All expression vector constructions were performed using the Gibson assembly method (Gibson et al., 2009). We prepared the backbone vectors from previously utilized single expression vectors, pPICZ A-AtFtsZ1_FL-mCerulean and pPICZ A-AtFtsZ2_FL-eYFP-MTS (Yoshida et al., 2016). These were then digested with EcoRI/XbaI or EcoRI/KpnI to excise the AtFtsZ1_FL or AtFtsZ2_FL fragment. For expression of the full-length FtsZ proteins in Pichia, the predicted transit peptides were excluded: AtFtsZ2 and AtFtsZ1 (1–48 aa and 1–57 aa, respectively), GsFtsZA and GsFtsZB (1–52 aa and 1–57 aa, respectively). The conserved core regions (CCRs) for FtsZs were defined as follows: AtFtsZ2_C (119–424 aa), AtFtsZ1_C (74–377 aa), GsFtsZA_C (121–424 aa), and GsFtsZB_C (91–404 aa). All the primers used in this study were listed in Supplementary Table S3 .

To generate G. sulphuraria FtsZA and FtsZB constructs, synthesized GsFtsZA and GsFtsZB sequences, which were codon-optimized for A. thaliana for better comparison to their Arabidopsis counterparts (TerBush et al., 2018), were used as templates. Full-length GsFtsZs were amplified using primer sets LY33/LY34 and LY242/LY244, while the CCRs were amplified using primer sets LY36/LY37 and LY243/LY246. The obtained fragments were inserted into the EcoRI/XbaI digested pPICZ A-mCerulean backbone. Similarly, the CCRs of AtFtsZ2 and AtFtsZ1 were amplified using LY332/LY333 and LY330/LY331, and were inserted into the digested pPICZ A-mCerulean backbone. To obtain constructs expressing MTS, the EcoRI/KpnI digested pPICZ A-eYFP-MTS was used as backbone. The primer sets LY332/LY335, LY330/LY334, and LY34/LY35 were used to amplify corresponding fragments in order to yield pPICZ A-AtFtsZ1_C-eYFP-MTS, pPICZ A-AtFtsZ2_C-eYFP-MTS, and pPICZ A-GsFtsZA_FL-eYFP-MTS.

The construction of the co-expression plasmid pPICZ A-AtFtsZ1_FL-mCerulean-AtFtsZ2_FL-eYFP-MTS was described previously (Yoshida et al., 2016). To generate pPICZ A-AtFtsZ1_C-mCerulean-AtFtsZ2_C-eYFP-MTS, the AtFtsZ1_C-mCerulean expression cassette was amplified using primers CC229/LY190 and then inserted into the BglII digested pPICZ A-AtFtsZ2_C-eYFP-MTS backbone. To obtain pPICZ A-GsFtsZA_FL-mCerulean-GsFtsZB_FL-eYFP, we first amplified the GsFtsZB_FL-eYFP-MTS expression cassette using primers CC229/LY190 and then inserted into the BglII digested pPICZ A-GsFtsZA_FL-mCerulean backbone. The obtained plasmid was further digested with BamHI/KpnI to excise the eYFP-MTS fragment and thus served as backbone. Finally, we used the primer pairs LY201/LY328 and LY329/LY192 to amplify the eYFP expression cassette lacking MTS, and inserted the cassette into the BamHI/KpnI digested backbone to yield pPICZ A-GsFtsZA_FL-mCerulean-GsFtsZB_FL-eYFP. Similar strategy was adopted to generate pPICZ A-GsFtsZA_FL-mCerulean-GsFtsZB_C-eYFP. In brief, the GsFtsZB_C-eYFP-MTS expression cassette was first amplified using primers CC229/LY190 and the obtained fragment was inserted into the BglII digested pPICZ A-GsFtsZA_FL-mCerulean backbone. The obtained plasmid was further digested with BamHI/KpnI to excise the eYFP-MTS fragment and used as backbone. The primer pairs LY410/LY328 and LY329/LY192 were used to amplify the eYFP expression cassette lacking MTS, and the resulting fragment was inserted into the digested backbone to obtain pPICZ A-GsFtsZA_FL-mCerulean-GsFtsZB_C-eYFP.

To generate the chimeric protein expression vector pPICZ A-AtFtsZ2_NT-Z1_C-Z2_CT-mCerulean, primer pairs of LY336/LY337, LY338/LY339 and LY340/LY341 were used to amplify fragments of AtFtsZ2_NT , AtFtsZ1_C and AtFtsZ2_CT , and then Gibson assembled into the digested pPICZ A-mCerulean backbone. Similarly, LY336/LY343, LY344/LY345, LY346/LY347 were used for amplification fragments of AtFtsZ1_NT , AtFtsZ2_C and AtFtsZ1_CT , and then Gibson assembled into the digested pPICZ A-mCerulean backbone to yield pPICZ A-AtZ1_NT-Z2_C-Z1_CT-mCerulean construct.

To obtain the pET11b-Z2NTZ1_CZ2CT vector, the fragment of Z2NTZ1_CZ2CT was amplified using the primer pair LY253/LY252 and then assembled into the BamHI/NdeI digested pET11b vector. Likewise, the fragment of Z1NTZ2_CZ1CT was amplified using the primer pair LY251/LY254 and then assembled into the BamHI/NdeI digested pET11b vector to yield pET11b-Z1NTZ2_CZ1CT.

Pichia pastoris (P. pastoris) X-33 strain was used in this study. The pPICZ A expression vector (Invitrogen), harboring the genes of interest, was transformed into the X-33 strain and integrated into the host genome via homologous recombination. The transformation of Pichia was conducted as previously described, with minor modifications (Chen et al., 2019). Briefly, 10 μg of the pPICZ A vector was linearized using enzyme restriction and introduced into the X-33 strain via electroporation. The pPICZ A vector contains a phleomycin resistant gene for selection in both Escherichia coli (E. coli) and P. pastoris. To select positive transformants in Pichia, 10 μg mL^-1 phleomycin (InvioGen) was added to YPDS (1% yeast extract, 2% peptone, 1 M sorbitol, 2% dextrose) plates containing 2% agar.

The pPICZ A vector features an inducible AOX1 promoter, which facilitates distinct levels of protein expression. During the growth phase, proteins are minimally expressed in BMGY medium (1% yeast extract, 2% peptone, 100 mM potassium phosphate at pH 6.0, 1.34% YNB, 0.00004% biotin, 1% glycerol), where glucose serves as the sole carbon source. In contrast, protein expression is induced upon transfer to BM medium (1% yeast extract, 2% peptone, 100 mM potassium phosphate at pH 6.0, 1.34% YNB, 0.00004% biotin), which lacks a carbon source.

To assess the expression of FtsZ proteins in P. pastoris, transformed Pichia cells were streaked out onto YPD (1% yeast extract, 2% peptone, 2% dextrose) plates. A single colony was selected, pre-cultured in 2.5 mL of BMGY medium overnight, and then centrifuge at 1,000 g for 2 min at room temperature. The pellet was resuspended in 500 μL of BM medium, and the culture was grown continuously in a shaker at 30°C until the indicated time.

To express the recombinant chimeric His-Z2NTZ1_CZ2CT and His-Z1NTZ2_CZ1CT proteins, we transformed the corresponding vectors into E. coli DE3 Rosetta cells. The transformed bacterial cells were cultured in a shaker at 37°C overnight. The following day, they were subcultured into fresh LB medium and allowed to grow at 37°C until the OD₆₀₀ reached between 0.6 and 0.8. Subsequently, the cultures were subjected to an ice shock for 10 min before the addition of Isopropyl β-D-1-thiogalactopyranoside (IPTG; 0.6 mM) to induce protein expression. After 36–42 hours of induction at 14°C, the cells were harvested by centrifugation and re-suspended in 20 mL of low salt buffer (LSB: 20 mM Tris pH 7.5, 50 mM NaCl, 10% glycerol). The re-suspended cell pellets were then stored at -80°C for further use.

Purification of the chimeric proteins was conducted according to the method previously described (Porter et al., 2021). In brief, the harvested cells were thawed and lysed using 1 mg/ml Lysozyme (Lab Scientific, Highlands, NJ) for 30 min at 4°C. After lysis, the cells were sonicated to further rupture the cells and release the expressed proteins. Following centrifugation to pellet the cell debris, the supernatant containing the soluble protein was loaded onto a Ni-NTA column (Qiagen). The column was then washed with a gradient of imidazole concentrations (from 20 to 50 mM) in LSB (20 mM Tris pH 7.5, 50 mM NaCl, 10% glycerol) to remove unbound proteins. The chimeric proteins were eluted using 300 mM imidazole in LSB. The eluted proteins were dialyzed against LSB to exchange the buffer and remove imidazole. Finally, the purified proteins were aliquoted and stored at -80°C for long-term preservation.

GTPase activity was measured following the procedures from Ingermann and Nunnari, with slightly modifications (Ingerman and Nunnari, 2005). The assay fundamentally measured the depletion of NADPH at 340 nm. The consumption of one NADPH molecule facilitates the regeneration of a GTP from a GDP, thereby reflecting the GTPase activities. All proteins used in the experiments were centrifuged at 80,000 g for 30 min at 4°C prior to the assay. In brief, a total volume of 180 µL was prepared, containing the desired protein concentrations in the reaction buffer (1 mM phosphoenolpyruvate, 0.4 mM NADH, and 20 U/mL pyruvate kinase/lactate dehydrogenase, in 50 mM HEPES–KOH, pH 7.5, 5 mM MgSO4, and 100 mM KCl). Finally, 20 µL of GTP (Sigma) was added to initiate the monitoring of absorbance at 340 nm using a SpectraMax M2 microplate reader (Molecular Devices). The GTPase activities were calculated according to the previous study (Porter et al., 2021).

All images were collected at room temperature using a spectral-based FluoView 1000 laser scanning confocal microscope (Olympus) equipped with a UPlanSApo 100× (NA 1.40) oil immersion objective. FV1000 ASW software (Olympus) was used to capture either snapshots or time-lapse images during the FRAP experiment. For proteins fused with eYFP in single expression cells, a 515 nm laser was utilized, while for proteins fused with mCerulean, a 458 nm laser was utilized. FRAP analysis was conducted as previously described (Yoshida et al., 2016; Chen et al., 2019), with slight modifications. Briefly, three images with a 10 s interval were taken as the pre-bleached control, and then a ROI (Region of Interest) of 20 pixels in diameter was selected for photobleaching for 20 ms using the Tornado scanning tool within the FV1000 ASW software. Recovery of the fluorescence signals was monitored at 10 s intervals for a total of 260 s. To correct and normalize the FRAP data, ROIs with the same size were recorded during each experiment: one at a region of the fluorescence signal away from the photobleached spot to account for photobleaching due to continual acquisition, and another at a region of the background signal to account for random noise. Curve fitting was performed with ProFit 7 software. The data were fit to the function of a two-binding-state model (Sprague et al., 2004; Yoshida et al., 2016): f(t) = (1 - r) (1 - C_eq1 e^-koff1*t - C_eq2 e^-koff2*t ), where t is time (s), C_eq1 and C_eq2 refer to the fractions of bound molecules, k_off1 and k_off2 refer to dissociation rate constants, and r is an additional parameter to account for incomplete recovery (Sprague et al., 2004). The corresponding parameters for each FRAP data were summarized in Supplementary Tables S1 , S2 . All image processing was conducted in Fiji.

The GenBank accession numbers for the FtsZ genes are as follows: cDNAs of A. thaliana FtsZ2 (AF089738) and A. thaliana FtsZ1 (AY113896) and G. sulphuraria FtsZA (BBAA82099), and G. sulphuraria FtsZB (BAA82091).

The conserved core regions of AtFtsZ1 and AtFtsZ2 largely determine the distinct assembly properties of the purified proteins in vitro (Porter et al., 2021). We therefore asked whether the core regions govern the unique turnover dynamics of their assembled filaments in living cells. To test this, we fused mCerulean (mC) to the C-termini of full-length AtFtsZ1 (AtFtsZ1_FL) and AtFtsZ2 (AtFtsZ2 _FL) lacking the TPs ( Figure 1a ) to create AtFtsZ1_FL-mC and AtFtsZ2_FL-mC, as well as to their core regions (AtFtsZ1_C and AtFtsZ2_C) (Porter et al., 2021) to create AtFtsZ1_C-mC and AtFtsZ2_C-mC. The fusion proteins were then expressed separately in P. pastoris. Throughout this paper, we describe data for AtFtsZ2 before AtFtsZ1 because AtFtsZ2 is a more typical FtsZ as described above (TerBush et al., 2013), and to facilitate comparisons between the Arabidopsis and red algal FtsZ proteins in experiments described below.

In control experiments, AtFtsZ2_FL-mC and AtFtsZ1_FL-mC both formed filaments in P. pastoris resembling those reported previously (Yoshida et al., 2016). AtFtsZ2_FL-mC filaments looped around the cell and were generally longer and more jagged in appearance than AtFtsZ1_FL-mC filaments, which were shaped like shallow arcs ( Figures 1b, c ). AtFtsZ2_C-mC and AtFtsZ1_C-mC both formed arc-shaped filaments ( Figures 1d, e ). The latter results validated that the core regions alone are sufficient for assembly of both AtFtsZs in living cells (TerBush et al., 2016).

To compare the turnover dynamics of the filaments formed in P. pastoris, we performed fluorescence recovery after photobleaching (FRAP) experiments. Recovery of fluorescence into the bleached region was monitored for 260 s and a two-binding-state equation was used to fit the recovery curves (Sprague et al., 2004; Yoshida et al., 2016). Because a half-time of recovery cannot be derived from such curves, we instead used the percent of fluorescence recovered 130 s after photobleaching (R₁₃₀) to allow direct statistical comparisons of dynamics between different FtsZ filaments, and indicated the statistics in Supplementary Figure S2 (Arabidopsis) and Supplementary Figure S6 (red alga). The R₁₃₀ values for AtFtsZ2_FL-mC and AtFtsZ1_FL-mC were 11% and 19%, respectively ( Figures 2a-c ; Supplementary Figure S2A ), indicating that AtFtsZ2_FL filaments are significantly less dynamic than AtFtsZ1_FL filaments, as shown previously in S. pombe (TerBush and Osteryoung, 2012). Similarly, AtFtsZ2_C-mC (R₁₃₀ = 17%) was significantly less dynamic than AtFtsZ1_C-mC (R₁₃₀ = 27%) ( Figures 2d–f ; Supplementary Figure S2A ), revealing that the conserved core regions contribute substantially to the differences between AtFtsZ2 and AtFtsZ1 turnover dynamics. R₁₃₀ was significantly lower for AtFtsZ2_FL-mC (11%) than AtFtsZ2_C-mC (17%) ( Figures 2b, e ; Supplementary Figures S2A , S3A ), suggesting the N- and/or C-terminal flanking regions may constrain the turnover of FtsZ2 filaments. R₁₃₀ was also lower for AtFtsZ1_FL-mC (19%) than AtFtsZ1_C-mC (27%) ( Figures 2c, f ), but the difference was not statistically significant ( Supplementary Figure S2A ). However, the fluorescence recovery curve of AtFtsZ1_FL-mC exhibited a higher recovery trend than that of AtFtsZ1_C-mC ( Supplementary Figure S3B ), implying that the N- and/or C-terminal flanking regions may also constrain the turnover of FtsZ1 filaments.

To study constriction of Escherichia coli FtsZ in vitro, Osawa et al. (2008) fused a membrane-tethering sequence (MTS) to the C-terminus of an FtsZ-eYFP fusion protein, which allowed it to form a membrane-tethered contractile ring inside liposomes. The chloroplast membrane proteins that interact with AtFtsZ2 are lacking in P. pastoris, but the same MTS also enabled full-length AtFtsZ2 fused to eYFP (AtFtsZ2_FL-eY-MTS) to form a contractile ring in P. pastoris that was attached to the plasma membrane, whereas AtFtsZ2_FL without the MTS was not attached (Yoshida et al., 2016). We used AtFtsZ2_FL-eY-MTS and constructed AtFtsZ2_C-eY-MTS to explore the effect of membrane tethering on the assembly and dynamics of the full-length and core AtFtsZ2 proteins. Unlike the proteins lacking the MTS ( Figures 1b, d ), both AtFtsZ2_FL-eY-MTS and AtFtsZ2_C-eY-MTS assembled into well-defined closed rings ( Figures 1f, g ). FRAP experiments showed that the R₁₃₀ values for AtFtsZ2_FL- eY-MTS and AtFtsZ2_C-eY-MTS were 9.5% and 8.7%, respectively ( Figures 2g–i ; Supplementary Figure S2B ). Statistical analysis demonstrated that the difference between the dynamics of AtFtsZ2_FL-eY-MTS and AtFtsZ2_C-eY-MTS was not significant ( Supplementary Figure S2B ), further indicating that AtFtsZ2 dynamics is dominated by its core region. To further explore the effect of membrane tethering on the assembly and dynamics of AtFtsZ1, we used AtFtsZ1_FL-eY-MTS (Yoshida et al., 2016) and constructed AtFtsZ1_C-eY-MTS. Unlike AtFtsZ1_C-eY-MTS, which was not successfully expressed, AtFtsZ1_FL-eY-MTS was successfully expressed and assembled into well-defined closed rings ( Supplementary Figure S4A ). The R₁₃₀ value for AtFtsZ1_FL-eY-MTS was 14% ( Supplementary Figure S4B ). Statistical analysis demonstrated that AtFtsZ1_FL-eY-MTS rings were more dynamic than AtFtsZ2_FL-eY-MTS rings ( Supplementary Figure S2B ), suggesting that membrane tethering does not alter the dynamic property difference between AtFtsZ2 and AtFtsZ1. Together, these results suggest that membrane tethering promotes the formation of AtFtsZ2_FL, AtFtsZ2_C and AtFtsZ1_FL into rings.

AtFtsZ2 is similar to bacterial FtsZs in that its primary function is to establish the structural framework of the chloroplast Z ring (Yoder et al., 2007; Erickson et al., 2010; Chen et al., 2018). In contrast, a critical role for AtFtsZ1 is to enhance Z-ring turnover dynamics by promoting AtFtsZ2 subunit exchange (TerBush and Osteryoung, 2012; Chen et al., 2018; TerBush et al., 2018). To ask whether the core region of AtFtsZ1 is sufficient for this activity, we coexpressed the full-length and core proteins in P. pastoris and compared the dynamics of the resulting filaments ( Figures 3a–h ). To mimic the membrane-tethering function of AtFtsZ2 in chloroplasts (Maple et al., 2005; Yoshida et al., 2016), in these experiments we only used AtFtsZ2_FL-eY-MTS and AtFtsZ2_C-eY-MTS.

In control experiments, AtFtsZ2_FL-eY-MTS and AtFtsZ1_FL-mC coassembled primarily into closed rings ( Figure 3a ) as observed previously (Yoshida et al., 2016). R₁₃₀ for AtFtsZ2_FL-eY-MTS was significantly higher when coassembled (13%) compared to when assembled separately (9.5%) ( Figures 3c, d , 2h ; Supplementary Figure S2B ), which is consistent with prior findings showing that AtFtsZ2 dynamics is accelerated by coassembling with AtFtsZ1. AtFtsZ2_C-eY-MTS and AtFtsZ1_C-mC also coassembled, forming either deeply curved arcs wrapped around the cell or closed rings ( Figures 3b, f ). R₁₃₀ for AtFtsZ2_C-eY-MTS was also significantly higher when coassembled (11%) with AtFtsZ1_C-mC compared to when assembled separately (8.7%) ( Figures 3g , 2i ; Supplementary Figure S2B ). These results demonstrate that the core region of AtFtsZ1 contains the primary information required for enhancing the turnover of AtFtsZ2 subunits from coassembled rings.

While our results above show that the core regions dominate the distinct turnover dynamics of AtFtsZ2 and AtFtsZ1, a previous study of truncated AtFtsZs in S. pombe suggested the N- and C-terminal regions influence the dynamics of filaments assembled by corresponding AtFtsZs (TerBush et al., 2016). To further explore the contributions of the N-terminal (NT) and C-terminal (CT) flanking regions to the relative difference of the AtFtsZs dynamics, we generated two chimeric proteins, AtFtsZ1_NT-AtFtsZ2_C-AtFtsZ1_CT (Z1NTZ2_CZ1CT) and AtFtsZ2_NT-AtFtsZ1_C-AtFtsZ2_CT (Z2NTZ1_CZ2CT), in which the flanking regions were swapped ( Figures 4a, b ). This swap strategy was designed to explore whether the flanking regions of AtFtsZ1 would make AtFtsZ2_C behaves like AtFtsZ1 ( Figure 4a ), and vice versa ( Figure 4b ). To evaluate their functionalities, we expressed them in E. coli, purified them, and measured their GTPase activities. The activities of Z1NTZ2_CZ1CT and Z2NTZ1_CZ2CT at 25°C were 0.39 ± 0.21 and 0.26 ± 0.10 GTP FtsZ^-1 min^-1, respectively, similar to the activities reported for AtFtsZ2 and AtFtsZ1, respectively (Porter et al., 2021), indicating both chimeras are functional GTPases.

The chimeric proteins were then fused to mCerulean and expressed in P. pastoris. Z1NTZ2_CZ1CT-mC formed long arced filaments while Z2NTZ1_CZ2CT-mC formed condensates ( Figures 4c, d ), indicating that the swap of AtFtsZ2’s flanking regions to AtFtsZ1_C appears to prevent the Z2NTZ1_CZ2CT chimeric protein from assembling into long filaments. In FRAP experiments, Z1NTZ2_CZ1CT-mC (R₁₃₀ = 10%) was less dynamic than Z2NTZ1_CZ2CT-mC (R₁₃₀ = 13%) ( Figures 4e–g ; Supplementary Figure S3C ). However, the difference in R₁₃₀ values was not statistically significant ( Supplementary Figure S2C ), unlike the difference between AtFtsZ2_FL-mC and AtFtsZ1_FL-mC ( Supplementary Figure S2A ). However, the dynamics of Z1NTZ2_CZ1CT-mC and Z2NTZ1_CZ2CT-mC were similar to those of AtFtsZ2_FL-mC and AtFtsZ1_FL-mC, respectively ( Supplementary Figure S2C ), implying the flanking regions of one protein did not alter the overall dynamics of the other. Both chimeric proteins ( Figures 4f, g ) were significantly less dynamic than their corresponding core proteins ( Figures 2e, f ; Supplementary Figure S2C ), suggesting that the flanking regions of AtFtsZs generally suppress the dynamics of both proteins. Collectively, these results are consistent with the conclusion that the core regions predominate in determining the differences in the relative dynamics between AtFtsZ2 and AtFtsZ1.

The dominant role of the core regions in controlling AtFtsZ filament turnover dynamics led us to ask if this also applies to other chloroplast FtsZ pairs. To address this, we tagged full-length Galdieria sulphuraria (Gs) FtsZA (GsFtsZA_FL) and GsFtsZB (GsFtsZB_FL) and their corresponding core regions (GsFtsZA_C and GsFtsZB_C) with mCerulean ( Figure 5a ). These constructs did not include the TP of corresponding GsFtsZs, and each construct was expressed individually in P. pastoris. GsFtsZA_FL-mC assembled into closed ring-like structures while GsFtsZB_FL-mC formed straight or arc-shaped filaments ( Figures 5b, c ). In FRAP experiments, the fluorescence recovery curve of GsFtsZA_FL-mC exhibited a slower recovery trend compared to that of GsFtsZB_FL-mC ( Figures 6a–c ; Supplementary Figure S5A ). The R₁₃₀ values for GsFtsZA_FL-mC and GsFtsZB_FL-mC were 32% and 46%, respectively. The difference between these values was statistically significant ( Supplementary Figure S6A ), indicating that GsFtsZA_FL-mC filaments are significantly less dynamic than GsFtsZB_FL-mC filaments. While the core proteins of GsFtsZA_C-mC and GsFtsZB_C-mC both assembled into shallow arcs ( Figures 5d, e ), R₁₃₀ values for GsFtsZA_C-mC filaments (18%) were significantly lower than for GsFtsZB_C-mC filaments (35%) ( Figures 6d–f ; Supplementary Figure S6A ), indicating that the GsFtsZ core proteins behaved similarly to the full-length proteins. Combined with the data from AtFtsZ, these findings suggest that the core regions of FtsZs dominate their unique dynamics in living cells. However, both GsFtsZ core proteins were less dynamic than their corresponding full-length proteins ( Supplementary Figures S5B, C ), although statistically the difference was only significant for GsFtsZA_FL-mC (R₁₃₀ = 32%) vs. GsFtsZA_C-mC (R₁₃₀ = 18%) ( Supplementary Figure S6A ). These differences are opposite those observed between the AtFtsZ core and full-length proteins, and suggest that the flanking regions of the GsFtsZ proteins, particularly for GsFtsZA, influence their absolute dynamics.

To investigate how membrane association affects the morphology and dynamics of the GsFtsZA full-length and core proteins, we generated GsFtsZA_FL-eY-MTS and GsFtsZA_C-eY-MTS. Both GsFtsZA_FL-eY-MTS and GsFtsZA_C-eY-MTS assembled into well-defined rings ( Figures 5f, g ), indicating that MTS fusions facilitate ring formation. R₁₃₀ values for GsFtsZA_FL-eY-MTS (12%) and GsFtsZA_C-eY-MTS (11%) were both reduced compared to those of the equivalent proteins lacking the MTS ( Figures 6b, e, g–i ), in particular for GsFtsZA_FL-mC (32%) though the fluorescent tags were not identical. Together with our AtFtsZ2 results, these data suggest that membrane tethering promotes the Z-ring assembly and likely constrains filament dynamics as well. However, the dynamics of GstFtsZA_FL-eY-MTS (R₁₃₀ = 12%) and GsFtsZA_C-eY-MTS (R₁₃₀ = 11%) were very similar ( Figures 6h, i ; Supplementary Figure S6A ), further supporting that the core regions determine the dynamics of GsFtsZA, as observed for AtFtsZ2 ( Figures 2h, i ; Supplementary Figure S2B ).

It has been proposed that FtsZB in red algae functions similarly to FtsZ1 in green lineage. In line with this, GsFtsZB has been reported to promote exchange of GsFtsZA subunits from coassembled filaments in S. pombe (TerBush et al., 2018). To test whether the core region of GsFtsZB is sufficient to enhance the dynamics of red algal chloroplast Z ring, we coexpressed the full-length of GsFtsZA with the full-length or core protein of GsFtsZB in P. pastoris and performed FRAP analysis on the reconstituted GsFtsZ rings ( Figures 7a–h ). Given that GsFtsZA_FL forms ring-like structures in the absence of MTS ( Figure 6a ), we only used GsFtsZA_FL-mC in the coexpression experiments. In contrast to singular expression, both GsFtsZB_FL-eY and GsFtsZB_C-eY coassembled primarily into ring-like structures when coexpressed with GsFtsZA_FL-mC ( Figures 7a, b ), further supporting it is GsFtsZA that dominates the morphology of red algal chloroplast Z ring, as AtFtsZ2 does in green lineage.

FRAP analyses showed that R₁₃₀ for GsFtsZA_FL-mC was significantly increased when coassembled (42%) with GsFtsZB_FL-eY compared to when assembled separately (32%) ( Figures 7c, d , 6b ; Supplementary Figure S6B ), demonstrating that GsFtsZB can enhance the turnover dynamic of GsFtsZA in the reconstituted red algal chloroplast Z ring, as observed for AtFtsZ1 in green lineage. Likewise, R₁₃₀ for GsFtsZA_FL-mC was significantly increased when coassembled (44%) with GsFtsZB_C-eY compared to when assembled separately (32%) ( Figures 7f, g , 6b ; Supplementary Figure S6B ), indicating an enhancement of GsFtsZA_FL-mC dynamics by GsFtsZB_C-eY in the coassembled rings. These findings demonstrate that the core region of GsFtsZB is sufficient to promote the turnover dynamic of GsFtsZA submits from their coassembled rings.