164

Ann. N.Y. Acad. Sci. 1014: 164–169 (2004). © 2004 New York Academy of Sciences.doi: 10.1196/annals.1294.017

Lipid Rafts and Apical Membrane Traffic

JOACHIM FÜLLEKRUG AND KAI SIMONS

Max-Planck-Institute of Molecular Cell Biology and Genetics, Dresden, Germany

ABSTRACT: Lipid rafts are dynamic assemblies floating freely in the surround-ing membranes of living cells. This membrane heterogeneity provides a usefulconcept for understanding processes as diverse as cell polarity, signal trans-duction, and membrane sorting. Individual rafts are small entities containingthousands of lipids but only a few proteins. Regulation of raft association andsize is an elementary feature of interactions at the molecular level. By cluster-ing small rafts into a bigger platform, proteins are brought together for modi-fication. Oligomerization might transform a monomeric weakly raft-associatedprotein into an assembly with higher raft affinity. Lectins are multivalentglycoprotein-binding proteins and are likely to be key players in mediating theclustering of rafts in vivo. Glycosylation-dependent surface delivery in a polar-ized fashion is a feature conserved across evolution, and we expect lectins to beat the heart of the molecular machinery responsible for lipid raft delivery to thecell surface. Currently, we are evaluating candidate proteins by affinity chro-matography, proteomics, and RNA interference.

KEYWORDS: polarity; rafts; epithelia; glycosylation; MDCK

LIPID RAFTS

Lipid rafts are dynamic microdomains in the membranes of living cells, com-posed of cholesterol and sphingolipids. Cholesterol and sphingolipids carryingsaturated hydrocarbon chains assemble to form tightly packed subdomains corre-sponding to liquid-ordered phases biophysically characterized in model mem-branes.1,2 These lipid rafts float freely in the surrounding membrane which is morefluid and analogous to the liquid-disordered phase. At the molecular level, the higherfluidity is the consequence of the high surface area occupied by unsaturated phos-pholipids compared to the dense packing of the sphingolipid-cholesterol assemblies.Cholesterol also condenses monounsaturated phospholipids but leaves them in theliquid-disordered phase.3

Proteins strongly associating with lipid rafts include glycosylphosphatidyl-inositol (GPI)-anchored proteins, doubly acylated proteins, and several transmem-brane proteins, especially palmitoylated ones.

Address for correspondence: Joachim Füllekrug, Ph.D., Max-Planck-Institute of MolecularCell Biology and Genetics, Pfotenhauerstrasse 108, D-01307 Dresden, Germany. Voice: (+49)351 2102536; fax (+49) 351 2101209.

fuellekr@mpi-cbg.de

165FÜLLEKRUG & SIMONS: LIPID RAFTS AND APICAL MEMBRANE TRAFFIC

CELL POLARITY

Epithelial cells form sheets that line surfaces, especially in tissues specialized forabsorption and secretion. Two different plasma membrane domains are separated bytight junctions: the basolateral surface is responsible for cell-cell and cell-matrixinteractions, and the apical membrane faces the lumen. The protein and lipidcomposition of these two membranes is strikingly different, or “polarized.”4 Theapical plasma membrane is highly enriched in glycosphingolipids and GPI-anchoredproteins.5

Cholesterol and ceramide, the precursor for sphingolipids, are synthesized in theendoplasmic reticulum (ER), but the steady-state concentration of these compoundsis too low to allow formation of rafts. Sphingomyelin and glycosphingolipids aresynthesized in the Golgi apparatus, and it is here that raft assembly takes place.Cholesterol and other raft lipids are excluded from retrograde traffic between theGolgi apparatus and ER,6 and the overall consequence is an increasing concentrationof cholesterol and sphingolipids from the ER to the plasma membrane.5

Biochemically, this corresponds to an increased resistance of membranes to non-ionic detergents in the cold.7 This method is a good first approach to investigate raftassociation of proteins, although not without pitfalls. Raft proteins acquire detergentresistance while transiting the Golgi apparatus, concomitantly with the enrichmentof raft lipids (FIG. 1).

FIGURE 1. Floatation analysis. Resistance to non-ionic detergents in the cold is a hall-mark of lipid microdomains enriched in sphingolipids and cholesterol and is widely used inassessing raft association of a given protein. Raft proteins would still be associated with lip-ids after detergent extraction and have a lower density than that of solubilized membraneproteins not associated with rafts. The different densities are exploited by ultracentrifuga-tion, leading to floatation of DRMs (fractions 2–4) versus solubilized proteins (fractions7,8). Two raft markers for MDCK cells, VIP17 and caveolin-1, localize to detergent-resistant membranes (DRMs) at the top of the gradient. Note that although gp114 is an apicalmarker protein derived from a membrane rich in glycolipids and cholesterol, it is solubilizedunder these conditions. However, crosslinking of proteins often strengthens a weak affinityfor rafts, and the same has been observed for gp114.18

166 ANNALS NEW YORK ACADEMY OF SCIENCES

At the level of the trans-Golgi network (TGN), a sorting process leads to transportcontainers with different protein and lipid composition. Apical containers arederived from sorting platforms enriched in lipid rafts, whereas basolateral sortingrelies on the binding of adaptor proteins to short amino acid sequences present in thecytoplasmic tails of proteins. This TGN-mediated sorting ensures the maintenanceof the polarized phenotype not only in epithelial cells but also in other types ofpolarized cells such as neurons. Cognate pathways retaining some characteristics ofpolarized sorting are also present in fibroblasts.8

CLUSTERING OF RAFTS

There is agreement that individual rafts are small, with a diameter of 50 nm9 (seeRef. 10 for a different view). This size corresponds to several thousands of lipid mol-ecules but, based on fibroblast measurements, only to about 10–20 protein mole-cules. This means that regarding protein content there are many different kinds ofrafts with a unique protein composition. Consequently, for efficient interaction ofdifferent raft-associated proteins, rafts need to form larger platforms. This processcan be manipulated experimentally by the use of antibodies in an approach referredto as clustering or crosslinking.

Lateral crosslinking of membrane proteins leads to multimerization of protein-lipid interactions. Weak interactions are amplified and cause stronger attracting orrepelling forces. Raft domains coalesce and segregate from non-raft lipids andproteins.

Experimentally, if two raft-associated proteins expressed in fibroblasts are sub-jected to antibody clustering, they colocalize in large patches readily observed bylight microscopy. By contrast, antibody clustering of a non-raft protein and a raftprotein leads to segregated patches. The crosslinking approach works only on intactrafts as evidenced by inhibition after cholesterol depletion.11

Some proteins are only weakly associated with lipid rafts in the monomeric state,but oligomerization, either by ligands in vivo or by antibodies in vitro, will increaseraft affinity. This is the case for the IgE receptor in the allergic immune response ofmast cells. In the resting state, the IgE receptor is outside rafts, but moves into raftsafter crosslinking by multivalent IgE-antigen complexes. The doubly acylated Lyntyrosine kinase is raft associated and is thought to initiate the signaling response byphosphorylating the subunits of the IgE receptor.12

GLYCOSYLATION AND APICAL SORTING

Glycosylation was shown to be crucial for apical delivery for a variety of differentproteins in mammalian cells, both soluble and membrane associated.13,14 Recentresults from our lab suggest that glycosylation-dependent polarized delivery is alsoa feature of yeast cells, underscoring the importance of glycosylation in this processacross evolution. The simplest model to integrate the findings on lipid microdomainsand glycosylation is the postulation of lectins binding specifically to glycoproteins(FIG. 2). Multivalent lectins would bind to several glycoproteins at the same timeand, by crosslinking, enhance low intrinsic raft affinities. We think that several

167FÜLLEKRUG & SIMONS: LIPID RAFTS AND APICAL MEMBRANE TRAFFIC

lectins (which could be raft associated themselves) will account for different kindsof cargo glycoproteins. Small rafts would be brought together into larger sortingplatforms to allow polarized sorting in the TGN. This would fit well with the obser-vation that at least in some cases, raft association is not sufficient for apical deliveryof GPI-anchored proteins, but that glycosylation has to be present to make theprocess more efficient.15

MDCK CELLS AND GLYCOSYLATION

MDCK cells have served as a model system in epithelial cell biology for years.The main reason is that these cells differentiate into fully polarized cells when grownon semipermeable filter supports. The cells form a monolayer, with tight junctionsseparating a basolateral domain (facing the filter support and neighboring cells) froman apical surface. Protein sorting, raft formation, and glycosylation have been stud-ied extensively in MDCK cells. Our current focus is the identification and character-ization of lectins in this epithelial polarity model system. Using glycoproteinspurified from MDCK cells, we identified an endogenous lectin by affinity chroma-tography, which is described elsewhere. Another approach focuses on marker glyco-proteins of MDCK cells. Monoclonal antibodies derived from earlier studies16 haveserved to identify apical or basolateral membranes in immunocytochemistry andbiochemical analysis (FIG. 3). However, the molecular identity of these proteins isunknown in most cases. Experimental findings relating to the marker proteins them-selves have remained isolated within the model system of MDCK cells. We haveidentified a frequently used apical marker protein, the sialoglycoprotein gp114, as amember of the carcinoembryonic antigen (CEA) family. Previous results on endo-cytosis and transcytosis of gp114 in MDCK cells17 should now lead to new insights

FIGURE 2. Lectins cluster glycoproteins for apical sorting. We suggest that there is afamily of lectins that would be crucial for establishing an apical sorting platform. Multi-valent lectins would crosslink different apical cargo proteins and glycolipids. This cluster-ing would stabilize weak individual interactions between proteins and lipid microdomains.The lectins are very likely glycoproteins, and by binding also to each other they would createa lattice of glycoproteins and lectins. Another prediction is that the lectins would have anaffinity for raft microdomains. Glycoproteins containing a basolateral sorting determinantwould be sequestered away from the apical lectin raft platform with the help of adaptorproteins.

168 ANNALS NEW YORK ACADEMY OF SCIENCES

FIGURE 4. Lectin-mediated clustering of rafts. Individual rafts are small, containingonly a few proteins. Clustering these rafts into larger assemblies enables molecular inter-actions between proteins. Lectins are major candidates for regulating raft size at the plasmamembrane.

FIGURE 3. MDCK cell polarity. MDCK cells were grown on permeable filters for 4days. (a) Indirect immunofluorescence using antibodies directed against gp114 shows gran-ular staining of the apical surface that is especially prominent towards neighboring cells. (b)A vertical section demonstrates the apical confinement of gp114. (c) A midsection throughan MDCK cell shows the localization of p58 to the lateral domain. In addition, there is apool of intracellular granules containing p58. (d) The vertical view depicts lateral and basalstaining.

169FÜLLEKRUG & SIMONS: LIPID RAFTS AND APICAL MEMBRANE TRAFFIC

regarding CEA biology and function. In addition, the p58 basolateral glycoproteinwas identified as the β subunit of Na+K+-ATPase. This identification of markerproteins is embedded in a proteomics effort on glycoproteins/lectins of MDCK cells.Candidate proteins are being tested by RNA interference to evaluate their role inapical sorting and epithelial traffic routes.

SUMMARY

The concept of lipid raft microdomains builds a framework to incorporate seem-ingly unrelated processes such as cell polarity, signal transduction, and membranesorting in a coherent fashion. Crucial to this concept is the regulation of raft associ-ation and size. By clustering small individual rafts together into a larger platform,efficient interaction of raft-associated proteins is realized. Weak raft affinities ofindividual proteins are strengthened by oligomerization. Apart from classic ligandsand cytoskeletal scaffolds, we believe that lectin-mediated clustering of glyco-proteins is an important way for cells to regulate lipid rafts in vivo (FIG. 4).

REFERENCES

1. BROWN, D.A. & E. LONDON. 1998. Annu. Rev. Cell Dev. Biol. 14: 111–136.2. SIMONS, K. & E. IKONEN. 1997. Nature 387: 569–572.3. SILVIUS, J.R. 2003. Biochim. Biophys. Acta 1610: 174–183.4. RODRIGUEZ-BOULAN, E. & W.J. NELSON. 1989. Science 245: 718–725.5. VAN MEER, G. 1989. Annu. Rev. Cell Biol. 5: 247–275.6. BRUGGER, B., R. SANDHOFF, S. WEGEHINGEL, et al. 2000. J. Cell Biol. 151: 507–518.7. BROWN, D.A. & J.K. ROSE. 1992. Cell 68: 533–544.8. KELLER, P. & K. SIMONS. 1997. J. Cell Sci. 110: 3001–3009.9. PRALLE, A., P. KELLER, E.-L. FLORIN, et al. 2000. J. Cell Biol. 148: 997–1008.

10. ANDERSON, R.G.W. & K. JACOBSON. 2002. Science 296: 1821–1825.11. HARDER, T., P. SCHEIFFELE, P. VERKADE & K. SIMONS. 1998. J. Cell Biol. 141: 929–942.12. HOLOWKA, D. & B. BAIRD. 2001. Semin. Immunol. 13: 99–105.13. SCHEIFFELE, P., J.PERANEN & K. SIMONS. 1995. Nature 378: 96–98.14. GUT, A., F. KAPPELER, N. HYKA, et al. 1998. EMBO J. 17: 1919–1929.15. BENTING, J.H., A.G. RIETVELD & K. SIMONS. 1999. J. Cell Biol. 146: 313–320.16. BALCAROVA-STANDER, J., S.E. PFEIFFER, S.D. FULLE & K. SIMONS. 1984. EMBO J. 3:

2687–2694.17. BRANDLI, A.W., R.G. PARTON & K. SIMONS. 1990. J. Cell Biol. 111: 2909–2921.18. VERKADE, P., T. HARDER, F. LAFONT & K. SIMONS. 2000. J. Cell Biol. 148: 727–739.