The actin homologue MreB organizes the bacterial cell membrane

Abstract

The eukaryotic cortical actin cytoskeleton creates specific lipid domains, including lipid rafts, which determine the distribution of many membrane proteins. Here we show that the bacterial actin homologue MreB displays a comparable activity. MreB forms membrane-associated filaments that coordinate bacterial cell wall synthesis. We noticed that the MreB cytoskeleton influences fluorescent staining of the cytoplasmic membrane. Detailed analyses combining an array of mutants, using specific lipid staining techniques and spectroscopic methods, revealed that MreB filaments create specific membrane regions with increased fluidity (RIFs). Interference with these fluid lipid domains (RIFs) perturbs overall lipid homeostasis and affects membrane protein localization. The influence of MreB on membrane organization and fluidity may explain why the active movement of MreB stimulates membrane protein diffusion. These novel MreB activities add additional complexity to bacterial cell membrane organization and have implications for many membrane-associated processes.

Figure 1. Disruption of the MreB cytoskeleton is accompanied by an aberrant membrane stain.

(a) Phase contrast images of B. subtilis cells (upper panel) expressing GFP-MreB (middle panel), and fluorescent Nile Red membrane stains (lower panel), are depicted in the absence or presence of the proton ionophore CCCP. Some of the Nile Red foci appearing with CCCP are highlighted with arrows (see Supplementary Fig. 5a for more examples). The CCCP-triggered local enrichment of MreB is not caused by the weak dimerization property of GFP, which was recently shown to stimulate protein clustering (Supplementary Fig. 5b), or by artificial overproduction (Supplementary Fig. 6). Strain used: B. subtilis YK405 (gfp-mreB). (b) Incubation of F₁F_o ATP synthase-deficient cells with CCCP also results in a rapid (within 2 min) emergence of Nile Red foci, ruling out that the appearance of Nile Red foci is due to a reduction of ATP levels (see Supplementary Fig. 1c for more examples). Strain used: B. subtilis HS13 (Δatp). (c) Comparison of Nile Red foci, and the localization of the integral membrane protein F₁F_o ATP synthase fused to GFP (AtpA-GFP). The lack of overlap in fluorescence signals rules out membrane invagination, or aberrantly formed septa (see arrows) as an explanation for the strong local Nile Red fluorescence (see Supplementary Fig. 4 for more examples). The reason that in some cases the Nile red stain appears to protrude into the cytoplasm is a consequence of high membrane fluorescence originating from slightly above and below the exact focal plane. Strain used: B. subtilis BS23 (atpA-gfp). (d) CCCP does not affect Nile Red fluorescent membrane stain in Staphylococcus aureus and Corynebacterium glutamicum which both lack MreB (see Supplementary Fig. 1d for more examples). Strains used: C. glutamicum RES167 and S. aureus RN4220. (e) Colocalization of GFP-MreB (upper panel) with Nile Red (middle panel) in B. subtilis strains encoding wild type MreD (wt), and a 15 amino acid C-terminal truncation of MreD (−15aa), which results in a mild delocalization of MreB in low Mg²⁺ medium (see Supplementary Fig. 1e for more examples). Strains used B. subtilis HS35 (gfp-mreB mreD wt) and HS38 (gfp-mreB mreD_1-471/−15aa). Scale bar, 2 μm.

Figure 2. Distorted lipid staining requires the MreB cytoskeleton.

(a) CCCP-dependent fluorescent Nile Red foci occur in B. subtilis strains deficient for single MreB homologues, but are absent in strains that lack all three MreB homologues (ΔmreB, Δmbl, ΔmreBH). Some of the Nile Red foci appearing with CCCP are highlighted with arrows. See also Supplementary Fig. 7 for the analysis of the foci frequency. Strains used: B. subtilis 3728 (ΔmreB), 4261 (Δmbl), 4262 (ΔmreBH) and 4277 (ΔmreB, Δmbl, ΔmreBH). (b) Colocalization of GFP-MreB (left panels) with Nile Red (middle panels) in B. subtilis cells depleted for RodA. The MreB cytoskeleton is disrupted with CCCP. Some of the Nile Red foci appearing with CCCP are highlighted with arrows. Strain used: B. subtilis HS36 (gfp-mreB Pspac-rodA). Scale bar, 2 μm.

Figure 3. Analysis of cell membrane fluidity.

(a) Nile Red fluorescence intensity of B. subtilis cells grown in room temperature, 30 °C or 48 °C followed by shift to 30 °C, was measured fluorometrically. To prevent rapid adaptation of membrane fluidity, the measurements were carried out with a lipid desaturase (des)-deficient strain. The curves represent average values with s.d. of three replicate measurements. Strain used: B. subtilis HB5134 (Δdes). (b) Microscopic analysis of fluid membrane domains using Laurdan GP. The images show a colour-coded Laurdan GP intensity map in which red indicates regions of increased fluidity. Membrane fluidity is determined for untreated B. subtilis cells (left panel) and cells treated with CCCP (right panel). See also Supplementary Fig. 10 for the measurement of the average fluidity change. (c) Colocalization of Nile Red foci and Laurdan GP in CCCP-treated B. subtilis cells. Strain used: B. subtilis 168 (wild type). Scale bar, 2 μm.

Figure 4. Localization of RIFs in untreated B. subtilis cells.

(a) RIFs can be visualized in normally growing untreated B. subtilis cells using the fluid membrane dye DiI-C12 (left panel). Staining with the long acyl-chain variant, DiI-C18, reveals a regular (smooth) membrane signal (right panel). Strain used: B. subtilis 168 (wild type). (b) Colocalization of DiI-C12 and GFP-MreB (see Supplementary Fig. 11 for more examples and statistical analysis of colocalization). Strain used: B. subtilis YK405 (gfp-mreB). (c) Fluorescence intensity correlation graphs are shown for the cells in panel b. The graphs display a pixel by pixel intensity correlation between DiI and GFP-MreB fluorescence, and shows the Pearson’s correlation coefficients (Rr) (see Supplementary Fig. 11 for details). It should be mentioned that some fluorescence correlation between DiI-C18 and GFP remains since both DiI-C18 and MreB are present in the cell membrane. Strain used: B. subtilis YK405 (gfp-mreB). Scale bar, 2 μm.

Figure 5. DiI-C12-stained RIFs are MreB dependent.

(a) DiI-C12 fluorescence of widefield (upper panel), and maximum intensity projection of a Z-stack (lower panel) of B. subtilis ΔmreC and ΔmreD cells, and cells that lack the three MreB homologues (ΔmreB, Δmbl, ΔmreBH). See Supplementary Movie 1 for 3D-reconstruction, Supplementary Movie 2 for corresponding raw and deconvolved images, and Supplementary Fig. 11c for more examples. Strains used: B. subtilis 3481 (ΔmreC), 4311 (ΔmreD), 4277 (ΔmreB, Δmbl, ΔmreBH, ΔrsgI). (b) Colocalization of DiI-C12 and GFP-MreB in the absence of MreBCD. The maximum intensity projections of Z-stacks are depicted. See Supplementary Movie 3 for 3D-reconstruction and Supplementary Movie 4 for corresponding raw and deconvolved images. Strain used: B. subtilis HS35 (gfp-mreB ΔmreBCD). Scale bar, 2 μm.

Figure 6. Membrane fluidity increases in the absence of MreB.

(a) Overall membrane fluidity of several cytoskeletal mutant strains was measured as Laurdan GP of logarithmically growing B. subtilis cells. As a positive control, an increase in fluidity (= decrease in Laurdan GP) was established by adding the membrane fluidizer benzyl alcohol (BA; 30 mM) to wild type cells. The diagram depicts the average values and s.d. of three independent measurements. (b) Ratios between chain lengths of the major fatty acids (C17 and C15) and ratios between the iso and anteiso forms of fatty acids are depicted for the different strains (see Supplementary Table 1 for the detailed composition). High C17/C15 or high iso/anteiso ratios indicate a reduced fluidity. The diagram depicts the average values and s.d. of two independent analyses. Strains used: B. subtilis 168 (wild type), 4264 (ΔrsgI), 3481 (ΔmreC), 4277 (ΔmreB, Δmbl, ΔmreBH, ΔrsgI).

Figure 7. Aberrant localization of different membrane proteins.

(a) The localization of several membrane proteins (FruA, P16.7, YhaP, YqfD and Rny) is changed after the MreB cytoskeleton was disturbed by addition of CCCP. Some of the foci appearing with CCCP are highlighted with arrows. The localization of other proteins (AtpA and SdhA) remains unaffected. See Supplementary Fig. 12 for more examples, and colocalization with Nile Red. Strains used: B. subtilis FruA-GFP, 110WA (p16.7-gfp), yhaP-gfp, yqfD-gfp, 3569 (rny-gfp), BS23 (atpA-gfp), BS112 (sdhA-gfp). (b) The CCCP-induced clustering of membrane proteins FruA, P16.7, YhaP and YqfD is absent in strains deficient for MreB homologues. Strains used B. subtilis HS43 (rny-gfp, ΔmreB*), HS44 (fruA-gfp, ΔmreB*), HS45 (P16.7-gfp, ΔmreB*), HS46 (yhaP-gfp, ΔmreB*), HS47 (yqfD-gfp, ΔmreB*). mreB* designates (ΔmreB, Δmbl, ΔmreBH, ΔrsgI). Scale bar, 2 μm.

Figure 8. MreB movement influences membrane protein diffusion.

(a) Kymographs visualizing the diffusion of fructose permease (FruA) and F₁F_o ATP synthase (AtpA) in the absence and presence of vancomycin (van), which blocks MreB movement. Time lapse series with 20 s length and 100 ms time resolution was acquired using TIRF microscopy. See Supplementary Movie 5 for corresponding raw image series. (b) In TIRF microscopy, only proteins within the vicinity of the evanescent light (<200 nm distance from the coverslip surface) are subject to fluorophore (GFP) bleaching. The graphs depict the averaged, normalized and background-subtracted fluorescent signals of fructose permease (FruA) and F₁F_o ATP synthase (AtpA) in the presence (− van) and absence (+ van) of MreB movement. A significant increase in bleaching is observed when movement of MreB is inhibited with vancomycin, which indicates a reduced diffusion of proteins in and out of the range of the evanescent light. The curve fits were performed using a two-phase exponential decay model composed of a rapid bleaching of GFP within the range of the evanescent wave, and slower (diffusion limited) bleaching of the whole cell fluorescence (see Supplementary Fig. 14 for details). (c) Rate constants and s.e. of slow (diffusion limited) bleaching kinetics for Fructose permease (FruA), RNaseY (Rny), F₁F_o ATP synthase (AtpA) and Succinate dehydrogenase (SdhA) in the absence and presence of vancomycin. Increased rate constant indicates a reduced exchange of protein between the TIRF-illuminated area and rest of the cell surface. The number of analysed cells, goodness-of-fit and s.e. are provided in Supplementary Fig. 14. Strains used: B. subtilis BS23 (atpA-gfp), BS112 (sdhA-gfp), FruA-GFP and 3569 (rny-gfp).