Bactofilin BacA is required for proper cell morphology in H. neptunium.

(A) Schematic representation of the two bactofilin genes present in the H. neptunium genome. bacA lies adjacent to the M23 peptidase gene lmdC. Arrows indicate the direction of transcription. (B) Domain organization of BacA and BacD from H. neptunium. The bactofilin polymerization domain (colored boxes) is flanked by non-structured N- and C-terminal regions. (C) Morphology of H. neptunium bactofilin mutants. Shown are representative cells of strains LE670 (wild type), EC28 (ΔbacA), EC23 (ΔbacD) and EC33 (ΔbacAD), imaged by differential interference contrast (DIC) microscopy. Bar: 4 µm. (D) Immunoblot analysis of the strains shown in panel C, performed using anti-BacA antibodies. (E) Transmission electron micrographs of ΔbacA cells at the early stalked-cell stage. Bar: 1 µm. (F) Quantification of the proportion of phenotypically abnormal stalked and budding cells in cultures of the strains analyzed in panel C (n = 100 cells per strain).

Lack of BacA leads to uncontrolled growth of the stalk and bud compartments.

(A) Immunoblot showing the levels of BacA in strain EC41 (ΔbacA PCu-bacA) over the course of BacA depletion. Cells were grown in copper-containing medium, washed and transferred to inducer-free medium. At the indicated time points, samples were withdrawn and subjected to immunoblot analysis. Strains LE670 (wild type) and EC28 (ΔbacA) were included as controls. The position of BacA is indicated. (B) Morphological defects induced by BacA depletion. Cells of strains LE670 (wild type) and EC41 (ΔbacA PCu-bacA) were grown in medium containing 0.5 mM CuSO4, washed, and incubated for 6 h in inducer-free medium before they were transferred onto ASS-agarose pads lacking inducer and imaged at the indicated time points. Bar: 2 µm. (C) Changes in the growth pattern of H. neptunium in the absence of bactofilins. Cells of strains LE670 (wild type) and EC33 (ΔbacAD) were stained with the fluorescent D-amino acid HADA prior to analysis by fluorescence microscopy. Shown are representative images of cells at different developmental stages. Bars: 2 µm.

BacA and BacD co-polymerize into filamentous structures.

(A) Model of a BacA trimer generated with AlphaFold-Multimer (Evans et al., 2022). Only residues P34-D138 are shown for each subunit. (B) Visualization of BacA polymers. Purified BacA-His6 was stained with uranyl acetate and imaged by transmission electron microscopy (TEM). Arrowheads point to BacA filaments. Asterisks indicate filament bundles and sheets. Bars: 200 nm. (C) Copolymerization of BacA and BacD after heterologous coexpression in E. coli. Cells of E. coli BL21(DE3) transformed with plasmid pEC121 (PT7-bacA-eyfp PT7-bacDecfp) were grown in LB medium containing 5 % glucose and induced with 0.5 mM IPTG prior to imaging. Bar: 3 µm.

BacA and BacD show a dynamic, cell cycle-dependent localization pattern.

(A) Localization pattern of BacA-YFP. Cells of strain EC61 (bacA::bacA-eyfp) were transferred to agarose pads and imaged at 20-min intervals. Shown are overlays of DIC and fluorescence images of a representative cell, with YFP fluorescence shown in red. Scale bar: 3 µm. (B) Demographic analysis of BacA-YFP localization in swimmer (left), stalked (middle), and budding (right) cells of strain EC61 (bacA::bacA-eyfp). The fluorescence intensity profiles measured for cells of each type were sorted according to cell length and stacked on each other, with the shortest cell shown at the top and the longest cell shown at the bottom (n=250 swimmer cells, 215 stalked cells and 49 budding cells). Dotted red lines indicate the positions to which cells were aligned. (C) Demographic analysis of BacD-Venus localization. Cells of strain EC67 (bacD::bacD-venus) were analyzed as described in panel B (n=416 swimmer cells, 248 stalked cells and 45 budding cells). (D) Co-localization of BacA and BacD in cells of strain EC68 (bacA::bacA-eyfp bacD::bacD-mCherry). Shown are DIC images and overlays of the YFP and mCherry signals of representative cells, arranged according to their developmental state. The individual signals are shown in Figure S6A. The Pearson’s Correlation Coefficient (PCC) of the two fluorescence signals in a random population of cells (n=454) is 0.506. Bar: 1 µm.

The single-particle dynamics of BacA change over the course of the cell cycle.

(A) Representative heat maps showing the sum of single-particle positions observed in live-cell single-particle localization microscopy studies of a swimmer, stalked and budding cell producing BacA-YFP (EC61) or a mutant cell producing the polymerization-deficient BacAF130R-YFP variant (MO78). Insets show the corresponding bright-field images. Bars: 1 µm. (B) Mean-squared displacement (MSD) analysis of the mobility of BacA-YFP in wild-type swimmer (n=100), stalked (n=105) and budding (n=101) cells (EC61). Cells (n=110) producing the BacAF130R-YFP were analyzed as a control (MO78). Error bars indicate the standard deviation. (C) Bubble plots showing the proportions of mobile and immobile particles (size of the bubbles) and the corresponding average diffusion constants (y-axis) in the cells analyzed in panel B.

Putative operons comprising adjacent BacA and M23 peptidase genes are widely conserved among bacteria.

(A) Arrangement of bacA and lmdC genes in species whose BacA homolog has a proven role in cell morphogenesis. (B) Co-conservation of BacA and M23 peptidase genes across different bacterial phyla. Bioinformatic analysis was used to identified genomes that contained a bactofilin gene located next to an M23 peptidase gene. After retrieval of the taxonomy IDs of the corresponding species from the National Center for Biotechnology Information (NCBI) website, a phylogenetic tree of the species was created using the iTOL server (Letunic and Bork, 2021). Abbreviations: DT (Deinococcus-Thermus group), FCB (Fibrobacteres-Chlorobi-Bacteroidetes group), β (β-proteobacteria), δ (δ-proteobacteria). The full list of species used and details on the genes identified are provided in Dataset 2.

LmdC is a peptidoglycan hydrolase with DD-endopeptidase activity.

(A) Predicted domain architecture of H. neptunium LmdC. The predicted positions of the transmembrane helix (TM), the three coiled-coil regions (CC) and the M23 peptidase domain are indicated. (B) Alignment of the amino acid sequences of multiple M23 peptidases showing the conservation of the catalytic residues in LmdC. Residues required to coordinate the catalytic Zn2+ ion of LytM from S. aureus (Firczuk et al., 2005) are indicated by arrow-heads. The proteins shown are EnvC from E. coli (UniProt: P37690), NlpD from E. coli (P0ADA3), LytM from S. aureus (O33599), MepM from E. coli (P0AFS9) and LmdC from H. neptunium (Q0BYX6). (C) Predicted molecular structure of H. neptunium LmdC, generated with Alphafold2 (Jumper et al., 2021). The different domains of the protein and the position of proline 66, which terminates the N-terminal fragment of LmdC used for the in vivo analyses in this study (LmdC1-66) are indicated. (D) HPLC traces showing the muropeptide profile of peptidoglycan treated with LmdC. Cell walls were incubated with the isolated M23 peptidase domain of LmdC at pH 5.0 and pH 7.5. Subsequently, muropeptides were released with cellosyl, reduced and separated by HPLC. A control sample lacked LmdC (No protein). The nature of the major products is indicated above the peaks. Tri, Tetra and Penta stand for N-acetylglucosamine-N-acetylmuramitol tripeptide, tetrapeptide and pentapeptide, respectively. (E) Structure of a TetraTetra muropeptide. Abbreviations: G (N-acetylglucosamine), M (N-acetylmuramic acid). The cleavage site of LmdC is indicated by a red arrowhead.

LmdC colocalizes and directly interacts with BacA.

(A) Domain architectures of the native LmdC protein and the LmdC1-66-mCherry fusion used for co-localization studies. The transmembrane helix is shown in purple. Abbreviations: CC (coiled-coil domain), M23 (M23 peptidase domain), CHY (mCherry). (B) Co-localization of BacA-YFP and LmdCN-mCherry in H. neptunium. Shown are representative cells of strain MO79 (bacA::bacA-eyfp PCu::PCu-lmdCN-mCherry that were induced for 2 h with 0.5 mM CuSO4 prior to analysis by DIC and fluorescence microscopy. (C) Demographic analysis of BacA-YFP and LmdCN-mCherry co-localization in swimmer (left; n=99), stalked (middle; n=103) and budding (right; n=22) cells from the culture imaged in panel B. Data were analyzed as described for Figure 4B. Corresponding fluorescence profiles are shown at the same relative position in the two graphs. (D) Bio-layer interferometric analysis of the interaction between cytoplasmic domain of LmdC and BacA. Sensors derivatized with a biotinylated synthetic peptide comprising the N-terminal cytoplasmic region of LmdC (amino acids 1-38) were probed with the indicated concentrations of BacA. After the association of BacA, the sensors were transferred to protein-free buffer to induce BacA dissociation. The interaction kinetics were followed by monitoring the wavelength shifts resulting from changes in the optical thickness of the sensor surface during association or dissociation of the analyte. The extent of non-specific binding of BacA to the sensor surface was negligible (control). (E) Affinity of the BacA-LmdC interaction. The maximal wavelength shifts measured in the experiment shown in panel D were plotted against the corresponding BacA concentration. The data were fitted to a one-site binding model, yielding an apparent Kd value of ~15 µM.

BacA and LmdC contribute to cell morphogenesis in R. rubrum.

(A) Schematic representation of the bacA-lmdC operon in R. rubrum. (B) Domain organization of BacA from R. rubrum. The bactofilin polymerization domain (green box) is flanked by non-structured N- and C-terminal regions. (C) Phenotypes of R. rubrum wild-type (S1), ΔbacARs (SP70), ΔlmdCRs (SP68) and ΔbacARs ΔlmdCRs (SP116) cells, imaged using phase-contrast microscopy. Bar: 5 µm. (D) Superplots showing the distribution of cell sinuosities in populations of the indicated R. rubrum strains. Small dots represent the data points, large dots represent the median values of the three independent experiments shown in the graphs (dark blue, light blue, grey). The mean of the three median values is indicated by a horizontal red line. Whiskers represent the standard deviation. *** p<0.005; ns, not significant; t-test). (E) Co-localization of BacARs-mCherry and LmdCNRs-mNeonGreen in the ΔlmdCRs background (SP119). The Pearson’s Correlation Coefficient (PCC) of the two fluorescence signals in a random subpopulation of cells (n=87) is 0.515. Scale bar: 5 µm. (F) Localization of BacARs-mNeonGreen in the ΔlmdCRs background (SP98). Bar: 5 µm. (G) Localization of LmdCRs-mNeon-Green in the ΔbacARs background (SP118). Bar: 5 µm.

Model of the roles of BacA and LmdC in H. neptunium and R. rubrum cell morphogenesis.

(A) Cell morphogenesis in H. neptunium wild-type (top) and ΔbacAD cells (bottom). In wild-type cells, BacA polymers form a complex with the DD-endopeptidase LmdC at the future stalked pole. The hydrolytic activity of LmdC helps to curve the cell wall at the incipient stalk base and thus determine stalk morphology. The zone of high positive inner cell curvature, and potentially the physical barrier constituted by the bactofilin polymer, prevent the movement of elongasome complexes from the mother cell body into the stalk, thereby limiting peptidoglycan biosynthesis to the stalked cell pole. At a later stage, the bactofilin-LmdC complex localizes close to the tip of the stalk and again generates a ring-shaped zone of positive cell curvature, thereby promoting the remodeling of the stalk tip into a spherical bud. At the onset of the budding process, elongasome complexes accumulate in the nascent bud by a so-far unknown mechanism. The positively curved bud neck and the bactofilin polymers prevent the movement of elongasome complexes from the bud into the stalk, thereby limiting cell growth to the bud compartment. In the ΔbacAD mutant, cells fail to concentrate LmdC at the future stalk base and bud neck. As a consequence, peptidoglycan biosynthesis is no longer limited to the different growth zones, leading to pseudo-stalk formation and unregulated bud expansion. (B) Bactofilin-mediated modulation of cell helicity in R. rubrum. BacARs (yellow) recruits LmdCRs (red) to the inner cell curvature. The hydrolytic activity of LmdC ultimately stimulates peptidoglycan biosynthesis at this position, leading to a reduction in cell helicity.

The dimorphic lifecycle of H. neptunium.

A motile, flagellated swimmer sheds its flagellum and forms a stalk at the opposite cell pole. A a defined point in the cell cycle, the terminal segment of the stalk dilates and develops into a new swimmer cell. After cell division, the newborn swimmer cell first needs to differentiate into a stalked cell to initiate a new round of cell division, whereas the stalked mother cell immediately enters the next replication cycle. The predominant growth zones (Cserti et al., 2017) are indicated in red.

Phenotypic analysis of H. neptunium bactofilin mutants.

(A) Immunoblot analysis of strains LE670 (wild type), EC28 (ΔbacA) and EC41 (ΔbacA PCu::PCu-bacA) grown in the absence (-bacA) and presence (+bacA) of 0.5 mM CuSO4, performed with an anti-BacA antibody. (B) Rescue of the phenotype of a ΔbacA mutant by ectopic expression of bacA. Cells of strain EC41 (ΔbacA PCu::PCu-bacA) were grown in copper-free medium, induced by the addition of copper, and imaged after the indicated time incubation times. Bar: 3 µm. (C) Quantification of the proportion of phenotypically abnormal stalked and budding cells in the cultures of strains EC41 (ΔbacA PCu::PCu-bacA) and EC43 (ΔbacA PZn::PZn-bacD) analyzed in panels B and E before (t=0 h) and 24 h after induction (n = 100 cells per time point). (D) Transmission electron micrographs of ΔbacD cells at the swimmer and stalk cell stage. Bar: 1 µm. (E) DIC images of a ΔbacA mutant overproducing BacD from a zinc-inducible promoter. Cells of strain EC43 (ΔbacA PZn::PZn-bacD) were grown in inducer-free medium, induced by the addition of 0.5 mM ZnSO4, and imaged after the indicated incubation times. Bar: 3 µm.

Growth of ΔbacAD cells in a microfluidic flow cell.

Cells of strain EC33 (ΔbacAD) were flushed into a microfluidic flow cell and imaged at the indicated time points. Bar: 3 µm.

Characterization of H. neptunium bactofilin mutants.

(A) Growth curves H. neptunium strains LE670 (wild type), EC28 (ΔbacA), EC23 (ΔbacD) and EC33 (ΔbacAD). Exponentially growing cultures were diluted into fresh media and monitored for 32 h using a microplate reader. Shown are representative data from one out of three independent experiments. (B) Muropeptide profiles of different H. neptunium strains. Cell walls of strains LE670 (wild type), EC28 (ΔbacA), EC23 (ΔbacD) and EC33 (ΔbacAD) were digested with cellosyl to release muropeptides, which were reduced and separated by HPLC. The identities of major muropeptide species, assigned based on the previously reported retention times (Cserti et al., 2017; Glauner, 1988), are given above the corresponding peaks. Tri, Tetra and Penta stand for N-acetylglucosamine-N-acetylmuramitol tripeptide, tetrapeptide and pentapeptide, respectively. “Anh” indicates muropeptides containing 1,6-anhydromuramic acid.

Stability of the fluorescent protein fusions used in this study.

Cells of strains LE670 (wild type), EC61 (bacA::bacA-eyfp), MO78 (bacA::bacAF130R-eyfp) and EC67 (bacD::bacD-venus) were subjected to immunoblot analysis using anti-GFP antibodies.

Localization dynamics of BacA-YFP.

Cells of strain EC61 (bacA::bacA-eyfp) were flushed into a microfluidic flow cell and imaged at 15-min intervals over multiple cell cycles. Shown are overlays of DIC and fluorescence images. Bar: 2 µm.

Dependence of BacD localization on BacA.

(A) Co-localization of BacA and BacD in cells of strain EC68 (bacA::bacA-eyfp bacD::bacD-mCherry). Shown are the individual fluorescence images used to generate the overlays in Figure 4D. (B) Random localization of BacD complexes in the absence of BacA. Cells of strain EC60 (ΔbacA PZn::PZn-bacD-venus) were grown in inducer-free medium and induced for 4 h with 0.5 mM ZnSO4 prior to imaging. Bar: 3 µm.

Localization and functionality of BacAF130R-YFP.

Cells of strain MO78 (bacA::bacAF130R-eyfp), producing the polymerization-deficient BacAF130R-YFP variant instead of the native protein, were grown to exponential phase and imaged. Bar: 3 µm.

Single-particle tracking analysis of BacA-YFP mobility.

Shown is a Gaussian mixture model (GMM) analysis of the mobility of the BacA-YFP variants analyzed in Figure 3B. The distributions of the frame-to-frame displacements in both x and y direction from single-particle tracking experiments were fitted to a two-component Gaussian function (sum in black), assuming a fast-diffusing mobile (dotted blue lines) and a slow-diffusing (dotted red lines) population.

Random localization of LmdC1-66-mCherry in the ΔbacAD background.

Cells of strain MO80 (ΔbacAD PCu::PCu-lmdC1-66-mCherry) were induced for 2 h with 0.5 mM CuSO4 prior to imaging. mCherry fluorescence is shown in yellow. Bar: 3 µm.

Analysis of BacA and LmdC from R. rubrum.

(A) Clearer visualization of the abnormal curvature of BacA- and LmdC-deficient R. rubrum cells by inhibition of cell division. Cells of strains SP68 (ΔlmdC), SP70 (ΔbacA) and SP116 (ΔlmdC ΔbacA) were treated with 5 μg/ml cefalexin (Cfx) for 6 h. Bar: 5 µm. (B) Muropeptide profiles of different R. rubrum strains. Cell walls of strains S1 (wild type), SP70 (ΔbacARs) and SP68 (ΔlmdC) were digested with cellosyl to release muropeptides, which were reduced and separated by HPLC chromatography. The identities of major muropeptide species, assigned based on the previously reported retention times (Glauner, 1988), are given above the corresponding peaks. Tetra stands for N-acetylglucosamine-N-acetylmuramitol tetrapeptide. “Anh” indicates muropeptides containing 1,6-anhydromuramic acid.

Cell lengths of R. rubrum strains.

Superplots showing the distribution of cell sinuosities in populations of strains S1 (wild type), SP105 (ΔbacARs PbacA-bacARs) and SP109 (bacARs::bacARs-mCherry). Small dots represent the data points, large dots represent the median values of the three independent experiments shown in the graphs (dark blue, light blue, grey). The mean of the three median values is indicated by a horizontal line. Whiskers represent the standard deviation. *** p<0.005; ns, not significant; t-test).

Lysis of E. coli upon heterologous expression of full-length R. rubrum lmdC.

Shown are cells of E. coli Rosetta(DE3)pLysS harboring pSP120 (PT7-lmdCRs) or the corresponding empty vector imaged 4h after the induction of gene expression with 0.5 mM IPTG. Arrowheads indicate ghost cells, cell debris, lysing cells or strongly elongated cells. Scale bar: 5 µm.

Analysis of R. rubrum strains producing fluorescent protein fusions.

(A) Levels of the LmdCNRs-mNeonGreen fusion in different R. rubrum strains. Strains SP98 (ΔlmdCRs bacARs::bacARs-mNeongreen), SP114 (bacARs::bacARs-mCherry PlmdC-lmdCNRs-mNeongreen), SP118 (ΔbacA ΔlmdC PlmdC-lmdC1-80-mNeongreen) and SP119 (bacA::bacA-mCherry ΔlmdC PlmdC-lmdC1-80-mNeongreen) were subjected to immunoblot analysis with an anti-mNeonGreen antibody. Predicted molecular weights: BacARs-mNeonGreen (44.8 kDa), LmdCNRs-mNeonGreen fusion (35.9 kDa) (B) Levels of the BacARs-mCherry fusion in different R. rubrum strains. Strains SP109 (bacARs::bacARs-mCherry), SP114 (bacARs::bacARs-mCherry PlmdC-lmdCN-mNeongreen), SP117 (bacA ::bacA-mCherry ΔlmdC) and SP119 (bacA ::bacA - mCherry ΔlmdC PlmdC-lmdCN-mNeongreen) were subjected to immunoblot analysis with an anti-mCherry antibody. The predicted molecular weight of BacARs-mCherry is 44.9 kD. (C) Co-localization of BacARs-mCherry and LmdCRs-mNeongreen in R. rubrum cells carrying the native lmdCRs gene. The Pearson’s Correlation Coefficient (PCC) of the two fluorescence signals in a random subpopulation of cells (n=177) is 0.54. Scale bar: 5 µm.