Dynamic confinement of NK2 receptors in the plasma membrane. Improved FRAP analysis and biological relevance.

A functional fluorescent neurokinin NK2 receptor, EGFP-NK2, was previously used to follow, by fluorescence resonance energy transfer measurements in living cells, the binding of its fluorescently labeled agonist, bodipy-neurokinin A (NKA). Local agonist application suggested that the activation and desensitization of the NK2 receptors were compartmentalized at the level of the plasma membrane. In this study, fluorescence recovery after photobleaching experiments are carried out at variable observation radius (vrFRAP) to probe EGFP-NK2 receptor mobility and confinement. Experiments are carried out at 20 degrees C to maintain the number of receptors constant at the cell surface during recordings. In the absence of agonist, 35% EGFP-NK2 receptors diffuse within domains of 420 +/- 80 nm in radius with the remaining 65% of receptors able to diffuse with a long range lateral diffusion coefficient between the domains. When cells are incubated with a saturating concentration of NKA, 30% EGFP-NK2 receptors become immobilized in small domains characterized by a radius equal to 170 +/- 50 nm. Biochemical experiments show that the confinement of EGFP-NK2 receptor is not due to its association with rafts at any given time. Colocalization of the receptor with beta-arrestin and transferrin supports that the small domains, containing 30% of activated EGFP-NK2, correspond to clathrin-coated pre-pits. The similar amount of confined EGFP-NK2 receptors found before and after activation (30-35%) is discussed in term of putative transient interactions of the receptors with preexisting scaffolds of signaling molecules.

A functional fluorescent neurokinin NK2 receptor, EGFP-NK2, was previously used to follow, by fluorescence resonance energy transfer measurements in living cells, the binding of its fluorescently labeled agonist, bodipy-neurokinin A (NKA). Local agonist application suggested that the activation and desensitization of the NK2 receptors were compartmentalized at the level of the plasma membrane. In this study, fluorescence recovery after photobleaching experiments are carried out at variable observation radius (vrFRAP) to probe EGFP-NK2 receptor mobility and confinement. Experiments are carried out at 20°C to maintain the number of receptors constant at the cell surface during recordings. In the absence of agonist, 35% EGFP-NK2 receptors diffuse within domains of 420 ؎ 80 nm in radius with the remaining 65% of receptors able to diffuse with a long range lateral diffusion coefficient between the domains. When cells are incubated with a saturating concentration of NKA, 30% EGFP-NK2 receptors become immobilized in small domains characterized by a radius equal to 170 ؎ 50 nm. Biochemical experiments show that the confinement of EGFP-NK2 receptor is not due to its association with rafts at any given time. Colocalization of the receptor with ␤-arrestin and transferrin supports that the small domains, containing 30% of activated EGFP-NK2, correspond to clathrin-coated pre-pits. The similar amount of confined EGFP-NK2 receptors found before and after activation (30 -35%) is discussed in term of putative transient interactions of the receptors with preexisting scaffolds of signaling molecules.
The general mechanisms underlying the activation of Gprotein-coupled receptors (GPCRs) 1 and the following attenuation of the cellular response seem well established at the molecular level. The prevailing view is that the intracellular signal results from a cascade of transient and sequential molecular interactions starting with the activation of the heterotrimeric G-protein upon ligand/receptor interaction. The biological effect then can be attenuated by three main mechanisms: (i) agonists removal; (ii) agonist-mediated desensitization of the receptor molecules by uncoupling of the activated receptors from heterotrimeric G-proteins; and (iii) agonist-stimulated receptor endocytosis followed by either recycling or degradation. The remaining key questions are to understand the architecture governing the molecular interactions of the signaling cascade and how these multiple interactions take place in space and time, notably starting with the lateral organization of the partners at the level of the plasma membrane (1)(2)(3)(4). Here we investigate these different aspects using the neurokinin NK2 receptor expressed in HEK293 cells as a model system.
Neurokinins form a family of neuropeptides acting at three distinct GPCRs, namely NK1, NK2 and NK3. The NK2 receptor shows preferential binding for the endogenous neurokinin A (NKA) (5) and is mostly found in peripheral nervous tissues where it participates mainly to smooth muscle contraction. In previous studies, a fluorescently modified receptor labeled with enhanced green fluorescent protein (EGFP) on the amino-terminal extracellular part (EGFP-NK2) was developed and expressed in a model cell system, the HEK293 cells. The binding properties of several fluorescent derivatives of NKA and their associated responses were studied using fluorescence resonance energy transfer (6 -8). In Vollmer et al. (6), the ligand binding to the receptor was measured on single cells after local application of fluorescent NKA. The observation, relevant to the present work, was that activation and desensitization of the response were compartmentalized at the level of the plasma membrane. We now investigate the possible dynamic confinement of the EGFP-NK2 receptors in the cell plasma membrane using fluorescence recovery after photobleaching at variable observation radius (vrFRAP).
FRAP was used extensively in the 1980s and 1990s (9 -11) to measure lateral diffusion of molecules in biological and model membranes. FRAP gathers information on the collective behavior of molecule diffusion. Typically, around 25,000 fluorescently labeled proteins are observed over an area of 4 -10 m 2 /cell and the process is repeated at least 30 -40 times on different cells.
An improved version of this technique is FRAP at variable observation radius. This approach aiming to analyze dynamic confinements is based on the assumption that the mobile fraction of a fluorescent molecule will depend on the radius of the area bleached by illumination. In a previous study, we validated this assumption first numerically with Monte Carlo simulations of FRAP experiments at variable beam radius per-formed on 150 ϫ 150 site triangular lattice (representing a membrane). These lattices were obstructed by fences delineating hexagonal closed areas of different sizes characterized by the radius of a circular disk of equivalent areas (12). Once the conceptual approach was validated, we set up vrFRAP in an experimental model of compartmentalized membranes consisting of a monolayer of apposed spherical silica bead-supported phospholipid bilayers of known radii (12). This technique has been successfully applied to biological specimens (12)(13)(14) indicating membrane organization in domains with a radius of ϳ250 nm, consistent with data obtained by single particle tracking experiments (4,15,16).
In this study, we used vrFRAP to investigate dynamic lateral confinement of the EGFP-NK2 receptor in the plasma membrane of stably transfected HEK293 cells. We then combined biochemical analysis and molecular co-localization approaches to get insight into the nature of the domains detected by vrFRAP.
Cell Culture and Transfection-Stable HEK293 cell lines overexpressing NK2 receptors or NK2 receptors fused at their amino-terminal part to the EGFP fluorescent protein were grown plated in minimal essential medium supplemented with 10% fetal calf serum and antibiotics (Invitrogen). These transfected cells were selected with 100 g/ml hygromycin B (Invitrogen) for 1 week/month (6). Stably transfected EYFP-GPI-anchored protein-expressing cells and EYFP-CD46-expressing cells were obtained upon transfection of HEK293 cells by calcium chloride methodology and 600 g/ml G418 selection (Invitrogen) with a pEYFP-GPI-N1 construct (17) or a pEYFP-CD46 construct (18), respectively, both cordially given by Dr. Patrick Keller.
HEK293 cells expressing or not expressing the non-fluorescent NK2 receptors were grown on a coverslip (22 ϫ 22 mm). The coverslips were placed on ice (4°C) for 10 min in culture medium and washed once with PBS Ca 2ϩ -Mg 2ϩ before labeling for 10 min at 4°C in PBS Ca 2ϩ -Mg 2ϩ supplemented with the C 6 -NBD-PC and DOPC lipids (19). They were then washed twice with PBS Ca 2ϩ -Mg 2ϩ before carrying FRAP experiments.
Inhibition of Protein Synthesis-Cells plated on a coverslip were washed twice with PBS Ca 2ϩ -Mg 2ϩ , incubated for 1 h at 37°C with 150 l of 10 g/ml cycloheximide (Sigma) in PBS Ca 2ϩ -Mg 2ϩ (20), and washed twice with PBS Ca 2ϩ -Mg 2ϩ .
Experimental Conditions for FRAP Experiments on Living Cells-FRAP experimental conditions must be well established to avoid the generation of artifactual immobile fractions and erroneous mobile fractions. (i) The surface of the illuminated area has to be small with respect to the cell surface (condition of infinite probe reservoir, value of cell surface Ͼ5ϫ the illuminated area) (21). HEK293 cells have a mean diameter of 15 m. Therefore, the maximal observation radius for this study was 3.45 m (21). (ii) The recovery half-time ( ϭ R 2 /4D, where R is radius and D is the diffusion coefficient) of the experiments must be calculated for each probe and for each radius of the illuminated area. The duration of the photobleaching phase must be Ͻ5% the fluorescence recovery half-time (21).
FRAP Experiments and Analysis of Fluorescence Recovery Data-The coverslip on which the cells were plated was washed with PBS Ca 2ϩ -Mg 2ϩ and covered with another coverslip (24 ϫ 36 mm) on a homemade stainless steel slide. The FRAP experiments were achieved on the apical pole of the cell where ligands have ready access to apical NK2 receptors.
Each preparation of cells was explored for 30 min only to circumvent potential problems due to changes in the viability and fluorescent labeling of cells. The temperature was systematically maintained at 20°C.
Experiments were carried out in "uniform disk illumination" conditions using an apparatus based on a Leitz Ortholux II fluorescence microscope (21). The expanded beam of the Argon laser ( ex ϭ 488 nm) was focused on a sample through a set of diaphragms (from 50 to 165 m) and through a Leitz ϫ63/1.3 oil immersion objective, making it possible to vary the radius of the observation area from 1.05 to 3.45 m. The bleaching rates of ϳ40% were obtained for a bleaching time of 50 -75 ms. Bleaching time was varied according to the probe studied to ensure consistency with the lipid or protein recovery half-time reported for the cellular plasma membrane (21). Ͼ30 fluorescence recoveries were accumulated for each experimental condition tested on various cells. All of the experiments were carried out at least in triplicate.
The diffusion coefficient (D) and the mobile fraction (M) were obtained by fitting diffusion equations to the experimental recovery curves as previously described (21). A minimization algorithm coupled to statistical analysis of the data was used to provide from a set of accumulated recovery curves the solution values for D and M and their maximum (D max and M max ) and minimum (D min and M min ) values for a confidence level of 95% (21). In addition, the analysis of the recovery curves for one or two diffusive populations was achieved according to the best fit criteria.
Analysis of vrFRAP Experiments-In our previous study, the mobile fractions (M) were observed to increase linearly with the increasing size radius (r) of the domain obstructed by hexagonal fences and corresponded to a unique master line leading to the relationship shown in Equation 1.
For closed areas, this line intercepts the mobile fraction axis at a value M Ϸ 0. D has to be recalculated according to Equation 2 (12).
When hexagonal fences of the diffusion areas were slightly open, M was found to be the sum of two contributions: (i) one that is an M value dependent on R and related to the fluorescent tracer remaining confined in the domain during the observation time and (ii) another permanent one, Mp, that is independent of R. Mp increases with increasing permeability of the fences, which delineate the hexagonal "closed areas." For the opening of 20.4, 10.2, 6.1, 4.1, and 0% of the diffusion area perimeter, the corresponding values of Mp are 0.75, 0.55, 0.45, 0.33, and 0 (12).
Thus, for membranes composed of slightly open domains, the fluorescence recovery curves can be fitted with two diffusion coefficients: (i) one that is independent of R and corresponding to particles that diffuse over long distances and (ii) another one that shows a quadratic dependence of D app on R and corresponding to the diffusion of particles inside the domains. The corresponding true diffusion coefficient value D real is calculated following Equation 2 (12).
As a control, it is important to note that, in egg-PC multilayers labeled with 1% fluorescent probe, a model membrane known to be laterally homogeneous (21), M and D are invariant with R (data not shown).
Activation of NK2 Receptors with Neurokinin A-A stock solution of 10 mM NKA in dimethylformamide was diluted in PBS Ca 2ϩ -Mg 2ϩ to give a final concentration of 1 M. HEK293 cells expressing or not expressing the non-fluorescent NK2 receptors and labeled with C 6 -NBD-PC or HEK293 cells overexpressing fluorescent EGFP-NK2 receptor were incubated on a coverslip (22 ϫ 22 mm) with 150 l of 1 M NKA. Cells were kept in contact with agonist solution at 20°C for the 30 min required to carry out FRAP experiments on one preparation. It was calculated that, in these experimental conditions (K i ϭ 64 Ϯ 25 nM) (6), Ͼ95% receptor binding sites were occupied.
Analysis of EGFP-NK2 Receptor Association with Lipid Microdomains Insoluble in Cold Triton X-100 -Three types of stable cell lines derived from HEK293 cells were tested: EGFP-NK2 receptors expressing cells; EYFP-GPI-anchored protein-expressing cells; and EYFP-CD46-expressing cells. 10 million cells for each sample were collected upon PBS, 5 mM EDTA treatment after 2 days of culture in 10-cm plates. EGFP-NK2 receptors expressing cells were incubated at 37 or 20°C in Eppendorf tubes with 1 ml of HEPES-BSA buffer (137.5 mM NaCl, 1.25 mM MgCl 2 , 1.25 mM CaCl 2 , 6 mM KCl, 5.6 mM glucose, 10 mM HEPES, 0.4 mM NaH 2 PO 4 , 1% bovine serum albumin (w/v), pH 7.4) for 5, 15, or 30 min in the presence or in the absence of 1 or 10 M NKA. The three types of cells were then evenly treated. Cells were briefly centrifuged at 4°C and suspended in 400 l of ice-cold TNE buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA) supplemented with 1% Triton X-100 and one tablet of Complete EDTA-free protease inhibitor (for 10 ml of buffer; Roche Applied Science). The samples were rotated in the cold room on a wheel for 1 h to allow detergent solubilization. The floatation of the lipid-enriched detergent-insoluble material was achieved in OptiPrep TM sucrose gradient (22). Seven fractions were collected from each gradient with fraction 1 having the lightest density. To analyze the content in fluorescently modified fusion proteins (either EGFP-NK2 receptors or EYFP-GPI or EYFP-CD46), each fraction of the density gradient was submitted to immunoprecipitation with polyclonal anti-GFP antibodies and the immunoprecipitated material was separated on polyacrylamide gels, blotted, and revealed with monoclonal anti-GFP antibodies (8). To analyze the content of each fraction of the density gradient in endogenous caveolin-1, each supernatant of the anti-GFP immunoprecipitation was trichloroacetic acid-precipitated, separated on polyacrylamide gels, blotted, and revealed with polyclonal anti-caveolin-1 (Santa Cruz Biotechnology) (22).
Texas Red-Transferrin Uptake of EGFP-NK2 Receptors Expressing Cells-EGFP-NK2 receptors expressing cells were split and grown for 2 days in 24-well plates on 12-mm glass coverslips coated with rat type I collagen. On the day of experiment, cells were starved by a 1-h incubation in serum-free medium. The cells were then incubated for periods ranging from 0 to 45 min in HEPES-BSA buffer supplemented with protease inhibitors (40 g/ml bestatin and bacitracin, 20 g/ml phosphoramidon, 50 g/ml chymostatin, 1 g/ml leupeptin) containing 20 g/ml Texas Red-transferrin and 1 M NKA at 37 or at 20°C. Transferrin uptake was stopped by placing cells on ice and washing them immediately with ice-cold HEPES-BSA buffer. The cells were then fixed in 4% paraformaldehyde in PBS for 15 min at 4°C and then incubated for 15 min in NH 4 Cl 50 mM. Coverslips were mounted onto microscope slides using an anti-fading agent, Mowiol (Calbiochem).
Cloning of ␤-Arrestin2 and Insertion into pCEP4 Expression Vector 5Ј to the EYFP Gene-␤-Arrestin2 was cloned by PCR on a -phage cDNA library made from HEK293 cells RNA as described below. A fraction (30 l) of the HEK293 cell cDNA library was denatured by boiling for 5 min and incorporated into a PCR reaction mixture using ␤-arrestin2 internal oligonucleotides to check for the presence of ␤-arrestin2. This finding was confirmed, and ␤-arrestin2-flanking oligonucleotides including a Kozak consensus sequence and sequences for flanking endonuclease sites (NotI and BamHI) for insertion into the pCEP4 expression vector were synthesized and included in a new PCR reaction on HEK293 cells cDNA. The resulting fragment was cloned into the pCEP4 vector and sequenced. From this construct, a ␤-arrestin2 NotI-NheI gene fragment was isolated and inserted into NotI-BamHI-digested pCEP4 5Ј to a NheI-BamHI EYFP fragment in a double cloning experiment.
Transient Transfection of Non-fluorescent NK2 Receptors Expressing Cells with ␤-Arrestin2-EYFP Construct-Non-fluorescent NK2 receptors expressing cells were transfected 2 days after plating on 10-cm plates by the calcium chloride technique with 0.8 g of ␤-arrestin2-EYFP construct. Medium was replaced the following day. On the second day after transfection, cells were plated on 12-mm glass coverslips coated with rat type I collagen and grown for an additional 2 days. On the day of experiment, cells were incubated for periods ranging from 0 to 45 min in HEPES-BSA buffer supplemented with protease inhibitors and with 1 M Texas Red-NKA at 37 or at 20°C. Fixation was followed as described for the transferrin uptake experiments with the exception that fixation was performed at 20°C.
Confocal Analysis-Cells were observed with a laser scanning microscope with upright stand SP1-UV type (Leica) using a PLApo 63 ϫ 1.3 numeric aperture oil immersion objective (Leica) at the Centre d'Imagerie of the IGBMC Institute in Strasbourg. Between 7 and 13 stacks (depending on the cell depth) of two-dimensional images were acquired sequentially in the "green" or "red" channels. Micrograph images in figures are taken from the middle of the cells.
Internalization of EGFP-NK2 Receptors Measured Using Anti-GFP Antibodies-Cells expressing EGFP-NK2 receptors were suspended in 12 ml of HEPES-BSA buffer at a concentration of 10 6 cells/ml. The cell suspension was separated into 2 ϫ 50-ml Falcon tubes with a magnetic stirrer and incubated at 37°C. At time point 0, 1 M NKA was added to one of the tubes. At time points 0, 5, 15, 30, and 45 min, 1 ml of the cell suspensions were collected from both tubes and transferred to a tube containing 10 l of sodium azide (10%) chilled in ice. At the end of the 45-min incubation, cells were pelleted by centrifugation (3 min, 2000 ϫ g) at 4°C and resuspended in 1 ml of ice-cold PBS, 1% BSA on ice for 15 min. Immunolabeling of EGFP-NK2 receptors expressed at the cell surface was done using a monoclonal mouse anti-GFP (1/100 dilution, Roche Applied Science) for 30 min on ice. Cells were then washed three times with 1 ml of ice-cold PBS, 1% BSA, and secondary labeling was done using a R-phycoerythrin-conjugated AffiniPure F(abЈ) 2 fragment goat antimouse IgG (1/100 dilution, Immunotech, Marseille, France). Samples were washed twice in 1 ml of PBS, fixed for 5 min in ice-cold PBS, 4% paraformaldehyde, resuspended in 1 ml of ice-cold PBS, and stored overnight. Phycoerythrin staining was quantified by flow-cytometric analysis (10,000 cells/sample) on a cytometer (FACStar, BD Biosciences). Mean phycoerythrin fluorescence intensity was calculated using FACStar data software after subtraction of nonspecific staining by the two antibodies measured on non-transfected cells. The ratio of NK2 receptors at the cell surface was directly correlated to the ratio of phycoerythrin fluorescence intensity for samples with NKA as compared with samples without NKA.

RESULTS
vrFRAP were carried out on EGFP-NK2 receptors expressing cells before and after activation by the agonist NKA to investigate a possible confinement of the receptors in the plasma membrane and to characterize the lateral diffusion coefficient of this receptor.
Before applying the FRAP technique to biological membranes, one essential point to check is the degree of intracellular labeling of cells by fluorescent probes (23). Indeed, the depth of focus when large areas are illuminated (r ϭ 3.45 m) was calculated to be ϳ4 m (23). For accurate determination of the fluorescence intensity of the plasma membrane, the fraction of the fluorescence intensity originating from the intracellular compartments must be Ͻ10% of the total fluorescence intensity. This is the case for EGFP-NK2 receptors expressing cells, which display intense fluorescence, originating almost exclusively from the plasma membrane ( Fig. 1, A and B) (6). However, we checked that the incubation of these cells with cycloheximide for 1 h at 37°C, which is known to block protein synthesis (20), has no effect on the repartition of the fluores- cence over the cells nor on the recorded mobile fraction during FRAP experiments (data not shown).
The temperature was maintained at 20°C during FRAP experiments to minimize endocytosis (6,24). At this temperature, ligand binding to the NK2 receptors as well as activation of at least two types of cellular responses (elevation of intracellular concentrations of both calcium and cAMP) were detected (7,8), showing that the receptor is functional. A confocal microscopy comparison of the localization of the EGFP-NK2 receptors before and after agonist activation is illustrated in Fig. 1. In the absence of the agonist, EGFP-NK2 receptor distribution at the plasma membrane is similar in cells incubated at 37 (Fig. 1A) or 20°C (Fig. 1B). After a 30-min exposure to agonist at 37°C, spots of intense fluorescence were clearly visible mostly in the cytoplasm (Fig. 1C). After a 1-min incubation with agonist, spots were detected in or near the plasma membrane. When incubation with agonist was carried out at 20°C, EGFP fluorescent signal was mainly maintained at the plasma membrane with occasional redistribution in plasma membrane-associated spots (Fig. 1D).
Confined Diffusion of Fluorescent EGFP-NK2 Receptors at the Plasma Membrane in the Absence of Agonist-The variation of the fraction of mobile receptors (M) and of the lateral diffusion coefficients (D) of EGFP-NK2 receptors as a function of the radius of illuminated area (R) has been determined in several experiments (n ϭ 30 -40) carried out over independent campaigns (n ϭ 3) of measurements. The results of each independent campaign were analyzed, and the results of one representative campaign are shown in Fig. 2. The mobile fraction (M) of receptors (filled triangles) increases linearly with the inverse of the radius of the illuminated area. This indicates possible lateral organization of the EGFP-NK2 receptors in micrometer domains. The slope of the linear regression is equal to 26.8 Ϯ 5.7 m and, thus, EGFP-NK2 receptors may be considered as compartmentalized according to Equation 1 in confined domains with a mean radius of around 420 Ϯ 80 nm. Moreover, the linear regression does not extrapolate to zero but instead intercepts the M axis at a mobile fraction value, Mp, close to 48 Ϯ 4% (filled triangles). This indicates as already described (12) that the domains are slightly open and that a fraction of NK2 receptors is allowed to exchange between the confined domains and the sur-rounding membrane (in proportion, ϳ6% of the perimeter of the confined surface allowed this exchange) (12).
Lateral diffusion coefficients can be compared and analyzed within a single set of experiments carried out on a same day ( Table I). The measured D app decreases with increasing 1/R. As the variation of M with 1/R indicates a confinement of NK2 receptors, the true value of the lateral diffusion coefficient D real can be calculated according to Equation 2 (12). Calculated values of D real (Table I) still varied with R. These series of D measurements have thus to be analyzed according to two populations of receptors participating to the global phenomenon of diffusion: one D1 app. corresponding to the diffusion of EGFP-NK2 receptors within domains and the other D2 corresponding to the long range lateral diffusion of EGFP-NK2 receptors, which is independent of R. All of the recorded recovery curves were reanalyzed and best fit-corresponded as follows. (i) D2 invariant with R (0.4 Ϯ 0.1 10 Ϫ9 cm 2 /s) corresponds to the long range lateral diffusion coefficient of receptors between confined domains (Table I). This value of D2 is consistent with the lateral diffusion coefficient reported for the free diffusion of proteins in the cellular plasma membrane (4,(25)(26)(27). The percentage of D2 decreases with observation area radius (Table I) as expected for the diffusion of receptors between confined domains (Ϸ65% NK2 receptors). (ii) D1 app. is an apparent lateral diffusion coefficient, and D1 real must be recalculated according to Equation 2 (Table I). The true lateral diffusion coefficient D1 real of NK2 receptors within domains corresponding to 35% of the receptors thus has a value of ϳ1.1 Ϯ 0.3 10 Ϫ9 cm 2 /s.

Effect of Activation by NKA on the Confinement of Fluorescent EGFP-NK2 Receptors at the Plasma Membrane-vrFRAP
experiments were carried out with EGFP-NK2 receptors expressing cells following activation of the NK2 receptors by the agonist NKA. NKA was used at a concentration of 1 M to ensure Ͼ95% receptor site occupancy (6). Cells were incubated for 15 min and then were mounted on a slide. The concentration of NKA was maintained during the 30 min required for the analysis of one cell preparation. Fig. 2 (filled squares) shows that M varies moderately with 1/R (slope Ϸ10.9 Ϯ 3 m), indicating confined diffusion of the NK2 receptors in the small domains of r Ϸ 170 Ϯ 50 nm following activation. From the intercept with the ordinate (Mp Ϸ 65 Ϯ 3%), it appears that NK2 receptors can exchange between these small domains and the surrounding membrane. It can be calculated that 15% of the perimeter of the small confinement areas (radius of 170 nm) is equivalent to 6% of the perimeter of domain with a radius of 450 nm. Thus, the fraction of NK2 receptors that were allowed to exchange is the same before and after activation.
The calculation of true D values corresponding to D app corrected for the diffusion area lead to D real values of ϳ10 Ϫ11 cm 2 /s still varying with the illuminated radius. The best fit of the recovery curves is obtained with two diffusion components (Table II). D2, independent of R, is close to 0.6 Ϯ 0.2 10 Ϫ9 cm 2 /s, a value similar to that observed in non-stimulated cells (0.4 Ϯ 0.1 10 Ϫ9 cm 2 /s). This lateral diffusion coefficient corresponds to long range diffusion of receptors. As expected, its proportion decreases with R (Table II)  display a pattern of confined diffusion similar to that of NK2 receptors in the plasma membrane of HEK293 cells. This was done by labeling the lipid phase with fluorescent phosphatidylcholine, C 6 -NBD-PC. After labeling at 4°C, only the plasma membrane was fluorescent and no uptake of the probe into the cytoplasm was observed (data not shown). This labeling remains stable over the period of FRAP experiments on one cell preparation (19).
Experiments are first carried out on cells expressing nonfluorescent NK2 receptors at the same density (2 ϫ 10 6 receptors/cell) than EGFP-NK2 receptors in the experiments described above. In the absence of receptor stimulation by NKA, the lateral diffusion of C 6 -NBD-PC is 1.8 Ϯ 0.2 10 Ϫ9 cm 2 /s for any illuminated radius tested with an invariant mobile fraction of around 75 Ϯ 3% (Table III). Similar D values (Ϸ1-2 10 Ϫ9 cm 2 /s) and M values (Ϸ70 -80%) have been reported for the lateral diffusion of lipophilic probes inserted into the plasma membrane of various cell systems (19,23,28). The non-negligible fraction of immobilized lipid probes (25 Ϯ 3%) has never been elucidated and suggested the existence of clusters of lipids with or without proteins displaying restricted diffusion (19). Moreover, the mobile fraction and lateral diffusion coefficient do not vary with R, the radius of illuminated area, indicating the absence of confinement for the lipid probes (12). Similar experiments were carried out following stimulation of NK2 receptors with 1 M NKA. Only two observation radii were tested. No variation of M with R was recorded, but higher D (Ϸ4 Ϯ 0.3 10 Ϫ9 cm 2 /s) and M (Ϸ86 Ϯ 3%) values were measured (Table III).
vrFRAP experiments were carried out on HEK293 cells which do not express NK2 receptors (Table III). In the absence of NKA, D and M were in the same range as for NK2 receptors expressing cells (D Ϸ 2 Ϯ 0.1 10 Ϫ9 cm 2 /s and M Ϸ 78 Ϯ 3%, Table III). In the presence of 1 M NKA, D did not change (D Ϸ 2 Ϯ 0.1 10 Ϫ9 cm 2 /s) but the mobile fraction increased to a value of 86 Ϯ 3%, similar to the M value reported for non-fluorescent NK2 receptors expressing cells incubated with NKA (Table III).
It can be concluded that the increase of C 6 -NBD-PC lateral diffusion coefficient is due to the lateral reorganization of NK2 receptors in the plasma membrane upon stimulation with NKA, whereas the increase of the mobile fraction of C 6 -NBD-PC probes could be due to a direct interaction of NKA, an amphiphilic peptidic agonist (critical micellar concentration Ϸ 10 Ϫ5 M (data not shown)) (5) with the lipid phase of the cell plasma membrane.
Analysis of the Raft Association of Fluorescent EGFP-NK2 Receptors before and after Activation by NKA-Several G-protein-coupled receptors have been shown to associate with lipid microdomains such as rafts and/or caveolae (29) by the criterion of insolubility in non-ionic detergent at 4°C (30) or by purification of caveolae-enriched membranes without detergent (31). To test whether the confined diffusion observed for the EGFP-NK2 receptors before and/or after activation by NKA was due to an association with one of these lipid microdomains, cold Triton X-100 extraction of cell membranes was performed on EGFP-NK2 receptors expressing cells. The insoluble lipidenriched material was separated by flotation in a sucrose OptiPrep density gradient upon centrifugation. Each fraction of the gradient then was tested by immunoblotting for the presence of proteins of interest. In the resting cells, EGFP-NK2 receptors were found at the bottom of the gradient together with the soluble material independently of the temperature tested, either at 20 or 37°C. Fig. 3A illustrates the detergent solubility of EGFP-NK2 receptors for cells incubated at 37°C for 5 min in the absence of NKA. The receptor protein is found in the highest density fractions, fractions 4 -6. Upon activation with NKA, the receptors are recovered in the soluble fraction at the bottom of the gradient at any time point and temperature tested. Fig. 3B illustrates the detergent solubility of EGFP-NK2 receptors for cells incubated with 10 M NKA for 5 min at 37°C. As positive controls, the endogenously expressed caveolin-1 protein and the overexpressed EYFP-GPI-anchored fusion protein (17) were found to be associated with the detergentinsoluble materials in the lightest fractions of the gradient, TABLE 1 Analysis of lateral diffusion coefficients and mobile fractions of EGFP-NK2 receptors in the basal state as a function of the size of the radius of the illuminated area, R As described under "Experimental Procedures," recovery curves can be fitted by one or two-diffusive populations. D and M are given at a 95% confidence level. Each set of experiments for one value of R contains ϳ30 -40 recovery curves. D app is obtained for one-D fit of the recovery curves. D real is the D recalculated from D app according to Equation 2. D1 app and D2 are obtained for two D fits of the recovery curves. D1 real is the D recalculated from D1 app according to Equation 2. D2 is the long range D of NK2 receptors between domains, whereas D1 real is the D of NK2 receptors inside domains. S 2 is the variance of each set of experiments. ⌬S 2 is the improvement of the variance after obtaining the best fit with two lateral diffusion coefficients. All of the experiments were carried out at 20°C.  (Fig. 3, C and E, respectively). As a negative control, the expressed EYFP-CD46 fusion protein (18) was tested and found to be completely solubilized by detergent (Fig.  3D). Thus, neither of the 70% of completely free to diffuse receptors nor the compartmentalized 30% appears to be in association with detergent-resistant lipid microdomains of the plasma membrane.

Co-localization of Fluorescent EGFP-NK2 Receptors with Texas
Red-Transferrin following NKA Activation-Upon stimulation, numerous G-protein-coupled receptors get internalized by clathrin-coated pits (32). In addition, clathrin-coated pre-pit domains have already been described in the literature (33-37). The observed confinement of part of activated EGFP-NK2 re-ceptors could be due to the re-localization of the receptors in such flat clathrin-coated compartments as described for the ␤ 2 -adrenergic receptors expressed in HEK293 cells (38) or for the thyrotropin-releasing hormone receptor-1 expressed in HeLa cells (39).
Therefore, the co-localization of EGFP-NK2 receptors with a fluorescent marker for clathrin-coated pits was investigated. Upon incubation at 37°C of EGFP-NK2-expressing cells with a buffer supplemented with both NKA (1 M) and Texas Redtransferrin (20 g/ml), there was a clear co-localization of EGFP-NK2 receptors with Texas Red-transferrin at all of the time points tested (from 5 to 15 to 30 and 45 min). Fig. 4 illustrates this co-localization for the incubation periods of 5 (Fig. 4, A-C) and 45 min (Fig. 4, D-F). In these representative confocal images, the red channel shows the Texas Red-transferrin staining of vesicular endocytic structures, which are close to the plasma membrane at first (Fig. 4A) and which later concentrate in a perinuclear compartment known as the recycling endosomes (see the cell in the center in Fig. 4D). The green channel shows the localization of the EGFP-NK2 receptors both at the periphery of the cell and in vesicular structures (Fig. 4, B and E). The vesicles containing the receptors are also positive for Texas Red-transferrin as the overlay images of the two channels indicate (Fig. 4, C and F). The early co-localization of activated NK2 receptors with internalized transferrin is a strong indication that these receptors internalize mainly via clathrin-coated vesicles. When the same type of experiments was performed at 20°C, EGFP-NK2 receptors were found to label mainly the plasma membrane at all of the time points tested as already described in Fig. 1. The green channel in Fig.  5 shows the plasma membrane localization of the EGFP-NK2 receptors after a 5-and 45-min activation (Fig. 5, B and E,  respectively). Texas Red-transferrin becomes detectable only in late incubation times (Fig. 5D, compare 45 min with 5 min in Fig. 5A). Texas Red-transferrin then gives a punctate staining at the periphery of the plasma membrane (Fig. 5D). These punctate structures are positive for the NK2 receptor as well as indicated by the overlay images of the two channels (Fig. 5F). This co-localization is in favor of the hypothesis that part of the activated EGFP-NK2 receptors is compartmentalized in clathrin-coated pre-pits at 20°C.
Co-localization of Non-fluorescent NK2 Receptors with ␤-Arrestin2-EYFP following NKA Activation-If the activated NK2 receptors are compartmentalized in clathrin-coated prepits at 20°C, one may expect to find early components of the endocytic machinery in these microdomains as well. ␤-Arres-tin2 translocates from the cytoplasm to the agonist-activated GPCRs and serves as an adaptor protein linking the receptor molecules to both the AP2 complex and the clathrin itself (40). This was particularly shown for NK1 and NK3 receptors (41).  3. Immunoblots illustrating that EGFP-NK2 receptors are not compartmentalized in detergent-insoluble lipid microdomains in HEK293 cells. EGFP-NK2 receptors expressing cells were treated (A) or not (B) for 5 min at 37°C with 10 M NKA prior to homogenization. EYFP-GPI-anchored protein (C) or EYFP-CD46 protein (D) expressing cells were directly processed to the homogenization step. The different cell homogenates were solubilized in 1% Triton X-100 at 4°C, and the lipid-enriched insoluble material was separated from the soluble protein fraction by centrifugation in an OptiPrep sucrose gradient. Six fractions were collected from each gradient with fraction 1 being the lightest sucrose fraction. The EGFP-NK2 receptors (A and B), the raft marker EYFP-GPI-anchored protein (C), and the non-raft marker EYFP-CD46 protein (D) were immunoprecipitated from each fraction of their corresponding gradient with monoclonal anti-GFP antibodies and revealed by SDS-PAGE followed by Western blotting with a polyclonal anti-GFP antibody. The presence of the positive control caveolin-1 (E) in the lipid microdomains was assessed by trichloroacetic acid precipitation of the different fractions of the gradient followed by SDS-PAGE and immunoblotting with anti-caveolin 1 antibodies.
In addition, the recruitment of ␤-arrestin2 has been reported to occur already at 16°C (38,42) in response to agonists of the ␤ 2 -adrenergic receptor expressed in HEK293 or rat basophilic leukemia (RBL) cells. Therefore, co-localization of non-fluorescent-activated NK2 receptors with ␤-arrestin2-EYFP was investigated. Non-fluorescent NK2 receptors expressing cells were transiently transfected with a ␤-arrestin2-EYFP-expressing vector. In the resting state, the ␤-arrestin2-EYFP localizes throughout the cytoplasm and is excluded from the nucleus (Fig. 6B). Non-fluorescent NK2 receptors were then activated with 1 M Texas Red-NKA, a ligand that allows to pinpoint membrane-bound NK2 receptors (7,8). At 37°C, ϳ50% of the cells expressing both the ␤-arrestin2-EYFP and the non-fluorescent NK2 receptors exhibited a translocation of the ␤-arres-tin2-EYFP from the cytoplasm to the plasma membrane at all of the time points tested (1,5,15,30, and 45 min). Translocation was total for ϳ75% of these cells and partial for the remaining 25% during the first 5 min of activation. At later time points (15,30, and 45 min), the amount of cells exhibiting a total membranous localization of ␤-arrestin2-EYFP dropped to 25%. In all of the cases where ␤-arrestin2-EYFP was localized at the periphery of membranes, there was total co-localization between the ␤-arrestin2 and Texas Red-NKA in internalized vesicular structures. Representative confocal images illustrate the co-localization of the two proteins after NK2 receptors activation for 5 (Fig. 6, D-F) and 30 min (Fig. 6, G-I).
The red channel shows the Texas Red-NKA localization in endocytic vesicles at a 5-min time point (Fig. 6D) and its accumulation in a perinuclear compartment after 30 min (Fig. 6G). For the same cells, the green channel shows transiently expressed ␤-arrestin2-EYFP. The arrestin has translocated from the cytoplasm to compartments totally at a 5-min time point (Fig. 6E) and partially at a 30-min time point (Fig. 6H). These compartments positive for arrestin contain the NK2 receptor as well as exemplified in the overlay of the two channels (Fig. 6, F and I). In conclusion, ␤-arrestin2 is recruited at sites of NK2 receptor internalization and co-localizes with internalized receptors. The maintenance of the interaction throughout the internalization pathway has already been described for some but not all of the GPCR/␤-arrestin2 interactions. In particular, it occurs with the NK1 receptor but not with the NK3 receptor overexpressed in KNRK cells (43).
When the experiment was performed at 20°C, a certain amount of ␤-arrestin2-EYFP was found to translocate at the periphery of the cell upon NK2 receptor activation. In contrast to incubation at 37°C, arrestin stayed at the periphery of the cell throughout the incubation period (from 1 to 45 min). During the first 15 min, ϳ30% of cells co-expressing ␤-arrestin2-EYFP and the non-fluorescent receptor displayed a translocation of ␤-arrestin2 at the periphery of the plasma membrane. At 30 min, 75% of the co-expressing cells displayed translocation of ␤-arrestin2-EYFP. At all of the time points, translocation was total for 25% of the cells. Whether the translocation was total or partial, all of ␤-arrestin2-EYFP present at the periphery of the plasma membrane co-localized with the NK2 receptors labeled with Texas Red-NKA. Representative confocal images illustrate the partial translocation of ␤-arrestin2-EYFP and its total co-localization with spots of activated NK2 receptors at the periphery of the cells after NK2 receptor activation for 5 (Fig. 7, D-F) and 30 min (Fig. 7, G-I). The red channel shows the Texas Red-NKA labeling (Fig. 7, D and G). The cell in the center of each image is transiently expressing ␤-arrestin2-EYFP visualized in the green channel (Fig. 7, E and  H), whereas the surrounding cells express only the non-fluorescent NK2 receptor. In the co-expressing cells, ␤-arrestin2-EYFP is both cytoplasmic and enriched in punctate structures at the cell periphery that contain the NK2 receptor (as exemplified in the overlay of the two channels in Fig. 7, F and I). This is in favor of the hypothesis that, at 20°C, part of the activated NK2 receptors is compartmentalized in clathrincoated pre-pits, which contain the adaptor ␤-arrestin2 protein.

DISCUSSION
Using a set-up of vrFRAP with improved data analysis (12,21,44), we studied the lateral diffusion and putative dynamic membrane confinement of neurokinin NK2 receptors expressed in HEK293 cells. We then combined biophysical study and biological approaches to define the nature of the detected confinement of the receptors.

Microdomains Confinement of EGFP-NK2 Receptors in the Resting State and after Activation by NKA-
The vrFRAP approach shows that the membrane lateral diffusion of NK2 receptors in the absence of agonist is confined to domains with a radius of 420 Ϯ 80 nm, which remains stable in the time scale of FRAP experiments (Fig. 2, filled triangles). An analysis of recovery curves with two lateral diffusion coefficients shows that 35% NK2 receptors diffuse within domains with a D1 real value of 1.1 Ϯ 0.3 10 Ϫ9 cm 2 /s, whereas the remaining 65% displays long range diffusion between domains with a coefficient D2 of 0.4 Ϯ 0.1 10 Ϫ9 cm 2 /s ( Table I). The D values obtained imply that a receptor can move around the entire cell in 15 min. Values of ϳ10 Ϫ9 cm 2 /s have been reported for lateral diffusion coefficients of GPCRs, estimated by classical FRAP experiments carried out with the ␤ 2 -adrenergic receptor expressed in HEK293 cells (45), the gonadotropin-releasing hormone receptor expressed in Chinese hamster ovary cells (46,47), and the luteinizing hormone receptor expressed in HEK293 cells (48). The single particle tracking technique, which analyzes individual trajectories, can provide complementary information concerning the compartmentalization of proteins (16). Diverse lateral diffusion coefficients obtained by single particle tracking have been reported, but most studies have described compartmentalization. This was shown for the -opioid GPCR expressed in NRK cells (4), for the endogenous metabotropic glutamate R5 GPCR expressed in neurons (3), for GPI-anchored proteins in C3H 10T1/2 cells (49,50), for the transferrin receptor expressed in NRK cells (51), and for the major histo- compatibility complex class 1 protein expressed in HeLa cells (52). The radii of the protein confinement zones reported range from 100 to 500 nm.
After stimulation of EGFP-NK2 receptors by 1 M NKA, vrFRAP experiments show that a fraction of NK2 receptors (30%) is still compartmentalized but in smaller domains (radius Ϸ 170 Ϯ 50 nm, Fig. 2, filled squares). The long range diffusion coefficient between domains found for the 70% remaining receptors is slightly higher than the one of unstimulated receptors (D2 Ϸ 0.6 Ϯ 0.2 10 Ϫ9 cm 2 /s, Table II). In con-trast, the diffusion coefficient corresponding to the diffusion of NK2 receptors within domains is two orders of magnitude lower (D1 real ϭ 0.8 Ϯ 0.15 10 Ϫ11 cm 2 /s). This finding indicates that these receptors are essentially immobilized in the time scale of FRAP experiments (30 -45 s). These data are consistent with a work reporting an immobilization of the gastrinreleasing peptide receptor, not reflecting internalization of the GPCR, after stimulation by the agonist GRP (gastrin-releasing peptide) (53). Similarly, FRAP experiments carried out on activated cholecystokinin GPCR endogenously expressed in pancreatic acinar cells report D values comprised between 10 Ϫ10 cm 2 /s when ligand was applied at 4°C and 10 Ϫ11 cm 2 /s when ligand was applied at 37°C (25). Although diffusion of the unstimulated receptor could not be followed in this work, the diffusion coefficients of the activated CCKR support the notion that receptors do not freely diffuse after stimulation. In the case of the gonadotropin-releasing hormone expressed in Chinese hamster ovary cells, the reduction of receptor diffusion at 37°C is also observed in the presence of agonist (2 ϫ 10 Ϫ9 cm 2 /s to 0.2 ϫ 10 Ϫ9 cm 2 /s after activation) (46,47) but to a lesser extent. Thus, comparing our data with the literature on experiments performed at 37°C, we think that measurements at 20°C are a good compromise since the major effect of lowering the temperature is to slow down the biological processes, either agonist binding or intracellular responses (6 -8), hence giving us more time to detect and accurately quantify/analyze these phenomena. In addition, the internalization process is almost non-existent at 20°C, which allows us to study activated NK2 receptors in a more synchronized manner, which is good with vrFRAP, analyzing mean population behavior.
Our results, obtained with the vrFRAP approach, are consistent with two models of membrane organization. (i) In one model, long range attractions between membrane proteins can generate protein confinement that does not involve the presence of physical fences as recently proposed for opioid receptors (4). In this model, membrane protein confinement is a dynamic process governed by physical parameters of protein-lipid interactions. Such dynamic confinement model predicts equilibrium between confined and freely diffusing protein species and can thus account for the escape of a fraction of NK2 receptors (Table I) as well as for the long range lateral diffusion of lipids (Table III). (ii) In a second model, confinement domains are limited by protein-protein interactions, similar to those in the cytoskeleton fence model (15), or by interactions with cytoplasmic scaffolding proteins (3).
Confinement of EGFP-NK2 from a Biological Point of View-Localization of GPCRs (29) and, more generally, of receptors (54) in the specific lipid microdomains called rafts has been described. Whatever the experimental conditions tested, we were not able to find an association of EGFP-NK2 receptors with lipid microdomains that resist to cold non-ionic detergent solubilization, a hallmark of raft domains (30). A slight decrease of diffusion coefficient is generally observed for proteins inside the liquid ordered phase created by preferential interactions between sphingolipids and cholesterol enriched in rafts as compared with those surrounded by liquid-disordered phase constituted of most lipids of the membrane (55). The large variation of diffusion coefficient observed upon receptor activation thus is not compatible with receptor redistribution from or into rafts.
Once activated, many GPCRs are internalized through clathrin-coated pits (39,56,57). In particular, this has been demonstrated for the neurokinin NK1 receptor by co-localization with the transferrin receptor (24, 58 -60). Not surprisingly, we have now found that activated EGFP-NK2 receptors expressed in HEK293 cells internalize at 37°C through clathrin-coated pits both by co-localization with fluorescently labeled transferrin (Fig. 4) and by blockade of clathrin-coated pit formation by sucrose (data not shown) (61,62). We also demonstrate that ␤-arrestin2, an adaptor for clathrin and AP2 complexes, is recruited at the periphery of the plasma membrane upon NK2 receptors activation certainly by binding to the receptors (40). At 37°C, ␤-arrestin2 co-localizes with internalized NK2 receptors. At 20°C, some ␤-arrestin2 is recruited at the periphery of the plasma membrane where it co-localizes with the activated NK2 receptors.
Thus, we propose that the small domains in which 30% of activated EGFP-NK2 receptors are immobilized might be precursors of clathrin-coated pits (63). Such precursors might correspond to preexisting clathrin-coated domains (33)(34)(35)(36)(37) in which ␤ 2 -adrenergic receptors (38) or thyrotropin-releasing hormone receptor-1 (39) have been shown to be targeted. It has also been proposed that these zones are flat when proteins reach them and that they invaginate following clathrin polymerization (37,(63)(64)(65). Considering that the individual spherical clathrin-coated pits have a radius of 75-100 nm (33,63,65), the radius of the corresponding flat zone (150 -200 nm) is close to the radius of the domains estimated in our vrFRAP experiments (170 Ϯ 50 nm, Fig. 2B). The immobilization of the activated EGFP-NK2 receptors in these precursors of coated pits might result from anchoring of receptor molecules by clathrin (56) and/or by the actin cytoskeleton (33). This could be addressed in a future work using fluorescent clathrin and fluorescent actin or by modulating actin dynamic using overexpressed actin-binding proteins.
The fraction of immobilized confined receptors detected after stimulation of cells by agonist is comparable to the fraction of FIG. 8. Schematic model describing lateral organization of NK2 receptors prior (state I) and after stimulation by NKA (cases IIA and IIB). Before activation by the agonist, 70% receptor molecules exhibit long range diffusion and 30% are confined in large (white) zones of confinement A with a radius of 420 nm. Taking into account that an HEK293 cell presents a radius of ϳ6.5 m, the total amount of NK2 receptors (2 ϫ 10 6 receptors) occupy 2.8% of the cell surface (receptor surface of Ϸ7 nm 2 ) (66, 67). Since lateral diffusion coefficient of NK2 receptors inside the zones of confinement A (1.1 Ϯ 0.3 10 Ϫ9 cm 2 /s) and in the surrounding membrane B (0.4 Ϯ 0.1 10 Ϫ9 cm 2 /s) are comparable, homogeneous distribution of NK2 receptors can be assumed. Thus, one can deduce that, in one large zone of confinement A (radius Ϸ 420 nm), 2.8% of the confinement surface is occupied by NK2 receptors. This corresponds to Ϸ2.000 NK2 receptors. Taking into account that 30% NK2 receptors are confined (Ϸ6 ϫ 10 5 receptors), the number of zones of confinement can be estimated around 300 by cell. In case IIA, The density of receptors at the cell surface is unchanged by NKA addition, but the radius of the confinement zone is reduced to 170 nm. These results show that the number of confinement zones C increases by a factor of 6. Each zone contains Ϸ330 -350 receptors. In case IIB, The number of confinement zones is constant. The density of receptors inside confinement zones C then reaches 17% of the confinement surface. The range of background colors of various circular patterns is related to the D value of NK2 receptors measured by vrFRAP: white, Ϸ1.1 ϫ 10 Ϫ9 cm 2 /s; dark gray, Ϸ0.4 ϫ 10 Ϫ9 cm 2 /s (in state I); light gray, Ϸ0.6 ϫ 10 Ϫ9 cm 2 /s; and black, Ϸ0.008 ϫ 10 Ϫ9 cm 2 /s in cases IIA and IIB. receptors confined before stimulation. Assuming homogeneous distribution of receptors at the cell surface (based on the same D inside and outside the zones of confinement A in state I) before agonist application, we estimate that there are ϳ300 large confinement zones each containing 2000 NK2 receptor molecules (Fig. 8, state I, see captions for the calculations). The membrane area covered by NK2 receptor molecules represents Ͻ3% membrane surface (2 ϫ 10 6 receptors with an area of Ϸ7 nm) (66,67). To maintain the 70/30 ratio of diffusing/confined receptors after agonist stimulation, two extreme cases can be envisioned. 1) Agonist application might lead to an increase of the number of confined zones keeping the surfacic density of receptors constant (case IIA), or 2) it might increase the density of receptors inside confined zones, keeping the confined zone numbers constant (case IIB) (Fig. 8). In case IIA, the number of confined zones would increase by a factor of 6. According to Schram et al. (44), lipid diffusion should significantly decrease due to the larger number of obstacles instead of increasing as is experimentally observed (Table III). In case IIB, the surfacic density of receptors inside confined zones C would reach 17%. Such a value is much below the maximal admitted protein density in biological membranes (ϳ50% in surface) (68). In addition, in this case, the rate of lipid diffusion is expected to increase as a result of the reduction of obstacle density (44). The reorganization of the plasma membrane, as illustrated in case IIB, is sufficient to account for the experimentally observed increase of the lateral diffusion coefficients of membrane partners and particularly of lipids (D Ϸ 4 Ϯ 0.3 10 Ϫ9 cm 2 /s) in the surrounding membrane.
Finally, the fact that similar amounts of EGFP-NK2 receptors (30 -35%) are compartmentalized before and after activation by NKA might have some important biological relevance. For example, the confinement observed for 35% resting EGFP-NK2 receptors might be due to transient interactions of the receptors with preexisting scaffolds of signaling molecules. This non-activated population of receptors can indeed diffuse within domains and exchange between domains and surrounding membrane. The preexisting scaffolds could be present in limited amount as compared with the total receptor number. In this work, this may for instance result in signaling and processing for internalization of only 30% of ligand-bound receptors. We have performed a quantification of the number of EGFP-NK2 receptors present at the surface of the cells before and after activation with 1 M NKA at different time points ranging from 5 to 45 min (data not shown). We have found that not Ͼ50% of the receptors get internalized over this period of time with 30% internalization occurring in the first 5 min. This supports our model in that not all of the receptors are capable of internalization. The reason why the amount of receptors internalized is higher than expected in our present model might be attributed to the fact that, at 37°C, most responses due to a cascade of protein interactions get amplified compared with 20°C. We have observed it previously with an enhancement of the maximal peak of calcium and cAMP at 37°C compared with 20°C upon NK2 receptor activation (8) and in this present study with a more pronounced recruitment of ␤-arrestin2 at 37°C compared with 20°C.
The preexistence of signaling scaffolds, called "transducisomes" (69,70), has already been proposed for receptors expressed in neurons at the level of synapses (3), but the existence of limited amount of preexisting signaling complexes in non-polarized cells is less documented. It would be interesting to test further this hypothesis by performing vrFRAP analysis on cells, which express variable amounts of receptors, and to characterize the dynamics of the interaction of the receptor with other proteins in an integrated temporal scheme.