J. Mol. Biol. (2005) 348, 491–501

Caveolin-1 Interacts Directly with Dynamin-2

Qing Yao1, Jing Chen1, Hong Cao1, James D. Orth1, J. Michael McCaffery2, Radu-Virgil Stan3 and Mark A. McNiven1*

Caveolin is the principal component of caveolae in vivo. In addition to a structural role, it is believed to play a scaffolding function to organize and inactivate signaling molecules that are concentrated on the cytoplasmic surface of caveolar membranes. The large GTPase dynamin has been shown to mediate the scission of caveolae from the plasma membrane, although it is unclear if dynamin interacts directly with caveolin or via accessory proteins. Therefore, the goal of this study was to test whether dynamin associates with caveolae via a direct binding to the caveolin 1 (Cav1) protein. Immunoelectron microscopy of lung endothelium or a cultured hepatocyte cell line stained with antibodies for Dyn2 and Cav1 shows that these proteins co-localize to caveolae. To further define this interaction biochemically, in vitro experiments were performed using glutathione-S-transferase (GST)-Dyn2 and GST-Cav1 fusion proteins, which demonstrated a direct interaction between these proteins. This interaction appears to be mediated by the proline-arginine-rich domain (PRD) of Dyn2, as a GST-PRD fragment binds Cav1 while GST-Dyn2DPRD does not. Further, in vitro binding studies using two Dyn2 spliced forms and Cav1 peptides immobilized on paper identify specific domains of Cav1 that bind Dyn2. Interestingly, these Cav1-binding domains differ markedly between two spliced variant forms of Dyn2. In support of these distinctive physical interactions, we find that the different Dyn2 forms, when expressed as GTPase-defective mutants, exert markedly different inhibitory effects on caveolae internalization, as assayed by cholera toxin uptake. These studies provide the first evidence for a direct interaction between dynamin and the caveolin coat, and demonstrate a selectivity of one Dyn2 form toward the caveolae-mediated endocytosis.

enlarge

Figure 1. Immunoelectron microscopy reveals an intimate association of dynamin with caveolae. (a)–(c) Immunogold localization of Dyn2 (MC61, polyclonal antibody) to caveolae in cultured mouse hepatocytes (HepG2). An accumulation of Dyn2 labeling can be seen only at plasma membrane-associated caveolae and not deeper within the section. Dyn2 was detected by goat anti-rabbit secondary antibody coupled to 5 nm gold. (d) and (e) Double immunogold labeling of rat lung endothelium using Dyn2 (polyclonal) and Cav1 (monoclonal) antibodies. Dyn2 was detected by goat anti-rabbit secondary antibody coupled to 10 nm gold, while Cav1 was detected by secondary antibody coupled to 5 nm gold. Large Dyn2 gold particles (arrows) can be seen along uncoated, flask-shaped caveolae at the plasma membrane, often associated with the neck of the vesicle, while numerous smaller Cav1 gold particles are seen decorating the caveolar membranes. The scale bars represent 50 nm.

enlarge

Figure 2. Reciprocal affinity pulldown experiments from tissue homogenates reveal a Dyn2 and Cav1 complex. (a) GST-alone and GST-Dyn2 (aa and ab spliced forms) beads were incubated with rat lung homogenate for two hours at 4 8C. After washing, the GST, GST-Dyn2 and associated proteins were eluted, and subjected to SDS-PAGE and immunoblotting with anti-Cav1 antibodies. Cav1 associates strongly with beads coated with different spliced forms of Dyn2 (aa and ab), while Cav1 did not associate with control beads coated with GST alone. Lung, Cav1 in the starting material. (b) GST-bound and GST-Cav1-bound glutathione agarose beads were incubated with purified Dyn2(aa), Dyn2(ab), respectively, for two hours at 4 8C. After washing, the bound protein on beads was eluted by SDS sample buffer and separated by SDS-PAGE, followed by immunoblotting with a Dyn2 antibody. Dyn2(aa) and Dyn2(ab) associated strongly with the GST-Cav1 beads, but not with GST-alone beads.

enlarge

Figure 6. Peptide mapping of Cav1–Dyn2 interactions. (a) To further define the physical interactions between Cav1 and Dyn2 proteins, 18 overlapping peptides representing the entire Cav1 protein (d) were robotically spotted onto nitrocellulose (Sigma) and overlaid with expressed full-length (b) Dyn2(aa) or (c) Dyn2(bb) proteins. Following extensive washing, incubation with a Dyn2 antibody and detection using ECL, sites of Cav1–Dyn2 interactions became readily apparent. The experiments were repeated three times each, using two different membranes and quantified by densitometry. (e) Sites of interaction between the two Dyn2 forms were similar but distinct with Dyn2(aa) showing enhanced binding at four of five putative Cav1 binding sites. (d) Histogram indicating relative strength of interaction sites aligned along peptide domains of the full-length Cav1 protein.

enlarge

Figure 7. Dyn2 spliced forms show distinct inhibitory effects on caveolae-based internalization of cholera toxin. Fluorescence double-labeling of a cultured rat fibroblast cell-line expressing distinct, GTPase-defective, mutant forms of Dyn2(aa) and Dyn2(bb) K44A that were then challenged 2.5 hours later to internalize rhodaminelabeled cholera toxin B-subunit. (a)–(d) Transfected cells stained with a Dyn2 antibody to identify mutant expressing cells. (a0)–(d0) Corresponding images of Dyn2 mutant expressing cells that were rinsed and fixed following 15 minutes of exposure to toxin at 4 8C. Asterisks (*) indicate transfected cells and arrows point to perinuclear Golgi and endocytic organelles to which internalized toxin is transported. (e) Fluorescence quantitation showing differential effects of the Dyn2 mutant proteins on toxin uptake compared to wild-type expressing cells. (a) and (a0), (b) and (b0) Cells expressing the Dyn2(aa)K44A mutant exhibited 80% less toxin uptake than cells expressing the other Dyn2 mutant forms. The scale bars represent 10 um.