A Novel Clathrin Adaptor Complex Mediates Basolateral Targeting in Polarized Epithelial Cells

Abstract

Although polarized epithelial cells are well known to maintain distinct apical and basolateral plasma membrane domains, the mechanisms responsible for targeting membrane proteins to the apical or basolateral surfaces have remained elusive. We have identified a novel form of the AP-1 clathrin adaptor complex that contains as one of its subunits mu1B, an epithelial cell-specific homolog of the ubiquitously expressed mu1A. LLC-PK1 kidney epithelial cells do not express mu1B and missort many basolateral proteins to the apical surface. Stable expression of mu1B selectively restored basolateral targeting, improved the overall organization of LLC-PK1 monolayers, and had no effect on apical targeting. We conclude that basolateral sorting is mediated by an epithelial cell-specific version of the AP-1 complex containing mu1B.

Full text

Cell, Vol. 99, 189–198, October 15, 1999, Copyright 1999 by Cell Press
A Novel Clathrin Adaptor Complex
Mediates Basolateral Targeting
in Polarized Epithelial Cells
a superficial relationship to signals that specify receptor
endocytosis via clathrin-coated pits (Matter and Mell-
man, 1994). Apical transport, on the other hand, occurs
only in the absence of a functional basolateral signal
and involves N- or O-linked carbohydrate moieties in a
Heike Fo
¨lsch,* Hiroshi Ohno,
‡
Juan S. Bonifacino,
†
and Ira Mellman*
§
*Department of Cell Biology and
Ludwig Institute for Cancer Research
Yale University School of Medicine
New Haven, Connecticut 06520-8002 protein’s ectodomain or as yet unspecified information
in the membrane-anchoring domain. These features are
†
Cell Biology and Metabolism Branch
National Institute of Child Health and Human thought to permit segregation in glycolipid rafts that are
incorporated into apical transport vesicles (Keller andDevelopment
National Institutes of Health Simons, 1997; Simons and Ikonen, 1997). The proteins
that decode either apical or basolateral signals, how-Bethesda, Maryland 20892-5430
‡
Division of Molecular Membrane Biology ever, have remained unidentified.
Since basolateral targeting signals are found in aCancer Research Institute
Kanazawa University membrane protein’s cytoplasmic domain, it seems likely
that they will be recognized by a cytosolic protein or13-1 Takaramachi
Kanazawa 920-0934 protein complex. Indeed, for that subset of proteins
which are localized at the basolateral surface due toJapan the presence of a cytoplasmic tail PDZ-binding domain,
interaction with cytosolic PDZ proteins has been found
essential to ensure polarity. However, PDZ interactionsSummary seem to be important for selective retention after arrival
at the basolateral surface as opposed to sorting intoAlthough polarized epithelial cells are well known to
maintain distinct apical and basolateral plasma mem- transport vesicles at the level of the TGN (Cohen et al.,
1998).brane domains, the mechanisms responsible for tar-
geting membrane proteins to the apical or basolateral In the case of the more common tyrosine-dependent
basolateral sorting signals, a role for clathrin adaptorsurfaces have remained elusive. We have identified a
novel form of the AP-1 clathrin adaptor complex that complexes has seemed a possibility, although no direct
evidence has yet emerged. It is well established thatcontains as one of its subunits m1B, an epithelial cell-
specific homolog of the ubiquitously expressed m1A. tyrosine-based and dileucine-based sorting signals in-
volved in endocytosis and lysosomal targeting are rec-LLC-PK1 kidney epithelial cells do not express m1B
and missort many basolateral proteins to the apical ognized by one or more heterotetrameric adaptor com-
plexes AP-1 (g,b1, m1, s1), AP-2 (a,b2, m2, s2), andsurface. Stable expression of m1B selectively restored
basolateral targeting, improved the overall organiza- AP-3 (d,b3, m3, s3) (Hirst and Robinson, 1998; Bonifa-
cino and Dell’Angelica, 1999). A fourth related complex,tion of LLC-PK1 monolayers, and had no effect on
apical targeting. We conclude that basolateral sorting AP-4 (e,b4, m4, s4), has recently been described (Dell’
Angelica et al., 1999a; Hirst et al., 1999). AP-2 mediatesis mediated by an epithelial cell-specific version of the
AP-1 complex containing m1B. endocytosis from the plasma membrane, while AP-1,
AP-3, and AP-4 are thought to mediate sorting at the
TGN and/or endosomal membranes. Interaction be-
Introduction tween the adaptor complexes and tyrosine-based sort-
ing signals is mediated by the msubunits (Ohno et al.,
The plasma membrane of polarized epithelial cells is 1995, 1996), while dileucine signals may interact with
differentiated into apical and basolateral domains, each the bsubunits (Rapoport et al., 1998).
with distinctive sets of proteins and lipids (Rodriguez- We have recently identified a closely related (79%
Boulan and Powell, 1992; Drubin and Nelson, 1996; Yea- identity at the amino acid level) isoform of the m1 subunit
man et al., 1999). Sorting of newly synthesized plasma of AP-1 whose expression is limited to cells of epithelial
membrane proteins typically occurs in the trans-Golgi origin (Ohno et al., 1999). mRNA for the new protein,
network (TGN), where apical and basolateral compo- designated m1B to distinguish it from the ubiquitously
nents are selectively packaged into specific transport expressed m1A, was found to be expressed in various
vesicles. The basic logic of polarized sorting is well polarized epithelial cell lines, including MDCK, Caco-2,
knownand involvesahierarchical interplaybetweentwo HT-29, Hec-1-A, and RL-95-2 cells. An exception was
types of sorting determinants. Basolateral transport is the kidney epithelial cell line LLC-PK1, thought to be
often dependent on the presence of a discrete cyto- derived from the renal proximal tubule. LLC-PK1 cells
plasmic domain targeting signal (Casanova et al., 1991; polarize but do not express detectable amounts of m1B
Hunziker et al., 1991; Le Bivic et al., 1991; Matter et al., mRNA. Interestingly, LLC-PK1 cells have recently been
1992). Basolateral signals often, but not always, involve reported to exhibit a potential “defect” in cell surface
criticaltyrosine ordileucine residues,suggesting atleast polarity (Roush et al., 1998), but the molecular nature
of the defect was not defined. We have taken advantage
§
To whom correspondence should be addressed (e-mail: ira.mellman
@yale.edu).
of the functional m1B knockout provided by LLC-PK1

Cell
190
Figure 1. Sorting of LDL Receptor in MDCK
Cells and LLC-PK1 Cells
MDCK cells (A) or LLC-PK1 cells (B) were
grown on Transwell filters and infected with
recombinant adenoviruses for 1 hr and cul-
tured for 2 days to allow for the expression
of LDL receptor. Viable cells were incubated
with antibodies against the LDL receptor,
fixed, and incubated with FITC-labeled sec-
ondary antibodies as described in Experimen-
tal Procedures. Specimens were analyzed by
confocal microscopy. Representative X-Y and
X-Z sections are shown.
cells to demonstrate that an epithelial cell-specific m1B- m1A (Figure 2A). To confirm the specificity of these anti-
bodies, they were tested by Western blot analysiscontaining AP-1 complex plays a critical role in basolat-
eral targeting. against recombinant m1A and m1B protein produced in
E. coli. As shown in Figure 2B, an antibody previously
generated against m1A reacted well with both m1A andResults
m1B. The new anti-peptide antibodies, however, were
specific for m1B. The C-terminal peptide antibody wasLDL Receptor Is Basolateral in MDCK Cells
and Apical in LLC-PK1 Cells subsequently used for detection of m1B expression in
LLC-PK1 transfectants.To determine whether m1B plays a role in polarized sort-
ing, we first compared the localization of the LDL recep- Nontransfected LLC-PK1 cells and cloned LLC-PK1
tor in m1B-positive MDCK cells and m1B-negative LLC-
PK1 cells. The LDL receptor is well known to localize
to the basolateral plasma membrane domain of MDCK
cellsby virtueof signals containedwithin its cytoplasmic
tail (Matter et al., 1992, 1994). Expression was achieved
byinfecting polarizedcells grownon filterswith a recom-
binant adenovirus encoding the receptor (adLDLR).
Two days after infection, the cells were immunola-
beled, fixed, and analyzed by confocal microscopy. In
the MDCK cells, LDL receptor was detected only at the
basolateralsurfaces, independently ofexpression levels
(Figure 1A). This was already apparent in X-Y confocal
sections (upper panel), where the marginal outlines of
expressing cells were stained. Basolateral localization
was confirmed in vertical (X-Z) optical sections, where
expressing cells displayed LDL receptor at the lateral
and basal, but not the apical, plasma membrane do-
mains (Figure 1A, lower panel). These control experi-
ments with MDCK cells demonstrated that virus infec-
tion did not interfere with polarized expression of the
LDL receptor. An entirely different picture was obtained
for infected LLC-PK1 cells, however (Figure 1B). In these
cells, LDL receptor was found only at the apical surface
regardless of expression level. This is the same pheno-
Figure 2. Analysis of LLC-PK1 Cells Transfected with m1A or m1B
type obtained in MDCK cells when both basolateral tar-
cDNAs
geting signals are deleted from the LDL receptor cyto-
(A) Peptides directed against the N-terminal part or C-terminal part
plasmictail (Matteret al.,1992). Theseobservations thus
of the human m1B protein (423 aa) and the corresponding regions
in mouse m1A (423 aa), which were used for immunization of rabbits
suggested a possible involvement for m1B in basolateral
for the production of m1B-specific antibodies.
targeting mediated by cytoplasmic tail signals.
(B) Recombinant m1A and m1B were produced in E. coli, partially
purified, and subjected to SDS–PAGE followed by a transfer onto
Generation of m1B-Expressing LLC-PK1 Cells
nitrocellulose. Immunodecoration with an antibody directed against
We then asked whether expression of m1B in LLC-PK1
m1A but cross-reacting with m1B (upper panel) and the two m1B-
cells would permit the basolateral targeting of LDL re-
specific antibodies (lower panels) were performed to characterize
the specificity of the antibodies.
ceptor. For this purpose, LLC-PK1 cells were first stably
(C) Nontransfected LLC-PK1 cells or LLC-PK1 cells transfected with
transfectedwith anexpression plasmid containinga full-
m1A or m1B cDNA were lysed in RIPA buffer, and 2.5, 5, or 10 mg
length m1B cDNA, or a m1A cDNA as a control. To follow
of total protein was used for SDS–PAGE and Western blotting (see
m1B expression at the protein level, two m1B-specific
Experimental Procedures for details). g-adaptin, m1A/B, and m1B
anti-peptide antibodies were produced in rabbits against
were detected by immunodecoration with anti-g-adaptin antibodies
either N-terminal or C-terminal peptides, which were
(clone 100/3, Sigma), a cross-reacting m1A/B antibody as in (B), and
the m1B-specific antibody raised against the C terminus of m1B.
only z50% identical to the corresponding peptides in

Basolateral Targeting Adaptor Complex
191
Figure 3. m1B Is an Alternative Subunit of the Adaptor Complex AP-1
(A) LLC-PK1::m1B transfectants (lanes 1 and 3) or LLC-PK1::m1A transfectants (lane 2) were lysed in Triton X-100 buffer and subjected to
coimmunoprecipitations using anti-g-adaptin antibodies. Immunoprecipitates were analyzed by SDS–PAGE and immunoblotting using anti-
b1/2-adaptin antibodies (clone 100/1), anti-g-adaptin antibodies (clone 100/3, Sigma), m1A/B cross-reacting antibodies, and m1B-specific
antibodies raised against the C terminus of m1B.
(B) LLC-PK1::m1B transfectants were metabolically labeled with [
35
S]methionine/cysteine overnight and lysed in Triton X-100 buffer. AP-1
complex was precipitated using anti-g-adaptin antibodies (clone 100/3). The precipitates were boiled in SDS buffer, and one-fourth was
directly subjected to SDS–PAGE analysis. The remaining extract was diluted in lysis buffer, and subunits of the AP-1 complex wererecaptured
usingpreimmune antibodies, anti-g-adaptinantibodies (clone 100/3),or anti-m1B antibodiesdirected against theC terminus ofm1B. Immunopre-
cipitates were analyzed by SDS–PAGE and autoradiography.
(C) Gel filtration analysis of LLC-PK1::m1B cell lysates. LLC-PK1::m1B transfectants were lysed in Triton X-100 buffer and subjected to gel
filtration analysis as described in Experimental Procedures. The proteins a-adaptin (subunit of AP-2), g-adaptin (subunit of AP-1), b4 (subunit
of AP-4), s3 (subunit of AP-3), and m1B were detected in the eluate fractions by Western blotting and immunolabeling using subunit-specific
antibodies.
cells transfected with m1A cDNA (LLC-PK1::m1A) or with in nontransfected LLC-PK1 or LLC-PK1::m1A transfec-
tants (Figure 2C, lanes 1–6). Although only one repre-m1B cDNA (LLC-PK1::m1B) were lysed and analyzed by
Western blotting. Both the parental and transfected cell sentative clone of LLC-PK1::m1A and LLC-PK1::m1B
is shown, analysis of several independently isolatedlines were found to express equivalent amounts of
g-adaptin, one of the large subunits of AP-1 (Figure 2C). clones yielded essentially the same results.
The cross-reacting anti-m1A antibody revealed a slight
increase of total m1 levels in both LLC-PK1::m1A (Figure m1B Is an Alternative Subunit of the AP-1
Adaptor Complex2C, lanes 4–6) and LLC-PK1::m1B transfectants (Figure
2C, lanes 7–9) as compared to the nontransfected LLC- Since m1A and m1Bare highly homologous (79% identity
at the amino acid level) (Ohno et al., 1999), it seemedPK1 cells (Figure 2C, lanes 1–3). In both cases, the in-
creases were 2- to 4-fold relative to endogenous m1A. likely that m1B might be a component of the ubiquitously
expressed AP-1 complex. To determine whether thisFinally, the m1B-specific antibody confirmed m1B ex-
pression only in LLC-PK1::m1Btransfectants (Figure2C, was the case, we immunoprecipitated AP-1 from LLC-
PK1::m1A transfectants or LLC-PK1::m1B transfectantslanes 7–9, lower panel). Virtually no signal was detected

Cell
192
using anti-g-adaptin antibodies. As shown in Figure 3A,
anti-g-adaptin antibodies coprecipitated g-adaptin, b1,
and m1A from lysates of LLC-PK1::m1A cells (Figure 3A,
lane 2). When cell lysate from LLC-PK1::m1B transfec-
tants was used, m1B was also coimmunoprecipitated
(Figure 3A, lane 3).
To further prove that m1B is a subunit of an AP-1
complex, we next metabolically labeled LLC-PK1::m1B
cells with [
35
S]methionine/cysteine. Cells were lysed,
andimmunoprecipitations using anti-g-adaptinantibod-
ies were performed. The primary coimmunoprecipitate
was then boiled in SDS to dissociate the AP-1 complex.
The boiled extract was diluted in lysis buffer and used
for a second round of immunoprecipitations using either
anti-gor anti-m1B antibodies.
As shown in Figure 3B, total immunoprecipitates con-
tained labeled bands corresponding to each of the four
expected AP-1 subunits: b1, g,m1A or m1B, and s1
(Figure3B, lane2). In thesecond precipitationof dissoci-
ated subunits, anti-g-adaptin antibodies precipitated
only g-adaptin (Figure 3B, lane 4), while antibodies to
the C-terminal m1B peptide precipitated only m1B (Fig-
ure 3B, lane 5); m1B could also be recaptured using the
antibodyagainst theN terminusof m1B (datanot shown).
Thesedata strongly suggestthat m1B becomesincorpo-
rated into an AP-1 complex in the m1B transfectants.
When LLC-PK1::m1A transfectants were used for this
experiment,m1B could not be precipitated in the second
round (data not shown).
Figure 4. AP-1B Mediates Basolateral Sorting of LDL Receptor and
Tfn Receptor
To exclude the possibility that m1B also assembled
into other known adaptor complexes (i.e., AP-2, AP-3,
LLC-PK1::m1A transfectants (left panels) or LLC-PK1::m1B transfec-
tants (right panels) were grown on Transwell filters for 4 days and
or AP-4), we performed gel filtration chromatography
infected with recombinant adenoviruses for 1 hr and cultured for 2
on cytosol from LLC-PK1::m1B transfectants using a
days to express LDL receptor (A) or human Tfn receptor (B). Cell
Superose 6 column. As indicated by Western blot analy-
surfacestaining with antibodies directed against LDL receptor or Tfn
sis of column fractions, m1B eluted in two peaks (Figure
receptor, respectively, was achieved as described in Experimental
3C). One peak coeluted with AP-1, as indicated by the
Procedures. Specimens were analyzed by confocal microscopy,
position of g-adaptin at an apparent molecular weight
and representative X-Y and X-Z sections are shown. The arrows in
(A) mark the position of the filters.
of 270 kDa. The second peak eluted with a molecular
weight of z65 kDa and thus presumably represented
unassembled and possibly monomeric m1B. Impor- allow LLC-PK1 cells to localize LDL receptors at the
tantly, m1B did not coelute with subunits indicative of basolateral surface. For this purpose, LLC-PK1 cells
any other adaptor complex, each of which exhibited transfected with the m1A or m1B cDNAs were infected
different apparent molecular weights as reported pre- with adLDLR and analyzed for cell surface appearance
viously (Dell’Angelica et al., 1997, 1999a). These data of the receptor. As shown in Figure 4A (left panel), LLC-
suggest that m1B can only assemble into AP-1 com- PK1 cells transfected with m1A expressed the receptor
plexes in the transfected LLC-PK1 cells, with overex- only at the apical plasma membrane, as found for paren-
pressed m1B existing as unassembled monomer rather tal LLC-PK1 cells (Figure 1B). LLC-PK1 cells transfected
than entering AP-2, AP-3, or AP-4 complexes. with m1B, however, exhibited a dramatic redistribution
Thus, two alternative AP-1 complexes can exist: one of LDL receptor to the basolateral plasma membrane
containing the ubiquitously expressed m1A (AP-1A), and (Figure 4B, right panel). Basolateral polarity was clearly
the other the epithelial cell-specific m1B (AP-1B). Unfor- evident in virtually all infected cells in both X-Y confocal
tunately, our m1B antibodies were not suitable for immu- images and X-Z vertical sections. These images were
nofluorescence microscopy, as found for all other anti-mindistinguishable from those obtained for infected MDCK
chainantibodies generatedto date(Simpson etal., 1996; cells that express m1B endogenously (Figure 1A).
Dell’Angelica et al., 1999a, 1999b). It was, therefore, Another well-studied basolateral receptor is the hu-
impossible to easily compare the intracellular distribu- man transferrin receptor (Tfn receptor), which also relies
tion of AP-1A and AP-1B. We were able to demonstrate, on a cytoplasmic tail signal for targeting to the basolat-
however, that the two subunits were associated with eral plasma membrane domain (Odorizzi and Trow-
markedly different targeting functions. bridge, 1997). We prepared a recombinant adenovirus
encoding the wild-type receptor (adTfnR), infected m1A-
m1B Confers Basolateral Polarity of LDL and and m1B-expressing LLC-PK1 cells, and then assayed
Transferrin Receptors in LLC-PK1 Cells for cell surface appearance of the receptor. As shown
Having demonstrated that m1B expression in LLC-PK1 in Figure 4B (left panel), m1A-expressing LLC-PK1 cells
cells results in its incorporation in an AP-1 adaptor com-
plex, we next asked whether m1B expression would exhibited Tfn receptor at both the apical and basolateral

Basolateral Targeting Adaptor Complex
193
encoding p75
NTR
and assayed for surface expression of
the receptor. As shown in Figure 5A, p75
NTR
was found
exclusively at the apical surface of both m1A- and m1B-
expressing LLC-PK1 cells.
We next examined the expression of a second protein
that is strictly apical in MDCK cells. A recombinant ade-
novirus was constructed encoding a chimeric mem-
brane protein consisting of the extracellular region and
membrane anchor of the murine Fc receptor (FcRII)
fused to the cytoplasmic domain of the LDL receptor,
truncated at position 22 (FcL [CT22] receptor). This trun-
cation has previously been shown to delete all basolat-
eral targeting information without affecting the recep-
tor’s endocytosis signal (dependent on the tyrosine at
position 18) (Matter et al., 1994). As found for p75
NTR
,
FcL (CT22) receptor was expressed largely at the apical
surface of LLC-PK1 cells regardless of whether they had
been transfected with m1A or m1B (Figure 5B).
Taken together, these results demonstrate that m1B
expression in LLC-PK1 cells was only able to induce
the basolateral targeting of proteins that contained ba-
solateral targeting signals. Thus, a m1B-containing AP-1
adaptor complex appears to be required and perhaps
responsible for signal-mediated basolateral sorting.
m1B Expression Enhances Overall Organization
of Polarized LLC-PK1 Cells
Not all membrane proteins depend on cytoplasmic do-
main targeting signals for sorting to the basolateral sur-
Figure 5. Apical Sorting Is Not Affected in LLC-PK1::m1B Transfec-
tants
face of epithelial cells. Some proteins, such as E-cad-
LLC-PK1 cells transfected with m1A cDNA (left panels) or m1B cDNA
herin or the Na,K-ATPase, achieve basolateral polarity
(right panels) were grown on Transwell filters and infected with
by homotypic interactions with proteins on adjacent
recombinant adenoviruses for 1 hr and cultured for 2 days to allow
cells or by interactions with the cytoskeleton at lateral
for expression of p75
NTR
(A) or FcL (CT22) receptor (B). Cell surfaces
surfaces subsequent to cell–cell contact (Mays et al.,
were stained with antibodies directed against the p75
NTR
or FcL
1995). Although parental LLC-PK1 cells were unable to
receptor, respectively, and analyzed by confocal microscopy as
decode at least some cytoplasmic tail basolateral tar-
described in Experimental Procedures. Shown are representative
confocal images of X-Y and X-Z sections.
geting signals, it was possible that they nevertheless
formed a basolateral domain by other means. We there-
fore next determined the polarity of endogenous Na,K-
ATPase in cells expressing m1A or m1B. In both cases,plasma membrane domains. This was consistent with
the “random” phenotype of a tail-minus Tfn receptor Na,K-ATPase distribution was largely restricted to the
lateralplasma membrane,corresponding tosites of cell–mutant expressed in MDCK cells (Odorizzi and Trow-
bridge, 1997). In contrast, when adTfnR was used to cell contact (Figure 6A). Thus, a protein whose polarity
apparently depends on its ability to interact with cy-infect m1B-expressing LLC-PK1 cells, surface receptor
was detected uniquely at the basolateral domain, a lo- toskeletal elements appeared to reach the basolateral
domain even in the absence of m1B.calization that was evident in every infected cell and
clearly seen in both X-Y and X-Z confocal sections (Fig- Closer inspectionof LLC-PK1 cellmonolayers stained
with Na,K-ATPase antibody suggested that m1Bexpres-ure 4B, right panel). Thus, m1B expression in LLC-PK1
cells confers the capacity for basolateral localization of sion had the effect of yielding cells with a more regularly
organized monolayer. To examine this more directly,another membrane protein that is well known to be
sorted to the basolateral surface in MDCK cells. m1A- and m1B-expressing LLC-PK1 cells were fixed and
stained with FITC-coupled phalloidin, which outlines the
overall shape of each cell by revealing the localizationPolarity of Apical Proteins Is Not Affected
by m1B Expression in LLC-PK1 Cells of filamentous actin. As shown in Figure 6B (left panel),
m1A-expressing cells (and their nontransfected parents)To determine whether m1B-induced basolateral sorting
is selective for membrane proteins that bear basolateral oftengrew on polycarbonatefilters asirregularly shaped
discontinuous cell sheets. Frequently, the cells piled uptargeting signals, we next infected the m1A- and m1B-
transfected LLC-PK1 cells with viruses encoding pro- on each other several deep in a disorganized fashion.
m1B-expressing LLC-PK1 cells, on theother hand, grewteins that are apical in MDCK cells. We first examined
the polarized expression of the p75
NTR
isoform of nerve in a far more regular fashion (Figure 6B, right panel).
Although individual cells were not of uniform size (as isgrowth factor receptor. p75
NTR
contains an O-glycosy-
lated stalk inits extracellular domain that is required for characteristic of MDCK cell cultures), they exhibited a
strict monolayer arrangement and covered the filter sur-apical targeting (Yeaman et al., 1997). Transfected LLC-
PK1 cells were infected with a recombinant adenovirus face in an essentially continuous fashion.

Cell
194
membrane. Nevertheless, it is highly likely that AP-1B
adaptors do act at the level of sorting as opposed, for
example,to selectiveretention ofproteins followingtheir
appearance at the basolateral surface. First, the closely
related AP-1A, AP-2, and AP-3 adaptor complexes are
well known to mediate membrane protein sorting rather
than retention (Mellman, 1996; Marks et al., 1997; Hirst
and Robinson, 1998). In addition, the signals apparently
utilized by AP-1B for basolateral targeting clearly act by
facilitating polarized sorting, both in the TGN and in
endosomes (Matter et al., 1993; Aroeti and Mostov,
1994). Indeed, the fact that the same signals are de-
coded at both sites suggests that the AP-1B adaptor
complex similarly works in both the secretory and endo-
cytic pathways. In this regard, it may be of interest that
g-adaptin has been found on clathrin-coated buds on
endosomes as well as the TGN but not at the plasma
membrane (Futter et al., 1998), although these earlier
experiments could not distinguish AP-1A and AP-1B
complexes.
AP-1B Complexes and the Decoding of Basolateral
Targeting Signals
It seems likely that m1B acts by directly recognizing
basolateral targeting signals. Although this point re-
Figure 6. Expression of m1B Induces More Regular Monolayer
mains to be directly demonstrated, it is by far the sim-
Growth
plest explanation and consistent with the ability of m
LLC-PK1::m1A transfectants (left panels) or LLC-PK1::m1B transfec-
chains to bind tyrosine-based signals for endocytosis
tants (right panels) were grown on Transwell filters for 4 days, fixed,
and labeled with antibodies directed against the Na,K-ATPase (A)
and lysosomal targeting (Bonifacino and Dell’Angelica,
or with FITC-phalloidin (B). Specimens were analyzed by confocal
1999). Indeed, we first became interested in m1B on the
microscopy, and representative X-Y and X-Z sections are shown.
basis of a yeast two-hybrid screen in which the LDL
receptor tail was challenged for interaction against all
known mchains. Only m1B was found to interact, albeit
These results suggest that LLC-PK1 cells are able to very weakly, with the LDL receptor tail; the interaction
achieve some measure of apical and basolateral polarity was dependent on a tyrosine residue at cytoplasmic tail
even in the absence of m1B. However, they are unable position 35 (R. C. Aguilar and J. S. B., unpublished data).
to correctly localize proteins containing basolateral tar- In addition, preliminary chemical cross-linking experi-
geting signals and also are unable to grow in regular ments also suggest that the LDL receptor interacts with
monolayers unless m1B was expressed by transfection. AP-1B, but not AP-1A, complexes in transfected LLC-
PK1 cells (H. F. and I. M., unpublished data).
Discussion Regardless of whether m1B functions by directly inter-
acting with basolateral targeting signals, it is remarkable
We have identified an important element of the mecha- that m1B can restore the polarized expression of such
nism responsible for targeting membrane proteins to a wide array of basolateral proteins. While the LDL re-
the basolateral surface of epithelial cells. Our results ceptor clearly relies on critical tyrosine residues for ba-
suggest that final localization of a wide variety of mem- solateral sorting (Matter et al., 1992), Tfn receptor may
brane proteins at the basolateral surface depends on not (Odorizzi and Trowbridge, 1997). Moreover, the fact
theexpression of m1B, an epithelial cell-specific compo- that m1B expression greatly improves the overall mono-
nent of the AP-1 clathrin adaptor complex. We propose layer organization of LLC-PK1 cells suggests that it also
that this AP-1B adaptor complex assembles with mem- induces the basolateral localization of one or more en-
branes at the level of the TGN, perhaps together with dogenous proteins (e.g., integrins) which play an impor-
clathrin, to form nascent secretory vesicles that selec- tant role in epithelial cell morphogenesis. In addition to
tively accumulate proteins destined for the basolateral such itinerant cargo proteins, m1B may also be required
surface. The selective accumulation of cargo into these forpackaging of specificv-SNAREs requiredfor efficient
vesicles is likely to reflect the interaction of the AP-1B basolateral fusion.
complex with at least a subset of distinct if degenerate In a yeast two-hybrid assay against peptides isolated
targeting signals found in the cytoplasmic domains of from a combinatorial library, m1B has been found to
many basolateral proteins. interact with a subset of tyrosine-containing motifs con-
Due to the leakiness of the monolayers of parental as forming to the canonical sequence YXXφ(Ohno et al.,
well as transfected LLC-PK1 cells, it was impossible to 1999). This finding is consistent with the conservation
perform the type of vectorial biotinylation experiments of a YXXφ-binding site on m1B, as predicted from the
needed to establish directly that expression of AP-1B crystal structure of m2 (Owen and Evans, 1998). How-
adaptors conferred the capacity to sort apical from ba- ever, while some basolateral targeting signals, particu-
larly those that are colinear with endocytosis signals,solateral proteins prior to their delivery to the plasma

Basolateral Targeting Adaptor Complex
195
appear to conform to this arrangement (Mellman, 1996), apical markers. This could explain why export of baso-
lateral proteins, such as VSV G protein, from the TGNthis seems not to be the case for the LDL or Tfn receptor
basolateral signals. The LDL receptor tail contains two of fibroblasts may not involve clathrin (Griffiths et al.,
1985), although a role for clathrin in TGN to plasmatyrosine-based, non-YXXφbasolateral targeting deter-
minants within its cytoplasmic tail (Matter et al., 1992). membrane transport has not yetbeen definitivelyinves-
tigated in any cell type. The coats on such vesicles mayThe membrane-proximal determinant comprisesthe en-
docytic signal NPVY and the acidic cluster EDE, while be extremely evanescent, making their identification dif-
ficult by conventional techniques.the membrane-distal determinant contains a GYSY se-
quence and an EDD acidic cluster. The Tfn receptor
does contain a YXXφ-type signal, YTRF, but this seems Polarized Sorting in the Absence of m1B
to mediate only endocytosis and not basolateral tar- Our results provide a plausible explanation for the re-
geting (Odorizzi and Trowbridge, 1997). Targeting of the cently described sorting “defect” in LLC-PK1 cells for
Tfn receptor to the basolateral surface is instead depen- two proteins bearing tyrosine-based sorting signals, the
dent, at least in part, on a GDNS sequence in the cyto- H,K-ATPase bsubunit and an influenza hemagglutinin
plasmic tail (Odorizzi and Trowbridge, 1997). Thus, it mutant. Although both proteins are sorted to the baso-
is likely that the mode of binding of these basolateral lateral plasma membrane in MDCK cells, they are apical
targeting determinants to m1B is distinct from that for in LLC-PK1 cells (Roush et al., 1998). Thus, the basolat-
YXXφsignals. The failure of m1A to substitute for m1B eraltransport ofthese proteinswould appear dependent
in basolateral targeting could be explained by the ab- on m1B.
sence of such a binding site for non-YXXφ-type signals Interestingly, AP-1B is apparently not required for the
on m1A. basolateral expression of another membrane protein,
anIgG Fc receptor FcRII-B2,which contains a dileucine-
type targeting signal (Hunziker and Fumey, 1994; MatterFormation and Fate of AP-1B-Coated Vesicles
Given the close relationship between m1A and m1B, it et al., 1994; Roush et al., 1998). At least in vitro, dileucine
signals interact with brather than msubunits (Rapoportseems likely that transport vesicles formed with AP-1B
adaptors should contain clathrin, as is the case for “con- et al., 1998). This observation suggests that, under some
circumstances, AP-1A adaptors may be able to mediateventional” AP-1A-coated vesicles involved in TGN to
endosome transport (Mellman, 1996; Hirst and Rob- basolateral targeting, or that there are yet additional
functional pathways remaining to be discovered.inson, 1998).Basolateral clathrin-coated vesicles might
thuscontain both AP-1Aand AP-1B adaptorcomplexes, That other, AP-1B-independent mechanisms for po-
larized targeting exist iswithout question. Bothrat hepa-suggesting their cargo might be similarly mixed. Indeed,
in MDCK cells, lysosomal enzymes and membrane pro- tocytes and hippocampal neurons are clearly polarized,
with both relying on basolateral-type targeting signalsteins when missorted to the plasma membrane are in-
variably targeted basolaterally, conceivably reflecting for transport to their sinusoidal and somatodendritic
surfaces, respectively (Weisz et al., 1992; Yokode et al.,the partial inclusion ofAP-1A-bound cargo in basolater-
ally targeted vesicles (Hunziker et al., 1991; Nabi et al., 1992; Jareb and Banker, 1998; Winckler and Mellman,
1999). Yet, neither cell type expresses m1B mRNA (Ohno1991). It remains possible that AP-1A and AP-1B com-
plexes differ in other ways that ensure a better subdivi- et al., 1999). It is possible that other cell type–specific
mor other adaptor subunits exist that mediate polarizedsion of their activities. For example, although the g,b,
and ssubunits that coprecipitate with m1B are similar sorting in hepatocytes and neurons. Another possibility
relates to the existence of different strategies for polar-to the corresponding subunits in m1A-containing com-
plexes, they may also represent as yet unidentified, AP- ized targeting. In hepatocytes, for example, basolateral
and apical proteins are not sorted in the TGN but rather1B-specific isoforms.
The existence of m1B in epithelial cells strongly sug- are delivered together to the sinusoidal (basolateral)
plasma membrane and then sorted in endosomes fol-gests that there are, in fact, some significant differences
between how polarized and “nonpolarized” cells sort lowing internalization (Hubbard et al., 1989; Wilton and
Matthews, 1996). Perhaps the ability to interact withmembrane proteins in the TGN. In epithelial cells, recog-
nition of basolateraltargeting signals by AP-1B appears AP-1A, or an entirely novel sorting complex, at the level
of endosomes precludes the need for AP-1B, which maycrucial for the formation of transport vesicles competent
for fusion with thebasolateral plasma membrane. Apro- be required for sorting only in the TGN. Thus, despite
the clear importance of AP-1B in the organization oftein not actively selected into these vesicles either be-
cause of lacking a sorting signal encoded in its cyto- many types of epithelial cells, its role is played in the
context of other mechanisms that act in concert to en-plasmic tail orbecause the respectiveadaptor complex
is not expressed in the given cell line often segregates sure the generation and maintenance of cellular asym-
metry.into glycolipid rafts and is subsequently delivered to the
apical membrane. Studies in fibroblasts have suggested
that the raft mechanism exists even in nonpolarized cells,
Experimental Procedures
leading to the segregation of cognate apical and baso-
Cloning and Expressing of m1B and m1A
lateral proteins into distinct transport vesicles (Mu
¨sch
All constructs were cloned by PCR using mouse m1A cDNA (Gen-
et al., 1996; Yoshimori et al., 1996). However, the ab-
Bank No. M62419) or human m1B cDNA (I.M.A.G.E. consortium,
sence of m1B expression in such cells would imply that
LLDL,ID 123283) astemplates and thePfu-polymerase (Stratagene).
the putative basolateral vesicles may, instead, actually
Both genes were cloned into the bacterial expression vector pET-
15b (Novagen) providing an N-terminal His
6tag
for overexpression in
be nonselective “default” vesicles, depleted of cognate

Cell
196
E. coli or into the CMV-based mammalian expression vector pCB6 min at room temperature. Subsequently filters were cut out and
incubated in a blocking solution (2% BSA [w/v], 0.1% saponin [w/v]
for transfection of LLC-PK1 cells. in PBS
11
) for 1 hr followed bya1hrincubation with the fluorescently
The N-terminal and C-terminal primers for cloning of m1A into labeled secondary antibody diluted in blocking solution in a wet
pET-15b were 59-GCGCGAATTCCTCGAGATGTCCGCCAGCGCCGT chamber. The filters were washed four times over a total period of
CTACGTA-39and 59-GCGCGGATCCTCACTGGGTCCGGAGCTG-39,30 min in blocking solution and finally mounted in a glycerol solution
respectively. The PCR product was cloned as a XhoI/BamHI frag- (50% glycerol [w/v], 10% DABCO [w/v] in PBS). For total staining,
ment. For cloning into pCB6, m1A was amplified using the N-terminal the cultures were first incubated with the primary antibody for 1 hr,
primer as above and the following C-terminal primer 59-GCGCAAG washed, and incubated with the secondary antibody as described
CTTTCACTGGGTCCGGAGCTG-39and cloned as an EcoRI/HindII above. The preparations were analyzed using a Zeiss confocal mi-
fragment. croscope (Microsystem LSM) with an Axiovert 100 microscope and
m1B amplification for cloning into pET-15b was achieved by using a Zeiss Plan-Neofluar 4031.3 oil immersion objective.
the following N-terminal and C-terminal primers 59-GCGCGAATTC
CTCGAGATGTCCGCCTCGGCTGTCTTCATT-39and 59-GCGCAGA Biochemical Procedures
TCTGTCGACCTAGCTGGTACGAAGTTGGTAATCGCC-39, respec- For immunoprecipitations or gel filtration analysis, LLC-PK1 cells
tively, and cloned as a XhoI/BglII fragment. Cloning of m1B into were split 1:1 or 1:2 in six-well plates 1 day prior to the experiment.
pCB6 was achieved using the same N-terminal primer as for cloning The cells were washed twice with ice-cold PBS
11
on ice. Buffer A
into pET-15b and 59-GCGCAAGCTTCTAGCTGGTACGAAGTTG-39(1% Triton X-100 [w/v], 0.3 M NaCl, 13protease inhibitors [Boeh-
as C-terminal primer. The PCR product was cloned as an EcoRI/ ringer], 50 mM Tris-HCl [pH 7.4], 0.1% BSA [w/v]) was added to the
HindII fragment. samples, the cells were scraped with a cell scraper, and then passed
For overexpression in E. coli, the genes cloned into pET-15b were four times through a 22 gauge needle and a 1 ml syringe. Lysis was
transformed into the E. coli strain BL21 (Novagen), overexpressed completed for 30 min on ice. A clarifying spin was done (15 min at
and partially purified under denaturing conditions using Ni-NTA 13,000 rpm, Eppendorf centrifuge at 48C for immunoprecipitations
chromatography. or 30 min at 100,000 3g for gel filtration analysis), and the superna-
For expression in LLC-PK1 cells, pCB6 harboring m1A or m1B tant was used for further analysis. For Western blot analysis, cells
cDNA was transfected using the calcium phosphate precipitation were lysed in RIPA buffer (50 mM Tris-HCl [pH 7.6], 150 mM NaCl,
method as described (Hunziker et al., 1991). Positive transfectants 1% Triton X-100 [w/v], 0.5% desoxycholate [w/v], 0.1% SDS [w/v],
were selected and maintained in growth medium supplemented with 13protease inhibitors [Boehringer]), and the protein concentration
1.8 mg/ml Geneticin. was determined.
For coimmunoprecipitations of m1B with antibodies against
Antibodies g-adaptin, the mouse monoclonal antibody 100/3 (Sigma) was
m1B-specific polyclonal antibodies were generated by injecting the bound to protein G–Sepharose. Lysis supernatant was added to the
peptides KGALAPLLSHGQVH and KEKEEVEGRPPIGV correspond- beads and incubated for 1 hr end-over-end at 48C. Coimmunopre-
ing to the N-terminal or C-terminal region of the human m1B protein, cipitates were washed two times with buffer A and once with buffer
respectively, into rabbits. Polyclonal antibodies against m1A were A without Triton X-100. The samples were analyzed by SDS–PAGE
a gift of Linton Traub (Washington University). Polyclonal anti-p75
NTR
and immunoblotting.
(9992) antibodies were obtained from Moses Chao (Skirball Insti- For recapturing experiments, cells were labeled with 2 mCi/ml
tute). s3 and b4 antibodies were as previously described (Dell’Ange- [
35
S]methionine/cysteine (Promix, Amersham) in a mixture of 91%
lica et al., 1997, 1999a). Monoclonal antibodies against b1/2 (clone RPMI medium without methionine/cysteine (110% [v/v] dialyzed
100/1), a-adaptin (clone 100/2), and g-adaptin (clone 100/3) were FBS) and 9% aMEM (110% FBS [v/v], 12 mM L-glutamine) for
purchased from Sigma. 14–16 hr prior to lysis. Immunoprecipitates obtained with anti-g-
adaptin antibody were washed five times in buffer A, and SDS bufferHybridoma cell lines producing the following monoclonal antibod- was added (0.1 M Tris-HCl [pH 7.4], 1% SDS [w/v], 10 mM DTT).
ies were purchased from the ATCC: anti-LDL receptor antibody (C7), The samples were vigorously shaken for 20 min at 48C and boiled
anti-FcL receptor antibody (2.4G2), and anti-human Tfn receptor for 5 minat 958C to releasethe immunoprecipitates from theantibod-
antibody (H68.4). Anti-Na,K-ATPase antibody (6H) was obtained ies and to denature the AP-1 complex. One-fourth of the extract
from Michael Caplan (Yale University). was directly analyzed by SDS–PAGE. The remaining extract was
FITC-phalloidin and FITC-labeled secondary antibodies were pur- diluted 20-fold with buffer A followed by a clarifying spin (15,000
chased from Sigma or Jackson Immuno Research, respectively. rpm, 48C, Eppendorf centrifuge) and incubated end-over-end at 48C
for 1 hr with anti-m1B antibodies, the corresponding preimmune
Cell Culture antibodies, or anti-g-adaptin IgGs coupled to protein A or protein
LLC-PK1 cells were grown in a-MEM containing 10% (v/v) fetal G–Sepharose, respectively. Immunoprecipitates were washed twice
bovine serum (FBS), 2 mM L-glutamine at 378C in a 5% CO
2
incuba- with buffer A and once with buffer A without Triton X-100. The
tor. Stably transfected cell lines were maintained with 1.8 mg/ml samples were analyzed by SDS–PAGE and autoradiography.
Geneticin in the medium. For polarity experiments, cells were plated For gel filtration analysis, cells from one chamber of a six-well
at a density of 8 310
5
cells per 24 mm filter on polycarbonate plate were lysed in 400 ml of buffer A. Two hundred microliters of
membrane filters (Corning-Costar Transwell units, 0.4 mm pore size) lysis supernatant was subsequently applied to a Superose 6 gel
and cultured for 4 days with changes of medium every day. filtration column (25 ml column volume, Pharmacia) equilibrated with
The construction of adenoviruses encoding for human Tfn recep- buffer B (0.5 mM EDTA, 0.5 mM PMSF, 1% Triton X-100, 0.3 M NaCl,
toror FcL (CT22) receptor(Matter et al.,1994) was doneas described 50 mM Tris-HCl [pH 7.4]). Fractions (0.5 ml) were collected and
(He et al., 1998). The adp75
NTR
and adLDL receptor were gifts from precipitated by adding TCA to a final concentration of 10% (w/v)
Moses Chao (Skirball Institute) and James Wilson (University of and analyzed by SDS–PAGE and Western blotting.
Pennsylvania), respectively.
For infection with defective adenoviruses, cultures were washed Miscellaneous
once in serum-free medium, and 50–100 plaque-forming units (pfu) SDS–PAGE was performed according to the published method of
of the viruses was added to the apical chamber. After 1 hr incubation Laemmli (1970). Protein determination and detection of proteins
at 378C, the medium was exchanged with serum-containing me- after blotting onto nitrocellulose were performed using the BCA
dium. The cells were prepared for immunofluorescence analysis 2 protein determination assay or the supersignal detection system,
days after the infection. respectively, according to the supplier’s instructions (Pierce).
For cell surface staining, the cultures were washed twice with
PBS
11
(PBS [2 g/l KCl, 2 g/l KH
2
PO
4
, 8 g/l NaCl, 1.15 g/l Na
2
HPO
4
,Acknowledgments
pH 7.4] plus 147 g/l CaCl
2
32H
2
O, 1 g/l MgCl
2
36H
2
O), and
antibodies were added to the apical and basolateral side. After an We would like to thank Michael Hull for his expertise and assistance
incubation of 7.5 min at room temperature, cultures were washed in preparing recombinant adenoviruses, Moses Chao (Skirball Insti-
tute) and James Wilson (University of Pennsylvania) for providing thetwice with PBS
11
and fixed in 3% paraformaldehyde/PBS
11
for 15

Basolateral Targeting Adaptor Complex
197
p75
NTR
and LDL receptor adenovirus stocks, respectively, Michael Keller, P.,and Simons, K.(1997). Post-Golgi biosynthetic trafficking.
J. Cell Sci. 110, 3001–3009.Caplan (Yale University) for providing the LLC-PK1 cells and anti-
bodies to Na,K-ATPase, and Linton Traub (Washington University) Laemmli, U.K. (1970). Cleavage of structural proteins during the
for anti-m1A antibodies. We are indebted to Michael Caplan and assembly of the head of bacteriophage T4. Nature 227, 680–685.
Graham Warren for helpful discussions, and to the entire Mellman/ Le Bivic, A., Sambuy, Y., Patzak, A., Patil, N., Chao, M., and Rodri-
Warren laboratory for their support. This work was funded by a guez-Boulan, E. (1991). An internal deletion in the cytoplasmic tail
postdoctoral fellowship from the Deutsche Forschungsgemein- reverses the apical localization of human NGF receptor in trans-
schaft (to H. F.) and research grants from the National Institutes fected MDCK cells. J. Cell Biol. 115, 607–618.
of Health (GM-29765) to I. M., and by Grant-in-Aids for Scientific Marks, M.S., Ohno, H., Kirchhausen, T., and Bonifacino, J.S. (1997).
Research from the Ministry of Education, Science, Sports and Cul- Protein sorting by tyrosine-based signals: adapting to the Ys and
ture of Japan (to H. O.). I. M. is an Affiliate Member of the Ludwig wherefores. Trends Cell Biol. 7, 124–128.
Institute for Cancer Research. Matter, K., and Mellman, I. (1994). Mechanisms of cell polarity: sort-
ing and transport in epithelial cells. Curr. Opin. Cell Biol. 6, 545–554.
Received August 10, 1999; revised September 10, 1999. Matter, K., Hunziker, W., and Mellman, I. (1992). Basolateralsorting
of LDL receptor in MDCK cells: the cytoplasmic domain containsReferences two tyrosine-dependent targeting determinants. Cell 71, 741–753.
Aroeti,B., and Mostov, K.E.(1994). Polarized sortingof the polymeric Matter, K., Whitney, J.A., Yamamoto, E.M., and Mellman, I. (1993).
immunoglobulin receptor in the exocytotic and endocytotic path- Common signals control low density lipoprotein receptor sorting in
waysis controlled bythe same amino acids.EMBO J. 13,2297–2304. endosomes and the Golgi complex of MDCK cells. Cell 74, 1053–
1064.Bonifacino, J.S., and Dell’Angelica, E.C. (1999). Molecular bases for
the recognition of tyrosine-based sorting signals. J. Cell Biol. 145,Matter, K., Yamamoto, E.M., and Mellman, I. (1994). Structural re-
923–926. quirementsand sequence motifs forpolarizedsorting and endocyto-
sis of LDL and Fc receptors in MDCK cells. J. Cell Biol. 126, 991–Casanova, J.E., Apodaca, G., and Mostov, K.E. (1991). An autono-
mous signal for basolateral sorting in the cytoplasmic domain of 1004.
the polymeric immunoglobulin receptor. Cell 66, 65–75. Mays, R.W., Siemers, K.A., Fritz, B.A., Lowe, A.W., van Meer, G., and
Nelson, W.J.(1995).Hierarchy ofmechanisms involvedin generatingCohen, A.R., Woods, D.F., Marfatia, S.M., Walther, Z., Chishti, A.H.,
Anderson, J.M., and Woods, D.F. (1998). HumanCASK/LIN-2 binds Na/K-ATPase polarity in MDCK epithelial cells. J. Cell Biol. 130,
1105–1115.syndecan-2 and protein 4.1 and localizes to the basolateral mem-
brane of epithelial cells. J. Cell Biol. 142, 129–138. Mellman, I. (1996). Endocytosis and molecular sorting. Annu. Rev.
Cell Dev. Biol. 12, 575–625.Dell’Angelica, E.C., Ohno, H., Ooi, C.E., Rabinovich, E., Roche, K.W.,
and Bonifacino, J.S. (1997). AP-3: an adaptor-like protein complex Mu
¨sch, A., Xu, H., Shields, D., and Rodriguez-Boulan, E. (1996).
with ubiquitous expression. EMBO J. 16, 917–928. Transport of vesicular stomatitis virus G protein to the cell surface
is signal mediated in polarized and nonpolarized cells. J. Cell Biol.Dell’Angelica, E.C., Mullins, C., and Bonifacino, J.S. (1999a). AP-4,
a novel protein complex related to clathrin adaptors. J. Biol. Chem. 133, 543–558.
274, 7278–7285. Nabi, I.R., Le Bivic, A., Fambrough, D., and Rodriguez-Boulan, E.
(1991). An endogenous MDCK lysosomal membrane glycoproteinDell’Angelica, E.C., Shotelersuk, V., Aguilar, R.C., Gahl, W.A., and
Bonifacino, J.S. (1999b). Altered trafficking of lysosomal proteins in is targeted basolaterally before delivery to lysosomes. J. Cell Biol.
115, 1573–1584.Hermansky-Pudlak syndrome due to mutations in the beta 3A sub-
unit of the AP-3 adaptor. Mol. Cell 3, 11–21. Odorizzi, G., and Trowbridge, I.S. (1997). Structural requirements
for basolateral sorting of the human transferrin receptor in the bio-Drubin, D.G., and Nelson, W.J. (1996). Origins of cell polarity. Cell
84, 335–344. synthetic and endocytic pathways of Madin-Darby canine kidney
cells. J. Cell Biol. 137, 1255–1264.
Futter, C.E., Gibson, A., Allchin, E.H., Maxwell, S., Ruddock, L.J.,
Odorizzi,G., Domingo, D., Trowbridge,I.S., andHopkins, C.R. (1998). Ohno, H., Stewart, J., Fournier, M.C., Bosshart, H., Rhee, I., Miya-
take,S., Saito, T., Gallusser, A.,Kirchhausen, T., and Bonifacino, J.S.In polarized MDCK cells basolateral vesicles arise from clathrin-
gamma-adaptin-coated domains on endosomal tubules. J. Cell Biol. (1995). Interaction of tyrosine-based sorting signals with clathrin-
associated proteins. Science 269, 1872–1875.141, 611–623.
Griffiths, G., Pfeiffer, S., Simons, K., and Matlin, K. (1985). Exit of Ohno, H., Fournier, M.C., Poy, G., and Bonifacino, J.S. (1996). Struc-
tural determinants of interaction of tyrosine-based sorting signalsnewly synthesized membrane proteins from the trans cisterna of the
Golgi complex to the plasma membrane. J. Cell Biol. 101, 949–964. with the adaptor medium chains. J. Biol. Chem. 271, 29009–29015.
Ohno, H., Tomemori, T., Nakatsu, F., Okazaki, Y., Aguilar, R.C.,He, T.C., Zhou, S., da Costa, L.T., Yu, J., Kinzler, K.W., and Vo-
gelstein, B. (1998). A simplified system for generating recombinant Fo
¨lsch, H., Mellman, I., Saito, T., Shirasawa, T., and Bonifacino, J.S.
(1999). Mu1B, a novel adaptor medium chain expressed in polarizedadenoviruses. Proc. Natl. Acad. Sci. USA 95, 2509–2514. epithelial cells. FEBS Lett. 449, 215–220.
Hirst, J., and Robinson, M.S. (1998). Clathrin and adaptors. Biochim.
Biophys. Acta 1404, 173–193. Owen, D.J., and Evans, P.R. (1998). A structural explanation for
the recognition of tyrosine-based endocytotic signals. Science 282,
Hirst, J., Bright, N.A., Rous, B., and Robinson, M.S. (1999). Charac- 1327–1332.
terization of a fourth adaptor-related protein complex. Mol. Biol.
Cell 10, 2787–2802. Rapoport, I., Chen, Y.C., Cupers, P., Shoelson, S.E., and Kirch-
hausen, T. (1998). Dileucine-based sorting signals bind to the beta
Hubbard, A.L., Stieger, B., and Bartles, J.R. (1989). Biogenesis of chain of AP-1 at a site distinct and regulated differently from the
endogenous plasma membrane proteins in epithelial cells. Annu. tyrosine-based motif-binding site. EMBO J. 17, 2148–2155.
Rev. Physiol. 51, 755–770. Rodriguez-Boulan, E., and Powell, S.K. (1992). Polarity of epithelial
Hunziker, W., and Fumey, C. (1994). A di-leucine motif mediates and neuronal cells. Annu. Rev. Cell Biol. 8, 395–427.
endocytosis and basolateral sorting of macrophage IgG Fc recep-
tors in MDCK cells. EMBO J. 13, 2963–2967. Roush, D.L., Gottardi, C.J., Naim, H.Y., Roth, M.G., and Caplan,
M.J. (1998). Tyrosine-based membrane protein sorting signals are
Hunziker, W., Harter, C., Matter, K., and Mellman, I. (1991). Basolat- differentially interpreted by polarized Madin-Darby canine kidney
eral sorting in MDCK cells requires a distinct cytoplasmic domain and LLC-PK1 epithelial cells. J. Biol. Chem. 273, 26862–26869.
determinant. Cell 66, 907–920. Simons,K., and Ikonen, E.(1997). Functional rafts incell membranes.
Jareb, M., and Banker, G. (1998). The polarized sorting of membrane Nature 387, 569–572.
proteins expressed in cultured hippocampal neurons using viral
vectors. Neuron 20, 855–867. Simpson, F., Bright, N.A., West, M.A., Newman, L.S., Darnell, R.B.,

Cell
198
andRobinson, M.S. (1996).A novel adaptor-related proteincomplex.
J. Cell Biol. 133, 749–760.
Weisz, O.A., Machamer, C.F., and Hubbard, A.L. (1992). Rat liver
dipeptidylpeptidase IV contains competing apical and basolateral
targeting information. J. Biol. Chem. 267, 22282–22288.
Wilton, J.C., and Matthews, G.M. (1996). Polarized membrane traffic
in hepatocytes. Bioessays 18, 229–236.
Winckler, B., and Mellman, I. (1999). Generation and maintenance of
neuronal polarity: controlling the sorting and diffusion of membrane
components. Neuron 23, 637–640.
Yeaman, C., Le Gall, A.H., Baldwin, A.N., Monlauzeur, L., Le Bivic, A.,
andRodriguez-Boulan, E. (1997).The O-glycosylated stalk domainis
required for apical sorting of neurotrophin receptors in polarized
MDCK cells. J. Cell Biol. 139, 929–940.
Yeaman, C., Grindstaff, K.K., and Nelson, W.J. (1999). New perspec-
tives on mechanisms involved in generating epithelial cell polarity.
Physiol. Rev. 79, 73–98.
Yokode, M., Pathak, R.K., Hammer, R.E., Brown, M.S., Goldstein,
J.L., and Anderson, R.G. (1992). Cytoplasmic sequence required for
basolateral targeting of LDL receptor in livers of transgenic mice.
J. Cell Biol. 117, 39–46.
Yoshimori, T., Keller, P., Roth, M.G., and Simons, K. (1996). Different
biosynthetic transport routes to the plasma membrane in BHK and
CHO cells. J. Cell Biol. 133, 247–256.