Association of the AP-3 Adaptor Complex with Clathrin

See allHide authors and affiliations

Science  17 Apr 1998:
Vol. 280, Issue 5362, pp. 431-434
DOI: 10.1126/science.280.5362.431


A heterotetrameric complex termed AP-3 is involved in signal-mediated protein sorting to endosomal-lysosomal organelles. AP-3 has been proposed to be a component of a nonclathrin coat. In vitro binding assays showed that mammalian AP-3 did associate with clathrin by interaction of the appendage domain of its β3 subunit with the amino-terminal domain of the clathrin heavy chain. The β3 appendage domain contained a conserved consensus motif for clathrin binding. AP-3 colocalized with clathrin in cells as observed by immunofluorescence and immunoelectron microscopy. Thus, AP-3 function in protein sorting may depend on clathrin.

The formation of vesicles for transport between organelles of the endocytic and secretory pathways and the selection of cargo for packaging into those vesicles are mediated by protein coats associated with the cytosolic face of the organelles (1). The best characterized coats contain clathrin and protein complexes termed “adaptors” (2). Clathrin is a complex of three heavy chains and three light chains that polymerizes to form the scaffold of the coats. The adaptors mediate attachment of clathrin to membranes and recruit integral membrane proteins to the coats. Two clathrin adaptors have been described to date—AP-1 and AP-2, which participate in protein transport to the endosomal-lysosomal system from the trans-Golgi network (TGN) and the plasma membrane, respectively.

Recently, another complex related to AP-1 and AP-2 has been identified. This complex, termed AP-3, is composed of two large subunits (δ and β3A or β3B), a medium-sized subunit (μ3A or μ3B), and a small subunit (σ3A or σ3B) (3-8). AP-3 has been localized to the TGN and endosomes (4, 5) and is involved in protein trafficking to lysosomes or specialized endosomal-lysosomal organelles such as pigment granules, melanosomes, and platelet dense granules (8, 9). Previous studies suggested that AP-3 is a component of a nonclathrin coat (3, 4). However, a critical question has remained unanswered: does AP-3 assemble with a structural coat protein that plays the role of clathrin?

We used glutathione S-transferase (GST) fusion proteins bearing the H (hinge) and C (COOH-terminal) segments of the “appendage” region of human β3A (6) to search for this putative protein (Fig. 1A). The reason for selection of these constructs was that both AP-1 and AP-2 bind clathrin via the appendage regions of the β1 and β2 subunits (10, 11). Affinity purification from a cytosolic extract of a human T cell line, Jurkat, resulted in isolation of a prominent 180-kD protein on GST-β3AC but not GST-β3AHor GST columns (Fig. 1B). Unexpectedly, microsequencing of eight tryptic peptides derived from the 180-kD protein identified it as the clathrin heavy chain (Fig. 1B) (12). Binding of clathrin to GST-β3AC could also be demonstrated by immunoblotting with antibodies to either the heavy (Fig. 1C, IB, and Fig. 1D, TD.1) or light (Fig. 1D, CON.1) chain of clathrin as well as by immunoprecipitation of metabolically labeled proteins (Fig. 1C, IP). Furthermore, the extent of clathrin binding to GST-β3ACwas comparable to that of a GST fusion protein having the appendage region of β2 (Fig. 1D, GST-β2592–951).

Figure 1

Interaction of β3A with clathrin. (A) Schematic representation of human β3A (6) and portions of it expressed as fusions with GST (20). (B) Isolation of clathrin by affinity chromatography on a GST-β3AC column. Cytosolic proteins that bound to the indicated GST-fusion proteins (21) were analyzed by SDS–polyacrylamide gel electrophoresis (SDS-PAGE). Arrow points to a protein that bound only to GST-β3AC. The sequences of tryptic peptides derived from this protein matched the segments of clathrin heavy chain indicated in the figure (12, 13). (C) Immunoblotting (IB, TD.1 antibody) (22) and immunoprecipitation (IP) (23) analyses of the binding of clathrin from Jurkat (IB) or HeLa (IP) cytosolic extracts to the indicated GST-fusion proteins. First lane in each panel corresponds to a clathrin-positive control. (D) Interaction of clathrin from Jurkat cell cytosol with the clathrin-binding domains of β3A and β2 as analyzed by immunoblotting with antibodies to the heavy (TD.1) or light (CON.1) chains of clathrin (22). (E) Binding of clathrin from bovine brain cytosol (upper) or [35S]methionine-labeled clathrin terminal domain (24) (lower) to GST-fusion proteins bearing portions of β3A or β3B (20).

The region of β3A that binds clathrin was further delineated by using additional GST-β3A constructs (Fig. 1A). Deletion of the first 19 or 76 residues from the C domain of β3A (GST-β3A829–1094or GST-β3A886–1094) abrogated interaction with clathrin (Fig. 1E, upper). Moreover, a construct having only residues 810 to 885 of β3A was able to bind clathrin. We also analyzed the interaction of the neuronal-specific β3B subunit (3) with clathrin. As observed for β3A, the C domain, but not the H domain, of β3B interacted with clathrin (Fig. 1E, upper).

Comparison of amino acids 810 to 885 of β3A to the corresponding segment in β3B revealed three clusters of conserved residues (Fig.2A). Groups of three or four residues within the clusters were mutated en bloc to alanine residues in the context of the GST-β3AC construct. Mutations of the sequences SLL817–819 and DLD820–822 within the first cluster (13) led to a dramatic decrease in clathrin binding, whereas other mutations had partial (DFN823–825) or no effect (PVS826–828, DLE845–847, LHL849–851, LLHR873–876) (Fig. 2A) (14). Thus, the segment SLLDLDDFN817–825 of β3A contained residues that were involved in clathrin binding. Similar amino acid sequences were found within the clathrin-binding regions of β1 and β2 (11) as well as within segments of arrestin3 (15) and amphiphysin II (16) that have been demonstrated to mediate interaction with clathrin (Fig. 2B). Alignment of these sequences defined a consensus motif for clathrin binding that consisted of acidic and bulky hydrophobic residues and conformed to the canonical sequence L(L, I)(D, E, N)(L, F)(D, E).

Figure 2

(A) Delineation of a clathrin-binding site within β3A. (Upper) Alignment of residues 810 to 885 of human β3A with residues 799 to 873 of human β3B (13). The indicated three-residue groups in the β3A sequence were mutated to alanine triplets in the context of the GST-β3AC construct (20). The resulting GST-β3AC mutants, as well as GST and wild-type GST-β3AC, were assayed for clathrin binding. (Middle) Immunoblotting of bound proteins from Jurkat cell cytosol with TD.1, the antibody to clathrin. (Lower) SDS-PAGE analysis of GST-fusion proteins. (B) Sequence alignment illustrating the existence of a potential clathrin-binding motif within residues 817 to 825 of human β3A and segments of human β3B, β1, and β2; bovine arrestin3 and β-arrestin; and human amphiphysins I and II. (C) Coassembly of adaptors with clathrin. A mixture of AP-1, AP-2, and AP-3, containing BSA and GST as irrelevant control proteins, was combined with purified clathrin and dialyzed against a buffer that promotes clathrin cage formation (18). Samples without clathrin were processed as controls. Samples were sequentially centrifuged at 16,000g for 5 min to remove aggregates (P1) and at 350,000g for 10 min to collect clathrin cages and associated proteins (P2). Both P1 and P2 pellets and the final supernatant (S) were analyzed by immunoblotting with antibodies to AP-1 (100/3; Sigma), AP-2 (100/2; Sigma), AP-3 (anti-μ3) (5), BSA (Cappel), and GST (prepared in our laboratory).

AP-2 and β-arrestin bind to the “terminal domain” at the NH2-terminus of the clathrin heavy chain (17). To test whether this was also the case for β3A and β3B, we analyzed the binding of radiolabeled clathrin terminal domain to various GST-fusion proteins. Like intact clathrin, the terminal domain fragment specifically bound to the C domains of β3A and β3B as well as to residues 810 to 885 of β3A (Fig. 1E).

To investigate whether the complete AP-3 complex interacted with clathrin, we used a coassembly assay in which clathrin cages were formed in the presence of a mixture of AP-1, AP-2, and AP-3 (18). We observed coassembly of AP-3 with clathrin cages to an extent comparable to that of AP-1 and AP-2 (Fig. 2C). Control experiments demonstrated that neither of the three adaptors sedimented in the absence of clathrin and that irrelevant proteins such as bovine serum albumin (BSA) and GST did not associate with clathrin cages (Fig.2C).

Confocal immunofluorescence microscopy analyses revealed localization of AP-3 in a juxtanuclear area as well as in scattered peripheral foci; 82 ± 6% of these peripheral foci coincided with clathrin-containing structures (Fig. 3). This colocalization was not likely due to random overlap because in many cases the same “constellations” of foci were recognizable for both AP-3 and clathrin (Fig. 3, insets). In contrast, the overlap of AP-3 with AP-2 amounted to only 12 ± 3% (14). Immunoelectron microscopy of thin cryosections revealed the presence of both AP-3 (15-nm gold particles) and clathrin (10-nm gold particles) on tubulovesicular membranes in the proximity of the Golgi stack (Fig.4A, arrowheads); labeling for AP-3 was also observed on clathrin-positive vacuolar structures reminiscent of endosomes (14). The percentage of AP-3–containing membranes that were also labeled for clathrin was 65 ± 5%. To ascertain whether AP-3 colocalized with clathrin on endosomes, we used a whole-mount immunoelectron microscopy technique that allows selective visualization of endosomes loaded with horseradish peroxidase–conjugated transferrin (19). We observed colocalization of AP-3 (10-nm gold particles) and clathrin (5-nm gold particles) on 60-nm-diameter buds at the ends of endosomal tubules (Fig. 4B, arrowheads) and, sporadically, on the vacuolar part of endosomes (14).

Figure 3

Confocal immunofluorescence microscopy analysis of colocalization of AP-3 with clathrin. Human HeLa cells were double-stained (5) for clathrin [X22 antibody followed by Cy3-conjugated anti-mouse immunoglobulin G; Jackson ImmunoResearch] (A) and AP-3 [β3C1 antibody (6) followed by Cy2-conjugated anti-rabbit immunoglobulin G; Jackson ImmunoResearch] (B). (C) Overlay of the images obtained for clathrin (red) and AP-3 (green); yellow indicates colocalization. Insets are magnified views of the smaller rectangular areas of the cytoplasm. Arrows point to structures where clathrin and AP-3 colocalize. Similar results were obtained with antibodies to other AP-3 subunits (14). Bar, 10 μm.

Figure 4

Immunoelectron microscopy analysis of the colocalization of AP-3 with clathrin. (A) Ultrathin cryosections of rat PC12 cells were double-labeled for AP-3 (β3A1 antibody) (6) (15-nm gold particles) and clathrin (polyclonal antibody provided by E. Ungewickell) (10-nm gold particles) as described (25). Arrowheads point to membranes displaying labeling for both AP-3 and clathrin. G, Golgi stack. (B) Human A431 cells were allowed to endocytose transferrin conjugated to horseradish peroxidase (19). After the peroxidase reaction, cells were permeabilized, immunolabeled for AP-3 (β3A1 antibody) (10-nm gold particles) and clathrin (X22 antibody) (5-nm gold particles), and examined by whole-mount transmission electron microscopy (19). Bars, 100 nm.

In conclusion, our observations demonstrated an association of the mammalian AP-3 complex with clathrin both in vitro and within cells. This association was mediated by interaction of the C domain of β3(A or B) with the terminal domain of the clathrin heavy chain. Residues in β3 that were involved in this interaction conformed to a consensus motif shared with β1, β2, nonvisual arrestins, and amphiphysins; conservation of this motif is likely to account for the ability of all these proteins to bind clathrin despite their structural diversity. Our findings imply that mammalian AP-3 is likely to function as a clathrin adaptor, just like AP-1 and AP-2. Thus, protein sorting events mediated by AP-3 may also depend on clathrin.

  • * To whom correspondence should be addressed. E-mail: juan{at}


View Abstract

Navigate This Article