|Home | About | Journals | Submit | Contact Us | Français|
Cells of Myxococcus xanthus will, at times, organize their movement such that macroscopic traveling waves, termed ripples, are formed as groups of cells glide together on a solid surface. The reason for this behavior has long been a mystery, but we demonstrate here that rippling is a feeding behavior which occurs when M. xanthus cells make direct contact with either prey or large macromolecules. Rippling has been observed during two fundamentally distinct environmental conditions: (i) starvation-induced fruiting body development and (ii) predation of other organisms. Our results indicate that case (i) does not occur in all wild-type strains and is dependent on the intrinsic level of autolysis. Analysis of predatory rippling indicates that rippling behavior is inducible during predation on proteobacteria, gram-positive bacteria, yeast (such as Saccharomyces cerevisiae), and phage. Predatory efficiency decreases under genetic and physiological conditions in which rippling is inhibited. Rippling will also occur in the presence of purified macromolecules such as peptidoglycan, protein, and nucleic acid but does not occur in the presence of the respective monomeric components and also does not occur when the macromolecules are physically separated from M. xanthus cells. We conclude that rippling behavior is a mechanism utilized to efficiently consume nondiffusing growth substrates and that developmental rippling is a result of scavenging lysed cell debris.
Myxococcus xanthus is a predatory -proteobacterium that is able to lyse a variety of other bacteria and grow on the nutrients released (7, 21). However, it is best known for its response to nutrient depletion in which growth halts and approximately 105 cells aggregate into fruiting bodies, where cells differentiate into metabolically quiescent spores (4). Predation and development reside on opposite ends of the M. xanthus life cycle, and there is little known about the connection between the two processes, yet they share some common mechanistic ground. There is evidence for both processes utilizing motility, intercellular communication, and of course, information about nutritional status (19). One striking commonality is that both predation and development have been observed to induce rippling motility behavior—the organization of cells into moving groups that resemble the movement of waves rippling on water (20).
The movement of M. xanthus cells on a solid surface is typically observed as a tangled collection of cells moving both individually and in streaming groups. Surface motility occurs through the use of two complementary gliding motility systems. Type IV pilus-based motility is thought to pull cells in a manner similar to the twitching motility of Pseudomonas aeruginosa (14), while a poorly understood slime extrusion mechanism is thought to be able to propel cells forward (8). Both systems require the chemotaxis-like frz pathway to modulate cell reversals and produce directed movement (17, 30). During rippling, cells are organized into nearly parallel lines of high cell density, with troughs of low cell density in between (24). Coordinated movement of cells perpendicular to the axis of the cell line creates the appearance of a traveling wave. However, when cells of neighboring parallel waves come into contact, cell reversals are induced such that the two waves reflect off each other (26). Thus, although cells aggregate into waves during rippling, there is no net cell displacement throughout the swarm, as each wave of cells oscillates between neighboring waves. Directed movement of cells can also be observed during fruiting body formation, but under these circumstances, cells tend to move in the direction of the stream, with net movement into aggregation centers producing the early fruiting body structure (27).
Although rippling has been observed during predation, predatory rippling is rarely mentioned in the literature and rippling has primarily been studied in pure cultures during starvation-induced development. Developmental rippling has been observed to occur spontaneously and sporadically in starving cultures of M. xanthus prior to and concurrent with fruiting body formation (6). Recent experimental and theoretical study of this process indicates that the rippling pattern can be produced through a minimal requirement of cell-cell contacts and an internal biochemical oscillation system (1, 9, 10, 26, 31). Developmental rippling has been proposed to rely on the levels of the starvation-induced C-signal, a 17-kDa form of the CsgA protein (12, 27). The C-signal is thought to be presented on the surface of one cell and to interact with an unidentified receptor on a neighboring cell to transmit the signal (27). The CsgA protein is induced by starving conditions, with CsgA levels rising throughout the developmental process and proposed to induce first rippling and then aggregation and sporulation, all as a function of CsgA concentration (6, 11).
The C-signal model, however, does not sufficiently explain why rippling behavior is observed during predation. Under predatory conditions, M. xanthus cells are digesting the available prey cells and growing on the nutrients released, yet they still demonstrate rippling behavior. It is unknown if predatory rippling behavior requires the starvation-induced C-signal or if it operates via an alternative mechanism. In this study, we demonstrate that rippling behavior is a general mechanism required for efficient predation and that developmental rippling only occurs under conditions which mimic predation, indicating that rippling is solely a predatory behavior in M. xanthus. We conclude that rippling behavior serves to maximize predation efficiency and nutrient scavenging.
M. xanthus strains DZ2 and DK1622, both commonly referred to in the literature as wild type, were used in this study. Strain DZ4469 (DZ2 ΔpilA::tet) was kindly provided by D. Zusman (28). Escherichia coli strain β2155 was used as prey. β2155 was used for quantitative predation assays, since it is a Kanr diaminopimelic acid (DAP) auxotroph and thus prey cell densities remain relatively constant in any media lacking DAP. For routine culturing, M. xanthus was grown in Casitone yeast extract (CYE) broth, which contains 10 mM morpholinepropanesulfonic acid (MOPS), pH 7.6, 10 g/liter Casitone, 5 g/liter yeast extract, and 8 mM MgSO4. E. coli was cultured in LB broth (23). Kanamycin was supplied when appropriate at 100 μg/ml and DAP at 100 μg/ml. CF agar was utilized for analysis of M. xanthus cells under low nutrient conditions. CF contains 10 mM MOPS, pH 7.6, 1 mM KH2PO4, 8 mM MgSO4, 0.02% (NH4)2SO4, 0.2% citrate, 0.2% pyruvate, and 150 mg/liter Casitone. MMC buffer was used for harvesting and washing of cells and consists of 10 mM MOPS, pH 7.6, 4 mM MgSO4, and 2 mM CaCl2. Saccharomyces cerevisiae strain PJ69-4A (provided by P. James) and Bacillus subtilis strain OI1085 (provided by G. Ordal) were cultured according to standard methods.
Cultures of M. xanthus and E. coli were harvested at mid-log phase and washed three times in MMC buffer. For qualitative assays, 30% India ink was added to M. xanthus cells for visualization, and M. xanthus and E. coli cells were pipetted 1 mm apart in 2-μl aliquots containing ~2 × 107 cells each. For quantitative analysis, cell suspensions were pipetted onto either CF or CYE media in 10-μl aliquots, with ~107 M. xanthus cells plated first and allowed to dry, followed by the addition of ~109 E. coli cells in 10 μl and incubation at 32°C. To measure the number of E. coli survivors, cells were harvested from plates at the time points indicated and resuspended in MMC buffer. The resuspension was then serially diluted and cells plated on appropriate media to determine the number of CFU. Addition of kanamycin to the media prevents the outgrowth of M. xanthus cells present in the suspension. Three samples were harvested and quantified at each time point. The same procedure was used for the analysis of components which induce rippling, with each substrate that was tested resuspended in both water and in melted agar to a final concentration of 1.5% immediately before plating. Substrates were tested at a concentration of 10 mg/ml. Prey were tested at densities of 108 cells/ml for B. subtilis, 1010 PFU/ml for P1 phage, and 107 cells/ml for S. cerevisiae. Each substrate was tested in the presence of at least three independent cultures.
Rippling behavior was observed using a Nikon SMZ10000 dissecting microscope and a Nikon Eclipse E400 phase-contrast microscope. Images were captured using QImaging camera and software.
Mid-log-phase cultures of M. xanthus cultures were harvested, washed in MMC buffer, and concentrated to ~108 cells/ml equivalent. Replicate 1-ml samples of each cell suspension were added to 12-well plates and incubated at 32°C. To quantify the number of intact cells, the suspension was aspirated, transferred to a 1.5-ml tube and sonicated at low power for 3 s to disperse large cell clumps. Direct cell counts were performed using a hemacytometer at the time points described in Results. Three independent cultures of each strain were examined, with three samples analyzed at each time point.
When incubated on low nutrient media at sufficiently high cell density, M. xanthus cells will glide into aggregates that become fruiting bodies. However, even on low nutrient media, not all of the cells congregate into fruiting bodies. Many cells glide away from the initial inoculum to colonize the agar surface. To analyze prey-induced rippling, ~2 × 107 M. xanthus cells mixed with India ink were pipetted adjacent to ~2 × 107 E. coli cells on low-nutrient CF media. As M. xanthus cells colonize the agar surface, the swarm expands in a thin, nonuniform layer of individuals and groups that results in a rough, tangled appearance (Fig. 1A and B). Direct contact of M. xanthus cells with suitable prey bacteria, such as E. coli, results in penetration of the prey colony by M. xanthus cells and subsequent lysis and digestion of the prey. Tangled behavior continues in the leading edge of M. xanthus cells as they migrate through the prey colony, but within 16 to 24 h of the initial contact with prey bacteria, rippling behavior is induced in trailing cells, resulting in aggregations of cells into nearly parallel waves which form perpendicular to the direction of the M. xanthus swarm migration (Fig. 1C and D). Under these conditions, rippling pattern formation is localized solely to the initial area occupied by the prey colony and continues for several days. Areas of the M. xanthus swarm adjacent to the prey locale, but outside of the prey colony, do not display the rippling phenotype at any time (Fig. 1E and F). Note that even as the swarm moves beyond the prey colony, and the nutrients contained therein, cells display tangled rather than rippling behavior. Also, several fruiting body aggregates can be observed forming near the initial M. xanthus inoculum; however, rippling behavior is not observed associated with these fruiting bodies.
The induction of rippling during predation indicates that rippling may be beneficial or even necessary for the efficient digestion of available prey substrates. To test whether rippling is required for predation, we incubated M. xanthus cells in the presence of E. coli on low- and high-nutrient medium. Preliminary results had indicated that rippling behavior is inhibited by excessive nutrients, although entry into the E. coli prey colony is not inhibited. To examine this process further, we used an E. coli DAP auxotroph (β2155) as prey, since the viable prey cell levels remain relatively constant in this strain on both CF and CYE media lacking DAP. Incubation of ~109 cells of β2155 in the presence of ~107 cells of M. xanthus strain DZ2 on CF media results in rapid lysis of prey, with 90% of the E. coli cells lysed in the first 24 h and the E. coli CFU dropping below detectable levels by 48 h (Fig. (Fig.2).2). By comparison, predation occurs much more slowly when the cells are incubated on ripple-inhibiting nutrient rich medium, with >90% of prey cell lysis not observed until 96 h. It should be noted that while viable E. coli cells are not detected after 48 h of incubation on CF in the presence of DZ2, rippling is still observed for at least 72 h beyond this point. This indicates that while live prey bacteria may induce rippling, complete lysis of the available prey bacteria does not correlate with the termination of rippling behavior.
The differences observed in predation under high- and low-nutrient conditions may be due to other factors besides rippling, such as differences in production levels of lytic enzymes. Analysis of a DZ2 ΔpilA mutant (DZ4469), which is unable to move by the type IV pilus motility system (32) and also unable to ripple (data not shown), indicates that it is defective in predation at both nutrient levels tested. On CF there are ~105 viable E. coli cells remaining after 48 h. After 96 h, E. coli survivors are not detected, indicating that lytic enzymes are still produced in this strain and that the E. coli cells are eventually lysed, but the rate is significantly slower than that of the parent strain. In the presence of high nutrient levels, E. coli lysis by the ΔpilA strain is indistinguishable from the basal rate of lysis of E. coli cells alone.
The observation of rippling continuing beyond the completion of prey lysis indicates that the substrate(s) which stimulate rippling may still be present. To determine what components of prey are capable of inducing rippling, we plated ~107 M. xanthus cells in a 10-μl suspension adjacent to a 10-μl aliquot of the test substrate on CF plates and allowed them to dry. In this assay, heat-killed E. coli cells and the insoluble fraction from lysed E. coli cells are both capable of inducing rippling behavior (Table (Table1).1). The soluble E. coli fraction was unable to induce rippling. A previous study by Shimkets and Kaiser demonstrated that M. xanthus rippling behavior is inducible by peptidoglycan extracts from M. xanthus as well as several other proteobacteria and gram-positive bacteria (24). Our data confirms this result, but further testing indicates that several other substrates also serve to induce rippling. We found that in addition to E. coli, B. subtilis, S. cerevisiae and P1 phage are all capable of inducing rippling. While B. subtilis has a peptidoglycan structure similar to that of E. coli, P1 phage and the yeast S. cerevisiae are two biological entities that do not contain peptidoglycan.
To determine if M. xanthus cells will ripple in the presence of molecules other than peptidoglycan, we incubated M. xanthus cells with bovine serum albumin and salmon testes chromosomal DNA. These substrates were chosen based on their large size and their representation of amino acid and nucleic acid composition. In both cases, rippling was observed, although the behavior was not as prolonged as with live cells or peptidoglycan. Examination of the monomeric components of these macromolecular substrates indicates that they fail to induce rippling under the conditions of this assay. Peptidoglycan components N-acetylglucosamine, diaminopimelic acid, Casitone (a mixture of amino acids and small peptides from hydrolyzed casein protein), and deoxynucleoside triphosphates were unable to induce rippling behavior. When physical separation of the M. xanthus cells from each substrate was performed through premixing of the substrate with melted agar prior to addition in the assay, the M. xanthus rippling response was not observed, indicating that direct cell-to-substrate or cell-to-prey contacts are required for this behavior.
Our analysis of rippling behavior during predation in M. xanthus strain DZ2 indicated that while rippling is readily induced by the presence of prey or prey components, it was not observed at all in our control samples lacking prey (Fig. (Fig.3).3). Other groups have reported that rippling will occur spontaneously on low-nutrient media in the absence of prey; in fact, this “developmental” rippling has been the focus of study on rippling behavior for a number of years (27). The apparent discrepancy may be due to the M. xanthus strain used, as DK1622 is most commonly used during experimental analysis of rippling behavior. The differences between the DK1622 and DZ2 have not been specifically characterized, but the DK1622 strain has undergone significant genetic modification, including the loss and subsequent restoration of both motility systems (2, 3, 29). The DZ2 strain has not been modified from the original isolate other than through culturing for strain propagation (D. Zusman, personal communication). Figure Figure33 shows that both DZ2 and DK1622 ripple when preying on E. coli, but only DK1622 was observed to ripple in pure culture. Rippling in DK1622 pure cultures is nonuniform, and the direction of wave propagation is unpredictable. The rippling pattern displayed during predation occurs uniformly and predictably for both strains, with the rippling pattern propagating out from the initial M. xanthus inoculum in concentric waves. However, there is still a consistent difference in the rippling pattern displayed by the two strains with DK1622 showing a shorter wavelength than DZ2. The basis for this difference is unknown.
Since rippling pattern formation occurs similarly when M. xanthus cells are in the presence of dead prey as it does with live prey, we felt that it was prudent to examine the role of developmental autolysis in connection with developmental rippling. There is contradictory evidence as to whether or not cell lysis is an essential component of the developmental program, and we considered that part of this discrepancy could be strain dependent (18, 22). We analyzed the autolytic rates of DZ2 and DK1622 cells that had been resuspended in MMC buffer and incubated in 12-well plates at 32°C. Figure Figure44 shows that DK1622 undergoes cell lysis at a faster rate than DZ2 after the onset of starvation, with >90% of the cells dying over the course of this assay. In strain DZ2, lysis is still observed but at a much slower rate. We hypothesize that the differences observed in autolytic rates account for the spontaneous developmental rippling observed in DK1622.
To test if increased cell death could induce developmental rippling in DZ2, we harvested and washed a culture of mid-log-phase DZ2 cells and split the cell suspension into two parts: one of which was left untreated and the other exposed to heat at 95°C for 10 min. We then mixed the untreated DZ2 cells with the heat-killed DZ2 cells in 10% increments ranging from 0 to 100% live cells and added the cell mixtures to CF plates to allow development. Under these conditions, we observed rippling to occur in all samples that contained 20% or more dead cell material (Fig. (Fig.5).5). Interestingly, rippling is only observable in the initial spot, indicating that live cells which leave the area containing the dead cell material are not induced to ripple.
In an accompanying experiment, aliquots of heat-killed DZ2 cells were mixed with water instead of live cells to yield increments of dead cell material as above. These preparations were then pipetted adjacent to 10-μl colonies of live M. xanthus DZ2 cells on CF. Under these circumstances, physical separation of developmental rippling and fruiting body formation can be observed, with M. xanthus cells located in the initial spot aggregating into fruiting bodies, while cells migrating into the region containing the dead cell material are induced to ripple. Rippling was observed in all samples containing the equivalent of 20% or more dead cell material.
When the phenomenon of rippling was first reported by Reichenbach in the 1960s, he noted that it occurred during both predation and development (20). In the early 1980s, Shimkets and Kaiser showed that rippling was inducible by peptidoglycan extracted from a variety of proteobacteria and gram-positive bacteria, including M. xanthus, raising the possibility that developmental rippling occurs in response to the peptidoglycan released from lysing M. xanthus cells (24). In this work, we demonstrate that rippling occurs as a general response to the presence of a variety of prey or macromolecular substrates. Physical separation from the macromolecular substrate or incubation in the presence of the monomeric subunits of these macromolecules does not induce rippling behavior. This result indicates that M. xanthus cells are capable of sensing a variety of macromolecules directly, rather than through release of extracellular digestive enzymes and subsequent sensing of the monomeric components. M. xanthus cells move at a rate which is slower than the diffusion of most water-soluble molecules, so it seems logical that their cell movements would be attuned to substrates that move more slowly then they do. Determining the mechanism for sensing these large molecules may be challenging, but there are other reported instances of bacteria utilizing mechanisms specifically designed for dealing with macromolecules, such as the cell-associated degradation of polysaccharides in Bacteroides spp. (25). Related predators of the Bdellovibrio genera (-proteobacteria) require direct cell contact and entry into the periplasm of their prey for predation to occur (13), so it is possible that M. xanthus predation functions best when in direct contact with suitable prey.
A recent study has shown that colliding rippling waves of M. xanthus cells reflect off each other, causing individual cells to reverse after contact such that any given wave of cells is trapped in an oscillation between neighboring waves. The result is that the entire swarm moves in a coordinated fashion that ensures multiple passages over a given area. The fact that oscillating behavior is induced when M. xanthus cells are located in the area of prey implies that rippling behavior is beneficial for predation. This is supported by the fact that genetic and physiological conditions which inhibit rippling also inhibit efficient predation. We propose a model in which the detection of macromolecules by M. xanthus stimulates the oscillatory behavior of moving cells as a method to ensure that all available growth substrate is consumed.
M. xanthus is a very slow moving bacterium confined to gliding along surfaces, and thus, it has been difficult to demonstrate chemotaxis by M. xanthus cells within a gradient of diffusible chemicals in a manner similar to that displayed by E. coli (5). However, directed movement of M. xanthus cells requires the che1 (frz) chemotaxis-like pathway, and cells carrying mutations in the frz pathway have been shown to exit E. coli prey microcolonies prior to complete lysis of prey (15, 16). This indicates that the Frz chemotaxis homologs are part of the molecular pathway necessary for M. xanthus to properly recognize and modulate cell movement in response to prey. Our observations of M. xanthus predatory behavior indicate that while M. xanthus cells do not appear to be specifically attracted to areas containing prey, cell movements are clearly altered in the presence of prey. It is possible that the ability of a given substance to induce rippling may be the best indication of its suitability as a chemoattractant for this species. Chemotactic movement allows E. coli cells to accumulate in locations supporting optimal growth. For a bacterium that is constrained, as M. xanthus is, to two-dimensional movement and adapted to the digestion of large molecules or prey cells, there is little advantage to a mass accumulation of cells when the growth substrate is nondiffusing. Cells which cannot make direct contact with the growth substrate may not gain any benefit. Thus, the induction of rippling may serve to trap some cells in a nutrient-rich environment while also serving to limit competition. Further examination of the process of rippling will be required to determine how it relates to typical chemotactic mechanisms.
Finally, we also demonstrate that the wild-type M. xanthus strain analyzed is critical for the interpretation of rippling behavior. Although DK1622 and DZ2 are both induced to ripple in the presence of prey, only DK1622 was observed to ripple in pure culture, a response that appears to be dependent on the elevated level of cell lysis detected in this cell line. Even in the presence of prey, rippling features such as wavelength are markedly different between the two strains. The observation of rippling behavior in the presence of nutrients released during predation challenges the idea that rippling behavior is stimulated purely by cell-cell transmission of the starvation-induced C signal. Also, the ability of dead cell material to stimulate rippling in developing colonies of DZ2 indicates that, while it is possible that C signaling is utilized to mediate the response to prey, it seems that both developmental and predatory rippling also require the presence of macromolecular growth substrates. Thus, a deeper understanding of rippling behavior and its requirements depends on a better understanding of the differences between the two M. xanthus wild-type strains and a careful dissection of the responses of this bacterium to the separate stimuli of starvation and predation.
We thank D. Bodenmiller for editorial comments. We also thank members of the Kirby lab for valuable comments and members of the Zusman lab for valuable communication and strains used in this work.
This work was supported by Georgia Tech start up funds and grant AI059682 from the National Institutes of Health to J.R.K.