The LipidWeb blank

Sphingolipids, Membrane Rafts and Caveolae

1. Introduction

It is impossible to understand the functions of sphingolipids in cells without some knowledge of their distinctive physical location within membranes. Sphingolipids are located only in the outer (exoplasmic) leaflet of the plasma membrane bilayer, while certain glycerophospholipids, including phosphatidylinositol, phosphatidylserine and phosphatidylethanolamine, occur only in the inner (cytoplasmic) leaflet under normal circumstances. Cholesterol is believed to occur in roughly equal proportions in both leaflets. Further, sphingomyelin and other sphingolipids together with cholesterol and trans-membrane proteins are located in an intimate association in specific sub-domains or 'rafts' or in related invaginated structures termed 'caveolae' of membranes. Rafts are considered to be laterally segregated regions that form transiently as a result of selective affinities between sphingolipids and specific membrane proteins, which act to compartmentalize and provide a platform for the latter and thereby separate different biochemical functions; more formal definitions are discussed below. Several distinct types of protein are associated with rafts and caveolae, including some with essential membrane functions (glycosylphosphatidylinositol-linked proteins, caveolins, flotillins), signalling proteins (e.g. Src family kinases), G protein-coupled receptors, and others that are lipid linked (palmitoylated, myristoylated, hedgehog). While the concept of rafts was developed for the Golgi and then the plasma membrane in eukaryotes, it has become evident that specialized membrane domains with signalling functions are also present in nuclei, endoplasmic reticulum and mitochondria. Functional membrane nanodomains are essential elements of prokaryotic cell membranes also.

Raft domains are present primarily in outer leaflets of asymmetric cell membranes, but they are believed to be coupled with lipids in the inner leaflet. It has become evident that such rafts do not represent a single monolithic structure, but a heterogeneous collection of domains differing in their protein and lipid compositions as well as in their stability with respect to size and time, i.e. length scales of tens of nanometers and time scales of milliseconds. Up to 50% of the plasma membrane in animal tissues may consist of rafts, and the apical membrane of epithelial cells especially may behave like a large raft. Many aspects of raft structure are uncertain and controversial, because of the technical difficulties involved in their study in membranes of living cells as opposed to model systems. Indeed, it has been argued by some that all of the evidence for the existence of rafts is indirect, and that alternative explanations may be possible. On the other hand, recent evidence obtained by using lipid-specific toxin fragments supports the concept of raft domains. The current controversy now appears to be mainly concerned with the mechanism(s) behind raft formation.

The terms 'rafts', 'microdomains', 'nanodomains' and 'transient nanodomains' in membranes are sometimes used interchangeably, although this may be too simplistic an approach. While the term 'raft' can be considered trivial and non-scientific, I will continue to use it here for convenience and until an alternative consensus has clearly emerged.

2. Raft Structure and Organization

Definitions: A formal definition of what constitutes a membrane raft was proposed after a conference on the subject, and at the time it appeared to be accepted by the lipid community as a useful working hypothesis (Pike, L.J. J. Lipid Res., 47, 1597-1598 (2006); DOI), i.e.

"Membrane rafts are small (10-200 nm), heterogeneous, highly dynamic, sterol- and sphingolipid-enriched domains that compartmentalize cellular processes. Small rafts can sometimes be stabilized to form larger platforms through protein-protein and protein-lipid interactions."

Others prefer "dynamic nanoscale sterol, sphingolipid-enriched, ordered assemblies of specific proteins, in which the metastable resting state can be activated to coalesce by specific lipid–lipid, protein–lipid, and protein–protein interactions" (Simons, K. and Sampaio, J.L. Cold Spring Harbor Persp. Biol., 3, a004697 (2011);  DOI). Simpler and perhaps more in tune with current thinking, that proposed by Sezgin, E. et al. Nature Rev. Mol. Cell Biol., 18, 361-374 (2017);  DOI) is -

Rafts are "transient, relatively ordered membrane domains, the formation of which is driven by lipid-lipid and lipid-protein interactions".

Earlier definitions tended to depend on the methods used experimentally to produce raft preparations and especially that derived from their resistance to non-ionic detergents, i.e. their insolubility in cold 1% Triton X-100, for example. While they were once described as detergent-resistant membranes or 'DRM', this has resulted in much confusion and controversy in the literature. DRM certainly contain raft material, but it is now considered by many experts in the field that they should no longer be treated as rafts per se, as more robust and definitive methods of studying raft formation are available. However, the transient nature of the phenomena mean that technical difficulties remain in studies of raft structure and formation in living cells that can confound interpretation of data and confuse the picture.

The sphingolipid-cholesterol hypothesis: Much of the early literature on raft formation is based on the concept of specific physical interactions between sphingolipids and cholesterol in membranes. According to this theory, the packing of the saturated acyl chains of sphingolipids with cholesterol is thermodynamically favoured over that with unsaturated acyl chains, and cholesterol is essential to the process of raft formation in the plasma membrane. If either the sphingolipid or cholesterol concentration is depleted by any means, the other follows. Indeed, there is evidence that in animal tissues sphingomyelin regulates the capacity of membranes to absorb cholesterol and thereby controls its flux between the plasma membrane and other regulatory pathways in the endoplasmic reticulum. Cholesterol interacts particularly strongly with sphingomyelin and much less with glycosphingolipids. The key to the phenomenon is that the two lipids form extensive intramolecular hydrogen bonds. It is noteworthy that epithelial cells of the kidney and stomach, which are especially enriched in cholesterol and sphingolipids, are highly impermeable to water and solutes.

Physical chemical studies have demonstrated that the conformation and orientation around the amide group of sphingomyelin are relatively rigid as is appropriate for an intermolecular hydrogen bond with a neighboring sphingomyelin molecule, while cholesterol enhances the order of the central hydrocarbon chains of sphingomyelin appreciably. To add to this, the sphingomyelin/cholesterol system is tolerant of temperature changes, even at low cholesterol concentrations. The mechanism of raft formation is believed to involve an enhancement of the order of the central sphingomyelin alkyl chains by the rigid cholesterol molecules by restricting the chain fluctuation and thus shortening the intermolecular distances between sphingomyelin molecules to facilitate the formation of hydrogen bonds between them thereby promoting the formation of a hydrogen bond network and a stable ordered phase that is resistant to temperature fluctuations. This tight packing leads to a smaller molecular surface area than would be predicted from the sum of those of the individual molecules. As a result, lateral diffusion occurs with separation of the lipids containing saturated fatty acyl chains from those that are highly unsaturated. Indeed, there is evidence that glycerolipids containing docosahexaenoic acid especially are incompatible with more saturated lipids and drive the segregation of the latter into raft domains.

As sphingolipids containing long, largely saturated acyl chains, they pack more tightly together, thus giving sphingolipids much higher melting temperatures than membrane glycerophospholipids. This tight acyl chain packing is essential for raft lipid organization, since the differential packing facility of sphingolipids and cholesterol in comparison with glycerophospholipids is believed to lead to a phase separation in the membrane that gives rise to sphingolipid-rich regions ('liquid-ordered' phase) surrounded by glycerophospholipid-rich domains ('liquid-disordered' phase). Such ordering is responsible for the resistance to attack by detergents. Sphingolipids tend to have more free hydroxyl groups, both in the long-chain bases and fatty acid components than glycerolipids, and these enter into hydrogen bonding and contribute to the stability of rafts, and the presence of very-long-chain fatty acid components (e.g. C26) is believed to be essential. One factor may be interleaflet coupling of liquid-ordered domains between the outer and the inner leaflets via interdigitation through such long acyl chains. As these rafts are relatively small (approximately 50 nm diameter and containing roughly 3000 sphingomyelin molecules) and mobile, they are not easy to study by microscopic methods; they are thicker than normal membranes (46 versus 40 angstroms). The rapid rate of formation-dissolution of rafts also hinders their study, although no accurate time-scale has been determined.

Lipid-protein model: There is a conflicting view of the raft hypothesis, based on studies by high-resolution secondary ion mass spectrometry, that is gaining traction. This suggests that sphingolipids are concentrated in micrometer-scale membrane domains while cholesterol is evenly distributed within the plasma membrane. In this model, sphingolipid distribution and raft formation in the plasma membrane is dependent on the cytoskeleton, but not on favorable interactions with cholesterol, and it does appear that the sphingolipid-cholesterol model may not hold when there are large amounts of protein in the membrane. A further confounding factor is that as many as 250 transmembrane proteins may interact directly with cholesterol at a consensus sequence termed the CRAC motif for "cholesterol recognition/interaction amino acid consensus" and this is independent of sphingolipids; many of these proteins function in organizing signalling hubs and include the caveolins (see below). It is entirely possible that formation of lipid rafts is a multifactorial process and that no single model will provide an adequate description.

cartoonAs an alternative to lipid-lipid interactions, raft formation is driven and stabilized by lipid-protein and protein-protein interactions, and membranes should be regarded as lipid-protein composites rather than a solution of protein in a lipid solvent. Indeed, membrane proteins are essential for raft formation, and the reggies/flotillins, which are sometimes described as molecular scaffolding, are especially important in that they promote the assembly of post-translationally modified proteins such as glycerophosphoinositol(GPI)-anchored proteins into rafts. These then recruit other proteins, which can include tyrosine kinases, phosphatases and other signalling proteins, and facilitate interactions among them. However, in general most proteins partition into the fluid-disordered rather than into the fluid-ordered nanodomains, so raft-related proteins are in the minority in membranes.

Palmitoylation or myristoylation is often required to target proteins to lipid rafts (see our web page on proteolipids), and for example flotillin is anchored in the lipid bilayer via N-terminal myristoyl and palmitoyl groups; it also binds to actin and the cytoskeleton and it is now evident that these are involved in compartmentalization of the membrane and raft formation. Other proteins with trans-membrane segments are targeted to rafts by specific amino acid sequences that provide much of the important biological properties of rafts, and are also required to maintain their stability. By means of this interplay of lipid-based raft units with protein-mediated assembly of specific protein complexes, functional domains are generated with high biological activity in the plasma membrane. From comparisons of membrane solubility in different detergents, it appears that there may exist subsets of membrane raft domains, which differ in their molecular compositions. In particular, they may contain distinct ganglioside species as well as the glycosyl phosphatidylinositol anchors. On the other hand, others have expressed doubt that GPI-anchored proteins occur in raft domains in plasma membranes of live cells.

While lateral diffusion may explain how the lipids aggregate in rafts, the introduction of proteins might require energy so other forces must be involved unless the protein is in a lower energy state in the raft. Then the process might be driven by entropy. Another question remains as to how the raft and non-raft phases remain relatively stable, and the answer may be that there is an interaction with cortical actin cytoskeleton underneath the plasma membrane that holds the raft in place. Further stability is introduced by the action of ceramide and lysosphingolipids, which can induce coalescence of lipid rafts into larger platforms or domains. For example, ceramides are generated from sphingomyelin in the plasma membrane by the action of sphingomyelinases by a well-regulated and tunable mechanism. Ceramides are able to form gel phases in membranes and displace cholesterol from its packing with sphingomyelin with those species containing fatty acids of medium chain-length (C12 to C18) as the most effective in promoting raft formation. Such modified rafts tend to have different protein contents and thence signalling functions from the cholesterol-rich rafts, so some consider these ceramide-rich platforms as a third type of membrane microdomain.

The role of other lipids: Rafts also contain some glycerophospholipids and these often consist of relatively simple molecular species with one saturated and one monounsaturated acyl chain. In addition, both ethanolamine plasmalogens, especially those containing arachidonic acid, and phosphatidylserine are enriched in rafts as compared to the plasma membrane as a whole. Other lipids can apparently form microdomains in membranes that are raft-like in their properties. One such is phosphatidylglucoside, in which the diacylglycerol component contains two saturated fatty acids, so is comparable to a ceramide in some respects. Similarly, phosphatidylinositol 4,5-bisphosphate is believed to interact with cationic residues of a large array of proteins together with cholesterol to form localized membrane domains. While a great deal of attention has been focused on rafts in the outer leaflet of the plasma membrane, relatively little is known of the structural organization and properties of the corresponding inner leaflet or how the two layers interact. However, it is believed that raft domains may sometimes form on the inner leaflet also and there is presumed to be interaction between the two.

Non-raft regions of membranes: A corollary of the existence of rafts, which are rich in sphingolipids and with low levels of polyunsaturated fatty acids, is that micro-domains must also exist in membranes that are depleted in sphingolipids/cholesterol and enriched in polyunsaturated fatty acids. Indeed, the rigid structure of cholesterol and the highly flexible chains of docosahexaenoic acid, for example, are incompatible and promote the lateral segregation of membranes into rafts. Micro-domains that are enriched in polyunsaturated lipids and cholesterol/sphingolipid-poor are technically even less easy to study than rafts, but these may also contain distinctive proteins and have important biological functions. There are also suggestions that the interface between rafts and non-raft regions may attract a specific range of proteins.

Plants, yeasts and bacteria: Specific micro-domains or rafts that are enriched in the characteristic plant sterols, sterol glucosides and sphingolipids, such as glucosylceramide and especially the glycosylinositol phosphoceramides, have been detected in the plasma membrane and Golgi apparatus of plant cells. They are especially important in the plasmodesmata, the narrow passages through the cell walls of adjacent cells that allows communication between them, where the membranes are enriched in sterols and sphingolipids containing predominantly very-long-chain fatty acids. These microdomains contain distinct sets of proteins, including those anchored by glycosyl phosphatidylinositol or glycosylinositol phosphoceramides. As in animal cells, such rafts may assist in positioning proteins in specific regions of the cell where they can function in development, membrane trafficking and signalling. The proteins flotillins and plant-specific remorins are seen as markers for distinct types of nanodomain. Membrane nanodomains act as signalling hubs also in plant pathogen-interactions and in root nodule symbiosis between plants and arbuscular mycorrhiza (fungi) or rhizobia (bacteria).

In the plasma membrane of the yeast Saccharomyces cerevisiae, raft-like structures have been identified that do not contain ergosterol but are composed primarily of sphingolipids, possibly inositol phosphorylceramide, and contain glycosylphosphatidylinositol-anchored proteins. However, micro-domains enriched in both ergosterol and sphingolipids have been found in the plasma membrane of other yeast species.

While bacteria in general lack much of the cellular compartmentalization present in eukaryotes, as well as both the sphingolipids and sterols characteristic of rafts, they arrange various signal transduction, protein secretion and transport processes in 'functional membrane microdomains'. These operate in much the same way as rafts in eukaryotes, and their formation is believed to involve interaction of farnesol and/or farnesol-derived polyisoprenoid lipids with flotillin-like proteins. Proteins associated with signal transduction, membrane trafficking and the regulation of metabolism are held in close proximity, and so increase the possibilities for interaction and the efficiency of related cell processes. Indeed, membrane organization may be of special importance in unicellular organisms, as this is the boundary between the organism and its environment and is vital for many cellular processes, including cell division and signal transduction. Borrelia burgdorferi is an extracellular tick-borne bacterium (spirochete) that causes Lyme disease and is the only prokaryote known to have cholesterol-rich microdomains (but not sphingolipids) with all the hallmarks of eukaryotic lipid rafts. It requires cholesterol from the host for growth and multiplication as it is unable to synthesise cholesterol itself.

3. Caveolae

Those subdomains in the plasma membrane related to rafts and termed ‘caveolae’ ("little caves") are flask-shaped invaginations 50 to 100 nm in diameter in membranes that provide a cholesterol- and (glyco)sphingolipid-rich environment, and they are found abundantly in vascular endothelial cells, adipocytes, smooth muscle cells and fibroblasts; for example, they can amount to 50% of the surface area of adipocytes. They are stabilized by particular membrane-spanning scaffolding proteins, the caveolins, a family hairpin-like proteins that bind strongly to cholesterol. Of these, caveolin-1 is the main structural component, i.e. a 24 kDa integral membrane protein with three palmitoylation sites at the C-terminal domain, which bind it to the inner leaflet of the plasma membrane; it also has phosphorylation, ubiquitination and SUMOylation sites. Caveolin-2 is required for lung functions, while caveolin-3 is specific to muscle. However, caveolin-2 does not form caveolae independently but associates with caveolin-1 in hetero-oligomers. The presence of caveolins is the defining property of caveolae, and it is believed that there may be different classes of caveolae with different compositions, physical forms and metabolic functions. In contrast to rafts, caveolae are clearly visible and their limits are recognizable by electron microscopy; rafts differ from caveolae in that they are self organized.

The genesis of caveolae is a complex process of stepwise assembly. In brief, it begins in the endoplasmic reticulum with the translation of caveolins and proceeds via the Golgi to the plasma membrane via a secretory pathway requiring the ganglioside GM1 and glycosylphosphatidylinositol-linked proteins. The caveolin proteins are inserted into the endoplasmic reticulum co-translationally and pass to the Golgi where they self-associate to form oligomers of 12 or 14 monomers with addition of cholesterol, which binds to the CRAC site in the caveolins. Experimental depletion of plasmalemmal cholesterol results in the loss of caveolae. As the complex nears the plasma membrane, palmitoylation of the caveolins occurs by means of palmitoyl acyltransferases on three cysteines in the C-terminal domains. At the plasma membrane, they can then assemble with more caveolin oligomers to form caveolae with the aid of further structural proteins termed 'cavins' (~50 kDa), of which four occur in vertebrates. These have been shown to bind to each other in vitro and to the anionic phospholipids phosphatidylserine and phosphatidylinositol 4,5-bisphosphate, which are concentrated in caveolae inner membranes. It is believed that these bind with the caveolins in a large caveolar coat complex in vivo, so for example, cavin-1 stabilizes caveolin-1 with the aid of adaptor proteins. The effect is to shape the caveolae, leading to invagination and stabilization of the complex.

Scottish thistleOther than a negligible content of glycosyl phosphatidylinositol, the lipid composition of caveolae is similar to that of rafts in general in that they contain appreciable amounts of sphingolipids and cholesterol. However, the gangliosides GM1 and to some extent GM3 appear to be concentrated in caveolae, while phosphatidylserine is the main anionic lipid and has a crucial role in their formation and stability. One hypothesis that cannot as yet be tested is that the neck region of caveolae with a strong positive curvature may be highly enriched in glycosphingolipids, which may be largely absent from their lumen. For example, there is evidence that the caveolins interact directly with glycosphingolipids especially, and that the latter may have regulatory effects in the signalling functions of caveolae. It has even been argued that caveolae best meet the accepted definition of a raft and should be used as a provisional standard against which other proposed raft structures should be judged.

A general picture has emerged of caveolin-1 and sphingolipids forming the scaffold for an integrated network of interactions, which contribute to the regulation of cellular functions via their influence upon receptor molecules in the plasma membrane. Caveolae are known to contain specific signalling proteins, including Ras proteins, G proteins and growth factor receptors, not present in other raft microdomains. They appear to have a role in controlling the level of free cholesterol in cells, and thence may affect signalling processes. They are especially abundant in endothelial cells, adipocytes, smooth muscle cells and fibroblasts, but have not been found in lymphocytes or mature neurons. In adipocytes, they may regulate the flux of fatty acids across the plasma membrane. Also in adipocytes, insulin is the main hormone that affects metabolism, and the receptor at the plasma membrane is located in caveolae with possible implications for diabetes, obesity and other metabolic disorders. It appears that caveolae may function as an important route by which nutrients, such as folate, glucose and fatty acids, are able to cross the plasma membrane. Although the mechanism is still obscure, one further suggestion is that caveolae may act as transport vesicles, which bud from the plasma membrane in response to specific cues. They may have a non-signalling role in the repair of membrane damage in response to physical stress, perhaps by acting as a reservoir of components that can flatten out under mechanical stress. Disassociation of caveolae is a highly complex process that involves restructuring of the caveolin oligomers, ubiquitination, internalization and eventually degradation.

There is some evidence that changes in caveolae create a catabolic microenvironment in tumours that supports oxidative mitochondrial metabolism in cancer cells and so contributes to the poor survival rates for cancer patients. They also facilitate the entry into cells of certain pathogens including Chlamydia, which enters host cells through membrane areas rich in cholesterol and ganglioside GM1 and co-localizes with caveolins-1/2.

4. Raft Function

It should be recognized that lipid rafts in general are dynamic structures, which can be formed or undergo compositional changes during signalling events and are short-lived (milliseconds or less). There may also be some form of cross-talk between different raft populations, which can coalesce during activity. Thus in resting cells, sphingolipids may exist in small and highly dynamic domains, which on stimulation can stabilize and grow. In the process, they initiate biochemical reactions by promoting interactions between proteins. Raft domains may also aid membrane trafficking, for example by facilitating the transfer of specific proteins, such as the glycosyl phosphatidylinositol-anchored proteins, from the endoplasmic reticulum or Golgi to the plasma membrane. As discussed above, they may have a role in cholesterol homeostasis.

Lipid rafts are believed to modulate signalling events in a number of different ways according to the composition of the specific subpopulations. Thus, the location of signalling molecules within one such micro-domain might in itself be a control mechanism, as a protein activated by phosphorylation within the raft might be prevented from interacting with an inactivating phosphatase in another region of the membrane, for example. Alternatively, interactions with specific raft lipids or with the distinctive biophysical environment could change the conformation of a resident protein and thence its activity. By concentrating all of the components of particular signalling pathways within one domain, lipid rafts could promote signalling in response to stimuli, while movement of signalling molecules in and out of the raft could control whether cells are able to respond to stimuli. Similarly, communication between different signalling pathways could be simplified if the relevant molecules were concentrated in the same lipid raft. In contrast, rafts might also regulate signals in a negative manner by sequestering signalling molecules in an inactive state. The physical properties of rafts may be key factors in these interactions. Thus, in response to receptor activation or other stimuli, sphingolipid compositions in rafts may be altered with effects on membrane architecture or morphology producing further downstream events.

Many of the signal transduction functions of rafts are related to cell adhesion, migration, survival and proliferation, but they can also support receptor-mediated apoptotic signalling. In effect, membrane rafts may act as supporting structures that enable segregation of pro- from anti-apoptotic molecules. As many oncogenic proteins are located in raft­like domains and mitogenic signalling stems from various cell surface receptors, it is believed that rafts may be involved in cancer development and progression and that rafts may be a novel target in the treatment of cancer. Similarly, the transition of macrophages into foam cells may be raft-dependent with the potential to affect cardiovascular disease, while in contrast ion channels in caveolae are essential for normal cardiac functions. A deficiency of flotillin in cells compromises many lipid raft-associated cell processes and is associated with Parkinson's and Alzheimer's diseases, for example.

cartoonRaft formation is believed to be particularly important to the activity of T cells, i.e. lymphocytes derived from the thymus gland that are intimately involved in antibody production. Thus, lipid rafts are a key element of membrane organization that appear to be crucial for the initiation of T cell signalling by enabling efficient interaction between antigens and receptors, and they provide a platform for the interaction of Toll-like receptors, which are critically involved in inflammatory responses, with their ligands in cells. Lipid rafts have an essential function in enabling tumor necrosis factor alpha (TNFα), one of the cytokines involved in systemic inflammation, to bind to its specific receptor; the activity of sphingomyelin synthase 2 is an important factor that modulates this activity. For example, lipid rafts have a vital role in the redox signalling that regulates the pathophysiology of many degenerative diseases in that large redox signalling molecules aggregate into lipid rafts to produce various reactive oxygen species (ROS). NADPH oxidase is considered especially important in this context and lipid rafts provide an essential platform to aggregate and assemble the necessary subunits of the enzyme into an active complex to produce ROS. In mitochondria, raft-like domains are involved in the response to oxidative stress and apoptosis, while in nuclei the signalling system that generates diacylglycerol and inositol-1,4,5-triphosphate are located in similar microdomains in the membrane.

It is apparent that the biological functions of neurotransmitters and their receptors, in particular acetylcholine and serotonin receptors, are highly dependent upon sphingolipids and cholesterol in raft domains. Among other events, the lipids interact with receptors to alter their conformation ("chaperone-like" effect), thus regulating neurotransmitter binding and signal transducing functions.

There is evidence that certain pathogens activate the acid sphingomyelinase that releases ceramide in membrane rafts transforming them into larger units, which can mediate the internalization of bacteria, viruses and parasites into host cells, to initiate programmed cell death (apoptosis) and release signalling molecules. They may also assist in the budding of viruses from infected cells. In effect, rafts and caveolae enriched in cholesterol re-organize the receptor and intracellular signalling molecules in the cell membrane and enable the interaction of pathogens with cells. The lipids of viruses are derived from the host membranes, and for example, it has recently been demonstrated that the lipids of the HIV virus are enriched in sphingolipids that appear to be derived very specifically from rafts. Further, these lipids are essential for the infectivity of the virus. In contrast, rafts can assist cells to defeat infection by activation of transcription factors and the release of cytokines.

Suggested Reading

Lipid listings Credits/disclaimer Updated: August 26th, 2019 Author: William W. Christie LipidWeb icon