Counterclockwise Rotation of a Gliding Cell. In our previous study, we proposed a helical loop track model for gliding motility of F. johnsoniae where SprB adhering to a substratum was propelled along a helical loop track on the cell surface, resulting in rotation and translocation of the cell 19. Here we investigated whether gliding cells actually rotate as proposed. To visualize cell rotation, we attempted to find a cell surface protein that did not change position with respect to the long axis during gliding, and observe its behavior by total internal reflection fluorescence (TIRF) microscopy using fluorescently-labeled antiserum. Cell surface proteins were isolated and collected by centrifugation. The isolated cell surface proteins were separated by SDS-PAGE (Fig. S1A). SprB with molecular mass greater than 250 kDa was detected by immunoblot analysis using anti-SprB antiserum (Fig. S1B). Major protein bands on the gel were identified by peptide mass fingerprinting analysis using MALDI-TOF-MS, which revealed that the cell surface protein fraction contained proteins with putative outer membrane protein (OMP) domains (Fjoh_0697, Fjoh_1311 and Fjoh_3514) and putative TonB-dependent receptors (Fjoh_0403, Fjoh_0736, Fjoh_4221 and Fjoh_4559) (Fig. S1A and Table S1). Amino acid sequence analyses for predicting signal peptides by SignalP 4.0 software25 and for predicting protein localization by CELLO v2.5 subcellular localization predictor26,27 suggested that most of the proteins were located at the outer membrane (Table S1). We generated rabbit antiserum against the cell surface protein fraction. Immunofluorescence microscopy using this antiserum and Alexa Fluor-555-conjugated secondary antibodies showed that signals were dispersed on the cell surface (Fig. S1C). TIRF microscopy with a translocating cell revealed that there were two types of signal movement (Movie S1). Type I signals, the most common, were located at a position a fixed distance away from a cell pole and periodically appeared on the surface with respect to the short axis when the cell moved on a substratum, indicating cell rotation. TIRF analysis showed that the type I signals always appeared from one side with respect to the short axis of the moving cell, demonstrating rotation in a counterclockwise (CCW) direction when viewed from the rear to the advancing direction of the cell (Movie S2). Type II signals, which comprised less than 20 % of otal signals, moved on the cell surface with respect to both axes like the SprB signal (Movie S3). This type of signal behavior was expected since SprB, which migrates from one pole to the other along an apparent helical track 19, was present in the surface protein fraction that was used to generate the polyclonal antiserum. We also generated antiserum against purified OMP domain-containing protein Fjoh_0697, which appeared to be a major cell-surface protein (Fig. S1A). TIRF analysis using anti-Fjoh_0697 antiserum revealed that the Fjoh_0697 signal did not migrate from pole to pole but was instead located at a position a fixed distance from a cell pole. In a translocating cell Fjoh_0697 periodically appeared on the surface with respect to the short axis, demonstrating CCW rotation of the gliding cell as observed from the rear of the cell (Fig. 1A, Movie S4).
SprB is the primary motility adhesin on the F. johnsoniae cell surface. Left-handed helical flow of immunolabeled SprB on a gliding cell was previously observed 19. In this study, for detailed analysis of the individual SprB signal movement on a gliding cell surface, limited numbers of SprB molecules on a cell were labeled with highly diluted antiserum (see Materials and Methods) to reduce signals and track SprB signals easily, and movement of signals on the surface of gliding cells was observed by fluorescence microscopy. Under these conditions, about 58% of cells had 1 to 10 SprB signals and the rest had none. Consistent with the previous report 19, the SprB signals were propelled between cell poles along an apparently helical loop track on gliding F. johnsoniae cells (Fig. 1B, Movie S5). Cells migrated 6.1 ± 1.1 µm (mean ± standard deviation, N = 63) when the Fjoh_0697 signal made one revolution in the direction of the short axis of a cell (helical pitch of Fjoh_0697) (Fig. 1C left and D), while the moving distance of SprB on the substratum during one revolution of SprB in the direction of the short axis of a cell (apparent helical pitch of SprB) (Fig. 1C right and D) was 4.55 ± 1.1 µm (N = 22). The rotational rate of Fjoh_0697 signals was directly proportional to the cell velocity (Fig. 1E). To show the relationship between movements of a cell and a labeled protein, velocities of labeled proteins and cells were determined. SprB (50 signals) and Fjoh_0697 (25 signals) from 12 cells were plotted on a graph (Fig. 1F). Plots showed that the velocity of Fjoh_0697 signals was positively correlated with the cell velocity. These data demonstrated that Fjoh_0697 signals are located at fixed positions on the cells. In contrast, SprB signals moved in both directions on the cells and the velocity of each SprB was not necessarily related to the cell velocity (Fig. 1F). In SprB as well, it should be more proportional, assuming that all SprB molecules move on a helical loop track of the cell surface at the same pace and that the movement of SprB causes the cell to move. The difference may be accounted for by the that the velocity of SprB varies from molecule to molecule, and SprB that advances the cell is only a part of SprB molecules. SprB, whose value on the horizontal axis is 0 in Fig. 1F, appears to actually move the cell. It is suggested from Fig. 1F that the other SprB molecules exist away from the substratum and move at varied velocities to some extent. The movement of SprB on a cell was then examined in more detail.
Multiple Lanes in Helical Loop Track for SprB Movement. Some SprB signals exhibited unexpected movements such as mid-cell U-turns. Figure 2A and Movie S6 show an SprB signal that moved toward the anterior (right) pole, looped around the anterior pole at approximately 1.8 sec and then moved toward the posterior pole. Interestingly, before the signal reached the posterior pole, it made a U-turn and moved toward the anterior pole (3.5–4.4 sec). The SprB signal then looped around the anterior pole and migrated all of the way to the posterior pole without conducting another mid-cell U-turn (5.2–9.5 sec). The frequency of mid-cell U-turns was about 0.07 per min per SprB, suggesting that the U-turn events do not happen frequently. Multiple SprB signals on a cell did not always move at the same speeds. In Fig. 2B and Movie S7, a cell that had two SprB signals traveling in the same direction is shown. The low intensity signal moved more slowly (about 1.2 µm/sec) than did the high intensity signal (about 2.4 µm/sec). The high intensity signal thus overtook and passed the low intensity signal while traveling to the anterior (right) pole. These results suggest that the track on which SprB traveled had multiple lanes.
Alteration of SprB Velocity during Translocation. Since we observed that some SprB signals moved with different speeds on the cell surface, we asked if SprB can change velocity during pole-to-pole movement. Since it was difficult to track individual SprB signals on cells of normal size for long periods of time, we observed SprB movement on filamentous cells generated by inhibiting cell septation with the antibiotic cephalexin. Addition of 20 µg/ml cephalexin to growth media resulted in filamentous cells that were about 100 µm long. The localization of SprB on glutaraldehyde-fixed cells as observed by immunofluorescence microscopy showed similar distributions on filamentous (cephalexin-treated) cells and on non-elongated control cells that had not been exposed to the antibiotic (Fig. S2). In both cases cells had about 1.5 SprB signals/µm on the cell surface under the labeling condition noted in the legend to Fig. S2. The average velocity of SprB on a filamentous cell was 2.3 ± 0.6 µm/sec (N = 96) which was similar to SprB movement on a normal-length cell (2.1 ± 0.6 µm/sec, N = 128) (Fig. 2G). We then determined whether the velocity of SprB along the cell surface varied. SprB movements on filamentous cells that were not themselves translocating on a glass surface are depicted by the kymographs in Fig. 2. On these nontranslocating cells the SprB molecules were apparently not attached to the substratum, since otherwise the action of the motility motors against these would have resulted in cell movement. The movement of four SprB signals over 9.4 sec on a filamentous cell is shown in Fig. 2C (the kymograph from Movie S8). Distance between signals marked with “b” and “c” did not change during the observation, resulting in parallel lines in the kymograph indicating that these signals moved with similar velocity. In contrast, signal “a” approached and overtook signal “b” within the first 5 sec of the recording. Similar to the observation of overtaking movement on a normal cell (Fig. 2B), this result suggests that SprB signals move with different velocities on a track with multiple lanes. Two other examples of SprB movement on individual filamentous cells are shown in Fig. 2D and Fig. 2E, which correspond to Movies S9 and S10 respectively. Parallel oblique lines appeared in these kymographs; however, some lines were not parallel, although SprB signals were moving in the same direction. SprB signals “d” and “e” moved toward the right with similar velocities of approximately 3.0 µm/sec, resulting in parallel oblique lines for about the first 8 sec of observation (Fig. 2D and Movie S9). After this point, signal “d” moved more slowly and eventually stopped, as indicated by the vertical line in the kymograph after 11 sec. As a result, the distance between signals “d” and “e” increased after 8 sec. In addition to this “slowdown and stop" behavior, we also observed ‘stay and go’ movement of SprB signals (Fig. 2E, Movie S10). Signal “f” did not move until about 4 sec after start of observation, as indicated by the vertical line. Subsequently, signal “f” began to move toward the right and drew an oblique line in the kymograph. In addition, the time course of SprB movement along the long axis of a cell by tracking analysis showed the alteration of SprB velocity during translocation (Fig. 2F). These results indicate that the velocity of SprB signals can change during translocation.
Visualization of Multi-Rail Structure for Gliding Machinery. Observation of SprB movement on the cell surface led us to hypothesize the presence of a multi-rail structure for the SprB “trains”. We visualized such a structure by electron microscopy. Cells were burst by osmotic shock to reduce cell thickness and were then negatively stained with 0.5% uranyl acetate. Multi-rail structures which formed bundles of 2–12 fibers were observed (Fig. 3A-D). The thickness of a fiber in a bundle was 7.5 ± 0.9 nm (N = 58). The multi-rail structure was easily peeled from the cell after osmotic shock. Thin filaments appeared to be attached to the multi-rail structure (Fig. 3D). Mutants deficient in gliding-related proteins (Gld proteins, Spr proteins and RemA protein) were examined for the presence of the multi-rail structure. The multi-rail structures were observed in each of the spr and remA mutants at more than 30% of osmotically shocked cells, but the structures were not observed in any of the gld mutants when at least 50 osmotically shocked cells of each gld mutant were examined (Fig. S3 and Table S2). These results suggest that the Gld proteins are required for formation of the multi-rail structure. F. johnsoniae gld mutants are completely deficient in gliding 28 whereas spr mutants retain limited ability to glide on some surfaces 29,30. The absence of the rails in the gld mutants suggested that these structures played an important role in gliding.
Association of the Multi-Rail Structure with SprB Filaments. For more detailed structural analysis of the multi-rail structure, the supernatant of an osmotically shocked sample was centrifuged and the resulting precipitate was suspended and analyzed by TEM. Consistent with the observation of multi-rail structures on osmotically shocked cell surfaces, thin filaments were attached to the multi-rail structures released from wild type cells, whereas there were no filaments attached to the multi-rail structure from the sprB mutant (Fig. 3E and F). Immuno-electron microscopic analysis with anti-SprB antiserum revealed that the thin filaments were SprB (Fig. 3G). These results suggest that the multi-rail structure is associated or interacted with the SprB filaments and forms a part of the gliding machinery.
The Multi-Rail Structure Underneath the Outer Membrane. The multi-rail structure was visualized by quick-freeze deep-etch electron microscopy, which provides images of membrane-attached or -embedded structures on a fractured surface. Figure 3H-J shows images of quick-freeze deep-etch replica with the wild type and gldNO mutant treated with osmotic shock. In the wild-type cells, multi-rail structures were observed on the periplasmic surface of the outer membrane and filaments (presumably SprB) extended from the outer membrane (Fig. 3H and I), as previously observed 31. In contrast, neither multi-rail structures nor SprB filaments were observed in cells of the gldNO mutant (Fig. 3J). To determine the presence of the multi-rail structure in an intact cell, plunge-frozen F. johnsoniae cells were analyzed by cryo-electron tomography. Similar to the previous observation by Liu et al. 32, section images from two representatives of 3D reconstruction of the wild-type cells showed SprB filaments extending from the outer membrane (Fig. 4A and B, Movie S11 and S12). In addition, long thin electron-dense structures were observed, apparently in the periplasm. Importantly, these structures seemed to form parallel lines. Analysis of stacks of section images confirmed the presence of apparently left-handed parallel lines, which may correspond to the multi-rail structure (Fig. 4C and D). Neither cell-surface SprB filaments nor the electron-dense rail structures were observed in cells of the gldJ, gldK, gldL, gldM or gldNO mutants (Movie S13-17). These mutants are completely nonmotile and are also completely deficient in secretion of SprB protein 15,16,33. Taken together, the electron microscopy results suggest that the multi-rail structure was located on the periplasmic surface of the outer membrane.
Components of the Multi-Rail Structure. Since the mutant study revealed that Gld proteins are required for formation of the multi-rail structure, we examined whether Gld proteins were components of the multi-rail structure by immunoelectron microscopy of osmotically shocked cells using antibodies against Gld lipoproteins (GldB, GldD, GldH and GldJ). Using anti-GldJ antibody, gold particles were accumulated on the multi-rail structure (Fig. 5A), suggesting that GldJ is a component of this structure. The other antibodies did not react to the multi-rail structure (data not shown).
Cells of the gldA, gldB, gldD, gldF, gldG, gldH, gldI, gldJ, gldK, gldL, gldM, and gldNO mutants showed no multi-rail structures (Fig. S3). GldJ was reported to be unstable in gldA, gldB, gldD, gldF, gldG, gldH, and gldI mutants 33. In contrast, GldJ was apparently stable in cells with mutations in gldK, gldL, gldM, and gldN 28. We reexamined the amount of GldJ in gldK, gldL, gldM, and gldNO mutants and obtained similar results, except for a partial reduction of GldJ levels in a gldK mutant (Fig. 5B). We also demonstrated that sprB mutant cells retained wild type levels of GldJ protein (Fig. 5B). Blue native PAGE analyses suggested that GldJ was part of a large complex in wild type and sprB mutant cells, whereas it appeared to be part of a smaller complex in cells of gldK, gldL, gldM and gldNO mutants (Fig. 5C). These results suggest that GldK, GldL, GldM and GldNO contribute to multimerization and/or complex formation of GldJ, and may explain the absence of multi-rail structures in cells lacking these proteins.
Helical Organization of GldJ in the Cell Envelope. The trajectory of SprB movement suggested the presence of a left-handed closed helical loop track spanning the cell from pole to pole. To elucidate the organization of GldJ in the cell envelope, we examined the localization of GldJ in fixed permeabilized cells by immunofluorescent TIRF microscopy. Images of wild-type cells showed multiple parallel oblique lines (stripe pattern) spanning the cell from pole to pole. (Fig. 6A). The average distance between the fluorescent lines in the direction of the long axis was 0.70 ± 0.14 µm (N = 126) (Fig. S4A). In contrast, the stripe pattern was not observed in the gldJ mutant cells (Fig. 6B). Due to TIRF microscopy that illuminates molecules near the contact regions with the glass, the stripe pattern, which tilts from the upper left to the lower right direction, represents the surface of the left-handed helical structure. Considering the trajectory of one molecule of SprB 23, the number of stripes observed per cell was not consistent with a single closed helical loop but suggested approximately four parallel closed helical loops from pole to pole for a typical 6 µm-long cell (Fig. 7A). The GldJ’s stripes observed here, which may form the four parallel closed helical loops, seem to be consistent with the trajectories of multiple SprB molecules in a nontranslocating cell in our previous study 19. F. johnsoniae wild-type and gldJ mutant cells expressing a GldJ-mCherry fusion protein were subjected to TIRF microscopy. The fusion protein appeared to form stripes, similar to the results using the GldJ immunofluorescent labeling (Fig. S4).
Presence of a Multi-Rail Structure in the Marine Gliding Bacterium S. grandis. Aizawa 34,35 reported that bundle fibers are found in osmotically shocked cells of the marine gliding Bacteroidetes S. grandis. S. grandis glides on solid surfaces at 5 µm/sec 35,36. S. grandis, a member of the class Sphingobacteriia 37, is not closely related to F. johnsoniae, a member of the class Flavobacteriia. Orthologs of F. johnsoniae Gld and Spr proteins were identified by BLAST search with the S. grandis genome 37 (Table S3). S. grandis had orthologs of gliding motility proteins (GldA, GldB, GldD, GldF, GldG, GldH, GldJ, GldK, GldL, GldM, GldN, SprA, SprB, SprC, SprD, SprE, and SprT). However, orthologs of the periplasmic lipoprotein, GldI and the cell-surface lectin, RemA 38 (which is not essential for gliding) were not found in S. grandis. TEM analysis revealed the presence of abundant bundle fibers in osmotically shocked cells as previously reported 34,35 (Figs. 3K and L). Negative staining with PTA revealed thin filaments, which may be SprB ortholog filaments, attached to the bundle fibers (Fig. 3L). The thickness of each fiber of the bundle fibers was 7.5 ± 0.8 nm (N = 73). Cryo-electron tomography also revealed the bundle fibers in intact S. grandis cells (Movie S18). Structural similarity of the multi-rail structures (bundle fibers) between F. johnsoniae and S. grandis suggest that the multi-rail structure may be a common component of the gliding machinery in diverse members of the phylum Bacteroidetes.