Direct MinE-membrane interaction contributes to the proper localization of MinDE in E. coli

Abstract

Dynamic oscillation of the Min system in Escherichia coli determines the placement of the division plane at the midcell. In addition to stimulating MinD ATPase activity, we report here that MinE can directly interact with the membrane and this interaction contributes to the proper MinDE localization and dynamics. The N-terminal domain of MinE is involved in direct contact between MinE and the membranes that may subsequently be stabilized by the C-terminal domain of MinE. In an in vitro system, MinE caused liposome deformation into membrane tubules, a property similar to that previously reported for MinD. We isolated a mutant MinE containing residue substitutions in R10, K11 and K12 that was fully capable of stimulating MinD ATPase activity, but was deficient in membrane binding. Importantly, this mutant was unable to support normal MinDE localization and oscillation, suggesting that direct MinE interaction with the membrane is critical for the dynamic behavior of the Min system.

Fig. 1

Evidence of direct MinE–membrane interaction. A. Quantitiative analyses of the percentage of pelleted full-length MinE, MinE^1–31 and MinE^32–88 with liposomes in the co-sedimentation assays. B–F. EM micrographs of free liposomes in solution (B and C) and MinE-associated tubules encircled by a coat (D–F; arrow). Micrographs (D–F) cover the same area and the boxed areas were viewed at higher magnification during acquisition. The width of this peripheral coat was approximately 9.4 nm.

Fig. 4

Abilities of the wild-type and mutant MinE proteins to interact with MinD, to stimulate MinD ATPase activity and to deform liposomes in vitro. A and B. The interaction between MinD and MinE was assayed using the bacterial two-hybrid system. Both liquid and plate assays are shown. Two dilutions (10⁻⁴, 10⁻⁵) of each culture were spotted on the same indicator plate (B). C. Stimulation of MinD ATPase activity by various MinE mutants in the presence or absence of liposomes. D. The full-length MinE carrying mutations in the N-terminal domain or in D45 and V49 did not co-sediment with liposomes. E–J. Degree of liposome deformation associating with the wild-type and mutant MinE proteins (C1, C2, C3 and D45A/V49A) was examined under EM.

Fig. 2

Electrostatic force is involved in mediating the MinE–membrane interaction. A. The co-sedimentation of full-length MinE with liposomes was disrupted by increasing concentrations of NaCl and KCl. The Tris-Tricine-SDS gel image exemplified an experiment with different NaCl concentrations that were used in the analyses. The lower arrow indicates the position of MinE. The higher arrow indicates an additional 28 kDa band induced by the presence of liposomes. S, supernatant; P, pellet. B. MinE showed a binding preference for the anionic phospholipids PG and CL. C. The MinE–membrane interaction was stabilized under high salt conditions by the presence of 50 mol% cardiolipin (PE : PG : CL = 36:14:50). The lower and higher panels show experiments with or without the presence of 18 µM BSA respectively. MinE was supplied at 6 µM in these reactions. Note that the 28 kDa band was relocated to the supernatant fraction in the presence of BSA, indicating its appearance in the pellet fraction was mostly non-specific. The additional bands in the lower gel were impurities of the BSA solution. D. Addition of cationic lysyl-DPPG to liposomes interfered with the co-sedimentation of MinE and MinE^1–31 with liposomes. The reactions were incubated for 60 min prior to centrifugation.

Fig. 3

Membrane association of various mutants of MinE^1–31-Yfp in Δmin cells. A. MinE^1–31 contains three clusters of basic residues (red). This region of MinE^1–31 was engineered to disrupt the positively charged residues in these clusters. The substituted residues are coloured in blue. B. Localization of wild-type (1) and mutant MinE^1–31-Yfp (4–9) as well as untagged Yfp (2) and Yfp-MinD (3) in Δmin cells. C. Western blot analyses on fractionated cells expressing the mutant forms of MinE^1–31-Yfp. The membrane fraction was estimated to be 30-fold more concentrated than the cytosolic fraction in this experiment to facilitate comparisons between different MinE^1–31-Yfp bound to the membrane. The arrow indicates MinE^1–31-Yfp. D. Analysis of fluorescence distribution across the cell width and G₂ of MinE^1–31-Yfp, Yfp and Yfp-MinD (top), mutant MinE^1–31-Yfp containing C1, C2 or C3 mutations (middle) and mutant MinE^1–31-Yfp containing R10G, K11E or K12E mutations (bottom).

Fig. 5

Effects of the N-terminal mutations on the localization of MinD and MinE. A. Colocalization of MinD with MinE lacking various positively charged residues (mutants C1, C2 and C3). p: peripherally localized MinD; d: diffusely distributed MinE; f: MinE focus. Yellow and cyan arrows indicate the MinD polar zones and the MinE rings respectively. The green arrow points out a fluorescent spot found in a minicell. B. Movement of MinD fluorescence in the presence of the C1 mutant MinE. The top cell started to disassemble MinD at one pole. White arrows indicate directions of polar zone growth and shrinkage. Another two cells showed atypical oscillation with a peculiar stage of MinD on the entire periphery of the cells (frames marked by the white dots) before re-initiating disassembly from the extreme cell pole (red arrows). C. Western blot analyses using the anti-Gfp antibody to detect levels of Yfp-MinD and MinE-Cfp in cells. The anti-Gfp antibody used here displayed different abilities to recognize Yfp and Cfp, thus the band intensity did not reflect the real ratio of Yfp-MinD to MinE-Cfp in cells.

Fig. 6

Model of MinE association with the membrane and a possible role of this membrane association in the Min system. A. The direct MinE–membrane interaction is facilitated at the membrane proximity through MinD's recruitment. However, this association is unstable on membranes where different phospholipids are randomly distributed. B. The MinE–membrane interaction can be stabilized by the presence of a high local concentration of cardiolipin at the midcell. A MinE ring subsequently appears at this site. The darker shade of the membrane represents the area enriched in cardiolipin. C. Three actions are required for dissociation of MinD from the membrane, including ATP hydrolysis in MinD, formation of the MinE ring, and MinE-induced membrane deformation. The degree of membrane deformation and the organization of MinE molecules in this process are under investigation.