Nonuniversal impact of cholesterol on membranes mobility, curvature sensing and elasticity

Abstract

Biological membranes, composed mainly of phospholipids and cholesterol, play a vital role as cellular barriers. They undergo localized reshaping in response to environmental cues and protein interactions, with the energetics of deformations crucial for exerting biological functions. This study investigates the non-universal role of cholesterol on the structure and elasticity of saturated and unsaturated lipid membranes. Our study uncovers a highly cooperative relationship between thermal membrane bending and local cholesterol redistribution, with cholesterol showing a strong preference for the compressed membrane leaflet. Remarkably, in unsaturated membranes, increased cholesterol mobility enhances cooperativity, resulting in membrane softening despite membrane thickening and lipid compression caused by cholesterol. These findings elucidate the intricate interplay between thermodynamic forces and local molecular interactions that govern collective properties of membranes.

Fig. 1. Lipid bilayer characteristics in the presence and absence of cholesterol.

Panel (a) presents snapshots of all-atom (AA) DOPC bicelles, without and with cholesterol molecules. DOPC within the central bicelle domain are colored in yellow (brown in bicelle rim domain), cholesterol in green, and DTPC molecules in blue. Rendering of the images was done with Blender v3.4.1 and Molecular Nodes v2.4.1. The study analyzes various features of lipid bilayers, including acyl chain order (b), area per lipid (c), and membrane thickness (d), using different setups (CG bicelles as circles and infinite, periodic bilayers as squares) and resolutions (open symbols for CG resolution and filled symbols for all-atom resolution). Error bars were smaller than the symbol size and therefore excluded.

Fig. 2. Coupling between membrane curvature and cholesterol distribution.

Mean curvature values within circular domains of radius 3 nm were analyzed in absence (black) and presence of cholesterol (blue), separately for DPPC (at 330 K), and DOPC bilayers (320 K) at all-atom resolution. Panels a, c show the time development and panels b, d histograms for the mean curvature $\overline{H}$ with a fit assuming a Gaussian distribution. Panel e shows the low-pass filtered mean curvature values (blue) together with the asymmetric distribution of cholesterol between the two leaflets within the analyzed central circular domain (red; difference in the number of cholesterol molecules between the upper and lower leaflets, normalized to the average number within one leaflet). Panel (f) displays the time-averaged mean curvature as a function of cholesterol asymmetry for both the DPPC and DOPC bilayers.

Fig. 3. Effect of cholesterol on membrane bending modulus.

Membrane bending moduli were determined from CG MD simulations (a) and from all-atom MD simulations (b) for different system setups (bicelles as circles and infinite, periodic bilayers as squares). The bending moduli were calculated using undulation analysis (for infinite systems) and the local fluctuation method (for bicelles). Panel c shows exemplarily the spectrum for DOPC bilayer systems with 0% and 40% cholesterol for different lateral box sizes (27 nm, 55 nm, and 110 nm; $⟨\mid z_{\mathbf{q}}{\mid }^2⟩$ are the amplitudes of the Fourier-transformed membrane surface). Displayed are mean values and errors as 95% confidence intervals employing parametric bootstrapping (N = 50,000 statistically independent samples) assuming Gaussian distributions of the mode-dependent amplitudes (infinite bilayers) or standard errors of the mean employing block averaging (at least N = 12 independent blocks; bicelles).

Fig. 4. Curvature-dependent cholesterol dynamics and distribution.

The sketch depicts the lateral positions of cholesterol molecules within the upper and lower leaflets (upper and lower panels, respectively) of a DOPC bicelle at atomistic resolution. Snapshots at distinct times, denoted as positive curvature (a, at time t₁) and negative curvature (b, at time t₂ > t₁), showcase the dynamic redistribution of cholesterol. The central domain, highlighted in green with a radius of 7 nm, serves as a reference. Cholesterol molecules departing from the central domain of the respective leaflet between these two snapshots are depicted in red, while molecules moving into the central domain of the upper/lower leaflet are represented in orange. The inset figures were created with BioRender.com.

Fig. 5. Effect of cholesterol dynamics on membrane bending modulus.

The bending modulus was analyzed for DPPC (a) and DOPC bilayers (b) based on unperturbed coarse-grained MD simulations in the absence of cholesterol (gray bars), in presence of 40% cholesterol, with blocked cholesterol flipping between the layers (noflip), and for simulations with blocked cholesterol flipping and restrained lateral cholesterol diffusion (restr.). Displayed are mean values and errors as 95% confidence intervals employing parametric bootstrapping (N = 50,000 statistically independent samples) assuming Gaussian distributions of the mode-dependent amplitudes.