Exomer: A coat complex for transport of select membrane proteins from the trans-Golgi network to the plasma membrane in yeast

Abstract

A yeast plasma membrane protein, Chs3p, transits to the mother-bud neck from a reservoir comprising the trans-Golgi network (TGN) and endosomal system. Two TGN/endosomal peripheral proteins, Chs5p and Chs6p, and three Chs6p paralogues form a complex that is required for the TGN to cell surface transport of Chs3p. The role of these peripheral proteins has not been clear, and we now provide evidence that they create a coat complex required for the capture of membrane proteins en route to the cell surface. Sec7p, a Golgi protein required for general membrane traffic and functioning as a nucleotide exchange factor for the guanosine triphosphate (GTP)-binding protein Arf1p, is required to recruit Chs5p to the TGN surface in vivo. Recombinant forms of Chs5p, Chs6p, and the Chs6p paralogues expressed in baculovirus form a complex of approximately 1 MD that binds synthetic liposomes in a reaction requiring acidic phospholipids, Arf1p, and the nonhydrolyzable GTPgammaS. The complex remains bound to liposomes centrifuged on a sucrose density gradient. Thin section electron microscopy reveals a spiky coat structure on liposomes incubated with the full complex, Arf1p, and GTPgammaS. We termed the novel coat exomer for its role in exocytosis from the TGN to the cell surface. Unlike other coats (e.g., coat protein complex I, II, and clathrin/adaptor protein complex), the exomer does not form buds or vesicles on liposomes.

Full text

THE JOURNAL OF CELL BIOLOGY
JCB: ARTICLE
© The Rockefeller University Press $8.00
The Journal of Cell Biology, Vol. 174, No. 7, September 25, 2006 973–983
http://www.jcb.org/cgi/doi/10.1083/jcb.200605106 JCB 973
Introduction
Vesicular traf cking provides a continuous exchange of pro-
teins and lipids between membranes in a eukaryotic cell (for re-
view see Rothman and Wieland, 1996; Schekman and Orci,
1996). Coat proteins are believed to confer much of the speci c-
ity associated with protein sorting into transport vesicles (Le
Borgne and Ho ack, 1998a,b; Springer and Schekman, 1998).
To date, three classes of coated vesicles have been identi ed:
clathrin/adaptor-coated vesicles mainly involved in traf c be-
tween the TGN and the endosomes (Robinson, 1994); coat pro-
tein complex I (COPI), which is responsible for both retrograde
transport from the Golgi back to the ER and intra-Golgi trans-
port (Orci et al., 1997; Spang and Schekman, 1998); and COPII,
which mediates anterograde transport from the ER to the Golgi
apparatus (Bednarek et al., 1996; Schekman and Orci, 1996).
Coat assembly is initiated by activation of the ADP ribo-
sylation factor (ARF) family of small G proteins (Arf1p and the
closely related Sar1p) by which membrane-selective nucleotide
exchange catalysts activate Arf1p or Sar1p for membrane at-
tachment (Sera ni et al., 1991; Donaldson et al., 1992; Palmer
et al., 1993; Barlowe et al., 1994). Arf1p regulates the recruit-
ment of COPI and most clathrin-containing coats, leading to
membrane deformation into coated buds and vesicles. Likewise,
COPII vesicles form when Sar1p-GTP recruits the inner coat
complex (the Sec23/24p heterodimer) and the outer coat (the
Sec13/31p heterotetramer; Matsuoka et al., 1998; Antonny
et al., 2003; Lee et al., 2004).
Although coat proteins account for much of the vesicular
traf c in a cell, no such involvement of coat proteins has been
documented in the formation of vesicles or tubules that convey
membrane and secretory proteins directly from the TGN to the
cell surface. As an example of this limb of the secretory path-
way, we have studied the transport of a cell wall biosynthetic
enzyme, Chs3p (chitin synthase III), from the TGN/endosome
membranes to the plasma membrane of the mother–bud junc-
tion in yeast. Chs3p is a multispanning transmembrane protein
Exomer: a coat complex for transport of select
membrane proteins from the trans-Golgi network
to the plasma membrane in yeast
Chao-Wen Wang,1,2 Susan Hamamoto,1,2 Lelio Orci,3 and Randy Schekman1,2
1Department of Molecular Cell Biology and 2Howard Hughes Medical Institute, University of California, Berkeley, Berkeley, CA 94720
3Department of Cell Physiology and Metabolism, University Medical Center, 1121-Geneva 4, Switzerland
A
yeast plasma membrane protein, Chs3p, transits
to the mother–bud neck from a reservoir compris-
ing the trans-Golgi network (TGN) and endosomal
system. Two TGN/endosomal peripheral proteins, Chs5p
and Chs6p, and three Chs6p paralogues form a complex
that is required for the TGN to cell surface transport of
Chs3p. The role of these peripheral proteins has not been
clear, and we now provide evidence that they create a coat
complex required for the capture of membrane proteins en
route to the cell surface. Sec7p, a Golgi protein required
for general membrane traffi c and functioning as a nucleo-
tide exchange factor for the guanosine triphosphate (GTP)–
binding protein Arf1p, is required to recruit Chs5p to the
TGN surface in vivo. Recombinant forms of Chs5p, Chs6p,
and the Chs6p paralogues expressed in baculovirus form
a complex of approximately 1 MD that binds synthetic
liposomes in a reaction requiring acidic phospholipids,
Arf1p, and the nonhydrolyzable GTPγS. The complex re-
mains bound to liposomes centrifuged on a sucrose density
gradient. Thin section electron microscopy reveals a spiky
coat structure on liposomes incubated with the full com-
plex, Arf1p, and GTPγS. We termed the novel coat exomer
for its role in exocytosis from the TGN to the cell surface.
Unlike other coats (e.g., coat protein complex I, II, and
clathrin/adaptor protein complex), the exomer does not
form buds or vesicles on liposomes.
Correspondence to Randy Schekman: schekman@berkeley.edu
Abbreviations used in this paper: ARF, ADP ribosylation factor; BFA, brefeldin A;
COP, coat protein complex; FPLC, fast protein liquid chromatography; GEF, gua-
nine nucleotide exchange factor; mArf1p, myristoylated Arf1p; NTA, nitrilotri-
acetic acid; PA, phosphatidic acid; PE, phosphatidylethanoamine.
The online version of this article contains supplemental material.

JCB • VOLUME 174 • NUMBER 7 • 2006 974
that is required for chitin ring formation during the G1 phase of
the cell cycle and, subsequently, in lateral cell wall chitin syn-
thesis (Shaw et al., 1991). Unlike other cell surface proteins,
Chs3p export is regulated in response to cell cycle and stress
signals (Shaw et al., 1991; Valdivia and Schekman, 2003). How-
ever, throughout the cell cycle, it is maintained in an intracellu-
lar reservoir by being recycled between the TGN and the early
endosomes. This recycling is mediated by clathrin and an adap-
tor protein complex (AP-1). Chs5p and Chs6p are peripheral
proteins that are required to transport Chs3p from the reservoir
to the cell surface (Santos et al., 1997; Ziman et al., 1998).
chs5
∆
and chs6
∆
mutants accumulate Chs3p in the TGN/endo-
some membranes, and the deletion of clathrin or subunits of
AP-1 restores Chs3p traf c to the cell surface by some unknown
bypass pathway (Valdivia et al., 2002). Thus, at least two mech-
anisms of traf c from the TGN/endosome membrane to the cell
surface are possible for Chs3p. Each pathway involves an unex-
plored protein-sorting event that packages Chs3p into secretory
vesicles that are delivered to the bud plasma membrane by the
standard secretory pathway (Valdivia et al., 2002).
The Chs5 and Chs6 proteins are restricted to yeast and
fungi, which may imply an organism-speci c role such as the
biosynthesis of yeast cell wall chitin. Yeast cells have three ad-
ditional Chs6-like proteins (Bch1p [YMR237W], Bud7p, and
Bch2p [YKR027W]), and their roles relative to Chs6p and the
traf c of Chs3p are not yet understood (Satchatjate and Schekman,
2006; Trautwein et al., 2006). To pursue the role of Chs5p and
Chs6p in the sorting and packaging of Chs3p, we have cloned
and characterized the gene products, including the three para-
logues of Chs6p. We have demonstrated that these Chs5 and
Chs6 proteins are associated with each other in a complex
that may make direct contact with Chs3p (Sanchatjate and
Schekman, 2006). In this study, we identi ed Sec7p, a TGN-
localized Arf1p nucleotide exchange factor that is required for the
membrane association of Chs5p and regulating the interaction
of Chs5p and Arf1p. We elucidated the biochemical require-
ments for membrane recruitment of a complex of Chs5p, Chs6p,
and the Chs6p paralogues and describe a novel coat structure
termed exomer, which forms when the complex is recruited to
synthetic membranes in the presence of Arf1p and GTPγS.
Results
Chs5p membrane association is mediated
by Sec7p
In a previous study, Santos and Snyder (2000) found Chs5p lo-
calized to puncta in cells marked by the TGN/endosome resi-
dent protein Kex2p. We con rmed the late Golgi localization of
Chs5p-RFP by comparing its localization in vivo with Sec7p-
GFP (late Golgi marker) and Anp1p-GFP (early Golgi marker)
using double staining live cell imaging (Fig. S1, A–C; available
at http://www.jcb.org/cgi/content/full/jcb.200605106/DC1). To
determine whether this localization is perturbed when Golgi
traf c is disrupted, we used a functional, integrated Chs5p-GFP
to examine its localization in wild type and in cells defective in
secretory protein traf c from the Golgi complex. In wild-type
cells, Chs5p-GFP localized to spots dispersed within the cyto-
plasm, as was determined previously (Santos and Snyder, 2000),
and this distribution was similar at 26 and 37°C.
We also expressed Chs5p-GFP in two temperature- sensitive
mutants in which the secretory function of the Golgi ap paratus
can be altered. A temperature-sensitive allele of the phosphati-
dylinositol 4-kinase (PIK1) gene blocks protein secretion and
accumulates exaggerated Golgi structures (Flanagan et al., 1993;
Figure 1. Chs5p is regulated by a Sec7p- and Arf1p-dependent machinery.
(A) Chs5p-GFP localizes to Golgi membranes, and this association is
Sec7p dependent. Wild-type (WT; CWY512), pik1-83 (CWY559), and
sec7-4 (CWY612) cells that bear chromosomally tagged Chs5p-GFP were
grown at 26°C to mid-log phase and either kept at 26°C or shifted to 37°C
for 40 min for fl uorescence microscopy. Arrowheads indicate exaggerated
Golgi. (B) Chs5p interacts with Arf1p-PA. CWY506 cells harboring Arf1p-
PA integrated at the chromosomal locus were harvested at mid-log phase
as described in Materials and methods. A clear cell lysate (input) was dis-
tributed in aliquots and incubated with 0.5 mM GTP, GMP-PNP, or GTPγS
for 10 min at 30°C. Arf1p–protein A was absorbed by IgG-coated Dyna-
beads at 4°C for 2 h. Beads were recovered using a magnet, and the un-
bound proteins were removed in the fl ow through (FT) followed by wash
(W) steps. Protein remaining bound (B) to the beads was resuspended in
buffer and analyzed by anti-Chs5p (also recognizes protein A [PA]) immuno-
blotting. (C) CWY506 was lysed as described in B followed by incubation
with 0.5 mM GTPγS in the presence of buffer (−) or 2 μg/ml brefeldin A
(BFA) at 30°C for 10 min. Samples were processed as in B except that only
the bound (B) samples are shown.

A NOVEL COAT FOR TRAFFIC FROM THE TRANS-GOLGI NETWORK • WANG ET AL. 975
Hama et al., 1999; Walch-Solimena and Novick, 1999; Audhya
et al., 2000). Similarly, temperature-sensitive alleles of SEC7 ar-
rest secretory traf c and accumulate large Golgi stacks at a re-
strictive temperature (Novick et al., 1980; Deitz et al., 2000). In
the pik1-83 strain at 37°C, Chs5p-GFP coalesced into large
punctae, as was previously observed for the Golgi marker Och1p
(Fig. 1 A; Strahl et al., 2005). In contrast, Chs5p-GFP dispersed
in a diffuse pattern in sec7-4 cells incubated at 37°C, although
Chs5p-GFP localized normally at 26°C in both mutant strains
(Fig. 1 A). The Golgi marker proteins Anp1p-RFP and Kex2p-
GFP accumulated in exaggerated structures at 37°C in sec7-4,
indicating that the Golgi membrane did not disperse (Fig. S1 D).
Attempts to localize Chs6p-GFP were unsuccessful because the
 uorescence of Chs6p-GFP was too weak for micro scopy as a
result of a low expression level (unpublished data).
SEC7 encodes a nucleotide exchange factor for Arf1p.
One explanation for the difference between pik1-83 and sec7-4
is that the sec7-4 mutation resides within the conserved Arf1
guanine nucleotide exchange factor (GEF) domain (Sec7 do-
main; Deitz et al., 2000). Thus, we considered the possibility
that Chs5p was restricted to Golgi membranes through an inter-
action with activated Arf1p. To test this possibility, we created a
strain harboring protein A fused to the C-terminal codon of the
chromosomal copy of ARF1. A cytosol fraction obtained from
this strain was incubated with GTP or one of two nonhydrolyz-
able analogues, GMP-PNP and GTPγS. Arf1p–protein A was
recovered by binding to IgG-coated Dynabeads, and samples
corresponding to the input, unbound wash fraction and bound
(bead) materials were separated and evaluated by SDS-PAGE
followed by Arf1p and Chs5p immunoblotting. A clear copuri-
 cation of Chs5p with Arf1p was detected in incubations con-
taining GTPγS and, to a lesser extent, with GMP-PNP (Fig. 1 B).
Nucleotide hydrolysis in the GTP sample may explain the poor
binding of Chs5p to Arf1p.
To further con rm the nucleotide-dependent interaction
between Chs5p and Arf1p–protein A and to support the role of
Sec7p in regulating this association, we next evaluated the re-
covery of bound Chs5p in incubations containing the ARF GEF
inhibitor brefeldin A (BFA). We observed a reduction in the re-
tention of Chs5p in an incubation containing GTPγS and 2 μg/ml
BFA (Fig. 1 C). A similar effect of BFA in vivo has been re-
ported by Trautwein et al. (2006). These experiments suggest
that activated Arf1p-GTP, presumably by contact with Sec7p,
recruits Chs5p to the Golgi membrane to initiate its role in the
traf c of Chs3p.
Lipid-binding and Chs6 interaction domains
of Chs5p
In addition to binding to activated Arf1p, Chs5p may interact by
additional contact with the Golgi membranes. To determine
whether Chs5p interacts with lipids, we expressed GST hybrid
forms of Chs5p and Chs6p in Escherichia coli, and the puri ed
GST hybrid proteins were probed for lipid interaction using an
overlay assay on PIP strips. GST-Chs5p but not GST or GST-
Chs6p showed signi cant interaction with most anionic lipids
(unpublished data). A similar spectrum was seen with 6× His-
tagged Chs5p expressed in yeast (unpublished data).
Fragments of Chs5p were fused to GST to de ne the region
interacting with lipids (Fig. 2 A). The potential anionic lipid inter-
action domain was mapped to the C-terminal one third of Chs5p.
This region may correspond to the C-terminal lysine-rich tail
(Santos et al., 1997). Although the C-terminal domain may facili-
tate the membrane recruitment of Chs5p, it appears to be dispens-
able for chitin synthesis because a truncated version of CHS5
corresponding to amino acid residues 1–401 complemented a
chs5
∆
strain based on the growth sensitivity of yeast cells making
chitin to the chitin-binding dye calco uor (unpublished data).
To understand the interaction between Chs5p and Chs6p,
fragments of Chs5p used in the lipid-binding assay were evalu-
ated for interaction with Chs6p using the yeast two-hybrid assay.
Binding domain–Chs5p (aa 1–79) and binding domain–Chs5p
(aa 1–260) showed two-hybrid interaction with activation
Figure 2. Functional mapping of Chs5p. (A) Mapping
Chs5p lipid interaction and Chs6p interaction regions.
Full-length Chs5p (aa 1–671) contains two motifs: FN3
(aa 79–160) and BRCT (aa 160–260). The fragments
positive for lipid interaction are shown by a plus sign
based on GST fusion constructs purifi ed from E. coli that
showed the lipid interaction profi le on PIP strips (Echenlon).
These fragments were also cloned into the yeast two-
hybrid construct pGBD to test interaction with pGAD-Chs6.
Growth on an SD-Leu-Ura-His plate is shown by a plus
sign. (B) Wild-type (WT; SEY6210) cells harboring GST,
GST-Chs5 (aa 1–79), and GST-Chs5 (aa 401–671)
cloned into pRS424 (2μ; tryptophan) were streaked on
SD-Trp plates in addition to 50 (CF 50) or 100 μg/ml (CF
100) calcofl uor. For each construct, two colonies were
restreaked and examined on the plates.

JCB • VOLUME 174 • NUMBER 7 • 2006 976
domain–Chs6p (Fig. 2 A). These two domains (the lipid-binding
domain within the C terminus of Chs5p and the Chs6 interac-
tion domain within the N terminus of Chs5p) were compared
by a competition assay evaluating their functional importance
in chitin synthesis. The overexpression of GST hybrids con-
taining the N-terminal domain of Chs5p interfered with Chs3p
traf c as judged by the calco uor growth test, whereas a GST
hybrid containing the C-terminal domain of Chs5p did not im-
pair chitin synthesis (Fig. 2 B). Thus, the interaction of Chs5p
and Chs6p may be crucial for the transport of Chs3p.
A purifi ed recombinant complex of Chs5p,
Chs6p, and Chs6 paralogues
To further investigate the role of activated Arf1p in Chs5p mem-
brane recruitment (Fig. 1), we sought to isolate the Chs5p- and
Chs6p-containing complex (Fig. 2) for functional tests. Efforts
to express the proteins in stable oligomeric forms in E. coli and
yeast resulted in poor yields. However, expression in baculovi-
rus proved more reliable. We created baculovirus vectors con-
taining N-terminally 6× His-tagged Chs5p and untagged
versions of one or more copies of Chs6p and its paralogues
(Bch1p, Bud7p, and Bch2p; Fig. 3 A). Recent evidence has sug-
gested that Chs5p interacts with Chs6p and each of the Chs6
paralogues and that complexes include more than one copy of
Chs6p and its paralogues (Sanchatjate and Schekman, 2006;
Trautwein et. al., 2006). The baculovirus system allowed us to
evaluate complex formation by coexpressing multiple combina-
tions of these recombinant Chs5p and Chs6p proteins. Consistent
with previous observations (Sanchatjate and Schekman, 2006;
Trautwein et. al., 2006), we found that Chs5p copuri ed with
each Chs6p or its paralogue when the two were coexpressed and
that multiple paralogues of Chs6p copuri ed with Chs5p from
cells expressing two, three, or all Chs6p and Chs6p paralogues
(Fig. 3 A and not depicted).
In most cases, the apparent abundance of Chs5p, based on
Sypro red staining intensity, approximated the abundance of the
sum of the Chs6 and Chs6p paralogues. Bch2p was a notable
exception, perhaps because of a lower virus titer. Infection with
a larger Bch2 baculovirus stock increased the abundance of this
species in the isolated Chs5p complex (unpublished data). The
ratios of the Chs6p species in the complex differed from those
detected in the complex isolated from wild-type yeast cells
Figure 3. Purifi cation and characterization of the Chs5p and
Chs6p protein complexes. (A) Chs5p and all Chs6-like pro-
teins were purifi ed as complexes. Amplifi ed baculovirus
stocks were inoculated (+) or not inoculated (−) into the in-
sect culture cell line Sf-9. Cultures were harvested after 4 d
postinoculation, and proteins were purifi ed on a Ni-NTA col-
umn (for binding to 6× His-tagged Chs5p). Purifi ed proteins
were examined by SDS-PAGE followed by Coomassie blue
staining. MW, mol wt. (B) Protein complexes purifi ed on Ni-
NTA were size fractionated on a Superose 6 FPLC column.
Elution during 6–18 ml is shown, and the localization of mol
wt standards is indicated. Peak fractions of the complex (size
as indicated) were analyzed by SDS-PAGE and Sypro red
staining. As examined on gels, His-Chs5p and Chs6p-like
proteins such as Bch1p, Bud7p, and Chs6p are found to-
gether in the 1-MD fractions. (C) The purifi ed Chs5–Chs6[all]
complex was analyzed by a 10–50% sucrose gradient.
A total of 20 × 100-μl fractions were collected from the top,
and proteins were analyzed by SDS-PAGE and Sypro red
staining. Gels were visualized using a Typhoon imager.

A NOVEL COAT FOR TRAFFIC FROM THE TRANS-GOLGI NETWORK • WANG ET AL. 977
(Sanchatjate and Schekman, 2006), probably re ecting the dif-
ferent level of CHS gene expression in these circumstances. The
recombinant expression of Chs6p and its paralogues without
Chs5p also resulted in complexes including Chs6p and one or
more Chs6p paralogues (unpublished data).
Af nity-puri ed complexes were evaluated by gel  ltration
on a Superose 6 fast protein liquid chromatography (FPLC) col-
umn to determine the size and composition of the complex.
Complex isolated from cells expressing Chs5p and all four Chs6p
and its paralogues (Chs5–Chs6[all]) fractionated at a position con-
sistent with a size slightly >1 MD with coincident chromatography
of the most abundant Chs6p species (Fig. 3 B). Similar patterns
of  ltration were seen with complexes containing only one Chs6p
paralogue, although a complex of Chs5p and Chs6p fractionated
somewhat heterogeneously. An independent method of separa-
tion, velocity sedimentation on a sucrose density gradient, con-
 rmed that Chs5p, Chs6p, and Chs6p paralogues (Chs5–Chs6[all])
were present in a large complex (Fig. 3 C). This pattern of cosedi-
mentation was not altered in the Chs5–Chs6[all] samples treated
with 3 M urea, 1% Triton X-100, or 1 M KCl (unpublished data).
Overall, recombinant Chs5p, Chs6p, and Chs6p paralogues, like
those isolated from wild-type yeast cells, appear to be self-
organized into a large and stable complex.
Myristoylated Arf1p-GTP𝛄S recruits
the Chs5–Chs6[all] complex to liposomes
Because genetic evidence suggests that more than just one of
the Chs6p and its paralogues is required for traf c of Chs3p to
the cell surface (Sanchatjate and Schekman, 2006; Trautwein
et al., 2006), we used the complex including Chs5p, Chs6p, and
three Chs6 paralogues (Chs5–Chs6[all]) to evaluate the role of
activated Arf1p in membrane recruitment. Recombinant myris-
toylated Arf1p (mArf1p; Q71L; GTPase de cient) was puri ed
from E. coli and mixed with GTPγS and liposomes formulated
with synthetic phospholipids with various levels of selected
acidic phospholipids (Fig. 2 A; Spang et al., 1998). EDTA was
included to stimulate spontaneous GTP/GDP exchange on
Arf1p (Antonny et al., 1997). After 1 h at 30°C, MgCl2 was
added to stabilize Arf1-GTPγS, and the samples were supple-
mented with Chs5–Chs6[all] and incubated for a further 10 min
at 22°C. For each liposome formulation, three samples were
prepared: complete, without Arf1p, and without GTPγS.
Liposomes and bound proteins were separated from unbound
materials by  otation sedimentation on a sucrose density shelf.
Fig. 4 A documents the recovery of proteins bound to liposomes,
and Fig. 4 B displays a quantitative representation of proteins
recovered in the  oated fractions. In most samples, Arf1p bound
to liposomes in the presence or absence of GTPγS. However,
recruitment of the Chs5–Chs6[all] complex was optimum in
incubations that contained Arf1p and GTPγS.
Certain liposome formulations (e.g., phosphatidylcholine/
phosphatidylethanoamine [PE]/phosphatidylserine/phosphatidic
acid [PA] with a high concentration of PI(4)P or phosphati-
dylinositol-4,5-bisphosphate) recruited Chs5–Chs6[all] in the
absence of Arf1p or GTPγS (Fig. 4, A and B). Other formula-
tions (e.g., major-minor mix optimized for COPII assembly;
Matsuoka et. al., 1998) displayed a substantial GTPγS require-
ment for recruitment of the Chs5–Chs6[all] complex. From
these results, we conclude that Arf1p binds directly to the
Chs5–Chs6[all] complex and facilitates the association of the
complex with liposomes, particularly with certain formulations
containing one or more of several acidic phospholipids.
The major-minor mix formulation, which was optimized
for the recruitment of activated Sar1p and COPII to liposomes
(Matsuoka et al., 1998) and which also works well for the re-
cruitment of activated Arf1p and coatomer (COPI; Spang, et al.,
1998), was reexamined for wild-type Arf1p and Chs5–Chs6[all]
assembly in the presence of various nucleotides. Fig. 4 C shows
a substantial dependence on GTPγS (or GMP-PNP), with much
less of the complex recruited in the presence of GDP or GTP.
GTP hydrolysis during recruitment and liposome sedimentation
may explain the failure to retain comparable amounts of the
Chs5–Chs6[all] complex on liposomes incubated with Arf1p-
GTP. Likewise, a Golgi-enriched membrane fraction incubated
with the Chs5–Chs6[all] complex con rmed that the binding of
Chs5–Chs6[all] in a reaction was stimulated by GTPγS (unpub-
lished data).
We next compared recruitment of the Chs5–Chs6[all]
complex to the other subcomplexes formed with Chs5p, Chs6p,
or fewer paralogues of Chs6p (Fig. 4 D). Chs5p alone was very
inef ciently recruited to major-minor liposomes. Combinations
including one paralogue were recruited in a manner largely in-
dependent of GTPγS. However, two paralogues known to be
important in Chs3p traf c (Chs6p and Bch1p) were recruited to
membranes in an Arf1p-GTPγS–dependent manner comparable
with Chs5–Chs6[all] (Fig. 4 D).
Stoichiometric recruitment of the Chs5–
Chs6[all] complex to liposomes
To provide evidence that the recruitment is under control by a
speci c, regulated process, we varied the concentration of Arf1-
GTPγS and the Chs5–Chs6[all] complex and the time of incu-
bation to discover the optimum conditions of assembly on
major-minor mix liposomes. Two-stage recruitment assays were
performed. Liposomes were incubated with 1 μM mArf1p
and GTPγS in a  rst-stage binding reaction as in Fig. 4 and
were mixed with the Chs5–Chs6[all] complex for 15 min at
22°C. Membrane-bound protein complexes were collected by
 otation on a step sucrose cushion, and the protein content was
measured by Sypro red staining of gels (Fig. 5 A). Chs5–
Chs6[all] complex binding to liposomes was saturated at 0.5
μM in this experiment. Conversely, at a  xed Chs5–Chs6[all]
concentration of 0.8 μM, Arf1p binding increased nonsatura-
bly (Fig. 5 B, ii), whereas binding of the Chs5–Chs6[all] com-
plex appeared to approach saturation at around 1.5 μM Arf1p
(Fig. 5 B, i). Excess bound Arf1p may not be functional or
accessible to the Chs5–Chs6[all] complex. Titration at the low
range of Arf1p, where the linear membrane association of Arf1p
and Chs5p was shown (Fig. 5 B, iii), indicated an 7.5:1 molar
stoichiometry of the Arf1/Chs5–Chs6[all] complex, re ecting
either a substantial fraction of bound Arf1p that is not accessible
to the Chs5–Chs6[all] complex or incomplete density gradient
recovery of Chs5–Chs6[all] compared with activated Arf1p
bound to liposomes.

JCB • VOLUME 174 • NUMBER 7 • 2006 978
The relationship of Arf1p exchange and Chs5–Chs6[all]
recruitment in a reaction was evaluated in a time course
experiment. Protein binding to liposomes was conducted with
Arf1p at approximately the maximum Chs5–Chs6[all] ratio
achieved in the titration experiment in Fig. 5 A. Incubations
included EDTA to promote Arf1p nucleotide exchange.
Samples were collected at the indicated times at 30°C, mixed
with MgCl2, and chilled on ice for the duration. The kinetics
of binding paralleled the rate of EDTA-stimulated Arf1p
nucleotide exchange as measured by the change in tryptophan
 uorescence of activated Arf1p (Fig. 5 C and not depicted;
An tonny et al., 1997). These results support our conclusion
that GTPγS-activated Arf1p recruits the Chs5–Chs6[all] complex
to the membrane.
Chs5–Chs6[all] complex forms a coated
surface on liposomes
Given the similarity between the Chs5–Chs6[all] complex and
the COPs (coatomer and COPII) in regard to Arf1p-GTPγS
(or Sar1p–GMP-PNP)–dependent recruitment to liposomes, we
examined the in uence of this assembly on the buoyant density
of liposomes. COPII proteins that assemble on liposomes in the
presence of Sar1p–GMP-PNP cause membranes to shift to a
higher buoyant density, re ecting the formation of protein-
coated surfaces and synthetic COPII vesicles (Matsuoka et al.,
1998). We formulated major-minor mix liposomes with Texas
red–PE and conducted assembly incubations for the complete
reaction containing Arf1-GTPγS and Chs5–Chs6[all] as de-
scribed above. Samples were applied to a 10–50% linear su-
crose gradient and centrifuged to equilibrium for 16 h at 4°C. In
control incubations, in samples of liposomes incubated with the
Chs5–Chs6[all] complex without Arf1p or GTPγS or liposomes
incubated with Arf1p and GTPγS alone (unpublished data),  u-
orescent lipid peaked at the top of the gradient (fraction 1). In a
complete reaction, most of the  uorescent lipids and Arf1p sed-
imented into the gradient but remained in the low density lipo-
some fractions (Fig. 6, C and D; fractions 2–8), whereas most of
the unbound Chs5–Chs6[all] complex sedimented to a high
density with no apparent lipid cofractionation (Fig. 6, B and D).
Chs proteins and Arf1p were detected in fractions 2–8 but not in
a sample of Chs5–Chs6[all] incubated with liposomes in
the absence of Arf1p (Fig. 6, B and C). Thus, at least some
liposome-bound Chs proteins may coat membranes suf ciently
to in uence membrane buoyant density.
Figure 4. Recruitment of the Chs5–Chs6[all]
complex by mArf1p. (A) Liposomes composed
of various phospholipid formulations were
tested for recruitment of the Chs5–Chs6[all]
complex in the presence of GTPγS alone,
mArf1p(Q71L) alone, or mArf1p(Q71L) and
GTPγS. Liposomes were fl oated through a step
sucrose gradient as described in Materials
and methods. Liposome-bound proteins were
analyzed by SDS-PAGE followed by Sypro red
staining. Gels were visualized using a Typhoon
imager. (B) Comparison of binding as shown
in A. mArf1p recovery from different liposome
formulations as shown in A was set at 100%
(light gray bar), relative recoveries of the His-
Chs5p amount (wt/wt) are compared (gray
bar), and the amount of His-Chs5p fl oated in
the absence of mArf1p was subtracted (black
bar) from the gray bar. The y axis measures
the relative binding index. (C) Standard re-
cruitment assay. Major-minor liposomes were
incubated with 1 μM mArf1p in the presence
of buffer (−), 0.1 mM GDP, GTP, GTPγS, or
GMP-PNP at 30°C for 1 h in a chelating condi-
tion used to trigger nucleotide exchange. 0.5
μM Chs5–Chs6[all] complex was tested for
binding as described in Materials and methods.
Floated proteins were examined by Sypro red
staining. (D) The same mArf1p exchange con-
ditions were performed as in C, but different
proteins or protein complexes were tested for
their association with mArf1p in the presence
of 0.1 mM GDP (D) or GTPγS (T). PC, phos-
phatidylcholine; PS, phosphatidylserine; PI,
phosphatidylinositol.

A NOVEL COAT FOR TRAFFIC FROM THE TRANS-GOLGI NETWORK • WANG ET AL. 979
The unambiguous assignment of a membrane coat re-
quires inspection by thin section electron microscopy. Samples
were prepared from a complete incubation (Fig. 7 A), an incu-
bation of liposomes and Arf1p-GTPγS alone (Fig. 7 C), and
an incu bation with Arf1p, Chs5–Chs6[all] complex, and GDP
(Fig. 7 B). A spiky coat appeared uniformly distributed along
the surface of liposomes incubated under conditions in which
the Chs5–Chs6[all] complex binds to the membrane but not in
the control conditions. Although some membrane pro les ap-
peared elongated, no coated buds or small coated vesicles were
evident. Thus, the Chs5–Chs6[all] complex, which we now call
exomer, appears to form coats on the membrane. However, un-
like the COPs, exomer by itself does not deform membranes to
induce the formation of buds and transport vesicles.
Discussion
Vesicular traf c in several limbs of the secretory pathway is ini-
tiated by the GTP-binding protein-dependent recruitment of
coat proteins that sequester cargo molecules in a bud and pinch
the membrane to form a transport vesicle. Clathrin, COPI, and
COPII are well-known examples of this seemingly general fea-
ture (for review see Rothman and Wieland, 1996; Schekman
and Orci, 1996). However, one clear gap in our knowledge con-
cerns the mechanism of sorting and transport of membrane and
secretory proteins from the TGN to the cell surface. Although
some proteins use clathrin to traverse the endosome en route to
the cell surface (Ang et al., 2004), others do not, and, until now,
the general view has been that the direct path out of the TGN
may involve tubular carriers formed without the intervention of
coat proteins. Some regulatory proteins control the formation of
transport carriers at the TGN by modulating lipid composition
(for example, activation of phospholipase D by protein kinase C;
Simon et al., 1996). Other proteins, such as FAPPs (four-
phosphate adaptor proteins), are phosphoinositide-binding pro-
teins that are found associated with TGN carriers and also
interact with Arf1p (Godi et al., 2004). Protein kinase D is re-
cruited to the TGN through interaction with diacylglycerol and
Figure 5. Recruitment of the Chs5–Chs6[all]
complex via mArf1p in the presence of GTP𝛄S
is a saturable process. (A) Titration of the
Chs5–Chs6[all] complex. 1 μM mArf1p incu-
bated with GTPγS and various concentrations
of the Chs5–Chs6[all] complex at room tem-
perature for 15 min. Recruitment was deter-
mined by a step fl otation gradient as described
in Materials and methods. Sypro red staining
of one experiment is shown, and quantitative
results from several experiments are plotted
below. (B) Titration of mArf1p with a fi xed
amount of the Chs5–Chs6[all] complex. Various
amounts of mArf1p were incubated with GTPγS
for 1 h at 30 min. 0.8 μM of a fi xed concen-
tration of the Chs5–Chs6[all] complex was
then tested for binding at room temperature for
15 min. Recruitments were determined by a
step fl otation gradient as described in Materi-
als and methods. Sypro red staining of one
experiment is shown, and quantitative results
from several experiments are plotted below.
(i and ii) The membrane-associated His-Chs5p
(i) and mArf1p (ii) are quantifi ed. (iii) The
membrane-associated His-Chs5p versus
mArf1p (wt/wt) before the saturation concen-
tration is quantifi ed. (C) Recruitment time
course experiment. 1 μM mArf1p and 0.5 μM
Chs5–Chs6[all] complex were mixed at t = 0.
Reactions were incubated at 30°C and stopped
at the indicated times. Membrane association
of His-Chs5p versus mArf1p (wt/wt) was deter-
mined by a step fl otation gradient as described
in Materials and methods. Sypro red staining
of one experiment is shown, and quantitative
results from several experiments are plotted
below. Error bars represent SD.

JCB • VOLUME 174 • NUMBER 7 • 2006 980
is subsequently activated by phosphorylation to promote carrier
 ssion (Liljedahl et al., 2001).
We have investigated traf c of the plasma membrane en-
zyme that makes the chitin ring at the mother–bud junction in
growing yeast cells. This protein, Chs3p, has an interesting
itinerary and set of genetic requirements for its traf c that are
somewhat distinct from the requirements of other cell surface
proteins. Speci cally, the action of two peripheral membrane
proteins, Chs5p and Chs6p, suggest a special machinery to
convey Chs3p from the TGN to the cell surface (Valdivia et al.,
2002). Chs5p and Chs6p form subunits of a complex, includ-
ing one or more additional Chs6p paralogues that also facili-
tate the traf c of Chs3p (Sanchatjate and Schekman, 2006;
Trautwein et. al., 2006). Chs5p is required for the transport of
at least two other proteins in addition to Chs3p, one of which,
Fus1p, a cell surface protein required for yeast cell fusion, does
not depend on Chs6p (Santos and Snyder, 2003). By analogy to
the paralogues of Sec24p that form dimers with Sec23p in the
COPII coat (Shimoni et al., 2000), we propose that Chs5p,
Chs6p, and its paralogues may recognize distinct sets of mem-
brane cargo proteins at the TGN for packaging into mature
secretory vesicles.
Our strategy has been to identify conditions that promote
the recruitment of Chs5p and Chs6p to the TGN/endosomal
membrane and then to reconstruct this interaction with syn-
thetic membranes. Two lines of evidence support a role for ac-
tivated Arf1p in Chs5p–Chs6p assembly at the TGN. Trautwein
et al. (2006) found that Arf1p-GTPγS binds Chs5p and Chs6p
in a crude lysate of yeast. In vivo, nucleotide exchange on Arf1p
is promoted by several proteins sharing a GEF domain that was
 rst identi ed in a peripheral Golgi protein, Sec7p, which is
required for secretory traf c within the Golgi complex. Mutations
within the Sec7 GEF domain block traf c of all or most
secretory proteins and disperse Chs5p from its normal punc-
tate TGN/endosomal localization (Fig. 1 A). BFA blocks GEF
activity and reduces the recruitment of Chs5p to immobilized
Figure 6. Enrichment of coated membrane on a sucrose density gradient.
(A) Sucrose density index for the gradient centrifuged at 4°C for 16 h at
100,000 g in a TLS55 rotor. (B) Distribution of the Chs5–Chs6[all] com-
plex (11–20) alone. (C) Cofractionation of lipids, mArf1p, and the Chs5p–
Chs6[all] complex in the lighter fractions (1–10). (D) Quantifi cation of the
distribution in B and C.
Figure 7. Coating of the Chs5–Chs6[all] complex on liposomes. (A) Major-
minor liposomes were incubated with mArf1p and GTPγS at 30°C for 1 h,
and the Chs5–Chs6[all] complex was then added and incubated at room
temperature for 15 min followed by fi xation for thin section electron micros-
copy. An extensive coating on liposomes can be seen as spikes formed by
patches. The inset is a higher magnifi cation view of the boxed area.
(B) Major-minor liposomes were incubated with mArf1p-GDP for 1 h at 30°C,
and the Chs5–Chs6[all] complex was added at room temperature for 15
min. (C) Major-minor liposomes were incubated with mArf1p-GTPγS alone
at 1 h at 30°C, and buffer was added at room temperature for 15 min.

A NOVEL COAT FOR TRAFFIC FROM THE TRANS-GOLGI NETWORK • WANG ET AL. 981
Arf1p-GTPγS (Fig. 1 C). Arf1p-GTP may tether Chs5p to
membrane much as it does for coatomer in COPI vesicle for-
mation and adaptor proteins (AP-1 and -2) for clathrin-coated
vesicles and as Sar1p-GTP does for Sec23p–Sec24p in COPII
coat assembly.
Coat recruitment and assembly on arti cial membranes is
stimulated by acidic phospholipids. Recombinant Chs5p inter-
acts with Chs6p and with a variety of acidic phospholipids
(Figs. 2 and 4). Recombinant complex expressed in baculovirus-
infected Sf-9 cells as combinations of Chs5p, Chs6p, and one or
more of the Chs6p paralogues fractionates as an 1-MD com-
plex that binds to arti cial membranes in an Arf1-GTPγS (or
GMP-PNP)–dependent manner (Fig. 4). Optimal interaction
occurs on liposomes that are similar in composition to those
formulated for the recruitment and assembly of the COPII and I
coats (Matsuoka et al., 1998; Spang et al., 1998). Complexes
formed with binary combinations of Chs5p and Chs6p or one
other Chs6p paralogue are somewhat less strictly dependent on
Arf1p-GTPγS for membrane recruitment. Thus, we suggest
that the native mixed complex (i.e., Chs5–Chs6[all]) is the
likely form recruited to TGN membranes in vivo.
The Chs5–Chs6[all] complex binds to and perturbs the
density of liposome membranes (Fig. 6). Thin section micros-
copy of  xed, membrane-bound complex reveals an electron-
dense coat whose morphology is quite distinct from that seen
for other coating complexes associated with cargo traf c (Fig. 7).
Unlike COPII and I, which assemble onto and vesiculate arti -
cial liposomes, the Chs5–Chs6[all] coat forms a spiky struc-
ture but does not pinch the membrane into buds and small
vesicles. We propose to call the Chs coat the exomer to re ect
its role in the exocytosis of Chs3p and select additional proteins.
Although this coat also traf cs a subset of other cargo proteins,
the subunits are dispensable for normal cell viability, and the
CHS5 and CHS6 genes are largely restricted to fungi that make
chitin. It seems likely that other coats of similarly restricted
roles will be uncovered in yeast and in other organisms, but
the null phenotype of such coat subunits may not be nearly as
dramatic as one normally associates with a general block
in secretion.
The absence of coated buds and small vesicles in the prep-
aration of exomer-coated liposomes suggests that other factors
cooperate with the exomer to convey cargo proteins out of the
TGN donor compartment. Chs3p interacts with Chs5p–Chs6p
(Sanchatjate and Schekman, 2006; Trautwein et al., 2006), and
it is possible that cargo-coat contact may promote the shape
change that accompanies vesicle morphogenesis. Alternatively
or in addition, the exomer may engage elements of the cytoskel-
eton, perhaps actin directly, to draw cargo molecules into tu-
bules much as has been shown for cooperative interaction of
clathrin, actin, and the Arp2/3 complex in cell surface invagina-
tion and endocytosis in yeast (Kaksonen et al., 2003, 2005).
Indeed, genetic studies link Chs5p and Chs6p to elements of
the cytoskeleton (Tong et al., 2004; Lesage et al., 2005). A bio-
chemical reconstitution approach with exomer, liposomes, and
cytoskeletal proteins may now be used in an effort to recapitu-
late the formation of more fully developed TGN to cell surface
traf c intermediates.
Materials and methods
Yeast strains and materials
GFP was integrated at the C-terminal codon of the CHS5 locus by a
PCR-based one-step transformation procedure (Longtine et al., 1998).
Primers for C-terminal GFP integration were 5′-A A G A A G A A T A A G A A G A-
A T A A G A A G A A A G G G A A A A A G A A A C G G A T C C C C G G G T T A A T T A A and
3′-A T A A A A A A T A G A T T A T A T T T G C T G A G G G A T T C T C A G T C G G A A T T C G A G C T-
C G T T T A A A C . A similar approach was applied to generate the Arf1-PA
and chs5
∆
strains. Primers for C-terminal Arf1 integration were 5′-G G T-
T T G G A A T G G T T A A G T A A C A G T T T G A A A A A C T A A A C T C G G A T C C C C G G-
G T T A A T T A A and 3′-C T T T A T G T T T C A T T T A G T T T A T A C A A G C G T A T T T G A T C C-
G A A T T C G A G C T C G T T T A A A C . Deletion primers to generate chs5
∆
were
5′-G T C T T C A G T T G A T G T A C T G T T A A C A G T A G G T A A G T T G G A C G G A T C C C-
C G G G T T A A T T A A and 3′-A T A A A A A A T A G A T T A T A T T T G C T G A G G G A T T C T-
C A G T C G G A A T T C G A G C T C G T T T A A A C .
Yeast strains used in this study include the following: SEY6210 (MAT
α
,
leu2-3, ura3-52, his3-
∆
200, lys2-801, trp-
∆
901, suc2-
∆
9), CWY512
(Chs5-GFP::HIS3 SEY6210), CWY559 (MATa pik1-83:TRP1 ade2-
101och his3-200 leu2-1 lys2-801a trp1-d ura3-52, Chs5-GFP::HIS3,
pRS412), CWY612 (MATa ura3-1 leu2-3112 trp1-
∆
1 his3-11,15 sec7-4,
Chs5-GFP::KAN), CWY506 (Arf1-PA::HIS SEY6210), CWY624 (chs5
∆
::
LEU2 SEY6210), and PJ69-4A (MAT
α
, trp1-901, leu2-3, 112, ura3-52,
his3-200, gal4
∆
, gal80
∆
, LYS2::GAL1-HIS3, GAL2-ADE2, met2::
GAL7-lacZ).
All chemical reagents were purchased from Sigma-Aldrich unless
specifi ed. Lipids were obtained from Avanti Polar Lipids, Inc., and PIP strips
were purchased from Echenlon. Sypro red protein staining dye was pur-
chased from Invitrogen. Complete protease inhibitor cocktail was obtained
from Roche Molecular Biochemicals. E. coli BL21 (DE3) coexpressing yeast
N-myristoltransferase and Arf1 (wild type and Q71L) were provided by
R. Kahn (Emory University, Atlanta, GA). Anti-Chs5 antiserum was described
previously (Sanchatjate and Schekman, 2006). Anti-GST antibody was
purchased from Santa Cruz Biotechnology, Inc. DH10Bac Competent cells
and the Bac-to-Bac baculovirus expression system were purchased from
Invitrogen. Glutathione-Sepharose fast fl ow, pGEX vector, and the Superose
6 gel fi ltration column were obtained from GE Healthcare. Dynabead
M-500 subcellular was purchased from Dynal.
Protein purifi cation
GST, GST-Chs5, GST-Chs6, and other GST-Chs5 fragments shown in Fig. 1 B
were constructed in the pGEX vector (GE Healthcare), and proteins were
purifi ed from E. coli BL21. In brief, 1 L of cell culture was grown to A600 =
0.5–1.0 followed by 200 μM IPTG induction for 3 h at 22°C. Cells were
resuspended in 20 ml PBS buffer, lysed by sonication, and centrifuged at
12,000 rpm for 10 min. The clear cell lysate was incubated with 5 ml
glutathione-Sepharose (prewashed by PBS) at 4°C for 3 h. Beads and ad-
sorbed proteins were poured into a column, and 3 × 30 ml PBS aliquots
were applied to the column to remove nonspecifi c material. Bound proteins
were eluted with 12 ml PBS + 10 mM of reduced glutathione. For the lipid-
protein overlay assay, we adjusted purifi ed proteins to 150 μg/ml and fol-
lowed the procedures recommended by the manufacturer (Echelon).
mArf1p was purifi ed from an E. coli BL21 (DE3) strain that coex-
pressed yeast N-myristoyltransferase and either wild-type or dominant-
activated (Q71L) Arf1p. The purifi cation procedures have been published
previously (Randazzo et al., 1992), except a Sephacryl S-100 column was
used for the gel fi ltration step. Based on the mobility shift by electrophoresis
and in agreement with the literature, we confi rmed that >75% of the
E. coli–purifi ed mArf1p, either wild-type or the dominant-activated mutant,
was N-myristoylated. The purifi cation resulted in 80% pure mArf1p.
The Chs5p–Chs6p complex was purifi ed using the Bac-to-Bac bac-
ulovirus expression system (Invitrogen). To make the recombinant viruses
for the baculovirus expression system, we used pFastBac (Invitrogen) to
clone all Chs6p-like proteins, and pFastBac-HT was used to clone Chs5p,
resulting in an N-terminal 6× His tag. Genes were cloned independently
into constructs followed by transformation into DH10Bac competent cells
for the production of recombinant bacmid vectors. Bacmid DNA was then
transfected into Sf-9 insect cells, and viruses were amplifi ed separately
and kept frozen at −80°C for later use; the same virus stocks were
thawed from a small aliquot, and the virus titer remained constant. 500 ml
of insect cultures (Sf-9) were infected with the indicated virus stock; for
example, 500 μl of virus stock of His-Chs5, Chs6, Bch1, Bund7, and
Bch2 was added, respectively, into the same culture to assemble the
Chs5–Chs6[all] complex. Infected cultures were harvested after 4 d of
growth at room temperature.

JCB • VOLUME 174 • NUMBER 7 • 2006 982
Cells were lysed in 20 ml of lysis buffer I (50 mM Hepes, pH 7.4,
450 mM KOAc, 0.1 mM EGTA, 20 mM imidazole, and 10% glycerol)
supplemented with 1× protease inhibitor cocktail and 1 mM PMSF. The ly-
sate was centrifuged at 12,000 rpm for 10 min (SS34 rotor; Sorvall), and
the resulting supernatant was incubated with 5 ml Ni–nitrilotriacetic acid
(NTA) agarose at 4°C for 3–4 h. Beads were washed fi rst with 20 ml of ly-
sis buffer I followed by washing twice with wash buffer II (50 mM MES,
pH 6.3, 450 mM KOAc, 0.1 mM EGTA, 40 mM imidazole, 10% glyc-
erol, and 1 mM PMSF), three additional washes with wash buffer III (50
mM Hepes, pH 7.4, 450 mM KOAc, 250 mM sorbitol, 0.1 mM EGTA, 40
mM imidazole, 10% glycerol, and 1 mM PMSF), and one wash with HKG
(50 mM Hepes, pH 7.4, 50 mM KOAc, and 10% glycerol). Bound pro-
teins were eluted with HKGI buffer (50 mM Hepes, pH 7.4, 50 mM KOAc,
10% glycerol, and 0.5 M imidazole), and the elution was then dialyzed
against HKG with three changes of buffer. Precipitation occurred during
dialysis, and the protein aggregates were removed by centrifugation at
14,000 rpm at 4°C for 5 min. The protocol results in a >90% pure protein
complex at a protein concentration of 0.5 mg/ml. If necessary, the com-
plex was concentrated by loading a 1-ml aliquot of the Ni-NTA fraction on
a 200-μl, 2.2 M sucrose cushion followed by 100,000 rpm centrifugation
for 1.5 h (TLA100.3 rotor; Beckman Coulter). Most proteins were recov-
ered in the sucrose fraction. This procedure yielded approximately
>1 mg/ml of pure complex.
Arf1-PA pull-down experiment
CWY506 cells were harvested at A600 = 1.0 from 100 ml of culture.
Cells were lysed by agitation with glass beads in 2.5 ml of lysis buffer B88,
1× complete protease inhibitor cocktails, and 1 mM PMSF. Total cell ly-
sates were centrifuged at 20,000 g for 10 min at 4°C. To the resulting su-
pernatant, we added 0.5 mM of nucleotide where indicated and incubated
at 30°C for 10 min. 20 μl IgG-coated Dynabeads were added and incu-
bated at 4°C for 2 h. Procedures for IgG coating on Dynabeads were pro-
vided by the manufacturer. Dynabeads were recovered by binding to a
magnet, and beads were washed with 1 ml B88 buffer three times. Bound
proteins were eluted in 100 μl SDS-PAGE resuspension buffer, and 10 μl
was loaded on SDS-PAGE followed by immunoblot analysis.
Liposome recruitment assay
Liposomes were prepared as described previously except 1 mol percent-
age of Texas red–PE was substituted for NBD phospholipids (Matsuoka
et al., 1998; Spang et al., 1998). In brief, 2 mM of lipid mixtures based on
the lipid composition indicated in Fig. 4 B were prepared, and the organic
solvent was evaporated by a rotavapor. HK buffer (20 mM Hepes, pH 6.8,
and 160 mM KOAc) was applied to dried lipids, and the suspension was
incubated at room temperature overnight followed by 19 passages through
a 400-nm Nuclepore polycarbonate membrane.
20 μl of liposomes consisting of 1 mM of lipids were incubated for
1 h at 30°C with 1.9 μg of purifi ed mArf1p (fi nal concentration of 1 μM
in a 80-μl standard reaction) and 0.1 mM nucleotide (GDP, GTP, or GTPγS)
in a buffer composed of 20 mM Hepes, pH 7.4, 1.2 mM MgCl2, 2.5
mM EDTA, 50 mM NaCl, and 15 mM KOAc. Samples were returned to
ice, and 2 mM MgCl2 was added to stabilize the nucleotide-loaded
mArf1p. The reaction was then adjusted to 80 μl by the addition of 0.5
μM Chs5–Chs6[all] complex (mol wt of 300 kD) or as indicated. In a
second stage, complex recruitment was conducted in a buffer composed of
38 mM Hepes, pH 7.4, 1 mM EDTA, 1.7 mM MgCl2, 18.75 mM NaCl,
35 mM KOAc, and 6.25% glycerol and followed by the binding step at
room temperature for 15 min unless otherwise indicated. Samples were
then transferred to ice and mixed with 50 μl of 2.5 M sucrose in HKM buf-
fer (20 mM Hepes, pH 6.8, 160 mM KOAc, and 1 mM MgCl2). A 110-μl
sample was removed to a tube (TLA-100; Beckman Coulter) and layered
with 100 μl of 0.75 M sucrose in HKM and 20 μl HKM, sequentially.
Samples were centrifuged in a TLA-100 rotor (Beckman Coulter) at
100,000 rpm for 25 min at 24°C. The upper 30-μl fractions were carefully
removed after centrifugation, 5 μl of which was used for the determination
of lipid recovery based on the fl uorescence of Texas red–PE. The remaining
fl oated fraction (20 μl) was resuspend in SDS-PAGE resuspension buffer,
and the amount applied to SDS-PAGE was based on lipid recovery. Gels
were stained with Sypro red, images were taken using a Typhoon imager
(GE Healthcare), and quantifi cation was performed using ImageQuant
software (GE Healthcare).
Gel fi ltration and gradient analysis
The gel fi ltration experiment was performed using an AKTA FPLC system
(GE Healthcare). The mol wt standards for this experiment contained
1 mg/ml blue dextran (2,000 kD), 2.5 mg/ml thyroglobulin (669 kD), 0.15
mg/ml ferritin (440 kD), 1 mg/ml catalase (232 kD), 1 mg/ml aldolase
(158 kD), 4 mg/ml BSA (74 kD), 0.15 mg/ml ribonuclease (13.7 kD), and
1 mg/ml cytochrome C (12.4 kD). 200-μl samples (0.5–1.0 mg/ml) of
a purifi ed complex were injected into the system using a constant fl ow rate
at 0.2 ml/min and collected into fractions of 250 μl. A 10-μl aliquot of
each fraction was analyzed by SDS-PAGE followed by Sypro red staining.
Quantifi cation was performed using a Typhoon imager and ImageQuant
software. To check the integrity of the purifi ed Chs5–Chs6[all] complex, we
performed a sucrose velocity gradient analysis. A 120-μl sample contain-
ing 60 μg of purifi ed complex was loaded on top of a 1.89-ml 10–50%
linear sucrose gradient in HKM, and centrifugation was performed for 16 h
at 4°C in a rotor (TLS55; Beckman Coulter) at 55,000 rpm. Similar gradi-
ent conditions were used to check lipid and protein distribution in the stan-
dard incubation, and 160 μl of the reaction mixture was applied instead.
A total of 20 fractions (20 × 100 μl) were collected from the top of this
gradient, and the sucrose concentration was determined using a refractom-
eter (Fisher Scientifi c). To check lipid recovery, we removed a 50-μl sample
from each fraction into a microtiter plate for quantifi cation of the fl uores-
cence of the Texas red–PE. SDS-PAGE resuspension buffer was added to
the remaining fraction, and proteins were analyzed by SDS-PAGE and
Sypro red staining. Quantifi cation was performed using a Typhoon imager
and ImageQuant software.
Fluorescence and electron microscopy
All strains used for microscopy were grown in synthetic dextrose (SD)
medium to mid-log phase. Cells were incubated either at 26°C or shifted to
37°C for 40 min before examination. Microscopy was performed using a
fl uorescence microscope (Eclipse E600; Nikon). Images were captured by
a CCD camera (C4742-95; Hamamatsu) using Image-Pro software (Media
Cybernetics). For thin section microscopy, we fi xed a standard reaction as
described in the Liposome recruitment assay section with 2% glutaldehyde
and 1% osmium tetroxide in cacodylate buffer for 1 h on ice. Samples
were centrifuged using a TLA100.3 rotor (Beckman Coulter) at 55,000
rpm for 30 min. Membrane pellet fractions were processed for thin section
electron microscopy as described previously (Orci et al., 1993).
Online supplemental material
Fig. S1 shows that Chs5p colocalized with Sec7p to the late Golgi com-
partment. Although the Chs5p-GFP/RFP signal was diffusely distributed at
the restrictive temperature in the sec7-4 strain, other Golgi marker signals
were focused and more exaggerated, indicating that the Golgi membrane
did not disperse in the sec7-4 strain. Online supplemental material is avail-
able at http://www.jcb.org/cgi/content/full/jcb.200605106/DC1.
We thank Drs. Eugene Futai, Jinoh Kim, and other members of the Schekman
laboratory for discussion and encouragement and Bob Lesch and Crystal
Chan for technical assistance. We thank Pierre Cosson for the suggestion of
the name exomer as well as for improving the manuscript. We are grateful to
have had Drs. Robyn Barfi eld, Alenka Copic, Jinoh Kim, and Trevor Starr im-
prove the manuscript. We thank Drs. David Drubin and Georjana Barnes for
use of their FPLC system and microscope, Jasper Rine for access to his fl uores-
cence microscope, Jeremy Thorner for the pik1-83 strain, and Ann Fischer for
the cell culture facility. We thank the Pole Facultaire de Microscopie Ultrastruc-
ture at the University of Geneva Medical School for access to electron micros-
copy equipment.
This work was supported by the National Institutes of Health (grant
GM26755), the Howard Hughes Medical Institute (grant to R. Schekman),
and the Swiss National Science Foundation (grant to L. Orci).
Submitted: 16 May 2006
Accepted: 18 August 2006
References
Ang, A.L., T. Taguchi, S. Francis, H. Folsch, L.J. Murrells, M. Pypaert, G.
Warren, and I. Mellman. 2004. Recycling endosomes can serve as in-
termediates during transport from the Golgi to the plasma membrane of
MDCK cells. J. Cell Biol. 167:531–543.
Antonny, B., S. Beraud-Fufour, P. Chardin, and M. Chabre. 1997. N-terminal hy-
drophobic residues of the G-protein ADP-ribosylation factor-1 insert into
membrane phospholipids upon GDP to GTP exchange. Biochemistry.
36:4675–4684.
Antonny, B., P. Gounon, R. Schekman, and L. Orci. 2003. Self-assembly of mini-
mal COPII cages. EMBO Rep. 4:419–424.

A NOVEL COAT FOR TRAFFIC FROM THE TRANS-GOLGI NETWORK • WANG ET AL. 983
Audhya, A., M. Foti, and S.D. Emr. 2000. Distinct roles for the yeast phospha-
tidylinositol 4-kinases, Stt4p and Pik1p, in secretion, cell growth, and
organelle membrane dynamics. Mol. Biol. Cell. 11:2673–2689.
Barlowe, C., L. Orci, T. Yeung, M. Hosobuchi, S. Hamamoto, N. Salama, M.F.
Rexach, M. Ravazzola, M. Amherdt, and R. Schekman. 1994. COPII: a
membrane coat formed by Sec proteins that drive vesicle budding from
the endoplasmic reticulum. Cell. 77:895–907.
Bednarek, S.Y., L. Orci, and R. Schekman. 1996. Traf c COPs and the formation
of vesicle coats. Trends Cell Biol. 6:468–473.
Deitz, S.B., A. Rambourg, F. Kepes, and A. Franzusoff. 2000. Sec7p directs the
transitions required for yeast Golgi biogenesis. Traf  c. 1:172–183.
Donaldson, J.G., D. Finazzi, and R.D. Klausner. 1992. Brefeldin A inhibits Golgi
membrane-catalysed exchange of guanine nucleotide onto ARF protein.
Nature. 360:350–352.
Flanagan, C.A., E.A. Schnieders, A.W. Emerick, R. Kunisawa, A. Admon, and
J. Thorner. 1993. Phosphatidylinositol 4-kinase: gene structure and
requirement for yeast cell viability. Science. 262:1444–1448.
Godi, A., A. Di Campli, A. Konstantakopoulos, G. Di Tullio, D.R. Alessi, G.S.
Kular, T. Daniele, P. Marra, J.M. Lucocq, and M.A. De Matteis. 2004.
FAPPs control Golgi-to-cell-surface membrane traf c by binding to ARF
and PtdIns(4)P. Nat. Cell Biol. 6:393–404.
Hama, H., E.A. Schnieders, J. Thorner, J.Y. Takemoto, and D.B. DeWald. 1999.
Direct involvement of phosphatidylinositol 4-phosphate in secretion in
the yeast Saccharomyces cerevisiae. J. Biol. Chem. 274:34294–34300.
Kaksonen, M., Y. Sun, and D.G. Drubin. 2003. A pathway for association of
receptors, adaptors, and actin during endocytic internalization. Cell.
115:475–487.
Kaksonen, M., C.P. Toret, and D.G. Drubin. 2005. A modular design for the
clathrin- and actin-mediated endocytosis machinery. Cell. 123:305–320.
Le Borgne, R., and B. Ho ack. 1998a. Mechanisms of protein sorting and coat
assembly: insights from the clathrin-coated vesicle pathway. Curr. Opin.
Cell Biol. 10:499–503.
Le Borgne, R., and B. Ho ack. 1998b. Protein transport from the secretory to
the endocytic pathway in mammalian cells. Biochim. Biophys. Acta.
1404:195–209.
Lee, M.C., E.A. Miller, J. Goldberg, L. Orci, and R. Schekman. 2004. Bi-
directional protein transport between the ER and Golgi. Annu. Rev. Cell
Dev. Biol. 20:87–123.
Lesage, G., J. Shapiro, C.A. Specht, A.M. Sdicu, P. Menard, S. Hussein, A.H. Tong,
C. Boone, and H. Bussey. 2005. An interactional network of genes involved
in chitin synthesis in Saccharomyces cerevisiae. BMC Genet. 6:8.
Liljedahl, M., Y. Maeda, A. Colanzi, I. Ayala, J. Van Lint, and V. Malhotra. 2001.
Protein kinase D regulates the  ssion of cell surface destined transport
carriers from the trans-Golgi network. Cell. 104:409–420.
Longtine, M.S., A. McKenzie III, D.J. Demarini, N.G. Shah, A. Wach, A.
Brachat, P. Philippsen, and J.R. Pringle. 1998. Additional modules for
versatile and economical PCR-based gene deletion and modi cation in
Saccharomyces cerevisiae. Yea s t . 14:953–961.
Matsuoka, K., L. Orci, M. Amherdt, S.Y. Bednarek, S. Hamamoto, R. Schekman,
and T. Yeung. 1998. COPII-coated vesicle formation reconstituted
with puri ed coat proteins and chemically de ned liposomes. Cell.
93:263–275.
Novick, P., C. Field, and R. Schekman. 1980. Identi cation of 23 complementa-
tion groups required for post-translational events in the yeast secretory
pathway. Cell. 21:205–215.
Orci, L., D.J. Palmer, M. Ravazzola, A. Perrelet, M. Amherdt, and J.E. Rothman.
1993. Budding from Golgi membranes requires the coatomer complex of
non-clathrin coat proteins. Nature. 362:648–652.
Orci, L., M. Stamnes, M. Ravazzola, M. Amherdt, A. Perrelet, T.H. Sollner, and
J.E. Rothman. 1997. Bidirectional transport by distinct populations of
COPI-coated vesicles. Cell. 90:335–349.
Palmer, D.J., J.B. Helms, C.J. Beckers, L. Orci, and J.E. Rothman. 1993. Binding
of coatomer to Golgi membranes requires ADP-ribosylation factor.
J. Biol. Chem. 268:12083–12089.
Randazzo, P.A., O. Weiss, and R.A. Kahn. 1992. Preparation of recombinant
ADP-ribosylation factor. Methods Enzymol. 219:362–369.
Robinson, M.S. 1994. The role of clathrin, adaptors and dynamin in endocytosis.
Curr. Opin. Cell Biol. 6:538–544.
Rothman, J.E., and F.T. Wieland. 1996. Protein sorting by transport vesicles.
Science. 272:227–234.
Sanchatjate, S., and R. Schekman. 2006. Chs5/6 complex: a multi-protein com-
plex that interacts with and conveys chitin synthase III from the trans-
Golgi network to the cell surface. Mol. Biol. Cell. doi:10.1091/mbc.
E06-03-0210.
Santos, B., and M. Snyder. 2000. Sbe2p and sbe22p, two homologous Golgi pro-
teins involved in yeast cell wall formation. Mol. Biol. Cell. 11:435–452.
Santos, B., and M. Snyder. 2003. Speci c protein targeting during cell differen-
tiation: polarized localization of Fus1p during mating depends on Chs5p
in Saccharomyces cerevisiae. Eukaryot. Cell. 2:821–825.
Santos, B., A. Duran, and M.H. Valdivieso. 1997. CHS5, a gene involved in chi-
tin synthesis and mating in Saccharomyces cerevisiae. Mol. Cell. Biol.
17:2485–2496.
Schekman, R., and L. Orci. 1996. Coat proteins and vesicle budding. Science.
271:1526–1533.
Sera ni, T., G. Stenbeck, A. Brecht, F. Lottspeich, L. Orci, J.E. Rothman, and
F.T. Wieland. 1991. A coat subunit of Golgi-derived non-clathrin-coated
vesicles with homology to the clathrin-coated vesicle coat protein beta-
adaptin. Nature. 349:215–220.
Shaw, J.A., P.C. Mol, B. Bowers, S.J. Silverman, M.H. Valdivieso, A. Duran,
and E. Cabib. 1991. The function of chitin synthases 2 and 3 in the
Saccharomyces cerevisiae cell cycle. J. Cell Biol. 114:111–123.
Shimoni, Y., T. Kurihara, M. Ravazzola, M. Amherdt, L. Orci, and R. Schekman.
2000. Lst1p and Sec24p cooperate in sorting of the plasma membrane
ATPase into COPII vesicles in Saccharomyces cerevisiae. J. Cell Biol.
151:973–984.
Simon, J.P., I.E. Ivanove, M. Adesnik, and D.D. Sabatini. 1996. The production
of post-Golgi vesicles requires a protein kinase C-like molecule, but not
its phosphorylating activity. J. Cell Biol. 135:355–370.
Spang, A., and R. Schekman. 1998. Reconstitution of retrograde transport from
the Golgi to the ER in vitro. J. Cell Biol. 143:589–599.
Spang, A., K. Matsuoka, S. Hamamoto, R. Schekman, and L. Orci. 1998.
Coatomer, Arf1p, and nucleotide are required to bud coat protein com-
plex I-coated vesicles from large synthetic liposomes. Proc. Natl. Acad.
Sci. USA. 95:11199–11204.
Springer, S., and R. Schekman. 1998. Nucleation of COPII vesicular coat
complex by endoplasmic reticulum to Golgi vesicle SNAREs. Science.
281:698–700.
Strahl, T., H. Hama, D.B. DeWald, and J. Thorner. 2005. Yeast phosphatidylino-
sitol 4-kinase, Pik1, has essential roles at the Golgi and in the nucleus.
J. Cell Biol. 171:967–979.
Tong, A.H., G. Lesage, G.D. Bader, H. Ding, H. Xu, X. Xin, J. Young, G.F.
Berriz, R.L. Brost, M. Chang, et al. 2004. Global mapping of the yeast
genetic interaction network. Science. 303:808–813.
Trautwein, M., C. Schindler, R. Gauss, J. Dengjel, E. Hartmann, and A. Spang.
2006. Arf1p, Chs5p and the ChAPs are required for export of specialized
cargo from the Golgi. EMBO J. 25:943–954.
Valdivia, R.H., and R. Schekman. 2003. The yeasts Rho1p and Pkc1p regulate
the transport of chitin synthase III (Chs3p) from internal stores to the
plasma membrane. Proc. Natl. Acad. Sci. USA. 100:10287–10292.
Valdivia, R.H., D. Baggott, J.S. Chuang, and R.W. Schekman. 2002. The yeast
clathrin adaptor protein complex 1 is required for the ef cient retention of
a subset of late Golgi membrane proteins. Dev. Cell. 2:283–294.
Walch-Solimena, C., and P. Novick. 1999. The yeast phosphatidylinositol-4-OH
kinase pik1 regulates secretion at the Golgi. Nat. Cell Biol. 1:523–525.
Ziman, M., J.S. Chuang, M. Tsung, S. Hamamoto, and R. Schekman. 1998.
Chs6p-dependent anterograde transport of Chs3p from the chitosome
to the plasma membrane in Saccharomyces cerevisiae. Mol. Biol. Cell.
9:1565–1576.