GPI (glycosylphosphatidylinositol)-anchored proteins are characteristic components of biochemically defined lipid rafts. Rafts may be involved in T-cell stimulation, but it is not clear whether molecules involved in TCR (T-cell receptor) signalling are partitioned to T-cell synapses through raft microdomains or through specific protein–protein interactions. We have used FRET (fluorescence resonance energy transfer) analysis to study the distribution of GPI-anchored fluorescent proteins in the plasma membrane of live cells. Multiple criteria suggested that FRET between different GPI-anchored fluorescent proteins in COS-7 or unstimulated Jurkat T-cells is generated by a random, unclustered distribution. Stimulation of TCR signalling in Jurkat T-cells by beads coated with antibodies against TCR subunits resulted in localized increases in fluorescence of raft markers. However, measurements of FRET and ratio imaging showed that there was no detectable clustering and no overall enrichment of raft markers in these regions.
Results and discussion
Lipid rafts, microdomains enriched in cholesterol and sphingolipids, are supposed to be playing a role in many cellular processes by forming ‘organizing platforms’ for proteins . However, most of the evidence for existence and function of lipid rafts comes from indirect methods. Recent studies involving single-particle tracking, FRET (fluorescence resonance energy transfer) and other approaches have yielded contradictory results, pointing at possible methodological problems .
GPI-APs (glycosylphosphatidylinositol-anchored proteins) are well-characterized markers for the DRM (detergent-resistant membrane) fraction of the plasma membrane, which is commonly equated with rafts [1,3]. The studies of their cellular distribution might shed light on the fundamental properties of rafts. A convenient approach for detecting nanoscale distribution in vivo employs FRET, non-radiative transmission of fluorescence energy from a donor fluorophore to an acceptor that occurs at distances <12 nm.
We have studied the distribution of GPI-APs in live COS-7 and Jurkat T-cells using an assay for FRET between two spectral variants of GFP (green fluorescent protein): CFP (cyan fluorescent protein) and citrine fluorescent protein (CitFP) as donor and acceptor respectively . GPI-anchored monomeric CFP (mCFP–GPI) and CitFP (mCitFP–GPI) were expressed in COS-7 and Jurkat T-cells and FRET ratio  between them was measured. The dependency of FRET ratio on surface density of the fluorophores reflects the degree of clustering [6,7]. The efficiency of FRET between mCFP–GPI and mCitFP–GPI was linearly dependent on mean fluorescence intensity (proportional to the surface density of the proteins), indicating that mCFP–GPI and mCitFP–GPI are most probably randomly distributed (Figure 1A). The data were confirmed by direct measurement of FRET efficiency using an acceptor photobleaching-based approach .
Random distribution of GPI-FPs in COS-7 cells
Treatment with 10 mM β-methylcyclodextrin, which solubilizes membrane cholesterol and disrupts lipid rafts [1,2], had no effect on FRET between GPI-mFPs (Figure 1B). Also, FRET between GPI-mFPs was compared with FRET between GPI-mCFP and LmCitFPGT46, a chimaeric transmembrane protein that is excluded from DRMs , and was found to be identical, indicating that the GPI-mFPs and LmCitFPGT46 have equivalent distributions (Figure 1B). As a positive control, treatment with monoclonal antibodies against GFP conferred a significant effect on the FRET signal, reflecting the formation of GPI-mFP clusters too small to be seen by light microscopy (Figures 1C and 1D). Altogether, the above results strongly suggest that FRET between mCFP–GPI and mCitFP–GPI reflects a random, unclustered distribution in the plasma membrane.
Next, we addressed a situation where partitioning into biochemically defined lipid rafts would be deemed functionally significant, and found one in the case of TCR (T-cell receptor) activation . Localized activation of TCR in vitro was induced by exposure of Jurkat T-cells to beads coated with antibodies against TCR subunits and was confirmed by recruitment of the soluble kinase Zap70 . TCR activation in cells expressing GPI-mFPs frequently resulted in accumulation of fluorescence at the contact region (Figure 2). This is consistent with a local increase in surface density of raft components, but could also be explained by convolution of the plasma membrane generating an apparent increase in fluorescence. Our results favoured the latter possibility. The region of TCR activation did not manifest higher FRET when compared with the rest of the plasma membrane, indicative of the average distance between the fluorophores remaining the same (Figure 2A). Moreover, in some cells, apparent membrane convolution was visible (Figure 2C). As a positive control, LAT (linker for activation of T-cells) protein was used, as it has been shown to be biochemically recruited to regions of TCR activation through specific protein–protein interactions . Ratio imaging of mCFP–GPI and LAT–mCitFP showed significant enrichment of LAT in regions of active TCR signalling (Figure 2B).
FRET ratio imaging in activated Jurkat T-cells
Our results imply that different GPI-linked proteins are not clustered together in lipid rafts, undermining the notion that lipid rafts might function as organizing platforms. This conclusion is in good agreement with several other studies based around various assays, including FRET , measurement of rates of diffusion of different raft markers  and immunoelectron microscopy . On the other hand, a recent study in Chinese-hamster ovary cells suggested that a small proportion of total GPI-APs is present in clusters of at most four GPI-APs . The consensus that emerges is that GPI anchor is not sufficient to induce significant clustering of different proteins within small, compact, lipid rafts, and DRM incorporation is then not necessarily a reflection of clustering in the plasma membrane. Our observations are consistent with the view that depicts biochemically defined lipid rafts as shells of raft lipids surrounding individual raft proteins . Also, partitioning into biochemically defined lipid rafts does not correlate with recruitment to regions of TCR activation, and apparent increase in fluorescence of raft markers is most probably due to complex membrane topology and hence is a poor indicator of fluorophore surface density.
Lipids, Rafts and Traffic: A Focus Topic at BioScience2004, held at SECC Glasgow, U.K., 18–22 July 2004. Edited by G. Banting (Bristol, U.K.), N. Bulleid (Manchester, U.K.), C. Connolly (Dundee, U.K.), S. High (Manchester, U.K.) and K. Okkenhaug (Babraham Institute, Cambridge, U.K.)