The strong association of EzrA(Spn) and FtsZ(Spn) was demonstrated from co-IP experiments showing that EzrA as bait can pull down FtsZ in Spn extracts without the cross-linking procedure (Figure 1C). To determine the spatiotemporal relationship of EzrA relative to FtsZ during the Spn cell cycle, we performed EFm on live cells expressing fluorescent-protein fusions of FtsZ and EzrA and IFM on fixed cells expressing epitope-tagged FtsZ and EzrA, as described in section “Materials and Methods.” Unless otherwise noted, all strains in this study did not show detectable growth or cell shape defects when fusion proteins were expressed as the only copy in the cell from native promoters (Supplementary Table 1 and data not shown). Linkers in fusion constructs are omitted in the text for simplicity, but are listed in Supplementary Tables 1, 2.

In live cells of strain IU14153, C-terminal fusion proteins EzrA-mNeonGreen and FtsZ-CFP showed complete overlap by EFm (Figure 1A). Demographs generated with MicrobeJ (Ducret et al., 2016) of fields of these unsynchronized cells at different stages of division confirmed overlap of EzrA and FtsZ and indicated simultaneous movement of the proteins from septa to developing equators of daughter cells, including the stage at which three bands are present (Figure 1B). Likewise, 2D-IFM images of fixed cells expressing epitope-tagged EzrA-HA and FtsZ-Myc (strains IU7223 or IU9713) indicated overlap at all stages of cell division (Supplementary Figure 2A, left). Cells were retrospectively binned into four stages, and fluorescence signals of EzrA-HA, FtsZ-Myc, and DNA nucleoids (DAPI) were averaged as described previously (Land et al., 2013; Supplementary Figure 2B). This quantitative analysis confirmed overlap of EzrA and FtsZ throughout the cell cycle (Supplementary Figures 2B,C). Averaged images at each division stage indicated that the EzrA-ring is slightly larger than the FtsZ-ring, similar to the slightly larger FtsA-ring in relation to the FtsZ-ring (Mura et al., 2016). This conclusion was confirmed by Student t-test analysis of the distributions of band diameters in each cell (Supplementary Figure 2C) (see Tsui et al., 2014). This slight difference in diameters coincides with location of the C-terminus of EzrA closer to the membrane than FtsZ (Supplementary Figures 1C,D), whose apparent distance is likely enhanced by the length of the fluorescently labeled antibodies used in IFM.

High-resolution 3D-SIM IFM images (Supplementary Figure 3) confirmed and extended results from 2D-EFm. For each stage of division, > 20 cells were examined by 3D-SIM. Stage 1 cells show highly correlated spatial organization of EzrA and FtsZ, with the diameter of EzrA slightly larger than that of FtsZ in rotated images (Supplementary Figure 3). Most EzrA and FtsZ are visible at the septa of all Stage 2 cells examined, and in ≈35% of cells, faint rings containing EzrA and FtsZ are detected in nascent rings moving toward the equatorial regions of daughter cells (arrow, Supplementary Figure 3, Stage 2) (Perez et al., 2019). Later in division, EzrA and FtsZ locate in a characteristic three-ring pattern at the septum and the equators of daughter cells over the separating nucleoid (Supplementary Figure 3, Stages 2.5 and 3). In Stage 4 cells, most EzrA and FtsZ locate to the new equatorial rings of the daughter cells, sometimes leaving a small dot of each protein at the closing septum (Supplementary Figures 2B, 3).

We extended these analyses to further characterize FtsZ and EzrA localization in a ΔmapZ mutant. Certain FtsZ-GFP fusions are partially functional in ΔmapZ mutants, resulting in more severe morphological defects than in strains containing ΔmapZ alone (Perez et al., 2019). We did not observe these compounded defects in a ΔmapZ mutant expressing epitope-tagged EzrA-HA and FtsZ-Myc by PCm or growth curve analysis (Supplementary Figure 2A, right and data not shown). Previously, we showed that in the absence of MapZ, EzrA moves to daughter cells by a delayed and aberrant streaming mechanism that was rarely (1%) seen in WT cells (Perez et al., 2019). This secondary, failsafe mechanism is responsible for placement of Z-rings near the equators of most, but not all, ΔmapZ daughter cells (Supplementary Figure 2A, right). EzrA and FtsZ co-localized during delayed streaming in late-divisional ΔmapZ cells, as observed by 2D-EFm (Supplementary Figures 2A,B, stage 4) and at high resolution by 3D-SIM (Supplementary Figure 4). Together, these results confirm that a tight association between EzrA and FtsZ filaments/bundles is maintained during this alternate failsafe streaming mechanism.

Previously, we reported that EzrA or FtsA moves circumferentially at the same velocity as treadmilling FtsZ filaments/bundles in nascent rings or early equatorial rings during the Spn cell division (Perez et al., 2019). By contrast, the density of molecules was too great in septal rings or fully developed equatorial rings to determine dynamic patterns by TIRFm. We further evaluated how tightly associated dynamic patterns of EzrA are with FtsZ or FtsA in developing equatorial or nascent rings. In these experiments, we performed dual-color TIRFm where we tracked native locus-expressed FtsZ-sfGFP or GFP-FtsA alongside EzrA-HaloTag (EzrA-HT) labeled to saturation with the HaloTag JF549 ligand (Grimm et al., 2015; Perez et al., 2019). Kymographs along equatorial and nascent-ring planes indicate that for the most part, the movement of FtsZ filaments/bundles or FtsA in FtsZ filaments/bundles mirrored that of EzrA in time and space, indicating a strong dynamic association (Supplementary Figure 5). In addition, we observed some events consistent with unassociated EzrA, as indicated by lack of FtsZ or FtsA and vertical EzrA traces (magenta lines, Supplementary Figure 5). Together, we conclude that EzrA is tightly associated with circumferentially moving FtsZ filaments/bundles that also contain FtsA, although some EzrA is not associated with moving FtsZ filaments/bundles.

Several reports suggest that EzrA is associated with PBPs and PG synthesis in Bsu and Sau (Claessen et al., 2008; Jorge et al., 2011; Steele et al., 2011). In Spn, PG synthesis enzymes and activity localize late in division from septa to new equatorial FtsZ-rings in daughter cells (Tsui et al., 2010; Land et al., 2013; Perez et al., 2019). Given the strong association between FtsZ and EzrA described above, we postulated that EzrA would precede PG synthesis to the new equatorial rings of daughter cells. To localize EzrA relative to PG synthesis, we carried out short pulse (2.5 min) labeling with a fluorescent D-amino acid (FDAA) in a strain expressing EzrA-sfGFP from the ezrA native locus (strain IU10254) as described in Materials and Methods (Supplementary Figure 6). FDAAs are incorporated into the PG in regions of active PBP transpeptidase activity (Kuru et al., 2012; Tsui et al., 2014; Boersma et al., 2015). Cells were binned by division stage, and average fluorescence intensities were quantitated as described above for EFm images (Supplementary Figure 2). Statistical analysis of labeling widths shows that the FDAA labeling was greater than the EzrA width at all division stages (Supplementary Figure 6B). In early stage-1 and -2 cells, PG synthesis and EzrA are confined to division septa. In later stage-3 and -4 cells, EzrA shows the same pattern reported previously for FtsZ (Land et al., 2013). PG synthesis remains confined to the septum, while a substantial amount of EzrA has moved to the equators of daughter cells (Supplementary Figure 6A). By stage 4, little EzrA remains at the septum compared to small region of PG synthesis.

3D-SIM images of vertically oriented cells confirm the separation of the FDAA labeling and EzrA in early-stage 1 and 2 cells (Supplementary Figure 6C). In stage 1 cells, the separation is small and consistent with the fact that PG synthesis occurs outside of cells, whereas the sfGFP-fused C-terminus of EzrA is inside the cell near the membrane (Supplementary Figures 1D, 6C). The increased separation of labeling in stage 2 cells is consistent with the recent report that the contracting FtsZ-ring tracks with an inner ring of sPG synthesis at the leading edge of the septal annulus, while an outer concentric ring of pPG elongation synthesis remains at the outer edge of the septal disk (Perez et al., 2021). We conclude that EzrA is tracking in a pattern reported before for FtsZ and FtsA that is different from that of PG synthesis.

To directly test the essentiality of EzrA(Spn), we performed assays for recovery of colonies following transformation with amplicons containing null mutations, as described in section “Materials and Methods.” We transformed an amplicon containing a null ΔezrA<>aad9 allele, where the ezrA ORF is replaced by an ORF imparting spectinomycin resistance, into an unencapsulated derivative of strain D39W (IU1945; D39 Δcps) and compared the number of transformants with a positive-control amplicon (ΔpurR<>aad9) and a negative-control amplicon (ΔftsZ::aad9) (Table 1). Following 24 h of incubation, the ΔpurR control showed many transformants, while ΔezrA or ΔftsZ showed none (Table 1, column 1). After 48 h of incubation, a limited number of small, variable-sized transformant colonies, indicative of potential suppressor mutations, were observed for ΔezrA, but not ΔftsZ (Table 1). These potential ΔezrA transformants showed similar cell growth as the EzrA depletion strains described next; however, they were unstable and could not be recovered after storage and were not further characterized here.

EzrA(Bsu) is conditionally essential in the presence of certain antibiotics (Gamba et al., 2015). Consequently, we repeated the transformation assay with an amplicon containing a ΔezrA::P_c-erm deletion–insertion null mutation into the D39 Δcps Spn strain, selecting for erythromycin resistance. We obtained no ΔezrA::P_c-erm transformants after 24 of incubation (data not shown), similar to the results for the ΔezrA<>aad9 amplicon (Table 1). Finally, we repeated the transformation assay for the ΔezrA<>aad9 amplicon into laboratory strain R6 (strain EL59), which has acquired numerous mutations compared to its D39W progenitor strain (Lanie et al., 2007). Mutations in the R6 strain genetic background suppress the essentiality of numerous genes involved in PG synthesis, including ΔmreC, ΔgpsB, ΔmltG, and ΔstkP (Land and Winkler, 2011; Rued et al., 2017; Straume et al., 2017). However, transformation of the ΔezrA<>aad9 amplicon into strain R6 gave no colonies after 20 h of incubation (data not shown), similar to the D39 Δcps strain (Table 1). Attempts to delete ezrA in the commonly used Rx1 background were also unsuccessful (Massidda et al., 2013). Together, these results show that EzrA(Spn) is essential and not conditionally lethal.

To determine phenotypes caused by EzrA(Spn) depletion, we constructed a ΔezrA//P_Zn-ezrA⁺ merodiploid strain (IU8799) by moving a ΔezrA<>aad9 deletion into a ezrA⁺//P_Zn-ezrA⁺ strain (IU8795) containing an ectopic copy of ezrA⁺ under control of a Zn²⁺-inducible promoter in the bgaA gene (see Supplementary Table 1). Consistent with EzrA(Spn) essentiality, transformation of the ezrA⁺//P_Zn-ezrA⁺ merodiploid with the ΔezrA<>aad9 amplicon only resulted in many colonies when Zn (0.5 mM ZnCl₂ and 0.05 mM MnSO₄) was added to selection plates (Table 1, middle columns). Without added Zn, only a few putative suppressor mutants were detected. Likewise, transformation of a comparable ftsZ⁺//P_Zn-ftsZ⁺ merodiploid with a ΔftsZ::aad9 amplicon only resulted in colonies when Zn was added (Table 1, right columns).

To characterize EzrA depletion phenotypes, we performed growth curve analyses and PCm on cells depleted for EzrA in BHI broth, as described in section “Materials and Methods.” Growth rates and yields of an EzrA merodiploid (ΔezrA//P_Zn-ezrA⁺) strain were dependent on the concentration Zn (ZnCl₂ + 1/10 MnSO₄) in cultures, where Mn²⁺ was added to prevent Zn²⁺ toxicity (see section “Materials and Methods”; Figure 2A and Supplementary Figure 7A; Jacobsen et al., 2011). Depletion growth curves depended on OD₆₂₀ at which Zn was omitted from cultures containing 0.5 mM Zn (Supplementary Figure 7B), consistent with the requirement of EzrA for growth. Although cultures containing 0.3–0.5 mM Zn grew, similarly (Supplementary Figure 7A), only 0.5 mM fully complemented cell shape and was used in depletion experiments started at OD₆₂₀ ≈0.01–0.05 (Supplementary Figures 7B,C). Finally, a merodiploid strain ectopically expressing EzrA fused to a FLAG³ epitope tag (ΔezrA//P_Zn-ezrA-FLAG³) grew, similarly, to the ΔezrA//P_Zn-ezrA⁺ strain in 0.5 mM Zn and after Zn was removed, after which OD₆₂₀ stopped increasing in ≈3 h (Supplementary Figure 7C).

After depletion of EzrA or EzrA-L-FLAG³ for 1 h or longer, cell shapes became distorted and non-uniform (Figure 2B). With time, the median aspect ratio of EzrA-depleted cells decreased compared to the non-depleted control cells, and cells appeared as irregular spheroids with variable relative volumes (Figure 2B and Supplementary Figure 8). Quantitative Western blotting showed that ectopic EzrA-L-FLAG³ in the presence of 0.5 mM Zn was expressed at 85% of the cellular amount of EzrA-L-FLAG³ expressed from the native chromosomal locus (Figure 2C). Depletion of EzrA-L-FLAG³ for 1 to 4 h reduced the relative cellular amount to 66% and to 10%, respectively (Figure 2C). At 3 h of depletion, when cultures stopped growing (Figure 2A and Supplementary Figure 7C), the relative amount of EzrA-L-FLAG³ was ≈22% of that in cells expressing EzrA-L-FLAG³ from the native locus (Figure 2C). Distorted cells depleted of EzrA did not appear to lyse, and cell debris was not observed by PCm. Viability staining with the LIVE/DEAD BacLight procedure, as described in Materials and Methods, showed that > 90% of non-growing, distorted cells lacking EzrA were stained as viable for 3, 4, and 7 h after depletion (Supplementary Table 3). Consistent with this conclusion, CFUs of ΔezrA//P_Zn-ezrA cells depleted of EzrA for 4 or 8 h without Zn were recovered on plates containing Zn with no decrease or increase in CFU/mL throughout the depletion time course (data not shown). Thus, EzrA is required for Spn cell size and shape homeostasis, although prolonged depletion of EzrA under these culture conditions is bacteriostatic and not bactericidal.

Finally, we determined the effects of the EzrA(ΔTM) and EzrA(ΔQNR motif) or EzrA(QND motif) mutant variants, which mediate membrane anchoring and medial localization, respectively, in Bsu (Supplementary Figures 1C, 9A) (Land et al., 2014). We expressed EzrA(ΔTM) (deletion of amino acids 2-26; strain IU11123), EzrA(ΔQNR) (deletion of amino acids 510-516 (deletion of amino acids 510-516; strain 10909), and EzrA(QND motif) (ezrA(R515D); strain IU10901) from the native locus of a merodiploid strain ectopically expressing ezrA⁺ from the P_Zn promoter (Supplementary Table 1). Depletion of EzrA⁺ showed that ezrA(ΔTM) and ezrA(ΔQNR) caused the same growth defects as ΔezrA (Supplementary Figure 9B), whereas ezrA(QND) strains grew like WT (data not shown). sfGFP fused to the C-terminus of EzrA(QND), EzrA(ΔQNR), or EzrA(ΔTM) in comparable EzrA⁺ merodiploid strains showed that EzrA(QND)-sfGFP and EzrA(ΔQNR)-sfGFP were expressed at nearly the same level as EzrA⁺-sfGFP upon EzrA⁺ depletion, whereas EzrA(ΔTM)-sfGFP was expressed at only 30% of the WT level (Supplementary Figure 9C). Moreover, the ectopic expression of EzrA(ΔTM)-FLAG³ at the WT level did not complement a ΔezrA mutation, confirming that the TM domain is required for EzrA function (data not shown). EzrA(ΔQNR)-sfGFP and EzrA(ΔTM)-sfGFP were highly mislocalized upon depletion of EzrA⁺ in merodiploid strains (Supplementary Figure 9D). Remaining midcell rings containing EzrA(ΔQNR)-sfGFP or EzrA(ΔTM)-sfGFP may reflect dimer formation with residual EzrA⁺ after depletion. In contrast, EzrA(QND motif)-sfGFP largely localized normally to midcell and equatorial bands in irregularly shaped cells (Supplementary Figure 9D). In addition, the ezrA(QND)//P_Zn-ezrA⁺ mutant was temperature sensitive for growth at 42°C in the absence of Zn (data not shown). We conclude that the TM domain of EzrA is required for membrane localization and possibly protein stability, while the QNR motif localizes EzrA to the septal and equatorial rings of Spn cells. While not necessary for ring localization, amino acid R515 in the QNR motif of EzrA(Spn) is required for full function.

Depletion of EzrA often results in larger cells attached to smaller cells (Figure 2B), reminiscent of minicells containing guillotined DNA or no DNA produced by division site placement mutants (Wu and Errington, 2004; Bernhardt and de Boer, 2005). To test this idea, we stained the DNA of cells from WT, EzrA-complemented, or EzrA-depleted cultures with DAPI as described in Materials and Methods. We counted nucleated and anucleate cells that were at pre-divisional (stage 1) or post-divisional (stage 4) stages (n = 400 per condition) based on PCm (Supplementary Figure 10). Depletion of EzrA resulted in a relatively high number (3.25% of cells detected by PCm) of cells lacking nucleoids compared to WT (none) or EzrA-complemented strains (<0.25%) (arrows, Supplementary Figure 10; Supplementary Table 4). By comparison, cells lacking MapZ, which plays a role in positioning FtsZ-rings at the equators of daughter cells (Fleurie et al., 2014a; Holeckova et al., 2015; Perez et al., 2019), displayed only about 0.5% anucleate cells, which was greater than WT but far fewer than in EzrA-depleted cells (Supplementary Table 4). These results indicate that EzrA plays an important role in chromosome segregation, likely by modulating Z-ring placement and organization, analogous to Z-ring regulators in rod-shaped bacteria.

EzrA regulates FtsZ-ring number and position in Bsu cells. In the absence of EzrA, Bsu cells contain two extra parallel Z-rings per cell length (Levin et al., 1999). To determine the role of EzrA in regulating the presence and position of FtsZ-rings in Spn, we performed 2D-IFM using the anti-FtsZ(Spn) antibody in EzrA-depleted cells (Lara et al., 2005). The majority of pre-divisional (stage 1) cells of WT, EzrA-complemented (+Zn), and EzrA-depleted (1 h) cells contained normally placed FtsZ-rings at midcell (see Figures 1A, 3A). In contrast, after 2 h of EzrA depletion, the majority of cells lacked identifiable Z-rings (Figure 3A). Cells classified as “FtsZ other” did not contain obvious Z-rings but did show the presence of FtsZ-labeling, and cells classified as “no FtsZ detected” lacked FtsZ labeling and appeared opaque by PCm, which may indicate an inability to be permeabilized and take up antibodies (data not shown). Nevertheless, comparison of cells with the FtsZ signal indicates that at 2 or 3 h after EzrA depletion, less than half of pre-divisional cells have Z-ring structures, while 90% of WT and 70% of EzrA-complemented cells contain Z-rings (Figure 3A), indicating that EzrA is necessary for FtsZ-ring formation in Spn. 3D-SIM IFM confirmed this conclusion at higher resolution than 2D-IFM. Among the patterns of FtsZ labeling in cells depleted for EzrA for 3 h, we observed diffuse localization and aberrant, twisted Z-ring structures, as well as cells lacking labeling patterns (Figure 3C).

To determine if EzrA influences positioning of FtsZ-rings in Spn, we determined the relative position of Z-rings in cells from Figure 3A. We measured the distance of Z-rings to the nearest cell pole normalized to cell length, where 0.5 indicates exact midcell placement. Fifty percent of WT and EzrA-complemented (+Zn) cells had FtsZ-rings positioned between 0.45 and 0.5 (Figure 3B). Cells depleted of EzrA for 1 or 2 h maintained Z-rings near this midcell range. By 3 h of EzrA depletion, the median (0.38) and distributions started to trend downward. Z-rings were significantly out of the midcell region at 3 h, indicating that the presence of EzrA also influences placement of Z-rings in Spn.

We confirmed these conclusions by tracking FtsZ-GFP expressed from the native chromosomal locus in live cells complemented or depleted for EzrA (Figure 4 and Supplementary Figure 11). Defects in midcell localization and correct plane placement of FtsZ-GFP were apparent by 1 h of EzrA depletion (Figure 4B and Supplementary Figure 11B). By 3 h of EzrA depletion, polar placement of FtsZ-GFP band structures or condensed foci were observed (Figure 4B and Supplementary Figure 11C). Demographs of FtsZ-GFP in a population of EzrA-depleted cells revealed major delocalization of FtsZ-GFP as a function of relative cell length (Figure 4B). Localization of FtsZ-GFP to equatorial rings became aberrant at 2 h of EzrA depletion, while FtsZ-GFP localization in the whole population became aberrant at 3 h (Supplementary Figure 11B). We also observed that the culture density of strains expressing FtsZ-GFP began to drop after about 4 h of EzrA depletion, in contrast to strains expressing FtsZ⁺, indicating an aberration caused by the FtsZ-GFP fusion upon extended EzrA depletion (Supplementary Figure 11A). We conclude from these combined IFM and GFP imaging experiments that EzrA is required for the presence and localization of FtsZ in Spn, in contrast to Bsu where the absence of EzrA leads to the formation of additional FtsZ-rings (Levin et al., 1999).

In pre-divisional and early divisional Spn cells, PG synthesis occurs at a midcell ring initially organized by FtsZ and FtsA (Supplementary Figure 1) (see Briggs et al., 2021). As septation proceeds, FtsZ continues to organize the inner ring of sPG synthesis that closes the septum, while pPG remains in an outer ring that lacks FtsZ (Briggs et al., 2021; Perez et al., 2021). Given that FtsZ organizes PG synthesis at different stages of division in Spn, we predicted aberrant PG synthesis patterns upon depletion of EzrA that mirrored those of FtsZ described above (Figures 3, 4, and Supplementary Figure 11). To detect regions of active PBP TP activity, we labeled cells sequentially with two colors of FDAAs (Supplementary Figure 12A). Cells were labeled with a “long pulse” of HADA FDAA at the start of EzrA depletion or continued synthesis in merodiploid strains, as described in Materials and Methods. After 1, 2, or 3 h, cells were washed and a second TADA FDAA was added for a short pulse (5–18 min depending on strains or conditions, Supplementary Figure 12A), indicating regions of new PG synthesis. Cells observed by PCm and EFm showed patterns of TADA labeling (Figure 5A and Supplementary Figures 12A–C) similar to those of FtsZ with increasing EzrA depletion (Figure 3A). At 1 h of EzrA depletion, most TADA labeling was at midcell rings of pre-divisional and post-divisional cells, but by 2 h and beyond of EzrA depletion, TADA labeling was increasingly seen in polar foci and other aberrant and diffuse patterns (Figure 5A). 3D-SIM images captured details of these aberrant PG synthesis patterns, notably rings or puncta of TADA labeling at distal poles, diffuse TADA labeling, and aberrantly placed planes of TADA labeling in EzrA-depleted cells (Supplementary Figures 13B,C). As expected from these results, the placement of medial TADA-rings, when present, showed the same trend as FtsZ-rings in being located away from the exact midcell the longer EzrA was depleted (Figures 3B, 5B). Together, these results indicate that PG synthesis tracks with the mislocalization of FtsZ that occurs when EzrA is depleted, possibly contributing to the diverse, aberrant cell shapes that occur.

The essentiality of EzrA(Spn) and the dependence of FtsZ-ring formation on EzrA (Figures 3, 4) suggest that EzrA(Spn) is a positive regulator of FtsZ-ring formation, unlike EzrA(Bsu) which is a negative regulator (Levin et al., 1999). To test this hypothesis, we overexpressed EzrA to see if extra FtsZ-rings appear in Spn. To this end, we introduced copies of P_Zn-ezrA⁺ into two different ectopic sites in an Spn strain containing ftsZ-CFP at the native chromosomal locus (strain IU13700). In the BHI medium without added Zn, FtsZ-CFP localized normally in the ftsZ-CFP or ezrA⁺ merodiploid strain (-Zn, Figure 6), whereas overexpression of EzrA resulted in 90% of cells showing elongated, wider chains of cells containing multiple FtsZ-CFP rings (arrows, Figure 6). We corroborated this conclusion using a different construct grown in the C+Y medium. FtsZ-sfGFP was expressed at the native locus in a merodiploid strain containing an ectopic copy of P_Zn-ezrA⁺ (strain IU14224). The ftsZ-sfgfp control strain with or without Zn and the ftsZ-sfgfp//P_Zn-ezrA⁺ merodiploid strain without Zn showed the expected pattern of FtsZ-rings (Supplementary Figure 14). Overexpression of EzrA by the addition of 0.25 or 0.5 mM Zn again caused formation of chains of elongated, wider cells containing multiple FtsZ-sfGFP rings (arrows, Supplementary Figure 14). Note that in the C+Y medium, the addition of 0.5 mM Zn and 0.05 mM Mn led to cell debris, indicative of lysis in both ezrA⁺ and EzrA-overexpression cells. However, multiple FtsZ-sfGFP rings were only present in EzrA-overexpression cells and not in WT ezrA⁺ cells. These combined experiments confirm that the appearance of multiple FtsZ-rings upon EzrA overexpression is not tag, construct, or medium dependent. We conclude that overexpression of EzrA in Spn does indeed leads to increased formation of parallel FtsZ-rings, consistent with EzrA acting as a positive regulator of FtsZ-ring formation in Spn cells.

To study the relationship between FtsZ and EzrA in Spn, we further characterized the phenotypes caused by FtsZ depletion under our culture conditions. We constructed a strain expressing FtsZ-Myc from the chromosomal locus (IU7223) and a ΔftsZ merodiploid strain expressing FtsZ⁺ or FtsZ-Myc from an ectopic site (IU8124 or IU8237, respectively). Induction of FtsZ⁺ or FtsZ-Myc with 0.3 mM Zn fully complemented the ΔftsZ mutation in the merodiploid strains for growth and cell morphology (Figures 7A,B, and Supplementary Figures 15A,B). Western blotting indicated that ectopic FtsZ-Myc was expressed at 80% in the merodiploid compared to FtsZ-Myc expressed from the chromosome (Figure 7C). Depletion of FtsZ⁺ or FtsZ-Myc resulted in a rapid cessation of growth, rounding of cells to spheroids, enlargement of relative cell volumes (≈ 4 ×), and a decrease in OD₆₂₀ indicative of cell autolysis (Figures 7A,B and Supplementary Figures 15A,B). These phenotypes are similar to the aberrant, heterogeneous, exploding cells reported previously for CRIPSRi- or IPTG-regulated depletion of FtsZ in strain D39V (Liu et al., 2017; Gallay et al., 2021). Western blotting indicated that FtsZ-Myc cellular amount was reduced rapidly to < 10% within 1 h of depletion (Figure 7C). Titration with 0.135 mM Zn showed the relative expression of FtsZ-Myc at 3 h of growth to be at ≈23%, which was sufficient to allow continued growth without lysis (data not shown). Autolysis, but not cell shape defect upon FtsZ depletion, was abrogated by a ΔlytA mutation, indicating that autolysis of spheroidal lacking FtsZ was mediated by induction of the LytA amidase activity (Figure 7A). Nevertheless, CFU counting showed that the viability of the spheroid FtsZ-depleted ΔlytA cells decreased at a similar rate as in the lytA⁺ cells that were lysing (Supplementary Figure 15A). In FtsZ-depleted lytA⁺ and ΔlytA cells, CFU/mL counts at 7 h of depletion dropped to ≈1% to 3% compared to CFU counts at 1 to 3 h of depletion. We concluded that, as expected, FtsZ depletion in Spn is lethal and results in LytA-induced autolysis and loss of viability that is independent of LytA autolysis.

We determined the organization of EzrA upon FtsZ depletion in a doubly epitope-tagged merodiploid strain expressing EzrA-HA from the native chromosomal locus and FtsZ-Myc ectopically (strain IU8237). We averaged 2D-EFm IFM images of dually labeled cells at different stages of division (Figure 8A). After 1 h of FtsZ depletion, FtsZ localization started to become disorganized, while EzrA-rings remained at midcell septa but failed to appear at the equators of daughter cells in stage 3 and 4 divisional cells (arrows, Figure 8A), compared to WT cells (in strain IU7223; see Supplementary Figure 2B). After 2 h of FtsZ depletion, a band of EzrA remained at septa and only weakly appeared at equators compared to EzrA-complemented (+Zn) cells, while FtsZ became severely disorganized (Figure 8A and Supplementary Figure 16). 3D-SIM images confirmed these conclusions (Supplementary Figure 17). Although some cells were able to complete a cycle of division and chromosome segregation, EzrA remained at septa in rings or diffuse areas and was not recruited to the equators of daughter cells (bottom, Supplementary Figure 17). Enlarged, spheroidal cells depleted of FtsZ contained diffuse regions of nucleoid staining, consistent with defects in chromosome organization and segregation (Supplementary Figure 17). This lack of EzrA-rings at future division sites supports the conclusions that FtsZ is required for initial recruitment of EzrA to equatorial rings of daughter cells.

Although EzrA was absent from equatorial rings during FtsZ depletion, many pre-divisional cells still contained EzrA-rings but lacked FtsZ labeling, which was obscured by the image averaging used in Figure 8A. To circumvent this issue, we counted co-localization events of EzrA-rings and FtsZ-rings during FtsZ depletion in pre-divisional cells (Figure 8B). WT cells displayed Z-ring and EzrA-ring colocalization in nearly all cells, whereas FtsZ-complemented strains (+Zn) showed ≈60% EzrA and FtsZ co-localization in pre-divisional cells (Figure 8B), consistent with the slightly lower expression of FtsZ detected in Western blots (Figure 7C). Depletion of FtsZ for 1 or 2 h showed a large, increasing drop of EzrA-ring and FtsZ-ring colocalization (Figure 8B). Concomitantly, after severe depletion of FtsZ for 2 h, a majority (≈80%) of cells lacked both EzrA- and FtsZ-rings (Figure 8B). In addition, we examined the localization pattern of FtsA upon FtsZ depletion since FtsA, similar to EzrA, co-localizes with FtsZ at all stages of cell division (Mura et al., 2016). Co-localization experiments of FLAG-FtsA upon FtsZ-Myc depletion led to a similar result (Supplementary Figure 18). Together, these data indicate that once assembled, EzrA- and FtsA-ring structures can persist in the absence of FtsZ-rings, but assembly of new EzrA and FtsA-rings requires the presence of FtsZ (Figure 8C).

Finally, a similar conclusion was reached about the organization of PG synthesis upon FtsZ depletion. Merodiploid ΔftsZ//P_Zn-ftsZ⁺ cells labeled with one color of FDAA for a long pulse to indicate regions of “old” PG were washed and labeled with a short pulse of a second color of FDAA to indicate regions of “new” PG synthesis (Supplementary Figure 19A) and examined by 2D-EFm (Supplementary Figures 19B, 20) or by 3D-SIM (Supplementary Figure 19C). After 1 h of FtsZ depletion, new PG synthesis was still occurring in rings at midcell regions. By 2 h of FtsZ depletion, organized FDAA-rings of new PG synthesis were largely absent (Supplementary Figure 19B), and only aberrant, diffuse FDAA-labeling patterns at old division sites and elsewhere were present in FtsZ-depleted in cells (Supplementary Figures 19C, 20). We conclude that FtsZ is required for organizing PG synthesis at septa and equators of dividing Spn cells, but PG synthesis continues in a diffusive manner in the absence of this organization.

Previously, it was shown that pneumococcal EzrA interacts with FtsZ in B2H experiments and in biochemical assays using surface plasmon resonance detection (Fleurie et al., 2014b; Rued et al., 2017). We further determined complexes that contain EzrA in Spn cells by co-IP (Figures 1C, 9) and potential direct interactors of EzrA(Spn) by B2H assays (Supplementary Figure 21). For Co-IP assays, chromosomal expressed EzrA-L-FLAG³ in extracts of cells was the bait that bound magnetic FLAG-tag beads, and an extract from cells expressing untagged EzrA was used as the negative control (Figure 1C). Even without cross linking, EzrA pulled down FtsZ, indicative of a complex containing FtsZ and EzrA. B2H assays confirmed direct binding between EzrA and FtsZ in both B2H constructs (Supplementary Figure 21; Fleurie et al., 2014b; Rued et al., 2017).

EzrA was reported to interact with several cell division proteins in other bacteria, especially in B2H assays (Claessen et al., 2008; Jorge et al., 2011; Steele et al., 2011; Pompeo et al., 2015). To identify complexes that contain EzrA, we performed co-IP experiments on extracts of cross-linked Spn cells, as described in section “Materials and Methods.” All strains used in these experiments expressed untagged EzrA (negative control) or epitope-tagged EzrA-FLAG³ (bait) (Table 2) or other FLAG-tagged bait proteins (Table 3) in combination with prey proteins that were epitope tagged with -HA or -Myc (Figure 9; Supplementary Figures 22–24).

We tested whether EzrA(Spn) can be detected in complexes at some stage of cell division with FtsZ-ring regulators, including FtsA, SepF, and ZapA, as was determined previously by co-IP for EzrA in Bsu (Ishikawa et al., 2006). We also tested if EzrA(Spn) is in a complex with MapZ, which guides formation of equatorial Z-rings (Fleurie et al., 2014a; Holeckova et al., 2015; Perez et al., 2019). Consistent with a role for EzrA in regulating Z-ring dynamics, EzrA(Spn) was detected in complexes with FtsZ and the other inferred Z-ring regulators, FtsA, SepF, ZapA, and MapZ (Figure 9A and Tables 2, 3). Strong co-IP signals were detected for at least one bait:prey combination with EzrA and each of these proteins. For EzrA and ZapA, complex formation was detected for ZapA(bait):EzrA(prey) (Table 3), but not for EzrA(bait):ZapA(prey) (Table 2), perhaps indicating a detection limit of our co-IP assay. Interactions between EzrA and FtsZ, FtsA, SepF, and MapZ were corroborated by B2H assays (Supplementary Figure 21), indicating that EzrA directly interacts with these proteins at some stage of the cell cycle (Figure 9B). In contrast, we did not detect direct interactions in B2H assays between EzrA(Spn) and ZapA or its partner ZapJ, which is discussed below (Supplementary Figure 21). This negative result could reflect the inability of ZapA(Spn) and ZapJ to interact in the absence of each other in Eco or cross-binding of ZapA(Spn) to ZapAB(Eco) or FtsZ(Eco) in B2H assays.

We also probed for EzrA(Spn) interactions with PBPs and cell-cycle regulators. In Bsu and Sau, EzrA interacts with PBPs based on B2H assays (Claessen et al., 2008; Jorge et al., 2011; Steele et al., 2011). In Spn, we detected an unambiguous co-IP signal for complexes containing EzrA and aPBP1a (Supplementary Figures 23A,B; Table 3). In contrast, we detected a marginal co-IP signal for complexes containing EzrA and bPBP2x (Supplementary Figure 22C; Table 3) and no detectable complex between EzrA and aPBP2a or bPBP2b (Supplementary Figure 22C; Table 2). Consistent with the co-IP data, B2H assays detected interactions between EzrA and aPBP1a or bPBP2x and a lack of detection with bPBP2b, although an interaction with aPBP2a was detected using B2H (Supplementary Figure 21). We conclude that complexes containing EzrA(Spn) also contain certain PBPs (Figure 9B).

We further tested whether EzrA can be detected in complexes with the cell cycle and PG-synthesis regulators DivIVA (morphogenetic determinant Fadda et al., 2007; Fleurie et al., 2014b), StkP (serine/threonine kinase Novakova et al., 2010; Beilharz et al., 2012; Fleurie et al., 2014b), GpsB (regulator of PBP activity Cleverley et al., 2016, 2019; Rued et al., 2017), and FtsK (chromosome partitioning and dimer resolution Le Bourgeois et al., 2007; Burton and Dubnau, 2010; Massidda et al., 2013; Pinho et al., 2013). Previous B2H and surface plasmon resonance assays suggested that EzrA(Spn) interacts with GpsB, StkP, and DivIVA (Fleurie et al., 2014b). Previously, we reported complexes containing EzrA and GpsB in Spn cells (Rued et al., 2017). We detected strong co-IP signals for complexes in Spn cells containing EzrA and StkP or FtsK (Figure 9A and Table 3). In contrast, the signal for DivIVA(prey):EzrA(bait) complexes was marginal (Figure 9A and Table 3), as was the signal for DivIVA(bait):FtsZ(prey) complexes (Supplementary Figure 24A; Table 3), possibly indicating a detection limit of the co-IP assay. Interactions between EzrA and GpsB, StkP, or FtsK were corroborated by B2H assays (Supplementary Figure 21). We also tested for complex the formation between EzrA and MreC (PG elongasome/pPG synthesis regulator) by co-IP and B2H assays. We were unable to detect a complex containing EzrA and MreC in Spn by co-IP (Table 2), but we did detect an interaction by B2H assays (Supplementary Figure 21), possibly indicating a transient interaction in cells.

Finally, we used the B2H assay to test for interactions between EzrA(Spn) and several proteins not tested in co-IP assays. These assays indicate possible direct interactions between EzrA and MacP (positive regulator of PBP2a) (Fenton et al., 2018), RodA (SEDS GTase in pPG synthesis) (Meeske et al., 2016), MreD and RodZ (PG elongasome/pPG regulators) (Massidda et al., 2013; Briggs et al., 2021), MpgA (formerly MltG; glycosidase in pPG synthesis) (Taguchi et al., 2021), and FtsQ/L (divisome assembly proteins) (Noirclerc-Savoye et al., 2005; Briggs et al., 2021). Taken together, these results show that EzrA(Spn) forms complexes and interacts with many key proteins that mediate Z-ring regulation, cell division, and PG synthesis (Figure 9B and Supplementary Figure 21), consistent with its extended spectrin-like repeated structure (Supplementary Figure 1D) and the diverse interactions reported for EzrA in other bacteria. These multiple EzrA interactions support a model in which EzrA is not only required as a positive regulator of Z-ring assembly in Spn (above; Figure 6), but functions as a regulator that helps link FtsZ/FtsA filaments/bundles to PG synthesis and cell-cycle checkpoints.

EzrA(Spn) is in complexes with several proteins that regulate FtsZ-ring formation, including FtsA, SepF, and ZapA (Figure 9A and Tables 2, 3). FtsA and SepF have been characterized previously in Spn (Mura et al., 2016; Perez et al., 2019). FtsA is essential and always co-localizes with FtsZ in Spn, whereas ΔsepF deletion mutants form elongated, widened cells lacking septal constrictions, suggesting that SepF polymers mediate FtsZ-ring closure during septation, possibly by acting as a curved clamp at the leading edge of the closing septum (Wenzel et al., 2021). Despite these phenotypes, ΔsepF mutants grow, similarly, to WT in BHI broth, and the cell morphology defects of ΔsepF mutants are largely reversed by FtsA overexpression (Mura et al., 2016). In contrast to SepF, nothing has been reported about ZapA function in Spn or other ovococcal bacteria.

Co-immunoprecipitation experiments detected ZapA in complex with FtsZ, FtsA, and EzrA in Spn cells (Supplementary Figure 25; Table 3), so we compared the spatiotemporal location of ZapA with that of FtsZ. ΔzapA mutants did not show any overt growth or cell morphology defects compared to WT in BHI broth or C+Y media (Supplementary Figure 27B; data not shown). We constructed a strain expressing FtsZ-Myc and ZapA-L-FLAG from their chromosomal loci (strain IU10752) and compared their localization to FtsZ-Myc and EzrA-L-FLAG³ (IU8681) by 2D EFm (Figure 10B and Supplementary Figure 26A). Unlike EzrA, which co-localizes with FtsZ and moves to the equatorial rings of daughter cells in stages 3 and 4 of the division cycle, ZapA remains at the midcell septal ring until late in cell division (Figure 10B). Moreover, quantitative comparisons of ring widths of C-terminal-tagged EzrA- EzrA-L-FLAG³ or ZapA-L-FLAG with FtsZ-Myc indicated that EzrA-rings are slightly larger than FtsZ-rings (Figure 10B and Supplementary Figure 26A), as discussed above (Supplementary Figure 2). In contrast, ZapA-rings have a considerably smaller width than FtsZ widths (Figure 10B and Supplementary Figure 26A), indicative of separate sublocations of EzrA and ZapA in FtsZ-rings.

Late arrival of ZapA at equators compared to FtsZ was confirmed by 3D-SIM of the FtsZ-Myc and ZapA-L-FLAG epitope-tagged strains (Supplementary Figure 26B) and separately by demograph analysis of 2D-EFm images of live cells expressing ZapA-sfGFP (strain IU10065) or FtsZ-sfGFP (strain IU9985) (Figure 10C). These results indicate that ZapA is not present in the nascent MapZ/FtsZ/FtsA/EzrA-rings that move out from the septum to the equators during Spn division (Perez et al., 2019) or in early equatorial rings. Consistent with this conclusion, the velocity of FtsZ treadmilling in nascent and early equatorial rings was the same in a ΔzapA mutant as the WT zapA⁺ strain (Supplementary Figure 27A). Similar to ZapA, 2D-EFm and 3D-SIM showed that SepF remains at septa throughout most of the division cycle and arrives late at the equators of daughter cells (Supplementary Figures 28A,C; Mura et al., 2016). However, unlike ZapA, the width of the SepF ring at the midcell is very similar to that of FtsZ (Supplementary Figure 28B). Together, these results are consistent with the idea that the pneumococcal FtsZ regulators EzrA, FtsA, SepF, and ZapA form a spatially ordered network, analogous to the one proposed for Eco (Buss et al., 2015).

As shown above (Figures 3, 4, 6), EzrA acts as a positive regulator of FtsZ-ring formation in Spn. We wanted to examine whether ZapA and SepF contribute to this positive regulation. ΔzapA mutants do not exhibit growth or cell morphology defects under the culture conditions tested (Supplementary Figure 27B), so we determined the effects on growth when EzrA is depleted in a ΔzapA or ΔsepF mutant. As noted above, depletion of EzrA results in bacteriostatic spheroid cells for long periods of time (Figure 2; Supplementary Table 3). Lack of ZapA (Figure 10A) or SepF (Supplementary Figure 28D) prevents prolonged growth upon EzrA depletion, whereas lack of MapZ does not (Supplementary Figure 28D). Depletion of EzrA in the ΔzapA or in the ΔsepF mutant gave mostly viable, irregularly shaped spheroid cells, typical of EzrA depletion in the WT strain (Figure 2B and Supplementary Figure 29). However, ΔsepF cells during depletion of EzrA appear larger than ΔzapA cells under EzrA depletion (Supplementary Figure 29), consistent with the increased length and width of ΔsepF mutants (Mura et al., 2016). By comparison, ΔzapA ΔsepF mutants grew, similarly, to ΔsepF single mutants, forming elongated and widened cells (data not shown; Mura et al., 2016). Finally, lack of ZapA did not alter the growth stoppage and lysis when FtsZ was depleted (Supplementary Figure 27B), but ΔzapA ΔmapZ mutants grew slightly slower than a ΔmapZ mutant (Supplementary Figure 27C). Together, these results are consistent with ZapA and SepF, playing accessory roles to EzrA in positively regulating FtsZ-ring formation, separate from MapZ, in Spn.

In Eco, ZapA interacts with a partner protein, called ZapB (Ebersbach et al., 2008; Galli and Gerdes, 2012). Spn does not encode an obvious ZapB homolog. NCBI blast search of the gene (spd_0370) immediately downstream from zapA in the Spn D39 genome identified a colicin V superfamily protein as a homologue, but not zapB(Eco). spd_0370 was annotated as zapB (Slager et al., 2018) in D39V, and as a colicin V superfamily protein (Lanie et al., 2007) in D39W, and is a homologue of yshB(Bsu), a gene downstream of zapA(Bsu) (BioCyc). To test experimentally whether Spd_0370 might be involved in Z-ring dynamics in Spn, we fused the N- and C-termini of Spd_0370 to GFP and to the FLAG epitope tag and determined its localization in Spn cells. 2D-EFm showed that Spd_0370 localized to the membrane but did not localize to the midcell divisome region, despite confirmed expression by Western blotting (data not shown). Given its lack of homology to Eco ZapB and its localization, we conclude that spd_0370 does not encode an FtsZ-ring regulator, despite its co-transcription with zapA (Slager et al., 2018).

To identify a partner of ZapA(Spn), we turned to an unbiased co-IP/MS approach. We formaldehyde-cross-linked Spn cells expressing ZapA-FLAG, which was then enriched on magnetic beads, and potential interactors were resolved by SDS-PAGE followed by silver staining as described in section “Materials and Methods” (Figure 11A). A single faint band with a molecular mass of about 24 kDa was present in the extract from cells expressing ZapA-FLAG, but absent from control cells lacking any FLAG-tagged proteins. The region between ≈21 and 31 kDa was excised from both lanes and subjected to peptide analysis by mass spectroscopy. Comparison of peptides from the ZapA-FLAG sample with the untagged ZapA⁺ control revealed that ZapA-FLAG pulled down the protein Spd_1350, which we renamed ZapJ (calculated molecular mass = 23,844 kDa and pI = 9.11). zapJ appears to be in a single-gene operon (Slager et al., 2018) and is located upstream of murC (Supplementary Figure 30A) that encodes UPD-N-acetylmuramate-alanine ligase, which catalyzes the third reaction in Lipid II precursor synthesis (Supplementary Figure 30A). ZapJ was putatively annotated as a cystathionine γ-synthase (Slager et al., 2018), which is inconsistent with the genetic relationships of ΔzapJ mutants that we observed below. B2H assays confirmed a direct interaction between ZapA and ZapJ and a self-interaction of ZapA (Figure 11B).

For every phenotype examined, ZapJ completely phenocopied ZapA. Similar to ΔzapA mutants, ΔzapJ mutants grew like WT and did not show cell morphology defects under the culture conditions tested (Figure 10A; data not shown). ZapJ fused at its C-terminus to Halotag (HT) or sfGFP localized to the midcell (Figure 11C and Supplementary Figure 30B), and ZapJ co-localized with ZapA in cells at different division stages (Figure 11C). Like ZapA, demograph analysis showed that ZapJ is a late-arriving protein to the equators of daughter cells (Figure 10C and Supplementary Figure 30C). EzrA depletion in a ΔzapA or ΔzapJ mutant rapidly prevented continued growth (Figure 10A), and the impaired growth of ΔzapA or ΔzapJ in ΔmapZ mutant backgrounds was indistinguishable (Supplementary Figure 27C). Together, these results support the hypothesis that ZapA and ZapJ interact and act as accessory positive regulators of FtsZ-ring formation in Spn. Finally, ZapJ is confined to Streptococci, which appear to lack a ZapB homolog, and zapJ is genetically closely linked to murC across Streptococcus species (Supplementary Figure 31).