A structure-based mechanism for Arf1-dependent recruitment of coatomer to membranes

Abstract

Budding of COPI-coated vesicles from Golgi membranes requires an Arf family G protein and the coatomer complex recruited from cytosol. Arf is also required with coatomer-related clathrin adaptor complexes to bud vesicles from the trans-Golgi network and endosomal compartments. To understand the structural basis for Arf-dependent recruitment of a vesicular coat to the membrane, we determined the structure of Arf1 bound to the γζ-COP subcomplex of coatomer. Structure-guided biochemical analysis reveals that a second Arf1-GTP molecule binds to βδ-COP at a site common to the γ- and β-COP subunits. The Arf1-binding sites on coatomer are spatially related to PtdIns4,5P(2)-binding sites on the endocytic AP2 complex, providing evidence that the orientation of membrane binding is general for this class of vesicular coat proteins. A bivalent GTP-dependent binding mode has implications for the dynamics of coatomer interaction with the Golgi and for the selection of cargo molecules.

Figure 1. Assays for Coatomer Interaction with Arf1

(A) Schematic diagram of βδ/γζ-COP, the cargo-binding tetramer of the COPI coat, based on the expected close similarity to AP adaptor complexes. Small (30 kDa) appendage domains on β- and γ-COP are joined to the α-solenoid (also called trunk) domains via flexible linkers. β- and γ-COP are related in sequence and structure. Likewise ζ-COP is related to the N-terminal domain of δ-COP, but the C-terminal domain of δ-COP (equivalent to the C-terminal domain of the μ2-AP subunit of AP2) breaks the —symmetrical relationship of γζ-COP and βδ-COP. (B) The fluorescence GTPase assay. The graph shows time courses for four representative reactions. The fluorescence output of a solution of 0.5 μM soluble (lacking membrane-anchor residues 1–17) human Arf1-mant-GTP or Arf1-mant-GppNHp was monitored upon addition of various forms of coatomer, at a fixed concentration of GAP (40 μM human GAP1 catalytic core, residues 1–136). The left-hand portion of the curves, drawn as a dotted line, indicates the order of addition of the various protein components. A control experiment (orange curve) shows Arf1-mant-GppNHp in the presence of GAP and 0.1 μM βδ/γζ-COP; there is no fluorescence decrease with this non-hydrolyzable substrate. The experimental curve in magenta shows GTP hydrolysis on Arf1-mant-GTP caused by GAP alone. The curve in black shows rapid GTP hydrolysis on Arf1-mant-GTP catalyzed by GAP plus 0.1 μM βδ/γζ-COP. The blue curve shows that the cage-forming αβ′ε-COP complex (0.1 μM) has no effect on GAP-catalyzed GTP hydrolysis. Finally, the green curve shows that in the absence of GAP, βδ/γζ-COP alone has no affect on GTP hydrolysis. (C) The graph shows least-squares fits to first order exponentials (red lines), using data (black lines) from (B). In these experimental conditions (see Experimental Procedures), the rate constant is 0.016 s⁻¹ in the absence of βδ/γζ-COP, and 0.32 s⁻¹ in the presence of 0.1 μM βδ/γζ-COP. (D) Binding and elution (pull-down) assay to test the interaction between coatomer and yeast Arf1. GST or GST-Arf1 (residues 18–181) protein complexed with GDP, GppNHp or GTP was immobilized on glutathione sepharose beads. The beads were incubated with purified βδ/γζ-COP and then washed. (Total bound proteins are shown in Figure S1). GAP protein was added to trigger GTP hydrolysis and elution of βδ/γζ-COP. Eluted protein was analyzed by SDS-PAGE and Coomassie blue staining. The Arf1-GST bands released in the elution step represent only ~2.5% of the bead-bound Arf1-GST. See also Figure S1.

Figure 2. Identification of Arf1-binding sites on βδ/γζ-COP

(A) GTP-dependent binding of Arf1 to γζ-COP dimers. The pull-down assay was utilized to test the Arf1-binding capacity of truncated forms of γ-COP. In the experiments, γ-COP was present in a dimer with full-length ζ-COP (lanes 1–6) or truncated (1–153) ζ-COP (lanes 7–14). All of the dimeric proteins, from the longest to the shortest forms (left to right), bind to Arf1. (B) GTP-dependent binding of Arf1 to βδ-COP dimers. Arf1 binds to full-length βδ-COP (lane 2) and to a βδ-COP dimer lacking the β-COP appendage domain (lane 4). (C) Various forms of γζ-COP, including the minimal γζ-COP dimer (γ-COP 1–355 and ζ-COP 1–153) identified in (A), are fully active in the fluorescence assay—that is, all synergize with GAP to catalyze GTP hydrolysis on Arf1. For clarity, the curves have been offset incrementally along the x-axis. (D) The βδ-COP dimer can also synergize with GAP to catalyze GTP hydrolysis on Arf1. Curves have been offset incrementally along the x-axis.

Figure 3. Crystal Structure of γζ-COP Complexed with Arf1-GppNHp

(A) Ribbon diagram with γ-COP colored green, ζ-COP blue and Arf1 gold. The α helices of the γ-COP α-solenoid domain are labeled according to the scheme introduced by Collins et al. (2002) for the AP2 adaptor complex (γ-COP is related to the α-adaptin subunit of AP2). (B) Close up view in the same orientation as (A). The switch I and II elements of the G protein are indicated (red and cyan, respectively), and the side chains of several key interfacial residues are included (a more detailed analysis of interfacial residues is presented in Figure 4). (C) Structural overlap of γζ-COP and the corresponding region of the ασ-adaptin dimer of AP2 (Collins et al., 2002). γ- and ζ-COP are colored green and blue as before; σ2-adaptin is orange and α-adaptin is light brown. (D) A composite molecular model that includes the GAP catalytic domain of ASAP3, taken from the crystal structure of ASAP3 bound to Arf6 (Ismail et al., 2010). The model was created by superimposing the Arf6 and Arf1 structures. The picture is oriented as in (A).

Figure 4. Interacting Surfaces of γζ-COP and Arf1

(A) Close-up view showing the interface of γ-COP (green) and Arf1 (gold). Side-chain groups from γ-COP that reside at the interface (within 3.5Å of Arf1) are drawn in blue. One γ-COP residue not at the interface, Leu128, is colored purple; this residue was chosen as a control mutation for experiments shown in (C) and (E). The picture is rotated approximately 180° around a vertical axis relative to Figure 3A. (B) Close-up view in a similar orientation to (A), with side-chain groups from Arf1 colored brown. In the Arf1 mutagenesis experiments, residue I43 (colored cyan) was chosen as a control mutation since it is distant from the protein-protein interface. (C) Bar graph shows the effects of mutating γ-COP interfacial residues on the interaction between γζ-COP and Arf1, as measured by the pull-down assay. Mutations were introduced into full-length γ-COP. Binding is expressed as a percentage of the binding of wild type γζ-COP. The control γ-COP mutant, L128E, corresponds to the side chain colored purple in (A). (D) Bar graph shows the effects of mutations in S. cerevisiae Arf1 on the interaction with full-length γζ-COP. Note that the interfacial residues chosen for mutagenesis—F51, L77 and Y81—are identical in human Arf1. The Arf1 control mutant I43E (cyan) is a valine residue in the human Arf1 sequence. (E) The ability of γ-COP mutants to synergize with GAP in GTP hydrolysis was measured using the fluorescence assay. The effect of the γ-COP mutations is essentially the same as in the pull-down assay (C). Curves have been offset incrementally along the x-axis.

Figure 5. Mapping the Arf1-binding site on βδ-COP

(A) Sequence- and structure-based alignment of the large subunits of COPI and AP complexes, in the vicinity of the α4- and α6-helices of the α-solenoid domain. The three upper sequences are aligned based on a structural overlap of γ-COP (this study), α2-adaptin (from the crystal structure of AP2 determined by Collins et al. [2002]) and γ1-adaptin (from the AP1 crystal structure determined by Heldwein et al. [2004]). The sequence of β-COP is aligned to the others based on sequence homology including the conservation of helical repeats of the α-solenoid fold. The residues of γ-COP and β-COP selected for mutagenesis are colored red. The location of β-COP residue Pro56 implies that the α4 helix will start downstream of this residue in β-COP; consistent with this, the key interface residues of γ-COP are located toward the C-terminal end of the α4 helix. (B) Bar graph shows the effects of β-COP mutations on the interaction between βδ-COP and Arf1, as measured by the pull-down assay. Mutations were introduced into full-length β-COP protein. The β-COP residue L124 was selected as a site distant from the interface with Arf1; this residue corresponds to the mutation L128E on γ-COP (see Figure 4A). (C) Bar graph shows the effects of mutations in S. cerevisiae Arf1 on the interaction with full-length βδ-COP. This is the same set of mutations that were used to test γζ-COP interactions in Figure 4D. (D) The ability of β-COP mutants to synergize with GAP in GTP hydrolysis was measured using the fluorescence assay. The effect of the β-COP mutations is essentially the same as in the pull-down assay (B). Curves have been offset incrementally along the x-axis. (E) Summary of the mapping experiments. Arf1 probably binds in a similar manner to γ-COP and β-COP, whereby the Arf1 switch regions interact with residues of the α4- and α6-helices of the α-solenoid domain.

Figure 6. Model for Membrane Recruitment of Coatomer

(A) Left panel shows a composite model of βδ/γζ-COP bound to membrane via two molecules of Arf1-GTP. The γζ-COP/Arf1 crystal structure is colored (following the scheme used in Figure 3) as is the second molecule of Arf1. The remainder of βδ/γζ-COP, in grey, is modeled based on homology with the AP2 adaptor complex; specifically, we used the crystal structure of the open conformation of AP2 (Jackson et al., 2010). The N-terminal amphipathic α helices of Arf1 (colored red) are modeled in their expected locations as membrane anchors. The right panel shows the structure of the open form of AP2 (as described by Jackson et al. [2010]) modeled in an interaction with membrane via a PtdIns4,5P₂ headgroup (van der Waals spheres colored red) bound to the primary site on α-AP and a YxxΦ cargo sorting motif (stick representation colored red) that interacts with the μ2-AP subunit. For clarity, we have drawn AP2 in the same orientation as βδ/γζ-COP; this required only a slight shift of the membrane-bound orientation proposed by Jackson et al. (2010). See also Figure S2. (B) Effects of mutations in the Arf1-binding sites of full-length βδ/γζ-COP complex, measured using the pull-down assay. Single mutations reduce but do not abolish Arf1 interaction (lanes 3 and 4), whereas a double mutation binds to Arf1 at background levels (compare lane 5 to lane 1). (C) The effects of single and double mutations in the βδ/γζ-COP complex on GTP hydrolysis in the fluorescence assay.