Developmental Cell 10, 329–342, March, 2006 ª2006 Elsevier Inc.DOI 10.1016/j.devcel.2006.01.016
Molecular Switches Involving the AP-2
b2 Appendage Regulate Endocytic Cargo
Selection and Clathrin Coat Assembly
Amie L. Steinhauser,2Brett M. Collins,1Robyn Roth,3
John E. Heuser,3David J. Owen,1,*
and Linton M. Traub2,*
1Cambridge Institute for Medical Research
University of Cambridge
Cambridge CB2 2XY
2Department of Cell Biology and Physiology
University of Pittsburgh School of Medicine
Pittsburgh, Pennsylvania 15261
3Department of Cell Biology and Physiology
Washington University School of Medicine
St. Louis, Missouri 63110
the repertoire of endocytic cargo sorted into clathrin-
coated vesicles beyond the transmembrane proteins
that bind physically to the AP-2 adaptor. LDL and
GPCRs areinternalized byARHand b-arrestin,respec-
to the AP-2 b2 appendage platform via an a-helical
[DE]nX1–2FXX[FL]XXXR motif, and that this motif also
occurs and is functional in the epsins. In b-arrestin,
this motif maintains the endocytosis-incompetent
state by binding back on the folded core of the protein
in a b strand conformation. Triggered via a b-arrestin/
GPCR interaction, the motif must be displaced and
must undergo a strand to helix transition to enable
the b2 appendage binding that drives GPCR-b-arrestin
complexes into clathrin coats. Another interaction
surface on the b2 appendage sandwich is identified
for proteins such as eps15 and clathrin, suggesting
a mechanism by which clathrin displaces eps15 to
lattice edges during assembly.
Cargo selectivity is a hallmark of clathrin-mediated ve-
sicular trafficking. However, clathrin triskelia do not
bind cargo molecules directly, so packaging of selected
transmembrane cargo into coated vesicles requires
adaptor proteins (Owen et al., 2004; Robinson, 2004;
Sorkin, 2004). At the plasma membrane, the major cla-
thrin adaptor is AP-2, a heterotetrameric complex com-
posed of large a and b2, medium m2, and small s2 sub-
units(Collins etal., 2002).
underscore the pivotal role of AP-2 in endocytosis; in-
herited mutation or targeted disruption of AP-2 subunit
genes is homozygous lethal in C. elegans (Kamikura
and Cooper, 2003; Shim and Lee, 2000), Drosophila
(Gonzalez-Gaitanand Jackle,1997),and mice(Mitsunari
et al., 2005). The biochemical properties of AP-2 (re-
viewed in Owen et al., 2004) explain the critical role of
4, 5-bisphosphate (PtdIns[4,5]P2) via the a and m2 sub-
units, to YXXØ-type sorting signals (Bonifacino and
Traub, 2003) via m2 (Owen and Evans, 1998) in a manner
ing et al., 2005), and to [DE]XXXL[LI] dileucine motifs,
probably via an a/s2 subunit hemicomplex (Janvier
b2 appendage both bind clathrin (Lundmark and Carls-
son, 2002; Owen et al., 2000), and, together, they pro-
mote polymerization of soluble triskelia into the regular
polyhedral array typical of assembled clathrin (Owen
et al., 2000). Thus, AP-2 meshes cargo capture with cla-
thrin coat assembly events at the plasma membrane to
ensure the highly selective internalization of designated
AP-2-clathrin triad, however. At least 20 additional endo-
cytic ‘‘accessory’’ factors (Slepnev and De Camilli, 2000)
contribute to clathrin-mediated endocytosis, and many
of these factors are clathrin adaptors themselves. Still,
AP-2 acts as a central hub, coordinating the association
(McMahon and Mills, 2004; Traub, 2005). The appendage
domains, which project from the heterotetrameric adap-
tor core on flexible linkers, share a common fold (Owen
et al., 1999, 2000; Traub et al., 1999) and manage these
protein-protein interactions. To bind the appendages,
short stretches of linear sequence dock transiently onto
the appendage surface; single or tandemly arrayed com-
binations of DP[FW] (Brett et al., 2002; Owen et al., 1999),
FXDXF (Brett et al., 2002), or WXX[FW]X[DE] (Jha et al.,
2004; Ritter et al., 2003; Walther et al., 2004) interaction
motifs are found in various accessory proteins, and
each of these motifs binds to the a appendage in an ex-
tended conformation (Brett et al., 2002; Mishra et al.,
2004; Praefcke et al., 2004; Ritter et al., 2004).
A subset of the AP-2 appendage binding accessory
proteins appears to be dedicated clathrin-associated
sorting proteins (CLASPs) that expand the selection
capability of the lattice by binding, like AP-2, to
PtdIns(4,5)P2, clathrin, and a discrete class of cargo
(Robinson, 2004; Traub, 2005). Of the CLASPs, two, the
b-arrestins that sort ligand-activated G protein-coupled
receptors (GPCRs) and the autosomal recessive hyper-
cholesterolemia (ARH) protein that sorts the FXNPXY-
type sorting signal found in the LDL receptor, display
absolute selectivity for the b2 appendage over the a ap-
pendage (He et al., 2002; Laporte et al., 2002; Mishra
anisms responsible for these interactions; the crystal
structure of an b2 appendage-ARH peptide complex
shows that these, and the epsin CLASPs, utilize a con-
sensus [DE]nX1–2FXX[FL]XXXR sequence that adopts
an a-helical conformation to engage b2, explaining the
good affinity and selectivity. We also show that, in b-
arrestin, this interaction sequence functions as a pivotal
molecular switch, allowing only activated b-arrestin to
enter coated structures. Finally, we demonstrate a sec-
ond binding site on the b2 sandwich subdomain that
functions in clathrin lattice assembly.
Results and Discussion
The b2 appendage binding region of ARH is contained
within 16 residues near the C terminus of the 308 amino
understand the structural basis for the highly selective
association of ARH with b2, we crystallized the append-
age in complex with the ARH-derived peptide ligand
252DDGLDEAFSRLAQSRT and determined the structure
at 1.6 A˚resolution by molecular replacement with the
unliganded b2 appendage as a search model (Figure 1A;
ative to the central phenylalanine, designated F0, which
is invariant in all of the sequences examined.
A Helical ARH Peptide Binds the b2 Appendage
The structure of the b2 appendage-ARH complex re-
veals the molecular details for selective binding of the
The peptide can be unambiguously located on the b2
platform subdomain in the electron density map after
molecular replacement (Figures 1A and 1B; Figure S1
line] shows the initial experimental electron density
an a-helical structure that inserts several residues into
an extensive interaction surface (Figure 2). There is ex-
tremely good shape complementarity (Sc value of 0.8,
with a perfect fit being 1 [Lawrence and Colman, 1993])
between the ARH a helix and a deep groove on the top
of the platform subdomain that it occupies, burying
a total surface area of 1333 A˚2. Superposition of the
ARH-liganded and -unliganded structures of the b2
appendage (Figure S2) shows that the helical motif
engages a preformed binding site (233 Caatoms aligned
with an rmsd of 0.48 A˚).
Molecular Basis for b2 Platform Selectivity
Despite limited sequence identity, the b2 and a append-
ages have a similar bilobal structure (Owen et al., 1999,
of the platforms with respect totheir sandwich domains;
Figure S3). Comparison of the ARH b2 appendage bind-
ing site with those for DP[FW] and FXDXF motifs on the
a appendage shows that all three peptides engage
Figure 1. ARH-b2 Appendage Cocrystal
(A)Orthogonalviews of theb2appendage incomplexwith ARH pep-
bon representation that is colored from light to dark gray from the
N terminus to the C terminus. The ARH peptide is colored purple,
and side chains are shown in ball-and-stick representation.
(B) Refined 2Fo2 Fcelectron density map of the ARH peptide bound
to the platform subdomain (contoured at w1.3 s).
Table 1. Summary of Crystallographic Analysis and Refinement
a, b, c (A˚)
a, b, g (º)
Resolution range (A˚)a
37.75, 36.31, 98.98 37.65, 36.39, 97.19
90, 92.91, 90
90, 92.70, 90
Resolution range (A˚)
Number of reflections
in working set/test set
Number of atoms
Average B factors (A˚2)
Bond lengths (A˚)
Bond angles (º)
aNumbers in parenthesis refer to highest-resolution shell.
bRmerge= SjIo2 <I>j/SIo, where Iois the intensity measurement, and
<I> is the mean intensity for multiply recorded reflections.
cRcrystand Rfree= SkFoj 2 jFck/jFoj for reflections in the working and
test sets, respectively.
a similar region on their target platform subdomain. In
structuraloverlays ofthetwo appendages,thephenylal-
nine of the FXDXF bind a similarly located hydrophobic
pocket to that which accommodates Leu262 (F+3 posi-
tion) of the ARH peptide, which in b2 we term the [FL]
plementary hydrophobic pocket on the b2 appendage,
termed the F pocket. These cavities are visible in a sur-
face representation of the b2 appendage platform (Fig-
ures 2C and 2D). Previous biochemical results show
that ARH binding requires b2 platform residues Trp841
and Tyr888 (He et al., 2002; Mishra et al., 2005). A
Y888V mutation is more detrimental than a W841A
mutation, which is explained by the structure: Tyr888
contributes to both the F and [FL] pockets, whereas
Trp841 lines only the [FL] pocket (Figures 2B and 2E).
The F pocket does not exist in the a appendage, as it is
filled by residues 879–884, especially by the side chain
of Val880; thus, binding of an a-helical peptide is steri-
cally blocked. Also, Tyr888, which creates one side of
the F pocket in b2, is replaced with a smaller valine resi-
due in a.
The side chain of the ARH peptide F+7 residue Arg266
extends along a small channel on the surface of the b2
platform subdomain, which we term the R pocket, form-
tor to Y888V being more deleterious than a W841A sub-
sitionally stabilizing, Glu902, which itself contacts ARH
peptide (F+7) Arg266. The significance of electrostatic
interactions mediated by ARH peptide Arg266 was
probed with ITC by measuring the affinity of ARH for b2
appendage E849A and E902A mutants. Both mutations
effect (Figure 3). Glu849 is conserved between a and b2
appendages, but Glu902 is replaced by Arg905 in the
a appendage, where it contacts the aspartate residues
in a appendage platform binding motifs. This explains
the preference of the b2 appendage platform for motifs
containing a conserved basic arginine residue.
The F, [FL], and R pocket interactions can only occur
simultaneously when thehelical motiffits intoits binding
groove, providing the specificity for binding. The helicity
specificity-determining residues. Itisineffectanother 3-
pin plug in a socket interaction similar to ØXXYXXØ mo-
actions is in the 1–10 mM range (Honing et al., 2005).
utilized is a b strand, as opposed to the a helix for the b2
a and b2 appendages, the individual platform subdo-
mains have thus evolved to recruit distinct motif-con-
taining binding partners via predominantly hydrophobic
Figure 2. Structural Features of the ARH-b2
interacting residues from ARH, b-arrestins 1
and 2, and epsins 1 and 2. Residues in the
b2 appendage that make important interac-
tions with ARH side chains are shown below.
(B) The b2 appendage platform subdomain in
complex with the ARH peptide. Residues in
the b2 appendage and ARH that interact are
shown in ball-and-stick representation and
are colored with carbon atoms in gold and
(C) Electrostatic surface representation of the
b2 appendage platform subdomain (colored
from 27.8 to +7.8kT/e [red to blue]) in com-
plex with the ARH peptide (main chain and
side chain atoms are represented by ribbons
and bonds, respectively). The complex is
shown in the same orientation as (B).
(D) Surface representation shown in a similar
orientation as (B). The side chain binding
pockets are clearly visible.
(E) Magnification of the platform subdomain
peptide binding site highlighting residues
that line the F and [FL] pockets.
(F) Magnification of the platform subdomain
peptide binding site highlighting residues
E849 and E902 that coordinate the F+7
Arg266 of the ARH peptide, which sits in the
AP-2 b2 Appendage Function in Endocytosis
interactions with specificity deriving from a combination
of shape and electrostatic complementarity.
b-Arrestin Binding to the b2 Appendage
The ARH b appendage binding motif displays obvious
sequence similarity with the C termini of b-arrestins 1
and 2, which are known to bind the b2 appendage (Kim
and Benovic, 2002; Laporte et al., 2000; Milano et al.,
2002) (Figure 2A). All three regions are predicted to
form a short a helix embedded in an otherwise unstruc-
tured polypeptide segment, and the pivotal role of phe-
nylalanine (Phe391 in b-arrestin 1) and arginine (Arg394
tion of b-arrestins with b2 is well established (Kim and
Benovic, 2002; Laporte et al., 2000; Milano et al., 2002).
We thus expect this b-arrestin sequence to engage the
b2 appendage in a helical conformation, similar to the
ARH motif. Indeed, in a construct comprising the rele-
to GST (GST-barCT), introduction of helix-disrupting
proline residues at positions within the motif not ex-
pected to contact the b2 appendage (F+2 D390P and
F+5 R393P) strongly inhibits binding to the b2 append-
age (Figure 4A). This abrogation of binding is not due to
simple aggregation or misfolding of the GST-barCT, as
binding to clathrin via an upstream
box is unaltered (Figure 4A). The binary association of
two-hybrid interaction assay as growth on quadruple
(2Ade/2His/2Leu/2Trp) dropout plates (Figure 4B). A
helix-disrupting F+2 R393P mutation abolishes the
capability of the transformed yeast to grow under
these conditions. Also, ITC measurements show that
similar to the KDobtained for the ARH peptide (2.4 mM;
Figure 3) (Mishra et al., 2005).
Like the association with ARH, [FL] pocket mutations
(Y888V and W841A) on the b2 platform subdomain abro-
gate interactions with b-arrestins (Kim and Benovic,
(Figure 2A) predicts that b-arrestin 1 Phe391, like ARH
F+3 Leu262, occupies the [FL] pocket, making Phe391
a vital binding determinant (Figure 2A). Consequently,
Y888V or W841A disrupt the association of the two pro-
dicts that the proximal Phe388 in b-arrestin 1 (Phe389 in
b-arrestin 2) is the F0 residue and is, therefore, essential
for appendageengagementbecause ofitsability tobind
into the b2 F pocket. Indeed, mutating Phe388 in GST-
barCT abolishes interactions with the b2 appendage
(Figure 4A) and, in yeast two-hybrid assays, prevents
Figure 3. Binding of the b2 Appendage to
Various FXX[FL]XXXR Motif Peptides
(A) ARH binding to b2 appendage wild-type
and R834A, E849A, and E902A platform
(B)b-arrestin 1binding tob2appendage wild-
type and R834A, E849A, and E902A platform
(C) Epsin 1 binding to b2 appendage wild-
type and the E849A platform mutant.
(D) Equilibrium dissociation constants (KD)
and reaction stoichiometries (n).
Figure 4. b-Arrestin Binding to the b2 Appendage
(A) Aliquots of 100 mg GST (lanes a and b); GST-barCT (lanes c and d); GST-barCT F388A (lanes e and f),D390P (lanes g and h), R393P (lanes i and
Coomassie blue or transferred to nitrocellulose. The blot was probed with the anti-clathrin heavy chain (HC) mAb TD.1 and the anti-b subunit
(B) Yeast-two hybrid analysis of b2 appendage-partner interactions. Saccharomyces cerevisiae strain AH109 transformed with the indicated
pGBKT7bindingdomain (BD)andthe pGADT7 activationdomain(AD)plasmid combinationswerespottedonto SDminimalmediumplateslack-
ing either Leu and Trp or Ade, His, Leu, and Trp and grown at 30ºC.
(C) Intramolecular binding of the b2 appendage binding sequence (colored purple), which now adopts a b strand conformation, to the folded
domains of b-arrestin (colored from pale green [N terminus] to dark green [C terminus]) (PDB ID 1JSY). The polar core is circled (black, broken
line). Important side chains in the polar core are shown in ball-and-stick representation, while disordered regions of polypeptide not visible inthe
structure are indicated by the dotted, green lines.
(D) Molecular details of the interaction between the b2 appendage binding sequence of b-arrestin 1 and the folded portion of b-arrestin (PDB
code: 1G4M) (colored as in [C]). Residues important in the interaction are shown in ball-and-stick representation. During activation, Lys10
and Lys11 rotate toward the incoming receptor phosphates, destabilizing stand b1 as a prelude to ejection of the FXX[FL]XXXR sequence.
(E) Surface representation of the interaction shown in (C) shown in same orientation as (D); the b2 appendage binding sequence is again colored
(F) Aliquots of 50 mg GST (lanes a, b, e, and f) or the GST-b2 appendage (lanes c and d, g and h) immobilized on GSH-Sepharose were incubated
with lysate from HeLa cells transfected with either wild-type FLAG-b-arrestin 1 (a–d) or the activated R169E mutant (e–h). Portions of the super-
natant (S, 2%) and washed pellet (P, 10%) were resolved on SDS-PAGE and were either stained with Coomassie blue or transferred to nitrocel-
lulose. The blot was probed with an anti-b-arrestin 1 mAb. The asterisk indicates the FLAG-b-arrestin 1.
AP-2 b2 Appendage Function in Endocytosis
growth on quadruple dropout plates (Figure 4B). The
alignment also suggests that of the two potential argi-
nines in the b-arrestin sequences linked to binding to
the b2 appendage (Laporte et al., 2000), the second
(Arg395, at the F+7 position) is involved in directly con-
tacting Glu902 and Glu849 at the base of the R pocket
on the b2 appendage. Again, in support of this model,
we confirm that an R395A substitution in GST-barCT
completely abolishes binding to the AP-2 b2 subunit
(Kim and Benovic, 2002; Milano et al., 2002) (Figure 4),
and we show that the R pocket mutants E849A and
E902A reduce the affinities for the ARH and b-arrestin
data confirm that both ARH and the b-arrestins utilize
a helical FXX[FL]XXXR motif to engage the b2 append-
age platform with very similar, relatively high affinities.
Activation-Dependent Coupling of b-Arrestins
with the Clathrin Coat Machinery
Numerous time-resolved imaging studies show that
within seconds of agonist addition, diffusely distributed
b-arrestin translocates to preformed clathrin-positive
structures at the cell surface as the CLASP gathers acti-
vated GPCR into coated vesicles (Oakley et al., 2000;
Santini et al., 2002; Scott et al., 2002). This indicates
that, in the basal state, b-arrestins do not engage AP-
2/clathrin lattices effectively, despite the positioning of
a clathrin box motif (previously suggested as the b-ar-
restin endocytic trigger [Kim and Benovic, 2002]) within
a large unstructured (crystallographically invisible) loop
(Hanet al., 2001; Milano etal., 2002) that should beread-
ily accessible for interactions with the clathrin terminal
domain. The PtdIns(4,5)P2 binding site on b-arrestin
(Gaidarov et al., 1999; Milano et al., 2002) is likewise
available for membrane association and is therefore
also unlikely to drive GPCR incorporation into clathrin-
coated structures, although it may aid in localizing b-
arrestins near the plasma membrane.
b-arrestins are composed of two b sandwich subdo-
mains separated by a buried, polar hinge (Figure 4C)
that stabilizes b-arrestin in the structurally determined
‘‘closed’’ basal conformation (Gurevich and Gurevich,
2004; Milano et al., 2002). In particular, five charged
side chains in this hinge, including Arg169 and Arg393,
establish interdependent electrostatic interactions to
prevent inappropriate association of the b-arrestins
with GPCRs (Figures 4C–4E), and they also function as
the major phosphorylated peptide sensors for activated
GPCRs (Oakley et al., 1999). Importantly, our delineation
of the exact b2 appendage binding sequence in the b-
arrestins shows that, in the ‘‘closed’’ basal state (not
bound to a GPCR), this very same motif is sequestered
age binding motif is the only part of the 64 residue C-ter-
minal region of b-arrestin that is visible; in the full-length
b-arrestin 1 structures, Phe388 and Phe391, two main
specificity determinants for b2 appendage binding, en-
gage the b-arrestin core (Figures 4C–4E) (Han et al.,
2001; Milano et al., 2002). Remarkably, in the intact b-
arrestin structure, this motif does not adopt an a helix,
as we infer it does when complexed with the b2 append-
age, but forms an extra b strand augmenting (Harrison,
1996) strand b1 of b-arrestin 1. Thus, the b2 appendage
binding motif must undergo a major conformational
switch dependent upon its binding partner.
GPCRs bind a cognate site on the b-arrestin N-terminal
subdomain (Figure 4C). This interaction reorganizes the
polar core, locally unfolding b-arrestin strand b1, which
then destabilizes the DIVFEDFARQR motif/folded b-ar-
restin association (Gurevich and Gurevich, 2004). This
restructuring effect is exacerbated by the fact that F+5
Arg393 (italicized below and not involved in b2 append-
age binding) is an intrinsic part of the polar core and
plays a key role in the DIVFEDFARQR sequence binding
to the rest of b-arrestin (Figures 4D and 4E). The net re-
sult of GPCR-induced conformational changes is ejec-
tion of the b-arrestin FXX[FL]XXXR motif, freeing it for
interaction with the b2 appendage. In vitro, this recep-
tor-dependent conformational rearrangement can be
mimicked by disruption of the polar core by an R169E
mutation (Kovoor et al., 1999).
These observations suggest that release of the
FXX[FL]XXXR motif from the b-arrestin core, transition
to an a-helical conformation, and then binding to the
AP-2 b2 appendage is the primary trigger for incorpora-
tion of activated GPCR-b-arrestin complexes into
coated vesicles. In the ‘‘closed’’ basal state then, the
FXX[FL]XXXR motif should not be able to engage the
molecular interactions of the DIVFEDFARQR sequence
with its two discrete binding partners are generally com-
parable, both utilizing two phenylalanines and an argi-
nine in chemically similar pockets, along with several
backbone hydrogen bond interactions. Both interac-
cromolar range) and, therefore, able to compete with
each other. In this case, the sequence clearly could not
operate as the GPCR endocytic trigger, but, since the
interaction with the folded portion of b-arrestin is intra-
molecular, the concentration of this binding partner is
effectively increased into the low millimolar range (L.N.
Johnson and J. Ladbury, personal communication),
greatly favoring it over the intermolecular b2 appendage
interaction. To show that the FXX[FL]XXXR motif func-
tions as a GPCR-cargo-induced AP-2 b2 binding switch,
we measured the interaction of full-length or the ‘‘acti-
vated’’ R169E mutant b-arrestin 1 with the b2 append-
age. By ITC, no significant interaction is detectable
whereas the R169E mutant, in which the FXX[FL]XXXR
motif should be free, binds b2 with a KDof w1 mM (Fig-
ure S4). This value is comparable to that measured for
the isolated peptide with the b2 appendage and argues
strongly that intact b-arrestin binds b2 in an analogous
fashion to the peptide. Corroborating these results,
pull-down assays with immobilized GST-b2 appendage
with cell extracts derived from HeLa cells transfected
with FLAG-tagged wild-type or R169E b-arrestin 1
show that there is a dramatic difference in binding to
the b2 appendage. While the ‘‘activated’’ mutant binds
robustly, very limited association of the wild-type b-
arrestin 1 with b2 is apparent (Figure 4F), as reported
by Kim and Benovic (2002).
Finally, in support of the identification of the DIVFED-
FARQR sequence in b-arrestin being the endocytic trig-
ger, in Drosophila, sensory/visual arrestin (Arr2), despite
lacking a clathrin box, promotes clathrin-dependent
internalization of activated (phosphorylated) rhodopsin
Arr2381DDNIVFEDFAKMR sequence is 77% identical to
the mammalian b-arrestins, and it is therefore likely to
play an important role in the clathrin-mediated endocy-
tosis of rhodopsin in the fly compound eye. This block
of sequence is highly conserved in several insect sen-
sory arrestins without the presence of a clathrin box
(Merrill et al., 2003). Also significant is that, in mammals,
both rod and cone arrestins have the sequence FEE-
FARXN. This preserves the two aromatic side chains
and the basic residue important for the intramolecular
association that maintains the basal state, but the ab-
sence of the F+7 position arginine will preclude associa-
tion with AP-2, and, indeed, visual arrestins do not de-
liver GPCRs to clathrin-coated structures effectively
(Oakley et al., 2000).
A [DE]nX1–2FXX[FL]XXXR Motif in Epsin 1
Protein database searching with the program SIRW
(^P represents ‘‘not’’ proline) and a filter selected for do-
mains not represented in PFAM reveals that mammalian
epsins 1 and 2 also have putative FXX[FL]XXXR motifs
(Figure 2). These are positioned within polypeptide re-
gions predicted to be largely unstructured (Linding
et al., 2003) in which appendage and clathrin binding
motifs are characteristically found (Puntervoll et al.,
2003). Upstream of the proximal F0 residues, both epsin
sequences possess at least one acidic amino acid 2
residues N-terminal to the F0 phenylalanine, like the
ARH and b-arrestin sequences. This suggests that the
b2 appendage platform binding motif is best described
as [DE]nX1–2FXX[FL]XXXR. A construct comprising a 66
residue region of epsin 1 that encompasses the
[DE]nX1–2FXX[FL]XXXR motif fused to GST binds AP-2
in a manner that is mainly dependent upon the proximal
F0 phenylalanine, since a F403A substitution decreases
AP-2 binding significantly (Figure 5A). Yeast two-hybrid
(Figure 4B) and binary interaction assays with b2 ap-
pendage show that this region of epsin 1 binds to the
b2 appendage directly, in a manner now completely de-
pendent upon Phe403 (Figure 5B). From this, we infer
that the 66 residue epsin F403A mutant must also bind
cytosolic AP-2 weakly via the a appendage by using
the two DPW motifs present in this model protein. ITC
measurements confirm that the interaction of an epsin
1 peptide with the b2 appendage is direct but is about
10-fold weaker than for ARH and b-arrestin sequences
(Figure 3). The reduced affinity probably reflects both
the larger, polar aspartic acid-for-alanine substitution
Figure 5. A [DE]nX1–2FXX[FL]XXXR Motif in
Epsins 1 and 2
(A) Aliquots of 100 mg of either GST (lanes
a and b), GST-epsin 1 (356–421) (lanes c and
bilized on GSH-Sepharose were incubated
with rat brain cytosol. Portions of the super-
natant (S, w1%) and washed pellet (P, 10%)
were resolved by SDS-PAGE and either
stained by Coomassie blue or transferred to
nitrocellulose. The blot was probed with the
anti-clathrin heavy chain (HC) mAb TD.1 and
the anti-b subunit GD/2 antibody.
(B) Aliquots of 100 mg GST (lanes a and b),
GST-epsin 1 (356–421) (lanes c and d), or
on GSH-Sepharose were incubated with
60 mg/ml thrombin-cleaved b2 appendage in
the presence of 25 mM PPACK and 100 mg/ml
BSA. Portions of the supernatant (S, 0.5%)
and washed pellet (P, 10%) were resolved by
SDS-PAGE and either stained by Coomassie
blue or transferred to nitrocellulose. The blot
was probed with anti-b2 antibody.
(C) Aliquots of 50mg of eitherGST(lanes a and
b) or GST-b2 (lanes c–h) immobilized on GSH-
Sepharosewereincubatedwith ratbrain cyto-
sol in the absence (lanes a–d) or presence
of 4 mM epsin 1 (373DTEPDEFSDFDRLRTA)
(lanes e and f) or 0.8 mM b-arrestin 1
(383DDDIVFEDFARQRLKG) peptide (lanes g
and h). Portions of the supernatant (S, w1%)
and washed pellet (P, 10%) were resolved by
SDS-PAGE and either stained by Coomassie
blue or transferred to nitrocellulose. Sections
of the blots were probed with anti-epsin or
anti-eps15 antibodies or anti-AP180 mAb.
The asterisk indicates a nonspecific band
detected in all of the supernatant fractions by
the anti-eps15 serum used.
AP-2 b2 Appendage Function in Endocytosis
Figure 6. A Second Distinct Binding Site on the Sandwich Subdomain of the b2 Appendage
(A) Residues ‘‘AAF’’ were modeled into this side site. The b2 appendage is colored gray, and the peptide (purple) is shown in ball-and-stick
in ball-and-stick representation and are colored gold and purple, respectively. Also shown is the electron density (colored black) corresponding
to a single xenon atom, contoured at w3 s.
(C) Refined 2Fo2 Fcelectron density map of the peptide ‘‘AAF’’ modeled into the sandwich subdomain (contoured at w1.3 s). The peptide
(purple) is shown in ball-and-stick representation.
(D) Aliquots of 50 mg GST (lanes a and b); wild-type GST-b2 appendage (lanes c and d); the GST-b2 appendage Q756S (lanes e and f), Q804A
(lanes g and h), or Y815A (lanes i and j) sandwich mutants; the GST-b2 appendage W841A (lanes k and l) or Y888V (lanes m and n) platform
mutants immobilized on GSH-Sepharose were incubated with rat brain cytosol. Portions of the supernatant (S, w1%) and washed pellet (P,
10%) were resolved by SDS-PAGE and transferred to nitrocellulose. Sections of the blots were probed with anti-AP180 mAb, anti-CALM
mAb, anti-amphiphysin mAb, anti-eps15, or anti-epsin1 antibodies. The asterisk indicates a nonspecific band detected in all of the supernatant
fractions by the anti-eps15 serum used.
(E) Aliquots of 5, 10, 20, 50, or 100 mg the GST-aCappendage (lanes a–e), the GST-b2 appendage (lanes f–j), the GST-b2 appendage Y815A (lanes
k–o) or 100 mg GST (lane p) immobilized on GSH-Sepharose were incubated with rat brain cytosol. Portions (10%) of each washed pellet were
resolved by SDS-PAGE and were either stained with Coomassie blue or transferred to nitrocellulose. Sections of the blots were probed with
anti-AP180 mAb, anti-CALM mAb, anti-amphiphysin mAb, anti-clathrin-HC mAb TD.1, anti-eps15, anti-SNX9, or anti-epsin1 antibodies.
(F) Aliquots of 50 mg of either GST (lanes a and b), wild-type (lanes c and d) or Y815A mutant (lanes e and f) GST-b2 hinge+appendage (H+A),
wild-type (lanes g and h) or Y815A mutant (lanes i and j) GST-b2 H+A (LLN/AAA), wild-type (lanes k and l) or Y815A mutant (lanes m and n)
atthe F+4position inepsin (which is not optimal for plat-
epsin sequences compared with the ARH and b-arrestin
sequences as indicated by secondary structure predic-
tions. The properties of the peptide (large baseline sig-
nals) make it impossible to obtain a completely accurate
KD, but it is between 30 and 80 mM. Irrespective, a b2
appendage E849A platform mutation that strongly
perturbs ARH and b-arrestin binding to the b2 append-
age also abolishes the interaction of epsin with the b2
appendage (Figure 3).
Peptide competition experiments confirm that the
epsin 1 sequence binds the b2 platform with weaker
affinity, but in the same manner as ARH and b-arrestin
sequences. A 5-fold higher concentration of the epsin
peptide (4 mM) is less effective at preventing cytosolic
epsin 1 from binding to the GST-b2 appendage than
a b-arrestin 1 peptide (0.8 mM) (Figure 5C). Remarkably,
sociation of either AP180 or eps15 with the immobilized
b2 appendage (Figure 5C). This suggests the possibility
that these proteins must utilize a distinct interaction
surface to bind the b2 appendage.
A Functionally Distinct b Sandwich Binding Site
In the initial maps of the b2 appendage-ARH peptide
complex after molecular replacement, we observed
form subdomain bound full ARH peptide. The site is po-
sitioned on the sandwich subdomain, and the excellent
quality of the electron density makes it obvious that it
packed molecules in the crystal, it is likely essential for
crystallization, and we believe the polypeptide to be ei-
ther intact or degraded ARH peptide, as the b2 protein
was highly pure following gel filtration. The side chain
the peptide is thus most likely the EAF sequence of the
ARH peptide. This peptide is not necessarily an in vivo-
relevant binding partner, since the long time required
for crystal growth suggests low-affinity binding. How-
ever, opportunistic sequestering of this peptide indi-
cates that this sandwich site is available for binding sim-
ilar sequence peptides in vivo. The side chain of the
b7, and b8. The aliphatic portions of the side chains of
Gln756, Asn758, and Gln804 stack against the peptide
phenyl ring, as does the aromatic group of Tyr815
(Figure 6B). The binding of the peptide is further stabi-
lized by two hydrogen bonds involving the peptide
main chain of this phenylalanine and the side chains
of Gln756 (OE) and Tyr815 (OH), respectively. The posi-
tioning of this surface is analogous to the site upon the
platform-lacking AP-1 g subunit and GGA appendage
sents an aromatic residue) (Collins et al., 2003; Miller
sandwich that binds the WXX[FW]X[DE] motif (Mishra
et al., 2004; Praefcke et al., 2004; Ritter et al., 2004)
(Figure S5). Superposition of the GGA and b2 append-
ages reveals a striking overlay between the phenylala-
nine of the exogenous peptide and the Ø of the
JG[PED]Ø motif. This binding site is not present in the
a appendage sandwich since, crucially, Ile757 of the a
appendage (Asn758 in the b2 appendage) projects into
the analogous depression on the surface of a pocket
and because a small residue (Ala806 in the b2 append-
further blocking this pocket. Also, Ser815 (Tyr815 in b2
be unable to stabilize an interaction with the peptide
served on b appendages through metazoan evolution.
action surface, six are invariant from Caenorhabditis to
mammals and in Arabidopsis (Figure S6). Of the remain-
ing 2 residues, Ala758 is changed to methionine only in
Arabidopsis, and Val813 is conservatively substituted
with isoleucine in insects and alanine in worms.
To explore the biological significance of the b2 sand-
to alterthe molecular surface ofthis region. In pull-down
assays, wild-type GST-b2 appendage affinity isolates
AP180, CALM, eps15, amphiphysins I and II, and epsin
1 from brain cytosol (Figure 6D). Q756S, Q804A, or
Y815A b2 mutations interfere with both AP180 and am-
phiphysin binding. Although binding of eps15 is mini-
mally changed in the Q756S or Q804A mutants, the
Y815A substitution almost completely abolishes the
eps15 interaction. Crucially, epsin 1 binding is unaf-
fected by all of these sandwich subdomain mutations,
but platformchanges (W841Aand Y888V)severely com-
pected. These platform mutants still bind eps15 nor-
mally. The region of eps15 that engages the sandwich
subdomain is within the DPF triplet-rich C terminus be-
cause, fused to GST, residues 594–896 of eps15 effi-
ciently bind the purified b2 appendage (Figure S7).
Q756S, Y815A, or Y815W mutations each compromise
the ability of the b2 appendage to bind GST-eps15
(594–896). Yet, in the same assay, there is no difference
ages binding to either GST-b-arrestin 1 or GST-ARH
(Figure S7). The severely disruptive yet selective effect
of the GST-b2 Y815A sandwich site mutant on partner
engagement is best seen in titrations (Figure 6E). A hier-
archical pattern of partner association occurs as the
concentration of immobilized GST-a or GST-b2 append-
age increases. For both the a and b2 appendage, eps15
and epsin 1are strong binding partners, but onlyassoci-
ation of eps15 is >15-fold reduced in the b2 Y815A mu-
tant; epsin binding, as well as SNX9 (in which we identify
a probable b2 appendage platform binding motif
346ESEVFQQFLNFR), is insensitive to this sandwich
site mutation. These results strongly suggest that two
age preferentially engage distinct binding partners.
GST-b2 H+A DLLNLD, or the GST-b2 appendage (A) (lanes o and p) immobilized on GSH-Sepharose were incubated with rat brain cytosol. Por-
were probed with anti-clathrin-HC mAb TD.1, anti-clathrin light chain (LC) mAb Cl57.3, anti-AP180 mAb, or anti-eps15 or anti-epsin 1 antibodies.
The asterisk indicates a nonspecific band detected in all of the supernatant fractions by the anti-eps15 serum used.
AP-2 b2 Appendage Function in Endocytosis
The b2 appendage also binds to clathrin (Lundmark
and Carlsson, 2002; Owen et al., 2000), and the Y815A
substitution completely disrupts this interaction with
clathrin in the context of a GST-b2 appendage (Fig-
ure 6E). To probe the possible contribution of the sand-
wich site to the general clathrin binding properties of the
b2 subunit, we analyzed GST fusion proteins containing
both the appendage preceded by the unstructured
hinge, which houses the631LLNLD clathrin box, or the
C-terminal appendage alone. The hinge+appendage
protein binds soluble clathrin tightly, quantitatively re-
moving trimers from the cytosol (Figure 6F). In the con-
thrin binding substantially. The functional interplay
between the clathrin box and the b2 appendage sand-
wich site is plainly apparent when the LLNLD sequence
ing is markedly reduced compared to binding to the hin-
sandwich contact site, as no clathrin is bound by Y815A
mutants of GST-hinge+appendage constructs with cla-
thrin box mutation or deletion (Figure 6F). In all of these
binding studies, the Y815A substitution disrupts eps15
and AP180 binding to the appendage or hinge+append-
remains. Two conclusions can be drawn from this series
of experiments. First, optimal binding of clathrin to AP-2
occurs when the heavy chain engages both the clathrin
box and the b2 appendage sandwich site. Second, en-
gage the same surface on the appendage utilized by the
clathrin heavy chain, presumably by a yet-to-be defined
region of the distal leg.
Function of b2 Appendage Binding Sites In Vivo
In order to better understand the functional interactions
of the b2 appendage in vivo, we expressed in HeLa cells
a variety of point and deletion mutants of the b2 subunit
with YFP appended to the C terminus. The wild-type b2-
YFP construct targets to punctate structures (Figure 7A)
that colocalize with the AP-2 a subunit and clathrin (data
not shown), even at relatively high levels of expression.
This represents a mixture of b2-YFP incorporated into
AP-2 and free b2-YFP targeted to endocytic sites by vir-
tue of interactions with clathrin coat components. Dele-
fuse cytosolic distribution, although some puncta are
b2-YFP due to a small fraction of the tagged protein be-
ing incorporated into otherwise normal AP-2 adaptors.
This attests to the functional role of the hinge and
appendage in vivo. Similar results are obtained upon
transfection of a b2-YFP sandwich (Y815A) and platform
(Y888V) double mutant. Deletion of the LLNLD clathrin
box, by contrast, still allows the mutant b2-YFP to target
underlining the relative functional importance of the two
b2 appendage motif binding sites. Massive transient
overexpression of wild-type b2-YFP causes the forma-
tion of large intracellular aggregates that also sequester
ARH, eps15, and clathrin (Figure 7B). Based on fluores-
cence intensity, introduction of a single sandwich site
Y815A mutation onto the b2-YFP decreases the extent
to which eps15 agglomerates with the overexpressed,
tution, the double mutant, while still forming extensive
aggregates, now has a much diminished capacity to
trap either platform (ARH) or sandwich (eps15) binding
Our data showing that clathrin competes with eps15
and AP180 for the sandwich of b2 appendage provide
a molecular explanation for previous work documenting
that AP-2-driven assembly of clathrin cages displaces
eps15 from the AP-2 complex (Cupers et al., 1998;
Owen et al., 2000). If, by simultaneously engaging the
clathrin box in the b2 subunit hinge and the sandwich
subdomain, clathrin ejects eps15 from the b2 append-
age in vivo, then eps15 might be excluded from the as-
sembled region of the clathrin lattice and concentrated
at the periphery of lattices. In freeze-etch images of
NRK cell plasma membranes, eps15 is clearly restricted
to the lattice rim (Figure 7C), as first suggested in thin
sections (Tebar et al., 1996). This contrasts with the lo-
calization of both epsin, which binds both the a and b2
appendage platforms and to PtdIns(4,5)P2, and the cla-
thrin light chain; both of these proteins are distributed
throughout the lattice, at the periphery and upon invag-
an important feature of appendage-motif interactions:
the occurrence of discrete motifs for different append-
age binding partners within CLASPs or accessory fac-
tors dictates the positioning of these proteins within
the assembling lattice.
Finally, the anomalous difference maps of a xenon-
pressurized b2 appendage-ARH complex crystal reveal
a single xenon atom bound in a pocket adjacent to the
phenylalanine side chain-containing pocket (Figure 6B).
As xenon characteristically binds cavities lined with hy-
drophobic residues (Quillin et al., 2000), this adjacent
pocket might accommodate a bulky hydrophobic
(F,M,L,I,V) or Ø residue. The relative positioning of this
second potential Ø residue-accepting pocket suggests
that candidate ligands could be ØGXsmallØ, ØXGs-
sequence of the unfolded portion of human AP180 able
to bind the b2 appendage (Hao et al., 1999) has a candi-
date FGDAF sequence that, when mutated to AGDAA,
causes a severe reduction in binding to the b2 append-
age (Figure 4B). This FGDAF sequence is repeated ex-
actly again in the unstructured region, and it also occurs
within the C-terminal segment of eps15 that binds the
b2 sandwich. Multiple copies of motifs embedded in
largely unstructured regions are characteristic of other
appendage binding sequences, supporting the notion
that FGDAF is an authentic b2 appendage sandwich
AP-2 participates directly in assembly of the polyhedral
ses of transmembrane cargo (YXXØ and [DE]XXXL[LI])
through direct binding, while the appendages also serve
as interaction hubs that organize the flow of information
at the coat assembly zone (Mishra et al., 2004; Praefcke
et al., 2004). Here, we demonstrate that the b2 append-
age binds to a-helical [DE]nX1–2FXX[FL]XXXR motifs on
ity is conferred mainly by the anchor F, [FL], and R
motif is found in CLASPs, clathrin adaptors that bind
to specific families of transmembrane receptors: ARH
to LDL receptors (Traub, 2005), b-arrestins 1 and 2 to
GPCRs (Lefkowitz and Shenoy, 2005), and epsins 1
and 2 to polyubiquitinated receptors (Barriere et al.,
2006; Hawryluk et al., 2006). Thus, a major role of the
b2 appendage is to recruit CLASPs with associated
increasing the spectrum of endocytic clathrin-coated
vesicle cargo beyond proteins with motifs recognized
by the AP-2 adaptor core. The mode of b2 platform en-
pendage ligands. Instead of numerous, avidity-based
Figure 7. b2 Appendage Function In Vivo
(A) Representative single optical confocal sections of HeLa cells transiently transfected with either a wild-type (WT) b2 subunit-YFP construct,
a b2 subunit trunk (1-607)-YFP (Trnk), b2 DLLNLD-YFP, or b2 (Y815A/Y888V)-YFP sandwich and platform double mutant.
(B) Representative confocal sections of HeLa cells expressing a wild-type b2 subunit-YFP, a b2 (Y815A)-YFP sandwich, or a b2 (Y815A/Y888V)-
YFP double mutant. YFP fluorescence is shown in the left panels of each pair, and ARH or eps15 staining after incubation of fixed and permea-
bilized cells with appropriate primary antibodies is shown in the right panels. Insets show the clathrin staining pattern for the region boxed in the
(C) Adherent plasmalemmal sheets from NRK cells labeled with anti-eps15, anti-epsin 1, or anti-clathrin LC, followed by secondary antibodies
conjugated to15nmgold(pseudo-coloredred). Notethateps15 isfoundonlyatthe edgesofclathrinlattices(pseudo-coloredpaleyellow),while
epsin and clathrin are found throughout clathrin lattices, regardless of whether the lattices are rounded (arrows) or flat.
AP-2 b2 Appendage Function in Endocytosis
interaction motif repeats, b-arrestin, ARH, and epsin
each contain only a single [DE]nX1–2FXX[FL]XXXR motif,
structural and biochemical information was obtained by
using synthetic peptides; thus, future studies are re-
quired to determine whether any alterations in the
mode of engagement occur between the intact proteins.
and the b-arrestin 1 (R169E) protein binding to b2 argue
against this however.
Under basal conditions, it clamps b-arrestin in an endo-
that augments a preexisting b sheet such that it crosses
the b-arrestin surface and is anchored by the interac-
tions of the Phe388, Phe391, and Arg393 side chains.
Once freed by a conformational change in the folded
core driven by ligand-activated GPCR engagement, the
pendage (Kim and Benovic, 2002), and, along with the
bound GPCR, translocates to forming clathrin-coated
structures. To carry out this pivotal regulatory function,
the [DE]nX1–2FXX[FL]XXXR motif undergoes a large con-
formational transition and can therefore be regarded as
a flexible key that can fit two very different locks.
A second protein-protein interaction site also exists
on the sandwich subdomain of the b2 appendage that
interacts both with clathrin-associated accessory fac-
tors and clathrin itself, and it functions independently
of the platform CLASP binding site. The ability of poly-
meric clathrin to compete accessory factors off the
sandwich site provides a mechanism for the temporal/
spatial patterning of these proteins during clathrin coat
formation. After polymeric clathrin-induced displace-
ment, those factors still able to bind independently to
clathrin and PtdIns(4,5)P2, such as AP180 and epsin,
will remain at the center of clathrin structures and be in-
corporated into coated vesicles. Those such as eps15,
which bind neither effectively, are localized to the pe-
ripheral, circumferentially positioned ‘‘assembly zone’’
(Praefcke et al., 2004), where the density of polymeric
livery of ubiquitinated cargo bound via its own ubiquitin-
interacting motifs to epsin, which is concentrated within
the forming clathrin lattice through several interactions.
CLASPs utilizing the sandwich site and being subjected
to competition by assembled clathrin, is that the plat-
form interaction surface is reserved for relatively tightly
bound CLASPs. This prevents the expulsion of cargo-
loading CLASPs during coat fabrication. In the case of
b-arrestins, the availability of their privileged site on the
b2 appendage allows them and physically associated,
actively signaling GPCR superfamily members to be
rapidly recruited and subsequently internalized via pre-
existing clathrin-coated structures (Santini et al., 2002;
In summary, the b2 appendage is a versatile protein
interaction scaffold with two main functions. It helps
position proteins involved in clathrin coat assembly at
regions of clathrin formation through one site and,
of cargo-specific CLASPs, and their cognate cargo, into
the coat via specific and tight binding of a [DE]nX1–2
Protein expression and purification, ITC, binding assays, yeast two-
hybrid screens, cell culture, transfections, and imaging were per-
formed essentially as described in our previous publications, and
details, along with information on the DNA/plasmid preparations
used, are presented in the Supplemental Data.
Crystallization and Structure Determination
b2 appendage (2.1 mM) and ARH peptide (DDGLDEAFSRLAQSRT)
(8.1 mM) were incubated on ice for w30 min. The best crystals
were grown by sitting drop vapor diffusion against a reservoir con-
taining 18% PEG 8000, 100 mM HEPES (pH 7.5), and 4 mM DTT.
fore being mounted into cryoprotected buffer (20% PEG 8000, 100
mM HEPES [pH 7.5], 4 mM DTT, 18% glycerol, and 0.2 mM ARH)
and flash frozen at 100K. X-ray diffraction data were collected at
100K at ESRF beamline ID-23, were indexed and integrated in
MOSFLM, and were scaled with SCALA (CCP4, 1994) with cell pa-
rameters 37.750, 36.313, 98.982, a = b = 90.00, g = 92.91 in space
group P21.A molecular replacement solution was determined by us-
ing unbound b2 appendage (PDB code 1E42) as the search model in
AMORE (Navaza, 1994). A xenon derivative was prepared by placing
a cryoprotected crystal at 10 atmospheres pressure of xenon before
flash freezing. A data set was collected on this crystal at wavelength
7.1 keV, also on ID-23, and the location of the xenon sites was deter-
mined by difference Pattersons (CCP4, 1994). Model building was
705–937 of b2 appendage, residues 255–267 of ARH, 3 residues of
a second peptide modeled as the sequence AAF in the sandwich
side site, and 144 waters (coordinates deposited as PDB code
2G30). Accessible surface calculations were performed by using
SURFACES (CCP4, 1994). Figures were made with Aesop (M. Noble,
personal communication) and CCP4 mg (Potterton et al., 2002).
Supplemental Data including detailed Supplemental Experimental
able at http://www.developmentalcell.com/cgi/content/full/10/3/
inson for critical comments on the manuscript. We are also grateful
to all of our colleagues who provided important reagents. This
work was funded in part (D.J.O, B.M.C. and M.A.E) by a Wellcome
Trust Senior Research Fellowship in Basic Biomedical Science to
D.J.O and in part by National Institutes of Health grant R01
DK53249 and an American Heart Association (AHA) Established
Investigator Award (0540007N) to L.M.T. P.A.K. was supported by
AHA Predoctoral Fellowship Award 0415428U.
Received: November 18, 2005
Revised: December 30, 2005
Accepted: January 12, 2006
Published online: March 6, 2006
Alloway, P.G., Howard, L., and Dolph, P.J. (2000). The formation of
stable rhodopsin-arrestin complexes induces apoptosis and photo-
receptor cell degeneration. Neuron 28, 129–138.
Barriere,H., Nemes, C., Lechardeur,D.,Khan-Mohammad, M., Fruh,
K., and Lukacs, G.L. (2006). Molecular basis of Ub-dependent inter-
nalization of membrane proteins in mammalian cells. Traffic 7,
Bonifacino, J.S., and Traub, L.M. (2003). Signals for sorting of trans-
membrane proteins to endosomes and lysosomes. Annu. Rev. Bio-
chem. 72, 395–447.
Brett, T.J., Traub, L.M., and Fremont, D.H. (2002). Accessory protein
recruitment motifs in clathrin-mediated endocytosis. Structure
(Camb) 10, 797–809.
CCP4 (Collaborative Computational Project, Number 4) (1994). The
CCP4 suite: programs for protein crystallography. Acta Crystallogr.
D Biol. Crystallogr. 50, 760–763.
Collins, B.M., McCoy, A.J., Kent, H.M., Evans, P.R., and Owen, D.J.
(2002). Molecular architecture and functional model of the endocytic
AP2 complex. Cell 109, 523–535.
Collins, B.M., Praefcke, G.J., Robinson, M.S., and Owen, D.J. (2003).
Structural basis for binding of accessory proteins by the appendage
domain of GGAs. Nat. Struct. Biol. 10, 607–613.
Cupers, P., Jadhav, A.P., and Kirchhausen, T. (1998). Assembly of
clathrin coats disrupts the association between Eps15 and AP-2
adaptors. J. Biol. Chem. 273, 1847–1850.
(1999). Arrestin function in G protein-coupled receptor endocytosis
requires phosphoinositide binding. EMBO J. 18, 871–881.
Gonzalez-Gaitan, M., and Jackle, H. (1997). Role of Drosophila
a-adaptin in presynaptic vesicle recycling. Cell 88, 767–776.
Gurevich, V.V., and Gurevich, E.V. (2004). The molecular acrobatics
of arrestin activation. Trends Pharmacol. Sci. 25, 105–111.
Han, M., Gurevich, V.V., Vishnivetskiy, S.A., Sigler, P.B., and Schu-
bert, C. (2001). Crystal structure of b-arrestin at 1.9 A˚: possible
mechanism of receptor binding and membrane Translocation.
Structure (Camb) 9, 869–880.
Hao,W.,Luo, Z.,Zheng,L.,Prasad, K.,andLafer, E.M.(1999). AP180
and AP-2 interact directly in a complex that cooperatively assem-
bles clathrin. J. Biol. Chem. 274, 22785–22794.
Harrison, S.C. (1996). Peptide-surface association: the case of PDZ
and PTB domains. Cell 86, 341–343.
and Traub, L.M. (2006). Epsin 1 is a polyubiquitin-selective clathrin-
associated sorting protein. Traffic, in press.
He, G., Gupta, S., Yi, M., Michaely, P., Hobbs, H.H., and Cohen, J.C.
(2002). ARH is a modular adaptor protein that interacts with the LDL
receptor, clathrin and AP-2. J. Biol. Chem. 277, 44044–44049.
Honing, S., Ricotta, D., Krauss, M., Spate, K., Spolaore, B., Motley,
A., Robinson, M., Robinson, C., Haucke, V., and Owen, D.J. (2005).
Phosphatidylinositol-(4,5)-bisphosphate regulates sorting signal
recognition by the clathrin-associated adaptor complex AP2. Mol.
Cell 18, 519–531.
Janvier, K., Kato, Y., Boehm, M., Rose, J.R., Martina, J.A., Kim, B.Y.,
Venkatesan, S., and Bonifacino, J.S. (2003). Recognition of dileu-
cine-based sorting signals from HIV-1 Nef and LIMP-II by the AP-1
g-s1 and AP-3 d-s3 hemicomplexes. J. Cell Biol. 163, 1281–1290.
Jha, A., Agostinelli, N.R., Mishra, S.K., Keyel, P.A., Hawryluk, M.J.,
and Traub, L.M. (2004). A novel AP-2 adaptor interaction motif ini-
tially identified in the long-splice isoform of synaptojanin 1, SJ170.
J. Biol. Chem. 279, 2281–2290.
Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard. (1991). Im-
proved methods for binding protein models in electron density
maps and the location of errors in these models. Acta Crystallogr.
A 47, 110–119.
Kamikura, D.M., and Cooper, J.A. (2003). Lipoprotein receptors and
a disabled family cytoplasmic adaptor protein regulate EGL-17/FGF
export in C. elegans. Genes Dev. 17, 2798–2811.
Kim, Y.M., and Benovic, J.L. (2002). Differential roles of arrestin-2
interaction with clathrin and adaptor protein 2 in G protein-coupled
receptor trafficking. J. Biol. Chem. 277, 30760–30768.
Kiselev, A., Socolich, M., Vinos, J., Hardy, R.W., Zuker, C.S., and
Ranganathan, R. (2000). A molecular pathway for light-dependent
photoreceptor apoptosis in Drosophila. Neuron 28, 139–152.
Kovoor, A., Celver, J., Abdryashitov, R.I., Chavkin, C., and Gurevich,
V.V. (1999). Targeted construction of phosphorylation-independent
b-arrestin mutants with constitutive activity in cells. J. Biol. Chem.
Laporte, S.A., Oakley, R.H., Holt, J.A., Barak, L.S., and Caron, M.G.
(2000).The interaction ofb-arrestinwiththe AP-2adaptor isrequired
for the clustering of b2-adrenergic receptor into clathrin-coated pits.
J. Biol. Chem. 275, 23120–23126.
Laporte, S.A., Miller, W.E., Kim, K.M., and Caron, M.G. (2002).
b-arrestin/AP-2 interaction in G protein-coupled receptor internali-
zation. Identification of a b-arrestin binding site in b2-adaptin. J.
Biol. Chem. 277, 9247–9254.
Lawrence, M.C., and Colman, P.M. (1993). Shape complementarity
at protein/protein interfaces. J. Mol. Biol. 234, 946–950.
Lefkowitz, R.J., and Shenoy, S.K. (2005). Transduction of receptor
signals by b-arrestins. Science 308, 512–517.
Linding, R., Russell, R.B., Neduva, V., and Gibson, T.J. (2003). Glob-
Plot: exploring protein sequences for globularity and disorder.
Nucleic Acids Res. 31, 3701–3708.
Lundmark, R., and Carlsson, S.R. (2002). The b-appendages of the
four adaptor-protein (AP) complexes: structure and binding proper-
ties, and identification of sorting nexin 9 as an accessory protein to
AP-2. Biochem. J. 362, 597–607.
McMahon, H.T., and Mills, I.G. (2004). COP and clathrin-coated ves-
icle budding: different pathways, common approaches. Curr. Opin.
Cell Biol. 16, 379–391.
Merrill, C.E., Pitts, R.J., and Zwiebel, L.J. (2003). Molecular charac-
terization of arrestin family members in the malaria vector mosquito,
Anopheles gambiae. Insect Mol. Biol. 12, 641–650.
Milano, S.K., Pace, H.C., Kim, Y.M., Brenner, C., and Benovic, J.L.
(2002). Scaffolding functions of arrestin-2 revealed by crystal struc-
ture and mutagenesis. Biochemistry 41, 3321–3328.
Miller, G.J., Mattera, R., Bonifacino, J.S., and Hurley, J.H. (2003).
Recognition of accessory protein motifs by the g-adaptin ear do-
main of GGA3. Nat. Struct. Biol. 10, 599–606.
Mishra, S.K., Watkins, S.C., and Traub, L.M. (2002). The autosomal
recessive hypercholesterolemia (ARH) protein interfaces directly
with the clathrin-coat machinery. Proc. Natl. Acad. Sci. USA 99,
A., Heuser, J.E., Fremont, D.H., and Traub, L.M. (2004). Dual-
engagement regulation of protein interactions with the AP-2 adaptor
a appendage. J. Biol. Chem. 279, 46191–46203.
Mishra, S.K., Keyel, P.A., Edeling, M.A., Owen, D.J., and Traub, L.M.
(2005). Functional dissection of an AP-2 b2 appendage-binding
sequence within the autosomal recessive hypercholesterolemia
(ARH) protein. J. Biol. Chem. 280, 19270–19280.
Mitsunari, T., Nakatsu, F., Shioda, N., Love, P.E., Grinberg, A., Boni-
facino, J.S., and Ohno, H. (2005). Clathrin adaptor AP-2 is essential
for early embryonal development. Mol. Cell. Biol. 25, 9318–9323.
Navaza, J. (1994). AMoRe: an automated package for molecular re-
placement. Acta Crystallogr. A 50, 157–163.
Oakley, R.H., Laporte, S.A., Holt, J.A., Barak, L.S., and Caron, M.G.
(1999). Association of b-arrestin with G protein-coupled receptors
during clathrin-mediated endocytosis dictates the profile of recep-
tor resensitization. J. Biol. Chem. 274, 32248–32257.
Oakley, R.H., Laporte, S.A., Holt, J.A., Caron, M.G., and Barak, L.S.
(2000). Differential affinities of visual arrestin, b arrestin1, and
b arrestin2 for G protein-coupled receptors delineate two major
classes of receptors. J. Biol. Chem. 275, 17201–17210.
Owen, D.J., and Evans, P.R. (1998). A structural explanation for the
recognition of tyrosine-based endocytotic signals. Science 282,
Owen, D.J., Vallis, Y., Noble, M.E., Hunter, J.B., Dafforn, T.R., Evans,
P.R., and McMahon, H.T. (1999). A structural explanation for the
binding of multiple ligands by the a-adaptin appendage domain.
Cell 97, 805–815.
Owen, D.J., Vallis, Y., Pearse, B.M., McMahon, H.T., and Evans, P.R.
(2000). The structure and function of the b2-adaptin appendage
domain. EMBO J. 19, 4216–4227.
AP-2 b2 Appendage Function in Endocytosis
Owen, D.J., Setiadi, H., Evans, P.R., McEver, R.P., and Green, S.A.
(2001). A third specificity-determining site in mu 2 adaptin for se-
quences upstream of YxxF sorting motifs. Traffic 2, 105–110.
coats: structure and function. Annu. Rev. Cell Dev. Biol. 20, 153–191.
Potterton, E., McNicholas, S., Krissinel, E., Cowtan, K., and Noble,
M. (2002). The CCP4 molecular-graphics project. Acta Crystallogr.
D Biol. Crystallogr. 58, 1955–1957.
Praefcke, G.J., Ford, M.G., Schmid, E.M., Olesen, L.E., Gallop, J.L.,
Peak-Chew, S.Y., Vallis, Y., Babu, M.M., Mills, I.G., and McMahon,
H.T. (2004). Evolving nature of the AP2 a-appendage hub during cla-
thrin-coated vesicle endocytosis. EMBO J. 23, 4371–4383.
tingsdal, M., Cameron, S., Martin, D.M., Ausiello, G., Brannetti, B.,
Costantini, A., et al. (2003). ELM server: a new resource for investi-
gating short functional sites in modular eukaryotic proteins. Nucleic
Acids Res. 31, 3625–3630.
Size versus polarizability in protein-ligand interactions: binding of
noble gases within engineered cavities in phage T4 lysozyme.
J. Mol. Biol. 302, 955–977.
Ramu, C. (2003). SIRW: a web server for the simple indexing and re-
searches. Nucleic Acids Res. 31, 3771–3774.
Ritter, B., Philie, J., Girard, M., Tung, E.C., Blondeau, F., and
McPherson, P.S. (2003). Identification of a family of endocytic pro-
teins that define a new a-adaptin ear-binding motif. EMBO Rep. 4,
Ritter, B., Denisov, A.Y., Philie, J., Deprez, C., Tung, E.C., Gehring,
define the specificity of accessory protein binding to AP-1 and AP-2.
EMBO J. 23, 3701–3710.
Robinson, M.S. (2004). Adaptable adaptors for coated vesicles.
Trends Cell Biol. 14, 167–174.
Santini, F., Gaidarov, I., and Keen, J.H. (2002). G protein-coupled re-
ceptor/arrestin3 modulation of the endocytic machinery. J. Cell Biol.
Scott, M.G., Benmerah, A., Muntaner, O., and Marullo, S. (2002). Re-
cruitment of activated G protein-coupled receptors to pre-existing
clathrin-coated pits in living cells. J. Biol. Chem. 277, 3552–3559.
Shim, J., and Lee, J. (2000).Molecular genetic analysis of apm-2 and
aps-2,genesencoding themediumandsmallchainsofthe AP-2cla-
thrin-associated protein complex in the nematode Caenorhabditis
elegans. Mol. Cells 10, 309–316.
Slepnev,V.I., and De Camilli, P. (2000).Accessory factors in clathrin-
dependent synaptic vesicleendocytosis. Nat.Rev.Neurosci.1,161–
Sorkin, A. (2004). Cargo recognition during clathrin-mediated endo-
cytosis: a team effort. Curr. Opin. Cell Biol. 16, 392–399.
Tebar, F., Sorkina, T., Sorkin, A., Ericsson, M., and Kirchhausen, T.
(1996). eps15 is a component of clathrin-coated pits and vesicles
and is located at the rim of clathrin-coated pits. J. Biol. Chem.
Traub, L.M. (2005). Common principles in clathrin-mediated sorting
at the Golgi and the plasma membrane. Biochim. Biophys. Acta
Traub, L.M., Downs, M.A., Westrich, J.L., and Fremont, D.H. (1999).
Crystal structure of the a appendage of AP-2 reveals a recruitment
platform for clathrin-coat assembly. Proc. Natl. Acad. Sci. USA 96,
Walther, K., Diril, M.K., Jung, N., and Haucke, V. (2004). Functional
dissection of the interactions of stonin 2 with the adaptor complex
AP-2 and synaptotagmin. Proc. Natl. Acad. Sci. USA 101, 964–969.
The coordinates for the b2 appendage-ARH peptide complex have