The spatial organization of signalling proteins in the cell membrane is often ascribed to lipid-based ‘rafts’. But single-molecule tracking reveals that such organization probably arises by protein–protein interactions.
Signal transduction — the relay of signals from outside a cell to inside — frequently involves bewildering patterns of interactions between several different types of protein at the cell surface. Work attempting to make sense of this complexity suggests that specific lateral organization of the interacting proteins in the membrane is key to their signalling functions. But how is such organization generated? Work by Douglass and Vale, published in Cell1, begins to provide some answers, and emphasizes the utility of recently developed single-molecule imaging techniques in addressing the dynamic properties of signalling networks. Moreover, the authors' experiments directly address the contro- versy over the mechanisms that generate localized variations in the composition of the cell membrane.
Simple mixtures of lipids in artificial membrane bilayers can segregate into regions that differ in the way the acyl chains of the lipids are packed together. This can spontaneously generate heterogeneity in the membrane, as lipids that prefer different local environments tend to separate out from one another. More-ordered acyl-chain packing is associated with the presence of increasing amounts of cholesterol and sphingolipids — both found in natural membranes — and the resulting lipid domains tend not to be soluble in non-ionic detergents2. Detergent-insoluble fractions enriched for particular proteins and lipids can also be isolated from cells. Extrapolating from the artificial membrane data, it has been proposed that these detergent-resistant fractions might be derived from functional domains, or ‘lipid rafts’, in cell membranes, where self-organization of the membrane lipids leads to the recruitment of specific proteins3. This lipid raft hypothesis has received much attention and is certainly appealing, but the correlation between detergent resistance and domain formation in vivo is a topic of some debate4. Artificial membranes may not be good models for cell membranes that are rich in protein and have two asymmetric layers of hundreds of different types of lipid.
The case of signalling through T-cell receptors is particularly germane to this debate. At the start of an immune response, T cells are activated when antigen molecules bind to the receptors on their surface. Stimulation of T cells usually occurs when stable contacts — referred to as ‘synapses’ — form between the T cell and so-called antigen-presenting cells5. There is a striking degree of spatial organization within the synapse; for example, molecules involved in the adhesion of the interacting cells, and activators and inhibitors of the signalling cascade, segregate and take on highly specific patterns5,6. Moreover, because several of the proteins recruited to the T-cell synapse are found in detergent-resistant membrane fractions, various functions have been ascribed to lipid rafts in organizing these proteins during T-cell-receptor signalling7. This instance of a signalling machine in which the appropriate spatial distribution of its components is likely to be central to its function is a promising place to look for direct evidence of a physiological role for lipid rafts.
Douglass and Vale1 adopted a widely used method for stimulating T cells — cross-linking the receptors using specific antibodies. But they modified it so as to be able to view the stimulated surface of the cell using either conventional confocal microscopy or total internal reflection imaging; the latter is a highly sensitive approach that can detect single fluorescent molecules in a narrow focal plane (Fig. 1). This remarkable technical achievement allowed individual fluorescent proteins involved in T-cell-receptor signalling to be tracked directly. Several of these proteins clustered together, for example the stimulatory co-receptor CD2, the adaptor protein LAT and the enzyme Lck; the negative regulator CD45 did not occur in the clusters.
LAT seems to play a prominent role in generating clustering, as the proteins do not group together in cell lines lacking LAT. By using a laser to photo-bleach a spot on the membrane, the researchers followed fluorescent molecules as they moved into the bleached area. This showed that there was a constant exchange of individual protein molecules in and out of clusters that were themselves stable over time. Finally, ingenious experiments tracking individual LAT, Lck and CD45 molecules relative to clusters containing CD2 revealed that these clusters can temporarily trap freely diffusing LAT and Lck, but not the negative regulator CD45.
Douglass and Vale's system provided an excellent opportunity to unravel the mechanisms that generate the spatial organization of signalling molecules (Fig. 1). A mutant of LAT, which lacks the acylated amino-acid residues required for incorporating the protein into lipid rafts8, showed identical clustering and diffusional trapping to normal LAT. Conversely, a LAT mutant lacking residues required for specific protein–protein interactions did not cluster or show diffusional trapping. Similar results were obtained with a truncated form of Lck that retains only the portions required for membrane association and raft incorporation. This mutant also did not cluster or show diffusional trapping.
These data do not support a link between raft incorporation (detergent resistance) and spatial organization during T-cell-receptor signalling. The authors propose instead that the clustering and diffusional trapping are best explained by a network of protein–protein interactions between the relevant signalling molecules. Although such a network remains to be characterized directly, previous studies corroborate the primary role of protein–protein as opposed to protein–lipid interactions in T-cell-receptor signalling9. They also do not support the proposed link between raft incorporation and protein recruitment to membrane regions where the T-cell receptor has been activated10.
Now that our picture of protein dynamics during spatially organized signalling is beginning to reach single-molecule resolution, there seems no need to invoke the raft model to explain these dynamics. However, before summarily wielding Occam's razor, we should remember that model systems such as that used by Douglass and Vale do not fully replicate the properties of the T-cell synapse in vivo, and further subtleties doubtless remain to be investigated.
Douglass, A. D. & Vale, R. D. Cell 121, 937–950 (2005).
London, E. & Brown, D. A. Biochim. Biophys. Acta 1508, 182–195 (2000).
Simons, K. & Ikonen, E. Nature 387, 569–572 (1997).
Munro, S. Cell 115, 377–388 (2003).
Davis, D. M. & Dustin, M. L. Trends Immunol. 25, 323–327 (2004).
Kupfer, A. & Kupfer, H. Semin. Immunol. 15, 295–300 (2003).
Harder, T. Curr. Opin. Immunol. 16, 353–359 (2004).
Zhang, W., Trible, R. P. & Samelson, L. E. Immunity 9, 239–246 (1998).
Hartgroves, L. C. et al. J. Biol. Chem. 278 20389–20394 (2003).
Glebov, O. O. & Nichols, B. J. Nature Cell Biol. 6, 238–243 (2004).
About this article
Coupled Membrane Lipid Miscibility and Phosphotyrosine-Driven Protein Condensation Phase Transitions
Biophysical Journal (2020)
Novel alkyl chain-based fluorescent probes with large Stokes shifts used for imaging the cell membrane and mitochondria in different living cell lines
RSC Advances (2017)
Two-photon fluorescent probe for detecting cell membranal liquid-ordered phase by an aggregate fluorescence method
Journal of Materials Chemistry B (2017)
Biochimica et Biophysica Acta (BBA) - Molecular Cell Research (2015)
A combined fluorescence spectroscopy, confocal and 2-photon microscopy approach to re-evaluate the properties of sphingolipid domains
Biochimica et Biophysica Acta (BBA) - Biomembranes (2013)