Cell division in a minimal bacterium in the absence of ftsZ


  • Maria Lluch-Senar,

    1. Institut de Biotecnologia i Biomedicina and
    2. Departament de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain.
    Search for more papers by this author
  • Enrique Querol,

    1. Institut de Biotecnologia i Biomedicina and
    2. Departament de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain.
    Search for more papers by this author
  • Jaume Piñol

    Corresponding author
    1. Institut de Biotecnologia i Biomedicina and
    2. Departament de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain.
      E-mail jaume.pinyol@uab.cat; Tel. (+34) 93 5811278; Fax (+34) 93 5812011.
    Search for more papers by this author

E-mail jaume.pinyol@uab.cat; Tel. (+34) 93 5811278; Fax (+34) 93 5812011.


Mycoplasma genomes exhibit an impressively low amount of genes involved in cell division and some species even lack the ftsZ gene, which is found widespread in the microbial world and is considered essential for cell division by binary fission. We constructed a Mycoplasma genitalium ftsZ null mutant by gene replacement to investigate the role of this gene and the presence of alternative cell division mechanisms in this minimal bacterium. Our results demonstrate that ftsZ is non-essential for cell growth and reveal that, in the absence of the FtsZ protein, M. genitalium can manage feasible cell divisions and cytokinesis using the force generated by its motile machinery. This is an alternative mechanism, completely independent of the FtsZ protein, to perform cell division by binary fission in a microorganism. We also propose that the mycoplasma cytoskeleton, a complex network of proteins involved in many aspects of the biology of these microorganisms, may have taken over the function of many genes involved in cell division, allowing their loss in the regressive evolution of the streamlined mycoplasma genomes.


Mycoplasmas are characterized by having small genomes with a low G+C content and by the absence of a cell wall, which confers on them a pleomorphic appearance. These microorganisms, belonging to the Mollicutes class, are widespread in nature and they are parasites or pathogens of a very broad range of hosts, including humans. Although they have a minute and deceptively simple appearance, a closer study of mycoplasma cells reveals considerable intricacy. Despite their reduced genomes, many mycoplasmas have advanced systems required for a parasitic life such as adhesion and internalization inside host cells, antigenic variation, extensive membrane transport systems, or the ability to locomote over solid surfaces (gliding motility).

Mycoplasma genitalium is considered the aetiological agent of non-gonococcal and non-chlamydial urethritis (Jensen, 2004). The most relevant structure in the M. genitalium cytoplasm is a cytoskeleton organized around a rod shaped electrondense core (Hatchel and Balish, 2008; Pich et al., 2008) that defines a distinctive tip structure known as terminal organelle. The majority of cytoskeletal proteins present in the terminal organelle have been primarily identified as cytadhesins or cytadherence-associated proteins (Burgos et al., 2006; 2007; Pich et al., 2008). However, a variety of proteins involved in energy metabolism (dihydrofolate reductase, DHFR), translation/transcription (Elongation factor Tu), heat shock (DnaK) and cell division (FtsZ) were also identified as cytoskeletal proteins in the closely related species Mycoplasma pneumoniae (Regula et al., 2001). The terminal organelle is essential for cytadherence and further parasitism of host target cells (Mernaugh et al., 1993; Burgos et al., 2006) and has a key role in the guidance of the mycoplasma movement (Burgos et al., 2008). Compelling evidence also support that the terminal organelle is (or contains) the molecular motor for gliding motility in M. pneumoniae (Hasselbring and Krause, 2007).

Untreated M. genitalium infections can persist for a long time (Hjorth et al., 2006; Cohen et al., 2007). Among the mechanisms allowing for such persistence, there are highly sophisticated systems to evade the host immune response. The most obvious one is probably the intracellular residence of this microorganism, but mechanisms to generate antigenic variation can also be found. It is well known that variability occurs in the P140 and P110 main adhesins, which are essential for the terminal organelle development (Burgos et al., 2006). These adhesins are encoded, respectively, by ORFs mg191 and mg192 (also referred to as mgpB and mgpC), located inside the MgPa operon. Several non-coding sequences reminiscent of mg191 and mg192 also exist, scattered along the M. genitalium genome, and they are known as MgPa repeats or MgPa islands (Fraser et al., 1995; Peterson et al., 1995). Prominent in a densely packed genome of only 580 kb, these repetitive sequences provide a nearly unlimited source of antigenic variation by recombining among themselves and with the coding sequences in the MgPa operon (Iverson-Cabral et al., 2007; Ma et al., 2007). MgPa islands also endow a phase variation mechanism which can switch ON–OFF, in a reversible or irreversible way, the expression of P110 and P140 adhesins (Burgos et al., 2006). These phase variants are also generated by recombination and they lose the terminal organelle, in consequence being unable to adhere to host cells or other surfaces.

It is currently thought that mycoplasmas evolved from bacteria of the Firmicutes taxon by regressive evolution (Weisburg et al., 1989). A parasitic way of life in very constant environments likely led ancient mycoplasmas to lose many of their genes, in this way modelling the minimal genomes that these microorganisms exhibit. One of the clearest examples of such reductive evolution can be seen when examining the operons or clusters containing genes involved in cell division. While it is very common to find 15 or 16 genes in the division and cell wall synthesis cluster (DCW) of cell-walled bacteria, most sequenced mycoplasma genomes include only mraZ, mraW, ftsZ and one gene encoding a hypothetical protein. It is also noteworthy that some mycoplasma species may even lack several of these genes. Despite the fact that a few other genes may also be collaborating in cell division (Alarcón et al., 2007), genes in the DCW of mycoplasmas are expected to be essential for cell growth. This view was reinforced when no disruptions were identified in the corresponding operons of M. genitalium, M. pneumoniae and M. pulmonis by global transposon mutagenesis (Hutchison et al., 1999; Glass et al., 2006; French et al., 2008). However, in a previous work we found a transposon insertion inside the coding region of M. genitalium ftsZ (Lluch-Senar et al., 2007). This gene was shown to be transcribed in M. genitalium (Benders et al., 2005) and is considered essential for cell division by binary fission. The encoded tubulin-like FtsZ protein is the first component of the cell division apparatus that arrives at the division site and polymerizes at midcell, forming a filamentous ring (Z-ring) that recruits additional proteins required for the progression and the completion of cytokinesis (Weiss, 2004).

The transposon insertion previously identified in ftsZ was compatible with the presence of a truncated protein containing a significant amount (40%) of the amino acid sequence. Thus, we aimed to construct a M. genitalium ftsZ null mutant by gene replacement to better understand the role of this protein. The obtained mutant strain was characterized in terms of growth kinetics, ultrastructure, cell adhesion and gliding motility. The results confirm our previous expectations indicating a non-essential role of ftsZ for cell growth and reveal that, in the absence of the FtsZ protein, M. genitalium can still manage a feasible cell division and cytokinesis using the force generated by the gliding motility apparatus. To our knowledge, this is the first report depicting alternative mechanisms to perform cell division by binary fission in a microorganism that already contains the ftsZ gene.


Obtaining a M. genitalium ftsZ null mutant

Mycoplasma genitalium ftsZ (ORF mg224) was deleted by gene replacement. The pΔftsZTet suicide plasmid was designed to contain the tetM438 selectable marker enclosed by the flanking regions of the mg224 ORF (Fig. 1A). After transforming M. genitalium wild-type (WT) G37 strain with plasmid pΔftsZTet, recombinant clones were analysed by Southern blot (Fig. 1B). Two of ten clones showed the intended replacement as a result of a double-cross-over recombination event between pΔftsZTet and the mycoplasma genome, resulting in the complete deletion of the mg224 ORF (Fig. 1A). The absence of ftsZ in the null mutants, which were termed ΔftsZ, was also confirmed by PCR and Southern blot (Fig. 1C and E). Importantly, plasmid pΔftsZTet was designed to preserve promoter signals surrounding the mg224 ORF (Fig. 1A), in particular the previously identified promoter sequence upstream of mg225 ORF (Benders et al., 2005). To confirm that the deletion in ftsZ was not affecting the transcription of downstream genes, cDNA from mg225 ORF in ΔftsZ mutants was consistently amplified by RT-PCR (Fig. 1D).

Figure 1.

Engineering of ΔftsZ mutants.
A. Schematic representation of the double-cross-over recombination event between pΔftsZTet and the homologous regions flanking ftsZ. DFR and UFR represent the downstream and upstream flanking regions of mg224 in pΔftsZTet respectively. The size of fragments resulting from the EcoRV (E) digestion of WT genomic DNA and the size of fragments resulting from the KpnI (K) digestion of WT and ΔftsZ genomic are indicated. S1 and S2 are the probes used in Southern blots B and F respectively. Primers used in PCR reactions and RT-PCR are also shown with numbered arrows. P1: 5′BEKOftsZ; P2: 3′BEKOftsZ; P3: 5′ftsZ; P4: 3′ftsZ; P5: 5′RTmg225; and P6: 3′BDKOftsZ (Table S1). Additional restriction sites used for cloning: PstI (P), EcoRI (E1), ClaI (C) and BamHI (B).
B. Southern blot to demonstrate the replacement of ftsZ by the tetM438 selectable marker. The 4.4 kb band was obtained by hybridizing the KpnI digestion of ΔftsZ genomic DNA with probe S1.
C. PCR detection of ftsZ. Lanes 1–3: PCR amplification using primers 5′ftsZ (P3) and 3′ftsZ (P4) and genomic DNAs of WT (lane 1) and the ΔftsZ mutants obtained (lanes 2 and 3). Lanes 4–6: Positive controls of PCR amplifications using primers 5′BEΔftsZ (P1) and 3′BEΔftsZ (P2) and the same DNA templates used in lanes 1–3 respectively.
D. RT-PCR of mg225 transcript in the WT and ΔftsZ mutant. Total RNA from the M. genitalium G-37 (lane 2) and ΔftsZ mutant (lane 4) was retrotranscribed using primer 3′BDKOftsZ (P6). The resulting cDNAs were amplified by PCR using primers 5′RTmg225 (P5) and 3′BDKOftsZ (P6), obtaining a fragment of 450 bp. The negative controls corresponding to the amplification by PCR of total RNAs from the WT and the ΔftsZ mutant treated with DNase I are shown in lanes 1 and 3 respectively.
E. Southern blot to detect the presence of ftsZ and related sequences in the genome of the ΔftsZ mutant. EcoRV-digested genomic DNAs from WT and ΔftsZ (lanes 1 and 2 respectively) were hybridized with probe S2 (lanes 3 and 4 respectively). A 10.8 kb band is readily detected in lane 3 and has the size expected according to the EcoRV sites shown in (A). No bands were observed in lane 4 even after an extended detection time.

Growth kinetics of ΔftsZ cells

The growth kinetics of the ΔftsZ mutant was compared with that of the WT strain (Fig. 2). The cell mass from several cultures of both strains was measured by ATP luminometry (Stemke, 1995; Lluch-Senar et al., 2007). Cultures of both strains exhibited the same doubling time in log phase (11.8 h and 11.7 h respectively) and no significant differences were found when analysing additional parameters such as the cell mass in the stationary phase. These results indicate that cells not only remain viable upon deletion of ftsZ, but also that this deletion has no impact on growth kinetics of M. genitalium.

Figure 2.

Growth kinetics of WT and ΔftsZ strains. Semi-logarithmic plot with the growth curves of WT (squares) and the ΔftsZ mutant (triangles). Linear regression of ATP values from measures taken at 0 h, 16 h, 24 h and 39 h were used to determine the slopes corresponding to the exponential phase of growth. Error bars corresponding to the standard deviations of values obtained in three separate experiments are also represented.

Adhesion properties and cytoskeletal structure of ΔftsZ cells

Since FtsZ protein has previously been identified as a cytoskeletal protein in the related species M. pneumoniae (Regula et al., 2001), we were prompted to investigate the presence of defects in the adherence properties as well as downstream events affecting other cytoskeletal proteins in ΔftsZ mutants. Haemadsorption (HA) was qualitatively assessed, and colonies from ΔftsZ cells exhibited the same HA+ phenotype as colonies from WT cells. ΔftsZ cells did not show significant differences regarding their plastic adhesion properties (Table 1). Neither were there significant differences found when examining the protein profiles of ΔftsZ and WT cells by SDS-PAGE and Western blot. In particular, ΔftsZ cells exhibited WT levels of MG386, HMW1, HMW2, P140, P110, HMW3 and P41 proteins (data not shown), indicating that deletion of ftsZ does not give rise to detectable downstream events in most of the cytoskeletal proteins of M. genitalium. In addition, examination of ultrathin sections of ΔftsZ cells by transmission electron microscopy did not reveal appreciable ultrastructural changes in the electrondense core or other cytoskeletal components of the terminal organelle. However, long cell extensions were frequently observed in the pole opposite to the terminal organelle of ΔftsZ cells (Fig. 3A).

Table 1.  Parameters of WT and ΔftsZ strains.
Strain% Non-adherent cells ± SEDoubling time (h)Cytokinesis length (h)Gliding velocity (µm s−1) ± SENon-motile cells (%)Haemadsorption
  1. SE, standard error. ++: all (ΔftsZ) or most (WT) colonies in the HA assay were found to be evenly covered with erythrocytes (Fig. 5A).

WT33.3 ± ± 0.00210++
ΔftsZ34.7 ± ± 0.00339++
Figure 3.

Colony and cell morphology of WT and ΔftsZ strains.
A. TEM images of the WT and ΔftsZ strains. No significant differences are observed in the ultrastructure of the terminal organelle from cells of WT and ΔftsZ strains (electrondense cores are marked with a black arrow). The white arrow highlights the presence of a tail on the opposite side to the terminal organelle frequently found in ΔftsZ cells. Bar, 200 nm.
B. Colonies of WT and the ΔftsZ mutant grown in culture dishes covered with SP-4 medium containing 0.5% low-melting-point agarose. Bar, 250 µm.
C. Selected frames from microcinematographies of WT and ΔftsZ strains. The presence of many non-motile ΔftsZ cells is indicated by black arrows. Most of these cells appear frequently coupled to another non-motile cell and both are linked by a thin filament. A black circle indicates a single cell exhibiting a long filament. Bar, 5 µm. The complete cinematographies can be found in Supporting information as Movie S1 and Movie S2 respectively.

Gliding motility properties of ΔftsZ cells

Because no significant changes were observed in the cytoskeleton of the mutant cells, no changes were expected in the gliding motility properties of ΔftsZ cells. However, gliding motility was assessed by examining colony morphology in culture dishes covered with semi-solid medium. Under these conditions, colonies from cells with gliding motility defects are compact and do not exhibit the microsatellite colonies that are typically developed from gliding-proficient cells (Pich et al., 2006a). Surprisingly, colonies from ΔftsZ cells were more compact and with fewer satellite microcolonies than those from WT cells (Fig. 3B), suggesting that deletion of ftsZ interferes in some way with the gliding motion. Gliding motility was also monitored by microcinematography. The speed and movement patterns of ΔftsZ cells were essentially the same shown by WT cells, although it was noticeable that 39% of ΔftsZ cells versus 10% of WT cells remained non-motile during the observation period (Table 1). Interestingly, in the microcinematographies most non-motile ΔftsZ cells were found very close to another non-motile cell, and sometimes both cells seemed to be linked by a thin filament (Fig. 3C). Thus, the high ratio of non-motile ΔftsZ cells is consistent with the more compact phenotype of the colonies of this strain. However, the normal velocity and movement patterns exhibited by motile ΔftsZ cells suggest that the FtsZ protein is not directly involved in the gliding mechanics of mycoplasma.

Analysis of dividing ΔftsZ cells

To investigate the origin of non-motile cells detected in the ΔftsZ strain, the mutant strain was analysed by scanning electron microscopy (SEM). The images confirmed the presence of a large number of coupled cells linked by a thin filament (Fig. 4A and B) that were also observed in pictures from the cinematographies. Noteworthy, these coupled cells exhibited terminal organelles at both ends. Since the terminal organelle is the leading end in the mycoplasma motion (Bredt, 1968; Burgos et al., 2008), cells bearing two terminal organelles in the same direction but opposite senses are expected to be non-motile or barely motile. The images also suggest that these cells are dividing cells in the latter stages of cytokinesis, as has recently been demonstrated in M. pneumoniae (Hasselbring et al., 2006a). In addition, cells exhibiting a long tail in the pole opposite of the terminal organelle as well as chains of cells were also very common (Fig. 4A and B). Such morphologies were rarely observed when examining WT cells and are in agreement with the possibility that the force generated by the gliding motility apparatus in one or both daughter cells may be strong enough to separate dividing ΔftsZ cells.

Figure 4.

Stages in the cell division process of WT, ΔftsZ and TnGftsZ strains.
A. Histogram with the frequency of cells found in the different stages of cell division. ΔftsZ is indicated in black, WT in grey and in dark grey the complemented TnGftsZ strain. A picture with a representative cell from each stage is also shown. All pictures are at the same magnification (bar, 500 nm). Stages: 1, individual cells exhibiting a single terminal organelle and showing no tails in the opposite cell pole; 2, single cells exhibiting two terminal organelles, one of them migrating to the opposite cell pole; 3, dividing cells with two organelles in the opposite cell poles; 4, dividing cells with two organelles in opposite cell poles but joined by a filament; 5, chains of filamented cells; and 6, individual cells exhibiting a single terminal organelle and showing long tails on the opposite cell pole.
B. Representative SEM images of WT and ΔftsZ cells. Cells in latter stages of cell division are shown with white arrows. Individual cells after the division process with long tails on the opposite side of the terminal organelle are circled in white. Bar, 5 µm.

Introducing non-adherent mutations in the ΔftsZ strain

If the gliding apparatus is essential for cytokinesis in ΔftsZ cells, additional mutations abolishing motility in this strain should render cells unable to divide, thus being, in consequence, non-viable. Unfortunately, methods to construct strains bearing conditional lethal mutations are not currently available in mycoplasmas. However, we have taken advantage of the phase variation mechanism that deletes a region involving coding sequences of P110 and P140 adhesins (Burgos et al., 2006) to investigate the role of the gliding machinery in the cytokinesis of ΔftsZ cells. Since P140 and P110 proteins are required for cell adhesion, phase variants are also expected to be non-motile. Supporting this view, M. pneumoniae cells are non-motile after disruption of the gene coding for the P1 adhesin (Hasselbring et al., 2006b), homologous to the P140 adhesin. In addition, phase variants occur with high frequency in the WT strain and can be easily detected. Therefore, if gliding motility is essential for cell division in the absence of ftsZ, no phase variants should be detected among cells of the ΔftsZ strain.

Based on this rationale, the presence of HA- variants in serial culture passages of WT and ΔftsZ strains was screened. Viable, non-adherent cells were detected when they were able to develop HA- colonies on agar plates (Mernaugh et al., 1993). The frequency of phase variants was low at passage 0 of WT cultures (0.3%) but it increased rapidly until becoming stabilized with a frequency of around 2.8% in the last passages (Fig. 5A and B). In contrast, phase variants were never detected in the successive culture passages of the ΔftsZ strain. This result indicates that HA- phase variants are not viable in the genetic background of the ΔftsZ strain. The presence of the recombinant chromosomes from phase variants in the serial culture passages was also investigated. For this purpose, a quantitative real-time PCR (Q-PCR) was designed to detect a specific amplicon resulting from the recombination of the MgPa operon with MgPa island VI (Fig. 5C and D). Detection of this amplicon only reveals the presence of chromosomes from a fraction of non-adherent cells since additional recombination events rendering non-adherent cells are also possible with other MgPa islands (Burgos et al., 2006). The identity of this amplicon was confirmed by sequencing (data not shown). At passage 0, the frequency of recombinant chromosomes was very similar in the cultures from both strains (0.1%). This result indicates that recombination events deleting a portion of the MgPa operon also occur in the ΔftsZ strain and these events have a similar frequency in the WT strain. In WT cultures, the frequency of chromosomes with a deletion in the MgPa operon was progressively increased until reaching values close to 0.4% in the fourth passage and then remained stabilized until the end of the experiment (Fig. 5C). Cultures from the ΔftsZ strain did not show such increase in the number of recombinant chromosomes, indicating that cells bearing a deletion in the MgPa operon do not accumulate along the serial passages and mutations rendering non-adherent cells are lethal in the ΔftsZ strain.

Figure 5.

Quantification of non-adherent phase variants in serial culture passages.
A. Pictures of representative colonies from WT and ΔftsZ strains found after the HA assay. All pictures are at the same magnification. Bar, 50 µm.
B. Frequency of HA- colonies generated in serial passages. Black dots represent the percentage of HA- colonies in ΔftsZ in the selected passages. Black squares show the percentage of HA- colonies from the WT strain.
C. Frequency of chromosomes bearing deletions in the MgPa operon in serial cultures of WT (black squares) and ΔftsZ (black dots). These recombinant genomes are the result of a double cross-over between R3 and R5 boxes of MgPa VI island and the homologous regions in the MgPa operon (see D). Frequencies are the average of five replicate Q-PCR measurements and bars represent the respective standard errors.
D. Translocation of sequences from MgPa island VI to the MgPa operon in the M. genitalium genome. Schematic representation of a reversible, double-cross-over event between the R3′–R3 and R5′–R5 boxes from the MgPa operon (bases 220 000 to 230 000) to MgPa island VI (bases 310 000 to 320 000). Boxes R1 to R6 refer to the DNA repetitive sequences of the MgPa operon, which are also found in MgPa island VI (R1′ to R6′) located 90 kb downstream. P1 and P2 indicate the position of the primers used for Q-PCR. A number of additional double-recombination events are also possible but are not included in the drawing for clarity. The location of the different MgPa islands in the M. genitalium genome can be found in Lluch-Senar et al. (2007).

Complementation analysis of the ΔftsZ strain

Because only two ΔftsZ mutants were isolated by gene replacement, a complementation assay of the ΔftsZ strain was performed to exclude the presence of additional genetic defects in this strain. The WT ftsZ allele under the control of the mg438 constitutive promoter (Burgos et al., 2007) was reintroduced by transposition to ΔftsZ cells by using plasmid pTnGftsZ (Table S2). After transforming with this plasmid, 58% of the recovered colonies exhibited normal growth, with the presence of a large number of microsatellites (Fig. S1), suggesting that the WT ftsZ allele in ΔftsZ cells was able to restore the normal gliding phenotype. Four colonies were picked, filtered-cloned and the corresponding cultures were analysed by Southern blot to exclude the presence of multiple transposon insertions (data not shown). Microcinematographies from cells of two of these colonies confirmed that gliding motility parameters were restored to normal levels (Fig. S1). In the same way, SEM analyses demonstrated that frequencies of different morphologies were restored to values very close to those shown by WT cells (Fig. 4A). Finally, one of these transformants was submitted to successive passages and HA- phase variants were recovered again, with a very similar frequency to that observed in the WT strain (Fig. S1). Together, these data indicate that the presence of the WT ftsZ allele is able to restore the normal growth parameters in ΔftsZ cells, and exclude that genetic defects other than the absence of ftsZ are involved in the phenotype exhibited by the ΔftsZ strain.


The sequencing of mycoplasma genomes has revealed the presence of few genes orthologous to those involved in cell division of most cell-walled bacteria. Although genes involved in the segregation of the peptidoglycan layer are obviously useless in mycoplasmas, it is perplexing to see how these microorganisms can divide by binary fission in the absence of several genes crucial for the Z-ring architecture. Moreover, the absence of these genes is difficult to explain, since their function cannot be rescued simply by uploading external cell resources, as is the case of genes involved in metabolism, and suggests the existence of alternative mechanisms in the cell division of these microorganisms. The previous isolation of a ftsZ transposon-generated mutant prompted us to characterize the role of this gene and the encoded protein in M. genitalium in depth and thus gain insight into its biology and cell division systems. The data reported herein demonstrate that M. genitalium cells can divide by binary fission in the absence of FtsZ and highlight the role of the terminal organelle and its associated cytoskeletal structures in the cell division of this minimal microorganism. This finding is comparable to the L-forms of Bacillus subtilis dividing by a ‘budding’ or ‘extrusion-resolution’ mechanism in the absence of the ftsZ gene, probably reflecting the presence of ancient proliferative mechanisms in the modern microorganisms (Leaver et al., 2009). Our results suggest that in M. genitalium binary fission may be accomplished by two redundant mechanisms, the first one based on the well-known ftsZ machinery and the second one based on the pivotal roles of the mycoplasma cytoskeleton and gliding motility.

FtsZ has previously been identified as a component of the cytoskeletal fraction of M. pneumoniae cells (Regula et al., 2001), in accordance with the biochemical properties of eukaryotic cytoskeletal tubulins (Ramsby and Makovsky, 1999). Nonetheless, cells from the ΔftsZ strain do not exhibit adherence defects or detectable changes in their cytoskeletal ultrastructure and glide at the same velocity as do WT cells. Furthermore, ΔftsZ cells do not show downstream events in their levels of proteins involved in cell adhesion and motility. These functional impairments and such downstream events are commonly observed in strains with deletions or disruptions in genes coding for functional cytoskeletal proteins in M. genitalium (Burgos et al., 2006; 2008; Pich et al., 2008) and M. pneumoniae (Krause and Balish, 2001). While our results do not discard the possible association of FtsZ with the main structures of the mycoplasma cytoskeleton, they exclude a role for this protein in the functions traditionally linked to the terminal organelle and its associated cytoskeletal structures.

SEM analysis of the ΔftsZ strain reveals a high frequency of mutant cells in the latter stages of cytokinesis, suggesting that this process takes a long time in this strain (approximately 4 h). In contrast, the low frequency of dividing cells that can be observed when examining the WT population indicates that cytokinesis is a process that can be accomplished in 1 h and suggests that most of non-dividing cells are motile. Nevertheless, the long time spent to segregate dividing ΔftsZ cells has no impact in the overall duration of the cell cycle in M. genitalium since doubling times are very similar in both strains, suggesting that the temporal control of the cell cycle in mycoplasmas is independent of the length of cytokinesis. This is in contrast with previous observations indicating the presence of specific checkpoints controlling the cell cycle of Caulobacter crescentus (Hottes et al., 2005; Jensen, 2006; Shen et al., 2008). Alternatively, the time between two consecutive division rounds in M. genitalium may be long enough to provide a significant buffering capacity in the case of a delay in the cytokinesis, as has been observed in the ΔftsZ cells.

We have provided evidence that mutations rendering non-adherent cells might be lethal in the ΔftsZ strain. This was demonstrated by quantifying the appearance of viable non-adherent phase variants and chromosomes with deletions in the MgPa operon along serial culture passages. In these experiments, the frequency of non-adherent phase variants was very low in the first passages of WT cultures. This result can be explained by the fact that only collected cells growing attached to the plastic surface of culture flasks from routine cultures of M. genitalium WT cells were collected, thus avoiding the accumulation of non-adherent cells in the laboratory stocks. Since both adherent and non-adherent cells were used to start the next serial passages, the frequency of non-adherent cells progressively increased, reaching values close to 3%. This frequency is high enough to allow the detection of phase variants among the ΔftsZ cell population if such non-adherent cells are viable. In the same way, recombinant chromosomes bearing deletions in the MgPa operon do not accumulate in the successive serial passages, supporting that HA- phase variants are not viable in ΔftsZ cultures. Furthermore, SEM images suggest that the long tails in the mutant strain are produced as a consequence of the force generated by the gliding motility apparatus in the absence of FtsZ. Such tails become progressively longer and thinner until they are eventually broken. Taken together, these results strongly suggest that gliding motility is essential to segregate dividing ΔftsZ cells. However, such broken tails have to be rapidly resealed to maintain cell integrity. Because ΔftsZ cells grow at the same rate as WT cells, these data suggest that the mycoplasma cell membrane is very efficient when sealing such broken ends. A similar conclusion can be drawn from a previous work showing that terminal organelles may detach from the main cell body of M. pneumoniae cells deficient for P41 (Hasselbring and Krause, 2007).

Mycoplasmas lack genes coding for Min and SlmA or Noc proteins. These proteins, especially the Min group, have a very important role as negative effectors of FtsZ polymerization and are essential for the establishment of cell polarity and the correct placement of the division site at midcell. SlmA and Noc proteins provide a second mechanism, the nucleoid occlusion system, to prevent the formation of septa over nucleoids (Rothfield et al., 2005). On the other hand, M. genitalium strains with deletions in mg191 or mg192 lose the terminal organelle and exhibit a high frequency of cells of variable size and cells with a multilobed or a pleomorphic appearance, indicating the presence of defects in their cell division process (Burgos et al., 2006). Preliminary work also indicates the existence of dramatic changes in the electrondense core ultrastructure of the cells from those strains, thus suggesting a role for cytoskeleton in the cell division of M. genitalium (R. Burgos et al., unpubl. data). Because the cytoskeleton is the most notable polar component of this microorganism, this structure may also be involved in the placement of the Z-ring in WT cells. Therefore, the presence of a cytoskeleton in many mycoplasma species can also be seen as an opportunity to undertake many of the functions of cell division proteins missing in mycoplasmas.

One open question finally arises. If ftsZ is not essential for cell division in mycoplasma, why is this gene still present in the streamlined genome of M. genitalium? As is shown here, dividing ΔftsZ cells remain fixed at the same location for extended periods of time. Such lengthy time in the absence of movement may be detrimental for the in vivo survival of this mycoplasma, in this way providing a selective pressure against the loss of this gene. However, alternative explanations may also be possible, since dividing cells of M. pneumoniae, a related respiratory mucosal pathogen, can remain non-motile for several hours (Hasselbring et al., 2006a) despite the presence of a functional ftsZ. We suggest an alternative possibility, related to the intracellular residence of M. genitalium, also favouring the conservation of this gene. Soon after the internalization of M. genitalium inside eukaryotic host cells, mycoplasmas appear enclosed in vacuoles, many of them with a perinuclear location (Mernaugh et al., 1993; Jensen et al., 1994; McGowin et al., 2009). Since one of the prerequisites for gliding motility is the presence of a solid surface, it is plausible to think that the gliding force can be ineffective to support cell division in internalized mycoplasmas. In this scenario, ftsZ should be essential for the intracellular survival of M. genitalium. Fortunately, this hypothesis is fully testable and may reveal a new role for ftsZ as a virulence factor of mycoplasmas.

Experimental procedures

Bacterial strains and growth conditions

Escherichia coli strain XL-1 Blue was grown at 37°C in 2YT broth or LB agar plates containing 75 µg ml−1 ampicillin, 40 µg ml−1 X-Gal and 24 µg ml−1 Isopropyl-beta-thio galactopyranoside (IPTG) when needed. The M. genitalium G-37 WT strain was grown in SP-4 medium (Tully et al., 1979) at 37°C under 5% CO2 in tissue culture flasks (TPP). To select the pΔftsZTet M. genitalium transformant cells, plates were supplemented with 2 µg ml−1 tetracycline. To select pTnGftsZ-complemented cells (TnGftsZ strain), plates were supplemented with 2 µg ml−1 tetracycline and 100 µg ml−1 gentamicin.

DNA manipulations

DNA genomic of M. genitalium was isolated by using the E.Z.N.A. Bacterial DNA Kit (Omega BIO-TEK). Plasmidic DNA was obtained by using the Fast Plasmid Mini Eppendorf Kit. All primers and plasmids used in this work are summarized in Tables S1 and S2. The purification of PCR products and digested fragments from agarose gels was achieved using the E.Z.N.A. Gel Extraction Kit (Omega BIO-TEK).

Molecular cloning

A 1 kb PCR fragment encompassing the upstream region of mg224 was obtained by using primers 5′BEΔftsZ and 3′BEΔftsZ, which contain the PstI and EcoRI restriction sites, respectively, at their 5′ ends. Another 1 kb PCR fragment encompassing the downstream region of mg224 was obtained by using primers 5′BDΔftsZ and 3′BDΔftsZ that contain, at their ends, the BamHI and ClaI restriction sites respectively. Both PCR fragments were cloned into EcoRV-digested pBE (Pich et al., 2006b), excised with the corresponding enzymes (Roche), and ligated together with a 2 kb BamHI–EcoRI fragment encompassing the tetM438 selectable marker (Pich et al., 2006b) and a PstI–ClaI-digested pBSKII(+) (Invitrogen). This ligation mixture resulted in the pΔftsZTet vector that was used to transform M. genitalium G-37 and obtain the mg224 null mutant (ΔftsZ). To construct the pTnGftsZ plasmid, a 1.1 kb PCR fragment encompassing the mg224 gene was first obtained by using primers 5′TnftsZ and 3′TnftsZ. Primer 5′TnftsZ contains a PstI restriction site and the promoter sequence of M. genitalium mg438 ORF (Pich et al., 2006b). Primer 3′TnftsZ contains a EcoRI restriction site at its 5′ end. The PCR fragment was cloned into EcoRV-digested pBE, excised with EcoRI–PstI enzymes, and ligated to the pMTnGm EcoRI–PstI-digested vector (Pich et al., 2006b). The resulting vector was used to transform the ΔftsZ strain as described below.

Transformation of M. genitalium

Transformation of M. genitalium was performed by electroporation using 30 µg of DNA as previously described (Pich et al., 2006b).

Southern blotting and PCR assays

Genomic DNA of pΔftsZTet transformants were digested with KpnI enzyme and hybridized as previously described (Pich et al., 2006b) using probe S1, which was a 1 kb fragment obtained by digesting pΔftsZTet with PstI/EcoRI restriction enzymes (Fig. 1A). Genomic DNAs of ΔftsZ and WT were digested with EcoRV and hybridized using probe S2, which was a 1.1 kb fragment encomprisingftsZ amplified by PCR using 5′ftsZ and 3′ftsZ primers (Fig. 1A and Table S1). The ΔftsZ mutants were also checked by PCR. The positive control was performed by amplifying a 1 kb PCR fragment encompassing the upstream region of mg224 by using primers 5′BEΔftsZ and 3′BEΔftsZ (Table S1) and WT or ΔftsZ genomic DNAs as a template. To check the presence of ΔftsZ mutants, a 1.1 kb fragment of ftsZ was amplified using 5′ftsZ and 3′ftsZ primers (Table S1) and genomic DNA of WT or ΔftsZ as a template (Fig. 1C). Genomic DNAs of four TnGftsZ mutants were digested with EcoRV and hybridized using a probe of the gentamicin gene (data not shown). PCR products were sequenced to confirm the identity of the amplified fragments (data not shown).

RNA manipulations

Total RNA from 20 ml of mid-log-phase cultures of M. genitalium G-37 and ΔftsZTet mutants was extracted by using the RNAaqueous kit (Ambion). For RT-PCR assays, total RNA was treated with DNase I (New England Biolabs), and retrotranscribed using the SuperScript first-strand synthesis system kit (Invitrogen) and the 3′BDKOftsZ primer. The cDNA was amplified by PCR using the 5′RTmg225 and 3′BDKOftsZ primers, obtaining a fragment of 450 bp (Fig. 1D).

Electron microscopy

Samples were analysed by TEM as previously described (Pich et al., 2008). Samples were analysed by SEM according to the following procedure. Mid-log-phase cells grown in Lab-Tek chamber slides (Nunc) were washed three times with phosphate-buffered saline (PBS) and fixed with 1% glutaraldehyde for 1 h. Samples were washed three times with PBS and then dehydrated sequentially with 30%, 50%, 70%, 90% and 100% ethanol for 10 min each. Immediately, samples were critical point dried (K850 critical point drier; Emitech Ashfort; UK) and sputter coated with 20 nm thin gold layer. Samples were observed using a Hitachi S-570 (Tokyo, Japan) microscope. About 500 cells from each strain were analysed to determine the frequency of cells in the different stages. The percentages of morphologies showing a terminal organelle at both poles of mycoplasma cells (Fig. 4A, stages 3 and 4) were used to estimate the length of cytokinesis. This length was calculated as the product of the frequency of dividing cells by the doubling time derived from the slopes of growth curves (see below).


Samples with different cell densities were prepared as previously described (Pich et al., 2006a). Cell movement was examined at 37°C using a Nikon Eclipse TE 2000-E microscope and images were captured with a Digital Sight DS-SMC Nikon camera controlled by NIS-Elements BR software. Particular movements of 600 individual cells from WT and mutant strains were analysed to determine the frequency of motile cells during the observation period. Tracks from 50 individual motile cells corresponding to 2 min of observation and in six separate experiments were analysed to determine the gliding velocity and gliding motility patterns.

Quantitative plastic binding assay

Frozen stocks of the WT and ΔftsZ cells were diluted in 1 ml of SP-4 medium to give a final concentration of approximately 107 cfu ml−1. The binding assay was performed as previously described (Burgos et al., 2007).

HA activity

For qualitative assessment of the HA, colonies grown in SP4 plates were flooded with 2 ml of human erythrocytes diluted (1:50) in PBS and incubated for 1 h at 37°C. Plates were subsequently washed three times with PBS and observed using a LeicaMZFLIII microscope. Pictures were taken using a LeicaDC500 camera connected to the microscope.

Growth curves

To compare the growth kinetics of WT and ΔftsZ cells, two independent experiments were performed using 48 cultures of 1 ml inoculated with approximately 107 cfu ml−1 and grown in 24-well TPP plates at 37°C. ATP content from six wells corresponding to each one of the different growth intervals was measured using the ATP Bioluminescence Assay Kit HS II (Roche). Measuring ATP instead of turbidity is considered the most proper method to estimate mycoplasma cell mass (Stemke, 1995; Lluch-Senar et al., 2007). Briefly, 1 ml of lysis reagent buffer (Roche) was added to each well and 50 µl of samples were then placed in Wallac B&W isoplate polystyrene plates (Perkin Elmer). The ATP standard curve was constructed in the range of 10−6 to 10−12 M. Standards and samples were measured by duplicate after the automated injection of 50 µl of luciferase reagent in a Victor III luminometer (Perkin Elmer) using SP-4 medium as blank. ATP concentrations were calculated from a log–log plot of the standard curve data. The slopes corresponding to the exponential phase of growth were estimated from the linear regression of ATP values obtained from cultures grown for 0 h, 16 h, 24 h and 39 h. Mass doubling time of WT and ΔftsZ was calculated from the slopes of the growth curves.

Obtaining and detecting non-adherent mutants

To detect spontaneous, non-adherent phase variants, WT, ΔftsZ and TnGftsZ strains were grown in 25 cm2 flasks with 5 ml of SP-4 medium inoculated with approximately 107 cfu ml−1. After ∼80 h the cultures were scraped off from the flasks and 2 ml of samples of the suspensions were centrifuged to harvest cells, which were stored at −80°C for further DNA extraction. Successive passages were also performed in 25 cm2 cell culture flasks with 5 ml of SP-4 that were inoculated with 50 µl of the cell suspension from the previous passage. The remaining of the cell suspension was passed through a 0.45 µm low binding protein filter (Millipore) and 300 µl of samples of serial dilutions were spread on SP-4 plates, which were incubated 8 days at 37°C. Plates showing 400–500 colonies (about 1500 colonies per passage) were tested for HA as described above.

Quantitative PCR

The quantitative assay to detect the presence of recombinant chromosomes bearing deletions in the MgPa operon was performed using the LightCycler-FastStart DNA Master SYBR Green I (Roche) and the LightCycler.2 instrument (Roche). WT and ΔftsZ genomic DNAs were used for two single-step PCR reactions using the R3-5′ and R5-3′ primers (Table S1) and primers mg281-5′ and mg281-3′. R3-5′ and R5-3′ primers can detect the presence of a 558 bp DNA fragment (R3) resulting from the recombination between the MgPa operon and MgPa island VI (Fig. 5D) in both WT and ΔftsZ genomes. Primers mg281-5′ and mg281-3′ amplify a 621 bp DNA fragment from mg281 (T281) to quantify the number of M. genitalium genomes in the PCR reactions. The amplified fragments were detected by emission of fluorophore SYBR green at the 485 nm wavelength and the quantitative determination was carried out during the exponential phase of amplification. Multiple real-time data acquisition and analysis of the samples and of the known standard serial dilutions were performed and analysed by MyiQ software (Roche). Dilutions of the known-titre standards (6.05 ×  1013 molecules ml−1 for pBE-R3 and 4.13 × 1012 molecules  ml−1 for pBE-T281; Table S2) were prepared in DEPC-H2O, and 2 µl of each was used for the calibration curve in the amplification reaction. The calibration curves, obtained by quantification of the serial standard dilutions, were plotted in the dynamic range from 1.21 × 104 to 12 molecules ml−1 for pBE-R3 and from 8.26 × 106 to 8.26 × 102 molecules ml−1 for pBE-T281. The concentration of the unknown-titre samples was obtained by interpolation on the calibration curve suggested by the software. A negative water control was added in each run to avoid overestimation of the specific product due to primer–dimer formation. Finally, the PCR amplification products were checked by agarose gel electrophoresis. Data in Fig. 5C are derived from five replicate Q-PCR measurements.


This work was supported by Grant BIO2007-67904-C02-01 from the MCYT (Ministerio de Ciencia y Tecnología, Spain) to E.Q. and by the Centre de Referència de R+D de Biotecnologia (Generalitat de Catalunya). E.Q. is the Chair Holder of a ‘Cátedra de Transferencia de Tecnología Parc de Recerca UAB-Grupo Santander’. M.L.-S. acknowledges a FPI predoctoral fellowship from the MCYT. We acknowledge A. Sánchez-Chardi for his advice in transmission electron microscopy and Drs J.L. Arolas, A. Mozo and L. Strother for their helpful comments.