Fig. 1. Analysis of FRET on the cell membrane demonstrates the distinct levels of co-clustering of microdomain markers. (A) Illustration of our reductionistic approach to study activation-state-dependent microdomain localization of heterotrimeric G proteins. (Top) Activation followed by dissociation of the heterotrimer leads to a separation of G and G subunits. (Bottom) Only the lipid anchors of inactive heterotrimer and active G were fused to fluorescent proteins, and the microdomain localization of these G-protein-anchor constructs was studied in relation to microdomain markers (see B and C) using FRET between mCFP as a donor and mCit as an acceptor. (B) Schematic representation of microdomain markers with their lipid anchors. P, palmitoyl; G, geranylgeranyl; F, farnesyl; pb, polybasic sequence. Source refers to the proteins from which targeting sequences were derived. (C) Subcellular localization of microdomain markers imaged by confocal microscopy. All constructs were predominantly localized to the plasma membrane, with minor labeling of internal membranes or, in the case of the Rac1-derived construct, nuclear labeling. Only the mCit constructs are representatively shown. Bars, 10 µm. (D) By plotting the FRET efficiency (E) against the normalized acceptor surface concentration (cA) at a constant donor mole fraction (x_D=0.50±0.17), we obtained information about the clustering of donor and acceptor fluorophores. A random distribution of fluorophores (Wolber and Hudson, 1979) cannot describe our data (left, hashed curve). We found that the FRET efficiencies increased after a cA offset towards a plateau value E_max (indicated by solid horizontal line in the left plot). We therefore adapted the double exponential function of Wolber and Hudson (Wolber and Hudson, 1979), further taking the cA offset and the maximum efficiency E_max into account. This lead to fits which described all of our FRET data adequately (from left to right, ²: 22.6, 25.3 and 9.7). Each datapoint was calculated on a single cell. Representative examples of indicated FRET pairs are shown. (E) The E_max matrix of microdomain-marker FRET pairs. FRET values of mCFP-tH/mCit-tH and mCFP-tK/mCit-tK are significantly higher (bold) than those of tH-polybasic pairs (P<0.001 or P<0.05, respectively, 2-tailed Student's t-test). No significant differences were found for consistently high mCFP-polybasic/mCit-polybasic sequence pairs (bold). E_max values are given in percent ± s.d.; n, number of independent experiments.