Directly imaging emergence of phase separation in peroxidized lipid membranes

Abstract

Lipid peroxidation is a process which is key in cell signaling and disease, it is exploited in cancer therapy in the form of photodynamic therapy. The appearance of hydrophilic moieties within the bilayer's hydrocarbon core will dramatically alter the structure and mechanical behavior of membranes. Here, we combine viscosity sensitive fluorophores, advanced microscopy, and X-ray diffraction and molecular simulations to directly and quantitatively measure the bilayer's structural and viscoelastic properties, and correlate these with atomistic molecular modelling. Our results indicate an increase in microviscosity and a decrease in the bending rigidity upon peroxidation of the membranes, contrary to the trend observed with non-oxidized lipids. Fluorescence lifetime imaging microscopy and MD simulations give evidence for the presence of membrane regions of different local order in the oxidized membranes. We hypothesize that oxidation promotes stronger lipid-lipid interactions, which lead to an increase in the lateral heterogeneity within the bilayer and the creation of lipid clusters of higher order.

Fig. 1. Schematic of the lipid and dye structures used in this work.

Peroxidation of POPC results in the addition of a peroxide group (-OOH) which locates at either 9’ or 10’ position (see ESI). Membrane order is measured using the dyes Laurdan, BC10 and BC6 + + .

Fig. 2. Spectroscopic characterization of POPC/POPC-OOH large unilamellar vesicles (LUVs).

a Time resolved decay traces and (b) calculated microviscosity of BC10 labelled LUVs. c Emission spectra and (d) calculated GP of Laurdan-labelled LUVs. Instrument response function (IRF) is shown in black (a). Data shown as mean ± S.D. (n ≥ 3 independent repeats).

Fig. 3. Type II photo-oxidation of POPC GUVs, monitored by FLIM of BC10.

a–c FLIM images recorded at 0 s, 150 s and 450 s. Only the bottom left section was irradiated. Scalebar: 20 µm. d Changes in microviscosity recorded for irradiated (red) and non-irradiated (blue) GUVs. Data shown as mean ± S.D (N ≥ 3 GUVs in n = 3 independent repeats). e Lifetime distribution histogram and a zoomed in FLIM image of the irradiated GUV indicated by the white arrow in (c). Discrete color-coding of the inset FLIM image has been used to highlight areas of different microviscosity.

Fig. 4. X-Ray scattering intensity profiles for POPC/POPC-OOH bilayers.

a, b SAXS profile and interlamellar spacing calculated from the 1st Bragg peak. c, d WAXS traces and width of the fitted Lorentzian component. Data shown as mean ± S.D. S.D. in POPC-OOH containing membranes was estimated according to the S.D. of pure POPC samples, see ESI for details.

Fig. 5. FLIM imaging of POPC/POPC-OOH GUVs.

a–c Sample FLIM images of BC6 + + stained GUVs with increasing fractions of POPC-OOH. Scale bar: 20 µm. d Calculated membrane microviscosity using the reported rotor calibration. Box plots display the 25–75% range, error bars represent ± S.D., median is shown by a horizontal line and mean by a dot of N ≥ 30 GUVs from n = 3 independent repeats.

Fig. 6. MD simulations of POPC-OOH containing membranes.

a–c Lateral snapshots of the simulated bilayers containing increasing amounts of POPC-OOH. Colour coding: Green-POPC, Gray-oxPOPC9, Purple-oxPOPC10 (see Fig. S15 for details), Yellow-Phosphorous, Red-Oxygen. BC10 is also shown (cyan), within the leaflet regardless of the membrane’s composition, but closer to the phospholipid headgroups in the 100% POPC-OOH membrane. d, f Electron density profiles of membranes containing increasing amounts of POPC-OOH. In (e) the red traces correspond to all lipids (continuous trace), POPC (dotted) and POPC-OOH (dashed). g Area-thickness relationship for lipid peroxide containing bilayers. Data shown as mean ± S.D. from two independent simulations. h Relative variation of the 2-dimensional diffusion coefficient obtained from the MD simulations (⎕) and from molecular rotor data using the Saffman-Delbrück equation (∎). i, j Snapshots of a coarse-grained simulation showing the area per lipid (APL) maps of pure POPC (i) and POPC-OOH (j) membranes. Scalebar: 30 nm. k Bimodal APL distribution seen for the oxidized membranes (j), supporting the observation of lipid segregation in POPC-OOH membranes.

Fig. 7. Effect of lipid peroxidation on membrane architecture.

Snorkelling of the lipid tails leads to an increased lipid packing towards the interface, yet this causes the increase of the individual area per lipid. Concurrently, chain snorkelling creates a “void” at the membrane’s midplane, which lead to a lower membrane thickness and elastic properties.