FtsZ’s CTL is phosphorylated at S333 by PrkC

Previous phosphoproteomic screens identified two potential phosphosites in FtsZ’s CTL: S329 and S333 [37–39]. To test if FtsZ is phosphorylated in vivo, we harvested the lysate from log phase cells grown in S7 minimal medium and used PhosTag gels to resolve differentially phosphorylated species of fluorescently tagged FtsZ. For wild-type FtsZ, we observed the emergence of a second upper band, indicating that FtsZ is phosphorylated at one or more sites in vivo (Fig 1B, S2A). An FtsZ S329A-S333A double mutant showed complete loss of the upper band, confirming phosphorylation occurs at one of these residues (S2B Fig). To identify the specific phosphosite, we created FtsZ variants with alanine substitutions at each putative site. While the S329A mutant retained the upper phosphorylated band, the S333A mutant eliminated it (Fig 1B, S2C), demonstrating that FtsZ is singly phosphorylated at S333 under these conditions.

We next sought to identify the kinase responsible for this modification. While bacteria primarily use histidine kinases for signal transduction, some also encode a small number of Ser/Thr kinases that resemble eukaryotic kinases (so-called Hanks-type kinases) [29–32]. B. subtilis has four such kinases: PrkC, PrkD, YabT, and YrzF. We focused on PrkC because its homologs are known to phosphorylate FtsZ in other bacterial species (Table A in S1 Appendix) and it is expressed during vegetative growth [40]. Deletion of prkC eliminated the upper band (Fig 1B, S2D), indicating that this kinase is either directly or indirectly required for S333 phosphorylation in vivo.

S333 mutations alter FtsZ critical concentration and specific activity

To test whether S333 mutations affect FtsZ polymerization, we substituted S333 with glutamate (S333E) or alanine (S333A). Malachite green-based GTPase assays revealed that these mutations affected both FtsZ’s critical concentration and GTPase activity (S3A-D Fig). These mutations significantly reduced FtsZ’s critical concentration for assembly, from 0.51 ± 0.17 µM for wild-type to 0.41 ± 0.12 µM and 0.37 ± 0.14 µM for S333A and S333E, respectively (20% and 27% decreases, p < 0.001; mean ± 95% CI) (Fig 1C). The mutations also significantly enhanced FtsZ’s GTPase activity: compared to wild-type FtsZ’s rate of 4.0 ± 0.4 GTP/min/FtsZ, the S333A and S333E mutants showed significantly higher rates of 4.7 ± 0.2 and 5.1 ± 0.4 GTP/min/FtsZ, respectively (19% and 28% increases, p < 0.001; mean ± 95% CI) (Fig 1D). Taken together, these observations show that FtsZ’s polymerization and activity are sensitive to S333 mutations.

S333 mutations affect cell length

To determine whether S333 mutations affect cellular physiology, we measured cell lengths in strains expressing a modified ftsAZ operon (with wild-type ftsA and mutant ftsZ) as the sole copy from an ectopic locus under the native ftsAZ promoter (Paz, [41]) (S4A-C Fig). This design approximates native expression levels and ensures comparable protein levels across mutants, which is essential for phenotypic analysis since cell size is sensitive to FtsZ concentration. Western blot analysis confirmed that FtsZ protein levels were comparable across wild-type, S333A, and S333E strains (bWM200, bWM236, and bWM237, respectively; S4D Fig).

We measured cell lengths from fluorescence micrographs of membrane-stained cells. To accurately assess the impact of S333 mutations and control for batch effects, we analyzed multiple strains on each day, with measurements repeated both within and across days. We focused our analysis on cells within chains of three or more, as isolated cells and cell pairs showed systematically different length distributions. For each strain, this approach yielded approximately 100–300 cells per experiment.

In S7 minimal medium during early log phase, both S333A (bWM236) and S333E (bWM237) mutant cells were shorter than wild-type (bWM200): 2.37 ± 0.06 µm and 2.32 ± 0.07 µm compared to 2.49 ± 0.07 µm, respectively (5% and 7% decrease, p < .01, Student’s t-test; Fig 1E). The magnitude of these effects is notable, as previous studies have shown that doubling the amount of FtsZ only reduces cell length by 10% [42]. Our observation that a single mutation in the supposedly non-functional CTL can produce more than half of this effect size suggests strong involvement of the CTL in regulating cell division.

To test whether this phenotype was specific to minimal media conditions, we also measured cell lengths in CH rich medium. Surprisingly, S333 mutations produced the opposite effect in nutrient-rich conditions: S333A and S333E cells measured 2.79 ± 0.06 µm and 2.85 ± 0.17 µm compared to wild-type at 2.67 ± 0.14 µm (5% and 7% increase, p < 0.17 and p < 0.13, respectively; Fig 1F). This nutrient-dependent reversal of phenotype suggests that S333 phosphorylation modulates FtsZ’s response to different growth conditions, potentially through differential interactions with nutrient-dependent regulators of cell division such as UgtP [43].

FtsZ’s tail binds its core in vitro

Recent molecular dynamics simulations predicted that FtsZ’s tail might interact intramolecularly with its core, with the highest contact frequency in the region centered on S333 [34]. To test this prediction, we separately purified FtsZ’s globular core and tail (residues 1–315 and 316–382, respectively) and conducted in vitro binding assays using isothermal titration calorimetry (ITC) [44]. ITC experiments revealed specific, high-affinity binding between the FtsZ tail and core with 1:1 stoichiometry (K_d = 228 ± 24 nM, N = 1 ± 0.04; Fig 2A). Nuclear magnetic resonance (NMR) experiments provided additional clear evidence of complex formation, revealing striking changes in the ¹⁵N-labeled tail spectrum upon addition of unlabeled core (Fig 2B). These dramatic spectral changes are consistent with the sub-micromolar binding we measured via ITC. Together, these results demonstrate that FtsZ’s tail binds directly to its core with high affinity in vitro.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The FtsZ tail binds the core via the L330-H337 binding interface.

(A) Isothermal titration calorimetry demonstrates direct binding between the FtsZ tail (residues 316-382) and core (residues 1-315). The interaction shows high affinity (K_d = 228 ± 24 nM) and 1:1 stoichiometry. The continuous line represents the fit to a one-site binding model. (B) ¹H-¹⁵N HSQC NMR spectra of the 67-residue ¹⁵N-labeled tail alone (black) and in complex with unlabeled core (red, 1:1 molar ratio). Each peak represents a backbone amide, with its position and intensity reflecting its chemical environment. Spectra are displayed at equivalent contour levels. The dramatic changes in peak positions and intensities throughout the spectrum indicate extensive tail-core interaction. (C) Spectral changes for assigned residues in the S333 region upon core binding. Four regions from ¹H-¹⁵N HSQC NMR spectra show ¹⁵N-labeled tail alone (black) and bound to unlabeled core at 1:1 molar ratio (red). Blue dots mark the original peak positions in the free tail. Peak intensity changes reveal distinct binding behaviors: S333 shows dramatic intensity increase (141%); I334 and K335 show substantial decreases (32% and 40%); flanking residues L330, Q332, H337/N338, and V341 show modest changes (±10%). Quantification shown in S5B Fig. These response patterns indicate differential involvement of each residue in tail-core binding, with the S333-I334-K335 motif forming the binding hotspot.

https://doi.org/10.1371/journal.pone.0337820.g002

NMR identifies core binding residues on the tail

Our bioinformatic secondary structure predictions indicated α-helical propensity around S333, from L330 to H337 (Table B in S1 Appendix), in agreement with recent molecular simulations [34]. We therefore hypothesized that this region of the tail might function as an IDR molecular recognition element. To investigate this region, we synthesized a 15aa peptide spanning residues 328–342 for further NMR analysis, choosing these boundaries because proline residues typically disrupt helices. We collected ¹⁵N natural abundance ¹H-¹⁵N HSQC spectra of the 15aa peptide to observe its backbone amide peaks and used ¹H-¹H NOESY to assign their identities. To identify corresponding residues in the 67aa full-length tail, we overlaid the HSQC spectra of the 15aa peptide onto the 67aa tail. Peak overlap allowed us to assign eight residues in the 67aa tail spectrum: L330, Q332, S333, I334, K335, H337/N338 (overlapped peaks), and V341 (S5A Fig, S2 Table).

All eight assigned residues showed spectral changes upon binding (Fig 2C). To quantify these changes, we compared peak intensities between the free tail and the tail-core complex (S5B Fig, S3 Table). Three contiguous residues showed the most dramatic changes: S333’s peak intensity increased by 141%, while the adjacent I334 and K335 decreased by 32% and 40%, respectively. Neighboring residues showed more modest changes of around ±10% (Q332 and H337/N338 increased; L330 and V341 decreased). The striking intensity increase for S333 is consistent with this residue becoming more ordered upon binding, while the intensity decreases for I334 and K335 suggest these residues experience changes in their chemical environment, potentially through direct engagement with the core surface. Together, the S333-I334-K335 motif appears to form a binding hotspot at the center of the tail-core interface, with S333 potentially serving as a conformational switch flanked by I334 and K335 as key contact residues.

AlphaFold identifies binding residues on the core

To identify where these tail residues bind the core, we split FtsZ into its core and CTL sequences (315aa and 50aa, respectively) and co-folded them using AlphaFold Multimer [45,46]. Although the CTL is an IDR, AlphaFold consistently predicts that its S333 region binds the core on its C-terminal polymerization surface (Fig 3A, see Methods). Furthermore, out of the entire 50aa CTL, the S333-I334-K335 motif showed the highest prediction confidence relative to the core (lowest PAE, ~ 12Å; S6 Fig). Remarkably, this computationally predicted S-I-K site matches the one we identified via NMR.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Structural basis for phosphorylation-induced disruption of tail-core binding.

(A) AlphaFold model of the tail-core binding interface. AlphaFold Multimer predicts the tail’s C-terminal linker (CTL) binds the core’s C-terminal polymerization surface. Acidic residues (D210, D213) are shown in red, basic residues (K335, H337) in blue, tail hydrophobic residues (L330, I334) in salmon, and core hydrophobic residues (A279, A283, M292, I293, F294) in wheat. The S333 phosphosite is shown in cyan. (B) ITC demonstrates that S333 mutations abolish tail-core binding. Representative thermograms show no detectable interaction for S333E (magenta) or S333A (blue) tail variants when titrated into FtsZ core (K_d > 2 µM). (C-E) ¹H-¹H NOESY spectra reveal structural changes in the tail’s binding region in response to S333 modifications. Spectral strips show NOEs to (C) I334 HN, (D) K335 HN, and (E) K335 Hα for 15-residue peptides (P328-P342) with wild-type, S333A, S333E, or phospho-S333 sequences (left to right). (F) Fischer projection of the I334-K335 dipeptide showing newly appearing NOE connectivities as colored dashed lines: red (I334 Hγ2–HN intraresidue, based on panel C), green (K335 HN–I334 Hγ2 interresidue, based on panel D), and blue (K335 Hα–I334 Hγ12/Hγ13 interresidue, based on panel E). The K335 HN–Hβ2/Hβ3 intraresidue NOEs (purple) progressively coalesce across variants (panels C-E; S9 Fig). Conserved NOEs are not shown for clarity. The modifications induce progressive structural changes following an allelic series (WT < S333A < S333E ≈ pS333), with intensification of I334-K335 NOEs and emergence of new cross-peaks demonstrating that these residues collapse together. This compaction sequesters I334 and K335 from their core-binding partners, explaining the loss of tail-core binding observed in panel B.

https://doi.org/10.1371/journal.pone.0337820.g003

On the other side of the interface, AlphaFold consistently predicts that the CTL’s L330-H337 region interacts with two patches on the core: (i) α-helix H10/β-strand S9, and (ii) turn T7/α-helix H8. The predicted binding interface comprises three main features: (1) a hydrophobic interface, where I334 anchors into a pocket formed by M292, I293, and F294 on S9 (the “MIF pocket”), while L330 makes additional contacts on H10/S9; (2) an ionic interface, where K335 and H337 interact with acidic residues D213 (the GTPase catalytic residue) on H8 and D210 on T7, respectively; and (3) a polar interface, where N331, Q332, and S333 interact with residues across the H10-S9 and T7-H8 patches.

Since the CTL is directly attached to the core in full-length FtsZ, any predicted binding mode must be geometrically feasible as an intramolecular interaction. Though we modeled the core and CTL as separate proteins without any constraints on position or orientation, AlphaFold nevertheless converged on a binding mode that satisfies these geometric requirements. Using an X-ray crystal structure of the FtsZ core (PDB: 2VAM [47]), we measured the distance from G314 Cα (where the CTL exits the core) to I293 Cα (in the MIF pocket) as ~14 Å. This is well within reach of the 20aa linker connecting G314 to I334, which has an average end-to-end distance of 17–23 Å as a random coil (see Methods), providing ample length for the interaction.

These structural models extend our NMR findings in two important ways. First, they identify specific residues on the core for experimental validation: the MIF pocket (M292/I293/F294) and the acidic patch around the catalytic site (D210/D213). Notably, these residues map to the core’s C-terminal polymerization surface—the same surface where FtsZ monomers bind during filament assembly—suggesting that tail-core binding could regulate filament assembly by competing with monomer-monomer interactions. Second, the models reveal how phosphorylation might disrupt binding: S333 sits between hydrophobic anchors L330 and I334, and its phosphorylation would introduce a bulky negative charge that might prevent these residues from engaging the core.

Tail mutations abolish core binding

Having identified key binding residues via NMR and AlphaFold, we used ITC to test whether they are functionally required for binding. We started with the S333 phospho-site, which lies in the S-I-K motif at the center of the tail’s L330-H337 binding interface. Both phospho-mimetic (S333E) and phospho-dead (S333A) substitutions reduced tail-core binding below our detection limit (K_d > 2 µM; Fig 3B). This represents at least a 10-fold reduction in affinity compared to wild-type (K_d = 228 nM), showing that the S333 position is highly sensitive to modification.

We then targeted the hydrophobic anchor, I334, which AlphaFold predicts to insert into the MIF pocket. Substituting a bulky, charged residue at this position (I334E) also reduced binding below our detection limit (K_d > 2 µM; S7A Fig), confirming that hydrophobic packing is essential. Finally, we tested the ionic interface by mutating the two residues that AlphaFold predicts interact with D210/D213 on the core. A K335A/H337A double mutant also showed no detectable binding (K_d > 2 µM; S7A Fig), validating the importance of these predicted ionic interactions. These results confirm that the S-I-K motif identified by both NMR and AlphaFold is required for tail-core binding.

Core mutations abolish tail binding

Having demonstrated that specific tail residues are essential for binding, we next tested whether their predicted partners on the core are similarly required. We designed mutations targeting the hydrophobic and ionic interaction surfaces identified by AlphaFold. For the MIF pocket, we created a triple mutant (M292H/I293W/F294K) to introduce bulky and charged residues that would disrupt hydrophobic packing with I334. This mutation displayed aberrant binding with a biphasic isotherm (S7B Fig), confirming that the hydrophobic MIF pocket is essential for normal tail-core interaction. For the ionic interface, we neutralized the acidic patch (D210A/D213A) predicted to interact with K335/H337. This mutation weakened tail binding approximately two-fold, increasing the K_d from 228 ± 24 nM for wild-type to 501 ± 64 nM (S7C Fig), confirming that the D210/D213 acidic patch also contributes to the tail-core interaction. These observations confirm that the tail binds the core’s C-terminal polymerization surface, and they further pinpoint two specific structural elements of this interface: the hydrophobic MIF pocket and the D210/D213 acidic patch.

NMR reveals the structural basis of binding disruption

To understand the structural basis of binding disruption, we examined how S333 modifications affect the local structure of the tail’s binding region. We synthesized variants of our 15aa peptide (P328-P342) with S333A, S333E, and phospho-S333 modifications, and collected ¹H-¹H NOESY and ¹H-¹⁵N HSQC spectra to compare with the wild-type peptide characterized above (S8 Fig and S4–S6 Tables). Our analysis focused on understanding how modifications at position 333 affect the S-I-K motif that forms the core of the binding interface, as well as the broader L330-H337 region.

NOESY analysis revealed that S333 modifications cause the I334 and K335 sidechains to collapse and pack against each other, following a progressive series: WT < S333A < S333E ≈ pS333. Within individual sidechains, I334 showed dramatic emergence of the Hγ2-HN NOE that was essentially absent in WT but appeared strongly in all modified peptides (5-fold increase; Fig 3C), indicating the I334 γ-branch shifts from an extended conformation into closer proximity with the backbone. K335 displayed progressive coalescence of its Hβ2/Hβ3 peaks across peptide variants, with the characteristic doublet in WT merging into a single broad peak in pS333 (Fig 3D, S9), suggesting progressive restriction of the sidechain into a stable rotamer. Consistent with this, K335’s Hβ2-Hγ3 NOE intensity increased progressively across variants (S333A, + 34%; S333E, + 52%; pS333, + 61% relative to WT; S8 Fig), indicating the β and γ methylenes move closer together as the sidechain becomes more compact. Between the two sidechains, all S333 modifications drove I334 and K335 into closer proximity: K335 HN gained new NOEs to I334 Hγ2 in all three variants (Fig 3D); K335 Hα strengthened NOEs to I334 Hγ2 (Fig 3E); and K335 Hα gained new NOEs to I334 Hγ13 (A/E/pS) and Hγ12 (E/pS only) (Fig 3E). These patterns demonstrate that S333 modifications cause the extended I334 and K335 sidechains to collapse back toward the peptide backbone and pack against each other—sequestering them from their binding partners on the core.

To determine the full spatial extent of these structural perturbations, we analyzed HSQC spectra and calculated chemical shift perturbations (CSPs) for each residue relative to wild-type (S10A Fig, S5 Table). These analyses revealed backbone conformational changes spanning residues L330-H337 with the strongest effects on the S333-I334-K335 motif (S10B Fig). S333A caused the largest backbone perturbations at I334 and moderate perturbations at K335 (CSPs of 0.46 and 0.12, respectively), consistent with substantial local conformational reorganization following loss of the serine hydroxyl. In contrast, S333E and pS333 produced smaller backbone perturbations at I334 (CSPs of 0.09 and 0.08, respectively) and K335 (CSPs of 0.08 and 0.07, respectively), suggesting these two variants produce comparable changes in backbone conformation at the binding interface.

Peak intensity changes complemented the CSP analysis by revealing the dynamic consequences of these backbone conformational changes, with distinct stabilization patterns for each variant (S10C Fig, S5 Table). S333A primarily stabilized the L330-I334 region (~30% intensity gains at L330 and I334) while destabilizing H337-S340 (~20% losses), suggesting local hydrophobic stabilization from hydroxyl loss. In contrast, S333E and pS333 stabilized the broader L330-H337 region, with moderate gains at L330 and I334 (~10–30%) and dramatic stabilization of H337 (~30–40% gains). The I334 intensity gains across all variants corroborate the conformational restriction observed via NOESY (Fig 3C).

These complementary NMR analyses reveal how each variant achieves I334-K335 compaction through distinct mechanisms. S333A forms new hydrophobic contacts between Q332, A333, and I334, as evidenced by NOEs between Q332 Hβ2, Hβ3, and Hγ2 sidechain protons and A333 HN, as well as between A333 Hβ and I334 HN (S11A-B Fig). This local hydrophobic clustering drives I334-K335 compaction while destabilizing the C-terminal region. In contrast, S333E achieves comparable I334-K335 compaction (Fig 3C-E) without novel hydrophobic NOEs, likely through electrostatic effects that stabilize the entire binding element including H337. Phospho-S333 combines features of both: Q332 sidechain repositioning similar to S333A (S11C Fig) and stabilization of the L330-H337 region (S10C Fig), suggesting that S333E provides a conservative estimate of phosphorylation’s full structural impact.

Together, these NMR data demonstrate that S333 modifications disrupt the entire L330-H337 binding interface. Compaction of I334-K335 is the dominant structural change across all variants; based on our binding models and ITC results, this likely sequesters these two residues from their core partners. S333A achieves this compaction through hydrophobic contacts, while S333E and phospho-S333 achieve comparable I334-K335 compaction and stabilization of the broader L330-H337 region, likely through electrostatic mechanisms. These findings validate S333A and S333E as mechanistically distinct but functionally informative proxies for studying phosphorylation’s effects, with the structural gradient (WT < S333A < S333E) paralleling the functional disruption observed in our GTPase assays and cell length measurements. The structural similarities between S333E and phospho-S333 strongly suggest that phosphorylation would disrupt core binding at least as effectively as S333E, with the phosphate’s greater negative charge and steric bulk likely enhancing this disruption.

Tail-core binding modulates access to FtsZ regulators

The FtsZ core’s C-terminal polymerization surface is targeted by negative regulators of division. Having shown that the tail binds to this same surface, we wondered whether tail-core binding might regulate inhibitor access. To test this hypothesis, we focused on MinC, the classic regulator of division site placement. We noticed that the tail’s binding site on the core (H10/S9 and T7/H8) overlaps with the previously characterized binding site of MinC_N, the 92aa N-terminal domain of MinC [35] (Fig 4A-B, S12). We therefore used this protein fragment as a spatial probe in competitive binding experiments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The FtsZ tail binds the C-terminal polymerization surface and gates MinC_N access.

(A) Schematic representation of the FtsZ core highlighting the known MinC_N binding site (red) and the predicted FtsZ tail binding site (blue), which partially overlap. The tail is depicted with its core-binding region (L330-H337, purple) and the S333 phosphosite. (B) AlphaFold model of the MinC_N-FtsZ core interface, adapted from Machado et al. 2025 [35]. MinC_N is shown in pale green. Ionic interactions occur between MinC_N residues K12 and K15 (marine) and FtsZ residues D213 and D289 (red), respectively. MinC_N helix H1 (wheat) makes an additional hydrophobic contact with FtsZ helix H10 (salmon). (C) ITC demonstrates that the tail occludes MinC_N binding. MinC_N binds specifically to the isolated FtsZ core (magenta, K_d = 495 ± 110 nM, N = 1 ± 0.15) but shows no detectable binding to full-length FtsZ (blue, K_d > 2 µM), indicating the tethered tail blocks MinC_N access. (D) Competitive displacement experiments via ITC. The FtsZ tail can displace pre-bound MinC_N from the core (magenta), whereas MinC_N cannot displace pre-bound tail from the core (blue), demonstrating that the tail dominates access to the polymerization surface. (E) Model summarizing tail-mediated regulation of MinC_N access. The MinC_N binding site on the FtsZ core is denoted in red; the core-binding region of the tail is highlighted in purple. In its unphosphorylated state, the tail occupies the core’s polymerization surface, blocking MinC_N (pale green) binding. S333 phosphorylation releases the tail, exposing the MinC_N binding site.

https://doi.org/10.1371/journal.pone.0337820.g004

We performed ITC experiments, titrating MinC_N into either isolated FtsZ core (lacking its tail) or full-length FtsZ (including its tail). As expected, MinC_N exhibits specific binding to the tailless core (K_d = 495 ± 110 nM, N = 1 ± 0.15; Fig 4C), consistent with previous studies [35]. However, MinC_N binding was abolished when titrated into full-length FtsZ (K_d > 2 µM; Fig 4C). Since the tail is normally tethered to the core, this surprising result suggests two possibilities: either tail-core binding specifically occludes the MinC_N binding site, or the mere presence of the tail blocks MinC_N access through some other mechanism. A direct occlusion model is consistent with the tail’s two competitive advantages: two-fold higher affinity for the core (K_d = 228 nM vs 405 nM for MinC_N) and an enormous effective concentration advantage when tethered intramolecularly.

To distinguish between these possibilities, we tested whether our tail mutations that disrupt core binding would restore MinC_N access. MinC_N bound to full-length FtsZ variants containing S333A and S333E tail substitutions with affinities comparable to the tailless core (S13A-B Fig). On the other hand, the full-length K335A/H337A double mutant showed binding of multiple MinC_N molecules per FtsZ (S13C Fig), indicating significantly altered engagement of MinC_N with FtsZ. These results demonstrate that the wild-type tail directly occludes MinC_N’s binding site on the polymerization surface and can alter how molecules such as MinC_N engage FtsZ. Furthermore, they establish that each tail binds its own core intramolecularly (in cis), validating the physiological relevance of the trans binding we characterized in Fig 2A. This intramolecular tail-core interaction prevents binding of exogenous tails, as demonstrated by the inability of isolated wild-type tail to bind full-length FtsZ (K_d > 2 µM; S13D Fig).

To further probe the competitive relationship between the tail and MinC_N, we performed displacement experiments. First, we asked whether the tail could displace MinC_N from the core. We pre-formed a MinC_N:Core complex by mixing 45 µM MinC_N with 5 µM Core (9:1 ratio), achieving ~99% occupancy based on our measured K_d. Upon titrating Tail to a final concentration of 10 µM, we observed robust binding with 1:1 stoichiometry (Fig 4D). This demonstrates that the tail can completely displace MinC_N from the core, despite starting from a saturated complex.

We then tested the reciprocal experiment: can MinC_N displace a pre-bound tail? We pre-formed a Tail:Core complex using 30 µM Tail and 5 µM Core (6:1 ratio), again achieving ~99% occupancy. When we titrated MinC_N to 10 µM, we observed no detectable binding (Fig 4D). To ensure a fair comparison, we had adjusted the ligand:core ratios to account for their different affinities—using 9:1 for MinC_N:Core versus 6:1 for Tail:Core—so that both complexes started at equivalent occupancy. Under these conditions, the tail completely displaced MinC_N, while MinC_N could not detectably displace the tail. Together, these competition experiments demonstrate that the tail dominates access to the polymerization surface, effectively gating MinC_N’s access to the core (Fig 4E).

Culture conditions

Bacillus subtilis strains were stored as glycerol stocks at −80°C. For routine growth, cells were streaked onto Luria-Bertani (LB) agar plates containing appropriate antibiotics and incubated overnight at 37°C. For experimental cultures, cells were grown in the chemically defined S7 minimal medium [58] with the following modifications: 50 mM MOPS buffer (instead of the original 100 mM) [59], 20mM potassium glutamate (instead of sodium glutamate), and 1% (w/v) glucose.

Strain construction

Genetic constructs were generated using a multi-step assembly strategy involving polymerase chain reaction (PCR) [60], Overlap Extension PCR [61], and Gibson assembly [62,63]. Constructs were designed in Geneious [64] and typically involved anywhere from two to six fragments. Individual DNA fragments were amplified via standard PCR, then stitched together using Overlap Extension PCR to create 2–3 intermediate fragments of approximately equal size. These were assembled into complete genetic constructs via Gibson assembly, with approximately 1 kb homology arms at each end.

B. subtilis strains were derived from the prototroph PY79 [65]. To prepare competent cells, single colonies were inoculated into 1 mL of MC medium [66] and grown at 37°C with rolling aeration to an optical density (OD₆₀₀) of 1.10–1.30 (approximately 4 hours). Competent cells (200 µL) were transformed with either 20 µL of Gibson assembly reaction or 1 µL of purified genomic DNA and incubated with rolling at 37°C for 2 hours. Transformant cultures (50 µL) were then plated on appropriate selective media and incubated overnight at 37°C. Isolated colonies were streak-purified on selective plates with overnight incubation at 37°C. For long-term storage, single colonies were grown in LB medium to late logarithmic phase, and 20% glycerol stocks were prepared and stored at −80°C.

Strains generated for this study are listed in Tables C and D in S1 Appendix; oligonucleotides are listed in Table F in S1 Appendix. All strains were validated via Sanger (Genewiz) or Nanopore (Plasmidsaurus) sequencing.

Plasmid construction

Expression plasmids for protein purification were constructed as follows: Full-length B. subtilis FtsZ and variants were cloned into pET-28a with N-terminal 6xHis tags and TEV cleavage sites. FtsZ tail constructs (residues 316−382) were cloned into pGEX-6P-3 with N-terminal GST tags and PreScission protease cleavage sites. MinC_N was PCR amplified from B. subtilis genomic DNA and cloned into pGEX-6P-1 with an N-terminal GST tag and a PreScission protease cleavage site. Point mutations were introduced using the Q5 Site-Directed Mutagenesis kit (NEB) with primers designed using the NEBaseChanger tool. Detailed construction methods, including restriction sites, cloning strategies, and insert sequences, are provided in S1 Appendix. Plasmids generated for this study are listed in Table E in S1 Appendix; oligonucleotides are listed in Table F in S1 Appendix. All plasmids were validated via Nanopore sequencing (Plasmidsaurus).

Protein purification

One Shot® BL21 Star™ (DE3) Chemically Competent E. coli (Thermo) cells containing MinC_N or FtsZ constructs were inoculated in LB media containing 100 µg/mL kanamycin and grown overnight at 37°C with shaking at 200 rpm. In the morning, 20 mL of overnight culture was diluted into 2 L LB media containing 100 µg/mL kanamycin. Cells were allowed to grow at 37°C with shaking until reaching OD₆₀₀ ~ 1.2 where they were induced with 100 µM IPTG for 2 hours, after which they were pelleted at 4,000 × g for 20 minutes. A pellet corresponding to 2 L was resuspended in TKEG buffer (50 mM Tris pH 8, 50 mM KCl, 1 mM EDTA, and 5% glycerol) and sonicated using a 3/8” tip (200 Watt Ultrasonic Homogenizer) at a pulse-time setting with 60% amplitude for 8 rounds of 15 seconds on and 30 seconds off for each sample. Lysate was cleared by spinning at 21,000 × g for 20 minutes.

Lysate supernatant was added to Clontech Labs 3P TALON® Metal Affinity Resin and allowed to bind for 2 hours on a rocker at 4°C. Following binding, beads were washed twice with 100 mL of TKEG buffer on a column. Protein-bead complexes were placed in a 50mL Falcon tube in 10 mL of HMK buffer (50 mM HEPES, 100 mM KAc, 5 mM MgAc, pH 7.7) and 15 µL of Novex AcTEV Protease. The column was allowed to rock at 4°C for at least 8 hours prior to collecting the eluate.

For ¹⁵N-labeled FtsZ tail purification used in NMR studies, overnight cultures were inoculated and grown in LB up to an OD of 0.4 and pelleted at 4,000 × g for 20 minutes. Cell pellets were resuspended in pre-warmed M9 minimal medium and growth was continued as noted above.

Western blot analysis

Strains expressing FtsZ variants (Table C in S1 Appendix) were grown in 1 mL S7 minimal medium (modified with 50 mM MOPS and 50 µM iron citrate) at 37°C to mid-logarithmic phase (OD₆₀₀ ~ 0.5), normalized to OD₆₀₀ 0.05, grown again to OD₆₀₀ ~ 0.5, normalized again to OD₆₀₀ 0.05 in 5mL S7 medium, and grown to OD₆₀₀ 0.2. Pellets were harvested by centrifugation (5,000 × g, 10 minutes, 4°C), washed once with lysis buffer (20 mM Tris-HCl pH 7.5, 10 mM EDTA pH 8.0, 100 mM NaCl, 1 mM PMSF), and stored at −80°C overnight.

Cell pellets were resuspended in 50 µL lysis buffer supplemented with 1 mg/mL lysozyme, 10 µg/mL DNase I, and 100 µg/mL RNase A and incubated for 20 minutes at 37°C. Total protein concentration was determined using a NanoDrop 1000 spectrophotometer (Thermo) via A280. Lysates were mixed with Laemmli buffer containing 10% β-mercaptoethanol and 1 mM DTT, heated at 95°C for 10 minutes, and centrifuged (15,000 × g, 5 minutes, RT).

Protein samples (50 µg per lane) were loaded on a 10% SDS-polyacrylamide gel and separated at 120V for 1.5 hours. Proteins were transferred to a 0.2 µm nitrocellulose membrane (BioRad) using a Trans-Blot Turbo Transfer system (BioRad) at 25 V (1.0 A limit) for 30 minutes in Trans-Blot Turbo Transfer Buffer (BioRad). Membranes were stained with Ponceau S for 15 minutes at RT with shaking, rinsed five times with water until white, and imaged with a BioRad GelDoc. Membranes were then blocked with TBST (TBS with 0.1% Tween-20) with 5% non-fat milk for 30 minutes at RT, then incubated with rabbit polyclonal anti-FtsZ antibody (1:1,000 dilution; a gift from Jeff Errington) overnight at 4°C. After washing three times (10 minutes each) with TBST, membranes were incubated with HRP-conjugated goat anti-rabbit secondary antibody (1:10,000 dilution; LI-COR IRDye 680LT Goat anti-Rabbit IgG) for 2 hours at RT. After three TBST washes (10 minutes each), immunoreactive bands were imaged with a BioRad GelDoc. Band intensities were quantified using Python.

PhosTag gels

Strains expressing HaloTag-FtsZ variants (Table C in S1 Appendix) were grown in S7 minimal media (modified with 50 mM MOPS) with 20 µM IPTG at 37°C to OD₆₀₀ 0.5. Cultures were labeled with 50 nM HaloTag TMR (tetramethylrhodamine) ligand (Promega) for 30 minutes at 37°C (final OD₆₀₀ 0.7), harvested by centrifugation (5,000 × g, 5 minutes, 4°C), and stored at −80°C.

Cell pellets were lysed in 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 3 mM MgCl2, supplemented with protease and phosphatase inhibitors (1 mM PMSF, 50 mM NaF, 1 mM Na3VO4), 200 µg/mL lysozyme, and 1 µL SAN nuclease (ArcticZyme). After 30 minutes at 37°C, lysates were mixed with Laemmli buffer containing 10% β-mercaptoethanol and 1 mM DTT, heated at 95°C for 10 minutes, and centrifuged (15,000 × g, 5 minutes, RT).

Phosphorylation was analyzed using SuperSep Phos-tag gels (12.5%, Wako Chemicals). For each sample, eight 2-fold serial dilutions were prepared in sample buffer. Samples (10 µL) were separated alongside a molecular weight marker by electrophoresis in Tris-glycine-SDS running buffer at 15 mA (150V limit) for 3.5 hours. HaloTag-TMR labeled proteins were visualized using an Azure Sapphire biomolecular imager with the 520 nm laser line. Per-condition replicate counts were as follows: WT, n = 3; S333A, n = 2; ∆PrkC, n = 2; S329A-S333A, n = 1.

Microscopy: Culture conditions

Single colonies from LB overnight plates were inoculated into modified S7 medium supplemented with 50 µM iron citrate [67,68], and 4-fold serial dilutions were incubated overnight at 37°C with rolling aeration. This ensured all cultures underwent a minimum of 13 generations of growth before imaging, allowing cells to reach steady-state size after transitioning from stationary phase on LB plates to logarithmic growth in minimal medium [69]. The following morning, cultures that had reached an OD₆₀₀ between 0.10–0.30 were selected and further serial diluted into fresh S7 medium, and ten 2-fold serial dilutions were prepared for the day’s imaging. These cultures were grown at 37°C with rolling aeration for at least 4 hours prior to experimentation. All imaging was performed in early log phase, at OD₆₀₀ 0.12–0.25.

Microscopy: Sample preparation

Cultures were removed from 37°C and stained with FM 5-95 dye (1 µg/mL final, Thermo) by gentle swirling followed by 1 minute incubation at room temperature. Cells were immobilized by spotting 1 µL of stained culture onto a NaOH-washed coverslip and overlaying with a pad of S7 medium containing 2% agarose and FM 5-95 dye (0.5 µg/mL). Agarose pads were prepared by melting agarose in S7 medium, adding FM 5-95 dye, and casting between glass plates using plastic frames (1.5 cm × 1.5 cm × 1 mm) as spacers.

Microscopy: Flat field correction

Flat field correction was performed using 100 µM sodium fluorescein. For each correction set, 100 fields were collected without illumination (dark fields) and with 450 nm illumination through a FITC filter cube (flat fields). Dark and flat field composites were generated by averaging the respective 100-field sets. A pixel-wise gain map was then calculated as

Images were corrected by dark subtraction followed by multiplication with the gain map.

Microscopy: Epifluorescence imaging

Epifluorescence images were collected on a Nikon Ti-E microscope using a Nikon Plan Apo 100X Oil Ph3 objective (1.40 NA) with an Andor Clara DR-2136 CCD camera. FM 5-95 was excited at 550 nm using a CoolLED pE-300 Ultra LED source and imaged through a TRITC filter cube modified with an ET610lp emission filter (Chroma) to account for FM 5-95’s unusually large Stokes shift.

Cell length measurement

Images were segmented using Omnipose v0.5 [70], followed by manual correction of segmentation errors in Napari v0.4 [71]. Cell lengths were measured using custom Python code: Adjacent cell masks were merged into chain masks. Chain and cell boundaries were extracted from their masks using the marching squares algorithm, then refined with an active contour model aligning with membrane signals. Chain midlines were constructed by finding the longest path through Voronoi ridge vertices interior to the chain boundary. Chain midlines were then bisected with cell boundary contours to obtain cell midlines. Cell lengths were determined by measuring along their midline skeletons.

Analysis was restricted to cells in chains of three or more, as single and chained cells exhibited systematically different length distributions. To control for batch effects and biological variability, each strain was imaged multiple times, both within and between days. To account for the hierarchical nature of this experimental design, each replicate (approximately 100–300 cells) was collapsed to its median cell length. Each strain was represented by the arithmetic mean of these replicate medians, and statistical significance was assessed using a two-tailed Student’s t-test performed on the replicate medians.

AlphaFold analysis

The FtsZ core (residues 1–315) and CTL (residues 316–365) sequences were co-folded using AlphaFold2-Multimer v2.3 followed by Amber relaxation [45,72]. 100 models were generated (20 random seeds × 5 models per seed). Predicted aligned error (PAE) matrices were analyzed to assess prediction confidence at the core-CTL interface. PAE is shown as both a pairwise distance matrix and as a line plot of minimum PAE values per residue (S6 Fig). Interface residues were identified based on proximity (≤3 Å) between core and CTL atoms. Structural analysis and visualization were performed using PyMOL [73].

Analysis of geometric constraints on intramolecular binding

To measure the distance from the tail exit point to the center of the MIF pocket, we used PyMOL’s distance command to measure the Euclidean Cα-Cα distance between G314 and I293 on an X-ray crystal structure of the FtsZ core (PDB: 2VAM [47]). To estimate the reach of the 20aa linker (G314-S333), we applied the polymer scaling relationship R = R₀ × N^ν, where R₀ = 3.8 Å (Cα-Cα distance per residue), N = 20 residues, and ν = 0.5 (ideal chain) or 0.6 (self-avoiding chain), yielding a characteristic end-to-end distance of 17–23 Å.

Measurement of GTPase activity

FtsZ GTPase activity was measured using a malachite green phosphate assay. Proteins were purified via His6 affinity purification and eluted in HMK buffer (50 mM HEPES, 5 mM magnesium acetate, 100 mM potassium acetate, pH 7.7). Cellular phosphate was removed via buffer exchange into HMK buffer using Sephadex G-25 columns (Cytiva). Protein concentration was determined using a Pierce BCA Protein Assay Kit (Thermo). FtsZ was diluted to an initial concentration of 4 µM, followed by linear serial dilutions in one-eighth steps in HMK buffer.

FtsZ dilutions were kept on ice. GTPase reactions were initiated by the addition of GTP to a final concentration of 1 mM in a total reaction volume of 50 µL. Reactions were incubated at 37°C for 15 minutes in a PCR thermocycler. Following incubation, reactions were immediately quenched by 20-fold dilution into ice-cold 10 mM EDTA (10 µL of reaction into 190 µL of EDTA). The diluted samples were split into duplicate 100 µL aliquots. Using a BioMek FXp liquid handling system (Beckman Coulter), 80 µL of each diluted sample was simultaneously transferred into 20 µL of malachite green reagent (Malachite Green Phosphate Assay Kit, MAK307, Sigma) in a 96-well clear-bottom plate. Plates were incubated for 30 minutes at RT, during which the plates were spun at 3,000 × g for 15 minutes to eliminate bubbles. Absorbance was measured at 620 nm using an Epoch2 microplate spectrophotometer (BioTek). Phosphate concentrations were determined using phosphate standards (0–40 µM) prepared in 10 mM EDTA to match the buffer used for reaction quenching and sample dilution.

The phosphate production rate was calculated by dividing the total measured phosphate by the reaction time (15 minutes) and multiplying by the dilution factor (20). Since hydrolysis of one GTP molecule produces one inorganic phosphate, this phosphate production rate directly reflects the rate of GTP hydrolysis. For each FtsZ variant, GTPase rate was plotted against protein concentration. The slope of this line represents the specific activity (GTP/min/FtsZ), and the x-intercept defines the critical concentration (µM). Each variant was analyzed in a minimum of 3 independent experiments run on different plates.

Statistical analysis of GTPase activity and critical concentration

Data analysis was performed in Python. For both GTPase activity and critical concentration, bootstrap analysis was performed by randomly selecting one measurement at each concentration and performing linear regression. This process was repeated 1,000 times to generate distributions of slopes and x-intercepts. Means and 95% confidence intervals were calculated from these distributions. Statistical significance was assessed by comparing bootstrap distributions between conditions using two-tailed Student’s t-tests, with differences considered significant at p < 0.001.

Isothermal titration calorimetry

ITC experiments were performed in TKEG buffer (50 mM Tris pH 8, 50 mM KCl, 1 mM EDTA, and 5% glycerol) at 25°C. Full-length FtsZ or FtsZ core were diluted to 5 µM; FtsZ tail and MinC_N were diluted to 50 µM. For competition experiments, protein complexes were pre-formed by mixing ligand and core at ratios calculated to achieve ~99% occupancy based on measured K_d values, then immediately loaded into the ITC cell. For MinC_N displacement: 45 µM MinCN + 5 µM Core (9:1). For Tail displacement: 30 µM Tail + 5 µM Core (6:1).

ITC experiments were performed on a TA Instruments Affinity ITC by loading full-length FtsZ, FtsZ core, or pre-formed complexes into the cell, and FtsZ tail or MinC_N into the syringe. The cell temperature was equilibrated to 25°C with a stirring speed of 125 rpm and an initial delay of 600 seconds. Titrants were injected in 2 µL increments for a total of 20 injections with 180 second intervals between injections. Baseline correction, binding curve fitting, and parameter calculation were performed using NanoAnalyze. All ITC experiments were performed in at least duplicate. Binding parameters for all replicates are provided in S1 Table.

Nuclear magnetic resonance spectroscopy

NMR spectra were acquired on a Bruker Avance III HD 800 MHz spectrometer equipped with a TCI cryoprobe at 298K. Samples were prepared in 50 mM HEPES-d₁₈ (98% deuterated) pH 7.5, 50 mM KCl, 1 mM EDTA in 90% H₂O/10% D₂O. Chemical shifts were referenced to DSS. Spectra were processed with NMRPipe [74] and analyzed with Poky [75].

NMR: Data processing

Time-domain data were zero-filled and apodized with squared sine bell functions prior to Fourier transformation. For consistent quantitative comparison across samples, phase corrections were systematically optimized. NOESY spectra were phased to minimize row-wise intensity variance in the fingerprint region (7.85–8.50 ppm in F2; 0.5–5.0 ppm in F1), reducing artifacts from imperfect phase correction. HSQC spectra were phased to maximize the ratio of positive to negative signal intensity in the backbone amide region (7.85–8.50 ppm in F2; 113–127 ppm in F1).

NMR: Peptide variant experiments

Synthetic 15aa peptides corresponding to the FtsZ tail’s core-binding region (P328-P342) were synthesized and HPLC-purified to >95% purity at the Tufts University Core Facility (TUCF). Four variants were prepared: wild-type, S333A, S333E, and phospho-S333. Natural abundance ¹H-¹⁵N HSQC and ¹H-¹H NOESY spectra were collected. NOESY experiments used a mixing time of 120 ms with excitation sculpting for water suppression. Backbone amide resonances were assigned using sequential NOE walks and amino acid-specific patterns in NOESY spectra, using characteristic chemical shift values from the BMRB database [76] to predict approximate peak locations for each residue type. These assignments were then transferred to HSQC spectra via shared ¹H chemical shifts. HSQC peak assignments are provided in S4 Table. NOESY peak assignments are provided in S6 Table.

NMR: Chemical shift perturbation analysis

Chemical shift perturbations (CSPs) were calculated to quantify backbone structural changes upon S333 modification. To correct for sample-to-sample variation, spectra were aligned using three reference residues distant from the modification site (S329, Q339, S340). For each variant, reference residue Δδ_H were calculated relative to wild-type, averaged to obtain a global offset, and subtracted from all peaks. After alignment, CSPs were calculated for each variant peptide (S333A, S333E, and phospho-S333) relative to wild-type using the standard formula:

where Δδ_H and Δδ_N represent the chemical shift differences in the ¹H and ¹⁵N dimensions, respectively. The scaling factor of 5 accounts for the difference in chemical shift dispersion between ¹H and ¹⁵N nuclei. All CSP values are provided in S5 Table.

NMR: Peak intensity normalization

To account for differences in peptide sample concentration and acquisition time, peak heights were normalized using the median of selected peaks. For HSQC spectra, all backbone amide peaks excluding positions 333 (modification site) and 337/338 (overlapped peaks) were used. For NOESY spectra, diagonal peaks were excluded due to their high intensity and susceptibility to overlap artifacts. All off-diagonal intraresidue (i,i) and sequential (i,i + 1) NOEs observed consistently in all four samples were used. For NOEs annotated on both sides of the diagonal, peak heights were averaged to reduce measurement noise. Peak heights were divided by these median values and divided by 10⁶ for convenient numerical display.

For spectral display, contour levels were adjusted to maintain comparable signal across samples. A baseline contour threshold was set just above the noise floor in the wild-type spectrum (as determined by visual inspection), then divided by each sample’s normalization factor to establish equivalent contour thresholds for variant peptides. Contour level multipliers were kept constant at 1.4 across all spectra. Normalization factors and resulting contour thresholds are provided in S7 Table.

NMR: High-resolution NOESY cross-sections

To visualize changes in K335 Hβ2/Hβ3 cross-peak patterns, NOESY spectra were reprocessed with zero-filling to 4,096 points in the indirect dimension (8 × increase) to enhance digital resolution. One-dimensional cross-sections were extracted at the K335 HN chemical shift using nmrglue [77] and normalized as described above.

NMR: Tail-core binding experiments

¹⁵N-labeled FtsZ tail (67aa) and unlabeled FtsZ core (315aa) were purified. ¹H-¹⁵N HSQC spectra were collected using a TROSY-based pulse sequence with WATERGATE water suppression. Spectra were acquired for ¹⁵N-labeled tail alone and in complex with unlabeled core at a tail:core molar ratio of 1:1.

NMR: Tail-core binding quantification

To quantify changes in the 67aa tail upon core binding, peaks were manually picked using Poky [75]. Tail-core peaks were matched to the nearest free tail peak by CSP distance (threshold: 0.10 ppm; ties resolved by closest match). Spectra were normalized using median heights of stable reference peaks (defined by |log₂ fold-change| < 0.5), with all intensities scaled relative to free tail. Chemical shift perturbations and intensity ratios were calculated for each paired peak. Residues were assigned by matching tail peaks to peptide peaks by CSP distance (threshold: 0.05 ppm; ties resolved by closest match), yielding eight assigned residues: L330, Q332, S333, I334, K335, H337/N338, and V341. All peak assignments are provided in S2 Table; binding analysis is provided in S3 Table.

Software and data analysis

All data processing, computational image analysis, and statistical analysis was performed using Python 3 [78] with the NumPy [79], SciPy [80], Pandas [81], and Scikit-Image [82] libraries. Plots were generated using Matplotlib [83] and Seaborn [84]. Gels were analyzed using Fiji [85]. NMR spectra were processed with NMRPipe [74], analyzed with Poky [75], and visualized with nmrglue [77]. Protein structural models were generated with AlphaFold [72] and visualized with PyMOL [73].