Identification of residues critical for the membrane-binding activity of BacA in vivo.

(A) Schematic representation of the BacA-mVenus fusion protein used in this study. The proposed membrane-targeting sequence is highlighted in red. The sequence at the bottom shows that result of an amphipathic helix prediction for BacA using the AMPHIPASEEK software (Sapay et al., 2006). Residues predicted to be located in an unstructured, randomly coiled region are labeled with “c”. (B) Localization patterns of mutant BacA-mVenus variants. ΔbacAB cells producing BacA-mVenus or mutant variants thereof (strains LY84, LY89, LY90, LY97, LY111, LY112, LY113, LY119) were analyzed by phase contrast and fluorescence microscopy. The outlines of the cells are shown in the fluorescence images. Demographs summarizing the single-cell fluorescence profiles obtained from random subpopulations of cells are given next to the respective fluorescence images. The numbers of cells analyzed are: WT (130), Δ2-8 (292), F130R (156), F2Y (138), F2E (194), K4S-K7S (151), K4E-K7E (382), F2E-K4E-K7E (130). The vertical red line indicates the junction between cell body and the stalk. Scale bar: 2 μm. (C) Helical wheel diagram of the first eight amino acids of BacA. Residues are colored by properties: hydrophobic (gray), basic (blue), uncharged (yellow).

Verification of residues M1, F2 and K4/K7 as critical components of the BacA MTS.

(A) Mobility of the indicated BacA-mVenus fusion proteins. Shown is the average mean-squared displacement (MSD) (± SD) as a function of time, based on single-particle tracking analysis. The fitted lines were obtained by linear regression analysis. (B) Cell fractionation experiment investigating the membrane-binding activity of BacA-mVenus (WT) or a mutant variant lacking the predicted MTS (Δ2-8). Whole-cell lysates (W) as well as the soluble (S) and pellet (P) fractions of cells producing the indicated proteins were subjected to immunoblot analysis with an anti-GFP antibody, detecting the BacA-mVenus fusion protein. As controls, the same samples were probed with antibodies raised against the soluble cell division regulator MipZ (Thanbichler and Shapiro, 2006) or the membrane-bound flagellar L-ring subunit FlgH (Mohr et al., 1996) from C. crescentus. (C) As in panel B, but for cells producing mutant BacA-mVenus variants with single or multiple amino-acid exchanges in the predicted MTS. Shown are representative images (n=3 independent replicates). The strains used are given in the legend to Figure 1B.

Co-sedimentation analysis of the association of various BacA variants with liposomes.

The indicated proteins (20 µM) were incubated without (-) or with (+) liposomes (0.4 mg/mL) prior to ultra-centrifugation. The supernatant and pellet fractions of each mixture were analyzed by SDS gel electro-phoresis. Shown are scans of representative gels and a quantification of the average relative signal intensities (± SD) obtained for the different fractions (n=3 independent replicates).

Interplay between BacA assembly and membrane binding.

(A) Cell fractionation experiment investigating the membrane-binding activity of the polymerization-deficient F130R variant of BacA-mVenus in vivo (LY119). The analysis was performed as described for Figure 2B. (B) Co-sedimentation analysis of the association of BacA-F130R with liposomes in vitro, performed as described for Figure 3. (C) Role of polymerization in the membrane association of a BacA-mVenus variant carrying the membrane-targeting sequence of E. coli MreB. Shown are phase contrast and fluorescence images of ΔbacAB mutants (LY103, LY123) producing either a BacA-mVenus variant in which the MTS is replaced by two tandem copies of the N-terminal amphiphilic helix of E. coli MreB (MreBWT) or a polymerization-deficient variant thereof (MreBF130R). Demographs summarizing the single-cell fluorescence profiles obtained from random subpopulations of cells are given next to the respective fluorescence images. The numbers of cells analyzed are: MreBWT (126), MreBF130R (169). The vertical red line indicates the junction between cell body and the stalk. Scale bar: 2 μm. (D) Mobility of the indicated BacA-mVenus fusion proteins. Shown is the average mean-squared displacement (MSD) (± SD) as a function of time, based on single-particle tracking analysis. The mVenus, Δ2-8 and WT data are taken from Figure 2A and shown for comparison. (E) Cell fractionation experiment investigating the membrane-binding activity of MreBBacA-mVenus (MreBWT) and its polymerization-deficient F130R variant (MreBF130R) in vivo (LY103, LY123). The analysis was performed as described for Figure 2B.

Molecular dynamics simulation of the interaction between the BacA MTS and a model mem-brane.

(A) Snapshot of the molecular dynamics (MD) simulation system showing the 10-mer peptide MFSKQAKSNN (BacA1-10; red) after binding to the lipid bilayer. The water is shown in surface representation. K+ and Cl- counterions are not shown. (B) Close-up view of a representative snapshot from the MD simulation visualizing the binding mode of the peptide on the membrane surface. (C) Structural overlay of 40 snapshots from the MD simulation, taken after constant time intervals from the trajectory. (D) Density profiles of individual residues in the wild-type peptide along the membrane normal, i.e. the z-component of the distance vector from the center-of-mass (COM) of the bilayer, with the membrane midplane located at zero. The vertical dashed black line indicates the maximum of the density distribution of the lipid headgroup phosphates.

Contact numbers and interaction energies for different peptide-lipid bilayer interactions.

(A) The graph shows the total number of contacts between individual residues in the wild-type, K4S-K7S and F2Y peptides and the lipid bilayer as well as the number of contacts with PG lipids and GLY lipids. A contact between a peptide residue and a lipid was defined to exist if any two non-hydrogen atoms of the residue and a lipid molecule were within a distance of 0.5 nm to each other. Contacts were counted for each frame of the MD trajectories and averaged. Multiple contacts between a peptide and a lipid molecule were treated as a single contact, so that the number of contacts counted was either 1 or 0. The statistical errors plotted were obtained from the difference between the two different sets of 500-ns simulations, starting with peptides in an unfolded or α-helical conformation, respectively. (B) Energies of the interactions between individual residues in the wild-type, K4S-K7S and F2Y peptides and the lipid bilayer. The interaction energies plotted are the combined interaction energies of all Coulomb and van-der-Waals interactions in the force field averaged over the simulation trajectories.

Conservation of the N-terminal regions of bactofilin homologs in different bacterial phyla.

The pie chart in the middle shows the relative distribution of the 14337 unique bactofilin homologs analyzed among the indicated bacterial phyla. The sequence logos give the most widespread N-terminal motifs obtained either by a global analysis of all 14337 bactofilin sequences (global consensus) or by an analysis of subsets of these sequences from specific phyla.

Interaction of BacA with its client protein PbpC.

(A) Localization patterns of different PbpC variants. ΔbacB ΔpbpC cells producing mVenus-PbpC (LY75) or mutant variants thereof lacking region C1 (LY76) or carrying an unstructured region from C. crescentus DipM in place of the unstructured region connecting region C1 and the transmembrane helix (LY77) were analyzed by phase contrast and fluores-cence microscopy. The outlines of the cells are shown in the fluorescence images. Demographs summarizing the single-cell fluorescence profiles obtained from random subpopulations of cells are given next to the respective fluorescence images. The numbers of cells analyzed are: LY75 (158), LY76 (253), LY77 (119). Scale bar: 2 μm. (B) Biolayer interferometric analysis of the interaction between PbpC1-13aa and BacA. A synthetic peptide comprising the first 13 amino acids of PbpC (PbpC1-13) was immobilized on a biosensor and probed with increasing concentrations of BacA. After the association step, the sensor was transferred to a protein-free buffer to monitor the dissociation reaction. The graph shows a representative titration series (n=3 independent replicates). (C) Comparison of the interaction of PbpCaa1-13 with BacA and its poly-merization deficient F130R variant, performed as described in panel B. (D) Mapping of the PbpC binding site on BacA by hydrogen-deuterium exchange (HDX) mass spectrometry. The plots show the extent of deuterium uptake by three representative peptides obtained after peptic digestion of BacA protein (2.5 µM) that had been incubated in the absence or presence of the PbpC1-13 peptide (10 µM) for the indicated time periods (see Supplementary file 4 for the full set of peptides). (E) Mapping of the differences in deuterium uptake observed at t=1000 s onto the solid-state NMR structure of BacA (Shi et al., 2015).

Contribution of PbpC to BacA membrane association.

C. crescentus ΔbacAB ΔpbpC cells producing the indicated BacA-mVenus variants (WT, Δ2-8, F130R) under the control of a xylose-inducible promoter and PbpC1-132-mCherry under the control of a vanillate-inducible promoter (strains MAB575, MAB576 and MAB577) were grown in the presence of xylose (left) or both xylose and vanillate (right) prior to microscopic analysis. The images show representative fluorescence micrographs, with the cell outlines indicated in white. Arrowheads indicate polar PbpC1-132-mCherry foci. Demographs summarizing the single-cell BacA-mVenus fluorescence profiles obtained from random subpopulations of cells are provided next to the respective fluorescence images. The number of cells analyzed is shown in the top left-hand corner of each graph. The schematics on top illustrate the protein constructs used for the analysis (not to scale). Bar: 3 µm.

Ultrastructure of different BacA variants.

Wild-type BacA, its MTS-free Δ2-8 variant and its polymerization-deficient F130R variant (xx mg/ml) were stained with uranyl acetate and visualized by transmission electron microscopy. Bar: 100 nm.

Size-exclusion chromatography analysis of wild-type BacA and its F130R variant.

The indicated proteins were applied to a Superdex 200 size-exclusion column and detected photo-metrically at a wavelength of 280 nm. The following standard proteins were analyzed as a reference to calibrate the column: Thyroglobulin (669 kDa), Ferritin (440 kDa), Aldolase (158 kDa), Conalbumin (75 kDa), Ovalbumin (44 kDa), Ribonuclease A (14 kDa).

Stability of different BacA-mVenus variants.

Derivates of strain JK5 (ΔbacAB) carrying the indicated alleles of bacA-mVenus under the control of the xylose-inducible Pxyl promoter were grown overnight, diluted to an OD600 of ∼ 0.1 and incubated for another hour. The strains were then induced with 0.03% xylose for 1 h and subjected to immunoblot analysis with an anti-GFP antibody. Strain JK5 was analyzed as a negative control (NC). The positions of standard proteins (in kDa) are indicated on the left side of the images.

Localization patterns of different BacA-mVenus variants.

ΔbacAB cells producing the indicated BacA-mVenus variants (strains LY95, LY88, LY96, LY91, LY92) were analyzed by phase contrast and fluorescence microscopy. The outlines of the cells are shown in the fluorescence images. Demographs summarizing the single-cell fluorescence profiles obtained from random subpopulations of cells are given next to the respective fluorescence images. The numbers of cells analyzed are: S3A (109), K4S (150), Q5A (121), A6S (184), K7S (128). The vertical red line indicates the junction between cell body and the stalk. Scale bar: 2 μm.

Subcellular localization of the single-molecule tracks obtained for different BacA-mVenus variants.

The single-particle tracks determined for wild-type BacA-mVenus, free mVenus and the indicated BacA-mVenus variants were mapped onto phase contrast images of the cells in which they were recorded. The images show the results obtained from representative cells. Red lines indicate slow-moving particles, green lines indicate fast-moving particles (as defined in Supplementary file 1). Bar: 1 µm.

Solubility of different CreS-mNeonGreen variants.

(A) Schematics showing the CreS-mNeonGreen variants analyzed in this study. The membrane-targeting regions of CreS and BacA are shown in blue and red, respectively. (B) Cell fractionation experiment investigating the membrane-binding activity of CreS-mNeonGreen (WT) or mutant variants lacking the predicted membranetargeting region (Δ27) or containing the BacA MTS instead of the native membrane-targeting region (BacACreS). Whole-cell lysates (W) as well as the soluble (S) and pellet (P) fractions of cells producing the indicated proteins were subjected to immunoblot analysis with an anti-mNeonGreen antibody. As controls, the same samples were probed with antibodies raised against the soluble cell division regulator MipZ (Thanbichler and Shapiro, 2006) or the membrane-bound flagellar L-ring subunit FlgH (Mohr et al., 1996) from C. crescentus. Shown are representative images (n=3 independent replicates). (C) Quantification of the relative amounts of CreS-mNeonGreen or its mutant variants in the soluble and pellet fractions. The intensities of the signals obtained for the soluble and pellet fractions in the analysis described in panel B were quantified and normalized to the total intensity in these two fractions. Data represent the average (± SD) of three independent replicates.

Density profiles for different MTS variants determined by molecular dynamics simulation.

Density profiles of individual residues in the (A) wild-type, (B) F2Y and (C) K4S-K7S peptides along the membrane normal. The vertical dashed black line indicates the maximum of the density distribution of the lipid headgroup phosphates. The data shown in panel A are reproduced from Figure 5D for comparison.

Assessment of the conservation of the bactofilin membrane-targeting sequence.

(A) Analysis pipeline used to analyze the conservation of the N-terminal regions among bacterial bactofilin homologs. (B) Relative abundance of annotated bactofilin homologs in the different domains of life. (C) Abundance of bactofilins with predicted transmembrane helices in different bacterial lineages. (D) Distribution of bactofilin homologs without predicted transmembrane helices among different bacterial phyla. Small phyla contributing less than three percent of the total number of sequences have been aggregated in the category ‘Others’.

Conserved N-terminal motifs in bactofilin homologs from different phyla.

The sequence logos give the most frequent N-terminal motifs obtained either by a global analysis of all 14,337 bactofilin sequences (global consensus) or by an analysis of subsets of these sequences from specific phyla. The number of bactofilin homologs that show N-terminal sequences corresponding to a specific motif and the likelihood for the existence of this motif are indicated on the left side of each sequence logo.

Localization patterns of mVenus-PbpC and BacA-Venus in different strain backgrounds.

(A) Localization of mVenus-PbpC in the ΔbacB ΔpbpC (LY75) and the ΔbacA ΔpbpC (LY72) backgrounds. (B) Localization of BacA-Venus in the wild-type (MT256) and ΔpbpC (JK136) backgrounds. Cells were analyzed by phase contrast and fluorescence microscopy. The outlines of the cells are shown in the fluorescence images. Demographs summarizing the single-cell fluorescence profiles obtained from random subpopulations of cells are given next to the respective fluorescence images. The numbers of cells analyzed are: LY75 (158), LY72 (697), MT256 (264), JK136 (222). Scale bar: 1 μm.

Sequence alignment of the cytoplasmic tail of PbpC homologs.

The schematic at the top shows the structure of PbpC. Conserved domains are shown in different colors. The cytoplasmic tail of PbpC comprises conserved Region C1 (aa 1-13), a proline-rich region (aa 14-62), conserved Region C2 (aa 63-70), and a region rich in positively charged amino acids (aa 71-83), located adjacent to the transmembrane helix. Abbreviations: TG: transglycosylase domain, TP: transpeptidase domain. The NCBI identifiers of the proteins analyzed are given on the left.

Stability of the mVenus-PbpC fusion proteins used in this study.

ΔbacB ΔpbpC cells producing wild-type mVenus-PbpC (WT; LY75), an mVenus-PbpC variant lacking the conserved region C1 (Δ2-13; LY76) or an mVenus-PbpC variant containing an unstructured region from C. crescentus DipM in place of the unstructured region in between regions C1 and C2 (chimera; LY77) were grown overnight, diluted to an OD600 of ∼0.1 and incubated for another hour. Subsequently, the cells were induced for 1.5 h with 0.3 % xylose and subjected to immunoblot analysis with an anti-GFP antibody. A ΔbacAB mutant producing BacA-mVenus and the wild-type strain CB15N (NC) were analyzed as positive and negative controls, respectively. The positions of standard proteins (in kDa) are indicated on the left of image.

Relevance of BacA binding for the localization and functionality of PbpC under phosphate-limiting conditions.

(A) Phase contrast and fluorescence images of the C. crescentus wild-type (WT), a ΔbacB ΔpbpC mutant (LY71) and a ΔbacB ΔpbpC mutant producing either mVenus-PbpC (LY75) or an N-terminally truncated variant thereof lacking region C1 (Δ2-13; LY76) under the control of a xylose-inducible promoter after 24 h of cultivation in phosphate-limited (M2G-P) medium containing 0.3% xylose. The demographs at the bottom show the fluorescence profiles of a representative subpopulation of cells stacked on top of each other and sorted according to cell length. The numbers of cells analyzed are: 428 (WT), 415 (LY71), 659 (LY75), 753 (LY76). Bar: 3 µm. (B) Quantification of stalk lengths in the cultures described in panel A. Shown are bee swarm plots of the data. The red dot indicates the median, the lines indicate the standard deviation. (C) Stability of the indicated mVenus-PbpC variants under phosphate starvation. The indicated strains were cultivated for 24 h in phosphate-limited medium in the absence (uninduced) or presence (induced) of 0.3% xylose and subjected to immunoblot analysis with an anti-GFP antibody. The positions of mVenus-PbpC and free mVenus are indicated on the right.

Biolayer interferometry analysis of the interaction between PbpCaa1-13 and BacA.

(A) Control showing the interaction of wild-type BacA (50 µM) with an unmodified biosensor. (B) Affinity of BacA for the immobilized PbpC1-13 peptide. The final wavelength shifts measured for the different association curves in Figure 8B were plotted against the corresponding BacA concentrations. Data represent the average (± SD) of three independent replicates.

Mapping of the PbpC-binding site of BacA by hydrogen-deuteriumexchange (HDX) analysis.

(A) Schematic showing the domain structure of BacA. (B) HDX analysis of BacA. Purified BacA (2.5 µM) was incubated in deuterated buffer for the indicated time intervals either alone (Apo) or in the presence of PbpC1-13 peptide (10 µM). Shown is the degree of HDX along the primary sequence of BacA in the indicated conditions. The color scale is given on the right. The schematic at the top displays the predicted secondary structure of BacA. The black bars represent peptides of BacA that were analyzed for HDX. Residue-specific HDX information was obtained from these overlapping peptides by using the shortest peptide covering a given residue. Gaps indicate amino acid sequences not covered by any peptide. (C) HDX difference map. Shown are the residue-specific differences in HDX between BacA in the presence of PbpC1-13 peptide and BacA alone that were obtained after the indicated incubation times. The data are projected onto the primary sequence of BacA. The color code is given on the right. Blue color denotes regions showing reduced HDX in the presence of PbpC1-13.

Levels and stability the fluorescent fusion proteins used in colocalization studies.

C. crescentus ΔbacAB ΔpbpC cells producing the indicated BacA-mVenus variants under the control of a xylose-inducible promoter and PbpC1-132-mCherry under the control of a vanillate-inducible promoter (strains MAB575, MAB576 and MAB577) were grown in the presence of xylose (xyl) and/or vanillate (van) and subjected to immunoblot analysis with anti-GFP and anti-mCherry antibodies. The cells analyzed were from the same cultures as those in Figure 9. The positions of the fusion proteins and of the corresponding free fluorescent proteins are shown on the right.