Which Ras rides the raft?

To the editor

Biochemical and morphological analysis of plasma membranes has identified coherent lipid domains called rafts or caveolae that dynamically organize multiple membrane proteins. Ras, Ras activators and Ras effectors are found in caveolae/rafts, indicating that they may be sites of Ras action. An unresolved question of Ras function is the relative contribution of each of the three Ras isoforms (H, K and N) to the myriad cellular responses attributed to activated Ras. The importance of this question is highlighted by the specificity of activation of distinct Ras isoforms in various classes of human tumours, and by the dramatic difference each isoform contributes to mouse development.

Recent work by Hancock and colleagues^1,2 has generated the intriguing hypothesis that functional differences between Ras isoforms are due, at least in part, to localization in different plasma membrane compartments: H-Ras to caveolae/rafts and K-Ras to an unidentified, non-caveolar membrane domain. The primary difference between the Ras isoforms is their carboxy-terminal membrane targeting sequence. Both H- and K-Ras are prenylated but H-Ras is additionally modified by palmitoylation whereas K-Ras contains a basic, polylysine sequence.

Prior et al. suggest that the membrane anchors confer targeting specificity for distinct membrane domains¹. Despite the attractiveness of a multi-compartment model to explain how Ras isoforms can have different functions^3,4, it is important to consider a number of published observations that do not support this model. Prior et al.¹ and Roy et al.² used a pH 11 carbonate buffer to prepare caveolar membranes. However, when caveolae are prepared from sonicated plasma membranes using Optiprep gradients at physiological pH and ionic strength, both endogenous H-Ras and K-Ras are highly enriched in caveolae fractions⁵. Moreover, EGF stimulates recruitment to and activation of Raf exclusively in the caveolae fraction⁶. Disruption of caveolae/rafts by cholesterol depletion causes a loss of both K-Ras and H-Ras from caveolae fractions, and recruitment of Raf to caveolae is impaired⁵, indicating that both H- and K-Ras are in cholesterol-rich domains.

Native Ras has not been successfully immunolocalized in cells. Endogenous RhoA, however, a Ras superfamily member that is membrane anchored by a combination of a prenyl group and a polybasic domain similar to that found in K-Ras, has been successfully localized to caveolae by immunogold electron microscopy⁷.

Finally, there is compelling evidence that the entire Ras/Raf/Mek/Erk cascade is compartmentalized in caveolae/rafts and that Erk can be activated in purified caveolae fractions by PDGF^8,9. These findings are more consistent with a model where both H-Ras and K-Ras carry out critical signalling functions in caveolae/rafts. Methodology can have a dramatic impact on the nature of whatever system is under study. Clearly, observations of protein localization to plasma membrane compartments are particularly sensitive to the techniques used to make the observations. Further efforts are needed to understand how Ras is compartmentalized, and the consequences of this on Ras function.