in a single microorganism. Consequently, use of the term "gliding" should not be considered to imply the operation of a specific motility mechanism. Several models for gliding have been proposed for different organisms to explain the seemingly smooth advancement of cells on solid surfaces. These models include operation of contractile elements (22), directional propagation of waves along the cell surface (60), directional extrusion of slime (59, 99), rotary motors (91), controlled release of surfactants from poles of cells (34, 67), and movement of adhesion sites in the outer membrane along tracks fixed to the peptidoglycan of the cell wall (72). In most cases, the models were postulated as a result of observations on motility behavior of single cells. However, to date, no model has been verified by biochemical, molecular, and ultrastructural studies, and it seems unlikely that all gliding bacteria harbor the same motility mechanism. The focus of this review is on recent research which seeks to provide a mechanistic and molecular understanding for gliding. The single-cell gliding bacteria Myxococcus xanthus, Flavobacterium, and Cytophaga strain U67, as well as the filamentous organism Flexibacter polymorphus, serve as model organisms in these studies. Interestingly, gliding motility has so far not been reported for archaea. Diversity and ecological aspects of gliding motility have been reviewed recently (95, 96).

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Occurrence of gliding bacteria among the eubacteria. The figure depicts examples of bacteria (indicated in italics) that have been reported to move by "gliding." Note that "gliding" is an operational definition, and not all bacteria may glide by the same mechanism. No gliding archaea have been reported. Modified from reference 58.

Most of the research on gliding motility has been conducted with M. xanthus (50, 127, 143, 145). This microorganism is a Gram-negative soil bacterium which belongs to the delta subdivision of proteobacteria and is specialized to mineralize insoluble organic matter, specifically proteins (33, 95, 104). In common with other myxobacteria, M. xanthus exhibits an unusually complex life cycle during which gliding motility plays a crucial role (110). Under vegetative conditions, cells move as coordinated swarms. These swarms may contain thousands of cells, which secrete hydrolytic enzymes into their environment that lyse other cells and convert insoluble proteins into soluble, transportable amino acids. Metabolism of these amino acids and other compounds is strictly aerobic. The feeding strategy of moving in swarms on the metabolizable substrate has been termed the "wolf pack" effect to reflect the fact that large numbers of organized cells undoubtedly utilize insoluble nutrients more efficiently than a single cell (104). Furthermore, coordination of vegetative cells in swarms provides the basis for survival when nutrients become limiting. When cells are starved of nutrients, tens of thousands aggregate to form a fruiting body, within which cells differentiate into spores. The organized movement of vegetative cells ensures that sufficient cells are present to form this fruiting body. Subsequent dispersal of mature, spore-containing fruiting bodies guarantees that after spore germination, cells are present at a sufficiently high density to allow the formation of new vegetative swarms. It is therefore apparent that the motility of M. xanthus is a crucial prerequisite for this lifestyle and thus has attracted much research effort.

Over the few past decades, research has focused on genetic and molecular approaches, as well as on high-resolution motion analyses, to develop a cellular and mechanistic understanding of gliding motility in M. xanthus. Hodgkin and Kaiser initiated a genetic analysis by screening chemical- and UV-induced mutants for visible defects in colony swarming (56, 57). Subsequent analysis of these mutants revealed that motility in M. xanthus is controlled by two multigene systems: the A (adventurous) system, which controls gliding motility of individual, isolated cells, and the S (social) motility system, which is essential for cell movement in swarms and during aggregation and fruiting-body formation (56, 57). Both motility systems are required for wild-type swarming, because swarming is completely abolished in any AS double mutant. It should be noted that although gliding of isolated cells is observed only in A⁺ cells, cells in swarms can move by either A motility or S motility or by both systems at a time. The use of single-cell observations to infer involvement of a specific type of motility system can be misleading, and in a strict sense, A motility and S motility are defined only by the macroscopically visible nonswarming phenotype of a double-mutant colony (56).

As a result of significant progress in recent years, a refined picture of how M. xanthus moves is becoming apparent. Experimental evidence that is reviewed below suggests that A and S motility do not represent different modes of a single gliding mechanism but, instead, comprise two distinct motility mechanisms: pilus-independent single-cell gliding (A motility) and pilus-dependent movements (S motility), which may be related to another surface translocation mechanism, called twitching. Twitching is described as intermittent, jerky cell movement that seems to lack the degree of organization seen in gliding. Twitching, unlike S motility, can occur in directions other than the long axis of the cell (53). However, functional type IV pili are essential for both twitching motility and S motility. Genes of the S system have recently been shown to include genes necessary for the synthesis, processing, export, assembly, and function of type IV pili (102, 125, 139-141).

In this section, observations on the movement of wild-type swarms and on isolated single cells are summarized. Colonies of wild-type M. xanthus (DK1622, DZ2) expand as flat swarms on a 1.5% agar surface (Fig. 2A). Microscopy at the edges of these swarms shows that the advancing front of the swarm is composed predominantly of individual cells, with groups of cells forming behind (56) (Fig. 2A). The rate of swarm expansion depends on cell density in a type of first-order kinetics with a maximal rate of approximately 1.6 µm/min and a half-saturation cell density of approximately 2 × 10⁸ cells per ml (62). The swarming rate also depends on the concentration of the agar support (57). When the rate is measured as the expansion of a colony area, swarming appears to be faster on 0.4% agar and drops to about 1/10 the efficiency when the agar concentration increases to 2% (107). M. xanthus colonies also respond to stress forces in the (agar) surface, a phenomenon called elasticotaxis (37, 119). Small compressions in the agar surface lead M. xanthus colonies to swarm preferentially in the direction perpendicular to the stress force, which results in an elongated elliptical colony (37).

View larger version (81K): [in this window] [in a new window]   FIG. 2.   Two gliding motility systems are operative in M. xanthus. (Left) Colony morphology. (Right) Gliding speed plotted against cell-cell distance of individual cells (the detection limit for active movement was 1 µm/min [117]). (A) Wild-type DK1622 (A+S+). Gliding-speed data are from reference 117. (B) JZ315 (cglB, AS+). Gliding-speed data are from reference 101. Note that when the cells were separated by more than 2 µm, no active single-cell movement was observed. When the cells were in contact, the velocities observed were similar to those found when wild-type cells moved in close proximity (A). (C) DK3473 (pilR, A+S); cell movement of 50 individual DK3473 cells was examined as described previously (117). A total of 4,531 speed and cell-cell distance values were obtained and plotted. The speed of isolated cells was similar to that of wild-type cells. However, cells did not exhibit high-speed movements, as observed for wild-type and cglB cells (A and B) when in close proximity.

Videomicroscopy motion analyses have been used to determine the rates at which individual isolated cells glide on a 1.5% agar surface. The studies showed that the translocation velocity is highly variable for a given cell but also varies between different cells (117). Most cells were found to move at velocities between 1.5 and 6 µm/min, although occasionally speeds of up to 20 µm/min were observed (117). Cells change the direction of movement by reversing every 5 to 7 min, so that the leading end becomes the lagging end (12, 118). Bending of the highly flexible cell body during movement frequently results in deviations from the original track of less than 90°, thus also contributing to directional changes. Individual M. xanthus cells do not have a preferred direction of movement; i.e., a cell does not have a "head or tail" (106, 118). This is indicated by measurements of cellular gliding velocity taken in both the forward and backward directions, as well as by studies that timed the duration of movement in one or the other direction. Gliding speed is also independent of cell length. When M. xanthus cells were treated with cephalexin, which increases the cell length to up to 20 µm (compared to 6 µm untreated), no difference in gliding speed compared to untreated cells was noticeable (118a), suggesting that the gliding motor(s) operates at constant speed. Interestingly, the translocation velocity of individual cells varied with cell-cell distance (117) (Fig. 2A). When individual cells were separated by more than one cell diameter (0.5 µm) from the nearest neighbor, they moved with an average velocity of 3.8 µm/min. However, in close proximity, cells moved with an average velocity of 5.0 µm/min (117). These kinetically distinguishable modes of single-cell gliding can be separated genetically into A- and S-motility-related movements, respectively (Fig. 2B and C).

On rare occasions, individual wild-type cells have been observed to glide while one cell pole is fortuitously fixed on an agar surface. This event results in flexing of the cell as the end which is not fixed glides backward and forward (118). Long cells have also been observed to bend into a "U" and to move in one direction in this configuration. Because the cell poles would move away from each other if the cell was not a bent "U," this observation suggests that each cell pole can move independently in both directions. In addition, the region between the poles can exhibit movement, indicating that the subcellular elements that promote motility are localized at both cell poles and along the length of the cell body (117) (Fig. 3). Therefore, M. xanthus cells are proposed to contain not one single gliding motor which is localized to a specific site but, instead, multiple motor elements positioned along the entire cell surface (117) (Fig. 3). The activities of these multiple motor elements would normally be coordinated to generate overall movement in either the forward or the reverse direction. Currently, no data are available to differentiate between the possibility that cells (i) contain a single set of motor elements where each element is capable of operating in both directions or (ii) contain two sets of unidirectional motors that are arranged opposite to each other and differentially regulated (Fig. 3).

View larger version (10K): [in this window] [in a new window]   FIG. 3.   Models for localization of motility elements on the surface of M. xanthus cells. Multiple motor units are located along the cell body. (A) In this model, a single motor unit can generate displacement in two directions that are opposite to each other (--). (B) In this model, two types of unidirectional motors exist (- and -), each capable of generating force in only one direction. The motor elements are arranged opposite each other, so that selected activation of one or the other results in movement in one or the other direction.

Under certain conditions, M. xanthus cells have also been found to bend in the absence of any translocation of the cell body. Such bending was observed when dsp cells (see below) attached to nitrocellulose-coated coverslips (Fig. 4). While the majority of the cell body did not show any displacement, the ends of individual cells were found to flex until a certain degree of bending was achieved, when the end would "snap back" to form a straight cell. The flexing proceeded at a rate similar to that of gliding. These observations suggest that force can be generated relative to distal body sections in the absence of translocation of the entire cell.

View larger version (83K): [in this window] [in a new window]   FIG. 4.   Flexing of an M. xanthus cell. Cells of DK1680 (dsp) were grown in CTT liquid medium (117) to a density of about Klett 100 (~5 × 108 cells/ml), and 10-µl drops were placed on nitrocellulose-coated coverslips. Cell movement was monitored under an inverted microscope, and images were recorded during the observation period. In the absence of net movement, cells were observed to flex. This can be easily shown when superimposing two images that were recorded at different times. Dark areas indicate overlap of the cell during both recordings, and lighter areas show displacement of cell sections. The time difference between the two superimposed images was 1 min 48 s for the picture at top left (size bar included), 1 min 10 s for the picture at bottom left, 7 s for the picture at top right, and 27 s for the picture at bottom right. In the last picture, the lower cell was gliding in direction of the long axis of the cell while the upper cell flexed. Bar, 0.5 µm.

In contrast to other gliding bacteria, such as Cytophaga, M. xanthus cells have not been reported to rotate during translocation. However, cells have been observed to pivot around one cell pole which is tethered to a surface in a wet mount. Similar to studies with other gliding bacteria (72, 91), movement of beads along the surface of a cell has been examined in M. xanthus (118a). It appears that the direction of a moving bead does not always correlate with the direction of cell gliding. While a cell is moving forward, beads may be moving forward, backward, or not at all. Bead movement was also observed on cells which were not moving. Two beads on the surface of a single cell could be found to move in opposite directions. These observations are very similar to those made previously by Lapidus and Berg with Cytophaga strain U67 (72). Beads frequently seem to become "trapped" at cell poles. The velocity of bead movement is on the order of 2 to 4 µm/min (118a), which is comparable to the velocity of individual gliding cells, suggesting that bead movement along the cell surface may be related to gliding movement (72).

No study that identifies a source of energy for gliding movement in M. xanthus has been reported. The electrochemical membrane potential was proposed to power gliding motility in Flexibacter (see below). Because two gliding mechanisms operate in M. xanthus, each mechanism may utilize a different energy source (e.g., ATP versus electrochemical ion potential), thus complicating bioenergetic studies.

Several subcellular structures in M. xanthus are believed to be involved in promoting cell movement: (i) periodic chain-like structures associated with the outer membrane (38, 75) and (ii) polar type IV pili (61, 79).

Chain-like structures. Reichenbach and coworkers conducted detailed electron microscopy studies with whole cells and subcellular fractions of the gliding myxobacteria Myxococcus fulvus and M. xanthus (strains Mx f65-9 and DK1622). In both organisms, similar structures that are believed to be components of the gliding apparatus were identified (Fig. 5) (38, 75). The basic elements of this proposed gliding apparatus in M. xanthus DK1622 are linear chain-like strands. These strands are composed of multiple ring-like structures, which are threaded evenly along elongated elements (Fig. 5). Each ring element in a strand has an average outer diameter of about 16.4 nm and is composed of six or more peripheral protein masses and possibly three small central masses. The rings are connected to each other by two parallel strings of filamentous proteins, the elongated elements, which attach at the inner side of a ring. They separate two neighboring rings evenly at a distance of about 15.5 nm. Often, three chain-like strands appear to assemble into parallel, ribbon-like structures where rings of neighboring strands are in lateral contact. Several ribbons form a belt which wraps helically around the cell. The cells appear to be completely covered by these belts. The strand structures are located in the periplasmic space and are associated with the outer membrane. A significant amount of strand-like structures was released only after lysozyme treatment, which suggests that the strands are also associated with the thin peptidoglycan layer. It is hypothesized that the connection between a ring element and the elongated element may be flexible and may be the site of force generation during gliding (Fig. 5) (38, 75).

View larger version (101K): [in this window] [in a new window]   FIG. 5.   Chain-like structures proposed to be involved in gliding motility. The images are from Myxococcus fulvus, and similar structures were observed in M. xanthus. (A) Isolated aggregates of chain-like strands showing different orientations of periodic structural elements. Arrowheads indicate rings. Bar, 50 nm. (B) Regular spacing and positioning of rings normal to the longitudinal axis of the strands (large arrowheads). Three strands form a morphological unit, the ribbon (solid triangles). Long arrows indicate proposed contracted units resulting in a herringbone pattern. Bar, 50 nm. (C) Three-dimensional model of the architecture of strands localized in the periplasmic space and anchored to the peptidoglycan layer and outer membrane. Picture courtesy of Freese et al. Three strands are presented. Conformation of ring elements which are located normal to the outer membrane is shown. The model is not drawn to scale. RE, ring element, framed; EE elongated element; CM, cytoplasmic membrane; OM, outer membrane; cema, central mass; pm peripheral mass. For a description, see the text. Panels A and B reprinted from reference 75 with permission. Panel C reprinted from reference 38 with permission.

Type IV pili. In early motility studies on M. xanthus, pili were recognized to be required for S motility (61). Pili, which are comprised of proteinacious filaments 10 nm in diameter and 3 to 10 µm long, are localized predominantly at one cell pole (61, 78) (Fig. 6). Recent studies revealed that these pili belong to the class of type IV pili which are also required for twitching motility. Twitching has been observed in many gram-negative microorganisms, including Eiknella corrodens, Moxarella osloensis, Neisseria gonorrhoeae (52, 73), enteropathogenic Escherichia coli (114), Pseudomonas aeruginosa (28, 132), and many other pseudomonads (16, 17, 52, 53, 54). Type IV pili are also referred to as type 4 fimbriae. In addition to their role as potential motility organelles, the type IV pili are essential components of several important cellular functions including adhesion to surfaces and phage binding. Molecular details of assembly and function of type IV pili in M. xanthus are discussed below in the context of S motility.

View larger version (115K): [in this window] [in a new window]   FIG. 6.   Polar pili on an M. xanthus cell. Picture kindly provided by Dale Kaiser. Bar, 0.5 µm. Reprinted from reference 61 with permission.

The A-motility system controls single-cell gliding. Single cells are visible at the edge of A-motile colonies (Fig. 2C). Videomicroscopy analysis of A-motile cells shows that individual cells glide with wild-type speed when well isolated from other cells (Fig. 2A and C), suggesting that the A-motility system includes all components needed for the machinery and the control of gliding of single, isolated cells. However, compared to the wild type, fewer cells appear to organize into small groups at the perimeter (57) (Fig. 2C). The rate of swarm expansion of A⁺S cells depends on the cell density to a similar extent to that of A⁺S⁺ cells (half-saturation cell density about 10⁸ cells/ml). The maximal swarming rate is only 0.65 µm/min on 1.5% agar (62). A⁺S cells are capable of performing elasticotaxis (37). In fact, the elasticotaxis response in these strains is more pronounced that in the wild-type strain DK1622. The swarming of growing A⁺S colonies is strongly reduced on a low-percentage (0.3 to 0.5%) agar surface, and the swarming rate increases with increasing agar concentration (107).

View larger version (24K): [in this window] [in a new window]   FIG. 7.   Localization of motility genes on the M. xanthus genome. AseI (outer circle) and SpeI (inner circle) restriction maps of the DK1622 chromosome are shown. Modified from references 26 and 76 with permission.

View larger version (57K): [in this window] [in a new window]   FIG. 8.   Stimulation of motility of the agl and cgl A-motility mutants. Drops (on the left of each image) of a liquid culture of A-motility mutants (AS+) (recipient cells) were spotted on CTT agar plates and allowed to dry, and another drop of DK6204 (donor cells), a nonswarming mglBA mutant, was added to intersect with the former spot. (A) cgl recipient strain (AS+). After a few hours, single cells of DK1219 are visibly gliding as individual cells. (B) agl recipient strain (AS+). No stimulation of single-cell movement is observed. Bar, 20 µm.

The S-motility system controls gliding of cells in swarms. S-motile colonies show a clearly defined, undulating edge which is indicative of movement of cells within the colony (Fig. 2B). The cell density dependence of the swarming rate is decreased in these mutants compared to the wild type. The maximal swarming rate of three AS⁺ strains tested is approximately 0.45 µm/min, with a half-saturation cell density of 7 × 10⁸ cells per ml (62). Cells that are motile by S motility only are unable to respond to physical stress forces in the substratum (37). Accordingly, the absence of S motility causes a dramatic reduction in swarming of growing A⁺S colonies on a 0.3% agar surface, and the swarming rate increases with increasing agar concentration (107). Most of the S-motility mutants are also defective in fruiting-body formation (57, 77).

(i) pil genes. Analogous to defining A motility, genes belong to the S system if they result in a nonswarming double-mutant colony when they are mutated in an AS⁺ mutant (57). Numerous S-motility genes have been identified in mutant screens (Table 2). In the original genetic analysis by Hogkin and Kaiser, two major regions of S-motility genes were identified, the sglI and sglII regions (57). In recent research, it became clear that most of the sgl genes of the sglI region showed significant similarity to genes involved in pilin expression and processing and in export, assembly, and function of type IV pili in other bacteria. Therefore, these genes were concluded to be pil genes (126, 137-140, 141) (Fig. 9). Twitching motility is a type IV pilus-based mode of surface translocation and has been observed in many gram-negative microorganisms (16, 17, 28, 52, 53, 54, 132). Our current understanding of type IV pilus structure and biogenesis rests mostly on studies conducted with P. aeruginosa and N. gonorrhoeae (for recent reviews, see references 3, 30, and 39). Genes involved in type IV pilus export and assembly have multiple homologs to components of DNA uptake and protein secretion systems. These systems have in common that they catalyze a vectorial transport of larger polymers (polypeptides, DNA) across the outer and inner membranes of gram-negative bacteria. These transport systems can function either for import (DNA) or for export (pili, proteins). The strong similarity between the gene products in these systems suggests that these different systems may operate by a common mechanism. In the following discussion, only the model for type IV pilus formation and function is discussed, without elaborating the experimental evidence in support of that model (see volume 192 of Gene [1997] for numerous specific reviews).

View larger version (10K): [in this window] [in a new window]   FIG. 9.   Organization of M. xanthus pil genes. Compiled from references 102, 125, and 139 to 141. For details, see the text.

View larger version (16K): [in this window] [in a new window]   FIG. 10.   Models for generation of cell movement by type IV pili in M. xanthus. To illustrate the model of pilus-dependent S motility, individual cells are not drawn to be in contact with other cells. It should be noted that cell-cell contact is required for S motility. (A) A depolymerization-polymerization () shortens a pilus, thus generating displacement mostly in the direction of the long axis of the cell. (B) Pili attach to neighboring cells at "adhesion sites" (--), e.g., which are part of the gliding motor of A motility and are involved in A-motility-dependent gliding. Linked to the moving adhesion sites by a pilus, a neighboring cell is "dragged" along, thus performing a translocation in the direction of the long axis of the cell. (C) A combination of pilus retraction-extension and adhesion site-dependent displacement (A plus B).

(ii) tgl gene. In one S-motility mutant, tgl (for transient gliding), S motility and type IV pilus assembly can be stimulated by aligning tgl cells with cells of a tgl⁺ strain (57, 61, 102, 103, 124, 126). Because of the stimulation phenotype, tgl is, in a sense, the counterpart in S motility to the cgl genes in A motility. Similar to CglB, Tgl appears to be a lipoprotein which localizes to the outer membrane (102, 103). Tgl contains six tandem tetratricopeptide repeats, a motif which is known to be involved in protein-protein interactions. The observed stimulation by Tgl is believed to be due to Tgl acting as a pilus assembly factor rather than as a cell-cell signal (124).

(iii) Other S-motility genes. Next to the pil genes and tgl, there is a third group of S-motility genes that affects cell surface structures in M. xanthus: the genes of the dsp locus and the linked sglK locus. Both genes are also required for formation of extracellular fibrils (7, 108, 109, 111, 129). Mutants defective in dsp were isolated as S mutants and grow as dispersed cultures in liquid medium (88a). When crossed with an A mutant, Adsp double-mutant colonies show some reduced swarming after prolonged incubation. dsp mutants form pili but carry aberrant fibrils and are defective in cell cohesion and development (6, 7). Fibrils are extracellular, irregular branched structures of variable width and length (5, 7) and are required for cell-cell cohesion (5, 6). They are composed of polysaccharides and contain a class of integral fibril protein, IFP-1, which is defined as being released from fibril material only after treatment with sodium dodecyl sulfate and -mercaptoethanol (8, 9, 129). It has been suggested that the different sizes of IFP-1 proteins result from multimer formation of a single small-molecular-size subunit whose amino acid composition and N-terminal sequence were determined (9). Fibrils are found mostly on cells in groups where they establish cell-cell and cell-substratum contacts (7). A link between exopolysaccharides (fibrils) and type IV pilus-dependent S motility may exist, similar to the link between alginate production and twitching motility in P. aeruginosa (131). AlgR and FimS, which are a regulator and a sensor, respectively, are required for alginate production and for twitching motility. This sensor-regulator couple does not appear to regulate PilA production in P. aeruginosa (131), raising the possibility that the produced exopolysaccharides are mechanistically required for twitching movements. Accordingly, fibrils may have an analog function in S-motility gliding of M. xanthus.

AS double mutants. The nonswarming phenotype of AS double-mutant colonies has served as the conceptual basis for defining the two gliding systems in M. xanthus (56, 57). Motility mutants identified by the double-mutant test will most likely include those defective in structure, assembly, and function of the gliding motor elements (e.g., pil genes). Mutations in genes whose products play a modulating role on the activity of the gliding motors may only partially affect the swarming pattern of the A- or S-motility system when observed in a null mutant background of the other motility system. These mutants may exhibit reduced but not completely abolished movements. For example, mutations in frzCD, frzF, or frzE, when crossed into an A mutant, show a reduced S-motility swarming (37) (Tables 1 and 2; also see below). Also, genes of the dsp locus may play a regulatory role in S motility, because S-system swarming is severely reduced in Adsp double mutants.

Control of M. xanthus motility by the frz genes. The frz locus was identified by a set of mutants which display a unique developmental defect in aggregation and fruiting-body formation (144). Under starvation conditions, these frz mutants form entangled, "frizzy" aggregates. To date, seven frz genes have been identified within this locus: frzA, frzB, frzCD, frzE, frzF, frzG, and frzZ (13, 83, 123). In addition to the aggregation defect, the frz mutants carry a defect in colony swarming on low-percentage agar that somewhat resembles that seen in S-motility mutants (37, 107). Mutations in the frz genes also reduce the level of sporulation (105, 123) although this effect is strain dependent. Interestingly, single cells of most frz mutants have an altered frequency of reversing their direction of movement; wild-type cells reverse their direction once every 5 to 7 min (0.17 reversal min¹), while most frz mutant cells reverse on average once per hour (<0.02 reversal min¹). One notable exception was caused by a Tn5 insertion at the 3' end of the frzCD gene (a frzD mutant). Single cells bearing this mutation exhibit an increased (1.5 reversals min¹) rather than a decreased reversal frequency (12, 118). Gliding velocities, however, are unaffected in frzE mutants, as shown by high-resolution videomicroscopy studies (12, 118).

Motility and the mgl genes. In addition to isolating A- and S-motility mutants, Hodgkin and Kaiser identified one locus that, with a single mutation, rendered a colony completely nonswarming (57). Ten independent mutations were identified in this locus. Because of the colony-swarming defect, this locus appeared to be essential to both A and S motility and was called mgl (for mutual function for gliding). Therefore, it was reasoned that the mgl gene(s) might encode components of the gliding motor. In addition to abolishing swarming, mgl mutants are defective in fruiting-body formation and sporulation, presumably due to their inability to conduct C-factor signaling during development (70, 120). A molecular analysis of the mgl locus revealed that the mgl operon contains two genes, mglA and mglB (120, 121). A mutation in mglA causes the severe swarming and developmental defect, whereas mglB mutant colonies exhibit only partially reduced swarming, aggregation, and sporulation (49). This reduced swarming correlates with the reduced cellular level of the cytoplasmic MglA protein (49). A stabilizing interaction between MglA and MglB is suggested by the finding that in an mglB mutant, the cellular level of MglA is only 15 to 20% of that in wild-type cells but the level of mglBA mRNA is unaffected (49). MglA is not essential for growth.

1

Considering the above-discussed genetic, molecular, ultrastructural, and behavioral studies of M. xanthus, it is tempting to summarize these observations in the following speculative model for A-motility gliding of isolated M. xanthus cells (Fig. 11). This model also rests on observations and models, formulated for Cytophaga strain U67 and Flexibacter, that were conducted by Lapidus and Berg (72), Ridgway and Lewis (100), and other authors (see below) on other gliding bacteria. It should be kept in mind that because of the operational definition of gliding, the molecular mechanism(s) to achieve gliding movement of well-isolated single cells can differ among microorganisms.

View larger version (38K): [in this window] [in a new window]   FIG. 11.   Speculative model for single-cell gliding (A motility) in M. xanthus. The picture is a close-up of the interface of the surface of a gliding cell with the substratum and shows one of the numerous motor units in detail. The cell is moving from left to right. In this model, the force generator (molecular motor) is anchored to the cell wall (rigid peptidoglycan layer), which functions as a skeleton of a bacterial cell. A force of action is generated mainly in the direction of the long axis of the cell and is coupled to an adhesion site in the outer membrane, which interacts with the substratum. Because the coupling of the adhesion site to the surface is tight, the force generator moves to the right relative to the adhesion site, and the relative displacement is indicated by the arrows. The chemical energy that the force generator transforms into mechanical work is delivered by an energy transducer and is derived from the electrochemical ion potential across the cytoplasmic membrane.

Based on observations on the motility behavior of individual cells as well as on bead movement along the cell surface, it is generally believed that multiple motor elements exist along the cell body (see above) (45, 72, 100, 117). Because a gliding cell moves only when the force of action is generated at a rigid cellular structure such as the cytoskeleton or the cell wall, it is expected that the force-generating elements for gliding are connected with the cell wall as well as with the substratum (Fig. 11). This predicts that the motility apparatus catalyzing the conversion of chemical energy into mechanical energy may not be localized inside the cytosol. In general, ATP is not believed to be present in the periplasmic space, and in Flexibacter, gliding is most likely to be powered by the proton motive force (98). It is tempting to speculate that in M. xanthus, the energy for A-motility gliding is also directly derived from the proton motive force. A model for energy transduction between the cytoplasmic membrane and the periplasmic motor elements can be developed by analogy to the function of the energy transducer TonB. TonB, in a complex with ExbB and ExbD, mediates the active transport of iron siderophores across the outer membrane in E. coli (for a recent review, see reference 88). TonB is a cytoplasmic membrane protein that extends through the periplasmic space to contact the iron chelate receptor in the outer membrane. Energy is transduced to the receptor via proton motive force-induced conformational changes of the proteins involved. Accordingly, in A-motility system gliding, the energy of the proton motive force could be transduced by a TonB-like protein or protein complex to induce a conformational change in the gliding force generator.

It can be speculated that the products of A-motility genes include those that specify the structure of the gliding motor and gliding-specific adhesion sites, as well as those required for assembly of the structure, energy transduction, and regulation of the motor activity. A motility operates best on high-percentage agar surfaces, and A-motile cells respond to stress forces in the substrate matrix. These observations are consistent with the notion that cellular motor elements interact with rigid polymers in the substratum surface, e.g., in the process of force transmission. On a low-percentage agar surface, only few contact points are available, which could explain the reduced swarming by A motility in A⁺S cells on that surface. CglB could be required for force transmission to the substratum. The residual movement observed in the AS (cglB pilR) mutant could be due to ineffective force transmission by other, less specific molecules. Such function of the CglB protein is consistent with its property of being transferred between cells under certain conditions.

Although presently no experimental evidence exists that the chain-like structures are the motor elements for gliding, they nevertheless fit certain requirements expected for the gliding motor (Fig. 5 and 11). For example, the superstructure is anchored to the peptidoglycan layer and is in contact with the outer membrane. The chain-like strands that assemble in larger periodic ribbon-like and belt-like structures cover the complete cell in a longitudinal fashion. A ring could, in a sense, represent a single motor element. The fundamental step in force generation of this motor element could be a conformational change of the ring structure with respect to the elongated filaments in form of a tilt. Such a tilt would result in a small displacement of the outer perimeter of a ring with respect to the elongated filament. Because the outer perimeter of a ring is postulated to be in contact with the adhesion sites in the outer membrane, the vectorial force along the long axis of the cell is transduced to the substratum. Alternatively, a tilting of the rings in a way like that described by Freese et al. (38) could lead to a local contraction of the superstructure and thus transduce the force as well. A "relaxation" of that conformational state would be required for the molecular motor to return in its prestroke state. Because each strand contains a large number of ring-like structures, a M. xanthus cell carries multiple motor elements along the complete cell body. The presence of multiple motor elements were predicted from studies of U-shaped cells and of cells that are fortuitously attached with only one cell pole. The bead movement observed along the entire cell body may also support this hypothesis. Bead movement in the opposite direction of the cell translocation may indicate that the beads tagged the relaxation to the prestroke state.

Although the above-described thoughts are very hypothetical, some specific hypotheses can be derived and tested in experiments. These questions include the following. Is the superstructure involved in gliding? Is the observed cellular structure the molecular machinery that converts chemical energy into mechanical work? How is force transmitted to the substratum? How is the activity of a motor element regulated? How is gliding speed regulated? Is it due to adjustable activities of individual motor elements or to activation of different quantities of motors that operate in an on-only/off-only mode? Furthermore, how is the gliding movement of isolated single cells coordinated to result in macroscopic A motility, and how is type IV pilus-dependent movement coordinated to result in S motility?

Early studies on gliding motility focused on Cytophaga spp., since these cells are among the fastest of gliding bacteria, moving at speeds of up to 2 µm/s on a glass surface. Therefore, these gliders can be easily observed under the microscope. Although many similar gliding properties have been observed between Cytophaga johnsonae (which was recently reclassified as Flavobacterium johnsoniae [10]) and Cytophaga strain U67, the gliding mechanisms used by these species may be distinct. Based on studies of sensitivity to phage infection, Cytophaga strain U67 may be a close relative of F. johnsoniae (97).

Using video microscopy, Lapidus and Berg (72) conducted thorough behavioral studies on gliding movements of isolated Cytophaga strain U67 cells on the surface of glass slides. Slime trails, which were often postulated to be involved in active gliding in other organisms, were not detected in association with Cytophaga strain U67 movement. During gliding in either the forward or reverse direction, cells were observed to suddenly enter into abrupt clockwise or counterclockwise rotations (pivoting) around either cell pole at a rate of approximately 0.5 Hz; occasionally, faster pivoting was observed, sometimes for extended periods. Incomplete pivots, or flippings, where one cell pole is lifted and deposited somewhere else in the absence of complete rotations, have also been described. Cytophaga strain U67 cells were not observed to flex or to rotate during gliding. The gliding speed in Cytophaga strain U67 was demonstrated to be independent of the cell length, an observation which has also subsequently been made for M. xanthus cells (118a). Polystyrene latex beads that were found to adhere to these cells (72, 91) were used as tools to help identify potential subcellular motility elements. These spheres were observed to move from one cell pole to the other at speeds similar to the normal gliding speed of a cell. These events were independent of whether cells were in suspension or were attached to a surface (72). Bead movements could also stop and then resume in the opposite direction before they had reached the cell end. One particularly interesting observation was that multiple beads on the cell surface could move independently of each other. While one bead could be seen moving along the cell body, a second bead on the same cell could stop while a third could move in the opposite direction to the first bead. The speed of bead movement along the cell was independent of bead diameter, which ranged from 0.1 to 1.3 µm. Depletion of oxygen resulted in cessation of both gliding movements and movements of beads on cells, although, interestingly, attached spheres did not exhibit any noticeable Brownian motion. From these observations, a model of gliding for Cytophaga strain U67, in which adhesion sites in the outer membrane, which attach to the surface (or to polystyrene beads), are moved in the outer membrane on tracks that are fixed to the rigid cell wall, was proposed (72).

More recently, interference reflection microscopy was used to examine cell-substratum contact during gliding movement in Cytophaga strain U67. The entire length of the cell body was rarely seen in contact with the substratum (42). The observed cell-substratum sites were variable and moved relative to the glass substratum. Surface-exposed proteins that form physical contact with a solid substratum were reported after their identification by using substratum-immobilized ¹²⁵iodide (21). This set of currently uncharacterized proteins may contain those that mediate gliding-associated cell-substratum adhesion. Cytophaga strain U67 shows some degree of curvature in cell shape and was observed in this study to rotate during movement (42). The rotation of Cytophaga strain U67 cells around the long axis was sinistral when gliding on a glass surface, and a pitch of the helix about 79° ± 3° for every 8.3 µm of translational gliding movement was observed (42). Surfactants were shown to inhibit swarming and gliding (20).

In F. johnsoniae, sulfonolipids were detected as an unusual component of the cell membrane (1). The role of these sulfonolipids was speculated to be in presenting specific polysaccharides to the outer cell surface (41, 44). An uncharacterized mutant defective in sulfonolipid formation was unable to glide or to move polystyrene beads along the cell body. In addition, the mutant was insensitive to phage infection (45). When the sulfonolipid precursor cysteate was provided to the mutant, gliding was rapidly restored. However, bead movement and phage sensitivity took longer to be restored, suggesting that the cell surface features required for bead movement and phage sensitivity are different from those required for gliding (45). A causal relationship between phage sensitivity and bead movement, on the one hand, and gliding motility, on the other, was considered unlikely (43, 45). Recently, negative chemotaxis in F. johnsoniae in response to the repellents H₂O₂ and OCl, and N-chlorotaurine was reported (74).

Many motility mutants of F. johnsoniae that are defective in colony swarming have been isolated (25). One subclass of these mutants were nonmotile ("truly nonmotile") in wet-mount preparations and exhibited other pleiotropic phenotypes such as defects in chitin digestion, phage absorption, and cell adherence (25, 135). However, research on these and other F. johnsoniae mutants was hampered by the absence of a genetic system to facilitate further molecular analyses. A breakthrough came in 1996, when McBride and Kempf reported successful mutagenesis of F. johnsoniae by using Bacteroides transposon Tn4351 (85). A plasmid, based on the cryptic plasmid pCP1 (isolated from the closely related F. psychrophilus), that stably replicates in F. johnsoniae was developed. This plasmid served as the basis for constructing a genomic library of this microorganism. Using this library, a single open reading frame, named gldA, was identified and was found to be sufficient to complement the motility defect of 4 of 61 independently isolated nonmotile mutants (2). The predicted amino acid sequence of GldA shows strong homology to ABC transport proteins. GldB and GldD are predicted to be membrane proteins. Other mutations rendering cells unable to glide map to genes encoding proteins involved in cell surface lipopolysaccharide synthesis (85a). Further use of these tools for genetic investigations of gliding in F. johnsoniae will undoubtedly uncover new insights in the gliding mechanism of this bacterium. Comparing the molecular components identified in F. johnsoniae with those in M. xanthus will significantly enhance our understanding of this translocation process and its diversity.

In contrast to the above gliding bacteria, Flexibacter polymorphus is a filamentous bacterium that forms multicellular filaments which are enclosed by a common outer membrane. Extensive studies of gliding movements of F. polymorphus filaments were conducted by Ridgway and Lewin (100). Similar to M. xanthus and Cytophaga strain U67, F. polymorphus filaments start to glide immediately once in contact with a surface (100). When suspended in liquid medium and when forming contact only with each other, F. polymorphus filaments were observed to glide relative to each other (100). Gliding of filaments on an agar surface was accompanied by a sinistral rotation of the cell body. The gliding speed of a filament is approximately 12 µm/s and is independent of the filament length. F. polymorphus filaments were observed to spin or pivot around the long axis of the cell when only one cell end was attached to the substratum. By using dextran to manipulate the viscosity of the liquid medium, it was found that the gliding speed was inversely proportional to the viscosity, suggesting that the gliding motor operates at constant torque (100). Filaments exhibited a tactile response and reversed the direction of movement by gliding backward upon contact with an object. The frequency of reversal was observed to be inversely correlated with the filament length. Polystyrene beads were observed to move across the entire cell length at approximately the same speed as the filaments glide. Bead movement occurred in a more irregular fashion than did analogous movements in Cytophaga strain U67. Small beads tended to move in a more helical fashion, whereas particles of >0.3 µm were translocated less uniformly and did not circumvent the filament. Movement of multiple beads on a single filament seemed to be independent of each other. Bead movement was also observed on nongliding filaments and filaments suspended in liquid. Because bead movement was observed in nongliding filaments, it was concluded that bead movement may not necessarily be coupled to gliding movements. Extracellular fibrils that originate laterally from the cell body are involved in cell-substratum adhesion. The chemical composition of the F. polymorphus fibrils and associated exopolysaccharide is unknown, and these fibrils may very well be different from those observed in M. xanthus. The energy to power the gliding machinery is believed to be derived from the electrochemical proton potential (98). No genetic studies have been reported for F. polymorphus.

As indicated above, "gliding" is an operational definition, and considering the diversity of microorganisms capable of this type of surface translocation, different mechanisms may operate in different microbes. Two interesting observations on motility in cyanobacteria are important to mention when considering non-flagellum-based motility: swimming of nonflagellated Synechococcus, and slime extrusion in gliding of Phormidium unicatum and Anabena variabilis. In 1985, Waterbury et al. reported the isolation of several strains of the unicellular cyanobacterium Synechococcus that are capable of rapid swimming motility in the absence of flagella (128). Subsequent studies showed that their motility is dependent on a sodium gradient as the energy source (134). Furthermore, Ca²⁺ was found to be required for motility, and a highly abundant Ca²⁺ binding protein that is essential for motility was identified in the cell surface (19, 93). Self-electrophoresis was ruled out as a mechanism for this very interesting mode of movement (92). Based on theoretical considerations, a mechanism of traveling (longitudinal) surface waves was considered instead (35). A similar mechanism of wave propagation along the cell surface also presents an attractive model for single-cell gliding motility. Most behavioral observations made on gliding microorganisms (single-cell studies, bead movement, etc.), as well as the ultrastructural findings in M. xanthus, are consistent with such mechanism. However, gliding microorganisms have to be able to adhere reversibly to the substratum. Notably, M. xanthus, F. johnsoniae, Cytophaga strain U67, and F. polymorphus have not been reported to move by swimming. Research on this unusual mode of swimming in unicellular Synechococcus may provide unexpected insights into the diversity of bacterial translocation mechanisms that may affect our thinking on gliding motility as well.

Recently, the slime extrusion hypothesis of Ridgway (99) was revived by reports by Hoiczyk and Baumeister while studying the filamentous gliding cyanobacteria P. unicatum and A. variabilis (59). Using electron microscopy studies, these authors reported finding "junctional pore complexes" at the cross wall or septa of the filaments. These structures show some similarity to the type III secretion system apparatus in S. typhimurium (71). The junctional pore complexes seem to be involved in slime secretion in these cyanobacteria. As in many other gliding microorganisms, slime production is often associated with gliding. It was speculated that directional slime extrusion through the junctional pore complexes may be the motor for gliding in these cyanobacteria (59). However, no physiological evidence supports this hypothesis, and is it not clear whether slime formation is the result or cause of gliding motility.