|Home | About | Journals | Submit | Contact Us | Français|
The rod-shaped cells of the bacterium Myxococcus xanthus move uni-directionally and occasionally undergo reversals during which the leading/lagging polarity axis is inverted. Cellular reversals depend on pole-to-pole relocation of motility proteins that localize to the cell poles between reversals. We show that MglA is a Ras-like G-protein and acts as a nucleotide-dependent molecular switch to regulate motility and that MglB represents a novel GTPase-activating protein (GAP) family and is the cognate GAP of MglA. Between reversals, MglA/GTP is restricted to the leading and MglB to the lagging pole defining the leading/lagging polarity axis. For reversals, the Frz chemosensory system induces the relocation of MglA/GTP to the lagging pole causing an inversion of the leading/lagging polarity axis. MglA/GTP stimulates motility by establishing correct polarity of motility proteins between reversals and reversals by inducing their pole-to-pole relocation. Thus, the function of Ras-like G-proteins and their GAPs in regulating cell polarity is found not only in eukaryotes, but also conserved in bacteria.
Cell polarity is a fundamental property of all cells and involves establishing and maintaining the spatial asymmetry of macromolecules (Rafelski and Marshall, 2008). An important consequence of cell polarity is that the activity of asymmetrically localized proteins is spatially confined, thus, laying the foundation for processes that require the localized activity of a protein or protein complexes (Nelson, 2003; Gitai et al, 2005). Cell polarity touches on essentially every aspect of cell function and the processes in which polarity has a decisive function are remarkably similar in eukaryotic cells and bacteria and include cell growth, cell cycle control, division, differentiation, and motility (Etienne-Manneville and Hall, 2002; Gitai et al, 2005; Shapiro et al, 2009).
Major questions in understanding cell polarity are how proteins find their correct localization and how this localization may change dynamically over time. In eukaryotic cells, Ras-like G-proteins, which can be divided into five major subfamilies (Ras, Rho, Rab, Arf, and Ran) (Leipe et al, 2002), have essential functions in establishing and maintaining cell polarity. These proteins are binary molecular switches that cycle between an inactive GDP- and an active GTP-bound state (Bourne et al, 1991). The nucleotide-bound state is regulated by guanine nucleotide exchange factors (GEFs) that catalyse the intrinsically slow exchange of GDP for GTP, and GTPase-activating proteins (GAPs), which stimulate the low intrinsic GTPase activity (Bourne et al, 1991). GTP binding induces a conformational change in Ras-like proteins, thereby promoting their interaction with effectors that control or are part of downstream pathways to elicit a particular response (Wittinghofer and Nassar, 1996). Often the activity of Ras-like G-proteins is temporally and spatially regulated. For example, in Dictyostelium discoideum Ras activation occurs at the leading edge of cells exposed to a gradient of chemoattractant (Charest and Firtel, 2007) and in Saccharomyces cerevisiae, Cdc42, a master regulator of cell polarity conserved from yeast to mammals, is activated at the incipient bud site (Etienne-Manneville, 2004).
Over the last 10 years, it has become clear that bacteria are spatially highly organized and, thus, display cell polarity (Gitai et al, 2005; Shapiro et al, 2009). Spatially organized elements of bacteria include proteins as well as the chromosome (Viollier et al, 2004). Little is known about how polarity in bacteria is established and maintained. However, some principles are emerging for polarized proteins. First, protein localization may be positively regulated by trans-acting targeting factors (Shapiro et al, 2009) as shown for DivIVA, which directly recruits MinJ to the cell poles in Bacillus subtilis (Bramkamp et al, 2008; Patrick and Kearns, 2008). Second, localization may be negatively regulated as in the case of the cell division machinery. In this process, negative regulators inhibit FtsZ-ring formation in the entire cell with the exception of the incipient division site (Lutkenhaus, 2007). Third, some localized proteins interact with proteins that bind to specific sites on the chromosome (Thanbichler and Shapiro, 2008). As the chromosome is spatially organized, this interaction results in localization of the interacting protein. Finally, a mechanism for protein localization involving recognition of membranes of positive or negative curvature was proposed for some peripheral membrane proteins in B. subtilis (Shapiro et al, 2009). Except for proteins involved in cell cycle regulation and cell division, most polarized proteins in bacteria are not dynamically localized and do not change localization over time (Shapiro et al, 2009). The dynamic localization of these proteins depends on specific cell cycle events, which by largely unknown mechanisms trigger protein relocalization (Shapiro et al, 2009). Many dynamically localized proteins are restricted to the cell poles or the division site, and it has been proposed that proteins targeted to a cell pole by the earlier cell division could act as targeting factors for other polarly localized proteins as in the case of the TipN protein in Caulobacter crescentus (Huitema et al, 2006; Lam et al, 2006). Cells of the rod-shaped bacterium Myxococcus xanthus provide a simple experimental system to address how bacterial cell polarity and dynamic protein localization is accomplished because these cells switch their leading/lagging pole polarity axis frequently and independently of the cell cycle to regulate cell motility.
M. xanthus cells organize into two patterns, spreading colonies in the presence of nutrients and fruiting bodies in the absence of nutrients. Formation of both patterns depends on regulation of motility (Leonardy et al, 2008). M. xanthus has two motility systems, type IV pili (T4P) (Wu and Kaiser, 1995) and the A-engine (Hodgkin and Kaiser, 1979), which act synergistically (Kaiser and Crosby, 1983). M. xanthus cells move uni-directionally on surfaces; occasionally, however, cells stop and then resume motility in the opposite direction, with the old leading cell pole becoming the new lagging pole (Blackhart and Zusman, 1985). These events are referred to as reversals and at the cellular level a reversal corresponds to an inversion of the leading/lagging pole polarity axis. Cellular reversals are stochastic events, not coupled to the cell cycle, and the reversal frequency is regulated by the Frz chemosensory system (Blackhart and Zusman, 1985). Moreover, proper regulation of the reversal frequency is essential for formation of both cellular patterns (Blackhart and Zusman, 1985).
In M. xanthus, T4P are assembled at the leading pole, and during a reversal, this pole changes (Sun et al, 2000; Mignot et al, 2005; Bulyha et al, 2009). Between reversals, the ATPase PilT that energizes T4P retractions localize at the lagging pole, whereas the ATPase PilB, which catalyses T4P extension, and the FrzS protein, which has an unknown function in T4P function, localize at the leading pole. During reversals, these three proteins relocate between the poles. Several other T4P proteins localize in bipolar, symmetric clusters, which remain stationary at the poles during reversals (Bulyha et al, 2009).
It is not known how force is generated in the A-motility system. However, this motility system also depends on polar localization of proteins between reversals and their dynamic relocalization between poles during reversals as shown for the two A-motility proteins, AglZ and RomR, which have unknown functions in A-motility. Between reversals, AglZ localizes in a large cluster at the leading pole and smaller clusters, also referred to as focal adhesion complexes, along the cell body (Mignot et al, 2007) and RomR localizes with a large cluster at the lagging pole (Leonardy et al, 2007). During reversals, these polar clusters relocate in opposite direction between the poles. Thus, at the molecular level, a cellular reversal corresponds to the inversion of the polarity axis of proteins of T4P as well as of the A-engine.
The MglA protein is important for the function of both motility systems (Hodgkin and Kaiser, 1979). MglA shares homology to Ras-like G-proteins (Hartzell and Kaiser, 1991; Leipe et al, 2002) and is a member of the sixth major subfamily of these proteins (Leipe et al, 2002). Recently, it was found that MglA influences the polarity of motility proteins and in an mglA mutant RomR localizes in a single cluster at the ‘wrong pole', that is the pole containing T4P (Leonardy et al, 2007) and AglZ in a mostly diffuse pattern (Mauriello et al, 2010). It has also been shown that MglA has GTPase activity (Mauriello et al, 2010). These observations taken together with the knowledge that Ras-like proteins function in cell polarity in eukaryotes made us hypothesize that MglA is involved in regulating the polarity of motility proteins. Here, we show that MglA is a nucleotide-dependent molecular switch and that MglB, which is encoded in an operon with MglA, is a MglA-specific GAP. MglA/GTP represents the active form and stimulates motility by setting up the correct polarity of motility proteins. Further accumulation of MglA/GTP induces reversals by stimulating pole-to-pole relocation of motility proteins. MglA activity is controlled temporally and spatially by the Frz system and the MglAGAP MglB.
The molecular mass of MglA (22 kDa) is Ras like, and MglA contains most of the conserved canonical residues required for guanine nucleotide binding and GTP hydrolysis observed in Ras-like G-proteins. The major differences are the absence of a Thr residue in the switch I region and the absence of the highly conserved Asp in the DxxGQ motif in the switch II region (Supplementary Figure S1).
The analysis of mutants stabilized in the GTP-bound form as well as dominant-negative and inactive mutants have provided insights into the in vivo function of Ras-like proteins in eukaryotic cells. We used a similar approach to analyse the importance of the GTPase cycle of MglA and generated MglA mutants with presumably similar effects on nucleotide binding and/or hydrolysis. The MglAG21V mutant has a substitution in the P-loop corresponding to the oncogenic RasG12V mutant and is predicted to be permanently in the GTP-bound active state because both the intrinsic and even more significantly the GAP-stimulated GTPase activity is dramatically reduced (Scheffzek et al, 1997; Vetter and Wittinghofer, 2001). The MglAT26/27N mutant was designed based on the assumption that its properties are similar to the RasS17N mutant, which has a lower affinity for GTP than for GDP, and is in general much less stable because of the absence of the residue coordinating the Mg2+ ion. However, it is not necessarily locked in the GDP-bound state and its major defects seem to be the reduced binding of nucleotides and tight binding to GEF (John et al, 1993; Cool et al, 1999; Feig, 1999). In MglA, T26 corresponds to S17 in Ras. To exclude that T27 in MglA would take over the function of T26 and coordinate the Mg2+ ion, T26 as well as T27 were substituted to N in MglAT26/27N.
To confirm that MglA functions as a Ras-like G-protein and to characterize the mutant proteins, we attempted to purify the wild type (wt) and mutant M. xanthus proteins after overexpression in Escherichia coli. However, in our hands, all three proteins were mostly insoluble. Even the soluble fraction could not be purified to homogeneity and tended to aggregate at higher concentrations. Thermus thermophilus contains an mglBA operon encoding MglA and MglB proteins that are 62/81% and 28/52% identical/similar to MglA and MglB of M. xanthus, respectively (Supplementary Figure S1). The T. thermophilus MglA and MglB proteins (see below) were expressed and purified in soluble form and did not aggregate (Figure 1A). We, therefore, investigated the biochemical properties of the MglA/B proteins using the homologues from T. thermophilus.
Nucleotide binding is conveniently measured using N-methylanthraniloyl (m, mant)-labelled nucleotides. Although not all the canonical residues of eukaryotic Ras-like G-proteins are conserved, nucleotide-free MglA+ nevertheless bound mGDP and mGTP as well as the non-hydrolysable GTP analogue mGppNHp with nanomolar affinities, whereas mADP and mATP were bound with much lower affinities, which could not be quantified with confidence (Figure 1B). MglAG21V and MglAT26/27N were not significantly affected in nucleotide binding except that MglAT26/27N bound mGTP and mGppNHp with strongly reduced affinity (Figure 1B). To determine whether MglA has GTPase activity, the release of Pi from [γ-32P]GTP was measured. As is characteristic for Ras-like G-proteins, MglA+ showed a slow intrinsic GTPase reaction with a half-life of about 80 min, which is reduced about seven-fold in the case of MglAG21V (Figure 1C). This supports the hypothesis that MglA+ is a G-protein with properties similar to Ras-like proteins and that MglAG21V corresponds to the oncogenic-activating RasG12V mutant. The MglAT26/27N mutant showed an even more reduced intrinsic GTPase reaction (Figure 1C). In conjunction with the strongly reduced mGTP and mGppNHp affinities, we predict that M. xanthus MglAT26/27N is a loss of function and dominant-negative mutant of MglA because of loss of stability, tight interaction with a respective GEF, and/or disturbed interaction with downstream effectors similar to the corresponding mutants of Ras-like proteins (John et al, 1993; Cool et al, 1999; Feig, 1999).
mglA is in an operon with the upstream mglB gene (Stephens et al, 1989), and it has been reported that MglB is important for accumulation of MglA (Hartzell and Kaiser, 1991). Therefore, the mglAG21V and mglAT26/27N alleles were expressed together with mglB after integration at the M. xanthus phage Mx8 attachment site in both wt and in a ΔmglBA mutant. Using time-lapse microscopy, we found that wt cells moved with an average velocity of 3.2±0.3 μm min–1 and reversed their direction with an average reversal period of 13.9 min (Table Ia). The ΔmglBA mutant was non-motile. Expression of mglBA+ and mglBAG21V at native levels (Supplementary Figure S2) corrected the motility defects caused by ΔmglBA; however, mglAG21V cells reversed approximately three-fold more frequently than wt and with an average reversal period of 4.6 min. However, mglBAT26/27N did not complement the motility defects caused by ΔmglBA. Similar results were obtained when MglAG21V and MglAT26/27N were synthesized from the native chromosomal site. Merodiploid mglA+, mglAG21V cells moved at wt velocities, but hyper-reversed similarly to the mutant only accumulating MglAG21V (Table Ia). Merodiploid mglA+, mglAT26/27N cells also moved at wt velocities, but only reversed every 48 min. Thus, the mutant mglA alleles are dominant over mglA+ with respect to reversals. Taken together with the biochemical characteristics of the three MglA proteins, these data are in agreement with the hypothesis that MglA functions as a nucleotide-dependent molecular switch in which MglA/GTP represents the active form stimulating motility as well as reversals and MglAT26/27N, an inactive and dominant-negative form.
Frz regulates the reversal frequency, but is not required for motility per se. To determine whether Frz and MglA act in the same or in parallel pathways to induce reversals, we performed genetic epistasis tests (Table Ib). As earlier reported, cells containing frzlof (loss-of-function) or frzgof (gain-of-function) alleles rarely reversed and hyper-reversed, respectively. Importantly, the frzlof, mglAG21V mutant had a hyper-reversing phenotype similar to that of the frz+, mglAG21V mutant. In addition, the frzgof, mglAG21V mutant reversed with the same reversal period as frz+, mglAG21V cells. Moreover, all strains containing the mglAT26/27N allele were non-motile. Thus, mglAG21V bypasses frzlof for reversals and the frzgof and mglAG21V mutations do not cause an additive phenotype. These observations suggest that Frz and MglA act in the same pathway and that MglA acts downstream of Frz to induce reversals. These data also show that MglA/GTP has two separable activities. One activity is stimulation of motility and this activity is independent of Frz. The second activity is stimulation of reversals and this activity depends on Frz.
As MglA stimulates motility and reversals, we speculated that MglA activity is spatially regulated. Consistently, Mauriello et al (2010) found that a partially active MglA–YFP protein localizes to the leading pole. We generated an active YFP–MglA+ protein, which corrected the motility defects caused by the mglA9 mutation (Table Ic) or a ΔmglA mutation (data not shown). Immunoblots showed that YFP–MglA+ accumulated at levels similar to MglA+ in wt cells (Supplementary Figure S2). Degradation products similar in sizes to those of MglA+ and YFP also accumulated, suggesting that a fraction of YFP–MglA+ is cleaved near the fusion site. In moving cells, YFP–MglA+ localized in a cluster at the leading pole as well as diffusely to the cytoplasm (Figure 2A). This localization is in agreement with that observed using a partially active MglA–YFP protein except that we did not observe YFP–MglA+ localizing to focal adhesion complexes (Mauriello et al, 2010). We speculate that this difference could be caused by the partial activity of the MglA–YFP fusion used by Mauriello et al. Alternatively, the high background fluorescence observed with our YFP–MglA+ fusion, which might be caused by degradation of the fusion protein, masks the signal from the focal adhesion complexes. YFP–MglA+ localization did not change, whereas a cell was moving uni-directionally (Figure 2A). However, in reversing cells (Figure 2B), YFP–MglA+ localized dynamically: 3–4 min before a reversal, the fluorescence signal at the leading pole decreased and the cytoplasmic signal increased (Figure 2B″); subsequently, YFP–MglA+ accumulated at the lagging pole (Figure 2B′–B″). Once the YFP–MglA+ cluster had formed at the old lagging pole, a cellular reversal occurred (Figure 2B′–B″). This dynamic YFP–MglA+ localization was also observed during reversals in the presence of 25 μg ml–1 of the translation inhibitor chloramphenicol (data not shown). We conclude that YFP–MglA+ localization is dynamic. In all cells observed, the pole-to-pole relocation of YFP–MglA+ occurred within a 3–4 min interval and was initiated 3–4 min before a reversal and with the actual reversal, that is the change in the direction of movement, correlating with accumulation of YFP–MglA+ at the old lagging pole. A reversal occurs in two steps, first a cell stops for seconds to minutes and then the cell reverses. Depending on the duration of a stop, YFP–MglA+ relocation coincides partially or fully with the stop. We conclude that YFP–MglA+ localization correlates with uni-directional movement (leading pole cluster) and reversals (completion of transfer between poles).
To test whether the nucleotide-bound state of MglA is critical for localization, we generated strains synthesizing YFP–MglAG21V or YFP–MglAT26/27N at native levels (Supplementary Figure S2) as described for YFP–MglA+. The mutant MglA proteins fused to YFP exhibited the same characteristics as the mutant proteins without YFP (Table Ic). Moving cells containing YFP–MglAG21V (Figure 2C) reversed in a highly regular manner approximately every 4 min. YFP–MglAG21V formed a cluster that was continuously and regularly oscillating from the leading towards the lagging pole with a velocity of 1.3±0.2 μm min–1, each ‘arrival' of the cluster at the lagging cell pole correlating with reversal. Remarkably, after ‘arriving' at the lagging pole, the YFP–MglAG21V cluster immediately initiated the relocation towards the new lagging pole. Thus, in contrast to YFP–MglA+, YFP–MglAG21V did not form a stationary cluster at the leading pole, but continuously oscillated between the poles with a half-period of approximately 4 min. Inspection of YFP–MglAG21V localization in stalled cells (Figure 2D–D′) verified that the YFP–MglAG21V cluster relocated from pole-to-pole within a 4 min interval as opposed to remaining at a fixed position within a cell while the cell was moving.
The YFP–MglAT26/27N fluorescence signal was homogenously distributed throughout the cytoplasm (Figure 2E). The signal failed to segregate to the poles and localization did not change over time. The localization, the biochemical characteristics, and the in vivo activities of the three MglA proteins suggest that MglA localizing in a cluster at the leading pole as well as relocating MglA is MglA in the GTP-bound state and that the inactive form localizes diffusely to the cytoplasm.
As MglA acts downstream of Frz to induce reversals and the accumulation of MglA in a cluster at the old lagging pole correlates with reversal, we hypothesized that Frz stimulates the dynamic localization of MglA. To test this idea, we analysed the localization of YFP–MglA+ and YFP–MglAG21V in frzlof and frzgof mutants. In the non-reversing frzlof mutant, YFP–MglA+ localized in a cluster at the leading pole and did not display dynamic localization (Figure 3A). In the hyper-reversing frzgof mutant, YFP–MglA+ also accumulated in a cluster at the leading pole between reversals, and during the frequent reversals, the cluster relocated between the poles (Figure 3B–B′). In the frzlof mutant, the same localization pattern of YFP–MglAG21V was observed as in the frz+ strain and with YFP–MglAG21V forming a cluster that oscillated regularly between the poles with a 4 min half-period, each ‘arrival' of the cluster at the lagging pole correlating with reversal (Figure 3C–C′; cf. Figure 1C′). A similar localization of YFP–MglAG21V was observed in the frzgof mutant (data not shown). Thus, the Frz system is not necessary for polar localization of MglA+, but it is necessary and sufficient for pole-to-pole relocation of MglA+ during reversals. The two separable activities of MglA, their differential dependency on Frz, and the localization patterns of the mutant MglA proteins suggest that Frz by further stimulating accumulation of MglA/GTP induces the release of MglA/GTP from the leading pole and the subsequent relocation towards the lagging pole.
To examine whether MglB is involved in the regulation of MglA activity, we generated a ΔmglB mutant. This mutant accumulated normal levels of MglA (data not shown) and moved with normal velocity, but reversed two-fold more frequently than wt (Table Id). Furthermore, a fully active MglB–YFP protein (Table Id) accumulating at a level similar to that of MglB in wt cells (Supplementary Figure S3) localized in a cluster at the lagging pole in cells moving uni-directionally (Figure 4A). In reversing cells, the MglB–YFP cluster relocated from the old lagging to the new lagging cell pole in parallel with a cellular reversal (Figure 4B). In contrast to YFP–MglA+ (Figure 2B″), this relocation occurred during the reversal (Figure 4B″). Dynamic MglB–YFP localization was also observed during reversals in the presence of 25 μg ml–1 chloramphenicol (data not shown). Thus, MglB–YFP localization is dynamic during reversals, but relocation is initiated significantly later than in the case of MglA+.
To test whether MglA is involved in MglB localization, we analysed MglB–YFP in a ΔmglA mutant. In these non-moving cells (N=100), 84% contained MglB–YFP in bipolar, symmetric clusters and 16% contained a single MglB–YFP cluster (Figure 4C). Thus, compared with the localization of MglB–YFP in stalled mglA+ cells, MglB–YFP localization was shifted from a unipolar to a bipolar and symmetric pattern in the ΔmglA cells. Next, we analysed YFP–MglA+ localization in the ΔmglB mutant. In this mutant, YFP–MglA+ formed essentially symmetric and bipolar clusters (Figure 4D–D″) and generally with the larger cluster at the leading pole (Figure 4D′). Moreover, the larger YFP–MglA+ cluster relocated between the poles during reversals (Figure 4D′–D″).
Localization of MglA in the absence of MglB and MglB in the absence of MglA is shifted towards bipolar and symmetric suggesting that a mutually exclusive mechanism maintains MglA and MglB at opposite poles. As our data suggest that MglA localizing in a cluster at the leading pole is MglA/GTP and that the inactive form localizes diffusely to the cytoplasm, one mechanism by which MglB could inhibit MglA accumulation at the lagging pole would be to act as an MglAGAP.
To determine biochemically the function of MglB, we initially tested whether MglB is a nucleotide-binding protein and found that MglB neither binds mGDP/mGTP nor mADP/mATP (data not shown). The binding between a Ras-like G-protein and an interaction partner can be measured using fluorescent changes of mant-labelled nucleotides as shown for Ras and its GAP neurofibromin (Mittal et al, 1996; Ahmadian et al, 1997; Gremer et al, 2008) or Arl3 and its GAP RP2 (Veltel et al, 2008). A significant increase in fluorescence polarization was obtained on titrating MglB to MglA+/mGppNHp, but not to MglA+/mGDP (Figure 5A). The affinity to MglA+/mGppNHp was 60-fold higher (Figure 5A) compared with MglA+/GDP strongly suggesting that MglB senses and specifically binds to the active, GTP-bound conformation of MglA.
Ras-like proteins with bound GDP bind the transition state analogue aluminium fluoride (AlFx) only in the presence of a GAP (Mittal et al, 1996; Daumke et al, 2004; Gremer et al, 2008; Veltel et al, 2008). We used two approaches to test whether MglB can form a complex with MglA+/GDP in the presence of AlFx, as this can be considered the litmus test of whether a protein acts as a GAP (Gasper et al, 2009). First, as shown in Figure 5A, a significant increase in the polarization signal was detected with MglA+/mGDP in the presence of AlFx and MglB. Second, using size exclusion chromatography, we found that MglA+ elutes as a monomer at an elution volume corresponding to an apparent molecular mass of 22 kDa irrespective of the nucleotide-bound state, whereas MglB (17 kDa) elutes at a volume suggesting that it is a dimer or tetramer (Figure 5B, upper panel). No complex formation between MglB and MglA+ in its GDP-bound conformation was observed (Figure 5B; lower panel). However, addition of AlFx to this mixture resulted in complex formation as the proteins co-eluted in a single peak. Likewise, MglA+ in its mGppNHp-bound conformation co-eluted with MglB (Figure 5B, lower panel). These results strongly suggest that MglB is MglA's cognate GAP and has the capability of stabilizing the transition state of the GTPase reaction.
The GTPase reaction of a Ras-like G-protein is usually very slow and is accelerated in the presence of its cognate GAP, for example the slow intrinsic GTPase reaction of Ras is accelerated approximately 105-fold by RasGAP (Ahmadian et al, 1997), as is the GTPase activity of Rap by its cognate RapGAP (Daumke et al, 2004). To test whether MglB accelerates the GTPase reaction of MglA, we analysed GTP hydrolysis in the presence and absence of MglB. The slow intrinsic GTP-hydrolysis rate of MglA+ of 0.00009 s−1 was accelerated by catalytic amounts of MglB to 0.00027 s−1 and more than 100-fold by stoichiometric amounts of MglB (Figure 5C), confirming that MglB is an MglAGAP. In contrast, MglAG21V and MglAT26/27N showed no measurable GAP-stimulated GTP-hydrolysis with catalytic amounts of MglB (Figure 5C) and only a slight increase with stoichiometric amounts of MglB (data not shown). Moreover, binding of MglB to MglAG21V and MglAT26/27N in their active, GTP-bound conformation could still be detected (data not shown). Thus, we conclude that the substitutions in MglA inhibit the MglB-stimulated GTP hydrolysis rather than MglB binding showing the similarity to the Ras–RasGAP system, in which the G12V substitution inhibits GAP-mediated GTP hydrolysis, but not GAP binding (Gremer et al, 2008).
The earlier analyses suggested that MglA activity is temporally and spatially regulated by Frz and MglB. To identify downstream targets of MglA, we examined whether MglA activity has a function in the localization of three motility proteins, RomR, AglZ, and PilT.
In mglA+ cells, RomR–GFP localizes with a large cluster at the lagging pole and a small cluster at the leading pole (Figure 6A; Supplementary Figure S4). In parallel with a reversal, the large RomR–GFP cluster relocates from the old lagging to the new lagging pole. In the mglAG21V mutant, RomR–GFP localized in a bipolar, symmetric pattern (Figure 6B; Supplementary Figure S4). Moreover, during the frequent reversals, the symmetric localization of RomR–GFP did not change. In the mglAT26/27N mutant (N=100), RomR–GFP failed to localize at both poles. Rather in 86% of cells, RomR–GFP localized only at one pole. By staining of T4P, this pole was identified as the pole containing T4P, that is the ‘wrong' pole (Figure 6C), whereas the remaining 14% had RomR–GFP at the ‘correct' pole. In addition, RomR–GFP did not relocate between the poles in this mutant. Thus, MglA activity is important for establishing correct RomR polarity.
AglZ–YFP localizes in a large cluster at the leading pole and in small clusters along the cell body in mglA+ cells (Figure 6D; Supplementary Figure S5). In parallel with a reversal, the polar AglZ–YFP cluster relocated. In mglAG21V cells, AglZ–YFP localized as in mglA+ cells and the frequent reversals were accompanied by relocation of the large cluster (Figure 6E; Supplementary Figure S5). In the non-moving mglAT26/27N cells (N=100), 69% of cells displayed a homogeneous distribution of AglZ–YFP, 31% contained AglZ–YFP in a cluster at one pole, and no cells contained small clusters along the cell body (Figure 6F). This distribution is similar to the localization of AglZ–YFP in stalled wt cells (cf. Figure 6F). To resolve at which pole AglZ–YFP localizes in mglAT26/27N cells, we attempted to co-visualize T4P and AglZ–YFP in the same cells. However, under the conditions used for visualizing T4P, AglZ–YFP fails to localize to a pole (data not shown). Therefore, we co-visualized AglZ–GFP together with RomR–mDsRed in the same cells. In mglA+ cells, RomR–DsRed and AglZ–YFP localized to opposite pole, whereas in mglAT26/27N cells both proteins localized to the same pole (Figure 6G). Therefore, AglZ–YFP localizes correctly at the pole containing T4P in the mglAT26/27N mutant. Thus, MglA activity is not important for establishing correct AglZ polarity, but MglA regulates the dynamic localization of AglZ.
YFP–PilT localized with a large cluster at the lagging pole (Figure 6H; Supplementary Figure S6) and with a small cluster occasionally accumulating at the leading pole in mglA+ cells. In parallel with a reversal, the large YFP–PilT cluster relocated. In mglAG21V cells, YFP–PilT localized as in mglA+ cells and the frequent reversals were accompanied by relocation of the large YFP–PilT cluster (Figure 6I; Supplementary Figure S6). In the non-moving mglAT26/27N cells (N=100), YFP–PilT localization was shifted to a more unipolar pattern compared with the bipolar, symmetric pattern in stalled mglA+ cells (Figure 6J). By staining of T4P, this pole was identified as the pole not containing T4P (data not shown). Thus, MglA activity is important for establishing correct PilT polarity and MglA regulates dynamic localization of PilT.
In this study, we provide evidence that MglA is a Ras-like G-protein that functions as a nucleotide-dependent molecular switch to establish the correct polarity of motility proteins and to inverse their polarity during cellular reversals. Moreover, we show that MglA activity is regulated spatially and temporally by the MglAGAP MglB and the Frz system. According to the eukaryotic paradigm for Ras-like proteins, these proteins interact with three types of proteins: GEF and GAP, which have regulatory functions, and effectors that are components of downstream pathways involved in eliciting a particular response. Below we discuss our findings within the framework of this paradigm.
Genetic analyses showed that MglA/GTP represents the active form of MglA, suggesting that as in other G-proteins, MglA/GDP is the inactive form. MglA/GTP has two activities, stimulation of motility and stimulation of reversals. Two lines of evidence establish a link between MglA/GTP levels and MglA/GTP function. First, we directly showed that MglB functions as an MglAGAP, which converts the active MglA/GTP to inactive MglA/GDP, and an mglB mutant, which is predicted to accumulate increased amounts of MglA/GTP, moves with a normal velocity, but reverses more frequently than mglB+ cells. Second, the mutant protein MglAG21V, which is locked in the GTP-bound form, stimulates motility normally, but bypasses the requirement for Frz for reversals. On the basis of these observations, we propose that MglA/GTP functions in a concentration-dependent manner and that a low concentration of MglA/GTP stimulates motility and a high concentration stimulates reversals (Figure 7A).
By analysing the localization of three MglA proteins, we established a direct link between MglA/GTP localization and function. Between reversals, MglA/GTP localizes to a cluster at the leading pole, whereas the inactive form localizes diffusely to the cytoplasm. Several minutes before a reversal and in response to Frz activity, MglA/GTP is released from the leading pole and relocates to the lagging pole. On accumulation of MglA/GTP at the lagging pole, a reversal is induced. Together with our genetic analyses, these findings suggest that Frz activity further stimulates MglA/GTP accumulation at the leading pole. Once a threshold is reached, MglA/GTP is released from this pole.
Although the primary function of Frz is to induce the relocation of MglA, the primary function of MglB is in establishing spatial MglA/GTP asymmetry. Between reversals, MglA/GTP and MglB localize to opposite poles. However, each protein localizes in a bipolar and symmetric pattern in the absence of the other suggesting that the localization at opposite poles depends on a mutually exclusive mechanism. As we directly showed that MglB has MglAGAP activity, MglB likely excludes MglA/GTP from the lagging pole between reversals by locally converting MglA/GTP to MglA/GDP. How MglA/GTP excludes MglB from the leading pole remains to be clarified.
During reversals, the polarity of MglA/GTP and MglB is inverted. MglA/GTP relocation from the leading pole is initiated several minutes before a reversal (cf. Figure 2B), whereas MglB relocation from the lagging pole is initiated during the reversal (cf. Figure 4B). We infer that during a reversal, MglA/GTP ‘arriving' at the lagging pole at a high concentration interacts briefly with MglB causing a stimulation of MglA GTPase activity. As this results in a decrease in the MglA/GTP concentration, MglA/GTP can form a cluster at this pole. In parallel—and by an unknown mechanism—MglA/GTP excludes MglB from this pole causing its relocation to the new lagging pole. On the basis of the data showing that MglB is an MglAGAP at the lagging pole and that MglA/GTP relocation is initiated earlier than in the case of MglB, we suggest that the output of the Frz system is to directly or indirectly act as a GEF at the leading pole by converting MglA/GDP in a cytoplasmic pool to MglA/GTP. In current models of the Frz system, the output is the phosphorylated response regulator FrzZ (Inclan et al, 2007). We have been unable to detect direct interactions between MglA proteins or MglB and FrzZ, respectively (data not shown). Likewise, we have been unable to detect specific localization of FrzZ~P (data not shown). Although these results are all negative, they indicate that FrzZ~P may not act directly on MglA/GDP, but indirectly by stimulating an unknown GEF.
In total, we suggest that at the core of the regulatory circuit that sets up the leading/lagging polarity axis with MglA at the leading pole and MglB at the lagging pole is the mutually exclusive localization of MglA/GTP and MglB (Figure 7B). The series of events set in motion by Frz activity results in inversion of the leading/lagging polarity axis, and as a consequence, a cellular reversal ensues. Thus, the mutually exclusive localization of MglA and MglB sets up a stable polarity axis, which can be dynamically inverted in response to Frz activity. The asymmetric distribution of MglA and its cognate GAP is similar to that observed in eukaryotic systems involved in regulation of cell polarity, for example in the regulation of Ras by RasGAP in D. discoideum (Zhang et al, 2008), Cdc42 by Rga1 in S. cerevisiae (Tong et al, 2007) and Rho by RhoGAP in Caenorhabditis elegans (Anderson et al, 2008). Interestingly, in the latter case, this asymmetry also depends on a mutually exclusive mechanism. Although our model explains the basic circuit underlying the leading/lagging polarity axis and how it can be inverted, there are several unresolved questions. For instance, how does MglA/GTP exclude MglB from the leading pole? How does polar and dynamic MglA/GTP localization depend on concentration? What are the molecular nature of the landmarks recognized by MglA and MglB at the cell poles? Likewise, we do not know whether there is a GEF acting on MglA/GDP that regulate the accumulation of MglA/GTP between reversals. We are currently addressing these questions experimentally.
Analyses of the order of events during the pole-to-pole relocation of AglZ, FrzS, PilT, and RomR have shown that completion of relocation of the large polar clusters coincides with the reversal. The early release of MglA and the late release of AglZ, FrzS, PilT, and RomR suggested that MglA regulates the localization of motility proteins. Consistently, we found that MglA activity is required to establish the correct polarity of RomR and PilT, whereas correct AglZ polarity is established independently of MglA. This was a surprising observation given that AglZ and MglA interact directly (Yang et al, 2004; Mauriello et al, 2010). MglA locked in GTP-bound state induces frequent pole-to-pole relocation of AglZ and PilT and loss of asymmetric RomR localization. These data suggest that the two separable functions of MglA, that is stimulation of motility and stimulation of reversals, are manifested in these two distinct activities, establishment of polarity of some motility proteins and regulation of their dynamic localization. Accordingly, MglA is not part of the motility machineries and not a motility protein per se. Rather the motility defects in ΔmglA and mglAT26/27N mutants are caused by incorrect localization of motility proteins. The continuous oscillations of MglAG21V without the formation of a stationary cluster at the leading pole as well as the observation that MglA+ is released from the leading pole 3–4 min before the actual reversal and that cells continue to move in these 3–4 min support the idea that MglA is not a component of the motility machineries. The correct polarity of AglZ in the absence of MglA activity suggests that other mechanisms are also involved in establishing polarity of motility proteins. So far, the localization of five dynamically localized motility proteins has been shown to depend on Frz. We suggest that dynamic localization of motility proteins depend on MglA. Such a ‘one-for-all' mechanism would ensure that once established, the correct polarity of dynamically localized proteins is maintained over time guaranteeing that the two motility systems generate force in the same direction.
We currently do not know which effectors MglA interacts with to establish the correct polarity and induce pole-to-pole relocation of motility proteins. However, our data allow some speculations. PilT and RomR both localize with a large cluster at the lagging pole. As the large MglA+ cluster is at the leading pole, this suggests that MglA does not establish the correct polarity of these proteins by recruiting them to the lagging pole. Likewise, the accumulation of MglA/GTP at the lagging pole during a reversal induces the relocation of proteins from the leading to the lagging pole and vice versa. It is difficult to imagine how this would happen by direct interactions only. Recently, Mauriello et al (2010) observed that AglZ and FrzS cluster formation depends on the actin-like protein MreB, which forms a helix spanning the length of a cell in M. xanthus. Thus, it is a possibility that MglA regulates the polarity of MreB—or some other cytoskeletal element—to direct motility proteins in a vectorial manner to the correct poles. According to this scheme, relocation of MglA could result in an inversion of the polarity of MreB—or some other cytoskeletal element—and in this way, induce an inversion of the polarity of motility proteins. In this regard, it is interesting that directed motility of eukaryotic cells requires the polarized activity of Ras-like G-proteins of the Rho subfamily, which directly regulate re-organization of the actin cytoskeleton (Etienne-Manneville, 2004; Charest and Firtel, 2007). It is also interesting to note that translocation of MglA+ and MglAG21V from the leading to the lagging pole occurs in a highly directed manner (cf. Figure 2B–D) unlike that expected for freely diffusing molecules and on a timescale much slower than diffusion (Bulyha et al, 2009). In the absence of molecular motors such as kinesin, dynein, and myosin in bacteria, this relocation dynamics suggest that a cytoskeletal element pushes or pulls MglA from the leading to the lagging cell pole.
Each of the five major eukaryotic subfamilies of Ras-like G-proteins has its own cognate GEFs and GAPs with little if any sequence homology between GEFs and GAPs of different subfamilies (Vetter and Wittinghofer, 2001; Bos et al, 2007). GEFs and GAPs have in common that they are usually large multidomain proteins (Bos et al, 2007). MglB is a small protein of 17 kDa and does not show homology to known GAPs suggesting that it represents a novel GAP family. MglB is a member of the Roadblock/LC7 protein family implicated in regulation of NTPase activity and with members in all three domains of life (Koonin and Aravind, 2000). Thus, MglB GAP activity may be widely conserved.
To our knowledge, this is the first report showing a function of a Ras-like G-protein in establishing and maintaining cell polarity in bacteria. The finding that MglA functions and is regulated in a manner that shares many—if not all—of its characteristics with its eukaryotic counterparts provides evidence that the function of this family of proteins in regulation of polarity can be extended to include bacteria. MglA and MglB orthologues are found in many phylogenetically distant bacteria and archaea (Koonin and Aravind, 2000), suggesting that regulation of polarity by a Ras-like G-protein and its cognate GAP is a general feature in bacteria.
For microscopy, M. xanthus cells were grown and treated for time-lapse microscopy as described earlier (Leonardy et al, 2007; Bulyha et al, 2009). Cells were imaged at 30 s intervals for 10 to 15 min, and images were recorded and processed with Leica FW4000 V1.2.1 or Image Pro 6.2 (MediaCybernetics) software. Processed images were visualized in Metamorph 7.0r2 software (Molecular Devices). Quantification of fluorescence was performed as follows. Integrated fluorescence intensities (arbitrary units) of polar clusters and the cytoplasmic region between the polar clusters (Figure 2A′″) were quantified using the region measurement tool in Metamorph 7.0r2. Relative fluorescence intensities were calculated by dividing the integrated fluorescence intensities with the total fluorescent intensity of a cell. The linescan tool in Metamorph 7.0r2 was used to determine the position of the maximum fluorescence signal in a cell. This position was plotted as a function of time. T4P were visualized as described (Leonardy et al, 2007). To calculate reversal periods, the total number of moving cells was multiplied by the elapsed time and divided by the number of reversals. The velocity of moving cells was determined using the object track tool in Metamorph. Unless otherwise stated, 25 cells were analysed per experiment.
MglA proteins were made nucleotide free before determination of nucleotide affinities (Supplementary data). Fluorescence and polarization data were recorded with a Fluoromax-2 spectrophotometer (Jobin Yvon, Grasbrunn, Germany), with excitation and emission wavelengths of mant-nucleotides at 366 and 450 nm, respectively. GDP- and GTP-binding affinities of MglA+, MglAG21V, and MglAT26/27N were determined by titrating nucleotide-free MglA+, MglAG21V, or MglAT26/27N to 0.1 μM mGDP or mGTP or mGppNHp (Pharma Waldhof) at 37°C in Buffer M (50 mM Tris pH 7.5, 50 mM NaCl, 1 mM MgCl2, 1 mM DTE, 5% glycerol) monitoring the change in fluorescence. Obtained data points were fitted to a first-order reaction using Grafit5 (Erithacus software) to obtain the dissociation constant, Kd.
This was performed as described (Brinkmann et al, 2002). Briefly, a mix of 1 μM GTP and 60 nM [γ-32P]GTP in Buffer M was supplemented with 4 μM nucleotide-free MglA+, MglAG21V, or MglAT26/27N to start the intrinsic GTPase reaction at 25°C. For investigation of MglB-stimulated GTP-hydrolysis, catalytic amounts (0.05 μM) or stoichiometric amounts (4 μM) of MglB were added to start the reaction. Aliquots of 10 μl were taken at certain time points and mixed with 400 μl of charcoal solution (50 g l−1 charcoal in 20 mM phosphoric acid) to stop the reaction. The charcoal was pelleted and the amount of free 32Pi in the supernatant determined by scintillation counting. Data points were fitted to a first-order reaction to obtain kobs.
Binding affinities of MglB to MglA+, MglAG21V, and MglAT26/27N loaded with mGDP, mGppNHp, or mGDP/AIFx (Supplementary data) were determined by polarization measurements at 37°C in Buffer M. For this, 1 μM MglA loaded with the respective nucleotide was titrated with increasing amounts of MglB and the change in polarization signal monitored.
Complex formation was investigated by analytical gel filtration using a Superdex75 10/300 (GE Healthcare). A total of 1 mg MglA+ bound to GDP or GppNHp or GDP/2 mM AlFx in the presence or absence of 1.5 mg MglB were incubated 15 min at room temperature, applied and eluted with one column volume of Buffer M. The elution profile was recorded and eluted fractions analysed by SDS–PAGE.
We thank Hans-Ulrich Mösch, Peter Lenz, Eriko Takano and Anke Treuner-Lange for many discussions and practical help. The Max Planck Society, the graduate program ‘Intra- and intercellular transport and communication' funded by the German Research Foundation, and the LOEWE Research Center for Synthetic Microbiology supported this work.
The authors declare that they have no conflict of interest.