Cell division in the archaeon Haloferax volcanii relies on two FtsZ proteins with distinct functions in division ring assembly and constriction

Abstract

In bacteria, the tubulin homologue FtsZ assembles a cytokinetic ring, termed the Z ring, and plays a key role in the machinery that constricts to divide the cells. Many archaea encode two FtsZ proteins from distinct families, FtsZ1 and FtsZ2, with previously unclear functions. Here, we show that Haloferax volcanii cannot divide properly without either or both FtsZ proteins, but DNA replication continues and cells proliferate in alternative ways, such as blebbing and fragmentation, via remarkable envelope plasticity. FtsZ1 and FtsZ2 colocalize to form the dynamic division ring. However, FtsZ1 can assemble rings independent of FtsZ2, and stabilizes FtsZ2 in the ring, whereas FtsZ2 functions primarily in the constriction mechanism. FtsZ1 also influenced cell shape, suggesting it forms a hub-like platform at midcell for the assembly of shape-related systems too. Both FtsZ1 and FtsZ2 are widespread in archaea with a single S-layer envelope, but archaea with a pseudomurein wall and division septum only have FtsZ1. FtsZ1 is therefore likely to provide a fundamental recruitment role in diverse archaea, and FtsZ2 is required for constriction of a flexible S-layer envelope, where an internal constriction force might dominate the division mechanism, in contrast with the single-FtsZ bacteria and archaea that divide primarily by wall ingrowth.

Extended Data Fig. 1. Molecular phylogeny and comparison of archaeal FtsZ1 and FtsZ2 families.

(a) Maximum likelihood phylogenetic tree based on an alignment of the tubulin superfamily proteins identified in 60 diverse archaeal genomes (Supplementary Table 3), and the 13 bacterial and plant sequences used to identify them. Bootstrap support is shown for selected branches (%). (b) Domain organization and percent sequence identities for FtsZ1 and FtsZ2. The percentages over the domains (green and purple boxes) indicate the average sequence identity in that region for each H. volcanii FtsZ compared to all of the other members of the same family that were identified in the Archaea domain. The region between FtsZ1 and FtsZ2 represents the percent identity in the region between the H. volcanii FtsZ1 and FtsZ2 (%ID). The approximate location of conserved sequence motifs within the tail regions are indicated by vertical bars, coloured to indicate similarities between the two. (c) Aligned sequence regions containing conserved differences between the bacterial/plant FtsZ and the archaeal FtsZ1 and FtsZ2 families, labelled with the secondary structural elements. Boxed residues indicate conserved sites that are displayed in panel (e). (d) Crystal structure of FtsZ1 from Methanocaldococcus jannaschii (PDB: 1FSZ), with selected loops (T4-T7) involved in nucleotide binding and hydrolysis shown in pink. GDP is shown in orange, and the main domains are coloured as in panel (b). Boxed regions are expanded in panel (e), which displays some conserved residues that characteristically differ between the FtsZ1 and FtsZ2 families (grey space-filling models, with FtsZ2 consensus residues in parentheses) and cluster around the nucleotide-dependent polymerization surfaces.

Extended Data Fig. 2. Partial division phenotypes during depletion of FtsZ1 or FtsZ2.

(a) H. volcanii ID56 (p.tna-ftsZ1) was cultured in Hv-Cab + 2 mM Trp, and then loaded into a microfluidics platform and cultured with a flow of Hv-Cab (without Trp) over 15 h (0.5 p.s.i) to deplete FtsZ1. Shown is one cell that was identified to divide (unilaterally), even after ~9 h of depletion, and then one cell exhibited a budding-like process (arrows). Scale bars, 2 μm. (b) H. volcanii ID57 (p.tna-ftsZ2) was pre-cultured in Hv-Cab + 2 mM Trp, and then loaded into a microfluidics platform and cultured with a flow of Hv-Cab + 2 mM Trp for 3 h, followed by Hv-Cab (no Trp) for 10 h (2 p.s.i) to deplete FtsZ2. The zero timepoint represents the start of medium flow without Trp. During the early stage of depletion of FtsZ2, partial constrictions were sometimes observed, as seen in these two examples (i and ii), but these never completed division and the constriction eventually reversed over several hours (see arrows). Cells, however, retained some apparent ‘memory’ of the initial constriction often manifesting as a somewhat bilobed shape. The data shown is representative of at least two independent experiments. Scale bars, 2 μm.

Extended Data Fig. 3. Cellular DNA content during depletions of FtsZ1 and FtsZ2.

(a) SYTOX Green (SG) DNA staining of cells sampled from cultures 18 h after resuspension of mid-log cells in media without Trp. Stained cells were placed on an agarose pad and visualized by differential-interference contrast (DIC) and fluorescence microscopy (lower panels). Scale bars, 5 μm. (b) Flow cytometry analyses of cells sampled as per panel (a) (upper three panels), displaying side-scatter (as a proxy for cell size) versus SYTOX Green (SG)-DNA fluorescence. The lower three panels represent cultures treated in the same way, except 0.5 mM Trp was included in the medium. The data shown is representative of at least two independent experiments. The individual datapoints represent the area under the curve of each event detected; events were detected by a threshold of the side-scatter signal. After 18 h of ftsZ1 or ftsZ2 depletion, many very large cells with correspondingly high DNA content were observed, consistent with the images shown in panel (a). This indicates that DNA synthesis continues in proportion to the increase in cell volume during inhibition of cell division caused by depletion of FtsZ1 or FtsZ2.

Extended Data Fig. 4. Complementation of ΔftsZ1 ΔftsZ2.

(a) Phase-contrast images (left) and Coulter cytometry (right) of strains based on H. volcanii ID112 (ΔftsZ1 ΔftsZ2), plus pTA962-based plasmids expressing the indicated ftsZ genes, sampled during mid-log growth with the indicated concentrations of Trp. The same dataset for the wild-type control (H26 + pTA962) is shown in all graphs as a reference. Scale bars, 5 μm. (b) Corresponding western blot analyses of FtsZ1 and FtsZ2 protein levels in total cell extracts of the indicated strains. The data shown is representative of two independent experiments.

Extended Data Fig. 5. Comparison of ftsZ mutant cellular phenotypes in log and stationary phases.

The differing functions of FtsZ1 and FtsZ2 were also apparent when cultures of the knock-out, complementation and overexpression strains were compared in mid-log and stationary phases. Phase-contrast images (left) and Coulter cytometry distributions (right) of the wild-type and overexpression strains (a) and the indicated ftsZ knockout and complementation strains (b, c), all grown in Hv-Cab with the indicated concentration of Trp and sampled at mid-log and stationary phases. The data shown are representative of at least two independent experiments. Scale bars, 5 μm. Compared to mid-log cells, all the strains except the strains without a copy of ftsZ2, tended towards the wild-type size (smaller) and regular plate morphology in stationary phase. The ΔftsZ2 strains were somewhat smaller in stationary phase, but maintained greatly enlarged giant plate and elongated cells, suggesting a poor recovery as cell growth slows in stationary phase. These findings suggest that FtsZ2 confers a partial ability to divide and recover more normal cell sizes as the cell growth rate slows in stationary phase, whereas cells without FtsZ2 have a much stronger block to division that is maintained even as cells slow or stop growth in stationary phase.

Extended Data Fig. 6. FtsZ1 and FtsZ2 fluorescent fusions are not fully functional as sole copies but at moderate concentrations have minimal impact on cell division in the wild type.

(a-b) FtsZ1 and FtsZ2 fluorescent fusion proteins were functionally tested in their respective ΔftsZ1 or ΔftsZ2 backgrounds by phase-contrast and fluorescence microscopy (left) and by Coulter cytometry (right) for cell size. FtsZ1-GFP and FtsZ1-mCherry (0.2 mM Trp) partially complement ΔftsZ1. (b). FtsZ2-GFP was unable to complement the ΔftsZ2 background, whereas the untagged protein achieves full complementation (0.2 mM Trp). The same dataset for the wild-type control is shown in both graphs as a reference. (c-e) When expressed in wild-type cells, FtsZ1-mCherry or FtsZ2-GFP, or both, cause minimal effects on cell size and shape at a moderate level of expression (0.2 mM Trp), and show sharp midcell bands. These proteins are therefore useful localisation markers for division, although detailed analyses of FtsZ subcellular ultrastructure and dynamics await the development of functional complete labelling. The data shown are representative of at least two independent experiments. Scale bars, 5 μm.

Extended Data Fig. 7. Cell shape analyses for FtsZ localisation interdependency studies.

Cell area and shape (circularity) were determined for individual cells, as per Fig. 5 (0.2 mM Trp), and data were combined from two replicate experiments in for each plot. The plots are labelled with the strain’s relevant genomic background (left) and the ftsZ variant(s) expressed on the plasmid (right). The data shown are representative of at least two independent experiments.

Extended Data Fig. 8. FtsZ1-mCh localisation in ftsZ2-mutant strains.

(a) Demonstration of the automated image analysis procedure for determining FtsZ localisation parameters. Cell outlines were obtained (red), and the fluorescence (FtsZ1-mCherry in yellow) was quantified by averaging the intensity on the transverse axis to create a longitudinal intensity profile. Gaussian peaks were fitted to the significant localisations and a spline fit to the background. The localisation thickness (W) was taken as the width of the fitted Gaussian peaks at half height (μm), and the intensity (I) was taken as the integrated peak area (per μm across the cell). See Methods for further details. (b) Histograms for cells with the indicated number of localisations versus cell length for ΔftsZ2 + FtsZ1-mCh. Colored lines indicate the lengths of cells that have the indicated relative number of localisations per unit length. (c) Violin plots of the thickness of FtsZ1-mCh localisation in the indicated strain backgrounds; the median is indicated by a white dot, the thick bar is the interquartile range, and thin bar is the 9^th-91^st percentile range. The data shown are representative of at least two independent experiments. Explanation of the experiment using the ΔftsZ2 + FtsZ2.D231A-GFP + FtsZ1-mCh strain is given in the Supplementary results and discussion and Extended Data Fig. 10.

Extended Data Fig. 9. Localisation of FtsZ T7-loop mutants, ftsZ1.D250A-mCh and ftsZ2.D231A-GFP.

(a-b) The FP-tagged T7 mutants fail to complement their respective ΔftsZ strain. (c-d) Suppression of the severe dominant-inhibitory effects of the T7 mutants by the fluorescent tags, shown by phase-contrast and fluorescence microscopy (left) and Coulter cytometry (right). FtsZ1.D250A-mCh showed aberrant localisation and cellular distortions and envelope protrusions associated with the fluorescent filaments; small fluorescent particles detatched from cells are also evident (arrowheads). Yet the cell size distribution was only subtly affected. The data shown are representative of at least two independent experiments. Similar results were obtained with FtsZ1.D250A-GFP (H. volcanii ID153). FtsZ2.D231A-GFP shows similar localisation to wild-type (FtsZ2-GFP), with only a moderate increase in cell size increase observed at 1 mM Trp; note that FtsZ2-GFP had a much stronger influence (Extended Data Fig. 6d). Scale bars, 5 μm.

Extended Data Fig. 10. Co-localisation studies of wild-type FtsZ and T7-loop mutants.

Fluorescence microscopy (overlay of GFP and mCherry channels; co-localisation appears white) (left) and cell size/shape analysis plots (right) for the indicated strains (grown with 0.2 mM Trp) containing one tagged wild-type protein and the alternate tagged the T7-loop mutant. The data shown are representative of at least two independent experiments. In panels (b) and (d), results with cells grown with 1 mM Trp is shown in the lower insets; the morphology percentages for these were: (b) 95.4% wild-type-like, 3.2% giant plates, 0.9% filaments, and 0.5% debris (n = 439), and (d) 40.7% wild-type-like, 25.5% giant plates, 29.1% filaments, and 5.1% debris (n = 196). Scale bars, 5 μm.

Fig. 1. Depletion of FtsZ1 or FtsZ2 results in cell division defects.

(a) Phase-contrast microscopy of samples from cultures (H. volcanii ID56 - p.tna-ftsZ1, and ID57 - p.tna-ftsZ2) immediately prior to (0 h) or at the indicated time-points after resuspension of cells (to OD₆₀₀ = 0.25) in growth medium without Trp. Scale bars, 5 μm. (b) Western blots probed with FtsZ1 (left) or FtsZ2 (right) antisera during depletion with the corresponding Ponceau S pre-staining of total protein. The separate H26 wild-type (WT) lanes were from the same blots for each protein; bands for the ~46 kDa pre-stained marker (M) and FtsZ1 (expected ~40 kDa) and FtsZ2 (expected ~43 kDa) are indicated. The double band, clearer for FtsZ2, suggests modified or clipped forms. (c) In situ time-lapse imaging of FtsZ depletion. Cells cultured in Hv-Cab + 2 mM Trp were washed and placed onto agarose media pads without Trp and imaged over time. Morphology of giant cells differs from the liquid culture in (a), most likely due to the support provided by growth on agarose. (d) In situ time-lapse imaging of FtsZ restoration. The respective p.tna-ftsZ strains were initially grown for ~2 days without Trp inducer, and then cells were washed and placed onto agarose media (Hv-Cab) pads including 0.2 mM Trp for FtsZ1, and 2 mM Trp for FtsZ2. Scale bars (c-d), 2 μm.

Fig. 2. FtsZ1 and FtsZ2 are dispensable for survival but required for normal cell division.

(a) Box plots of agar colony counts (cfu/mL) of H. volcanii wild-type (H26), ΔftsZ1 (ID76), ΔftsZ2 (ID77), and ΔftsZ1 ΔftsZ2 (ID112), sampled during mid-log growth (OD₆₀₀ = 0.2) in Hv-Cab medium (+ 50 μg/mL uracil) (n = 4 separate cultures, 3 for ΔftsZ1 ΔftsZ2), 1 experiment). Box - interquartile range, whiskers - upper and lower limits, centre line - median. (b) Phase-contrast images (Scale bars, 5 μm) and (c) Coulter cell-volume frequency distributions (normalized to total count). (d) Confocal microscopy 3D-reconstructed images of cells stained with Mitotracker Orange and embedded in low melting point agarose. The reconstructions are shown with a ~45° rotation around the x-axis. The colour scale indicates the z-depth. White scale bars, 2 μm. Similar morphologies were seen with FM1-43 membrane staining (Supplementary Fig. 4b). (e) Live cell time-lapse images of ΔftsZ1 ΔftsZ2 growing on an agarose media pad, and (f) showing polar tubulation and budding-like processes (Scale bars, 5 μm). (g-i) Phase-contrast microscopy of the indicated strains expressing the ftsZ genes or containing vector only in steady mid-log cultures with Hv-Cab medium with the indicated concentrations of Trp. Scale bars, 5 μm.

Fig. 3. Overproduction of FtsZ1 or FtsZ2 differentially affects cell division and shape.

(a) Phase-contrast micrographs of H. volcanii wild-type (H98 + pTA962), ID25 (FtsZ1 overproduction, H98 + pTA962-ftsZ1) and ID26 (FtsZ2 overproduction, H98 + pTA962-ftsZ2) from steady mid-log cultures in Hv-Cab + 2 mM Trp. Scale bars, 5 μm. (b) Coulter cell-volume frequency distributions (normalized to total count) of the strains in panel (a). (c) Frequency distributions of the circularity of cell outlines, generated from the analysis of microscopy data from a representative experiment with the following numbers of cells: wild-type n = 280, + FtsZ1 n = 267, + FtsZ2 n = 541. (d) Western blots and Ponceau total protein staining of whole cell lysates of the wild-type, FtsZ1 overproduction and FtsZ2 overproduction strains, probed with FtsZ1 or FtsZ2 antisera, as indicated. The ~46 kDa size marker (M) is indicated.

Fig. 4. Midcell localisation of FtsZ1-mCh and FtsZ2-GFP.

Mid-log cultures of (a) H. volcanii ID49 (WT + FtsZ1-mCh), (b) ID17 (WT + FtsZ2-GFP) and (d) ID67 (WT + FtsZ1-mCh + FtsZ2-GFP dual localisation), cultured with 0.2 mM Trp, were imaged by phase-contrast and fluorescence microscopy. The mCh (magenta) and GFP (green) fluorescence was quantified along the long axis of cells, by plotting the relative (normalized) median intensity perpendicular to the long axis. (c) 3D-SIM micrographs of mid-log H. volcanii ID16 (WT + FtsZ1-GFP) (Hv-Cab + 0.2 mM Trp). The upper two panels show a ‘side-on’ view (~70° tilt of the xy view), and the lower four panels show a ~20° xy tilt. (d) Co-localisation of FtsZ1-mCh and FtsZ2-GFP in H. volcanii ID67, including selected examples of dividing cells (lower left). The ring containing FtsZ1 and FtsZ2 closes down with division constrictions that range between unilateral to partially and equally bilateral. Correlation coefficients were plotted as a frequency distribution (80% of the cells had a correlation coefficient greater or equal to 0.7). The total number of cells analysed (n) were obtained with data from two independent experiments. Intensity data are presented as median values with error bars showing upper and lower limits. Scale bars (panels a, b and d), 5 μm (larger field of view) and 2 μm (enlarged insets and smaller fields). Scale bars (panel c), 300 nm.

Fig. 5. Localisation interdependency and the effect of FtsZ mutations on FtsZ1-mCh and FtsZ2-GFP localisation.

Fluorescence and phase-contrast microscopy of mid-log H. volcanii strains (wild-type or ΔftsZ backgrounds) carrying plasmid copies of the indicated ftsZ mutants and fusion proteins. Strains were grown with 0.2 mM Trp unless otherwise specified. Insets show either a magnified view at the arrowhead in the main panels, or represent another culture grown with the indicated concentration of Trp. Scale bars (main fields), 5 μm. (a) Localisation of each FtsZ in the absence of the other FtsZ (insets scale bar, 2 μm). (b) Localisation of each FtsZ in the presence of the T7 mutant as the sole copy of the other. (c) Localisation of each FtsZ in the wild-type genomic background expressing the T7 mutant of the other. (d) Ratio of individual FtsZ1-mCh localisation intensity over the mean cellular background fluorescence versus localisation thickness (μm) in rods or filaments in the indicated backgrounds. Data-points are coloured with a proximity heatmap. Representative images of the wild-type background can be seen in Fig 4a.

Fig. 6. Model of FtsZ1 and FtsZ2 function at the archaeal division site.

Schematic models of FtsZ1 (magenta) and FtsZ2 (green) localisation and functions (arrows) are based on our results, summarized here. Most archaea, including H. volcanii, have an envelope composed of a lipid membrane and an S-layer glycoprotein semi-crystalline array on the surface. Our results raise questions about the ultrastructure and dynamics of the two FtsZ proteins in the division ring, the identity of an unknown number of other putative divisome components (x, y, and a SepF homolog), and how the ring could functionally associate with the envelope growth (Sec/PssA/PssD/ArtA) and cell shape regulation (CetZ1) machineries, which appear to be spatially and functionally linked to midcell^,.