Cell division protein CdpA organises and anchors the midcell ring in haloarchaea

Abstract

Many archaea appear to divide through the coordinated activities of two FtsZ homologues (FtsZ1 and FtsZ2) and another bacterial cell division homologue (SepF), which are part of the midcell division ring. Here, we identify an additional protein (HVO_0739, renamed CdpA) that is involved in cell division in Haloferax volcanii, with homologues in other Haloarchaea. CdpA localises at the midcell division ring, and this requires the presence of the ring-assembly protein FtsZ1. The division constriction protein FtsZ2 also influences the proper midcell assembly and structure of CdpA. In the absence of CdpA, cells frequently fail to divide properly, and FtsZ1 formed poorly condensed pseudo-helical structures spanning across a broad region of the cell, whereas FtsZ2 showed mispositioned foci, nano-rings, and filaments. The rate of directional movement of FtsZ1 and FtsZ2 structures around the division ring appears minimally affected by loss of CdpA, which resulted in continual repositioning of the aberrant FtsZ structures in the cells. In contrast to the FtsZ proteins, CdpA formed relatively immobile foci around the ring. Protein domain function studies, pull-down assays, and multimer structure predictions suggest that CdpA is part of a membrane complex that tethers FtsZ2 and other division proteins to the midcell membrane. Our discovery of an archaeal FtsZ organisation and midcell anchor protein offers new insights into cell division mechanisms that are similar across the tree of life.

Fig. 1. CdpA is a cell division protein in haloarchaea.

a Phylogenetic distribution of CdpA (orange circles), CdpA-like (pink), CetZ (blue), FtsZ1 (light green), and FtsZ2 (dark green) on a schematic reference phylogeny of the Archaea, expanding on Halobacteriota families. The presence/absence is shown at the indicated taxonomic level and is marked as present if identified in at least one genome within that taxon. b Alphafold2 prediction of CdpA monomer structure coloured by model confidence (blue - pLDDT > 90%, orange - pLDDT <50%) and predicted overall membrane topology. c Phase-contrast microscopy of wild-type (WT) and ΔcdpA strains in steady mid-log cultures and the corresponding cell shape analysis. n= number of cells examined over one biologically independent experiment. The data shown is representative of at least two independent experiments. Source data are provided as a Source Data file. d, Live-cell time-lapse images of ΔcdpA growing on an agarose media pad and showing cell lysis, fragmentation, and division processes (arrows). Phase-contrast and fluorescence micrographs of wild-type cells expressing CdpA-GFP (e) and ΔcdpA expressing CdpA-GFP (f) under control of the p.tnaA promoter with 0.2 mM Trp induction. Demographic analysis (righthand panels) show fluorescence intensity along the long axis ordered by cell lengths, top to bottom. n = number of cells examined over one biologically independent experiment. The data shown in d–f is representative of at least two independent experiments. All scale bars, 5 µm.

Fig. 2. Sub-cellular and sub-ring localisation of CdpA with FtsZ1 or FtsZ2.

a Phase-contrast and fluorescence images of H. volcanii strains YL6 (H26 + CdpA-GFP + FtsZ1-mCherry dual expression) and YL7 (H26 + CdpA-GFP + FtsZ2-mCherry dual expression) cultured with 0.2 mM Trp during mid-log growth. The middle graphs show the intensity profile of the fluorescence signal (magenta for mCherry, and green for GFP) by plotting the normalized median intensity perpendicular to the long axis of cells. The right graphs show correlation coefficients plotted as a frequency distribution. n = number of cells examined in one independent experiment. The data shown are representative of at least two independent experiments. Scale bars, 5 µm. Source data are provided as a Source Data file. b Dual-colour of 2D-SIM images of mid-log H. volcanii YL6 and YL7 grown with 0.2 mM Trp. mCherry (magenta) and GFP (green) fluorescence was quantified by plotting normalized intensity in the respective channel along the ring band. Scale bars, 1 µm. Source data are provided as a Source Data file. c 3D-SIM images (xy-tilted) of mid-log H. volcanii YL29 (H26 + CdpA-GFP), YL24 (H26 + FtsZ1-GFP), and YL26 (H26 + FtsZ2-GFP) grown with 0.2 mM Trp. Scale bars, 0.5 µm. The data shown in panel b and c are representative of at least two independent experiments.

Fig. 3. CdpA, FtsZ1 and FtsZ2 show different localization interdependencies and strongly influence each other’s midcell structures.

All strains were grown in Hv-Cab medium with 0.2 mM Trp, and mid-log cultures were sampled for microscopy. a Phase-contrast and fluorescence composite images of CdpA-GFP in the absence of FtsZ1, FtsZ2, or both. Scale bars, 5 µm. The micrographs shown are representative of at least two independent experiments. b Left: localization of FtsZ1-mCherry in wild-type H26 and ΔcdpA backgrounds. The arrowhead indicates a poorly condensed FtsZ1-mCherry structure in ΔcdpA. Scale bars, 5 µm. The dot-plots show quantification of the degree of FtsZ1-mCherry localization versus individual localization thickness as a measurement of protein condensation along the long axis in the indicated backgrounds. The data points are coloured with a proximity heatmap. c localization of FtsZ2-GFP in H26 (scale bar, 5 µm) and ΔcdpA backgrounds (scale bars, 2 µm). The histograms show the number of FtsZ2-GFP rings detected versus cell length in the indicated backgrounds. Blue diagonals (“All”) represent 1 localization per 4 µm in length (i.e. approximately equivalent to all division sites detected at the normal frequency per cell length in wild-type cells), green (“−1”) indicates one fewer rings/cell length, yellow is two fewer, and red is 3 fewer. n_x=number of cells for the indicated number(x) of localizations examined over one biologically independent experiment. d Upper: localization of FtsZ1-mCherry and FtsZ2-GFP in H26 and ΔcdpA strains. Scale bars, 2 µm, Bottom: correlation coefficient between FtsZ1-mCherry and FtsZ2-GFP plotted as a frequency distribution. The majority (97%) of H26 cells had a correlation coefficient of ≥0.8, while only 32% of ΔcdpA cells had a correlation coefficient of ≥0.8. For b–d, n = number of cells examined in one independent experiment, and the data shown is representative of at least two independent experiments. Source data are provided as a Source Data file.

Fig. 4. Division protein localization dynamics and the effect of cdpA deletion on FtsZ1 and FtsZ2 movement.

Cells were grown in Hv-Cab without tryptophan induction to minimize the expression level of fluorescent fusion proteins. Mid-log cultures were sampled for TRIF imaging. The data shown is representative of at least two independent experiments. a Maximum intensity projections (MIPs) of two representative wild-type H26 cells expressing FtsZ1-GFP (i and ii) are shown on the left, with the corresponding kymographs at positions indicated by red arrows in the MIPs shown on the right. b MIPs of two representative wild-type H26 cells expressing FtsZ2-GFP (i and ii) and their corresponding kymographs. c, d MIPs of wild-type H26 cells expressing CdpA-GFP and their corresponding kymographs with different imaging time-window (4 s intervals for 15 min, c; 10 s intervals for 40 min, d). e MIP of one representative ΔcdpA expressing FtsZ1-GFP and its kymograph computed from the intensity along the long axis of cell (between the two red arrows). The relative position (0-1) along the long axis of cell is labelled below the kymograph. f Montages of FtsZ1-GFP dynamics in ΔcdpA and the integrated fluorescence time trace for the ROI region, as shown as a small red box around the centre of ring at the start of imaging (Time 0 min). The red curve is the moving average (every 5 points) of the raw intensity (grey dots). Source data are provided as a Source Data file. g The mean power spectral density (PSD) curves (± SEM) of FtsZ1-GFP in wild-type and ΔcdpA cells (n = 21). Source data are provided as a Source Data file. Montages of representative FtsZ2-GFP dynamics in two ΔcdpA cells and their corresponding kymographs indicate the changes in ring position (h) and orientation (i). All scale bars are 1 µm.

Fig. 5. The three domains of CdpA have different contributions to the role of CdpA in of the organisation of FtsZ1 and FtsZ2.

a Alphafold2 models of CdpA showing domain boundaries of the GFP tagged deletion constructs, including the N-terminal domain (NTD), disordered linker, and C-terminal domain (CTD). b Fluorescence microscopy of H26 wild-type strains expressing the indicated constructs during mid-log growth in Hv-Cab medium with 0.2 mM Trp. Arrows indicate regions of weak midcell localisation. Scale bars, 5 µm. The lower panels show demographic analyses of fluorescence along the long axis, ordered from shortest to longest cells (left to right). n=number of cells examined over one representative experiment. c Phase-contrast and fluorescence micrographs of the indicated cdpA deletion strains producing either FtsZ1-mCherry or FtsZ2-GFP, sampled during mid-log phase growth in Hv-Cab + 0.2 mM Trp. Arrows indicate some faint foci of FtsZ2-GFP. Scale bars, 5 µm. d Fluorescence images of the indicated cdpA deletion strains co-producing FtsZ1-mCherry and FtsZ2-GFP. Scale bars, 2 µm. e Distributions of cellular correlation coefficients between FtsZ1-mCherry and FtsZ2-GFP and (f) ratio of localisation area for FtsZ1-mCherry:FtsZ2-GFP in the background strains shown below the lower panel. The sample sizes analysed for each background strain in panel e were as follows: 587 for H26, 665 for ΔcdpA, 397 for cdpA_ΔNTD, 424 for cdpA_ΔLinker, and 386 for cdpA_ΔCTD. The statistical test in panel e was performed using an unpaired t-test (two-tailed), with **** indicating a significant difference (P < 0.0001). The sample sizes analyzed for each background strain in panel f were as follows: 471 for H26, 320 for ΔcdpA, 349 for cdpA_ΔNTD, 329 for cdpA_ΔLinker, and 447 for cdpA_ΔCTD. (112 data points fell outside the axis limits). The data shown are representative from one of two independent experiments. Source data are provided as a Source Data file.

Fig. 6. The interdependency of CdpA and SepF localisation.

a Phase-contrast and fluorescence micrographs of wild-type H26 and cdpA deletion strains producing SepF-GFP, sampled during mid-log phase growth in Hv-Cab + 2 mM Trp. Microscopic images of the strains grown under 0.2 mM Trp can be found in Supplementary Fig. 21. At higher expression levels, SepF-GFP formed branched and less condensed midcell bands in the wild-type-like cells in the absence of CdpA. b Phase-contrast and fluorescence micrographs of SepF deletion strain (HTQ239) expressing CdpA-GFP under its native promoter. The histograms show the number (#) of CdpA-GFP rings detected versus cell length. Coloured lines indicate a nominal WT frequency of rings (All), or how many fewer rings were observed, as per Fig. 3c. n = number of cells examined over one biologically independent experiment with indicated ring number, and the data shown is representative of at least two independent experiments. Source data are provided as a Source Data file. c Phase-contrast and fluorescence micrographs of three CdpA domain deletion strains producing SepF-GFP, sampled during mid-log phase growth in Hv-Cab + 2 mM Trp. White arrows indicate small extracellular fluorescent particles. All scale bars, 5 µm. The micrographs shown are representative of at least two independent experiments.

Fig. 7. CdpA is associated with FtsZ2 and other division proteins in a putative membrane anchor complex.

a Western blots of one of three replicate co-immunoprecipitation experiments. The cell lysate (L), unbound protein flow through (FT) and bound protein elution fraction (E) from control (H26 + HA vector only) and ΔcdpA + CdpA-HA were analysed by 15% SDS-PAGE and western blotting. The total protein stain (Ponceau S) is shown for the same membrane used for the anti-HA probe, and replicate membranes were probed with anti-FtsZ1 or anti-FtsZ2 as indicated. The control (H26 + HA vector only) was not detectable within the size limits of the gel. b Alphafold2 model of the midcell anchor complex generated from a sequence input of 2 CdpA, 2 CdpB1, 2 CdpB2, 2 FtsZ2, and 2 SepF. The C-terminal regions of CdpA and the N- and C-terminal extensions of FtsZ2 that are largely disordered are not shown for clarity. c The Alphafold2 predicted aligned error (%) heatmap plot of all input sequences (concatenated), indicating the degree of confidence in pairwise interactions between residues. d The same structure and orientation as in panel b shown with surface electrostatics, indicating the approximate CdpA transmembrane domain (TMD) and SepF membrane targeting sequence (MTS). Unprocessed western blots and the Alphafold2 coordinates are provided in the Source Data file.