One important issue is therefore to understand how transmembrane proteins become integrated into raft domains. We have started analysis of this issue by purifying raft transmembrane proteins and reconstituting them into proteoliposomes. The first protein that we studied was BACE, the ß-secretase that catalyzes cleavage of the membrane-spanning ß-amyloid precursor protein. Our cell biological studies indicated that a raft clustering process takes place in endosomes that facilitates BACE cleavage. We then showed that BACE activity reconstituted the activity in proteoliposomes was stimulated by the addition of raft lipids including cerebrosides and cholesterol.

Until recently there were no rapid assays to study how transmembrane proteins can associate with liquid-ordered rafts. Now with the plasma membrane spheres that we have developed we have an ideal system to analyse how proteins partition into raft domains. Our future aims will be to define how partitioning in and out of raft domains is regulated. We have started a collaboration with Ilpo Vattulainen and his modelling group at the University of Tampere. Finland. We would use the knowledge emerging from our experiments to establish realistic models for how proteins could become raftophilic.

To gain more information of the lipid composition of raft domains we plan to use enveloped viruses as tools for this purpose. Many viruses bud out from host cell plasma membrane several viruses such as influenza and HIV wrap a raft domain around them while others exit from the liquid-disordered bilayer. Viruses are easy to purify. Thus, by analyzing the lipidomics of different enveloped viruses (influenza, HIV, Semliki Forest virus and Vesicular stomatitis virus) growing in the same host or different host cells we gain access to how different raft and non-raft domains are composed. This research relies on methodology to analyze the diversity of lipid species in cell membranes with high precision and sensitivity. This is now becoming possible due to the development of quantitative mass spectrometry, including practically all lipid species. Andrej Shevchenko (MPI-CBG) has together with my group developed this technology so that we today can determine the lipidomes of yeast and mammalian membranes quantitatively. The mass spectrometry at MPI-CBG in Dresden is now a pioneering leader worldwide in this field.

The next step in these subprojects will be to identify the raft clustering principle. Both for MDCK cells and for yeast, it has been demonstrated that glycan side chains of raft cargo proteins play a role as sorting determinants. There are two possibilities, either clustering proteins such as lectins function as nucleating agents for specific raft coalescence or then domain formation is driven by increased concentration of raft lipids and cis interactions of glycan chains promoting phase separation in the TGN. As an important step to elucidate the sorting mechanism we have recently completed an EMAP screen for the post- Golgi pathways in yeast with the lab of Nevan Krogan at USCF. This screen will give us a catalogue of almost all possible players in the raft pathway that we want to dissect. Ultimately we want to reconstitute domain-induced budding in vitro.

What remains then is to get an overall view on how to construct a raft phase in cells. Epithelial cells form collectives, in which the cells are connected by junctions and each cell is polarized into an apical and basolateral plasma membrane domain. So far lipid analyses have not been carried out for any tissue culture epithelial cell line. We have recently shown that the apical membrane behaves like a large connected raft membrane at 25°C. Raft-associated proteins were found to diffuse freely over the apical membrane while non-raft protein demonstrated confined diffusion. The interpretation that best fits the data is that the apical membrane is a phase-separated system with a continuous percolating raft phase at 25°C, in which isolated domains of the non-raft phase are dispersed. The basolateral membrane is assumed to behave like most other plasma membranes with fluctuating small and dynamic rafts.
We have now worked out methods to isolate both apical and basolateral plasma membrane domains. The lipidomes of the isolated plasma membrane domains from MDCK cells are being determined. Our studies revealed that the Forssman glycolipid with a pentasaccharide head group is a major component of the apical membrane in MDCK cells. Interestingly, our data demonstrate that the level of the Forssman glycolipid is almost nil in the unpolarized state and increases with polarization. In contrast MDCK cells undergoing an epithelial–mesenchymal transition (in collaboration with Hartmut Beug IMP) the reverse is true: Forssman levels go down. This subproject capitalizes on our unique ability to analyze mammalian lipidomes comprehensively and quantitatively and would illustrate the compositional principles involved in construction of a biologal raft membrane phase.

Lipid rafts play a central role in many disease processes. We have been heavily involved in one potential raft disease: Alzheimer’s disease. A key molecule in the pathogenesis of Alzheimer’s disease is the amyloid ß-peptide (Aß). Aß is liberated from the membrane-spanning ß-amyloid precursor protein (APP) by sequential proteolytic processing employing ß- and γ-secretases. Recent work shows that ß-secretase cleavage of APP occurs predominatly in endosomes and that endocytosis of APP and ß-secretase is essential for ß-cleavage and Aß production. We have obtained evidence for a raft clustering process that facilitates BACE cleavage in the endosomes. Owing to its critical role in ß-amyloid formation, ß-secretase lends itself as a therapeutic target for Alzheimer’s disease. Current inhibitor design focuses largely on active site binding but neglects the issue of the sub-cellular localization where the enzyme is active. We have addressed this issue by synthesizing a membrane-anchored version of a transition-state inhibitor by linking it to a sterol and other lipophilic moieties. Thus we not only target the inhibitor to active ß-secretase found in endosomes, but also reduce the dimensionality of the inhibitor thereby increasing its local membrane concentration. This inhibitor reduced enzyme activity much more efficiently than the free inhibitor in cultured cells. Assaying for cellular ß-secretase activity, we found that the raftophilicity of the anchor correlated with potency. These results suggest that raftophilic anchors of ß-secretase inhibitors enhance their inhibitory potential. To determine whether sterol-linked inhibitors were also effective in vivo, we used mice expressing an APP Swedish mutant. The data showed that the sterol-linked inhibitor directly injected into the brain was effective in inhibiting Aß production. We are continuing this work to find out how we can cross the blood brain barrier in animal experiments.