ATP-driven membrane binding and polymerization of bacterial actin MreB promotes local membrane fluidization Authors: Ingrid E. Adriaans, Ana Álvarez-Mena, Céline Dinet, Estelle Morvan, Kee Siang Lim, Erick J. Dufourc, Richard Wong, Arnaud Chastanet, Alphée Michelot, Rut Carballido-López, Birgit Habenstein Published: 2026 ABSTRACT The bacterial actin homolog MreB plays a key role in rod cell shape determination. We recently showed that MreB from the Gram-positive bacterium Geobacillus stearothermophilus (MreBGs) polymerizes into straight pairs of protofilaments in the presence of both ATP and a lipid surface. Membrane interaction is thought to be mediated by electrostatic interactions with anionic lipids, with final anchoring relying on two spatially close hydrophobic motifs that protrude from the MreBGs monomers, forming a putative membrane-insertion domain. Here, we determined the binding properties of ATP and ADP to MreBGs using fluorescence anisotropy and monitored ATP-mediated binding and polymer formation on lipid bilayers using liposome-binding assays and atomic force microscopy, respectively. Finally, we used solid-state NMR to visualize the interaction between the membrane and MreBGs at the atomic level. Our findings reveal that divalent cations are required for nucleotide binding and that, unlike eukaryotic actin, MreBGs has similar affinity for both ATP and ADP. We also show that monomeric MreBGs establishes peripheral contacts with the membrane likely through electrostatic interactions, while Mg⋅ATP-induced MreBGs filaments insert into the lipid bilayer without interfering with the membrane lamellar phase and have a significant local fluidifying effect. Significance Bacteria rely on the actin-like protein MreB to determine their cell shape, like actin does in eukaryotic cells. To perform its tightly regulated morphogenetic function, MreB forms membrane-associated nanofilaments, which control the cell wall biosynthetic machinery. We recently demonstrated that, in vitro, MreB from the Gram-positive bacterium Geobacillus stearothermophilus requires ATP and lipids to polymerize into pairs of filaments. Here, we show that in the presence of ATP, Geobacillus MreB forms membrane-bound filaments that directly impact local membrane fluidity, which could translate into a regulatory effect on cell wall synthetic enzymes. We reveal that MreB binds both ATP and ADP with similar affinity, unlike eukaryotic actin, and speculate that this could be a mechanism modulating the pool of polymerization-competent MreB. Introduction Filament-forming cytoskeletal proteins are a hallmark of cells across all domains of life, orchestrating essential processes such as cell shape determination, mechanical stability, cytokinesis, chromosome segregation, and intracellular transport. In most rod-shaped bacteria, the intracellular actin-like MreB proteins and the extracellular peptidoglycan cell wall are the major determinants of cell shape.1 In vivo, MreB forms membrane-associated polymeric assemblies that move around the rod circumference together with proteins of the peptidoglycan elongation machinery, forming the so-called Rod complex.2,3,4 MreB filaments orient this circumferential motion, thereby directing the insertion of new cell wall material in radial hoops and promoting cylindrical growth.2,3 In vitro, MreB polymerizes into straight pairs of protofilaments.5 Polymerization of MreB from the Gram-positive Geobacillus stearothermophilus (MreBGs), and possibly from other Gram-positive bacteria, requires the presence of lipids and ATP.6 Notably, ATP, but not ADP, promotes both MreBGs membrane binding and polymerization into the canonical pair of straight protofilaments.6 It was assumed that, like actin molecules, MreBGs binds both ATP and ADP, and it was hypothesized that only ATP binding or hydrolysis triggers conformational changes that modulate the affinity of MreBGs for the membrane.6 ATP-bound MreBGs membrane binding is mediated by two adjacent hydrophobic regions, the N terminus and a surface-exposed loop, which together form a putative membrane-insertion domain (Figure 1), and it is further facilitated by electrostatic interactions between charged residues surrounding these regions and anionic lipids.6 Direct membrane binding, via an amphipathic helix and/or a hydrophobic insertion loop, has also been shown for MreB filaments of several Gram-negative species.6,7,8,9 However, the insertion of the MreB molecule into the lipid layer has not yet been demonstrated, and the potential effects of MreB filaments on membrane structure or dynamics remain unexplored. In vivo data, using lipid staining and spectrometry approaches, suggested that MreB filaments induce the formation of regions of increased fluidity (RIFs) within the membrane of Bacillus subtilis.10 Whether this effect was direct or indirect could not be determined within the complex cellular environment, and the mechanism by which MreB was proposed to create fluid lipid domains remains elusive.Figure 1Crystal structure of MreB from G. stearothermophilus (PDB: 7ZPT), showing the adjacent hydrophobic N-terminal region (red) and the hydrophobic α2–β7 loop (blue), both involved in membrane binding and polymerization on lipid surfaces. Crystal structure of MreB from G. stearothermophilus (PDB: 7ZPT), showing the adjacent hydrophobic N-terminal region (red) and the hydrophobic α2–β7 loop (blue), both involved in membrane binding and polymerization on lipid surfaces. Here, we quantified the binding kinetics of ATP and ADP to MreBGs using fluorescence anisotropy, and we characterized the binding of ATP-MreBGs to a lipid bilayer using a combination of liposome-binding assays, atomic force microscopy (AFM), and wideline solid-state NMR (ssNMR). Notably, we found that Mg-MreBGs binds both ATP and ADP with similar affinities—unlike eukaryotic actins—and that ATP promotes the insertion of Mg-MreBGs into the lipid bilayer. This insertion significantly increases the fluidity of the hydrophobic core region of the membrane without interfering with the lamellar membrane phase. Furthermore, inorganic phosphate (Pi) was not detected in our ssNMR measurements, and signals of the α, β, and γ positions of the ATP were all detected in the presence ATP-MreBGs, suggesting that membrane-bound MreB filaments may predominantly be in either the ADP-Pi6 or the ATP state under these conditions. Together, our findings suggest that modulation of the intracellular ATP/ADP ratio could be a regulatory mechanism for MreB polymerization and support a model in which ATP-induced MreBGs filaments locally increase lipid-chain disorder, thereby modulating membrane fluidity at the sites of action of the Rod complex. Materials and methods Reagents and materials 1-palmitoyl-d31-2-oleoyl-sn-glycero-3-phosphocoline (POPC-d31), 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dioleoyl-sn-glycero-3-phospho-(1′-rac-glycerol) (DOPG), and E. coli polar extract were purchased from Avanti Polar Lipids (Alabaster, AL, USA). Chloroform and methanol used for liposome preparation were purchased from Sigma-Aldrich Chemicals (Saint-Quentin-Fallavier, France). Fluorescent nucleotides used in this study were purchased from Jena Bioscience (https://www.jenabioscience.com/) under references NU-805-488 (for N6-(6-Aminohexyl)-ATP-ATTO-488, hereafter ATP∗) and NU-247-488 (for N6-(6-Aminohexyl)-ADP-ATTO-488, hereafter ADP∗). Fluorescent nucleotide analogs were diluted in 20 mM HEPES (pH 7.5). Protein expression and purification His-tagged MreBGs was expressed in Escherichia coli and purified following a 2-step procedure with affinity purification (Ni-NTA agarose resin) and size exclusion chromatography. The protocol followed mainly the one previously described in Mao et al.6 Two specific modifications were introduced for the production of protein for ssNMR: proteins eluted from the Ni-NTA affinity column dropped in 1 volume of buffer B (20 mM Tris [pH 7], 500 mM KCl, 1 mM DTT, 2 mM EDTA), and size exclusion chromatography (HiLoad 16/60 Superdex 200 pg column from Cytiva) was pre-equilibrated and run with buffer C (20 mM Tris [pH 7]; 200 mM KCl). The protein purity was above 94%, as estimated by Coomassie staining followed by densitometry quantification. Fluorescence anisotropy Nucleotide exchange experiments were initiated by adding fluorescent nucleotides to MreBGs in buffer B in presence or in absence of 5 mM CaCl2 or MgCl2. The binding of fluorescent nucleotide analogs to MreBGs was monitored by measuring changes in anisotropy using excitation and emission wavelengths of 504 nm and 521 nm, respectively. Steady-state values were recorded 5 min after the start of the experiment. The experiments were performed using a Safas Xenius XC spectrofluorometer (Safas Monaco) controlled by SP2000 software, version 7.8.13.0. For binding affinity measurements, steady-state fluorescence anisotropy values were measured at constant ATP∗ or ADP∗ concentrations in the presence of an increasing concentration of MreBGs. Data were fitted and plotted using Igor Pro (version 9.0.5.1) to estimate Kd:r=rmax∗[MreBGs]Kd+[MreBGs]+rmin,where r represents the fluorescence anisotropy readout, rmax represents the maximum anisotropy value when MreBGs is saturated with the fluorescent nucleotide, [MreBGs] is the concentration of MreBGs, Kd is the dissociation constant, and rmin is the minimum anisotropy value corresponding to the fluorescent nucleotide alone in solution. Each data point represents the mean of at least three independent measurements, with error bars indicating the standard deviation. For kinetic experiments, the main reaction controlling the rate of the first nucleotide exchange (when MreBGs and ATP∗ or ADP∗ are mixed together) is the association of the fluorescent nucleotide with MreBGs (kon,ATP∗ or kon,ADP∗). The main reaction controlling the rate of the second nucleotide exchange (when an excess of ATP outcompetes ATP∗) is the dissociation of the fluorescent nucleotide from MreBGs (koff,ATP∗). The two reactions involved are described by the following equations:P+N=PNP+N∗=PN∗Here, P represents MreBGs, N the unlabeled nucleotide, N∗ the labeled nucleotide, and PN and PN∗ the protein-nucleotide complexes. These equations involve four rate constants: konN∗, konN, koffN, and koffN∗. The binding rate constants (kon) cannot be determined by this assay as they are too fast. However, the dissociation rates corresponding to the slowest reactions are accessible and can be mathematically described by the following differential equation:d[PNorPN∗]dt=−koff[PNorPN∗] This equation can be resolved by fitting an exponential decay function to estimate the koff value.r(t)=rmin+(r0−rmin)e−koff,N∗.t,where r(t) represents the fluorescence anisotropy at time t, rmin is the minimum anisotropy value for the fluorescent nucleotide alone, r0 is the initial fluorescence anisotropy before adding the excess unlabeled nucleotide, and koff,N∗ is the dissociation rate constant of the fluorescent nucleotide. Liposome pelleting assays Purified MreBGs was mixed at a final concentration (Cf) of 1.34 μM (0.05 mg/mL) in a sedimentation buffer (20 mM Tris-HCl [pH 7], 300 mM KCl, 5 mM MgCl2), in the presence or absence of ATP (Cf 2mM) and liposomes (Cf 0.5 mg/mL). Liposomes of E. coli polar extract from Avanti (Merck) were prepared freshly by rehydrating desiccated lipids with H2O under strong vortexing. Mixtures were incubated for 1 h at room temperature prior to ultracentrifugation at 100,000 × g for 1 h at 18°C. After careful recovery of the supernatants, the pellets were dispersed with an identical volume of sedimentation buffer supplemented with Tween20 at 0.1% final concentration. Aliquots of pellets and supernatants were loaded for analysis on a 12% SDS-PAGE, subsequently stained by Coomassie. Liposome preparation for AFM Liposomes were prepared by dehydrating a lipid chloroform mixture of 20 mol % DOPG and 80 mol % DOPC in a desiccator for >3 h at room temperature. Lipids were rehydrated with citrate buffer (20 mM citrate [pH 4.6] and 300 mM KCl). Liposomes were formed by vortexing and incubating at 40°C. Lipid mixtures were passed 21 times through a lipid extruder with a 100-nm pore size on a heated plate at 50°C (aimed internal temperature around 30°C) to produce small uniform-sized liposomes. A final concentration of 1 mM of liposomes was used. 3 mM of CaCl2 was added to the liposome suspension just prior to deposition. Nanofabrication of AFM cantilevers BL-AC10DS-A2 cantilevers were purchased from Olympus (Tokyo, Japan) and used as the scanning probe to visualize MreB polymerization. The spring constant (k) was 0.1 N/m, while the resonance frequency (f) was 0.6 MHz in water (1.5 MHz in air). Its length, width, and thickness were 9, 2, and 0.13 μm, respectively. Electron-beam deposition (EBD) was performed to fabricate a cantilever with a long and sharp tip, which is important to acquire high-resolution images. First, UV/O3 treatment was conducted to clean cantilevers, and then the cantilevers were soaked in piranha solution (containing sulfuric acid and hydrogen peroxide). After that, EBD was carried out on the cantilevers (30 kV accelerating voltage, 2-min irradiation) using a field emission scanning electron microscope (ELS-7500, Elionix, Tokyo, Japan). AFM nanoimaging AFM nanoimaging was performed using a laboratory-built microscope, as previously reported.11,12,13 A laser beam with a 670-nm wavelength was focused on an EBD-processed cantilever tip through a 20× objective lens (CFI S Plan Fluor ELWD, Nikon, Tokyo, Japan). Dynamic cantilever deflection was detected by sensing the position of laser beam reflected by the cantilever with a position-sensing two-segmented photodiode. An optimal tapping force was generated by adjusting the free oscillation amplitude of the cantilever (A0) to 1.5–2.5 nm and the set point to 80%–90% of the free amplitude. A glass stage glued with a stack of flat muscovite mica layer was attached to the HS-AFM scanner. Mica surface was treated with undiluted poly-L-Lysin for 10 min. The treated mica surface was washed with Milli-Q water and a liposome suspension of 20 mol % DOPG and 80 mol % DOPC (Avanti lipids) in 20 mM citrate buffer (pH 4.6) with 300 mM KCl, and 3 mM CaCl2 was added. After 25 min of incubation the supported lipid bilayer (SLB) was washed with 25 mM HEPES (pH 7.5). The experiment was performed in 25 mM HEPES (pH 7.5), 100 mM KCl, and 1 mM MgCl2. Purified His-MreBGs was centrifuged at 23,000 × g for 10 min at 4°C to remove any precipitates. MreB was added to the SLB with or without 0.2 mM of ATP. Several locations of 0.7–1.3 μm2 were imaged on the SLB after 15 min to ascertain the presence and architecture of the filaments. MreB polymerization in the presence of membrane vesicles Liposomes composed of POPC-d31, DOPC, and DOPG (35:45:20, w/w) were prepared by mixing lipid powders dissolved in chloroform/methanol (2:1 v/v). After solvent evaporation under a gentle air stream, the lipid film was rehydrated in 0.5 mL of Milli-Q water and lyophilized overnight. The resulting powder was resuspended in 100 μL of buffer A (20 mM Tris-HCl (pH 7.0), 200 mM NaCl, 1 mM DTT, and 2 mM EDTA) and homogenized by three cycles of vortexing, rapid freezing in liquid nitrogen (−196°C, 1 min), and thawing at 40°C for 10 min, yielding a milky multilamellar vesicle (MLV) suspension. MreB in buffer A (23 μM) was added to preformed liposomes at a protein-to-lipid mass ratio 1:2, and incubated for 30 min at 30°C. Next, when required, MgCl2 (cf 5 mM) and ATP (cf 2mM) were added, and the mixture was incubated for another 10 min at 30°C. The turbid proteoliposome suspensions (∼0.1–10 μm) were finally pelleted by ultracentrifugation (42,000 rpm, 1 h, 4°C; MLA 80 rotor, Beckman Coulter). Solid-state NMR 2H NMR experiments were performed on a Bruker Avance III spectrometer operating at 77 MHz with a 4-mm double-resonance HX-CPMAS probe. A phase-cycled quadrupolar echo sequence (90°x-t-90°y-t-acq) was used to perform the 2H NMR experiments.14 Acquisition parameters were as follows: spectral window of 500 kHz, π/2 pulse width of 3.50 μs, interpulse delays (t) of 50 μs, recycle delays of 2 s for 2H, and between 4,096 and 20,480 scans. Samples were equilibrated for 20 min at a given temperature before data acquisition. The temperature was regulated to ±1°C. Lorentzian line broadening was applied before Fourier transformation, starting from the top of the echo. To record undistorted 31P NMR spectra, we applied a Hahn spin echo sequence, 90°-t-180°-t-acq at the 31P frequency of 162 MHz on a 400 MHz (9.4 T) Bruker Avance III HD spectrometer,15 with a 90° pulse of 8 μs, an echo delay of 40 μs, a recycle delay of 5 s, a spectral window of 64 kHz, and a number of scans ranging from 1,024 to 4,096. All spectra were processed and analyzed using Bruker Topspin 4.0 software. Spectra were processed using a Lorentzian line broadening of 200 or 300 Hz for 31P NMR spectra. Computational analysis of 2H solid-state NMR spectra Spectral processing was carried out in TopSpin 4.3.0 (Bruker), where an exponential window of 200 Hz line broadening was applied before performing the Fourier transform. De-Pake-ing16,17 and the calculation of the first-order spectral moment (M1) were carried out using NMR Depaker (provided by Dr. Sébastien Buchoux)18; each M1 value was determined in triplicate, and the resulting standard deviations are shown as error bars in the plots. We assigned a default ±5% error to cover procedural uncertainties.19,20 To determine the SCD order parameters along the POPC-d31 acyl chain, we simulated wideline deuterium solid-state NMR spectra using a FORTRAN code developed by Erick Dufourc and accessed via a Microsoft .NET interface created by Arnaud Grélard. For each deuterated site along the deuterated lipid chain, we report the experimentally determined quadrupolar splitting (ΔνQ) and the count of deuterons at that specific acyl position based on the molecular structure of the lipid (Table S1). The simulations incorporated liposome deformation (reflected by a c/a parameter) to account for deviations from spherical bilayer normal distributions relative to the magnetic field.21 We adjusted the simulation to align with the experimental spectra through successive comparisons until optimal superposition was reached. From the best-fit simulation, we extracted the quadrupolar splitting for each labeled carbon and calculated the corresponding SCD order parameters in the bilayer. We assigned a default ±5% error to all SCD order parameters to cover both procedural uncertainties and intrinsic simulation variability; respecting the complexity of the system (three components), a safety margin chosen to exceed the experimental deviations reported in previous work.19,20 Similarly, an analogous approach was applied to the 31P NMR spectra, using chemical shielding anisotropy (Δσ) and intrinsic linewidth as fitting parameters. Results MreBGs nucleotide-binding and dissociation kinetics We first characterized the nucleotide-binding properties of MreBGs using fluorescent anisotropy, which is more sensitive and convenient for kinetic or rapid equilibrium studies than more traditional techniques such as isothermal titration calorimetry. A family of sensitive fluorescent nucleotides known as N6-(6-amino)hexyl-ATP derivatives was previously shown to bind to different types of eukaryotic actin, including muscle actin in mammals22 and the distant cytoplasmic actin of the parasite Leishmania23 while preserving nucleotide binding affinity and polymerization. We first determined whether these nucleotides also bind MreBGs. The addition of N6-(6-amino)hexyl-ATP-ATTO-488 (hereafter referred to as ATP∗) to MreBGs resulted in a rapid increase in fluorescence anisotropy (Figure 2A). No change was observed in the absence of Mg2+ or Ca2+, indicating that pre-binding of divalent cations to MreBGs is essential for nucleotide binding (Figure 2A), like in the case of actin.24 Purified MreBGs is presumably free of bound nucleotide in its active site since nucleotide binding was too fast to detect initial changes in anisotropy values with our instrument and thus could not be fitted accurately. However, we could measure plateau values that corresponded to steady-state binding (Figure 2B). The higher plateau values observed in the presence of Mg2+ indicate that ATP∗ has a stronger affinity for Mg-bound MreBGs than for Ca-bound MreBGs. Additionally, ATP∗ binding occurred more slowly in the presence of Ca2+, as steady state was reached at later time points than with Mg2+ (Figure 2A). Based on these results, all subsequent experiments were conducted in the presence of Mg2+. We first quantified the binding affinity of ATP∗ by titrating increasing concentrations of Mg-MreBGs against a fixed concentration of ATP∗ (0.2 μM) and measuring fluorescence anisotropy at steady state. Binding curves revealed sub-micromolar dissociation constants (Kd,ATP∗ = 18 ± 5 nM; Figure 2C). Identical measurements using the fluorescent equivalent ADP analog (N6-(6-amino)hexyl-ADP-ATTO-488, ADP∗) indicated that ADP binds MreBGs with similar—only slightly lower—affinity than ATP (Kd,ADP∗ = 27 ± 7 nM; Figure S1A; Table 1). The Kd for both nucleotides is in the nanomolar range, like the reported Kd of ATP-monomeric actin (globular actin, G-actin) under physiological Mg2+ conditions22,24,25 and indicating very high affinity. However, G-actin has a marked higher affinity for ATP than for ADP.26Figure 2Analysis of MreBGs-ATP binding and dissociation by fluorescence anisotropy. (A) Binding kinetics of N6-(6-amino)hexyl-ATP-ATTO-488 (0.2 μM, referred to as ATP∗) to MreBGs (1 μM), monitored by fluorescence anisotropy over time, in the presence (closed circles) or absence (open circles) of Ca2+ (red) or Mg2+ (blue) cations (5 mM). (B) Steady-state fluorescence anisotropy values of ATP∗ (0.2 μM) binding to MreBGs (1 μM) under the indicated ionic conditions. Bars represent mean ± standard deviation (n = 3 per condition). (C) Fluorescence anisotropy measurements of ATP∗ binding to MreBGs. Binding was assessed by titrating increasing concentrations of MreBGs against a fixed concentration of ATP∗ (0.2 μM) and measuring fluorescence anisotropy at steady state (5 min after mixing). Dissociation constants (Kd) were determined by fitting the binding curves. Each data point represents the mean ± standard deviation from at least three independent measurements. (D) Binding kinetics of ATP∗ (0.2 μM) to MreBGs (1 μM), followed in real time by fluorescence anisotropy. At 230 s, excess unlabeled ATP (1 mM) was added (dashed line), displacing bound ATP∗. Red line: fitting of the dissociation phase (II) yields the koff,ATP∗ value for ATP∗ release.Table 1Affinity and rate constants for nucleotide exchange on Mg-MreBGs measured in this studyChemical ReactionKd (μM)koff (s−1)Mg-MreBGs + ATP∗ = Mg-MreBGs-ATP∗0.018 ± 0.0050.0043 ± 0.0005Mg-MreBGs + ADP∗ = Mg-MreBGs-ADP∗0.027 ± 0.0070.0028 ± 0.0005 Analysis of MreBGs-ATP binding and dissociation by fluorescence anisotropy. (A) Binding kinetics of N6-(6-amino)hexyl-ATP-ATTO-488 (0.2 μM, referred to as ATP∗) to MreBGs (1 μM), monitored by fluorescence anisotropy over time, in the presence (closed circles) or absence (open circles) of Ca2+ (red) or Mg2+ (blue) cations (5 mM). (B) Steady-state fluorescence anisotropy values of ATP∗ (0.2 μM) binding to MreBGs (1 μM) under the indicated ionic conditions. Bars represent mean ± standard deviation (n = 3 per condition). (C) Fluorescence anisotropy measurements of ATP∗ binding to MreBGs. Binding was assessed by titrating increasing concentrations of MreBGs against a fixed concentration of ATP∗ (0.2 μM) and measuring fluorescence anisotropy at steady state (5 min after mixing). Dissociation constants (Kd) were determined by fitting the binding curves. Each data point represents the mean ± standard deviation from at least three independent measurements. (D) Binding kinetics of ATP∗ (0.2 μM) to MreBGs (1 μM), followed in real time by fluorescence anisotropy. At 230 s, excess unlabeled ATP (1 mM) was added (dashed line), displacing bound ATP∗. Red line: fitting of the dissociation phase (II) yields the koff,ATP∗ value for ATP∗ release. Table 1Affinity and rate constants for nucleotide exchange on Mg-MreBGs measured in this studyChemical ReactionKd (μM)koff (s−1)Mg-MreBGs + ATP∗ = Mg-MreBGs-ATP∗0.018 ± 0.0050.0043 ± 0.0005Mg-MreBGs + ADP∗ = Mg-MreBGs-ADP∗0.027 ± 0.0070.0028 ± 0.0005 Affinity and rate constants for nucleotide exchange on Mg-MreBGs measured in this study We next estimated the dissociation rates of both ATP∗ and ADP∗. We pre-incubated MreBGs with the labeled nucleotide until equilibrium was reached and triggered dissociation by addition of ∼200-fold excess unlabeled nucleotide (1 mM). The resulting decrease in fluorescence anisotropy was monitored over time, and the dissociation curves of ATP∗ and ADP∗ from the MreBGs-ATP∗ and MreBGs-ADP∗ complexes, respectively, were fitted using a mono-exponential decay model (koff,ATP∗ = 0.0043 ± 0.0005 s−1; koff,ADP∗ = 0.0028 ± 0.0005 s−1; Figures 2D and S1B, respectively; Table 1). Taken together, these findings indicate that in the presence of Mg2+, MreBGs binds both ATP and ADP with very high affinity, indicating that under physiological conditions (in vivo) and at the experimental ATP concentrations used here, MreBGs is predominantly nucleotide bound. Moreover, MreBGs exhibits only a slightly higher affinity for ATP∗ than for ADP∗, yielding an ATP/ADP affinity ratio substantially lower than that observed for eukaryotic actin. ATP promotes polymerization of MreBGs on supported lipid bilayers. Though both ATP and ADP bind similarly to MreBGs (Table 1), only ATP was previously found to efficiently promote the formation of pairs of MreBGs filaments on lipid monolayers, as visualized by transmission electron microscopy (TEM), and the absorption of MreBGs onto lipid bilayers in QCMD experiments.6 We first confirmed the ATP requirement for binding to a lipid bilayer using a liposome pelleting assay. In the absence of liposomes, MreBGs remained mainly soluble, regardless of the presence of ATP (Figure 3A). A fraction of MreBGs associated with liposomes in the absence of ATP, which may reflect the ion bridge-mediated intrinsic affinity of the protein for lipids. Consistent with this, increasing the concentration of salts (KCl), which reduces the contribution of electrostatic interactions to membrane binding by screening the charges of anionic lipid, abolished detectable binding of MreBGs to liposomes (Figure S2). In the presence of both liposomes and ATP, most MreBGs sedimented as expected (Figure 3A). We next turned to AFM to visualize MreBGs on an SLB in the presence and in the absence of ATP. We used SLBs made of DOPC doped with the anionic DOPG, which mimic Bacillus membranes and enable us to form SLBs on planar substrates.6 In the absence of added ATP, no MreBGs filaments were observed on the SLB (Figure 3B). Some puncta could be seen, which could be either monomeric or low oligomeric MreB species or small protein aggregates. In the presence of ATP, MreBGs rapidly formed a dense network of membrane-bound straight filaments (Figure 3B), similar to what has been observed by TEM on lipid monolayers.6 Taken together, these findings confirmed that ATP promotes binding and polymerization of MreBGs on biomimetic lipid bilayers.Figure 3ATP-dependent binding and polymerization of MreBGs on lipid bilayers. (A) Liposome pelleting assay. Representative 12% SDS-PAGE showing sedimentation of MreBGs (1.3 μM) in sedimentation buffer (20 mM Tris-HCl [pH 7], 300 mM KCl, 5 mM MgCl2) in the absence or in the presence of ATP (2 mM) or liposomes (0.5 mg/mL). S, supernatant; P, pellet. (B) AFM images of an SLB made with DOPC:DOPG 80:20 (molar ratio), previously shown to give a high absorption of MreBGs into SLBs,6 imaged before addition and 15 min after addition (min:sec) of MreBGs (1.2 μM) with or without 0.2 mM ATP. Scale bar, 200 nm. (C) Quantification of the presence of filaments on AFM fields. n indicates the number of areas evaluated. ATP-dependent binding and polymerization of MreBGs on lipid bilayers. (A) Liposome pelleting assay. Representative 12% SDS-PAGE showing sedimentation of MreBGs (1.3 μM) in sedimentation buffer (20 mM Tris-HCl [pH 7], 300 mM KCl, 5 mM MgCl2) in the absence or in the presence of ATP (2 mM) or liposomes (0.5 mg/mL). S, supernatant; P, pellet. (B) AFM images of an SLB made with DOPC:DOPG 80:20 (molar ratio), previously shown to give a high absorption of MreBGs into SLBs,6 imaged before addition and 15 min after addition (min:sec) of MreBGs (1.2 μM) with or without 0.2 mM ATP. Scale bar, 200 nm. (C) Quantification of the presence of filaments on AFM fields. n indicates the number of areas evaluated. ATP-dependent MreBGs interactions with the membrane detected by wideline ssNMR We next monitored the interactions of ATP-bound MreBGs with the membrane using wideline ssNMR. Biological membranes can adopt different physical states, including solid-ordered gel, liquid-ordered, and liquid-disordered fluid phases, which differ in lipid mobility and packing. Wideline ssNMR is a precise, nondestructive method to probe lipid membrane mesophase, mostly a lamellar phase in the plasma membrane, membrane order, phase transitions, and interactions under near-physiological conditions, typically detected on deuterium (2H) or phosphorus (31P) nuclei.27,28 Membrane properties are analyzed by measuring the residual quadrupolar splittings of deuterated lipid acyl chains (ΔνQ) or the chemical shift anisotropy of phosphorus headgroup.27,29,30 Because these anisotropic interactions are averaged out by motion, changes in lipid mobility translate into traceable variations in the observed 2H and 31P signals. Membrane MLVs are preformed and tuned with respect to the lipid composition, allowing to reproduce a native membrane-mimicking environment. Subsequently, wideline 2H and 31P NMR report on local changes in mobility and interactions with molecular partners in the hydrophobic membrane core region and at the phosphorus headgroup, respectively. We doped vesicles of DOPC:DOPG 80:20 with 2H-labeled POPC (palmitoyl-oleyl-phophatidyl-choline-d31) to produce POPC-d31:DOPC:DOPG (35:45:20 w:w) liposomes. To monitor the exclusive membrane interactions of polymerizing MreBGs in comparison with soluble MreBGs, we analyzed four analogously prepared samples of MLVs incubated with (1) buffer only, (2) buffer and MreB, (3) buffer and ATP, and (4) buffer with both MreB and ATP. We first recorded the quadrupolar splittings ΔνQ (kHz) from the acyl chains of POPC-d31 at 303 K and 263 K, which report on the order and mobility of different segments of the lipid tails (Figure 4B). The largest splittings typically correspond to the plateau region of the first seven C-2H positions along the acyl chain, near the glycerol headgroup, which is sterically constrained and relatively rigid. The smaller splittings originate from the more flexible regions in the center of the membrane and from the terminal methyl group, which is the most disordered and gives rise to the narrowest doublet in the ssNMR spectrum. At 303 K, the ssNMR patterns indicate a fluid, liquid-disordered membrane, while at 263 K, closer to the fluid-to-gel phase transition temperature of the DOPC and DOPG acyl chains (Tm ≈ −16°C), the lipid acyl chains become more ordered and rigid. In addition, a small central signal corresponds to 2HOH water molecules, detected at approximately 3 ppm.31 Differences between the samples concerning the isotropic signal intensity could thus arise from distinct final hydration levels despite identical sample preparation. Together, these measurements allow us to characterize how lipid mobility varies along the acyl chain and how it changes with the temperature or in contact with a molecular partner.Figure 4Effect of MreB on membrane dynamics. (A) Chemical structures of 2H31-POPC-d31, DOPC, and DOPG, used in a 35:45:20 mass ratio to reconstitute multilamellar vesicles. In 2H31-POPC-d31, the palmitoyl chain is perdeuterated, with 14 methylene CD2 groups and one methyl CD3 group for 2H solid-state NMR. (B) Comparison of 2H quadrupolar echo spectra of POPC-d31:DOPC:DOPG MLVs at 303 K and 263 K, recorded under buffer alone, ATP, MreB, and MreB+ATP conditions. Vertical dashed lines mark the regions corresponding to the largest quadrupolar splitting in the control condition. (C) First spectral moment (M1) plotted as a function of temperature for multilamellar vesicles with buffer alone, ATP, MreB, or both MreB and ATP, at a protein-to-lipid mass ratio of 1:2. Error bars represent ±5% variation in M1 values due to methodological and measurement uncertainties. Effect of MreB on membrane dynamics. (A) Chemical structures of 2H31-POPC-d31, DOPC, and DOPG, used in a 35:45:20 mass ratio to reconstitute multilamellar vesicles. In 2H31-POPC-d31, the palmitoyl chain is perdeuterated, with 14 methylene CD2 groups and one methyl CD3 group for 2H solid-state NMR. (B) Comparison of 2H quadrupolar echo spectra of POPC-d31:DOPC:DOPG MLVs at 303 K and 263 K, recorded under buffer alone, ATP, MreB, and MreB+ATP conditions. Vertical dashed lines mark the regions corresponding to the largest quadrupolar splitting in the control condition. (C) First spectral moment (M1) plotted as a function of temperature for multilamellar vesicles with buffer alone, ATP, MreB, or both MreB and ATP, at a protein-to-lipid mass ratio of 1:2. Error bars represent ±5% variation in M1 values due to methodological and measurement uncertainties. The visual aspect of the signals recorded at 303 K was very similar for the conditions of phospholipid vesicles alone and in the presence of either ATP or MreB only, whereas the 2H spectral fingerprint was altered in presence of both MreB and ATP (highlighted by the dotted lines in Figure 4B for eye guidance). We lowered the experimental temperature to 263 K, near the membrane phase transition, to examine the impact of MreB on the membrane in this altered membrane state. When comparing the spectra at 263 K, the three conditions of vesicles alone and with either ATP or MreB gave again similar spectra, in contrast to the spectrum of vesicles incubated with both MreB and ATP. To evaluate the impact of the interactions of polymerizing MreB with the membrane, we next calculated the first spectral moment M1 of the deuterated membrane lipids. This value represents an average of the order parameters along the lipid acyl chain and can be interpreted as variations in the mobility of the membrane phase.18,32,33 Higher M1 values correspond to more ordered, less mobile lipid chains, whereas lower values indicate increased chain flexibility and membrane fluidity. A significant reduction in the global order of the membrane hydrophobic core was observed in the presence of both MreB and ATP, suggesting clear interactions between MreB filaments and the membrane as well as a local fluidifying effect (Figure 4C). The membrane disordering effect was maintained when shifting the membrane to lower temperatures approaching the membrane fluid-to-gel phase transition (Figure 4C), potentially due to a continuous local membrane destabilizing effect of MreB filaments, favoring the fluid liquid-disordered membrane phase. MreBGs filaments promote fluidization of the lipid bilayer core We next calculated the carbon-deuterium order parameters (SCD) at each carbon position of the lipid acyl chains using the experimentally measured ΔνQ. The SCD value describes how much the residual quadrupolar couplings are averaged by the mobility of the C-D bond at each position of the acyl chain with respect to the bilayer normal.27,33 In other words, it provides a direct measure of how restricted in motion each segment of the lipid chain is on the sub-microsecond timescale. To obtain the SCD, the quadrupolar splitting measured at each carbon position k (ΔνQk) is converted into an order parameter using the relation ΔνQk = 0,75∗AQ SCDk, where AQ is the static quadrupolar coupling constant (167 kHz).33,34,35The order parameter of the acyl chain typically varies between disordered (2∗|SCD| = 0) to fully rigid systems (2∗|SCD| = 1). To interpret our experimental data, we simulated the expected 2H NMR spectra using established algorithms. These simulations take into account both the “de-Paked” oriented-like spectra16,18,32 as well as the MLV ellipsoidal deformation in the magnetic field represented by the c/a value (where c represents the axis of the ellipsoid parallel to the magnetic field). The spectra showed no significant ellipsoidal deformation of the vesicles in the magnetic field under any of the four tested conditions (Figure 5A). We also observed the common SCD plateau region,20,33 ranging from carbon position 2–7, for which SCD values remain rather constant (Figure 5B; Table S1), reflecting the sterically constrained, more rigid carbon-deuterium bonds near the phospholipid glycerol backbone. Neither ATP nor MreB alone affected the order parameter SCD of the acyl chain relative to the buffer-only control. However, MreB in the presence of ATP induced a clear reduction of the SCD considering the carbon positions 8 to 16, corresponding to the part of the acyl chains located in the inner, hydrophobic region of the lipid bilayer (Figure 5B; Table S1). This destabilizing effect, reflected by increased mobility of the membrane core acyl chains, likely arises from peripheral protein segments inserting into the lipid headgroup region, perturbing lipid packing and generating additional space for acyl chain fluctuations. Similar effects have been observed for membrane-binding peptides.36,37Figure 5Impact of MreB on membrane order. (A) Spectral simulations (black dashed lines) overlaid with the experimental 2H quadrupolar echo spectra acquired at 303 K under the four experimental conditions: buffer alone, ATP, MreB, and MreB + ATP. (B) Carbon-deuterium order parameters 2∗|SCD| of POPC-d31 in POPC-d31:DOPC:DOPG multilamellar vesicles (35:45:20) at 303 K, measured in the four experimental conditions. Error bars indicate a ±5% uncertainty in SCD values, reflecting both methodological and simulation-related variability. Impact of MreB on membrane order. (A) Spectral simulations (black dashed lines) overlaid with the experimental 2H quadrupolar echo spectra acquired at 303 K under the four experimental conditions: buffer alone, ATP, MreB, and MreB + ATP. (B) Carbon-deuterium order parameters 2∗|SCD| of POPC-d31 in POPC-d31:DOPC:DOPG multilamellar vesicles (35:45:20) at 303 K, measured in the four experimental conditions. Error bars indicate a ±5% uncertainty in SCD values, reflecting both methodological and simulation-related variability. ssNMR shows association of ATP-MreB to the lamellar membrane phase, with no detectable Pi release We next sought to characterize the nucleotide-bound state of ATP-induced MreBGs filaments on the preformed lamellar membrane. Previous results using nonhydrolysable analogs of ATP suggested that ATP hydrolysis was required for MreBGs membrane binding and polymerization.6 In agreement with this, Pi release by ATP-bound MreBGs was favored in the presence of lipids.6 We turned to 31P-detected wideline ssNMR under static conditions and analyzed the nucleotide phosphorus shifts and the residual chemical shift anisotropy (CSA) of phosphorus in the phospholipid headgroups. The CSA reflects the orientations and dynamics of the phosphate groups, providing information about the bilayer phase and its structural organization. Across the four experimental conditions tested, the 31P NMR line shapes were consistent with a lamellar phase20,29 (Figure 6). In a lamellar membrane phase, lipid molecules arrange in flat, parallel sheets (lamellae) that are typical biological membrane arrangements38 and at the origin of a characteristic phosphorus spectral envelope.39 The CSA was measured from the width of the 31P signals and reflects the range of orientations of the phosphate groups relative to the magnetic field. From our spectra, we obtained a CSA of approximately 40 ppm (6.4 kHz), in agreement with previous reports20,29 (Figures 6 and S3). The contributions of the individual 31P CSA components of the different lipids cannot be distinguished, and an overall pattern is detected, reflecting a very similar CSA value (Figure S4). Addition of MreBGs and/or ATP alone introduced minor contributions of an isotropic peak (2%) around 0 ppm (Figure 6), reflecting the orientation averaging of the phospholipid headgroup, as in occurs in rapid tumbling lipids, micelles, or small vesicles, potentially generated through electrostatic interactions of MreB or ATP with the lipids. Upon the addition of either ATP alone or MreB + ATP, the interpretation was complexified due to the superposition of 31P signals originating from free ATP and bound ATP (or a post-hydrolytic ADP-Pi species). 31P chemical shifts were similar to previously reported values.40,41 Notably, the 31P signals of ATP corresponding to the α, β, and γ positions were all detected but shifted by 1 ppm, 0.9 ppm, and 0.8 ppm, respectively, in the presence of MreB + ATP (Figures 6 and S4; Table S2), and the CSA pattern still reflected a lamellar membrane phase. No isotropic signal expected for free, rapidly tumbling 31P (typically observed around 3.7 ppm in water42) was observed in the MreB + ATP sample (Figure 6), suggesting that no continuous ATP hydrolysis occurred and thus no detectable Pi was released from ATP-bound filaments under these conditions.Figure 6Evaluation of MreBGs-induced changes in phosphorus properties of lipid headgroups and ATP. 31P NMR spectra of POPC-d31:DOPC:DOPG multilamellar vesicles (35:45:20) recorded at 303 K in the presence of buffer alone (blue), supplemented with ATP alone (cyan), MreBGs alone (green), or with both MreBGs and ATP (orange). Signals resulting from the three phosphate groups of ATP are indicated by the corresponding Greek letters. Magnified regions (boxed) are shown to discriminate the γ and α phosphate signals of ATP. In spectra recorded in the presence of both MreBGs and ATP (orange), the α-phosphate and the γ-phosphate positions are indicated with an asterisk, as the resonances may derive from the α and γ in ATP or in trapped ADP+Pi due to ATP hydrolysis.6 Evaluation of MreBGs-induced changes in phosphorus properties of lipid headgroups and ATP. 31P NMR spectra of POPC-d31:DOPC:DOPG multilamellar vesicles (35:45:20) recorded at 303 K in the presence of buffer alone (blue), supplemented with ATP alone (cyan), MreBGs alone (green), or with both MreBGs and ATP (orange). Signals resulting from the three phosphate groups of ATP are indicated by the corresponding Greek letters. Magnified regions (boxed) are shown to discriminate the γ and α phosphate signals of ATP. In spectra recorded in the presence of both MreBGs and ATP (orange), the α-phosphate and the γ-phosphate positions are indicated with an asterisk, as the resonances may derive from the α and γ in ATP or in trapped ADP+Pi due to ATP hydrolysis.6 Discussion In this study, we used three highly sensitive techniques—fluorescence anisotropy, AFM, and ssNMR—to characterize the binding of ATP to MreBGs and the effect of MreB binding to a membrane surface with unprecedented resolution. Our results reveal a high-affinity interaction between ATP and Mg-MreBGs, which is essential for filament formation on lipid bilayers and results in disruption of lipid packing and local fluidization of the membrane. While ssNMR analysis of 2H quadrupolar splitting is standardly used to probe membrane binding of peptides or proteins as well as changes in membrane phase and order,20,21,36,37,43,44 to our knowledge, this is the first report demonstrating its ability to detect nucleotide-induced membrane-binding and lipid-disordering by a polymerizing protein, highlighting its exceptional sensitivity to detect changes in the chemical environment and its applicability. In the absence of ATP, a small fraction of MreBGs co-pellets with liposomes in our liposome-binding assays. Increasing the salt concentration abolishes this association, likely by screening membrane negative charges, indicating that the interaction is electrostatic. Consistent with this, a very small fraction of lipids could be extracted from the lipid bilayer, potentially through electrostatic contacts between monomeric MreBGs and the membrane surface, resulting in an isotropic signal close to the noise level in the 31P NMR spectra. However, a small isotropic signal contribution is also observed in the presence of only ATP or ATP + MreB. In both cases, the signal could reflect newly formed rapidly tumbling lipids as in micelles or small vesicles. However, neither the membrane order nor the lamellar phase or phosphorus headgroup environment is perturbed, indicating the absence of significant protein-membrane interactions. This is further supported by the close similarity of the 2H-detected ssNMR spectra obtained with vesicles and with vesicles incubated with MreBGs alone. Taken together, these data suggest that monomeric MreBGs has peripheral contact with the membrane through electrostatic interactions, but it does not insert into the hydrophobic core of the bilayer. In the presence of ATP, the high affinity and rapid binding kinetics of Mg-MreBGs for the nucleotide ensure that, as observed for actin, MreBGs rapidly becomes fully ATP-bound and polymerization-competent. ATP-bound MreBGs assembled into filaments on SLBs as shown by AFM, and it induced clear changes in the 2H-detected ssNMR spectra, revealing membrane interactions that likely result from the insertion of peripheral protein segments into the lipid bilayer. Peptides and proteins can exhibit a varying effect on membrane order and thereby lipid chain mobility.20,21,36,37,43,45 Our data indicate that the disordering effect of ATP-bound MreBGs on acyl chain dynamics is moderate but significant. The membrane-insertion region likely corresponds to the hydrophobic spatially close N terminus and α2-β7 loop of MreBGs (Figure 1), which have been shown to be required for lipid interaction.6 We previously proposed that the transition from soluble MreBGs to membrane-bound polymers could be driven by a structural change triggered by ATP hydrolysis, which would render this region prone for insertion in the membrane.6 We did not detect an isotropic free phosphate signal in our 31P NMR spectra of MreBGs in the presence of ATP, which would be expected if ATP hydrolysis was required for membrane binding and free Pi was continuously released from the membrane-bound filaments, as previously reported.6 Although it cannot be formally excluded, it is unlikely that a large part of MreB-bound ATP was hydrolyzed to ADP and all Pi molecules were lost in the ssNMR ultracentrifugation step. The lack of detectable free Pi in our ssNMR experiments, while AFM showed a dense network of membrane-bound filaments in similar biochemical conditions, rather suggests absence of significant ATP hydrolysis under NMR conditions, or else trapping of the hydrolyzed Pi in the nucleotide-binding cleft during the course of the ssNMR experiment. Binding of either ATP or ADP-Pi-to the nucleotide-binding pocket would be consistent with the 31P chemical shift changes observed for all three ATP phosphate groups relative to unbound ATP, indicating a different chemical environment. The ADP-Pi-bound intermediate was previously suggested to be the long-lived nucleotide-bound form within MreB filaments.6 It cannot be ruled out, however, that ATP was the form detected in our 31P NMR spectra, and thus that ATP hydrolysis is not strictly required for membrane binding and polymerization, but occurs later, within membrane-bound MreB filaments, as is the case for actin filaments. Under this alternative model, ATP binding itself, rather than ATP hydrolysis, would trigger the structural change involved in the switch from the soluble to the membrane-affine form. Interestingly, while the crystal structure of the ATP-bound form and the nucleotide-free (apo) form of MreBGs (PDB: 8AZG and 7ZPT, respectively) are highly similar,6 ATP binding induces changes in the two membrane-anchoring domains (Figure S5). The end of the N-terminal region is predicted to be intrinsically disordered and was not resolved in the crystal structures (aa 1–5 and 1–3 missing, respectively). However, both the α2-β7 loop and the segments of the N-terminal extremity present in the crystals have a high B-factor (also known as the atomic displacement parameter) in the ATP-bound form of MreBGs (Figure S5), which reflects a higher disorder and likely a higher flexibility,46 favoring structural adaptations such as membrane insertion. While MreBGs does not polymerize on a lipid surface in the presence of ADP,6 our anisotropy data show that soluble MreBGs binds both ATP and ADP with comparable affinities under physiological Mg2+ conditions, with affinity for ATP being less than 2-fold higher than for ADP. In contrast, G-actin has a markedly higher affinity for ATP versus ADP, about 4-fold under physiological Mg2+ conditions (and 200-fold in the presence of Ca2+).25,26 Both eukaryotic and bacterial cells have similar (high) millimolar intracellular ATP concentrations (average in the 1–5 mM range) and maintain a similar (high) ATP/ADP average ratio of ∼5–10.47,48,49 Because of actin’s substantially greater affinity for ATP compared with ADP, coupled with the prevalence of G-actin bound to profilin, an essential actin binding protein (ABP) that catalyzes ADP-to-ATP exchange,24 the high majority of monomeric actin in eukaryotic cells is predicted to exist in the ATP-bound, polymerization-competent state. In bacterial cells, which lack known counterparts of ABPs, the similar affinities between Mg-MreBGs and both ATP and ADP found here suggest that the fraction of ATP-bound MreBGs is substantially lower than the fraction of ATP-bound actin, with a nonnegligible fraction of monomeric MreBGs bound to ADP. Although the intracellular ATP/ADP ratio ([ATP] = 2–2.6 mM and [ADP] = 0.4–0.8 mM measured in Escherichia coli49,50) would imply that most MreBGs monomers are ATP bound, this ratio is known to vary depending on growth conditions such as nutrient availability, growth rate, or environmental stress. Thus, the similar affinities of MreBGs for ATP and ADP may provide a mechanism for modulating the pool of polymerization-competent MreB in bacteria. Finally, our data show that MreBGs in polymerizing conditions impacts the membrane stability, locally increasing the mobility of the acyl chain in the membrane inner core. In line with this, recent molecular dynamics simulations suggest that E. coli MreB filaments induce local membrane bending by lipid displacement and physical distortion via the N-terminal amphipathic helix of E. coli MreB, which mediates membrane binding.51 Both shallow insertion of membrane-anchoring segments and membrane bending are mechanisms that can increase lipid chain mobility and thus fluidify membranes. Our finding that MreBGs filaments fluidify the membrane along membrane-bound MreB filaments is in agreement with previously reported in vivo data that suggested that MreB filaments generate fluid lipid domains (RIFs) in the B. subtilis plasma membrane.10 Thus, in Gram-positive bacteria at least, membrane-associated MreB filaments may directly contribute to local fluidization of the phospholipid membrane. Interestingly, AFM imaging demonstrated that the membrane-bound cell wall building block Lipid II co-separates in the liquid-crystalline (also known as liquid-disordered) more fluid regions of SLBs.52 In agreement with this, Lipid II as well as the Lipid II synthase MurG were reported to preferentially colocalize with RIFs in B. subtilis.53,54 Although the mechanisms behind the fluidifying effect measured in vitro might differ from those at the origin of the effect observed in vivo, it is tempting to speculate that increased membrane fluidity along membrane-bound MreB filaments might locally increase the concentration of Lipid II and/or modulate the activity of integral membrane proteins of the Rod complex,55 by increasing the rates of conformational changes or protein-protein or protein-substrate interactions. Future studies will be required to explore whether local, MreB-dependent, membrane fluidization influences the activity of the cell wall elongation machinery. In-cell solid-state NMR studies on bacteria and utilizing dynamic nuclear polarization might also be helpful to understand MreB activity in situ.56,57 Data availability Research data will be deposited on Zenodo after publication or be available from the corresponding authors upon request. Acknowledgments We thank Lukas Kalvoda and Martin Lenz, along with members of the ProCeD lab, for helpful discussions. This project has received funding from the 10.13039/501100000781European Research Council (ERC) under the Horizon2020 research and innovation program (10.13039/100010663ERC CoG No 772178 to R.C.-L.), the 10.13039/501100001665French National Research Agency (ANR-23-CE11-0005-01 to B.H.), a 10.13039/100018502Kanazawa University WPI-NanoLSI Bio-SPM collaborative research grant, the 10.13039/501100001321National Research Foundation (NRF) Singapore under its Mid-Sized Grant (10.13039/100005773MSG) (NRF-MSG-2023-0001 to A.M.), the 10.13039/501100001352National University of Singapore through the Mechanobiology Institute (A-0003467-00-00 to A.M.), and the Bordeaux University Health sciences and technologies department. C.D. was funded by a CENTURI postdoctoral fellowship. This study has also received financial support from the French government in the framework of the France 2030 program IdEx université de Bordeaux and has benefited from the facilities and expertise of the Biophysical and Structural Chemistry platform at IECB, 10.13039/501100004794CNRS UAR3033, 10.13039/501100001677Inserm US001, Univ. Bordeaux. Author contributions E.A., A.A.-M., C.D., E.M., K.S.L., and A.C. performed the research. I.E.A., A.A.-M., C.D., E.M., E.J.D., B.H., and A.C. analyzed data. R.W., A.M., R.C.-L., and B.H. provided infrastructure and scientific supervision. R.C.-L. and B.H. designed the research, funded the research, and wrote and revised the manuscript, with input from all authors. Declaration of interests The authors declare no competing interests.