Some Borrelia species are the causative agents of tick-borne Lyme disease responsible for different disabilities depending on species and hosts. Borrelia are highly motile bacterial cells, and light microscopy shows that these spirochetes can associate with each other during movement. Using cryo-electron tomography, we observed closely associated Borrelia cells. Some of these showed a single outer membrane surrounding two longitudinally arranged cytoplasmic cylinders. We also observed fusion of two cytoplasmic cylinders and differences in the surface layer density of fused spirochetes. These processes could play a role in the interaction of Borrelia species with the host's immune system.
Borrelia species, the causative agents of Lyme disease and relapsing fever, evade the adaptive immune response by antigenic variation of surface lipoproteins as well as temporal and spatial changes in expression of outer-surface proteins (Hefty et al., 2002; McDowell et al., 2002; Norris, 2006). The progression of Lyme disease is different among Borrelia species, with B. burgdorferi predominantly causing arthritis, B. garinii neuroborreliosis and B. afzelii a form of chronic dermatitis (Steere, 2001). Comparison of the genome sequences of three different B. burgdorferi strains suggests that Borrelia undergo genome-wide genetic exchange, which might form the basis of the observed differences in pathogenicity (Qiu et al., 2004). However, no mechanism is known to date that could explain how Borrelia achieve this exchange between individual cells. The Borrelia genome does not contain genes with homology to known pili-protein-encoding genes (Fraser et al., 1997), suggesting that Borrelia species do not normally transfer genes by conjugation. However, B. burgdorferi has a complete set of genes encoding the proteins necessary to carry out homologous recombination, although it lacks insertion sequences, transposons and restriction–modification systems (Fraser et al., 1997). Therefore, the major means of horizontal gene transfer of DNA in Borrelia is the homologous transfer of short DNA fragments (Dykhuizen and Baranton, 2001). The only exception to this seems to be the prophage, cp32, which could potentially transfer genes by generalized transduction, although only under artificial conditions (Eggers and Samuels, 1999; Eggers et al., 2001; Stevenson and Miller, 2003). The only evidence for frequent horizontal gene transfer and recombination has been found for ospC alleles within different Borrelia species (Livey et al., 1995; Dykhuizen and Baranton, 2001).
Horizontal transfer of DNA fragments in Borrelia could also be mediated by DNA-containing vesicles, as described for Escherichia coli (Yaron et al., 2000). Small vesicles have been found to bleb off the outer-membrane surface in Borrelia, but they are unlikely to contain DNA (Kudryashev et al., 2009). Larger vesicles have been observed that contain cytoplasm (Kudryashev et al., 2009) and DNA (Garon et al., 1989). Whether these vesicles play a role in horizontal gene transfer remains to be seen. Initial evidence for the existence of an intergeneric conjugal system in B. burgdorferi was recently demonstrated by conferring the erythromycin-resistant phenotype to Bacillus subtilis or Enterococcus faecalis (Jackson et al., 2007). Borrelia have also been shown to adopt cyst-like forms (Garon et al., 1989; Preac Mursic et al., 1996; Brorson and Brorson, 1997; 1998), but both the mechanism of cyst formation and the biological role ofcysts remain unclear. It is not clear how efficient homologous recombination occurs in Borrelia, and these spirochetes probably use a specialized mechanism for DNA transfer. Here, we present an ultrastructural study of Borrelia that reveals direct fusion of two closely apposed spirochetes and discuss its potential relevance for bacterial conjugation.
Using cryo-electron tomography (Lucic et al., 2005; Leis et al., 2008), we recently found that the overall cellular architecture is similar for the three species B. burgdorferi, B. garinii and B. afzelii (Kudryashev et al., 2009). These bacteria were harvested in log phase and were found to be highly motile when visualized by live staining with Hoechst and fluorescent microscopy imaging (Fig. 1 and Movies S1–S4). However, subtle and notable differences were found in the overall architecture of these Borrelia species and the related spirochetes Treponema primitia (Murphy et al., 2008) and Treponema denticola (Izard et al., 2008). These differences include some potentially important features such as the presence of cytoplasmic filaments and vesicles budding from the outer-membrane sheath as well as two distinct arrangements of periplasmic flagellar filaments (Kudryashev et al., 2009). In contrast to classical EM preparations, cryo-tomography revealed that the flagellar filaments line up next to each other to form a single ribbon rather than a bundle (Charon et al., 2009). When more than seven flagella were present in a bacterium, the flagellar filaments can pile up in what appears like a bundle of multiple small flagellar ribbons (Kudryashev et al., 2009). We and others have also noted the presence of an additional surface layer surrounding the outer membrane of Borrelia that we first called amorphous slime layer, based on previous statements (Hovind-Hougen, 1984; Barbour and Hayes, 1986). Subsequently, the disappearance of this layer upon surface proteolysis using proteinase K demonstrated that it is composed of outer-surface proteins (Liu et al., 2009). In the studies reported here, we observed closely apposing Borrelia cells both in preparations of live cells viewed with the light microscope and in images captured by electron tomography. The tomographic images show examples of bacterial fusion events, which are discussed in the context of their potential biological functions.
Close apposition of motile Borrelia
When observing the different Borrelia species with light microscopy, we noted that some Borrelia were swimming in pairs and chains (Fig. 1, Movies S1–S4), as was previously observed (Goldstein et al., 1994). Two bacteria could be seen apposing by either swimming in the same direction (70%) or by encountering each other while swimming in opposite direction (30%). These ratios were similar for all three Borrelia species: 75%, 70% and 74% for B. afzelii, B. garinii and B. hermsii respectively. Closely apposed pairs were found at a similar rate (1.1% of observed bacteria per time frame) as pairs that were concatenate, and thus likely dividing (2.8%; Table 1). Closely apposing bacteria usually stayed together for less than 10 s, although about 10% stayed closely apposed for more than 1 min. In contrast, bacteria that were seen in the final step of division (Fig. 1 and Movies S3 and S4) could be filmed for several tens of seconds before they either split (Movies S3 and S4) or moved out of the field of observation. In the electron microscope, we also found bacteria closely apposed to each other (Fig. 2). The ultrastructure of such apposed bacteria appeared similar to that of the individual ones. However, we do not know if these bacteria corresponded to pairs of bacteria as observed by light microscopy or simply individual bacteria that came close to each other during centrifugation or blotting.
Table 1. Number and percentage of observed apposing and dividing bacteria.
Total number observed
Number (percentage) apposing
Number (percentage) dividing
The numbers represent ‘specimen time points’, i.e. the number of image frames multiplied by the number of bacteria observed within the individual frames (total number), the number of bacteria observed being apposed to each other (Fig. 1) or undergoing cell division as evidenced by two bacterial bodies being linked at their poles. A total of 244 bacteria were tracked from five movies of B. burgdorferi yielding 29 280 specimen time points. From the other species, 475 bacteria from 10 movies of B. garinii, 422 bacteria from 10 movies of B. afzelii and 531 bacteria from 9 movies of B. hermsii were observed.
Fusion of the outer membranes of paired Borrelia
Occasionally, we also found two closely apposed B. burgdorferi cells that were partly wrapped around each other in a helical fashion (Fig. 3A), similarly to what was observed in the light microscope (Fig. 1A and B). These bacteria showed a single outer-membrane sheath that wrapped around the two cytoplasmic cylinders, indicating that the outer membranes of the individual bacteria had fused (Fig. 3). The flagellar filaments of the wrapped bacteria formed a common bundle (Fig. 3B–D and F). The bundled flagellar filaments in B. burgdorferi appeared at one side of the closely apposed cytoplasmic cylinders (Fig. 3C and D). Figure 3F shows that flagellar filaments emerging from basal bodies of the individual bacteria contribute to the common bundle. Both basal bodies are located within the previously reported range of distance from the pole of the cytoplasmic cylinder. Fusion of the outer-membrane sheath was also observed with B. garinii (Fig. 4). In B. garinii cells, which generally have more flagellar filaments, the majority of flagellar filaments could be seen at one side of the cytoplasmic cylinders, and only a few filaments (Fig. 4B and C) were located on the opposite side. Curiously, the two bacteria sharing a common outer-membrane sheath shown in Fig. 4 exhibited 29 periplasmic flagellar filaments, which represents an unusually high number compared with most single B. garinii cells (Kudryashev et al., 2009). In contrast to what was seen with individual spirochetes, the flagellar filaments in the fused cell were clearly arranged in a bundle rather than in a stack of sheets (Figs 3 and 4).
Fusion of the outer-membrane sheath of longitudinally aligned bacteria could be observed at any part of the cell bodies. The outer-membrane sheath of the two B. burgdorferi in Fig. 3A fuse only at the overlapping tips of the two bacteria, whereas for the B. garinii the outer-membrane sheath fused for several micrometers along most of the length of their cell bodies (Fig. 4). Outer-membrane fusion was also observed in B. hermsii (data not shown), a spirochete causing relapsing fever. Out of approximately 110 spirochetes located close to each other we found seven with shared outer membranes (Table 2).
Table 2. Number of observed cell fusion and division events.
Total number of observed cells
Total number of closely apposed cells
Fusion of the OMS
Fusion of the cytoplasmic cylinders
For comparison with Table 1, note that 2.6% of the total number (250) of observed bacteria were found undergoing outer-membrane sheath (OMS) fusion and 2.4% were undergoing cell division.
Fusion of the cytoplasmic cylinders
We further imaged bacteria with fused outer-membrane sheaths and found one striking example that showed the cytoplasm of the two bacteria fused over a distance of 3.15 µm. The outer-membrane sheath was fused over a much larger area (Fig. 5A and B). Analysis of a tomogram recorded from the region between the cytoplasmic cylinders of these fused cells showed a periplasmic flagellar filament located between two layers of peptidoglycan (Fig. 5C–E). Closer to the region of cytoplasmic fusion, the flagellar filament moved out of the fusion plane. Even closer to the zone of fusion, the two peptidoglycan layers could be seen to form either one or two closely apposed layers separating the two cytoplasmic cylinders (Fig. 5D and E).
In some bacteria with a fused outer membrane the surface layer appeared diminished (Figs 4A and C and 5B) as compared with cells without fused outer membranes (Fig. 3B). We thus compared the intensity of the surface layer and outer-membrane sheath from the electron-density profiles across the tomograms (Fig. 6). This analysis showed that Borrelia cells that fused only part of their outer-membrane sheaths exhibited a surface layer indistinguishable from individual bacteria, whereas those that fused over large surface areas showed a diminished, less dense surface layer (Fig. 6C). There was no difference in the ratio of the electron density of surface layers versus outer-membrane sheaths between bacteria that fused large proportions of their outer-membrane sheaths and the one bacterial pair that also fused their cytoplasmic cylinders (Fig. 6C). The bacteria that fused only part of the outer-membrane sheath were found to have a double periplasmic layer at one side of one of the cells, a single observation from our complete data set (Fig. 6B, II).
In this study, we have shown that the human pathogenic spirochetes causing Lyme disease can harbour two cytoplasmic cylinders covered by a single outer membrane and that, in addition, fusion of cytoplasmic cylinders can also occur. The fact that we detected fused outer membranes in B. burgdorferi, B. hermsii and B. garinii suggests that this may constitute a common feature of Borrelia. The observation of two cytoplasmic cylinders enclosed in one outer membrane is somewhat reminiscent of the situation seen during cell division, during which two cytoplasmic cylinders can be accommodated under a shared outer membrane prior to complete separation of the daughter cells (Kudryashev et al., 2009). The question thus arises whether the fused bacteria reported here originate from two separate bacteria undergoing close apposition (Fig. 1) or from bacteria that underwent cell division but in which one of the daughter cells then changed its direction of movement to push the not-yet-separated cytoplasmic cylinders towards each other. Similarly, it is not clear if fusion of the outer membrane (and possibly the cytoplasmic membrane) can occur during brief apposition. Considering the fluidity of the outer membrane (Kudryashev et al., 2009), we think that this type of fusion is possible, although this is in no way conclusively proven by our current work.
Are the observed events, which we interpret as fusion, biologically relevant? Electron microscopy has a long history of revealing important findings, but it also has a record of preparation artefacts. In these experiments, artefacts could include fusion during the centrifugation step or during blotting of excess liquid before plunge freezing. The forces generated during centrifugation could potentially cause neighbouring bacteria to fuse. However, the same step was applied before imaging with the light microscope, after which between 50% and 90% of bacteria were motile. Those motile cells that were closely apposed to each other remained in contact for only several seconds to several dozens of seconds. During blotting, shear forces occur in the direction perpendicular to the cell–cell contact suggesting that this treatment should not push cells together (Fig. 2). We recently showed that blotting can cause membrane rupture in mammalian cells (Lepper et al., 2010). However, the distance between the cytoplasmic membrane and the outer membrane in cells with fused outer membranes was the same as in single cells, suggesting that there were no re-arrangements in the overall membrane architecture. Furthermore, unlike chemical fixation, plunge freezing happens within milliseconds, thereby limiting the time available for rearrangement of proteins and other macromolecules. Finally, no examples were found of more than two cells sharing a common outer membrane. These considerations suggest that the fusion of the outer membranes is probably not an artefact of preparation. In any case, the fusion of the cytoplasmic cylinders during a preparation that does not disrupt any other aspect of the bacterial architecture seems highly unlikely.
Two fused cells, if derived from different clones, could exchange different immunoreactive antigens on their shared surface. We thus speculate that this fusion event could help these bacteria, which are known to vary their exposed antigens (Singh and Girschick, 2004), to evade the immune response of the host. This could be tested by generating strains expressing different fluorescent fusions of outer-membrane proteins and investigating mixtures of these bacteria. Considering the similarity between some Treponema and Borrelia, it would be interesting to see whether these two pathogens can fuse with one another.
The bacteria with fused cytoplasmic cylinders showed a diminished surface layer compared with non-fused bacteria, although they otherwise appeared normal. Bacteria that fused their outer-membrane sheaths over a large area also had a diminished surface layer. This decrease was only observed in a few bacteria undergoing extensive outer-membrane fusion. The bacteria next to the pairs with diminished surface layers showed normal layers, as did Borrelia that were treated with high concentrations of antibiotics or antibodies (data not shown and Kudryashev et al., 2009). This leads us to wonder if the fusing Borrelia shed part of their surface antigens, likely OspA, B, C and D (Liu et al., 2009), in preparation to express new ones. A major limitation of this analysis is that fusion of the cytoplasmic cylinders occurred over a small fraction of the length of the fused outer-membrane sheath, suggesting that such events are hard to capture even when the bacteria are carefully screened to identify those lying in close proximity.
It also remains unclear how fused bacteria can separate again (Fig. 7). Separation along the long axis is unlikely, as a large area of membrane would need to be pulled apart. The images presented in Fig. 1 show that apposing bacteria can change their direction of movement after apposition. Thus, separation could occur at the poles, as it does during cell division (Kudryashev et al., 2009). During cell division, long tubes of outer membranes can be seen between two dividing cells (Kudryashev et al., 2009). Perhaps the forces generated by the motility of the separating bacteria pull the outer membrane into ever-thinner tubes until the membranes pinch off and the cells separate. In this scenario, no dedicated scission protein(s) would be necessary.
Although our data are not conclusive, the observations presented here could reveal a two-step process that could lead to the exchange of genetic material between individual spirochetes. Fusion of the outer membrane could be followed by the fusion of small portions of the cytoplasmic membrane to allow genetic exchange. From over a hundred closely apposed pairs of bacteria examined, seven showed fused outer membranes, and only one pair showed a fusion of the cytoplasmic cylinders. Notably, the number of the observed outer-membrane fusion events is similar to the number of the observed cell-division events. Cells usually divide every 12 h (in our data set six bacteria were dividing, Table 2). This means that, within a population, the number of cells undergoing cell division might be about the same as the number of cells that are fused.
Closely apposed bacteria observed by light microscopy typically stay together for a shorter period of time (less than 10 s) than bacteria in the final step of division (Fig. 1 and data not shown). However, around 10% of closely apposed bacteria stay together for over 1 min, potentially time enough for cytoplasmic fusion to occur. It will now be important to develop methodologies to detect genetic exchange, for example, by direct observation of gene products encoded by the exchanged DNA (Babic et al., 2008) or by fluorescent in situ hybridization, as has recently been employed for the study of genetic exchange between two eukaryotic nuclei in a single Giardia cell (Poxleitner et al., 2008). A great advance would be the ability to image motile spirochetes in the light microscope and then image the same cells in the electron microscope. Such approaches have been successful for adherent neuronal cells (Sartori et al., 2007). However, we recently showed that the highly motile forms of a rodent malaria parasite, Plasmodium berghei sporozoites, could not be successfully correlated between light and cryogenic electron microscopy (Lepper et al., 2010). Transient associations of swimming spirochetes are therefore unlikely to be imaged by correlative microscopy without further technical improvements. One could, however, envision a set-up where some spirochetes are linked to the EM grids and others are freely swimming. If these bacteria then associate with each other for long enough, such a correlative approach could be successful. Moreover, it will be important to probe if these events can happen in vivo, either within the tick or within the host, where bacteria might not be as concentrated as in vitro. The Leptospira, related spirochetes, have been shown to form biofilms with masses of densely aggregating bacteria (Ristow et al., 2008), and it would be interesting to investigate these organisms for potential fusion events.
In conclusion, we present evidence that Lyme disease spirochete pairs can harbour two cytoplasmic cylinders, and thus house two genomes under a single outer membrane. We speculate that the fusion of the enclosed cytoplasmic cylinders could lead to the exchange of genetic material between individual cells. If so, this process could have important implications for the spirochete's life cycle and the pathogenesis of Lyme disease as well as contributing to the understanding of bacterial gene transfer mechanisms in general. The possibility of genetic exchange could be experimentally tested if two strains of the same species with enough genetic markers to test the progeny are incubated together. However, it will remain a challenge to show that such genetic exchange occurs via the cell-fusion events reported here.
Borrelia culture and treatment
Borrelia burgdorferi (strain ZS7), B. afzelii (strain MMS) and B. garinii (strain ZQ1) were cultured essentially as described previously (Wallich et al., 2005). Briefly, the Lyme disease and relapsing fever spirochetes B. hermsii (strain HS1) were cultivated in Barbour–Stoenner–Kelly (BSK)-H medium (PAN Biotech) supplemented with 5% rabbit serum (Cell Concept) at 30°C. Cells were harvested in the log phase during rapid growth. Cells were enriched by a centrifugation step at 8000 r.p.m. for 5 min followed by resuspension of the pellet in 20 µl of BSK (for cryo-EM preparation) or in a slightly larger volume for light microscopy. Living cells were viewed after staining the bacteria with Hoechst (1 pg ml−1) (Molecular Probes 34580) for 10 min. Imaging of living cells was performed on an inverted Axiovert 200M Zeiss microscope in an air-conditioned imaging suite at room temperature (24°C). Images were collected with a Zeiss Axiocam HRm at 2 Hz using the Axiovision 4.7 software and a 40× (NA 0.6) objective lens. Live microscopy revealed that well over 50% were actively moving, while most other bacteria were attached to the glass but also appeared alive as judged by their flexing behaviour (see Movies S1–S4). In some samples, kanamycin was added at final concentration of 100 µg ml−1 to the bacterial cultures and left for 24 h.
Images were exported and processed in ImageJ and arranged using the Adobe Creative Suite package. Note that less dense samples were imaged for light microscopy than for tomography, although in both cases bacteria underwent the same centrifugation steps. Results presented in Table 1 are from randomly selected movies. Although no examples of apposed cells were seen in the movie chosen for B. burgdorferi, in other movies such apposition events were seen with that species.
Cryogenic electron tomography was performed essentially as described before (Cyrklaff et al., 2007; Kudryashev et al., 2009). Borrelia in serum- and gelatin-free BSK medium or PBS were essentially treated the same way (cultivation, harvesting) as for light microscopy but were transferred onto EM carbon grids and incubated for 5–10 min. After removal of excess liquid, grids were rapidly plunged into liquid ethane and stored in liquid nitrogen. Grids were mounted in a Gatan cryo-holder (model: 626) and imaged in a CM 300 electron microscope (FEI) (accelerating voltage: 300 keV, TWIN objective lens) equipped with field emission gun and Gatan post column energy filter. Tilt series were made in two-degree increments. Sixty to 65 low-dose images, corresponding to a cumulative dose of under 20 000 electrons nm−2 filtered at zero energy loss, were recorded on a 2048 × 2048 pixel Gatan CCD camera, at a magnification of 43 000 (0.82 × 0.82 nm2 per pixel), and an objective lens defocus of between −3 and −12 µm. For this and our previous study (Kudryashev et al., 2009), a total of 46 tomographic reconstructions containing parts of 80 individual Borrelia were calculated by r-weighted back-projection using the ‘EM-image processing package’ (Hegerl, 1996). For visual presentation, tomograms were filtered using non-linear anisotropic diffusion (Frangakis and Hegerl, 2001). Visualization, volume rendering and segmentation were performed using the Amira package (TGS Europe S.A., France). Quantitative analysis was performed using TOM toolbox for Matlab (Nickell et al., 2005). Measurements of the electron density of the surface layer were compared with the density profile of the outer membrane above the background level outside the bacteria, as shown in Fig. 6E. To test the statistical significance of the densities of the outer surface layer for different spirochetes, we used a two-tailed T-test.
We thank Christiane Brenner for help with Borrelia culture, Andrew Leis, Simone Lepper and Markus Simon for discussions and comments on the manuscript, and Jürgen Plitzko and Günter Pfeiffer for help with electron microscopy. We also thank the reviewers for improving the manuscript text. The work was funded by grants from the German Federal Ministry for Education and Research (BMBF, BioFuture) to F.F., the German Research Foundation (DFG) to F.F. and M.C. (SPP 1128) and to RW (WA 533/7-1) and a postdoctoral fellowship of the Cluster of Excellence CellNetworks to L.L. We thank the Max Planck Society, the Institute for Computational Modeling of the Siberian Branch of the Russian Academy of Sciences, the Medical School at the University of Heidelberg, and the Chica and Heinz Schaller Foundation for support.