Dynamic Interaction of Amphiphysin with N-WASP Regulates Actin Assembly*

Amphiphysin 1, an endocytic adaptor concentrated at synapses that couples clathrin-mediated endocytosis to dynamin-dependent fission, was also shown to have a regulatory role in actin dynamics. Here, we report that amphiphysin 1 interacts with N-WASP and stimulates N-WASP- and Arp2/3-dependent actin polymerization. Both the Src homology 3 and the N-BAR domains are required for this stimulation. Acidic liposome-triggered, N-WASP-dependent actin polymerization is strongly impaired in brain cytosol of amphiphysin 1 knock-out mice. FRET-FLIM analysis of Sertoli cells, where endogenously expressed amphiphysin 1 co-localizes with N-WASP in peripheral ruffles, confirmed the association between the two proteins in vivo. This association undergoes regulation and is enhanced by stimulating phosphatidylserine receptors on the cell surface with phosphatidylserine-containing liposomes that trigger ruffle formation. These results indicate that actin regulation is a key function of amphiphysin 1 and that such function cooperates with the endocytic adaptor role and membrane shaping/curvature sensing properties of the protein during the endocytic reaction.

The dynamic nature of the actin cytoskeleton is crucial for a variety of cellular events, including cell morphogenesis, cell migration, and intracellular membrane traffic (1). Actin polymerization is stimulated by a variety of actin regulatory proteins, prominent among which are WASP family proteins that function by triggering Arp2/3-mediated actin nucleation (2). Activation of WASP family proteins, in turn, is controlled by factors that bind these proteins and release an autoinhibitory intramolecular interaction that prevents their VCA domain from interacting with the Arp2/3 complex. As extensively shown for N-WASP, many such factors are proteins that bind to the N-WASP proline-rich region via the SH3 4 domain.
Recently, we have found that the SH3 domain containing protein amphiphysin 1 stimulates actin polymerization during phagocytosis in testicular Sertoli cells, and this effect requires interactions of its C-terminal SH3 domain (3). Amphiphysin 1 is an endocytic adaptor present at high levels in brain at neuronal synapses but is also expressed at significant levels in Sertoli cells (4,5). In addition to a C-terminal SH3 domain, known to bind the GTPase dynamin and the phosphoinositide phosphatase synaptojanin (6), amphiphysin 1 contains an N-terminal BAR domain, a curved protein module that binds lipid bilayers and generates and senses curvature (7,8). It also contains binding motifs for clathrin and for the clathrin adaptor AP-2 (9). Hence, amphiphysin 1 was primarily studied as an endocytic protein capable of assembling at the neck of endocytic pits and of coupling clathrin-mediated budding to dynamin-mediated fission (10,11). However, regulatory roles of amphiphysin 1 in actin cytoskeleton have also been suggested by studies of neuronal growth cones (12) and by the function of the amphiphysin homologue in yeast, Rvs167 (13)(14)(15).
We have searched for a mechanistic explanation of the stimulatory action of amphiphysin 1 on actin dynamics. We now show that amphiphysin 1 acts as an activator of N-WASP, and we provide evidence for the occurrence of this activation in living cells.
Animals and Cell Culture-Wild type mice were purchased from Shimizu Laboratory Supplies Co. (Kyoto, Japan). Amphiphysin 1 knock-out mice (amphiphysin 1 Ϫ/Ϫ ) were generated by gene targeting in embryonic stem cells as described previously (16). Ten-week-old wild type or amphiphysin 1 Ϫ/Ϫ brains were used for cytosol preparation. All animals were maintained in clean conditions with free access to food and water. They were allowed to adapt to their environment for more than 1 week before initiating the experiments. Ser-W3 cells, a rat Sertoli cell line, were cultured with Dulbecco's modified Eagle's medium containing 10% fetal bovine serum at 37°C under 5% CO 2 (3).
Protein Purification-N-WASP was expressed in Sf9 cells by a Bac-to-Bac baculovirus expression system (Invitrogen) with a His 6 tag. Recombinant virus-infected Sf9 cells were lysed, and His 6 -N-WASP was purified with Ni-NTA-agarose (Qiagen). GST fusion VCA was purified as described previously (17,18). Arp2/3 complex was purified as described previously (19). Actin was purified from rabbit skeletal muscle, and monomeric actin (G-actin) was isolated by gel filtration on Superdex 200 (GE Healthcare) in G buffer (2 mM Tris-HCl, pH 8.0, 0.2 mM CaCl 2 , 0.5 mM dithiothreitol, 0.2 mM ATP). Full-length amphiphysin 1, 1-306 amino acids (N-BAR-PRS), 1-626 amino acids (⌬SH3), 226 -695 amino acids (⌬N-BAR), and ⌬248 -601 amino acids (N-BAR-SH3) were subcloned into the plasmid pGEX-6P as described previously (8). The SH3 domain was subcloned into the pGEX-2T vector. The nucleotide sequences of the constructs were verified using a DNA sequence analyzer. The expression of GST fusion proteins was induced by 0.1 mM isopropyl 1-thio-D-galactopyranoside at 37°C for 3-6 h in LB medium supplemented with 100 g/ml ampicillin at A 600 ϭ 0.8. The purification of GST fusion proteins was performed as described previously (9), and the cleavage of the GST with Pre-Scission protease was carried out according to the manufacturer's instruction. Finally, the proteins were purified on Mono Q column equilibrated in 20 mM Tris-HCl, pH 7.7, and 0.2 M NaCl. The protein solution (1 mg/ml protein) was stored at Ϫ80°C and thawed at 37°C before use. The purity of these proteins was confirmed by SDS-PAGE.
cDNA Constructs and Transfection-Full-length amphiphysin 1 or 1-626 amino acids (⌬SH3) containing the BamHI and EcoRI restriction sites were subcloned into a pEGFP-C1 vector (Clontech) or mCherry-C1 kindly gifted by Dr. Tsien (Howard Hughes Medical Institute, University of California, San Diego). In another case, amphiphysin 1 containing XhoI and EcoRI restriction sites was subcloned into mCherry-N1. The plasmids pEF-BOS-myc-N-WASP and pEGFP-N-WASP were described previously (20). Full-length N-WASP and 265-391 amino acids (PRD of N-WASP) containing XhoI and EcoRI restriction sites were subcloned into mCherry-C1. The nucleotide sequences of the constructs were verified using DNA sequence analysis. The constructs were transfected into Ser-W3 cells using a Lipofectamine 2000 transfection system (Invitrogen). For FRET-FLIM analysis, Ser-W3 cells (7 ϫ 10 4 cells per well) were co-transfected with 3 g of plasmids containing cDNA for GFP-tagged protein and 1 g of cDNA for mCherry-tagged protein. Four g of GFP-amphiphysin 1 were transfected as a negative control. Twenty four hours after transfection, cells were subjected to FRET analysis.
Quantification of Membrane Ruffle Formation-Ser-W3 cells (1 ϫ 10 4 cells/coverslip) in serum-free Dulbecco's modified Eagle's medium were stimulated with 0.25 mM PS-containing liposomes and incubated at 37°C for 10 min. The cells were then washed with PBS containing 1.5 mM CaCl 2 and 1 mM MgCl 2 (PBS(ϩ)) three times, fixed with 4% paraformaldehyde, permeabilized, and stained with Alexa 488-phalloidin or antic-Myc antibodies. Formed ruffles were characterized as thick actin filament accumulation at the cell periphery (3). To quantify ruffle formation, cells that had no ruffles were scored as negative, whereas cells that had one or more ruffles were considered to be positive. The number of ruffle-positive cells were counted and expressed as a percentage of total number of cells analyzed. At least 100 cells in different areas of the wells were counted in each experiment.
Microscopy-Ser-W3 cells (1 ϫ 10 4 cells/coverslip) were fixed with 4% paraformaldehyde in PBS(ϩ) at room temperature, permeabilized with 0.1% Triton X-100, and double stained by immunofluorescence as described previously (22). The samples were examined using a spinning disk confocal microscope system (CSU10, Yokogawa Electric Co., Japan) combined with an inverted microscope (IX-71, Olympus Optical Co., Ltd., Japan) and a Cool-SNAP-Pro camera (Roper Industries, Sarasota, FL). The system was steered by Metamorph software (Molecular Devices). When necessary, images were further processed using Adobe Photoshop and Illustrator software.
Preparation of Brain Cytosol-Brain cytosol was prepared as described previously (23). Briefly, 20 brains of wild type or amphiphysin 1 knock-out mice were homogenized in 5 ml of XB buffer (10 mM Hepes, 100 mM KCl, 2 mM MgCl 2 , 0.1 mM CaCl 2 , 5 mM EGTA, 50 mM sucrose, 1 mM dithiothreitol, 1 g/ml leupeptin, 5 g/ml pepstatin, and 0.4 mg/ml phenylmethylsulfonyl fluoride), pH 7.4. The homogenate was centrifuged at 3,000 ϫ g for 20 min and 10,000 ϫ g for 20 min. The resultant supernatant was diluted with XB buffer up to 4-fold and centrifuged at 400,000 ϫ g for 1 h. The clear supernatant was carefully collected and re-concentrated to one-fourth the volume using Centriprep-10 concentrators (Amicon Corp.). A final concentration of the cytosol was 40 -50 mg/ml. Amounts of actin in wild type or amphiphysin 1 Ϫ/Ϫ brain cytosol were estimated to be equal by Western blotting.
Preparation of Synaptosomes-Synaptosomes from adult mouse cortices were purified from P2 fractions by centrifugation on discontinuous Percoll gradients as described previously (24) with 0.5 mM EGTA in the initial homogenization buffer. The final synaptosomal pellet was resuspended in resting buffer (20 mM Hepes, 145 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 0.1 mM EGTA, 10 mM glucose, pH 7.4) to yield a synaptosome suspension with an OD 750 ϭ 0.75-0.85.
Determination of Actin Levels in Synaptosomes-G-actin/F-actin cycling was evaluated using a procedure described previously (25) with some modifications. One-ml aliquots of the synaptosome stock were preincubated at 30°C for 20 min and collected by centrifugation at 10,000 ϫ g for 20 s. Synaptosomes were resuspended in 70 l of depolarizing buffer (20 mM Hepes, 75 mM NaCl, 75 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, pH 7.4) and fixed by the addition of 120 l of 2.5% glutaraldehyde at various times of depolarization (from 3 to 120 s). For the zero time point, depolarizing buffer and glutaraldehyde were combined before the resuspension of the synaptosomal pellet. Synap-tosomes were sedimented and subsequently permeabilized in 0.1% Triton X-100 (v/v) and 1 mg/ml NaBH 4 in resting buffer for 2-3 min. The buffer was removed, and rhodamine-phalloidin or Oregon Green-DNase I (Invitrogen) in 5 mM KCl, 145 mM NaCl, 2 mM CaCl 2 was added (final volume 100 l). Staining for 20 min in the dark at room temperature was followed by 1-2 washes with 500 l of resting buffer. Labeled synaptosomes were resuspended in 0.32 M sucrose and stored in the dark at 4°C. The fluorescence associated with the samples was measured 24 h later using an LS50 spectrofluorometer (PerkinElmer Life Sciences) at the excitation and emission wavelengths of 540 and 566 nm for rhodamine-phalloidin and 497 and 524 nm for Oregon Green-DNase I, respectively (with 2.5 and 5 nm excitation/emission slits).
In Vitro Actin Assembly Assay-For quantitative analysis of actin assembly using cytosol, pyrene-actin assay was carried out according to Ma et al. (26). Briefly, diluted cytosol (8 mg/ml) with XB buffer supplemented with 0.4 mg/ml pyrene-actin (Cytoskeleton Inc.), 1.3 mM MgCl 2 , 0.1 mM EGTA, and ATP-generating system (1 mM ATP, 8 mM creatine phosphate, 8 units/ml phosphocreatine kinase) was incubated in a quartz cuvette at room temperature for 10 min. Lipid membrane at the indicated concentration was added, and pyrene fluorescence was then measured at 407 nm with excitation at 365 nm in an F-2500 fluorescence spectrophotometer (Hitachi Co. Ltd., Japan) with a 10-nm slit width. Pyrene-actin assays for N-WASP-dependent Arp2/3 activation were performed as described previously (20). Actin dynamics is reduced in amphiphysin 1 ؊/؊ in synapse and brain cytosol. A, liposome-induced actin polymerization measured by pyrene fluorescence. Wild type brain cytosols were treated with 100 M liposomes composed of 50% PS and 50% PC (ϩPS-lipo.) or 90% PC and 10% PI(4,5)P 2 (ϩPI(4,5)P 2 -lipo.). As controls, the incubation was carried out in the presence of 100 M 100% PC (ϩPC-lipo.) or in the absence of liposomes (ϩBuffer). Time 0 indicates the time when the liposomes were added. B, inhibition of PS-induced actin polymerization by wiskostatin. Brain cytosol was pretreated with 5 or 10 M wiskostatin (Wisko.) for 10 min. Then, the cytosol was stimulated with PS-liposomes or PC-liposomes. C, reduction of actin polymerization in amphiphysin 1 Ϫ/Ϫ brain cytosol ((Ϫ/Ϫ) ϩ PS-lipo.). The reduction was recovered in the presence of PS by adding back 1 or 2 M of recombinant full-length amphiphysin to the amphiphysin 1 Ϫ/Ϫ cytosol. The incubation was carried out as in B. D, actin, but not dynamin 1 and synaptophysin, was decreased in amphiphysin 1 Ϫ/Ϫ synaptosomes. Quantitative comparison of actin levels in total brain cytosol (left panel, n ϭ 8 for both genotypes) and synaptosomes (right panel, n ϭ 4 for both genotypes) from WT or amphiphysin 1 Ϫ/Ϫ mice. Twenty g of each fraction per lane were analyzed. Representative Western blots using antibody directed against actin, synaptophysin, or dynamin 1 were shown (upper panel). The amount of actin was measured by densitometry. Statistical significance was determined by Student's t tests (**, p Ͻ 0.01). E, cycling of actin assembly in depolarized synaptosomes. Synaptosomes from of WT (closed circles) and amphiphysin 1 Ϫ/Ϫ mice (open circles) were depolarized with high K ϩ for the indicated times. Amounts of F-actin (left panel) and G-actin (right panel) were fluorometrically measured by rhodamine-phalloidin and Oregon Green-DNase I, respectively. Each data point represents the mean Ϯ S.E. of the fluorometric readings performed from 10 independent experiments.  DECEMBER 4, 2009 • VOLUME 284 • NUMBER 49

JOURNAL OF BIOLOGICAL CHEMISTRY 34247
Multifocal Multiphoton Fluorescence Lifetime Imaging Microscopy and Data Analysis-The FRET-FLIM system was described previously (27). Briefly, the FRET-FLIM apparatus combines multifocal multiphoton excitation (TriMscope, LaVision Biotec, Bielefeld, Germany) and a fast-gated CCD camera (Picostar, LaVision Biotec, Bielefeld, Germany). Twophoton multifocal excitation was carried out using the Tri-MScope connected to an inverted microscope (IX 71, Olympus, Tokyo, Japan). A mode-locked Ti:Sa laser at 950 nm for the excitation of GFP (Spectra Physics, France) was split into 2-64 beams by utilizing a 50/50 beam splitter and mirrors. A line of focus was then created at the focal plane, which can be scanned across the sample. A filter wheel of spectral filters (535AF45 for GFP) was used to select the fluorescence imaged onto a fastgated light intensifier connected to a CCD camera. All instrumentation was controlled by IMSpector software developed by LaVision Biotec. Analysis of the data was done by using Image-J. Quantitative analysis was carried out as described previously (27). Briefly, the images from a timegated stack are first smoothed by a 3 ϫ 3 mask to decrease the noise. After that, Equation 1 was applied on the resulting background-subtracted time-gated images to recover mean lifetime pixel by pixel.
Finally, Equation 2 was applied on mean lifetime images using fixed lifetime donor values ( D ) to recover mf D . The quantity mf D stands for the minimal percentage of donor engaged in FRET.
The mf D was an interesting parameter because it retrieves information about a known threshold of interacting donor protein. In this study, mf D stands for the minimal percentage of amphiphysin 1/amphiphysin 1 interaction and the minimal percentage of amphiphysin 1/N-WASP interaction.

RESULTS
Amphiphysin 1 Is Implicated in N-WASP-dependent Actin Assembly-Incubation of liposomes containing acidic phospholipids with cytosol in the presence of ATP results in a powerful polymerization of actin that can be monitored by pyrene fluorescence assay (supplemental Fig. S1) (3,28). Time courses of the PS-or PI(4,5)P 2induced actin polymerization are shown in Fig. 1A. The actin polymerization in the presence of PS, but not PC, was inhibited by wiskostatin, an N-WASP inhibitor (Fig. 1B) (29). To determine the impact of amphiphysin 1 on this lipid bilayer-induced actin polymerization, we compared the effect of brain cytosol obtained from WT or amphiphysin 1 knock-out mice (amphiphysin 1 Ϫ/Ϫ ). Amphiphysin 1 Ϫ/Ϫ cytosol is almost completely devoid of not only amphiphysin 1 but also amphiphysin 2, a major brain-specific isoform that forms a heterodimer with amphiphysin 1 (16). PS-dependent actin polymerization in amphiphysin 1 Ϫ/Ϫ cytosol was reduced by ϳ60% compared with wild type cytosol at the 1500-s time point (Fig. 1C). The decrease was rescued by supplementing the amphiphysin 1 Ϫ/Ϫ cytosol with recombinant amphiphysin 1 (Fig. 1C). In the pres- ence of PC-liposomes, the supplementation of amphiphysin 1 showed no effect (Fig. 1C). Furthermore, addition of the same amount of amphiphysin 1 to WT cytosol did not alter PS-dependent actin polymerization (supplemental Fig. S2). Thus, amphiphysin is implicated in PS-dependent actin polymerization, and N-WASP is likely to be involved in this process.
Because both amphiphysin 1 and N-WASP are concentrated at the synapse (30, 31), we investigated whether the lack of amphiphysin affects the formation of F-actin in synaptosomes. The amount of actin in total brain cytosol was essentially the same in WT or amphiphysin 1 Ϫ/Ϫ mice (Fig. 1D, upper left  panel). In synaptosomes, however, actin was selectively reduced in amphiphysin 1 Ϫ/Ϫ (Fig. 1D, upper  right panel). Despite the reduced level of actin, the activity-dependent cycling of actin assembly was not changed in amphiphysin 1 Ϫ/Ϫ synaptosomes (Fig. 1E). Indeed, in both WT and amphiphysin 1 Ϫ/Ϫ synaptosomes, F-actin levels showed an early peak (10 s after depolarization), a drop, and a later peak (30 -40 s after depolarization), although the fluorescence levels were constantly and significantly lower in amphiphysin 1 Ϫ/Ϫ synaptosomes than in WT ones (Fig. 1E,  left panel). The parallel assay of the cycling of DNase I-labeled G-actin revealed, in both genotypes, an opposite pattern of cycling and lower G-actin levels in amphiphysin 1 Ϫ/Ϫ synaptosomes (Fig. 1E, right  panel). These data strongly support an implication of amphiphysin 1 in actin assembly and dynamics in the synapse.

Amphiphysin 1 Directly Stimulates N-WASP-dependent Arp2/3
Actin Nucleation-We next explored a potential interaction of amphiphysin 1 with N-WASP using pulldown assays from neurons and testicular Sertoli cells. Amphiphysin 1 pulled down N-WASP both from extracts of rat brain synaptosomes ( Fig. 2A) and from extracts of Ser-W3 cells, a rat Sertoli cell line, expressing Myc-tagged N-WASP (Fig. 2B). The interaction was mediated by amphiphysin 1 SH3 domain, as was the case for interaction with dynamin, a physiological binding partner of amphiphysin (Fig. 2, A and B) (11). A trace amount of N-WASP binding was detected by GST-Amph-⌬SH3 (Fig. 2B), which can be attributed to dimer formation of GST-Amph-⌬SH3 with endogenous amphiphysin 1. Direct interaction between N-WASP and amphiphysin 1 was demonstrated by pulldown assay using purified proteins (Fig. 2C).
To determine whether amphiphysin 1 directly regulates N-WASP-dependent Arp2/3 actin nucleation, formation of F-actin was quantified in an in vitro assay, in which pyreneconjugated actin was incubated with N-WASP, G-actin, Arp2/3 complex, and amphiphysin. The N-WASP-triggered actin assembly was enhanced by amphiphysin 1 in a dose-dependent manner (Fig. 3A). In absence of N-WASP, amphiphysin 1 had no stimulatory effect on the actin assembly (Fig. 3A). Furthermore, in vitro actin polymerization assay using amphiphysin 1-deletion mutants revealed that both the N-BAR and SH3 domain of amphiphysin 1 were required for the stimulation of actin assembly. Neither the SH3 alone nor an amphiphysin 1 construct lacking the N-BAR domain (⌬N-BAR) stimulated actin assembly (Fig. 3C).
Amphiphysin 1/N-WASP Interaction Takes Place at Cell Periphery-To elucidate physiological significance of amphiphysin 1/N-WASP interaction, in vivo interaction of these molecules was examined in Ser-W3 cells. Previously, we have reported that amphiphysin 1 stimulates actin polymerization, which in turn supports ruffle formation during phagocytosis. Such ruffle formation can be induced in Sertoli cells by stimulating surface PS receptors with PS-containing liposomes (3). In PS-stimulated Ser-W3 cells, amphiphysin 1 localized at ruffles (Fig. 4) as reported previously (3). Double immunofluorescence staining of the cell revealed co-localization of amphiphysin 1 and myc-N-WASP at ruffles in particular at their very edge (Fig. 4).
Next, physiological relevance of the amphiphysin 1-N-WASP complex was examined in PS-stimulated Ser-W3 cells.
Dysfunction of N-WASP by wiskostatin drastically reduced PSdependent ruffle formation (Fig. 5A). Expression of mCherry-PRD of N-WASP inhibited recruitment of amphiphysin 1 to the cell periphery and reduced the ruffle formation more than 50% (Fig. 5B). Expression of ⌬N-BAR of amphiphysin 1 also strongly inhibited the ruffle formation (Fig. 5C). These results suggest that amphiphysin 1 and N-WASP, at least partially, contribute to PS-dependent ruffle formation.
To investigate the temporal and spatial association of amphiphysin 1 with N-WASP in Ser-W3 cells, FRET-FLIM was performed (27,32). First, we proved sensitivity of the method by demonstrating the homotypic interaction of amphiphysin 1 to form a homodimer. Homodimerization of amphiphysin 1 has been shown by in vivo and in vitro experiments (33). This assay allows monitoring the spatio-temporal decrease of the mean fluorescence lifetime of GFP-tagged amphiphysin 1 because of the interaction of this protein with an mCherry-tagged partner. GFP-amphiphysin 1 and amphiphysin 1-mCherry were co-expressed in Ser-W3 cells, and the fluorescence lifetime of GFPamphiphysin 1 was acquired. The average mean lifetime of GFP-amphiphysin 1 decreased from 2.48 Ϯ 0.01 ns in cells expressing GFP-amphiphysin 1 alone (n ϭ 10) to 2.40 Ϯ 0.03 ns in co-transfected cells (n ϭ 14) (Fig. 6). This decrease in lifetime is characteristic of FRET between GFP and mCherry. Considering the fact that amphiphysin 1 acts as both donor and acceptor, the decrease of the mean lifetime of GFP could be underestimated more than the actual homotypic interaction. The high resolution map of the homotypic interaction can be displayed by using the minimal fraction of donor protein involved in FRET (mf D ) (Fig. 6) (27) (see "Experimental Procedures"). The mf D map represents the spatial distribution of the relative variation of the effective amount of interaction. In this case, the quantification is underestimated, and the three-dimensional representation highlights the differences from the control (Fig.  6A, right panel).
Next, Ser-W3 cells expressing GFP-amphiphysin 1 and mCherry-N-WASP were imaged using the FRET-FLIM assay. In the absence of PS, only one co-expressing cell showed FRET (n ϭ 4). The average mean lifetime decreased from 2.48 Ϯ 0.01 ns in cells expressing GFP-amphiphysin alone to 2.44 Ϯ 0.01 ns in the FRET-positive cell (S.D. calculated from the spatial distribution within the single cell). In the presence of PS, nine cells expressing GFP-amphiphysin 1 and mCherry-N-WASP showed FRET (n ϭ 10), and the average mean lifetime decreased to 2.42 Ϯ 0.02 ns (S.D. calculated from the spatial distribution within the nine cells).
A representative FRET-positive cell in the presence of PS is shown in Fig. 7A (arrowhead in right bottom panel). This demonstrates that the interaction is localized at a restricted area of the cell periphery. The specificity of this interaction is revealed by the fact that cells co-expressing GFP-amphiphysin with N-WASP-mCherry (position of the acceptor changed) did not show FRET either in the absence or in the presence of PS. In this configuration, when mCherry is placed C-terminal to the N-WASP, the acceptor is likely too far away from GFP-amphiphysin or in a wrong orientation respective to the donor to allow FRET. The fact that we did not observe a significant change in the lifetime upon PS addition (2.48 Ϯ 0.01, n ϭ 5) using GFP-amphiphysin/N-WASP-mCherry shows the specificity of PS addition upon the FRET signal occurring between GFP-amphiphysin and mCherry-N-WASP.
Specificity of this method was further strengthened, by confirming that the FRET signal was absent in other negative control cells as follows. In Sertoli cells co-expressing GFPamphiphysin and mCherry, the average mean lifetime after PS stimulation was 2.47 Ϯ 0.01 ns (n ϭ 5). Thus, it was very likely that the FRET signal occurring between GFP-amphiphysin and mCherry-N-WASP represents the interaction between amphiphysin and N-WASP in living cells.
Using this approach, we were able to quantify and follow the interaction in time in the presence of PS. For each cell analyzed, 20 FRET-FLIM images were sequentially acquired every 12 s, the time required for one image acquisition. From these FRET-FLIM time lapses, the spatio-temporal variations of the amount of the interaction were imaged by mf D time lapses (for one example see supplemental Movie S1). Snapshots of the interaction are displayed in Fig. 7C for sequential time lag numbers 5-7 over the 20 acquisitions. These mf D images reveal the localized and transient interaction between GFP-amphiphysin 1 and mCherry-N-WASP. Even if we measure the minimal fraction of donor engaged in FRET, because the GFP/mCherry ratio likely does not change during time lapse course (4 min), the spatiotemporal changes of mf D within single cells represent the dynamic changes of the amount of interacting donor. The time axis profile of one representative time-lapse acquisition of a co-transfected cell in the presence of PS are shown in Fig. 7D.
This profile shows that the mean amount of interacting donor increases by a factor of 2 within 1 min (Fig. 7D, 1st peak) followed by a decrease in 12 s. After a stable period for 1 min, rapid increase to the second peak is observed (Fig. 7D). The rough correlation of this stimulated interaction with the time course of ruffle formation, within 10 min, suggests a role in the control of the actin nucleation that underlies their generation and dynamics. Thus, these findings provide a mechanistic explanation for our previous observation that PS-stimulated ruffling and phagocytosis was strongly impaired in Sertoli cells of amphiphysin 1 Ϫ/Ϫ mice.

DISCUSSION
Until this study, although a functional link between amphiphysin and actin had been identified, the underlying mech- were obtained using TriMscope as described under "Experimental Procedures." Three-dimensional representation of mf D images was obtained using a threshold limit given by the control (0.2). GFP-lifetime decreased in the presence of amphiphysin 1-mCherry. Bar, 10 m. B, lifetime (left panel) and mf D histograms (right panel) of the control (green) and the co-transfection (red) cells show a decrease of the GFP lifetime from 2.48 ns (control) to 2.41 ns (co-expression) and the mean value of mf D from 0 (control) to 0.11 (co-expression).
anistic link had remained elusive (12)(13)(14)(15). Our results prove that amphiphysin, like other SH3 domain-containing proteins, functions as an activator of N-WASP, although additional more indirect connections to actin function are also possible. Several other SH3 domain-containing proteins that, like amphiphysin, also contain modules of the BAR domain superfamily were previously shown to stimulate actin nucleation via N-WASP and Arp2/3 in vitro (34 -37). Furthermore, we have used FRET-FLIM to provide evidence that amphiphysin 1/N-WASP interaction occurs in a living cell and that it is enhanced at the cell periphery by a stimulus that induces ruffles formation. The multiplicity of factors that regulate the activation of N-WASP is explained by the many different contexts in which N-WASP must function.
The profound impact of amphiphysin 1 on actin nucleation in brain and Sertoli cells, as genetically shown by studies in synaptosomes (this study) or in Sertoli cells (3) of amphiphysin 1 Ϫ/Ϫ mice, likely reflects the abundance of amphiphysin 1 in these cells. Amphiphysin 1 Ϫ/Ϫ mice exhibit cognitive defects (16). It is therefore of interest to note that mutations in several genes encoding actin-regulatory proteins, including the BAR and SH3 domain-containing protein oligophrenin 1 (7,38,39), are responsible for inherited cases of mental retardation in humans.
Stimulation of N-WASP-dependent actin assembly required, surprisingly, not only the SH3 domain but also the N-BAR domain (Fig. 3C). The BAR domain is responsible both for homo-or heterodimerization of amphiphysin (33) and for lipid bilayer binding (7,8). The property of amphiphysin 1 stimulates N-WASP-dependent actin assembly even in the absence of liposomes (Fig. 3, A and C). These results strongly suggest that dimerization but not membrane binding may be crucial for the activation. One potential explanation of this finding is that dimerization may in turn promote the clustering and dimerization of N-WASP. A recent study has shown that full activation of N-WASP requires its dimerization following its allosteric relief of autoinhibition by regulatory factors such as SH3 domains and that dimeric SH3 domains are much more powerful activators than monomeric SH3 domains (40). In any case, because full N-WASP activation requires its binding to PI(4,5)P 2 in the plasma membrane, the physiological sites of these interactions is the cell cortex. Several previous studies of amphiphysin 1 have focused on its role in endocytosis.
The current model is that the curvature-generating and curvature-sensing properties of the N-BAR domain of amphiphysin mediate its accumulation at the neck of endocytic pits. In endocytosis, amphiphysin 1 couples clathrinmediated budding to fission and clathrin uncoating. In the former process, amphiphysin 1 binds to clathrin and AP-2, the clathrin adaptor, and in the latter process it binds to dynamin and to the PI(4,5)P 2 phosphatase synaptojanin via its SH3 domain (6,11). However, it is now clear that actin and endocytosis, including clathrin-mediated endocytosis, are intimately interconnected and that actin also assembles at the neck of at least a subset of clathrin-coated pits (Fig. 8) (15,41). Thus, amphiphysin may function in cooperation with other BAR superfamily proteins that also contain SH3 domains to induce curvature-dependent actin polymerization at the neck of endocytic pits (34 -37). Evidence for a role of amphiphysin in ruffles and the occurrence of isoforms of amphiphysin that lack the clathrin and AP-2-binding motifs indicate that endocytosis-independent actions may occur (42,43). Our identification of a role for the BAR domain of amphiphysin independent from bilayer binding further supports this possibility. In conclusion, we have shown here that amphiphysin 1 is an important regulator of actin polymerization through its interaction with N-WASP and that this interaction is regulated in time and space.