Cell, Vol. 90, 1137±1148, September 19, 1997, Copyright 1997 by Cell Press
Visualization of ER-to-Golgi Transport
in Living Cells Reveals a Sequential Mode
of Action for COPII and COPI
Suzie J . Scales, Rainer Pepperkok,*
and Thomas E. Kreis²
Department of Cell Biology
University of Geneva
30 Quai Ernest-Ansermet
CH-1211 Geneva 4
membrane structures thought to constitute an interme-
diate station of transport between the ER and the Golgi
complex can be visualized under certain conditions
(e.g., reduced temperature or virus infection) (Saraste
and Kuismanen, 1984; Lotti et al., 1992; Krijnse Locker
et al., 1994; Bannykh et al., 1996). Among the terms
proposed to describe these membranes are vesicular±
tubular clusters (Balch et al., 1994) and IC (Schweizer
et al., 1990) (for a more comprehensive list, see Hauri
and Schweizer, 1992). It is still not clear whether this
IC does indeed constitute a bona fide compartment,
despite having a protein composition that appears dis-
tinct fromthatofthe ER andGolgicomplex (Schweizeret
al., 1991), since electron microscopy (EM) has revealed
direct connections of such tubulo-vesicular structures
with the ER (Hauri and Schweizer, 1992; Krijnse Locker
et al., 1994; Stinchcombe et al., 1995) orcis-Golgi(Tang
et al., 1993; Griffiths et al., 1994), suggesting that the IC
could instead represent subdomains of the ER or cis-
Golgi. Alternatively, the IC could act as the transport
intermediate between these two compartments (e.g.,
Saraste and Svensson, 1991). Understanding the nature
of the IC should help clarify the vesiculartransport steps
between the ER and Golgi complex and thus the roles
of COPI and COPII.
COPIIwas initally identified in yeast as secretory mu-
tant genes (Novick et al., 1980; Kaiser and Schekman,
1990), then characterized by reconstituting vesicle bud-
ding from ER membranes with the proteins Sec23/24p,
Sec13/31p, andSar1p(Barloweetal., 1994).SinceCOPII
vesicles can fuse with Golgi membranes, it was con-
cluded that COPII transports cargo from the ER to the
Golgi complex (Barlowe et al., 1994; Bednarek et al.,
1995). Mammalian homologs of Sec23p and Sec13p
have been localized to the ER and vesicles budding
from it (Orci et al., 1991a; Shaywitz et al., 1995), and
moreover, ts-O45-G, the temperature-sensitive glyco-
protein of vesicular stomatitis virus (VSV), has been
shown to be sorted and concentrated into COPII vesi-
cles at specific exit sites from the ER (Balch et al., 1994;
Bannykh etal., 1996). COPIItherefore has a definite role
in ER export, but it is not clear whether in mammalian
cells the vesicles are destined for the Golgi (Bednarek
et al., 1995) or IC (Aridor et al., 1995; Pepperkok et al.,
The situation is less clear for the COPI coat, which
consists of sevenCOPs (coatomer)and ARF1. Identified
as a coat on vesicles formed from mammalian Golgi
membranes in vitro, it was originally thought to mediate
anterograde transport of cargo through the Golgi stack
(Malhotra etal., 1989), consistentwithits apparentlocal-
ization by immunofluorescence (Duden et al., 1991).
However, EM shows the bulk of coatomer to be on early
exocytic membranes, including the ER (Krijnse Locker
et al., 1994; Orci et al., 1994; Pind et al., 1994), the IC
(Griffiths et al., 1995), and membranes adjacent to the
cis-Golgi (Oprins et al., 1993). This agrees with recent
experiments where COPI was shown to be necessary
for ER-to-Golgi transport of VSV-G in vitro (Peter et al.,
1993; Dascher and Balch, 1994) and in vivo (Pepperkok
Exocytic transport from the endoplasmic reticulum
(ER) to the Golgi complex has been visualized in living
cells using a chimera of the temperature-sensitive
glycoprotein of vesicular stomatitis virus and green
fluorescent protein (ts-G-GFPct). Upon shifting to per-
missive temperature, ts-G-GFPct concentrates into
COPII-positive structures close to the ER, which then
build up to form an intermediate compartment or
transport complex, containing ERGIC-53 and the
KDEL receptor, where COPII is replaced by COPI.
These structures appear heterogenous and move in a
microtubule-dependent manner toward the Golgi
complex. Our results suggest a sequential mode of
COPIIand COPI action and indicate that the transport
complexes are ER-to-Golgi transport intermediates
fromwhich COPImay be involved in recycling material
to the ER.
Proteins destined for secretion, the plasma membrane,
or intracellular organelles are synthesized in the endo-
plasmic reticulum (ER) and transported along the exo-
cytic pathway by coated vesicular carriers. Two types
of coats on vesicles have so far been identified in medi-
ating transport between the ER and the Golgi complex,
COPIand COPII(forreviews see Kreis etal., 1995; Roth-
man and Wieland, 1996; Schekman and Orci, 1996;
Schekmanand Mellman, 1997), buttheir individualroles
in these steps are still debated. The three basic models
forroles of COPs in early exocytic membrane traffic are:
(1) COPII mediates anterograde and COPI retrograde
traffic; (2) both COPs regulate parallel but independent
anterograde pathways, perhaps transporting different
cargo; and (3) COPII directs ER-to-intermediate com-
partment (IC) transport and COPI directs subsequent
IC-to-Golgi transport. These models are not mutually
exclusive, and so far there is no consensus resolving
One reason for the uncertainty of the role of COP-
coated vesicles in ER-to-Golgi traffic is that the mem-
braneboundaries inthis regionareunclear. Pleomorphic
*Present address: Light Microscopy Laboratory, ICRF, 44 Lincoln's
Inn Fields, London WC2A 3PX, United Kingdom.
²To whom correspondence should be addressed.
et al., 1993; Scheeletal., 1997;Pepperkok etal., submit-
ted), suggesting a role for COPI in transport between
the ER or IC and the Golgi complex (see also Orci et al.,
1997).SomeCOPImutants inmammals (Guoetal., 1994)
and yeast (Duden et al., 1994) are defective in ER-to-
Golgitraffic, althoughitis notclearifthis is adirecteffect
(Gaynor and Emr, 1997). Coatomer is also necessary
for the retrieval of KKXX-bearing proteins to the ER
(Letourneur et al., 1994) and interacts directly with this
retrieval signal in vitro (Cosson and Letourneur, 1994;
Lowe and Kreis, 1995). Since some retrieval-defective
coatomer mutants display normal anterograde traffick-
ing, COPI could be involved exclusively in retrograde
transport fromthe Golgito the ER, withany anterograde
blockage resulting indirectly from the lack of recycling
of necessary vesicle components (Pelham, 1994).
ts-045-G is a useful marker for exocytic membrane
traffic in mammalian cells. It misfolds at 39.5?C (nonper-
missive temperature) and fails to exit the ER but is rap-
idly folded upon shifting to 31?C and transported from
the ER throughthe Golgito theplasma membrane (Berg-
mann and Singer, 1983). This transport can be further
synchronized by accumulation of the protein in the IC
at 15?C and in the trans-Golgi network (TGN) at 20?C
(Griffiths and Simons, 1986). To determine whether
COPII and COPI act independently or are functionally
linked in ER-to-Golgi transport, we extended our previ-
ous studies of the anterograde transport of ts-045-G in
vivo by fusing it to the green fluorescent protein (GFP).
This chimera (ts-G-GFPct) is transported identically to
the unmodified ts-045-G and can be used to visualize
exocytic membrane transport. We find that COPII acts
prior to COPI in this process and that COPI remains
associated with transport intermediates during their
movement toward the Golgi complex. Furthermore, we
demonstrate that the IC is not a stable compartment
but rather a transport intermediate arising from the exit
site of the ER and travelling along microtubules to the
expression, coinciding with anti-VSV-G and anti-GFP
shown).Endo H analysis and quantitation of cell surface
fluorescence showed that C-terminal GFP had no effect
on the kinetics of ts-045-G transport from the ER to the
plasma membrane, while the other two constructs were
either not (ts-G-GFPnt) or very inefficiently (ts-G-GFPlm)
transported (data not shown). Furthermore, incubation
of injected cells at 39.5?C, 15?C, or 20?C results in the
accumulation of ts-G-GFPctin the ER, IC, or TGN, as
shown by colocalization with calnexin, ERGIC-53, and
galactosyltransferase, respectively (Figure 1). Interest-
ingly, mostof the ts-G-GFPct, whethervisualized by GFP
fluorescence or with antibodies against the lumenal or
cytoplasmic domain of VSV-G, is concentrated (in con-
trast to calnexin, for example) in the perinuclear part of
the ER near the Golgi complex at 39.5?C (Figure 1a). A
subsequent shiftto 31?C results in a synchronized wave
of ts-G-GFPcttransport to the cell surface. Thus, ts-G-
GFPctis clearly a valid tool for studying transport of ts-
045-G in vivo.
(Figure 1; data not
COPII Acts Closer to the ER Than COPI in Early
Exocytic Transport of ts-G-GFPct
Triple-labeling immunofluorescence was performed to
examine the roles of COPI and COPII in early exocytic
transport of ts-G-GFPct. ts-G-GFPct-expressing cells
were held at 39.5?C or15?C, released to 31?C forvarious
times,and the extentofcolocalizationof ts-G-GFPctwith
COPI- and COPII-positive structures was quantified. At
39.5?C, there are only a few bright spots of ts-G-GFPct
inthe ER (?10/cell), but many moreappearaftershifting
to the permissive temperature (up to 113 ? 56 by 10
min in the average cell, although there was significant
variationinthese numbers amongcells ofdifferentsizes;
Figure 3a). The extent of colocalization of ts-G-GFPctin
these structures with COPI agrees with our previous
results using cells infected with ts-045 VSV (Griffiths et
al., 1995), again indicating that the GFP does not alter
the normal transport of ts-045-G. However, the bulk
of ts-G-GFPcttook 2 min longer to appear in transport
intermediates thants-045-G, perhaps because GFP ren-
ders the protein larger, and so it takes longer to refold
or package into vesicles upon shifting to 31?C.
A typical distribution of ts-G-GFPct6 minafter release
from the ER block shows it in structures positive for
COPI, COPII, or both, with very few having neither coat
(Figure 2). Quantitative analysis revealed that the peak
of COPII only colocalizing with ts-G-GFPctis earlier (?2
min) than the peak of COPI (Figure 3b). COPI only in-
creases inits colocalizationwithcargo as COPIIcolocal-
ization is on the decline and peaks 10 min ormore after
release.Sincethese peaks are dueto thesuddenrelease
ofcargo accumulatedintheER,while normallythe levels
of COPI and COPII vesicles are steady, this strongly
suggests thatCOPIIacts earlierthanCOPIin ts-G-GFPct
transport. The transport intermediates colocalizing with
both COPs are presumably at the IC stage, since triple
labelings show that ?95% of the COPII- and ?98% of
the COPI-coated structures associated with ts-G-GFPct
also label for ERGIC-53, a widely used marker for these
membranes (Schweizeret al., 1990). Similarly, ?90% of
We have generated a fluorescent ts-O45-G to visualize
membrane traffic between the ER and Golgi complex in
vivo and to examine more directly the roles of COPIand
COPII in these steps.
Construction of a ts-O45-G Chimera with GFP That
Is Temperature-Sensitive and Transported
Normally to the Cell Surface
GFP was fused to the cytoplasmic C terminus of ts-045-G
(ts-G-GFPct)and (not shown)the Nterminus (ts-G-GFPnt)
or close to the lumenal transmembrane domain (ts-G-
GFPlm). For better control of expression levels, microin-
jection of the DNA into the nucleus of Vero cells was
preferred to transient transfection, and injected cells
were then incubated at 39.5?C for 8 to 12 hr (sufficient
to obtain a clearly visible signal, while avoiding excess
viral glycoprotein production that might interfere with
normal exocytic membrane traffic). ts-G-GFPct was
strongly fluorescent and remained arrested in the ER at
the nonpermissive temperature even after long times of
Sequential Mode of COP Action
Figure 1. Colocalization of ts-G-GFPct with
Exocytic Membrane Traffic Markers
Vero cells expressing microinjected ts-G-
GFPctDNA for 8 hr at the nonpermissive tem-
peraturewere eitherdirectly fixed inmethanol
(a and b) or kept for a further 3 hr with cyclo-
heximide at 15?C (c and d) or at 20?C (e and
f) before fixation. Cells were immunolabeled
with antibodies against calnexin (b), ERGIC-
53 (d), and galactosyltransferase (f) to visual-
ize the ER, IC, and Golgi complex, respec-
tively.ts-G-GFPctfluorescence is shownin(a),
(c), and (e). Arrowheads indicate colocalizing
structures. Bar ? 10 ?m.
COPI-associated ts-G-GFPctcolocalizes with the KDEL
receptor, another itinerant IC marker (data not shown;
Tang et al., 1993). Consistent with this presumption,
the number and average size of these doubly positive
structures increases after accumulation of ts-G-GFPct
in the IC at 15?C. Upon release of ts-G-GFPctfrom the
IC, the number of COPI-positive structures increases
as the number of doubly positive structures decreases
(Figure 3c). COPIIhas only asmall peak of colocalization
within 2 min. This indicates that not only does COPII
act before COPI, but also that they mainly act before
and after the IC, respectively.
However, it cannot be entirely ruled out that COPI
does not bud directly from the ER in a parallel pathway
to COPII (Bednarek et al., 1995), simply requiring more
time to package cargo than COPII. If only COPII buds
directly from the ER, it should be present on ER mem-
branes to a greater extent than COPI, since COP coats
appear mainly localized at their donor membranes. We
found that this was indeed the case, both at steady-
state, using protein disulphide isomerase (PDI) or ts-G-
GFPntas ER markers at 37?C,andduring early ts-G-GFPct
transport: 66% of COPII-coated ts-G-GFPct-containing
structures were associated with the ER, compared to
only 34% of COPI-coated ones (although note that the
total number of COPII-coated ts-G-GFPct-containing
structures is very low compared to COPI or the doubly
positive structures after the firstfew minutes; see Figure
3a). Furthermore, almost all the doubly positive struc-
tures were associated with the ER (86%), suggesting
that the IC is at or very close to the exit site of the ER.
The 34% overlap of COPIstructures with the ER is most
likely fortuitous, since ER overlap of p200, a protein
associated with TGN vesicles (Narula and Stow, 1995),
was 35%. Taken together, these results strongly impli-
cate COPII in transport out of the ER to the nearby IC
and indicate a subsequent role for COPI.
If COPII and COPI activities really are sequential,
transport intermediates should form but not deliver
cargo to the Golgi complex in the absence of COPI. We
tested this prediction using the drug brefeldin A (BFA),
which is known to rapidly prevent membrane binding of
COPI (Donaldson et al., 1990)and to prevent COPIvesi-
cle formation in vitro (Orci et al., 1991b) but not to affect
COPII(Bednarek etal., 1995; Shaywitz etal., 1995). Cells
expressing ts-G-GFPctin the ER were incubated with
Figure 2. Colocalization of ts-G-GFPctwith COPI and COPII After Release from the ER
Cells expressing ts-G-GFPctfor 8 to 9 hr at 39.5?C were shifted to 31?C for 6 min with cycloheximide before fixing in 3% paraformaldehyde
and indirect immunofluorescence. COPI was visualized with MAb CM1A10 and Cy3-conjugated anti-mouse (b), COPII was detected with
rabbit anti-Sec13p and Cy5-conjugated anti-rabbit (c), and ts-G-GFPctfluorescence was recorded in the fluorescein channel (a). ts-G-GFPct-
containing structures labeling for COPII only are indicated by small arrowheads, for COPI only by large arrowheads, for both coats by large
arrows, and for neither coat by small arrows. (a?)±(c?) show a 3? magnification of (a)±(c). Bar ? 10 ?m.
BFA for 5 min at 39.5?C before shifting to 31?C; ts-G-
GFPctstarted to concentrate into discrete spots on or
near the ER with normal kinetics, but in no instance
did any of these spots move to the perinuclear region.
Immunolabeling verified that most of COPI but none of
COPII had redistributed, and that the ts-G-GFPctstruc-
tures mainly label for COPII (data not shown). Although
underthese conditions the Golgi has started to relocate
to the ER (Lippincott-Schwartz et al., 1989), this cannot
explain the complete inhibition of movement, since
movement of ts-G-GFPctis not completely blocked by
BFA in cells where it was previously accumulated in the
IC at 15?C (data not shown); COPI, once bound to the
IC during this incubation, can probably effect one round
of transport. In this case, starting 2.5 min after the shift
to 31?C,many tubules appear, along whichcargo is seen
tomove inbothdirections (seealsoLippincott-Schwartz
et al., 1990). Only very short tubules are occasionally
seenin the absence of BFA.Furtherevidence that COPII
and COPI act before and after the IC, respectively, is
that transport intermediates form but do not move to-
ward the Golgi in a CHO cell line defective in ?-COP
(ldlF; Guo et al., 1994), even when accumulated in the
IC at 15?C priorto shifting to the permissive temperature
(data not shown).
COPIIs Associated with Cargo Moving Anterogradely
toward the Golgi Complex
Tocharacterize furtherthe site of COPIaction, transport
of ts-G-GFPctwas visualized in living cells. At 39.5?C,
Sequential Mode of COP Action
the dynamic state of the ER is evident, but no vesicular
transport is seen (data not shown). A few minutes after
shifting to 31?C, however, vesicle-like structures start
to appear. Many spend up to a few minutes hovering
around their site of formation (probably being buffeted
by the movements of surrounding membranes and cyto-
plasm), before they suddenly move toward the perinu-
clearregionina rapid, directedsaltatory motionsugges-
tive ofmicrotubule-dependent transport.During the first
10 min after release of ts-G-GFPctfrom the ER, all di-
rected motions were from the ER toward the Golgi com-
plex (i.e., in the anterograde transport direction). Immu-
nofluorescence showed that ?95% of these hovering
structures label forbothCOPIIandERGIC-53 and?98%
for both COPI and ERGIC-53, hence it can be assumed
that virtually all the doubly COP-positive structures also
label for ERGIC-53. Thus, we presume that they are not
single vesicles but either clusters of vesicles at the IC
stage or the IC itself, where extra quality control may
occur before further transport ensues. The relatively
long duration of the hovering step and subsequent
transport, compared to the initial rapid COPII-only-
dependent step, most likely explains why there are so
many more COPI- than COPII-only-associated struc-
tures after the first few minutes (Figure 3a), even though
the two coats act sequentially. Since these structures
eventually move toward the Golgi complex, they will
hereafter be called transport complexes (TCs). Occa-
sionally a TC will stop en route to the Golgi complex,
perhaps undergoing a further round of quality control,
before continuing its anterograde transport (Figure 4).
Within 9 s, this particularTC moves in a straight line for
?17?m (atanaveragespeed of 1.9?m/s). Itthenhovers
again for ?2.5 min before continuing its directed move-
ment for ?12 ?m at an average speed of 1.4 ?m/s. The
average velocity of ten other TCs in this same cell (and
others) was 1.2 ? 0.5 ?m/s. A few TCs chanced to have
been fixed during this process of directed movement
observed in vivo; they immunolabel for COPI but not
COPII (Figure 5), suggesting that COPI is associated
with TCs during their anterograde transport to the Golgi
in vivo. Thus, TCs are associated with both COPs while
hovering near the ER and lose COPII before or during
directed transport to the Golgi complex.
ts-G-GFPctaccumulates in numerous structures simi-
lar to or larger and brighter than those shown in Figure
4 when cells are incubated for 2 to 3 hr at 15?C. Upon
release of the 15?C block, all these TCs move toward
and disappear into the Golgi complex in a very similar
fashion to those formed uponrelease of ts-G-GFPctfrom
the ER (mostlikelyassociatedonly withCOPI; see statis-
tics inFigure3c).Importantly, we neverobserve budding
of vesicles carrying fluorescent cargo from TCs. How-
ever, we cannot completely exclude the possibility that
vesicles that bud are too small or contain too little fluo-
rescent cargo to be detected above the fluorescence
of the cargo remaining in the ER or that although TCs
Figure 3. Quantification of ts-G-GFPct-Positive Structures Colocal-
izing with COPs
ts-G-GFPctwas expressed in Vero cells as described in the legend
for Figure 2. After accumulation of ts-G-GFPctat 39.5?C, cells were
shifted for various periods of time to 31?C, fixed, and immunola-
beled. The unambiguously identifiable vesicular structures con-
taining ts-G-GFPct were scored for colocalization with COPI or
(a) Area graph showing the total numbers of ts-G-GFPct-containing
structures in the flat peripheral regions of the cell at the different
time points. The number of each type of COP-coated vesicles con-
taining ts-G-GFPctatany timepoint canbe calculatedby subtracting
the lower from the upper edge of each region. The total number of
ts-G-GFPctvesicles is thus the sum of all the regions at each time
point, i.e., the uppermost line of the graph. The error bars show the
mean standard deviation of 15 to 19 cells counted for each time
point and illustrate the intercellular heterogeneity with respect to
overall numbers of vesicles (smaller cells have fewer vesicles than
(b) Shows the relative COP coat distributions of the ts-G-GFPctstruc-
tures, expressed as a percentage of the total number of ts-G-GFPct-
positive structures counted in a selected area of the flat region of
a cell (such as the boxed area in Figure 2) for each time point for
the experiment in (a).
(c) Cells were treated and quantified as in (b), except that they were
incubated for 3 hr at 15?C with cycloheximide to accumulate ts-G-
GFPctin the IC before shifting to the permissive temperature. Error
bars represent the mean standard deviation (n ? 15 to 19 cells).
Figure 4. Movement of Vesicular Structures
Containing ts-G-GFPctIn Vivo
Cells expressing ts-G-GFPct in the ER at
39.5?C were shifted to 31?C (with cyclohexi-
mide) on the microscope stage and images
recorded every 3 s. The movement of one TC
from its starting point (arrowheads) is shown
(arrows) at the indicated time points after the
shift.Note the 2 1/2mintime intervalbetween
the left- and right-hand panels, during which
allmovement was undirected hovering.Bar?
are transported as single units to the Golgi complex,
vesicles bud and dock separately from it to the Golgi
once there, since the general fluorescence in this region
is too high to detect individual vesicles. Surprisingly,
TCs of different sizes can aggregate (or fuse) if they
come into contactwhile hovering.Any fusionis probably
incomplete, since subsequent immunolabeling shows
that they have both COPs associated (data not shown).
Again, this implies that the IC is not a stable compart-
ment as such, but a transient transport intermediate,
conveying cargo from the ER toward the Golgi complex
and probably fusing with it or other TCs as one unit.
The structures accumulating ts-G-GFPctat 15?C also
move toward the Golgi complex, albeit much more
slowly, hence their gradual accumulation in the perinu-
Sevento eightminutes aftershifting from15?C to 31?C,
ts-G-GFPct-containing transport intermediates start to
move from the Golgi complex toward the cell periphery,
coincident with the decline in association with COPs
(see Figure 3c). Since this does not occur when cells
are shifted from15?C to 20?C (when transport is blocked
in the TGN), it probably represents Golgi-to-plasma
membrane transport rather than retrograde transport
from the Golgi to the ER. These results further support
a sequential mode of COP action, with COPII acting
at the ER and COPI acting afterward during transport
toward the Golgi complex.
Transport between the ER and Golgi Complex
Occurs along Microtubules
Directed movement of ts-G-GFPctinTCs is almost com-
pletely inhibited in cells with depolymerized microtu-
bules. When cells with ts-G-GFPct accumulated in the
IC had their microtubules depolymerized at 4?C (a tem-
perature at which energy-dependent transport is ar-
rested and the Golgi complex left intact) and were then
shifted to 31?C in the presence of nocodazole, directed
movements of TCs carrying ts-G-GFPctwere abolished.
(In fact, observation of the appearance of vesicular
structures was facilitated under these conditions; they
appearednot tobud fromthe accumulatedandenlarged
IC but presumably from ER tubules.) These TCs were
largerthan normal 15?C ones (1.45 ? 0.34 ?m in appar-
ent diameter instead of 0.94 ? 0.42 ?m) and less accu-
mulated in the Golgi region. Moreover, if nocodazole
was removed, the microtubules rapidly repolymerized,
and TCs moved along them toward the microtubule or-
ganizing center (MTOC), where the Golgi complex is
positioned (Figure 6). Control experiments confirmed
that the Golgi complex was not scattered by the incuba-
tion of cells at 4?C and that no ts-G-GFPcthad reached
the Golgi during that treatment (data not shown), thus
this transport cannot be reclustering of Golgi fragments.
Some fluorescent viral membrane protein did however
appearat the cellsurfaceafter 1 hrat31?C, albeitsignifi-
cantly less than in control cells withintact microtubules
Sequential Mode of COP Action
Figure 5. COPI but not COPII Is Associated with Transport Complexes Traveling toward the Golgi Complex
ts-G-GFPct-expressing cells were shifted from 39.5?C to 4?C for 1 hr (ts-G-GFPctfolds at 4?C, and so more TCs form upon a subsequent shift
to the permissive temperature) before warming to 31?C in normal medium and recording of the movement of fluorescent vesicular structures.
The movement of a vesicular structure carrying fluorescent cargo is indicated with arrowheads on the frames taken in vivo between the first
(12?24??) and the last (13?21??) time points; at 13?24??, the cell was fixed and immunolabeled for COPI (??-COP) and COPII (Sec13). Note that
COPI but not COPII is associated with the moving TC. The distribution of ts-G-GFPct, ??-COP, and Sec13p is shown in the whole cell in the
lower panel at lower magnification. Bar ? 10 ?m.
(data not shown; see also Rogalski et al., 1984). Depo-
lymerization of the actin cytoskeleton with cytochalasin
D had no obvious inhibitory effect ondirected transport
between the ER and Golgi complex (data not shown).
Finally, COPI was stabilized on membranes by treat-
ment of cells with AlF4. AlF4did not significantly affect
velocity or directionality of movement of TCs. However,
fewer of the TCs formed from the ER moved toward the
Golgi complex, and they appeared not to fuse with it
(data not shown), consistent with a failure to uncoat.
Such structures indeed appeared to associate with
COPI more rapidly than normal, and both coats re-
mained on them for longer (?10 min), consistent with a
partial inhibition at a post-ER stage of exocytic trans-
port. Whilewecannotexclude thatAlF4is notalso affect-
ing other proteins than COPI, it does not inhibit move-
ment of TCs accumulated at 15?C toward the Golgi
region. Thus, COPI may be necessary for recycling of
COPII components from TCs to the ER to allow further
rounds of transport. Taken together, these results sug-
gestthatCOPIonTCs does not interferewiththeirmove-
ment toward the Golgi (only with delivery of cargo to
this compartment) and thus probably does not inhibit
microtubule-based motor function (e.g., by masking a
motor receptor on the TC membrane). Our attempts
to find coatomer±microtubule interaction in vitro have
failed, suggesting thatcoatomerprobably does not bind
directly to microtubules (data not shown).
We have tagged ts-045-G withGFP to visualize exocytic
membrane traffic. The chimeric ts-G-GFPctbehaves like
the normal viral glycoprotein and is thus a powerful
marker for monitoring the pathway of secretory cargo
in living (noninfected) mammalian cells. We show here
that newly synthesized ts-G-GFPctcongregates in COPII-
coated structures at the ER membrane, which subse-
quently build up to form larger TCs with which COPI
then associates. COPI, but not COPII, remains associ-
ated with these TCs as they move as entire units along
microtubules toward the Golgi complex to deliver their
cargo. No budding from TCs of vesicles containing fluo-
rescent cargo could be detected. Instead, these TCs
may aggregate oreventually fuse witheach otherbefore
arriving at the Golgi complex.
Use of ts-G-GFPctrevealed that ER-to-Golgi transport
is probably mediated by multiple mechanisms. At the
nonpermissive temperature, a significant fraction of ts-
G-GFPctconcentrates in the ER in the juxtanuclear re-
gion of the Golgi complex. This may be due to diffusion
Figure 6. Microtubules Facilitate Early Exo-
cytic Membrane Transport
Cells expressing ts-G-GFPctthat had been
accumulated in the IC by incubation at 15?C
were incubated for 1 hr at 4?C to depolymer-
ize their microtubules without causing the
Golgi complex to disperse or the ts-G-GFPct
to leavethe IC.They werethenshifted to31?C
on the microscope stage, and images were
recorded every 3 s. Within 6 min, TCs began
tomove inadirected saltatorymannertoward
a juxtanuclear region from which the newly
polymerized microtubules emanated (visual-
ized afterward by imunofluorescence with
MAb 1A2 [d]). The first (a) and last (b) frames
of the sequence at the indicated times after
the shift to 31?C are shown. (c) The last 20
frames (aperiod of1 min)were superimposed
to reveal the path taken by TCs (arrowheads)
toward theMTOC. The microtubulepathdoes
notexactly matchthat takenbythe TCs, since
the microtubules probably alsomoved during
the last minute of the sequence. Bar ? 10 ?m.
within the ER membrane, but it is unclear what mecha-
nisms retain it there. We also observed concentrated
spots of ts-G-GFPctbeing thrown along ER tubules as
they sway,suggesting that someformoffacilitated diffu-
sion or active transport occurs. Finally, ts-G-GFPctap-
pears to be buddedfromthe ER by COPII,withindividual
vesicles or larger TCs moving it toward the Golgi com-
plex in subsequent association with COPI.
Within the first few minutes of formation, TCs show
hovering movements resembling Brownian motion. This
is followed by rapid directed movements along microtu-
bules (average and maximal velocities of ?1.2 and 2
?m/s, respectively), initially exclusively directed toward
their minus ends (and the Golgi), consistent with the
involvement ofcytoplasmic dynein(Holzbaurand Vallee,
1994). While the extension of BFA-induced retrograde
tubules from the Golgi complex to the ER has been
demonstrated to be microtubule-dependent (Lippin-
cott-Schwartz et al., 1990), evidence for microtubules
in early anterograde traffic has beencontroversial. Endo
H analyses indicate that microtubule depolymerization
has only a minor effect on ER-to-Golgi transport, while
peripheralstructures accumulatedat15?C do notcluster
into the Golgi region upon warming in the absence of
microtubules, suggesting that Golgi elements scatter
toward these peripheral structures (Saraste and Svens-
son, 1991;Cole etal., 1996).Theseand ourinvivo results
strongly suggest that microtubules do facilitate trans-
port toward the Golgi complex.
By statistical analyses of the extent of colocalization
of ts-G-GFPctwith COPII and COPI at different time
points after release from 39.5?C, we show (1) that COPII
is the first to associate with cargo in the ER; (2) that
during the next few minutes, both COPs are associated
with a structure (TC) assumed to be the IC stage; and
(3) that subsequently COPI remains associated during
anterograde movement of TCs toward the Golgi com-
plex. We have also shown that COPII is closer to the
ER than COPI, both during ts-G-GFPcttransport and
at steady-state, consistent with EM data showing the
steady-state localization of COPII to ER export sites,
the transitional elements, and IC (Shaywitz et al., 1995;
Bannykh et al., 1996; Tang et al., 1997), and of COPI to
the IC and the Golgi complex (Oprins et al., 1993; Grif-
fiths et al., 1995). While colocalization is not definite
proof of site of function, our results suggest that COPI
acts temporally after COPII. This is supported by the
observation that ts-G-GFPctcanaccumulate into COPII-
coated vesicular structures but not move to the Golgi
complex upon shifting to the permissive temperature in
the presence of BFA, conditions under which COPI but
not COPII is affected. Likewise, in ldlF cells with defec-
tive ?-COP (Guo et al., 1994), TCs form, but their move-
ment toward the Golgi is inhibited. A sequential model
ofCOPIIthenCOPIactionhas previously beenproposed
forts-045-G transport, where COPIwas initially thought
to be involved in anterograde transport, after COPII but
before the IC stage (Aridor et al., 1995), and more re-
cently in retrograde transport from the IC to the ER
(Bannykh et al., 1996). Differential effects of microin-
jected GTP?S on early exocytic membrane transport
in vivo (GTP?S or anti-EAGE do not inhibit ER-to-IC
transport but block IC-to-Golgi transport) (Pepperkok
et al., submitted) provide additional evidence for the
Sequential Mode of COP Action
sequential action of COPII and COPI. Furthermore, mi-
croinjection of a recombinant GTP-restricted mutant
of SAR1 (Sar1a[H79G]p) (Aridor et al., 1995), appears
to accumulate COPIIonto membranes and cause COPI
to become predominantly cytosolic, as if COPII fails to
generate the TC membranes to which COPI can bind
(R. P. and T. E. K., unpublished data).
What are the TCs that move from near the ER toward
the Golgi complex? We believe them to be clusters of
vesicles (perhaps the vesicular-tubular clusters) (Balch
et al., 1994), since it is assumed that the COPs do not
both coat one vesicle simultaneously (Barlowe et al.,
1994; Aridor et al., 1995; Bednarek et al., 1995), and TCs
often label for both. Furthermore, most TCs appeared
largerthan expected for a single vesicle (although fluo-
rescence does not reflect true size). COPIIlabeling may
be due to vesicles that have docked or are forming
from the closely apposed ER exit site, while COPI is
presumably on budding structures (e.g., Oprins et al.,
1993; Pind et al., 1994; Griffiths et al., 1995). Since the
IC formed at 15?C resembles TCs formed after release
fromthe 39.5?C ER block, wethink that itis an expanded
versionofthe regulartransport intermediates,especially
as virtually all TCs label for the KDEL receptor and ER-
GIC-53/p58. Importantly, we never observed vesicular
structures containing ts-G-GFPctforming from TCs. In-
stead,allthe fluorescent TCs moved as entire structures
toward the Golgi complex. Thus, we propose that the
IC, rather than being a true individual compartment, is
a transient TC arising fromthe aggregationand eventual
fusion of ER-derived COPII-coated vesicles and with
which COPI later associates to perform a subsequent
functionessential fordelivery of cargo to the Golgi com-
plex (Pepperkok et al., 1993).
Based on ourresults, we can propose two models for
the role of COPs in early exocytic membrane traffic. In
both models, COPIIforms vesicles at defined ER export
sites (Barlowe et al., 1994; Bannykh et al., 1996), which
stay close to their budding sites. Such vesicles may
cluster and fuse to form a TC, the functional equivalent
of the IC. While the TCs hovernear their sites of forma-
tion, COPII is probably gradually exchanged for COPI,
allowing the TC to interact with a microtubule for trans-
port toward the Golgi complex (it is unclear whether
COPI also plays a role in regulating TC±microtubule in-
teractions). In the first model, COPI would be directly
involved in anterograde transport, namely in budding of
vesicles delivering cargo from TCs (which eventually
may build up the cis-Golgi network [CGN] by lateral
fusion)to the cis-Golgicisternae afterreaching the Golgi
complex. In the second model, COPI is associated with
anterogradely moving TCs but retrieves material to the
ER, consistent with conclusions from yeast genetics
(Letourneuretal., 1994; Cossonet al., 1996; Gaynorand
Emr, 1997). Some COPI-coated vesicles are probably
also involvedinretrievalof materialfromwithinthe Golgi
complex and TGN, as COPIhas been localized to these
compartments (Oprins et al., 1993; Orci et al., 1993;
Griffiths et al., 1995; Orci et al., 1997), and some KKXX-
bearing proteins have Golgimodifications (e.g.,J ackson
et al., 1993). In fact, preliminary data indicate that only
50% of the COPI vesicles in normal cells contain the
KDEL receptor, and in ts-G-GFPct-expressing cells,
while virtually all the COPI-coated vesicular ts-G-GFPct
structures contain the KDEL receptor, about half the
total KDEL receptor±positive COPI-coated vesicles do
not transport ts-G-GFPct(S. J . S. and T. E. K., unpub-
lished data; see also Griffiths et al., 1995). In this model,
COPI is indirectly necessary foranterograde membrane
traffic betweenthe ER andthe Golgicomplex;itretrieves
earlier acting factors from TCs (e.g., v-SNAREs, ERGIC-
53/p58, chaperones, COPII-associated membrane pro-
teins), which, if not removed, would prevent transfer of
cargo to the Golgi complex. This retrograde model of
COPIaction betterexplains our results, inparticularthat
no budding of fluorescent cargo is observed from the
TCs (probably most ts-G-GFPcthas been controlled for
quality in the ER) (Hammond and Helenius, 1995).
Whether or not this second model lends support to a
maturation model for Golgi biogenesis (e.g., Lippincott-
Schwartz, 1993) depends on whether TCs fuse to build
up the CGN, which then matures into Golgi cisternae
as a consequence of retrieval, or from where coated
vesicles may transport cargo anterogradely to the me-
dial Golgi (Orci et al., 1997).
A remarkable symmetry between early secretory and
early endocytic membrane traffic becomes apparent.
Like early endosomes (for reviews, see Gruenberg and
Maxfield, 1995; Mellman, 1996), early exocytic TCs build
up by the clustering of coated vesicles (clathrin/AP2,
COPII), appear to be dynamic transport intermediates
capable oflateralfusion,and maymature during ensuing
movement along microtubules toward their next desti-
nation nearer the MTOC. We have recently shown that
proteins immunologically related to COPI are essential
for endosome function (Whitney et al., 1995; see also
Aniento et al., 1996), so it is tempting to speculate that
their role in the endocytic pathway may be analogous
to the one we propose here forearly exocytic transport,
namely retrieval of proteins to promote maturation of
TCs. Simultaneous covisualization of cargo and COPI
in living cells will help to clarify the precise role of COPI
function in membrane traffic.
Cell Culture and Microinjection
Vero cells (ATCC CCL81) were cultured in GIBCO BRL solutions as
described (Kreis and Lodish, 1986) and kept at 39.5?C after ts-G-
GFPctDNA uptake. 15?C, 20?C, and 31?C incubations were in low
carbonate medium with 100 ?g/ml cycloheximide from Sigma. AlF4
(30mMNaF plus 50?MAlCl3),nocodazole(10?M), and cytochalasin
D (1 ?Mfor 1 hr) were also from Sigma.BFA (BoehringerMannheim)
was used at 10 ?M.
DNA was microinjected into nuclei using an automated microin-
jection system (Zeiss AIS) as described (Pepperkok et al., 1993).
Microinjected cells were incubated at 39.5?C for 8 to 12 hr prior to
visualization and analysis of ts-G-GFPct.
Antibodies and Immunofluorescence
The following antibodies were used: MAb against the lumenal do-
main of VSV-G (anti-VG; from Kai Simons, EMBL, Heidelberg, Ger-
many); MAb P5D4 and anti-P4 against the C terminus of VSV-G
(Kreis and Lodish, 1986); MAb CM1A10 (anti-??-COP; Palmer et
al., 1993); anti-mammalian Sec13p (Tang et al., 1997); anti-calnexin
(Hammond and Helenius, 1994), MAb G1/93 (anti-ERGIC-53;
Schweizer et al., 1988); MAb 1D3 (anti-PDI; from Stephen Fuller,
EMBL, Heidelberg, Germany); anti-KDEL-receptor (Tang, 1993);
MAb against galactosyltransferase (Kawano et al., 1994); MAb 1A2
against tubulin (Kreis, 1987); and MAb AD7 against p200 (Narula et
al., 1992). Antibodies against mouse and rabbit IgG (Cappel) were
labeled with Cy3 and Cy5 (Amersham) according to the manufactur-
For the labeling of viral glycoprotein at the cell surface and for
use of anti-Sec13p, cells were fixed for 20 min at RT with 3% para-
formaldehyde in a 1:1 mixture of PBS and culture medium; in all
other experiments, cells were fixed with methanol at ?20?C for 4
min. Immunofluorescence was then carried out as described (Kreis
and Lodish, 1986).
Aridor, M., Bannykh, S.I., Rowe, T., and Balch, W.E. (1995). Sequen-
tialcoupling between COPIIand COPIvesicle coats in endoplasmic
reticulum to Golgi transport. J . Cell Biol. 131, 875±893.
Balch, W.E., McCaffery, J .M., Plutner,H., and Farquhar,M.G. (1994).
Vesicular stomatitis virus glycoprotein is sorted and concentrated
during export from the endoplasmic reticulum. Cell 76, 841±852.
Bannykh, S.I., Rowe, T., and Balch, W.E. (1996). The organization
ofendoplasmic reticulumexport complexes. J .CellBiol. 135, 19±35.
Barlowe, C., Orci, L., Yeung, T., Hosobuchi, M., Hamamoto, S., Sa-
lama, N., Rexach, M.F., Ravazzola, M., Amherdt, M., and Schekman,
R.(1994). COPII: a membranecoat formed bysec proteins that drive
vesicle budding from the endoplasmic reticulum. Cell 77, 895±907.
Bednarek, S.Y., Ravazzola, M., Hosobuchi, M., Amherdt, M., Perre-
let, A., Schekman, R., and Orci, L. (1995). COPI- and COPII-coated
vesicles bud directly from the endoplasmic reticulum in yeast. Cell
Bergmann, J .E., and Singer, S.J . (1983). Immunoelectron micro-
scopic studies of the intracellulartransport of the membrane glyco-
protein (G) of vesicular stomatitis virus in infected Chinese hamster
ovary cells. J . Cell Biol. 97, 1777±1787.
Brewer, C.B., and Roth, M.G. (1991). A single amino acid change in
the cytoplasmic domain alters the polarized delivery of influenza
virus hemagglutinin. J . Cell Biol. 114, 413±421.
Cole, N.B., Sciaky, N., Marotta, A., Song, J ., and Lippincott-
Schwartz, J . (1996). Golgi dispersal during microtubule disruption:
regeneration of Golgi stacks at peripheral endoplasmic reticulum
exit sites. Mol. Biol. Cell 7, 631±650.
Cormack, B.P., Valdivia, R.H., and Falkow, S. (1996). FACS-opti-
mized mutants of the green fluorescent protein (GFP). Gene 173,
Cosson, P., and Letourneur, F. (1994). Coatomerinteraction with di-
lysine endoplasmic reticulum retention motifs. Science 263, 1629±
Cosson,P., De Âmollie Á re, C.,Hennecke, S.,Duden, R., and Letourneur,
F. (1996). ?- and ?-COP, two coatomer subunits homologous to
clathrin-associated proteins, are involved in ER retrieval. EMBO J .
Dascher, C., and Balch, W.E. (1994). Dominant inhibitory mutants
of ARF1 block endoplasmic reticulum to Golgi transport and trigger
disassembly of the Golgi apparatus. J . Biol. Chem. 269, 1437±1448.
Donaldson, J .G., Lippincott-Schwartz, J ., Bloom, G.S., Kreis, T.E.,
and Klausner, R.D. (1990). Dissociationof a110-kD peripheral mem-
brane protein from the Golgi apparatus is an early event in brefeldin
A action. J . Cell Biol. 111, 2295±2306.
Duden, R., Griffiths, G., Frank, R., Argos, P., and Kreis, T.E. (1991).
?-COP, a 110 kd protein associated with non-clathrin-coated vesi-
cles and the Golgi complex, shows homology to ?-adaptin. Cell 64,
Duden, R., Hosobuchi, M., Hamamoto, S., Winey, M., Byers, B., and
Schekman, R. (1994). Yeast ?- and ??-coat proteins (COP). Two
coatomer subunits essential for endoplasmic reticulum-to-Golgi
protein traffic. J . Biol. Chem. 269, 24486±24495.
Gallione, C.J .,and Rose, J .K.(1985).A single aminoacid substitution
in ahydrophobic domain causes temperature-sensitive cell-surface
transport of a mutant viral glycoprotein. J . Virol. 54, 374±382.
Gaynor, E.C., and Emr, S.D. (1997). COPI-independent anterograde
transport: cargo-selective ER to Golgi protein transport in yeast
COPI mutants. J . Cell Biol. 136, 789±802.
Griffiths, G., and Simons, K. (1986). The trans Golgi network: sorting
at the exit site of the Golgi complex. Science 234, 438±443.
Griffiths, G., Ericsson, M., Krijnse Locker, J ., Nilsson, T., Goud, B.,
Soling, H.-D., Tang, B.L., Wong, S.H., and Hong, W. (1994). Localiza-
tion of the Lys, Asp, Glu, Leu tetrapeptide receptor to the Golgi
complex and the intermediate compartment in mammalian cells. J .
Cell Biol. 127, 1557±1574.
Griffiths, G., Pepperkok, R., Locker, J .K., and Kreis, T.E. (1995).
Immunocytochemicallocalization of ?-COP to the ER-Golgi bound-
ary and the TGN. J . Cell Sci. 108, 2839±2856.
Gruenberg, J ., and Maxfield, F.R. (1995).Membrane transport in the
endocytic pathway. Curr. Opin. Cell Biol. 7, 553±563.
A bright, red-shifted GFP (pCGFP2, V68L, S72A; Cormack et al.,
1996) was kindly provided by D. Shima and ts-045-G DNA (R-ts-R;
Gallione and Rose, 1985) by C. Machamer.
ts-G-GFPctconstruction: the ts-G-GFPctconstruct contains the
SacII fragment of pCGFP2 at the C terminus of ts-045-G, where a
SacII site (underlined in the primer below) created by PCR replaced
the stop codon. Primers were 5?-CAT CAG GTG TCT GGT TCG AGA
TGG C-3? and 3?-CTA CTT GGC TGA ACC TTT CGG GCG CCA TCC
TAG GAA-5?. PCR products were verified bysequencing (Pharmacia
T7 kit) and expressed in pCB6 (Brewer and Roth, 1991).
Fluorescence Microscopy and Quantitation
Microscopy was done with a Zeiss inverted fluorescence micro-
scope (Axiovert TV135)with fluorescein, Cy3, and Cy5 filters.Images
were recorded and quantified as described (Pepperkok et al., 1993;
Griffiths et al., 1995), except that the overlays were only 3 pixels in
diameter. Up to 120 ts-G-GFPctstructures were counted per cell,
and the meanpercentages (Figures 3b and 3c) orabsolute numbers
(Figure 3a) of vesicles counted per cell (n ? 15 to 19) were plotted
using Cricket Graph 1.3 software. ER overlap was quantified in the
same way, using ts-G-GFPctor PDI as the ER marker.
For in vivo analyses, images were taken for 0.2 s every 3 to 5 s,
sometimes with attenuating filters to minimize UV exposure, al-
though the cells displayed no obvious symptoms of overexposure.
The temperature of the microscope stage was controlled as de-
scribed (Pepperkok et al., 1993).
Endo H Digestion
For endo H experiments, cells were either infected with ts-O45 VSV
(control; Kreis and Lodish, 1986) or transfected using the calcium
phosphate method and incubated at 39.5?C following precipitate
removal. TheCMV promoterwas furtherinduced intransfected cells
using 3 mM sodium butyrate (Sigma) 24 to 38 hr after transfection.
During the last 5 to 6hrof induction,transfected cells were depleted
of methioninefor20mininmet-freemedium at39.5?C thenmetaboli-
cally labeled for5hrwith 80?Ci[35S]methionine/cysteine(inaratio of
10:4.3) at 39.5?C. Lysis, immunoprecipitation, and endo H digestion
were performed and analyzed as described (Kreis and Lodish, 1986;
Scheel et al., 1997).
We would like to thank Drs. B. Cormack, C. Machamer, K. Matter,
and D. Shima for providing DNAs; M. Krieger for the ldlF cells; D.
Rifat foroligos; B.Burke, S.Fuller, H.-P.Hauri, A. Helenius, W. Hong,
J . Rothman, K. Simons, T. Suganuma, and B. L. Tang for antibodies;
and J . Lippincott-Schwartz for sharing unpublished results. Special
thanks go to the Kreis lab for stimulating discussions, in particular
to F. Perez and the readers of the manuscript, H. Goodson and A.
Whitney, and F. Abbet for technical assistance. This research was
supported by grants to T. E.K. from the Fonds Nationale Suisse, the
Canton de Gene Á ve, and the International Human Frontier Science
Program. R. P. was supported by an EMBO long-term fellowship.
Received J une 25, 1997; revised August 13, 1997.
Aniento, F., Gu, F., Parton, R.G., and Gruenberg, J . (1996).An endo-
somal ?-COP is involvedinthe pH-dependent formationoftransport
vesicles destined for late endosomes. J . Cell Biol. 133, 29±41.
Sequential Mode of COP Action
Guo, Q., Vasile, E., and Krieger, M. (1994).Disruptions inGolgi struc-
ture and membrane traffic in a conditional lethal mammalian cell
mutant are corrected by ?-COP. J . Cell Biol. 125, 1213±1224.
Hammond, C., and Helenius, A. (1994). Quality control in the secre-
tory pathway: retention of a misfolded viral membrane glycoprotein
involves cycling between the ER, intermediate compartment, and
Golgi apparatus. J . Cell Biol. 126, 41±52.
Hammond, C., and Helenius, A. (1995). Quality control in the secre-
tory pathway. Curr. Opin. Cell Biol. 7, 523±529.
Hauri, H.P., and Schweizer, A. (1992). The endoplasmic reticulum-
Golgi intermediate compartment. Curr. Opin. Cell Biol. 4, 600±608.
Holzbaur, E.L.F., and Vallee, R.B. (1994).DYNEINS: molecular struc-
ture and cellular function. Annu. Rev. Cell Biol. 10, 339±372.
J ackson, M.R., Nilsson, T., and Peterson, P.A. (1993). Retrieval of
transmembrane proteins to the endoplasmic reticulum. J . Cell Biol.
Kaiser, C.A., and Schekman, R. (1990). Distinct sets of SEC genes
govern transport vesicle formation and fusion early in the secretory
pathway. Cell 61, 723±733.
Kawano, J ., Ide, S., Oinuma, T., and Suganuma, T. (1994). A protein-
specific monoclonalantibody to rat liver?-1-?4 galactosyltransfer-
ase and its applicationto immunohistochemistry. J . Histochem. Cy-
tochem. 42, 363±369.
Kreis, T.E.(1987). Microtubules containing detyrosinated tubulin are
less dynamic. EMBO J . 6, 2597±2606.
Kreis, T.E., and Lodish, H.F. (1986). Oligomerization is essential for
transportof vesicularstomatitisviral glycoproteintothe cellsurface.
Cell 46, 929±937.
Kreis, T.E., Lowe, M., and Pepperkok, R. (1995). COPs regulating
membrane traffic. Annu. Rev. Cell Dev. Biol. 11, 677±706.
Krijnse Locker, J ., Ericsson, M., Rottier, P.J ., and Griffiths, G.(1994).
Characterization of the budding compartment of mouse hepatitis
virus: evidence that transport from the RER to the Golgi complex
requires only one vesicular transport step. J . Cell Biol. 124, 55±70.
Letourneur, F., Gaynor, E.C., Hennecke, S., Demolliere, C., Duden,
R., Emr, S.D., Riezman, H., and Cosson, P. (1994). Coatomer is
essentialforretrieval ofdilysine-tagged proteins to the endoplasmic
reticulum. Cell 79, 1199±1207.
Lippincott-Schwartz, J . (1993). Bidirectional membrane traffic be-
tween the endoplasmic reticulum and Golgi apparatus. Trends Cell
Biol. 3, 81±88.
Lippincott-Schwartz, J ., Yuan, L.C., Bonifacino, J .S., and Klausner,
R.D. (1989). Rapid redistribution of Golgi proteins into the ER in
cells treated with brefeldin A: evidence for membrane cycling from
the Golgi to ER. Cell 56, 801±813.
Lippincott-Schwartz, J ., Donaldson, J .G., Schweizer, A., Berger,
E.G., Hauri, H.P.,Yuan, L.C., and Klausner, R.D. (1990).Microtubule-
dependent retrograde transport of proteins into the ER in the pres-
ence of brefeldin A suggests an ER recycling pathway. Cell 60,
Lotti,V., Torrisi, M., Pascale,M., and Bonatti,S. (1992).Immunocyto-
chemical analysis of the transfer of vesicular stomatitis virus glyco-
protein from the intermediate compartment to the Golgi complex.
J . Cell Biol. 118, 43±50.
Lowe, M., and Kreis, T.E. (1995). In vitro assembly and disassembly
of coatomer. J . Biol. Chem. 270, 31364±31371.
Malhotra, V., Serafini, T., Orci, L., Shepherd, J .C., and Rothman, J .E.
(1989). Purification of a novel class of coated vesicles mediating
biosynthetic protein transport through the Golgi stack. Cell 58,
Mellman, I. (1996). Endocytosis and molecular sorting. Annu. Rev.
Cell Dev. Biol. 12, 575±625.
Narula, N., and Stow, J .L. (1995). Distinct coated vesicles labeled
forp200budfromtrans-Golgi networkmembranes.Proc. Natl.Acad.
Sci. USA 92, 2874±2878.
Narula, N., McMorrow, I., Plopper, G., Doherty, J ., Matlin, K.S.,
Burke,B., and Stow,J .L. (1992). Identificationofa 200-kD, brefeldin-
sensitive protein on Golgi membrane. J . Cell Biol. 117, 27±38.
Novick, P., Field, C., and Schekman, R. (1980). Identification of 23
complementation groups required for post-translational events in
the yeast secretory pathway. Cell 25, 205±215.
Oprins, A., Duden, R., Kreis, T.E., Geuze, H.J ., and Slot, J .W. (1993).
?-COP localizes mainly to the cis-Golgi side in exocrine pancreas.
J . Cell Biol. 121, 49±59.
Orci, L., Ravazzola, M., Meda, P., Holcomb, C., Moore, H.P., Hicke,
L., and Schekman, R. (1991a). Mammalian Sec23p homolog is re-
stricted to the endoplasmic reticulum transitional cytoplasm. Proc.
Natl. Acad. Sci. USA 88, 8611±8615.
Orci, L., Tagaya, M., Amherdt, M., Perrelet, A., Donaldson, J .G.,
Lippincott-Schwartz, J ., Klausner, R.D., and Rothman, J .E. (1991b).
Brefeldin A, a drug that blocks secretion, prevents the assembly of
non-clathrin-coated buds on Golgi cisternae. Cell 64, 1183±1195.
Orci, L., Palmer, D.J ., Ravazzola, M., Perrelet, A., Amherdt, M., and
Rothman, J .E. (1993). Budding from Golgi membranes requires the
coatomer complex of non-clathrin coat proteins. Nature 362,
Orci, L., Perrelet, A., Ravazzola, M., Amherdt, M., Rothman, J .E.,
and Schekman, R. (1994). Coatomer-rich endoplasmic reticulum.
Proc. Natl. Acad. Sci. USA 91, 11924±11928.
Orci, L., Stamnes, M., Ravazzola, M., Amherdt, M., Perrelet, A., So Èl-
lner, T.H., and Rothman, J .E. (1997). Bidirectional transport by dis-
tinct populations of COPI-coated vesicles. Cell 90, 335±349.
Palmer, D.J ., Helms, J .B., Beckers, C.J ., Orci, L., and Rothman, J .E.
(1993). Binding of coatomer to Golgi membranes requires ADP-
ribosylation factor. J . Biol. Chem. 268, 12083±12089.
Pelham, H.R. (1994). About turn for the COPs? Cell 79, 1125±1127.
Pepperkok, R., Scheel, J ., Horstmann, H., Hauri, H.P., Griffiths, G.,
and Kreis, T.E.(1993). ?-COP is essentialfor biosynthetic membrane
transport from the endoplasmic reticulum to the Golgi complex in
vivo. Cell 74, 71±82.
Peter, F., Plutner, H., Zhu, H., Kreis, T.E., and Balch, W.E. (1993).
?-COP is essential for transport of protein from the endoplasmic
reticulum to the Golgi in vitro. J . Cell Biol. 122, 1155±1167.
Pind, S.N.,Nuoffer, C., McCaffery, J .M., Plutner, H., Davidson, H.W.,
Farquhar, M.G., and Balch, W.E. (1994). Rab1 and Ca2?are required
for the fusion of carrier vesicles mediating endoplasmic reticulum
to Golgi transport. J . Cell Biol. 125, 239±252.
Rogalski, A.A., Bergmann, J .E., and Singer, S.J . (1984). Effect of
microtubule assembly status on the intracellular processing and
surface expression of an integral protein of the plasma membrane.
J . Cell Biol. 99, 1101±1109.
Rothman, J .E., and Wieland, F.T. (1996). Proteinsorting bytransport
vesicles. Science 272, 227±234.
Saraste, J ., and Kuismanen, E. (1984).Pre- and post-Golgi vacuoles
operate inthe transport ofSemliki Forest virus membrane glycopro-
teins to the cell surface. Cell 38, 535±549.
Saraste, J ., and Svensson, K. (1991).Distribution ofthe intermediate
elements operating inER toGolgitransport.J .CellSci.100, 415±430.
Scheel, J ., Pepperkok, R., Lowe, M., Griffiths, G., and Kreis, T.E.
(1997). Dissociation of coatomer from membranes is required for
brefeldin A-induced transfer of Golgi enzymes to the endoplasmic
reticulum. J . Cell Biol. 137, 319±333.
Schekman, R., and Orci, L. (1996). Coat proteins and vesicle bud-
ding. Science 271, 1526±1533.
Schekman, R., and Mellman, I. (1997). Does COPI go both ways?
Cell 90, 197±200.
Schweizer, A., Fransen, J .A.M., Bachi, T., Ginsel, L.,and Hauri, H.-P.
(1988). Identification, by a monoclonal antibody, of a 53-kD protein
associated with a tubulo-vesicular compartment at the cis-side of
the Golgi complex. J . Cell Biol. 107, 1643±1653.
Schweizer, A., Fransen,J .A.M., Matter, K., Kreis, T.E.,Ginsel, L., and
Hauri, H.-P. (1990). Identification of an intermediate compartment
involved in protein transport from the endoplasmic reticulum to the
Golgi apparatus. Eur. J . Cell Biol. 53, 185±196.
Schweizer, A., Matter, K., Ketcham, C.M., and Hauri, H.-P. (1991).
The isolatedER-Golgiintermediate compartmentexhibits properties
that are different from ER and cis-Golgi. J . Cell Biol. 113, 45±54.
Shaywitz, D.A., Orci, L., Ravazzola, M., Swaroop, A., and Kaiser,
C.A. (1995). Human SEC13p functions in yeast and is located on
transport vesicles budding from the endoplasmic reticulum. J . Cell
Biol. 128, 769±777.
Stinchcombe, J .C., Nomoto, H., Cutler, D.F., and Hopkins, C.R.
(1995). Anterograde and retrograde traffic between the rough endo-
plasmic reticulum and the Golgi complex. J . Cell Biol. 131, 1387±
Tang, B.L., Wong, S.H., Li Qi, X., Low, S.H., and Hong, W. (1993).
Molecular cloning, characterization, subcellularlocalization and dy-
namics of p23, the mammalian KDEL receptor. J . Cell Biol. 120,
Tang, B.L., Peter, F., Krijnse Locker, J ., Low, S.H., Griffiths, G.,
and Hong, W. (1997). The mammalian homolog of yeast Sec13p is
enriched in the intermediate compartment and is essential for pro-
tein transport from the endoplasmic reticulum to the Golgi appara-
tus. Mol. Cell. Biol. 17, 256±266.
Whitney, J .A., Gomez, M., Sheff, D., Kreis, T.E., and Mellman, I.
(1995). Cytoplasmic coat proteins involved in endosome function.
Cell 83, 703±713.