Reply

We have used multiple techniques to address the issue of Ras localization and shown for the first time that H- and K-ras have different distributions across caveolae, raft and non-raft domains (as defined in ref. 10). Anderson and White, however, argue for the exclusive caveolar localization of Ras proteins based solely on their biochemical studies showing that H-Ras and K-Ras cofractionate in low-density membranes in what they refer to as a caveolae fraction. Many of their comments follow from the inadequate resolution of their fractionation procedure. Given that they find that two proteins cofractionate and yet we can resolve them (by whatever method), it is self-evident that they have poor separation. Also how pure is the Anderson lab's caveolae fraction? It is important to note that recent work11 questioned their caveolar-localization of EGF receptors. Careful inspection of the Optiprep method that they used6 raises additional questions. Although there is overlap between caveolin and H-Ras on their initial gradient, the distributions of the two proteins are definitely not concordant and look very similar to the relative distributions we observed1. Second, they discard a substantial number of Ras- and Raf-containing fractions and disordered membrane fractions before isolating a caveolae fraction. These technical deficiencies would make it impossible to observe the differential Ras localizations shown in Prior et al.1 because so much of the plasma membrane is excluded from analysis. Although we emphasize the role of rafts in H-Ras localization and function, we did find a small proportion of K-Ras cofractionating with raft markers1. The significance of this is unknown.

Importantly, we correlated the raft/non-raft association of Ras, determined biochemically, with a detailed electron microscopy and functional analysis1. Our data are completely concordant with respect to: the different plasma membrane localizations of H-Ras and K-Ras (including endogenous H-Ras and K-Ras); the effect of GTP loading on H-Ras lateral segregation; and functional studies showing that H-Ras but not K-Ras signalling is cholesterol dependent2. The comments on Rho proteins emphasize another tenet of our paper: protein sequences adjacent to the membrane anchor have a profound influence over lipid raft association. These sequences are non-conserved between Rho and Ras so extrapolating from one protein to another is presently impossible. In addition, although certain components of the MAPK cascade have been found in caveolae, only our study provides quantitative electron microscopy data on the localization of Ras proteins, an essential technique for assessing caveolar localization.