ABSTRACT
Bacterial cell division and peptidoglycan (PG) synthesis are orchestrated by the coordinated dynamic movement of essential protein complexes. Recent studies show that bidirectional treadmilling of FtsZ filaments/bundles is tightly coupled to and limiting for both septal PG synthesis and septum closure in some bacteria, but not in others. Here we report the dynamics of FtsZ movement leading to septal and equatorial ring formation in the ovoid-shaped pathogen, Streptococcus pneumoniae (Spn). Conventional and single-molecule total internal reflection fluorescence microscopy (TIRFm) showed that nascent rings of FtsZ and its anchoring and stabilizing proteins FtsA and EzrA move out from mature septal rings coincident with MapZ rings early in cell division. This mode of continuous nascent ring movement contrasts with a failsafe streaming mechanism of FtsZ/FtsA/EzrA observed in a ΔmapZ mutant and another Streptococcus species. This analysis also provides several parameters of FtsZ treadmilling in nascent and mature rings, including treadmilling velocity in wild-type cells and ftsZ(GTPase) mutants, lifetimes of FtsZ subunits in filaments and of entire FtsZ filaments/bundles, and the processivity length of treadmilling of FtsZ filament/bundles. In addition, we delineated the motion of the septal PBP2x transpeptidase and its FtsW glycosyl transferase binding partner relative to FtsZ treadmilling in Spn cells. Five lines of evidence support the conclusion that movement of the bPBP2x:FtsW complex in septa depends on PG synthesis and not on FtsZ treadmilling. Together, these results support a model in which FtsZ dynamics and associations organize and distribute septal PG synthesis, but do not control its rate in Spn.
Significance This study answers two long-standing questions about FtsZ dynamics and its relationship to septal PG synthesis in Spn for the first time. In previous models, FtsZ concertedly moves from midcell septa to MapZ rings that have reached the equators of daughter cells. Instead, the results presented here show that FtsZ, FtsA, and EzrA filaments/bundles move continuously out from early septa as part of MapZ rings. In addition, this study establishes that the movement of bPBP2x:FtsW complexes in septal PG synthesis depends on and likely mirrors new PG synthesis and is not correlated with the treadmilling of FtsZ filaments/bundles. These findings are consistent with a mechanism where septal FtsZ rings organize directional movement of bPBP2x:FtsW complexes dependent on PG substrate availability.
Introduction
Cell division in most bacteria is mediated by the tubulin homolog, FtsZ, which polymerizes into dynamic filaments and bundles at the middle or toward the pole of dividing cells (1, 2). Polymerization of FtsZ filaments/bundles initiates sequential binding of a series of proteins that ultimately assemble into a controlled divisome machine for septal peptidoglycan (PG) synthesis leading to cell division (3, 4). The assembly of this machine involves the binding of FtsZ filaments to membrane-anchoring and filament-stabilizing and –bundling proteins (1). An ensemble of conserved FtsZ-ring component and regulator proteins then interact sequentially followed by a class B penicillin-binding protein (bPBP; transpeptidase (TP) of PG peptides), FtsW (glycosyl transferase (GT) that builds glycan chains), and MurJ (Lipid II substrate flippase) (5–7). The exact composition of FtsZ-ring divisomes, the mechanism of timing and triggering septal PG synthesis, and the involvement of PG remodeling by hydrolases is only partly understood and varies widely among different bacterial species (8–10).
Biochemical work demonstrates that FtsZ filaments move directionally by a treadmilling mechanism, similar to that first found for eukaryotic tubulins (11). In treadmilling, FtsZ-GTP monomers add to the growing (+) end of advancing filaments, and FtsZ-GDP, produced by FtsZ-catalyzed GTP hydrolysis, dissociate from the other disappearing (−) end of the FtsZ filament (12). The net result is directional movement of filaments, in which central FtsZ-GTP subunits are stationary. Total internal reflection microscopy (TIRFm) of moving FtsZ filaments/bundles and single-molecule (SM)-TIRFm experiments have recently established FtsZ treadmilling in the Gram-positive and –negative rod-shaped bacteria, Bacillus subtilis (Bsu) and Escherichia coli(Eco), respectively (12, 13). In these studies, the velocity of filaments/bundles was shown to depend on the FtsZ GTPase activity, but was independent of the addition of TP-inhibiting antibiotics. Both studies also concluded that bidirectional treadmilling of FtsZ filaments/bundles plays a role in organizing and distributing the septal PG synthesis apparatus. In Bsu, treadmilling is tightly coupled to and limiting for septal PG synthesis and septum closure, such that the velocity of septal bPBP2b movement correlates with the velocity of treadmilling of FtsZ filaments/bundles (12). This mode of PBP movement differs from that of MreB-mediated side-wall elongation that depends on PG synthesis and is blocked by antibiotics in Bsu and other rod-shaped bacteria (14, 15). Likewise, the velocities of bPBP3 (FtsI) and FtsZ treadmilling are correlated in Eco, but curiously, treadmilling velocity does not limit the rate of septal PG synthesis determined by incorporation of fluorescent D-amino acids (FDAAs) or the rate of septum closure (13). In contrast, after septal PG synthesis is initiated in Staphylococcus aureus (Sau), cytokinesis to close the septum does not depend on FtsZ treadmilling and is likely driven by PG synthesis (16).
Compared to these model rod-shaped and spherical bacteria, much less is known about FtsZ ring dynamics in ovoid-shaped bacteria, such as the human respiratory pathogen, Streptococcus pneumoniae(pneumococcus; Spn). Newly divided ovococcus bacteria form prolate ellipsoid-shaped cells containing equatorial rings composed of FtsZ and other proteins (bottom, Fig. S1A) (17, 18). These equatorial rings become the mature septa at the start of division (bottom, Fig. S1A) (19, 20). Mature FtsZ rings contain all of the proteins required for the stabilization and placement of FtsZ protofilaments and for PG synthesis during the next round of division (21). Spn lacks conventional nucleoid occlusion mechanisms, and high-resolution microscopy shows that FtsZ protofilaments are distributed in nodal patterns around mature septal FtsZ rings that surround the undivided nucleoid marked by its origin of replication (bottom, Fig. S1A) (22–24).
To construct an ellipsoid shape, two modes of PG synthesis are organized by the septal FtsZ rings in Spn(25). Septal PG synthesis mediated by Class b PBP2x (bPBP2x) and other proteins closes inward to separate cells, whereas peripheral PG synthesis mediated by bPBP2b and other proteins emanates outward from midcells to elongate cells (top, Fig. S1A). Early in division, a ring composed of MapZ (LocZ) splits (Fig. S1B, top; Fig. 1A) and is moved by peripheral PG synthesis toward the equators of the daughter cells (26, 27), preceded by the origin of replication (top, Fig. S1A) (23). MapZ movement precedes migration of FtsZ, FtsA (FtsZ membrane anchor and peripheral PG regulator in Spn (20)), and EzrA (FtsZ assembly modulator in Bsu(28) and FtsZ assembly positive regulator in Spn (A. Perez, in preparation)) to the equators (Fig. 1B-1F). During middle to-late cell division, FtsZ, EzrA, and FtsA are observed at the closing septum as well as at both developing equators, resulting in a distinctive three-band pattern (Fig. S1A, middle; Fig. S1B bottom three panels, Fig. 1B-F). After FtsZ, EzrA, and FtsA relocate to equators, proteins involved in PG synthesis, including DivIVA (negative-curvature binding protein that determines cell shape) (29), MltG (endolytic transglycosylase in peripheral PG synthesis) (30), GpsB (regulator that distributes septal and peripheral PG synthesis) (31), StkP (Ser/Thr protein kinase that regulates PG synthesis) (32), and bPBP2x (19), remain at the septum, and migrate to equators right before cells divide (Fig. 1G-L).
Little is known about how FtsZ moves from the septum to the MapZ rings that have arrived near the equators of daughter cells. Current models postulate that FtsZ migrates en masse from the septum to the equatorial MapZ rings at a later stage in division (e.g., (23)). A recent study used TIRFm to demonstrate treadmilling of FtsZ filaments/bundles in equatorial rings of Streptococcus mutans (Smu) (33), which is evolutionarily distant from Spn(33). In this study, en masse streaming of FtsZ from septa to equatorial rings was detected in a minority (≈7%) of dividing S. mutans cells (33). Here, we report that key proteins involved in FtsZ ring assembly and in septal and peripheral PG synthesis have different dynamics during pneumococcal cell division. We demonstrate and describe several parameters of FtsZ treadmilling in Spn. Furthermore, we show for the first time that nascent rings containing FtsZ, FtsA, and EzrA move out from mature septa guided by MapZ throughout the cell cycle. Streaming of EzrA filaments was only observed in ΔmapZ mutants as a possible division failsafe mechanism. In contrast, several other proteins were confined to mature septa and showed little dynamic movement within the limits of conventional TIRFm. Finally, we show that bPBP2x interacts with FtsW and that both proteins show directional movement along mature septal rings, independent of FtsZ treadmilling. Together, these findings reveal new aspects about the movement and assembly of FtsZ/FtsA/EzrA filament/bundles in dividing Spn cells and show that septal bPBP2x:FtsW complexes require PG synthesis for movement.
Results
Relocation of Spn cell division and PG synthesis proteins occurs in three stages and is dependent on pH. To compare the dynamics of pneumococcal cell division and PG synthesis proteins, we constructed and vetted a large set of fluorescent- and HaloTag (HT) protein fusions expressed from single-copy genes at their native chromosome loci (Tables S1). Each protein fusion contains a linker region specified in Table S1, but omitted in the text and figures to simplify designations. An unencapsulated derivative (Δcps) of serotype 2 strain D39 was used for these studies, because encapsulated D39 forms short chains (Fig. S1A) that make microscopy more difficult, and capsule tends to mask morphology defects of constructs (34). None of the final fluorescent- and HT-protein fusions ostensibly altered growth or cell morphology, and each showed localization of labeled proteins at septa and new equators of dividing cells grown exponentially in C+Y liquid medium, pH 6.9 (5% CO2) (Fig. 1 and S2), consistent with previous localization studies (see below).
Demographs generated by MicrobeJ (35) from fields of exponentially growing cells (e.g., Fig. S2) supported and extended the conclusion that Spn division and PG synthesis proteins relocate from the septa of single, early divisional cells (left side of demographs) to the equators of new daughter cells (right side of demographs) in three distinct stages (Fig. 1 and S1). MapZ relocates early, before FtsZ, FtsA, and EzrA (23, 26, 27). Residual MapZ remained between new equatorial rings until the migration of FtsZ and its associated proteins, FtsA and EzrA (Fig. 1A-1F), but a third sepal ring of MapZ reported previously (26) was not detected in cells grown under these conditions (see also (23)). FtsZ, FtsA, and EzrA next relocate to new equators at approximately the same time, with residual EzrA and FtsA remaining at septa when most of FtsZ has migrated (Fig. 1B-1F). Other cell division and PG synthesis proteins, including DivIVA, MltG, GpsB, StkP, bPBP2x, and FtsW, remain at septa after most FtsZ, FtsA, and EzrA has departed and move to the equators of daughter cells late in the division cycle (Fig. 1G-1L and Fig. S2G) (21, 22, 29, 30). The localization of StkP, bPBP2x, and FtsW is more diffuse away from septal and equatorial rings than that of the other proteins examined throughout the cell cycle (Fig. 1J-1L and S2G). Western blot control experiments did not detect cleavage of the GFP or HT reporter domains from GFP-StkP and HT-bPBP2x (Fig. S4). As shown later, diffusiveness in demographs corresponds to diffuse movement detected by TIRFm.
During these experiments, we unexpectedly noticed that the size and shape of wild-type Δcps cells depends on pH in C+Y liquid medium. At pH ≈7.6 (5% CO2), which supports natural competence (36), pneumococcal cells are markedly longer and larger than at pH ≈6.9 (5% CO2), which is the physiological pH at the surface of epithelial cells in the human respiratory tract (Fig. S3A and S3B)(37). Underscoring the effects of higher pH, strains expressing GpsB-sfGFP or the GFP-StkP showed morphological defects characteristic of reduced GpsB or StkP function, respectively, in C+Y at pH 7.6, but not at pH 6.9 (Fig. S2 and S3C). Effects of pH on cell length and aspect ratio of wild-type cells were not observed in BHI broth (Fig. S3A), which we used in previous studies, but cannot be used here because of autofluorescence.
Dynamics of FtsZ in nascent rings that form parallel to mature FtsZ septal rings. TIRFm and epifluorescence microscopy showed that FtsZ filament motion was detected both inside and outside of mature septal rings in E. coli(13, 38) and B. subtilis cells (12). To determine the patterns of FtsZ movement in Spn cells, we performed comparable TIRFm, which limits illumination to an 100–150 nm slice and removes out-of-focus background fluorescence light (39). TIRFm of cells was performed on agarose pads containing C+Y, pH 7.1 (no CO2). Newly separated pneumococcal cells contain a mature midcell septal ring that appears as a prominent fluorescent band composed of multiple overlapping FtsZ filaments (Fig. 2A and 2B). FtsZ filament/bundle motion is detected by fluctuations in kymographs of TIRFm images (Fig. S5A and S5B), but it was not possible to reliably quantitate FtsZ filament/bundle velocities by TIRFm in densely packed mature septal rings (Fig. 2B and 2D; Movie S1). FtsZ filament/bundle speeds in mature septal rings were determined by wide-field imaging of vertically oriented cells as described below.
We detected the initial stages of formation of nascent FtsZ rings on either side of mature septal rings (Fig. 2A-2D; Movie S2). FtsZ in nascent rings was detected as oblong spots moving in both directions parallel to mature septal rings (B in Fig. 2A and 2B-2D; Movie S2). Nascent FtsZ rings first appear very close to mature septal rings, and this distance increases as the nascent FtsZ filaments move outward toward the equators of daughter cells, eventually resulting in the characteristic pattern of three parallel FtsZ rings in mid-to-late divisional Spn cells (Fig. 2A and 2D). Summations of TIRFm images taken over 180 s movies indicate that the diameters of nascent rings start out approximately equal to those of mature septal rings (Fig. 2A and 2D). Kymographs through the long axis of cells show that nascent FtsZ rings form asynchronously on both sides of mature septal rings, with one nascent FtsZ ring detected slightly before the other ≈50% of the time (e.g., Fig. 3A). Nascent FtsZ rings move away from septal rings, add more filaments/bundles, and develop into early equatorial rings, in which directional velocities of FtsZ filaments/bundles are still detected (A in Fig. 2A; Fig. 3B). The diameters of equatorial rings became larger than those of residual septal rings, and the number of overlapping FtsZ filaments/bundles within new equatorial rings continue to increase (Fig. 2A and 3A). As the density of FtsZ filaments/bundles increases in new equatorial rings, motion is indicated by fluctuations in TIRFm kymographs (longer times, Fig. 2D; Fig. S5; Movies S1-S2). We show below that there is a correspondence between the position of nascent rings of FtsZ, FtsA, and EzrA and movement of the MapZ protein ring out from mature septal rings to the new equatorial rings of daughter cells.
Velocities of Spn FtsZ filaments/bundles moving in either direction in nascent rings were determined from kymographs (Fig. 2A-2D). FtsZ filament velocities were similar in nascent (31.5 ± 13.0 (1.0) nm/s (SD (SEM)) and early equatorial rings (34.4 ± 13.7 (1.5) nm/s), with a combined average FtsZ filament velocity of 32.4 ± (13.3) (0.8) nm/s in cells in C+Y, pH 7.1 (no CO2) (Fig. 2A-E). FtsZ filament velocities were comparable in cells in C+Y, pH 7.8 medium (no CO2) (33.0 ± 0.9 (10.0 nm/s); Fig. S3D). The velocities of Spn FtsZ filaments are similar to those reported previously for FtsZ filament/bundle movement in septal rings of Eco (27.8 ± 17.1 (SD)) (13) and Bsu(32 ± 7.8 (SD)) (12). We also analyzed the time between FtsZ filament appearances moving in the same direction (Fig. 2F). The relative frequency of appearance of FtsZ filaments moving in the same direction for the most part followed a random distribution, except between 30–40 s (Fig. 2F). The diameter of FtsZ-sfGFP rings in these live pneumococcal cells was determined by 2D-deconvolution epifluorescence microscopy to be 0.80 ± 0.06 μm, which corresponds to a circumference of ≈ 2,500 nm. Thus, the frequency of FtsZ filament appearance at intervals of 30–40 s cannot be caused by circumferential periodicity of FtsZ filaments moving at ≈34 nm/s, but may be related to an average clocked initiation of new FtsZ filaments.
FtsZ filament/bundle dynamics and processivity in mature septal rings. To determine the speed, processivity, and lifetime of FtsZ filaments/bundles in mature septal rings, individual Spn cells expressing FtsZ-sfGFP were oriented vertically in a micro-hole device described previously (Fig. 4A and S6) (12). FtsZ movement in the imaging plane was recorded by wide field time-lapse microscopy of mature septal rings with a range of diameters (Fig.4A and S7; Movie S3). Images were de-noised, and kymographs were generated (Fig. S6) (12). Lengths and angles of ≈600 FtsZ filament tracks from 29 cells were quantitated and used to compute FtsZ filament/bundle speeds, processivity, and lifetimes (Fig. 4B-4E). FtsZ filament/bundles move bi-directionally around Spn mature septal rings at an average speed of 30.5 ± 9.3 nm/s, which is comparable to the average velocity of FtsZ filaments/bundles in nascent and new equatorial rings (32.4 ± 13.3 nm/s; Fig. 2E) and independent of cell diameter (Fig. S7B). We conclude that the dynamic properties of FtsZ filaments/bundles in nascent and early equatorial rings match those of FtsZ filaments/bundles in mature septal rings. We also measured the total distance travelled by FtsZ filaments/bundles within septa. This gives a processivity distribution with an average of 515 ± 331 nm (Fig. 4C and 4E), meaning that an FtsZ filament typically traverses about 1/5 of the circumference of a Spn cell. Related to processivity, the time that FtsZ filaments/bundles exist in tracks is distributed with an average of 17.1 ± 9.4 s (Fig. 4D and 4E).
Treadmilling of FtsZ filaments/bundles in mature and nascent rings and dependence of FtsZ filament velocity on GTP hydrolysis. Previous studies show that FtsZ filaments move by a treadmilling mechanism in Eco and Bsu (12, 13). To demonstrate treadmilling of FtsZ filaments/bundles in Spn mature and nascent rings (Fig. 2), we constructed a functional FtsZ-HT construct (Fig. 1C) in a strain expressing EzrA-mNeonGreen (mNG) (Fig. 1E). A limiting concentration of HT substrate was added to approach single-molecule detection of FtsZ-HT by TIRFm (red, Fig. 5) in cells whose outlines were delineated by bright-field microscopy (blue, Fig. 5). EzrA-mNG was used as a fiducial marker for the locations of rings (green, Fig. 5). As presented below, FtsZ and EzrA exhibit similar patterns of movement in nascent and mature rings. In mature septal, nascent, and equatorial rings in daughter cells, single molecules of FtsZ appear as stationary foci that persist before disappearing (red spots, Fig. 5; Movie S4). We interpret these transient, static foci of single FtsZ molecules as representing non-moving FtsZ molecules within the cores of FtsZ filaments/bundles that are translocating by a treadmilling mechanism (Fig. 5B). The average lifetime of FtsZ foci detected in mature and nascent rings was 11.9 ± 9.1 (SD) s, with some foci persisting for 15–20 s (Fig. 5C). The average length of a treadmilling filament is set by the subunit lifetime and average filament speed, since subunits will bind to the plus end of a filament, and then depolymerize from the minus end. Thus, the estimated FtsZ filament length is 386 ± 335 (SD) nm.
In addition, we confirmed that the velocity of Spn FtsZ filament/bundle movement depends on GTP hydrolysis by FtsZ, as reported previously for other bacteria and in biochemical reactions (12, 13, 40). For these experiments, we constructed a Spn mutant expressing FtsZ(G107S), which likely is defective in GTP binding based on homologues in other bacteria (Fig. S8A) (41). The ftsZ(G107S) mutant is temperature sensitive for growth and lyses at 42° C (Fig. S8B). Following a shift from 32° C to 42° C, the ftsZ(G107S) mutant formed larger, more spherical cells than the ftsZ+ parent strain (Fig. S5C and S5D), although the relative cellular amount of FtsZ(G107S) was comparable to that of FtsZ+ in cells at 42° C (Fig. S5E). Strains expressing FtsZ(G107S)-sfGFP are not viable. Therefore, we constructed a ftsZ(G107S)//bgaA::PZn-ftsZ-sfgfp merodiploid strain (Table S1), in which we expressed and tracked the movement of low levels of ectopically expressed FtsZ-sfGFP by adding limited concentrations (0.1/0.01 mM) of Zn2+/Mn2+ at the still-permissive temperature of 37° C (Fig. S9). Under these conditions, the ftsZ+//PZn-ftsZ-sfgfp and ftsZ(G107S)//PZn-ftsZ-sfgfp strains show overall similar growth (Fig. S9A) and FtsZ-sfGFP localization (Fig. S9B), although slightly aberrant cells with mislocalized FtsZ-sfGFP were occasionally observed for the ftsZ(G107S)//PZn-ftsZ-sfgfp strain. TIRFm of nascent FtsZ rings revealed that FtsZ-sfGFP filaments still move bidirectionally, but with significantly reduced velocity in the ftsZ(G107S) mutant compared to the ftsZ+ parent strain (Fig. S10; Movie S5). Overexpression of another mutant allele, FtsZ(D214A) that is defective in GTPase activity, also severely decreases FtsZ filament/bundle velocity (see below; Fig. S16A). We conclude that Spn FtsZ filament/bundle velocity produced by treadmilling is dependent on GTP binding and hydrolysis by FtsZ, consistent with previous studies in other bacteria (12, 13, 16).
FtsA and EzrA form nascent rings with FtsZ in Spn. We examined the movement of several proteins involved in FtsZ filament formation and stabilization (FtsA and EzrA) and in septal peptidoglycan synthesis and cell division (MapZ, GpsB, MltG, DivIVA, StkP, bPBP2x, and FtsW) (see Introduction). Of this set, only FtsA and EzrA localize with FtsZ throughout the Spn cell cycle (Fig. 1B-1F) and form nascent rings in early divisional cells (Fig. 6; Movies S6 and S7). In summations of TIRFm movies, FtsZ, FtsA, and EzrA localize distinctly in mature septal, nascent, and equatorial rings and are not detected elsewhere in Spn cells (Fig. 6A and 6B). Like FtsZ, FtsA and EzrA form nascent rings moving out from mature septa (compare Fig. 2 and 6). The average velocity of EzrA in nascent rings (29.6 ± 15.3 nm/s) (Movie S6) was the same as that of FtsZ filaments (Fig. 6C). Unexpectedly, Spn FtsA consistently moved about 25% faster (41.7 ± 16.2 nm/s) in nascent rings than FtsZ filaments or EzrA (Fig. 6C; Movie S7), which contrasts with Bsu FtsA that moves at a velocity similar to FtsZ (12). We attempted to determine whether this faster velocity of FtsA is a property of the FtsA’-sfGFP-FtsA’ sandwich fusion used; however, to date, all FtsA reporter fusions constructed show growth and morphology defects. Nevertheless, we conclude that the FtsA and EzrA proteins that associate with and stabilize FtsZ filaments throughout the Spn cell cycle have similar overall dynamics as FtsZ filaments/bundles, including nascent ring formation (Fig. 6).
MapZ location corresponds to positions of nascent FtsZ and EzrA rings in early divisional Spn cells. We wondered whether nascent ring formation of FtsZ, FtsA, and EzrA was coincident with movement of MapZ protein rings, which emerge from either side of mature septal rings concomitant with the start of peripheral PG synthesis (Fig. S1A) and move perpendicular to the long axis of cells to the equators of the new daughter cells (Fig. S1B and S11) (26, 27). In demographs and summations of movies, MapZ is localized primarily in mature septa or in two rings adjacent to septa, although a slight haze of MapZ remains between equatorial rings until FtsZ had fully exited from septa (Fig. 1A and S11C; Movie S8). No directional movement of MapZ or fluctuations of MapZ signal was observed in rings in kymographs (Fig. S11C and S11D), consistent with minimal MapZ movement reported previously for Smu MapZ (33). This conclusion was confirmed directly by SM-TIRFm of i-tag-HT-MapZ (iHT-MapZ), which unlike HT-MapZ, did not cause cell morphology defects (Fig. S11A and S11B)). Single molecules of iHT-MapZ that appeared in MapZ rings remained static for as long 60 s to >100 s before disappearing due to motion out of the TIRF plane or photobleaching (Fig. S11E and S11F; Movie S9).
In high-resolution 3D-SIM images of cells co-expressing tagged MapZ and FtsZ, low amounts of FtsZ are detected in early divisional cells at positions corresponding to nascent rings observed by TIRFm (Fig. 2; arrowhead, Fig. 7A (i) and (ii)). These nascent FtsZ rings overlap with MapZ rings moving away from septa. Likewise, EzrA in nascent rings overlaps with the parallel MapZ rings adjacent to the septum in early divisional cells (dotted box, Fig. 7A(iii)). In later divisional cells, EzrA remains at constricting septa surrounding segregating nucleoids, when all MapZ has moved to the equators of daughter cells, which also contain some EzrA (box, Fig. 7A(iv)). These results are consistent with MapZ acting as a guide for the nascent rings of FtsZ, FtsA, and EzrA that initially delivers some, but not all, of FtsZ, FtsA, and EzrA to the equators of daughter Spn cells.
If MapZ is a guide for formation of nascent rings, then we would expect aberrant movement of FtsZ/FtsA/EzrA filaments in ΔmapZ mutants. In the D39 Spn genetic background, ΔmapZ mutants are viable and form nearly normal looking cells with some distortions and frequent misaligned division planes (23, 42). However, FtsZ-GFP fusions in ΔmapZ mutants exhibit a severe synthetic defect in growth and morphology that precludes their study in Spn (23), but that was not commented upon in Smu (33). In contrast, EzrA-GFP fusions in ΔmapZ mutants lack this defect and appear similar to ΔmapZ mutants (Fig. 7B). TIRFm of EzrA-GFP movement in a ΔmapZ mutant indeed revealed aberrant, untimed streaming of EzrA filaments, presumably in association with FtsZ, from parent to daughter cells, often resulting in rings that are not perpendicular to the long axis of cells (arrows, Fig 7B). Nevertheless, the rate of EzrA streaming was similar in ΔmapZ and mapZ+ strains (Fig. 7C). Altogether, these results are consistent with MapZ acting as a continuous guide for the orderly movement of FtsZ/FtsA/EzrA filaments from mature septal rings to new equatorial rings in daughter cells. However, in the absence of MapZ, a second streaming mechanism aberrantly distributes FtsZ/FtsA/EzrA filaments into daughter cells. bPBP2x is dynamic compared to other proteins that mediate PG synthesis. We next examined the motion of several other proteins involved in PG synthesis in Spn. GpsB (regulator), DivIVA (regulator), MltG (endo-lytic transglycosylase), StkP (Ser/Thr kinase), bPBP2x (TP), and FtsW (GT) (see Introduction) remain at mature septa until late in the division cycle after FtsZ, FtsA, and EzrA have largely moved to the equatorial rings of daughter cells (Fig. 1G-1I; Fig. S2G). Unlike FtsZ, FtsA, and EzrA (Fig. 2, 3, and 6), TIRFm analysis did not detect GpsB, MltG, DivIVA, StkP, and bPBP2x joining nascent rings and showed that these proteins are largely confined to mature septal and equatorial rings (Fig. S12 and 8A; Movies S10-S14). In these mature septal and equatorial rings, fluctuation of GpsB or MltG signal indicative of ordered movement is not evident (Fig. S12A and S12B), whereas DivIVA and bPBP2x are actively moving, especially in equatorial rings (Fig. S12C and 8A), and bPBP2× motion is more diffusive around cells (Movie S14). By contrast, the motion of StkP is distinctively different from that of the other proteins examined. StkP locates in mature septal rings, where signal fluctuations are not readily apparent (Fig. S12D), but at the same time, StkP moves rapidly and diffusively throughout whole cells, which is captured as “clouds” of protein in summations of TIRFm movies (S12D; Movie S13). Cells expressing GFP-StkP or HT-bPBP2× did not show obvious defects in growth or cell morphology (Fig. 1 and S2), where HT-bPBP2x was labeled by an excess of HT-JF549 substrate. We conclude that of this set of proteins, only FtsZ and its ring stabilizers, FtsA and EzrA, form nascent rings and that dynamics of the other proteins varies from minimally detectable by TIRFm to diffusive. bPBP2x and its partner FtsW move at the same velocities along septal rings. SM-TIRFm experiments were performed to delineate the motion of bPBP2x relative to the FtsZ in mature septal rings (Fig. 8B; Movie S15). SM-TIRFm detection of HT-bPBP2× was approximated by addition of a limited amount of HT-JF549 substrate (red, Fig. 8B) that gave the same rate of bPBP2x movement when titrated downward to where labeled cells were barely detectable. FtsZ-sfGFP (green) and/or midcells determined from cell outlines (blue) were used as fiducial markers for the location of mature septal rings (Fig. 8A). Some bPBP2x molecules are detected as moving rapidly around cells in a sporadic fashion (Movie S15), consistent with TIRFm (Fig. 8B). The movement of these diffusive bPBP2x molecules was not analyzed further due to their lack of continuous tracks in SM-TIRFm. Other bPBP2x molecules attach onto mature septal rings and move directionally for at least 18 s (montage, Fig. 8B) and in some cases >30 s. The velocity of bPBP2x molecules in septal rings (21.9 ± 12.8 nm/s) is significantly slower than that of treadmilling by FtsZ filaments/bundles (32.4 ± 13.3 nm/s) (Fig. 8C). Control experiments showed that the velocity of single molecules of HT-bPBP2× is the same in strains that express FtsZ-sfGFP (Fig. 8C) or that express FtsZ+ (Fig. 8F).
To further demonstrate that the velocity of HT-bPBP2x is not dependent on the fusion construct, we tracked the velocity of FtsW-HT in Spn (Fig. 8F; Movie S16). New results demonstrate that the biochemical GT activity of FtsW depends on its interaction with its cognate Class B PBP (43). We confirmed this interaction in Spn cells by co-localization of FtsW and bPBP2x (Fig. 8D) and by co-immunoprecipitation of FtsW with bPBP2× as bait in a 1:1 complex based on molecular mass (Fig. 8E). Consistent with a complex of bPBP2×:FtsW, FtsW and bPBP2× move along mature septal rings at the same velocity (Fig. 8F; Movie S16), which is slower than that of FtsZ treadmilling.
Movement of bPBP2x and FtsW depends on PG synthesis and not FtsZ treadmilling in Spn. Last, we examined whether the velocity of bPBP2x and FtsW is strongly correlated with FtsZ treadmilling, as was demonstrated in Eco and Bsu (12, 13). We found that there is minimal correlation between the velocity of bPBP2x movement on septa and the rate of FtsZ treadmilling (Fig. 9A). To perform these experiments we determined the velocity of bPBP2x at septa by SM-TIRFm in FtsZ(GTPase) mutants that slowed down FtsZ treadmilling by ≈2X (FtsZ(G107S); Fig. S13 and S14; Movie S17) or ≈10X (overexpression of FtsZ(D214A); Fig. S15 and S16; Movies S18 and S19) and that lead to a percentage of cells with aberrantly placed division rings. Strikingly, reduction of FtsZ treadmilling velocity by ≈2X or ≈10X does not reduce bPBP2x velocity or reduces it only slightly (≈1.3X), respectively (Fig. 9A, S14B, and S17C). Notably, in the FtsZ(D214A) mutant, bPBP2x moves ≈5X faster than the FtsZ filaments/bundles. Likewise, FtsW velocity is only reduce by only ≈1.4X in the FtsZ(D214A) mutant (Fig. S17D). Finally, reduction of FtsZ treadmilling velocity over this range does not affect the net level of PG synthesis, as determined by incorporation of FDAA label for 2.5 m (Fig. 9B). These results contrast sharply with those for Bsu, where inhibition of FtsZ treadmilling significantly reduces FDAA labeling (12).
We next tested whether bPBP2×:FtsW velocity depends on PG synthesis. In Spn, there are two MurA (UDP-N-acetylglucosamine enolpyruvyl transferase) homologs that catalyze the first committed step of PG synthesis (44, 45). Deletion of murA1 (spd_0967; also called murZ) does not significantly alter growth, cell morphology, or FtsZ treadmilling velocity in C+Y, pH 7.1 (Fig. S18). However, the velocity of bPBP2x and FtsW is consistently reduced by ≈1.5X in the ΔmurA1 mutant compared to the murA1+ parent (Fig. 9C). We conclude that limitation of PG synthesis slows down bPBP2×:FtsW velocity without detectably affecting FtsZ filament/bundle velocity. Last, we added the β-lactam methicillin at a concentration that inhibits most of bPBP2x TP activity almost specifically (22). Methicillin addition did not inhibit FtsZ treadmilling velocity (Movie S20), but nearly completely stopped the movement of bPBP2x (Fig. 9D and S19; Movie S21). Together, these combined results indicate that movement of bPBP2×:FtsW complexes along septal rings depends on PG synthesis and is independent of the movement of FtsZ filaments/bundles.
Discussion
Partitioning of FtsZ filaments/bundles into daughter cells occurs in ovoid-shaped (ovococcus) bacteria, such as Spn, by a mechanism that is fundamentally different from the Min and/or nucleoid occlusion systems present in rod-shaped bacteria (46, 47). Early in division, a ring containing MapZ protein splits from and moves parallel to the initial septal ring toward the site of the new equators in the daughter cells (Fig. S1). Movement of MapZ rings is presumably driven by outward peripheral PG synthesis from the midcell septum and is preceded by movement of the origin of replication during chromosome segregation, which is promoted by transcription and unknown mechanisms (23, 48). Spn MapZ is a bitopic membrane protein, whose 40 amino-terminal amino acids of its cytoplasmic domain can bind to FtsZ and whose extracellular carboxyl-terminal domain binds to PG (26, 49). However, the amino-terminal FtsZ binding amino acids are not conserved in Smu MapZ (33). MapZ is phosphorylated by the StkP Ser/Thr kinase; but, a requirement for MapZ phosphorylation on cell division and morphology seems to depend on Spn genetic background (26, 27). Likewise, in one laboratory strain, MapZ forms a third ring at division septa (26), whereas in other laboratory strains and in the progenitor D39 background of most laboratory strains, this third ring is rarely detected (Fig. 1, S7A, and S1B) (23, 27), making it unlikely that it plays an obligatory role in Spn division.
The interesting conjecture was made that as MapZ reaches the equators of daughter cells, it serves as a “beacon” for re-localization of FtsZ from mature septal rings (26). A recent paper proposes a concerted streaming mechanism in which FtsZ moves late in division from septa to equators in some Smu cells (33). In contrast, here we report that FtsZ transport to equators in Spn is a continuous process throughout the cell cycle (Fig. 10). Early in Spn division, nascent filaments/bundles of FtsZ are detected near and moving parallel to mature septal rings (Fig. 2 and 10). In the ≈10 m interval (1/3 of a generation) between the initial movement of MapZ and the migration of most FtsZ to equators (Fig. 1A and 1B), nascent FtsZ filaments/bundles move outward and become more dense until they reach equators, after which the remainder of FtsZ migrates to form mature equatorial rings (Fig. 3). Progressive nascent ring formation was detected in both Δcps derivatives and in the progenitor encapsulated cps+ parent D39 strain. These nascent FtsZ rings also contain EzrA and FtsA, which bind to, membrane anchor, and stabilize FtsZ filaments/bundles (Fig. 6), but none of the other PG synthesis proteins analyzed in this study was detected moving in nascent rings. FtsZ and EzrA move at the same velocity, whereas FtsA seems to move faster (Fig. 6C). Because strains expressing other FtsA fusions showed phenotypes, we could not distinguish whether this faster velocity of FtsA was construct specific. However, Spn FtsA has the unique biochemical property of readily polymerizing into filaments (50), and the faster velocity of FtsA could reflect this additional polymerization in cells.
High-resolution microscopy and effects of a ΔmapZ mutation suggest that nascent FtsZ/FtsA/EzrA filaments/bundles use MapZ as a guide, as opposed to a beacon, to reach the equators of daughter cells. 3D-SIM IFM images detect FtsZ and EzrA together with MapZ in early and later nascent rings (Fig. 7A). Furthermore, in the absence of MapZ, orderly nascent EzrA rings are lost, and EzrA abruptly streams between daughter cells, often resulting in the aberrant ring orientation reported previously (Fig. 7B) (23). Streaming rarely was observed in wild-type Spn cells (≈1%) and represents a second mechanism for translocation of FtsZ and its associated proteins to daughter cells. In this respect, streaming is a “failsafe” mechanism that accounts for the lack of lethality of Spn ΔmapZ mutations.
In Smu cells, continuous FtsZ nascent ring formation was not reported and streaming, which was detected in ≈7% of cells, is proposed as the primary mechanism for FtsZ movement from septal to equatorial rings (Fig. 6 in (33)). A possible reason for this difference is that Spn and Smu are evolutionarily distant Streptococcus species (33), and ovococcus bacteria exhibit differences in the relative timing of septal and peripheral PG synthesis (51). On the other hand, technical or strain differences may underlie the different results. In particular, the Smu cells in movies shown in (33) appear to be in middle-to-late divisional stages and contain prominent equatorial rings, whereas movement of Spn FtsZ was recorded in early to late stages of division in this study (Fig. 2 and 3). Together, our results indicate that treadmilling FtsZ/FtsA/EzrA filaments/bundles are components of migrating MapZ rings throughout the cell cycle in Spn (Fig. 10) and may thus, play an important role in assembly and organization of these rings, about which little is known. This transport mechanism also moves part of the cellular FtsZ population to the equatorial rings, where the FtsZ filaments/bundles may serve to nucleate the transport of the remainder of septal FtsZ later in the division cycle (Fig. 10). This movement likely occurs mainly by depolymerization of FtsZ filaments/bundles and possibly by some new protein synthesis, since concerted streaming of FtsZ filaments/bundles was rarely observed in wild-type Spn cells.
In the course of these studies, we also determined a set of basic parameters about FtsZ dynamics in Spn cells. Filaments/bundles of wild-type FtsZ always move with a velocity of ≈31 nm/s in rings at all stages of division and while streaming in ΔmapZ mutants (Fig. 2E, 4B, and 7C). The same velocity of FtsZ filament/bundle movement was determined by TIRFm (Fig. 2E) and independently by wide-field observation of vertically immobilized cells (Fig. 4). A treadmilling mechanism of FtsZ filament/bundle movement was confirmed directly by SM-TIRFm (Fig. 5), and SM-TIRFm and immobilized cell measurements indicated the lifetime of FtsZ subunits in filaments/bundles as ≈12 s (Fig. 5C) and the lifetime of entire FtsZ filaments/bundles as ≈17 s (Fig. 4D and 4E). Based on the average subunit lifetime and the velocity of filament/bundle movement, the average length of treadmilling FtsZ filaments/bundles in Spn cells is ≈390 nm. In addition, our measurements show for the first time that the processivity of treadmilling of FtsZ filaments/bundle is ≈500 nm (Fig. 4C), indicating that FtsZ filaments/bundles traverse ≈20% of the circumference of Spn cells on average. Finally, as expected from previous precedents (12, 13), mutations that decrease GTP binding or the GTPase activity of FtsZ reduce the velocity of FtsZ treadmilling by as much as 10X (Fig. 9A). FtsZ(GTPase) mutations also disrupt the placement of division planes compared to wild-type cells (Fig. S9, S13, and S15).
Besides FtsZ, FtsA, and EzrA, none of the PG synthesis proteins tested in this study was a member of translocating MapZ rings (Fig. 8A and S12). Within the limits of standard TIRFm, some of these proteins showed minimal movement in mature septal and equatorial rings (i.e., GpsB (regulator), MltG (endo-lytic glycosylase), and StkP (Ser/Thr kinase)) (Introduction; Fig. S12), whereas DivIVA (regulator) and bPBP2x (septal TP) showed obvious dynamic movements (Fig. 8A and S12C). We therefore determined the role of FtsZ treadmilling on the motion of bPBP2x and its partner FtsW in mature septa of Spn cells. A new biochemical study reports that FtsW GT activity depends on interactions with its cognate Class B PBP (43). In support of this interaction in Spn cells, bPBP2x and FtsW colocalize at all stages of Spn division (Fig. 8D) and bPBP2x pulls down FtsW in a likely 1:1 complex (Fig. 8E). In addition, SM-TIRFm showed that bPBP2x and FtsW move at the same velocity on septa (Fig. 8F). Moreover, besides possibly stimulating FtsW transglycosylase activity in Spn cells, the TP activity of bPBP2x is required for septal PG synthesis, because a pbp2×(S337A) active-site mutant is not viable (H.-C. Tsui, unpublished result).
Five pieces of evidence support the conclusion that movement of the bPBP2x:FtsW complex in septa of Spn cells depends on PG synthesis and not on FtsZ treadmilling. First, the velocity bPBP2x and FtsW is slower than that of FtsZ treadmilling in wild-type Spn cells (Fig. 8C). Second, the decreased velocity of FtsZ treadmilling in FtsZ(GTPase) mutants is not correlated with a decrease of bPBP2x velocity (Fig. 9A). In fact, in the slowest mutant (FtsZ(D214A) overexpression), bPBP2x is moving about 5X faster than FtsZ treadmilling. Third, severe reduction in FtsZ treadmilling velocity does not markedly decrease PG synthesis indicated by FDAA incorporation (Fig. 9B). Fourth, a decrease in PG synthesis precursors caused by a ΔmurA1 mutation, decreases the velocity of bPBP2x and FtsW by the same amount, but does not decrease FtsZ treadmilling rate (Fig. 9C). Last, addition of methicillin at a concentration that mainly inhibits bPBP2x TP activity stops the movement of bPBP2x, but does not decrease the velocity of FtsZ treadmilling (Fig. 9D).
These results strongly support the conclusion that the movement of the bPBP2x:FtsW complex in septal PG synthesis in Spn cells depends on and likely mirrors new PG synthesis and is not correlated with the treadmilling of FtsZ filaments/bundles. By contrast, the velocities of the septal Class B PBPs of Bsu and Eco are coupled to and limited by FtsZ treadmilling, resulting in a correlation between septal bPBP and FtsZ treadmilling velocities (12, 13). The mechanism(s) underlying this coupling and its relationship to the rate of PG synthesis in Bsu and Eco are not understood. On the one hand, FtsZ treadmilling is further coupled to and limiting for septal PG synthesis and the constriction of Bsu cells (12). On the other hand, the velocity of FtsZ treadmilling is not correlated with the rate of PG synthesis determined by FDAA incorporation or the rate of septum closure of Eco cells (13, 52). These differences suggest that additional metabolic (e.g., PG precursor pools) and structural (e.g., PG width and outer membrane synthesis) constraints may influence the relative rates of FtsZ treadmilling, bPBP complex movement, and PG synthesis in different bacteria (53, 54).
Besides the sidewall Rod complexes of rod-shaped bacteria (14), there is another precedent for the dependence of PBP movement on PG synthesis. Recent results show that septal PG synthesis continues to close division septa of Sau after FtsZ treadmilling is inhibited by addition of a drug (PC190723) (16). This finding is again consistent with an FtsZ treadmilling-independent mechanism by which PG synthesis itself drives for PBP motion (13, 52). The dependence of bPBP2x:FtsW movement on PG synthesis can be rationalized by a model proposed for the dependence of PBP movement on PG synthesis in sidewall elongation of rod-shaped bacteria (14). It was proposed that MreB filaments form tracks that direct the linear motion of PBP complexes (14, 55). At any point in a track, the PBP complex has used substrate behind it to synthesize PG, and the utilization of available substrate in front of it drives its motion. In the case of Spn bPBP2x:FtsW, it is possible that FtsZ filaments/bundles themselves or other proteins in septal FtsZ rings provide tracks that couple movement to PG synthesis. In this model, FtsZ treadmilling acts to dynamically distribute and spatially organize directional PG synthesis, possibly through indirect interactions, as suggested by recent high-resolution microscopy (M. Boersma, (56)). Future studies will test this and related models to provide an understanding about the relationships among FtsZ treadmilling, PBP complex movement, and PG synthesis rate and location in different bacteria.
Experimental Procedures
Detailed experimental procedures are described in Supporting Information (SI) Experimental Procedures, including bacterial strains and growth conditions; Western blotting; 2D-epifluorescence microscopy and demograph generation; growth and imaging of live cells by TIRFm; TIRFm image acquisition and processing; periodicity analysis of FtsZ filaments/bundles in nascent rings; culture growth and sample preparation for micro-hole immobilization of Spn cells; image acquisition, processing, and data analysis of vertically oriented cells in micro holes; single-molecule (SM) TIRFm; TIRFm of ftsZ(G107S)//PZn-ftsZ-sfgfp merodiploid strains; TIRFm of HT-bPBP2x in an ftsZ(G107S) mutant; TIRFm of PZn-ftsZ(D214A) or PZn-ftsZ(D214A)-sfgfp merodiploid strains; 3D-SIM immunofluorescence microscopy (IFM); coimmunoprecipitation (co-IP) of FtsW-GFP with bPBP2x-FLAG; labelling of FtsZ(GTPase) mutant cells with FDAAs; and TIRFm of methicillin treated cells.
Author Contributions
Conception or design of the study (AJP, YC, SLS, YT, MJB, SH, MEW); performed research (AJP, YC, YAY, SH, MEW); contributed reagents and tools (AJP, JBV, HCTT, CD), analyzed data (AJP, YC, SLS, MJB, SH, MEW); wrote manuscript (AJP, SLS, SH, MEW).
Acknowledgements
We thank Jack Ryan and Jiaqi Zheng for strain constructions, members of the Winkler laboratory and Ethan Garner for critical discussions, Jim Powers for help with OMX microscopy, Luke Lavis for Fluor JF549, Michael VanNieuwenhze for TADA, Jan-Willem Veening and Yves Brun for strains and constructs, Orietta Massidda for anti-FtsZ antibody, and Dalia Denapite, Reinhold Bruckner, and Regine Hackenback for anti-PBP2x antibody. This work was supported by NIH grants R01GM113172, RO1GM114315, and RO1GM127715 (to MEW); NSF grant MCB1615907 (to SLS); Wellcome Trust and Royal Society Sir Henry Dale Fellowship grant 206670/Z/17/7 (to SH); NIH predoctoral Quantitative and Chemical Biology (QCB) training grant T32 GM109825 (to AJP); and NIH equipment grant S10RR028697 (to the Indiana University Bloomington Light Microscopy Center).