Opening Statement

17 October 2001

Michael Edidin

from the Department of Biology, Johns Hopkins University, Baltimore, MD 21218, USA.

In the past few years, a number of experiments have given credence to the idea that lipid rafts, membrane microdomains, organize and regulate important cellular processes. However, the nature of these microdomains, including their size, composition, organization, and even their very existence, remains highly controversial. This controversy is driven, at least in part, by the fact that most characterization of lipid rafts involves the disruption of membranes and by uncertainties about the effects of probes, such as antibodies or toxins, on the organization of raft components in intact membranes.

A consideration of the size and stability of pure lipid domains,mainly defined in liposomes, suggests that lipid rafts in native membranes are likely to be small (consisting of a few hundred to a few thousand lipid molecules) and transient, unless stabilized by proteins (1). Such small domains may be observed in lipid monolayers [see, for example (2,3)] but are not readily imaged in cell membranes (4). Two different approaches have been used in order to identify lipid rafts in cell membranes. One approach is to observe the random or directed motion of membrane lipids and lipid-anchored proteins in the plane of the bilayer. This approach can report the size of membrane rafts. A second approach is to measure the proximity of raft components by the method of fluorescence energy transfer (FRET). FRET can report clusters of proteins and lipids, which can indicate their confinement in rafts, but cannot define the size of the clusters.

Both nanometer-sized gold beads and fluorescent lipid analogs have been used to define rafts and other lipid domains (5-8). Whatever the label, changes in position of single beads or fluorescence from single molecules can be resolved to ~10 nm, allowing definition of domains as small as 25 to 50 nm in diameter (6). However, the data of other experiments report raft domains hundreds of nanometers in diameter (5,8). The differences in estimated sizes turn on the choice of probes and the assumptions about their diffusion in a membrane bilayer containing lipid rafts.

Because FRET between suitably labeled molecules decays as the sixth power of the distance between them (9), the distance for detectable FRET is limited to 10 nm at most. Thus, FRET observed between labeled molecules thought to be in rafts confirms that they are separated by less than 10 nm. Even in a population of randomly distributed molecules, some will be within FRET distance of one another. Hence, FRET in itself need not report the concentration of molecules in rafts. However, the relationship between the efficiency of FRET, the concentration of the fluorescent acceptor, and the ratio of donor fluorophore to acceptor fluorophore can indicate that some or all molecules of a population are clustered (10). Using a different approach, Varma and Mayor (11) found that all molecules of the glycophosphatidyl inositol (GPI)-anchored folate receptor were clustered and estimated the raft size as ~70 nm in diameter. FRET between labeled folate receptor molecules diminished when the membrane was depleted of cholesterol (11) or when the concentration of nonlabeled lipid-anchored protein increased (12). Sharma et al. reported that FRET between GPI-anchored green fluorescent protein (GFP) molecules also indicated their clustering, but that FRET diminished when the membrane concentration of other GPI-anchored proteins was increased (12). These results are all consistent with small rafts that have a limited capacity for their components (whatever fraction of cell surface area consists of rafts). It may also explain why we (13) found random distributions of endogenous GPI-anchored proteins and of the ganglioside GM1. All of these molecules are expressed at high levels in the cells studied. If only a small fraction of any raft component is actually in rafts (because of competition for rafts by other components), then the FRET signal of this fraction would be swamped by the FRET between randomly distributed molecules.

Reviewing the conflicting and incomplete data on rafts in native membranes, it appears likely that any persistent structures are small and transient. Receptor activation (14,15) or changes in cholesterol content (16) may cause fusion of small rafts into structures hundreds of nanometers in diameter that are observable by light microscopy. A basis for this is suggested by some model and experimental studies showing that cholesterol can form small reversible complexes with lipids containing long saturated acyl chains; for example, with sphingolipids (17). The complexes contain only 15 to 30 molecules, which is 1-10th to 1-20th the number in the small lipid rafts reported by Pralle et al. (6). In mixed lipid monolayers, a fluid phase enriched in these lipid and cholesterol complexes can form. In cells, this phase could contribute to raft formation by fusion of these very small lipid and cholesterol complexes, perhaps driven by aggregated receptor proteins. These complexes also regulate the chemical activity of cholesterol and its availability to other membrane processes. Thus, when membrane cholesterol increases, at first more cholesterol molecules are sequestered in cholesterol and lipid complexes. With further increases in cholesterol, membrane cholesterol is available to interact with cholesterol binding proteins or to form cholesterol-rich domains. In turn, this may drive cholesterol and lipid complexes to fuse to form rafts. It may well be that this broader approach is the way to understand rafts as part of a complex of cholesterol metabolic and functional pathways for cholesterol in membranes (18,19).

Refereces

Return to Topics