Biophysical studies of lipid nanodomains using different physical characterization techniques

Abstract

For the past 50 years, evidence for the existence of functional lipid domains has been steadily accumulating. Although the notion of functional lipid domains, also known as "lipid rafts," is now widely accepted, this was not always the case. This ambiguity surrounding lipid domains could be partly attributed to the fact that they are highly dynamic, nanoscopic structures. Since most commonly used techniques are sensitive to microscale structural features, it is therefore, not surprising that it took some time to reach a consensus regarding their existence. In this review article, we will discuss studies that have used techniques that are inherently sensitive to nanoscopic structural features (i.e., neutron scatting, nuclear magnetic resonance, and Förster resonance energy transfer). We will also mention techniques that may be of use in the future (i.e., cryoelectron microscopy, droplet interface bilayers, inelastic x-ray scattering, and neutron reflectometry), which can further our understanding of the different and unique physicochemical properties of nanoscopic lipid domains.

Figure 1

Cartoon of membrane evolution. As models for biological membranes have evolved, they have yielded a more detailed picture of the plasma membrane (PM). (A) The evolution of the 1920s Gorter and Grendel membrane (1,4) by Danielli and Davson, where the lipid bilayer was sandwiched between adsorbed protein films (5) in order to address some of the PM’s physicochemical properties. (B) Protein-inclusive models advanced to include cytoskeletal elements and biomolecular diversity, culminating in the fluid mosaic model by Singer and Nicholson (6). Although evidence for lipid domains existed when the fluid mosaic model of the PM was proposed (e.g., (7,8,9,10), it was not until the late 1990s, when Simons and Ikonen put forth the lipid raft model (C) (11), that researchers started to carry out systematic studies to characterize them in vivo. To see this figure in color, go online.

Figure 2

Neutron contrast variation. At given H₂O/D₂O ratios, different biomolecules (e.g., proteins, lipids, polysaccharides, DNA) can be made “invisible” to neutrons (i.e., when the water line intersects the line associated with a given class of biomolecules). For example, proteins and lipids have different match points, namely at 8% and 41% H₂O/D₂O ratios, respectively. To see this figure in color, go online.

Figure 3

Schematic of changes in scattering as a function of domain size. Heberle et al. studied the four-component phase-separating lipid mixture of DSPC, DOPC, POPC, and cholesterol using SANS. Chain perdeuterated DSPC (DSPC-d₇₀) was used to introduce neutron contrast, and the H₂O/D₂O solvent ratio was adjusted to match an ideally mixed lipid membrane, i.e., no lipid domains at an elevated temperature. The presence of a SANS signal (right) indicated the existence of lipid domains, which grew in intensity with increasing domain size due to an increasing DOPC:POPC ratio. Domain size was determined by fitting the SANS data using a modified coarse-graining method (28,36). To see this figure in color, go online.

Figure 4

Motional averaging and chemical exchange. In NMR, the precession of the nuclei in a magnetic field is sensitive to their orientations and motions. (A) Orientational dependence of the interaction between nuclei and their environment typically results in a so-called “Pake” powder line shape. Rapid reorientation of the nuclei causes the line shape to narrow. Since ²H is a spin-one nucleus, there are two possible spin transitions, resulting in a symmetrized powder pattern. (B) Nuclei in different magnetic environments will produce distinct resonances in their spectra. If the nucleus is exchanging between the different environments, it experiences chemical exchange. At intermediate rates of exchange, the distinct resonance for each environment will begin to average, and the line shapes will broaden. At faster rates of exchange, their resonances will merge into a single homogeneous resonance. To see this figure in color, go online.

Figure 5

²H NMR spectra from PSM-$d_{31}$/PDPC/cholesterol bilayers can be analyzed by “dePaking” the spectra. The methyl resonances in each lipid domain can then be fitted to determine domain composition (77). (A) Here, the palmitoyl chain of PSM is perdeuterated, thus the NMR spectrum appears as a superposition of symmetric “Pake” powder patterns, where the methyl resonances are the most disordered. (B) A “dePaking” deconvolution algorithm was used to convert the anisotropic Pake powder patterns to isotropic peaks (78). (C) The methyl region resonances for the liquid-disordered (inner pair of peaks) and liquid-ordered (outer peaks) phases were fitted to determine domain composition. The figure was adapted from Kinnun et al. (77). To see this figure in color, go online.

Figure 6

Experimental evidence for the emergent behavior of phonon (E, Q) gaps in the vibrational spectra of lipid membranes as revealed by inelastic x-ray scattering. (A) Schematic of the inelastic x-ray scattering geometry where an incident x-ray beam scatters from a phospholipid membrane (121). (B) Longitudinal and transverse acoustic phonon excitations of a DPPC membrane at 20$°$C (gel phase) and 45$°$C (fluid phase) (121). (C) Schematic of a transient nanoscopic lipid domain surrounded by transient nanopores, effectively forming a transient void ring (122). The existence of transient void rings around transient lipid nanodomains across the membrane is responsible for the passive transmembrane transport of small molecules and solutes (122), molecular-level mechanical stress propagation (123), and self-diffusion of lipids (122). (D) The lateral self-diffusion coefficient $D\left(T\right)$ of a DPPC membrane calculated from Eq. 8 and compared with measured values (124,125,126) from different lipid phases (gel, ripple, and fluid) (122). To see this figure in color, go online.

Figure 7

Imaging giant and large unilamellar vesicles by fluorescence microscopy and cryo-ET using trimeric his6-mCherry. (A and B) Fluorescence microscopy images of giant unilamellar vesicles prepared with different lipid mixtures to tune the size of the lipid domains. Trimeric his6-mCherry was used to image lipid domains due to its fluorescent properties. (C–F) Cryo-ET slices showing regions where trimeric his6-mCherry associates with the membrane (yellow) or where it is absent from the membrane (green). Here, trimeric his6-mCherry’s electron-dense properties were exploited in order to observe where the probe interacts with the membrane. Figure adapted from (144). To see this figure in color, go online.

Figure 8

Visualization of lipid domains using cryo-EM. (A) Large unilamellar vesicles composed of DPPC/DOPC/palmitoyl oleoyl phosphatidaglycerol (POPG)/cholesterol (40/35/5/20) showing clear variations in membrane thickness (D_TT) visually (left) or by color coding (right), where the blue shading indicates regions with a thicker membrane. The scale bar is 100 nm. (B) Histograms showing the thickness distributions of two different lipid domain-forming membrane compositions, DPPC/DOPC/cholesterol (40/40/20) (top) or DPPC/DOPC/POPG/cholesterol (40/35/5/20) (bottom). It was observed that the inclusion of POPG, a negatively charged lipid commonly used to prevent multilamellar vesicle contamination in extruded samples (140), enhances the separation between the two bilayer thickness distributions. Figure adapted from (143). To see this figure in color, go online.

Figure 9

Geometry of a reflectometry experiment highlighting some of the scattered components from a membrane projected onto the plane of the detector. For a given measurement, either the incident angle is fixed with a varying $\lambda$, or vice versa. The neutrons then interact with the membrane with one of four outcomes shown. 1) The neutrons are unaffected, impinging directly on the detector, labeled direct; 2) the neutrons refract, landing between the direct spot and the sample horizon; 3) the neutrons specularly reflect at an angle of incidence $\theta$; the specular reflections are dominated by chemical order along the thickness of the membrane; and 4) the neutrons reflect at an angle that differs from the incident angle $\theta$ and above the sample horizon. In this scenario, the signal is dominated by lateral correlations of the membrane, such as the distribution and possible ordering of lipid domains. To see this figure in color, go online.

Figure 10

(A) The droplet interface bilayer technique is composed of two aqueous lipid droplets suspended in oil, in which their point of contact forms a lipid bilayer. Electrodes in each droplet allows for resistance and capacitance measurements. (B) Capacitance and temperature curves from cooling DPPC/DPhPC/cholesterol (20/50/30 mol %) and brain total lipid extract droplet interface bilayers (C16 alkane solvent) reveal low-enthalpy phase transitions. Here, the starting point of the cooling leg is indicated by arrows with circular end points, while the cooling leg of the response is indicated by dashed arrows. To see this figure in color, go online.