Copyright C Munksgaard 2001
Traffic 2001; 2: 820–830
Munksgaard International Publishers
Sgf1p, a New Component of the Sec34p/Sec35p
Dong-Wook Kim1,2, Thomas Massey1,2,
Michael Sacher1,2, Marc Pypaert1 and
Department of Cell Biology and 2Howard Hughes Medical
Institute, Yale University, New Haven, CT, USA
* Corresponding author: Susan Ferro-Novick,
Here we report the identification of SGF1 as a highcopy suppressor of the sec35–1 mutant. SGF1 encodes an essential hydrophilic protein of ⬃100 kDa.
Using the yeast two-hybrid system and coprecipitation
studies, we demonstrate that Sgf1p is a new subunit
of the multiprotein Sec34p/Sec35p complex. Reduced
levels of Sgf1p lead to the accumulation of a variety of
membranes as well as a kinetic block in endoplasmic
reticulum to Golgi traffic. Immunofluorescence studies
demonstrate that Sec34p is found throughout the
Golgi, with a high concentration on early Golgi. Although an earlier study suggested that Sec34p
(Grd20p) is not required for protein secretion, we show
here that the sec34–2 and sec35–1 mutations lead to
a pleiotropic block in the secretion of all proteins into
the growth medium.
Key words: membrane traffic, secretion, vesicle
Received 16 May 2001, revised and accepted for publication 16 August 2001
The composition of organelles is maintained by the proper
targeting of different classes of vesicles to their correct destination. Vesicle targeting reactions involve recognition of the
vesicle by the target membrane followed by two different
types of membrane interactions. The first, called tethering,
does not bring the vesicle and acceptor membrane in close
enough proximity to allow for membrane fusion. The second
reaction, referred to as docking, leads to a more intimate interaction between the two membranes so that fusion can
Tethering reactions at different stages of the exocytic and
endocytic pathways appear to be effected by multicomponent complexes that a; re peripherally associated with the
target membrane (1), while the docking reactions are mediated by a family of integral membrane proteins called the
SNAREs. SNAREs are related to proteins found on the synaptic vesicle or the neuronal plasma membrane. They cycle between the donor and acceptor compartments and are highly
conserved from yeast to man (2). Unlike the tethering complexes, which do not resemble each other, all SNAREs are
homologous to either synaptobrevin/VAMP, syntaxin or
SNAP-25. The pairing of a SNARE on the vesicle with its
cognate partner on the target membrane brings the vesicle
in direct contact with the acceptor compartment, allowing
these two membranes to fuse (3).
ER to Golgi transport is unusual in that two large tethering
complexes have been implicated in the tethering of COP II
vesicles to the Golgi, TRAPP I (4,5) and the Sec34p/Sec35p
complex (6,7). Surprisingly, it was recently reported that
GRD20, which is identical to SEC34, is required for protein
sorting in the trans-Golgi/endosomal system, but not protein
secretion (8). Mutations in grd20 exhibited only a modest
reduction in the secretion of invertase, while the vacuolar protease carboxypeptidase Y (CPY) was secreted (missorted)
into the medium. In light of these findings, we have re-evaluated the role of Sec34p and Sec35p in membrane traffic.
Using an assay that measures the transport of all proteins
secreted into the growth medium, we show here that the
sec34–2 and sec35–1 mutants pleiotropically block secretion. Previous studies have shown that the entire cellular pool
of Sec34p is in a complex with Sec35p (6). Interestingly,
immunofluorescence studies have revealed a gradient of
Sec34p throughout the Golgi with the majority concentrated
on early Golgi. In an effort to identify additional members of
this complex, we have performed a high-copy suppressor
screen with the sec35–1 mutant. This screen has led to the
identification of Sgf1p, a new component of the complex.
Reduced levels of Sgf1p lead to a kinetic delay in ER to Golgi
transport and the accumulation of a variety of membranes.
The implications of these findings are discussed.
The sec34–2 and sec35–1 mutants pleiotropically
As a previous study suggested that Sec34p is not required
for secretion (8), we re-evaluated the role of Sec34p in the
secretory pathway using a different type of assay. A specific
set of proteins are secreted into the medium in wild-type
cells (see Figure 1, lane 1) (9), and mutations in genes whose
products are directly involved in anterograde transport block
the secretion of these proteins. In contrast, mutations in
genes whose products are indirectly involved in secretion
may block the traffic of some, but not all, proteins secreted
by wild-type cells. For example, in sec21–3, which blocks
retrograde Golgi to ER transport, a subset of proteins (CPY
Figure 1: Sec34p and Sec35p are required for secretion. Two
OD599 units of cells were pulse-labeled with 150 mCi of [35S] ProMix
in 0.4 ml of minimal media for 15 min at 37 æC before the cells were
pelleted by centrifugation. Proteins secreted into the media were
precipitated with TCA and visualized by SDS-PAGE and autofluorography. Molecular markers are shown to the right. The bands secreted into the medium are marked with asterisks.
and a-factor) are completely blocked in the ER, whereas
other proteins (invertase and HSP150) are secreted normally
(10). The block in CPY transport in sec21–3 may be due to
the limited retrieval of transport factors such as cargo receptors that are required for the packaging of certain cargo into
anterograde ER to Golgi COP II vesicles (10).
To assay for a general defect in secretion, sec34–2 and
sec35–1 as well as sec21–1, sec23–1, sec18–1 and
sec1–1, were preshifted to 37 æC for 20 min and pulse-labeled
for 15 min. Cells and media were separated by centrifugation
and proteins in the medium were precipitated with TCA and
resolved by SDS-PAGE. In wild-type, eight protein bands
were visible in the medium (Figure 1, lane 1), while few were
apparent in sec21–1 (Figure 1, lane 4). No proteins were detected in the medium of three different mutants that block
anterograde transport, sec23–1, sec18–1 and sec1–1
(Figure 1, lanes 2, 3 and 5). Interestingly, in sec34–2 and
sec35–1 most secretory proteins failed to be transported into
the medium (Figure 1, lanes 6 and 7). A faint protein band
was seen in sec34–2 (Figure 1, lane 6), which may be due
to an incomplete block in secretion in this mutant at 37 æC.
These findings clearly demonstrate that mutations in subunits
of the Sec34p/Sec35p complex lead to a robust block in
Traffic 2001; 2: 820–830
Sec34p resides primarily on an early Golgi compartment
Previous localization studies, using a Sec34p-HA construct,
have shown that Sec34p resides in the cytosol and on the
Golgi (8). Immunofluorescence studies revealed that 51% of
the Golgi-localized Sec34p colocalized with the trans-Golgi
marker A-ALP. A-ALP contains the cytoplasmic domain of
dipeptidyl aminopeptidase fused to the transmembrane and
lumenal domains of alkaline phosphatase (8). The localization
of over half the Sec34p to the trans-Golgi seemed counterintuitive to its proposed role in ER to Golgi tethering (7). To
re-examine the distribution of Sec34p, we localized Sec34pmyc in strains expressing the cis-Golgi marker Och1p-HA,
and a different late Golgi marker, Sec7p-GFP (11,12). We
found that 79% of the Golgi puncta staining for Sec34p-myc
colocalized with puncta staining for Och1p-HA (Figure 2A),
whereas only 28% colocalized with puncta staining for
Sec7p-GFP (Figure 2B). Looking at the reverse colocalizations, we found that 73% of the Golgi puncta staining for
Och1p-HA and 27% of the puncta staining for Sec7p-GFP
also stained for Sec34p-myc. Hence, although Sec34p-myc
can be found throughout the Golgi, it predominantly localizes
to an early Golgi compartment. Interestingly, a human ortholog of Sec34p has recently been reported to colocalize well
with cis/medial Golgi markers and partially with a trans-Golgi
Identification of high-copy suppressors of the sec35–1
The Sec34p/Sec35p complex contains several putative members (6). To identify new subunits or genes whose products
may interact with the Sec34p/Sec35p complex, we transformed the sec35–1 mutant with a 2-mm yeast genomic library and
screened for genes that suppress the growth defect of sec35–
1 at 37 æC. A total of 81 suppressors were isolated from 77 000
transformants and each showed plasmid-dependent suppression. Restriction analysis and DNA sequencing revealed
that these plasmids could be divided into five groups, each
with a different region of genomic DNA. The SEC35 and
SEC34 structural genes comprised two of the groups. The third
group was found to contain the SBP1 and RPL8A genes. SBP1
encodes a single-stranded nucleic acid-binding protein,
whereas RPL8A encodes the ribosomal protein L8A. This
group was also found to suppress sec34–2 (6). Previous
studies suggested that this suppression was indirect, as members of this group were found to suppress mutations that block
membrane traffic at all stages of the exocytic pathway. The insert in the fourth group was subcloned to 3.75 kb (data not
shown). It contained a hypothetical open reading frame (ORF)
(YPR105C) that conferred suppression. YPR105C encodes a
hydrophilic protein of 861 amino acids with a predicted molecular weight of ⬃98 kDa. It lacks a signal sequence as well as
transmembrane domains, and does not have any known motifs that facilitate membrane attachment. Thus, this ORF is predicted to encode a cytosolic or peripheral membrane protein
that we have named SGF1 (suppressor gene of sec35). A
search of the database revealed homologs of SGF1 in S.
pombe (CAA19344) and humans (AL050101). The insert in
Kim et al.
Figure 2: The
Sec34p-myc colocalizes with
the early Golgi marker Och1pHA. (A) Subcellular localization of
Sec34p-myc and Och1p-HA. A
strain carrying integrated Sec34pmyc
Och1p-HA was fixed. The cell wall
was removed and the sample was
probed with mouse anti-myc and
rabbit anti-HA antibodies. Cy3(anti-mouse) and FITC- (anti-rabbit) conjugated secondary antibodies were used to localize the
primary antibodies by fluorescence microscopy. Representative cells are shown. Plain arrowheads highlight colocalization and
arrowheads with asterisks indicate
a lack of colocalization. (B) Subcellular localization of Sec34p-myc
and Sec7p-GFP. A strain containing both Sec34p-myc and Sec7pGFP was fixed, converted to
spheroplasts and probed with
mouse anti-myc antibody. Cy3conjugated secondary antibody
was used to localize Sec34p-myc
by fluorescence microscopy,
whereas GFP marked Sec7p-GFP
through its intrinsic fluorescence.
Representative cells are shown.
Plain arrowheads highlight colocalization and arrowheads with asterisks indicate a lack of colocalization.
the fifth group was subcloned to a 2.5-kb fragment that retained suppression activity (data not shown). This fragment contained a hypothetical ORF (YML068W) that encodes a protein
with a molecular weight of ⬃54 kDa which is predicted to have
at least two zinc finger domains. The N-terminus of this protein
is also predicted to have a coiled-coil region and its human ortholog (AF060544) associates with a known transcription factor (14). This finding and the experimental results presented
below have led us to conclude it suppresses sec35–1 indirectly.
Sgf1p interacts with Sec34p and Sec35p in the yeast
To begin to investigate whether Sgf1p physically interacts
with Sec34p and Sec35p, we used the yeast two-hybrid system. This system, which takes advantage of the properties
of the Gal4p transcriptional activator, is a genetic assay for
detecting protein–protein interactions (15). Gal4p contains
two functionally essential domains: a DNA-binding domain
(BD) and a transcription activation domain (AD). The interaction of two proteins fused to Gal4-BD and Gal4-AD, respectively, results in the transcriptional activity of a reporter gene
(here, b-Galactosidase). Full-length Sgf1p, YML068W-p,
Sec34p and Sec35p were fused to Gal4-AD in a pACT2 fish
vector or Gal4-BD in a pAS1-CYH2 bait vector (16). Activation
of the reporter b-Galactosidase gene by the Gal4p transcriptional activator was assessed by the ability of yeast colonies
to produce a blue color when X-gal was used as a substrate.
We found, as expected, that Sec34p and Sec35p interacted
with each other in this assay (Table 1). In addition, both
Sec34p and Sec35p interacted with Sgf1p, but not with
YML068W-p. Thus, Sgf1p, Sec34p and Sec35p specifically
interact with each other in the yeast two-hybrid system.
Sgf1p is a new component of the Sec34p/Sec35p complex
The finding that Sgf1p interacts with Sec34p and Sec35p in
the two-hybrid assay suggested that Sgf1p may be a memTraffic 2001; 2: 820–830
Table 1: Sgf1p interacts with Sec34p and Sec35p by yeast two
ber of the Sec34p/Sec35p complex. To characterize this interaction further, we performed coprecipitation experiments.
Lysates were prepared from yeast strains containing myctagged Sgf1p or YML068W and an untagged strain. The lysates were immunoprecipitated with anti-c-myc antibody,
subjected to Western blot analysis and probed for the presence of Sec35p (Figure 3A). Sec35p was only detected in
the immunoprecipitate from the Sgf1p-myc tagged lysate,
but not the untagged lysate (Figure 3A, lanes 1 and 2). When
the lysate from the strain containing myc-tagged Sgf1p was
treated with 1% SDS prior to immunoprecipitation, only
Sgf1p-myc, but not Sec35p, was detected (lane 3). These
results indicate that Sec35p only coprecipitates with Sgf1p
under nondenaturing conditions. In a similar experiment,
myc-tagged Sec35p did not coprecipitate with YML068W
(data not shown), indicating that there is no direct interaction
between the two.
To analyze the protein that SGF1 encodes, we raised polyclonal antibody to Sgf1p. Anti-Sgf1p antibody recognized a
polypeptide of ⬃100 kDa (Figure 3B, lane 3) that was not
detected by preimmune serum (lanes 1 and 2) and was overproduced in a strain that overexpresses SGF1 (lane 4). This
antibody was then used to probe Western blots in the reciprocal of the coprecipitation experiment described above.
Sec35p-myc was precipitated from a lysate and Western blot
analysis was used to analyze the precipitate for Sgf1p and
Sec34p (Figure 3C). Sgf1p and Sec34p coprecipitated with
Sec35p-myc under nondenaturing conditions, but not from
untagged or Sec35p-myc tagged lysates under denaturing
conditions (Figure 3C, compare lane 2 with lanes 1 and 3).
When wild-type lysates were analyzed on a Superdex-200
gel filtration column, all of the Sgf1p was found to cofractionate with Sec34p and Sec35p (Figure 4). Furthermore, unlike
the TRAPP complex which is found in two forms (5), only
one form of the Sec34p/Sec35p complex, migrating between TRAPP I (⬃ 300 kDa) and TRAPP II (⬃ 1000 kDa), was
detected. Previous studies have shown that Sec34p and
Sec35p quantitatively bind to a Mono Q anion exchange column and are eluted together as a complex (7). If the Sec34p,
Sec35p and Sgf1p that cofractionate on the Superdex-200
column are components of the same complex, they should
co-elute with each other on a Mono Q anion exchange colTraffic 2001; 2: 820–830
Figure 3: Sgf1p is a component of the Sec34p/Sec35p complex. (A) Sec35p coprecipitates with Sgf1p-myc. Lysates prepared
from a strain containing tagged Sgf1p-myc (lane 2) and an untagged strain (lane 1) were incubated with anti-c-myc antibody and
the immune complexes were collected onto protein A-Sepharose
beads as described previously (6). One sample was pretreated with
1% SDS before the addition of antibody (lane 3). The beads were
washed and treated with sample buffer. The solubilized precipitates
were resolved by SDS-PAGE (10%) and subjected to Western blot
analysis using anti-c-myc and anti-Sec35p antibodies. (B) Production of anti-Sgf1p antibody. Lysates prepared from wild-type without (lanes 1 and 3) or with SGF1 on a high-copy vector (lanes 2
and 4) were resolved by SDS-PAGE (10%) and blotted with either
preimmune serum (lanes 1 and 2) or anti-Sgf1p serum (lanes 3 and
4). (C) Sgf1p and Sec34p coprecipitate with Sec35p-myc. Lysates
prepared from a strain containing tagged Sec35p-myc and an untagged strain were treated as in (A). Lane 1, untagged sample; lane
2, Sec35p-myc-tagged sample; lane 3, Sec35p-myc-tagged
sample pretreated with 1% SDS.
umn. Fractions 9 and 10 from the Superdex-200 column (see
Figure 4A) were pooled and applied to a Mono Q column in
the presence of 100 mM KCl and then eluted with a linear salt
gradient (100–500 mM). Western blot analysis of the column
fractions revealed the presence of Sec34p, Sec35p and
Sgf1p in fractions 9 and 10 (see Figure 4B), supporting the
Kim et al.
Figure 4: Sgf1p cofractionates
with Sec34p and Sec35p. (A)
Cytosol (5 mg of total protein), prepared from wild-type cells was
fractionated on a Superdex200 gel filtration column and 25
(1 ml) fractions were collected.
Each fraction was probed with
antibodies directed against Sgf1p,
Sec34p, Sec35p or Trs33p (a subunit of TRAPP I and TRAPP II). Molecular size standards used to calibrate the column were thyroglobulin (669 kDa), ferritin (440 kDa),
catalase (232 kDa), bovine serum
albumin (67 kDa) and chymotrypsinogen A (25 kDa). (B) Fractions 9
and 10 from the Superdex-200
column were loaded onto a Mono
Q column that was eluted with a
linear salt gradient (100 mM to
500 mM KCl). Fractions (15 ¿ 1 ml)
were collected and analyzed for
the presence of Sec34p, Sec35p
and Sgf1p by Western blot analysis. Fractions 9–10 correspond to
a salt concentration of ⬃325 mM.
hypothesis that Sgf1p is a member of the Sec34p/Sec35p
containing medium. After ⬃10 h at 25 æC, the growth rate of
SFNY1031 decreased compared to wild-type (Figure 5A).
Reduced levels of Sgf1p lead to a delay in CPY transport and the accumulation of membranes
To evaluate whether SGF1 is essential, we constructed a diploid strain in which one allele of SGF1 was replaced with the
S. pombe his5π gene (17). This strain was sporulated and
dissected, and the tetrads were incubated on YPD plates at
25 æC. After 5 days, the 12 tetrads examined clearly showed
2 π :2 – segregation for viability. The colonies that grew well
were all his–, indicating that they contained the wild-type
copy of the gene. This finding demonstrates that SGF1 is
essential for growth.
CPY transport was analyzed in SFNY1031 and wild-type cells
that were incubated for 12 h in glucose-containing medium.
In wild-type, CPY is translocated into the ER where it is coreglycosylated (p1 form). Upon transit to the Golgi, outer chain
carbohydrate is added to the core-glycosylated form to yield
p2CPY. Finally, in the vacuole CPY is processed to the mature
form (20). Cells were pulse-labeled for 4 min at 30 æC and
chased for up to 30 min. Wild-type cells efficiently transported CPY to the vacuole during the 30-min chase (Figure
5C). In contrast, at early time points SFNY1031 showed a
clear delay in processing p1 CPY. This delay in CPY transport
correlated with a decrease in Sgf1p (Figure 5B). The amount
of Bos1p, an ER membrane protein, remained unchanged
under these conditions.
Since previous work has implicated Sec34p and Sec35p in
ER to Golgi traffic (7,18,19), we reasoned that Sgf1p, as a
component of the Sec34p/Sec35p complex, may also play
a role at this stage of the secretory pathway. To test this hypothesis, we constructed a strain (SFNY1031) in which the
sole copy of SGF1 was placed behind the glucose-repressible GAL1 promoter. Since SGF1 is essential, this strain
should grow in galactose-containing medium, but not in the
presence of glucose. Using this strain, we tested whether
reduced levels of Sgf1p would result in a membrane traffic
defect. To repress the expression of SGF1, cells that grew in
galactose-containing medium were inoculated into glucose824
To characterize the morphological consequences of reduced
levels of Sgf1p, thin-section electron microscopy of
SFNY1031 (Figure 6B) was compared with that of wild-type
(Figure 6A). In wild-type, tubules of ER were occasionally
found at the periphery of the cell or in contact with the nuclear envelope. In contrast, SFNY1031 cells incubated for
12 h in glucose-containing medium accumulated membranes. The ER lumen and the nuclear envelope were modestly dilated in these cells, and small vesicles (50–60 nm in
Traffic 2001; 2: 820–830
Figure 5: Reduced levels of Sgf1p lead to a delay in CPY
transport. (A) Reduced levels of Sgf1p result in a growth defect in
SFNY1031. Cells were grown at 25 æC in YP medium containing
0.5% galactose and 2% raffinose to an OD599 of 18 and then inoculated into YP medium containing 2% glucose (final OD599 Ω 0.03).
The OD599 was measured at each time point. (B) The amount of
Sgf1p in SFNY1031 was significantly reduced after a 12-h incubation in glucose-containing medium. Cell lysates prepared from wildtype and SFNY1031 were resolved on SDS-PAGE (10%) and subjected to immunoblot analysis using anti-Sgf1p and anti-Bos1p
antibodies. (C) The p1 form of CPY accumulates in SFNY1031. After
a 12-h incubation, cells were pelleted, washed and pulse-labeled
with [35S] ProMix at 30 æC in minimal medium containing 2% glucose.
size) that are comparable in size to ER to Golgi transport vesicles were also observed. Interestingly, large vesicles of the
size seen in mutants blocked in post-Golgi transport and
membrane structures that resembled aberrant Golgi (21)
were sometimes observed, suggesting that Sgf1p might
have a role at multiple stages of the secretory pathway. Thinsection electron micrographs of the sec34–2 (Figure 6C) and
sec35–1 (Figure 6D) mutants have similar phenotypes.
Traffic 2001; 2: 820–830
sec34–2 displays synthetic lethal interactions with
mutations that disrupt Golgi and post-Golgi traffic
Several lines of evidence suggest that the Sgf1p/Sec34p/
Sec35p complex may function at more than one stage of the
secretory pathway. First, the overexpression of SEC34 has
been reported to inhibit the growth of the sec9–4 mutant,
which is defective in post-Golgi secretion (6). Overexpression
of SNC2, a post-Golgi v-SNARE, suppresses sec35–1 (19).
Second, the loss of Grd20p/Sec34p function results in the
missorting of CPY and the mislocalization of a trans-Golgi
membrane protein, Kex2p (8). Third, endocytosis of the afactor receptor is defective in grd20 mutants (8). Finally, mutants with defects in subunits of the Sgf1p/Sec34p/Sec35p
complex accumulate a variety of membranes (Figure 6). This
finding prompted us to examine whether sec34–2 displays
genetic interactions with other mutations that block protein
transport at different stages of the secretory pathway
(Table 2). Synthetic lethality or sickness of double mutants
results when the effect of combining two mutations in the
same haploid cell causes lethality or sickness under normally
permissive conditions. Such interactions suggest that the
proteins encoded by the mutated genes are functionally related (22,23). The sec34–2 mutant was crossed to mutants
that block vesicle budding (sec13–1, sec23–1, sec24–1 and
sec31–1), ER to Golgi traffic (bet3–1), retrograde transport
from the Golgi to the ER (sec21–1, ret2–1, sec27–1 and
arf1D), intra-Golgi traffic (trs130ts2) and post-Golgi transport
(sec2–41 and sec4–8) (5,21,23–26). For each cross, 12–24
tetrads were dissected and the viability of the segregants was
investigated. In combinations of sec34–2 with sec23–1,
sec24–1, sec21–1, ret2–1, sec27–1, sec4–8 and bet3–1,
three viable colonies were found in the majority of tetrads,
with some having two or four viable colonies at 25 æC or
30 æC. Inviable colonies were confirmed to have double mutations by scoring the viable segregants for temperature sensitivity and the appropriate auxotrophic marker. In crosses of
sec34–2 with sec13–1, sec2–41 and trs130ts2, but not with
sec31–1 or arf1D, the double mutants were sick. These genetic interactions, together with other results, support the hypothesis that the Sgf1p/Sec34p/Sec35p complex may act in
more than one membrane traffic event.
Previous studies have shown that Sec34p and Sec35p are
members of a multiprotein complex (6). Here we show that
Sgf1p is a high-copy suppressor of the sec35–1 mutant.
Sgf1p coprecipitates with Sec34p and Sec35p and interacts
with both of these proteins in the yeast two-hybrid system.
Sgf1p, which is approximately the same size as Sec34p
(⬃ 100 kDa), is larger than the other putative members of this
complex (91, 73, 68 and 51 kDa) that were identified in a
precipitate of Sec35p-myc from a radiolabeled lysate (6).
Interestingly, two different high-throughput two-hybrid
screens have shown that Sec35p interacts with Sgf1p
Kim et al.
Figure 6: Reduced levels of
Sgf1p lead to the accumulation of membranes. EM
analysis of wild-type (panel A)
and SFNY1031 (panel B) after a
12-h incubation at 25 æC in glucose-containing medium, and
sec34–2 (panel C) and sec35–1
(panel D) after a 2-h shift at 38 æC.
Samples were prepared for electron microscopy as described
previously (34). ER (large arrowheads); small vesicles (small arrowheads); large vesicles (full arrows); aberrant membranes that
resemble Golgi (open arrows).
The bars in each panel represent
In vitro transport studies have implicated the Sec34p/Sec35p
complex in the tethering of ER-derived COP II vesicles to the
Golgi (7). Several other tethering factors have also been implicated in this process. These factors include TRAPP I, Uso1p,
and the small GTP-binding protein Ypt1p [for reviews, see
(1,29)]. TRAPP I specifically binds to COP II vesicles and
stimulates guanine nucleotide exchange on Ypt1p (5,30). As
the recruitment of Uso1p to membranes is dependent on
Ypt1p (31), the activation of Ypt1p by TRAPP I (5) may be
important for this event. Uso1p is a large protein with a
globular head and a coiled-coil tail (32). Its participation in
vesicle tethering may be a prerequisite to SNARE pairing as
p115, the mammalian ortholog of Uso1p, has been shown to
bind to the SNAREs (33).
A recent report, however, has questioned the role of Sec34p
(also called Grd20p) in the trafficking of secretory proteins
(8). This claim was prompted by the observation that mutations in grd20 result in the mislocalization of resident transGolgi proteins, such as Kex2p, with only modest affects on
the secretion of invertase. To address whether the Sec34p/
Sec35p complex is required for the secretion of other proTraffic 2001; 2: 820–830
Table 2: Synthetic growth defects between sec34–2 and other
teins, we analyzed the secretion of all proteins into the
growth medium. Mutants that block retrograde Golgi to ER
traffic only block a subset of these proteins, while mutations
in bona fide components of the anterograde secretory apparatus pleiotropically block the secretion of all these proteins
(10). Our findings indicate that sec34–2 and sec35–1 pleiotropically block the secretion of this class of proteins into the
medium. Consistent with these observations, a significant
block in invertase secretion was previously reported for
The localization of Sec34p to the trans-Golgi and its proposed role in the localization of trans-Golgi proteins (8)
seems to be in contradiction with a role in tethering ER-derived COP II vesicles. In an effort to resolve this apparent inconsistency, we localized Sec34p with respect to early
(Och1p) and late (Sec7p) Golgi markers. Our findings revealed that most of the membrane-bound Sec34p is found
on early Golgi. That Sec34p can be found throughout the
Golgi together with the observation that CPY is missorted and
Kex2p is mislocalized in grd20 (sec34) mutants (8), suggests
that Sec34p may be a general transport factor that mediates
more than one vesicle trafficking event. Consistent with this
hypothesis, we find genetic interactions between sec34–2
and mutations that block ER to Golgi, Golgi, and post-Golgi
Stage-specific mutants that block membrane traffic between
the ER and Golgi complex accumulate an extensive network
of dilated ER (5,21,24,34). In contrast, mutants that harbor
defects in components of the Sec34p/Sec35p complex accumulate modest amounts of ER and the lumen does not
appear to be as dilated as other known ER-accumulating mutants (21). Mutations in subunits of the Sec34p/Sec35p complex also lead to the accumulation of a variety of membranes
including structures that resemble aberrant Golgi. Thus, it is
possible that the morphological and kinetic affects observed
on ER to Golgi traffic in sec34, sec35, and sgf1 mutants may
be an indirect consequence of blocking the flow of traffic
through the Golgi. Similar phenotypes have been reported
before for mutations in vti1 (35). The yeast v-SNARE Vti1p
Traffic 2001; 2: 820–830
mediates multiple transport events, including retrograde traffic to the cis-Golgi (35–37). Like sec34, some alleles of vti1
do not block secretion.
Thus far only one complex, TRAPP II, has been implicated in
the tethering of Golgi vesicles (5). As traffic through the Golgi
involves multiple tethering events, it is likely that other tethers
are involved in these events. An attractive possibility is that
the Sec34p/Sec35p complex is one of these tethers. Interestingly, other mutants that disrupt Golgi traffic have been
shown to display genetic interactions with mutations that
block ER to Golgi, Golgi and post-Golgi transport (5,34,38).
An alternate explanation of the data is that Sec34p is present
in two different complexes that tether different classes of
transport vesicles. This has recently been reported for the
TRAPP complex which is found in two forms, TRAPP I and
TRAPP II (5). The two complexes share seven subunits
(Bet5p, Trs20, Bet3p, Trs23p, Trs31p, Trs33p and Trs85p),
while three subunits (Trs65p, Trs120p and Trs130p) are
unique to TRAPP II. Mutational analysis and in vitro transport
studies have revealed that these two complexes mediate different transport steps. The two forms of TRAPP can be separated by fractionating cytosol on a Superdex-200 column.
However, using gel filtration conditions that have allowed us
to separate TRAPP I from TRAPP II, and by anion exchange
chromatography, we only detect one form of the Sec34p/
Sec35p complex. Thus, this hypothesis seems unlikely but
we cannot definitively rule it out. Additional experiments will
be needed to determine if the Sec34p/Sec35p complex mediates multiple tethering events.
Materials and Methods
Strain constructions and growth conditions
The strains used in this study are listed in Table 3. SFNY1091 and
SFNY1092 were constructed by transforming SFNY772 with the Och1HA
plasmid (pOH) (39), or a plasmid containing Sec7p-GFP (pSSEC7EGFPx3; a gift from Benjamin Glick). pSSEC7-EGFPx3 was first digested
with SpeI and the linear fragment was transformed into SFNY772.
SFNY1006 was constructed as previously described (6). The SGF1 gene
was disrupted by replacing the ORF (bp 1–2586) with the S. pombe his5π
gene (17). Briefly, a hybrid sequence containing the his5π gene flanked by
part of SGF1 was amplified by polymerase chain reaction (PCR). This product was then transformed into wild-type diploid cells (NY1523). Hisπ transformants (SFNY1033) were selected and the disruption was confirmed by
PCR before the strain was subjected to tetrad analysis. SFNY1031 was
constructed as follows: the amino terminal portion (1–348 bp) of the SGF1
gene was amplified by PCR, digested with BamHI and HindIII, and inserted into the same sites of pNB529, placing the amino terminal portion
of SGF1 behind the GAL1 promoter. The plasmid was digested with MscI,
which cuts within the SGF1 gene, and the linear fragment was transformed
into NY604 to construct a strain in which full-length SGF1 was placed
under the control of the regulatable GAL1 promoter. Leuπ transformants
were selected and the chromosomal fusion of SGF1 to the GAL promoter
was confirmed by PCR. Cells expressing Sgf1p from the GAL1 promoter
were grown to an OD599 of 18 in YP medium containing 0.5% galactose
and 2% raffinose and then inoculated into YP medium containing 2% glucose.
Kim et al.
Table 3: Yeast strains used in this study
MATa sec1–1 ura3–52
MATa sec13–1 ura3–52
MATa sec21–1 ura3–52
MATa GALπ ura3–52 leu2–3, 112
MATa sec2–41 leu2–3, 112
MATa sec4–8 leu2–3, 112
MATa sec23–1 ura3–52
MATa sec23–1 leu2–3, 112
MATa sec23–1 leu2–3, 112
MATa sec18–1 leu2–3, 112
MATa/a GALπ/GALπ ura3–52/
ura3–52 leu2–3, 112/leu2–3,
MATa bet3–1 leu2–3, 112
MATa ura3–52 leu2–3,
112 his3-D200 lys2–801 arf1D::URA3
MATa sec34–2 lys2–801
MATa sec34–2 ura3–52 leu2–3,112
MATa sec34–2 ura3–52
MATa ura3–52 SEC34
(with three c-myc tags)
MATa sec34–1 ura3–52
MATa sec35–1 ura3–52
MATa ura3–52 SEC35 (with three c-myc tags)
MATa ura3–52 leu2–3,
112 his3-D200 trs130ts2
MATa sec27–1 ura3–42
MATa ret2–1 ura3–52 leu2–3, 112
his3-D200 lys2–801 suc2-D9
MATa sec24–1 ura3–52 leu2–3, 112
MATa sec31–1 ura3–52 leu2–3, 112
MATa ura3–52 L-A-o SGF1
(with three c-myc tags)
MATa ura3–52 YML068W
(with three c-myc tags)
MATa GALπ ura3–52 leu2–3, 112
MATa/a GALπ/GALπ ura3–52/
ura3–52 leu2–3, 112/leu2–3, 112
MATa ura3–52 SEC34
(with three c-myc tags)
pOH(OCH1-HA URA3 CEN)
MATa ura3–52 SEC34
(with three c-myc tags)
Cell labeling and immunoprecipitation
Secretion of [35S] radiolabeled proteins was measured by a modification
of the method of Gaynor and Emr (10). Cells grown at 25 æC to early exponential phase in minimal medium were harvested, resuspended in 400 ml
of minimal medium and preshifted to 37 æC for 20 min before 2 OD599
units were radiolabeled for 15 min at 37 æC with 150 mCi of [35S] ProMix
(Amersham, England). The radiolabeled cells (380 ml) were transferred to
an ice-cold tube containing 38 ml of 500 mM NaN3/500 mM NaF. To separ-
ate medium from cells, cell suspensions were centrifuged at 14 000 g for
1 min. The supernatant (300 ml) was transferred to a tube containing 19.5 ml
of 100% TCA and 1.5 ml of 2% deoxycholate, and incubated for at least 1 h
on ice. The TCA-precipitated proteins were washed three times with icecold acetone, air-dried, resuspended in SDS sample buffer and resolved
on an 8% SDS-polyacrylamide gel.
For pulse-chase analysis of CPY, wild-type and SFNY1031 were grown at
25 æC in YP medium containing 0.5% galactose and 2% raffinose to an
OD599 of 18 and then inoculated into YP medium containing 2% glucose
(final OD599 Ω 0.03). After a 12-h incubation at 25 æC, cells (10 OD599 units)
were pelleted, washed and resuspended in 3.5 ml of minimal medium containing 2% glucose. The cells were preincubated for 20 min at 30 æC and
then pulse-labeled with 250 mCi of [35S] ProMix. Following a 4-min pulse
labeling, 700 ml of cells were removed (0 min chase) and cold methionine
and cysteine were added to a final concentration of 10 mM. After 5, 10 and
30 min, samples were transferred to a cold tube containing NaN3 and NaF
(final concentration of 10 mM) and CPY was immunoprecipitated as described previously (34).
The precipitation of myc tagged proteins was performed as described previously (6).
Immunofluorescence and electron microscopy
Five OD units of cells of an overnight culture that was grown to
OD599 Ω 1 were centrifuged, fixed and converted to spheroplasts as described previously (40). Spheroplasts were adhered to poly-L-lysine-coated
slides for 15 min before they were washed three times with PBST (PBS
buffer, pH 7.4, 0.1% tween-20, 10 mg/ml BSA and 0.1% NaN3). Lysed
spheroplasts were incubated overnight with 30 ml primary antibody in a
humid chamber at 4 æC, washed 10 times in PBST and then incubated with
30 ml of secondary antibody for 90 min in a humid chamber at 4 æC. Double
immunofluorescence of Sec34p-myc and Och1p-HA was achieved by incubating with a mixture of mouse anti-myc monoclonal antibody 9E10
(1 : 100 dilution; Babco, CA, USA) and rabbit anti-HA polyclonal antibody
(1 : 500 dilution; Babco, CA, USA) and then with a mixture of donkey antimouse IgG antibody conjugated to Cy3 (1 : 1000 dilution; Jackson Immunoresearch Laboratories, PA, USA) and donkey anti-rabbit IgG antibody
conjugated to FITC (1 : 1000 dilution; Jackson Immunoresearch Laboratories, PA, USA). Double immunofluorescence of Sec34p-myc and Sec7pGFP required primary incubation with mouse anti-myc monoclonal antibody 9E10 (1 : 100 dilution) followed by secondary incubation with donkey
anti-mouse IgG antibody conjugated to Cy3 (1 : 700 dilution). Following the
antibody incubations, the slides were washed 30 times with PBST and
three times with PBS (pH 7.4) before the coverslips were applied with
Fluoromount G (Southern Biotechnology Associates, AL, USA). Samples
were viewed under a fluorescence microscope, and images were obtained
using a charge-coupled device camera and manipulated with Openlab
software (Improvision, MA, USA). Adobe Photoshop and Illustrator software were used to produce figures.
To quantify the colocalization of Sec34p-myc and Och1p-HA, 316 Sec34pmyc-positive puncta and 340 Och1p-HA-positive puncta were scored in
73 cells by eye after image capture. A total of 251 puncta stained for both
proteins. In order to quantify the colocalization of Sec34p-myc and Sec7pGFP, 251 Sec34p-myc-positive puncta and 257 Sec7p-GFP-positive
puncta were scored in 70 cells. A total of 70 puncta stained for both proteins.
For electron microscopy, wild-type and SFNY1031 were incubated for 12 h
at 25 æC in YP medium containing 2% glucose, while sec34–2 and
sec35–1 mutants were shifted to 38 æC for 2 h, and processed as described before (34).
Traffic 2001; 2: 820–830
Screen for high-copy suppressors of sec35–1
To isolate high-copy suppressors of sec35–1, the sec35–1 mutant
(SFNY 816) was transformed with a yeast genomic high-copy library
and Uraπ transformants were selected at 25 æC and then screened for
growth at 37 æC. Of the 77 000 transformants screened, 81 were found
to grow at 37 æC. Plasmids were retrieved from these 81 transformants,
amplified in E. coli and reintroduced into sec35–1 to confirm that the
suppression was plasmid dependent. Plasmids containing the SEC34
and SEC35 structural genes were identified by restriction analysis.
SEC35 has a unique 717 bp PshAI and BstEII fragment, while SEC34
contains a unique 2.2 kb BstEII and BglII fragment. Those plasmids that
did not contain SEC34 or SEC35 were sequenced. A combination of restriction analysis and DNA sequencing revealed that the 81 suppressors
contained five different regions of genomic DNA. Two of the genomic
regions of DNA contained the SEC35 and SEC34 structural genes. The
third was found to contain the SBP1 and RPL8A genes. The insert
(8.14 kb) in the fourth genomic region was digested with KpnI and SpeI
and the resulting 3.75 kb fragment was inserted into the KpnI-SpeI sites
of pRS426. It contained a hypothetical open reading frame (YPR105C)
that conferred suppression. We named this ORF SGF1 (suppressor gene
of sec35–1). The insert (8.65 kb) in the fifth genomic region of DNA
was digested with AseI and the resulting two fragments were inserted
into the same site of YEp24. One (2.5 kb) of these fragments contained
YML068 which retained full suppression activity.
Antibody to Sgf1p was prepared against an MBP-Sgf1p recombinant form
of the protein. To construct the plasmid encoding MBP-Sgf1p, the SGF1
ORF was amplified by PCR, placing a BamHI site before the start codon
and a HindIII site after the stop codon (sense primer [5ƒª3ƒ], CAT CTA
GGA TCC ATG GAA GGG CAA AAA TCG AAT G; antisense primer [5ƒ-3ƒ],
CAT CTA AAG CTT CTA TTA CTG TGT TCT ATC AAT CTT C). The PCR product was digested with BamHI and HindIII and then ligated into the same
sites of pMAL-c2 (New England Biolabs, MA, USA). The construct was
transformed into BL21 cells (Novagen, WI, USA), and the fusion protein
was purified as described previously (6).
The ORFs of SGF1, YML068W, SEC34 and SEC35 were amplified by PCR,
sequenced, digested with NcoI and BamHI, and inserted into the same
sites of pACT2 (fish) or pAS1-CYH2 (bait) (16). Two different constructs
were introduced into the yeast strain Y190, and Leuπ and Trpπ transformants were selected. The transformants were smeared onto filter paper
(Whatmans, England) and the filter paper was frozen in liquid nitrogen,
thawed, and overlaid onto a filter soaked with Z buffer (0.1 M sodium phosphate buffer, pH 7.0, 0.01 M KCl, 1 mM MgSO4.7H2O, and 30 mM b-mercaptoethanol) containing X-gal (Boehringer Mannheim, IN, USA). After an
overnight incubation at 30 æC, ⬃300 colonies were screened for a change
from white to blue color. If the majority of the colonies turned blue, it was
designated as a positive interaction.
Partial purification of the Sec34p/Sec35p/Sgf1p complex
A yeast lysate was prepared as described before (6), except the cytosol
was centrifuged at 84 000 g for 1 h. A total of 5 mg of protein was loaded onto a 25-ml Superdex-200 gel filtration column, pre-equilibrated
with buffer A (25 mM Tris, pH 7.6, 100 mM KCl, 1 mM DTT, 1¿Pic), at a
flow rate of 0.4 ml/min. Fractions (1 ml) were collected and the peak of
Sec34p, Sec35p and Sgf1p, as determined by Western blot analysis,
was loaded onto a 1-ml Mono Q column at a flow rate of 1 ml/min.
The column was washed with 5 column volumes of buffer A and the
protein was eluted with a linear gradient (15 ml) of buffer A to buffer B
(same as buffer A with 500 mM KCl). Proteins were detected by Western
Traffic 2001; 2: 820–830
We thank Yueyi Zhang, Elaine Downie and Lora Cavallo for excellent technical assistance and Benjamin Glick for providing the Sec7p-GFP plasmid.
We are grateful to Jemima Barrowman for comments on the manuscript
and to Joyce Anquillare for help in the preparation of the manuscript. D.-W.
K., T. M. and M.S. are supported by the Howard Hughes Medical Institute.
1. Guo W, Sacher M, Barrowman J, Ferro-Novick S, Novick P. Protein complexes in transport vesicle targeting. Trends Cell Biol 2000;10:251–255.
2. Ferro-Novick S, Jahn R. Vesicle fusion from yeast to man. Nature
3. Hanson PI, Roth R, Morisaki H, Jahn R, Heuser JE. Structure and conformational changes in NSF and its membrane receptor complexes
visualized by quick-freeze/deep-etch electron microscopy. Cell
4. Barrowman J, Sacher M, Ferro-Novick S. TRAPP stably associates
with the Golgi and is required for vesicle docking. EMBO J 2000;19:
5. Sacher M, Barrowman J, Wang W, Horecka J, Zhang Y, Pypaert M,
Ferro-Novick S. TRAPP I implicated in the specificity of tethering in
ER-to-Golgi transport. Mol Cell 2001;7:433–442.
6. Kim DW, Sacher M, Scarpa A, Quinn AM, Ferro-Novick S. High-copy
suppressor analysis reveals a physical interaction between Sec34p
and Sec35p, a protein implicated in vesicle docking. Mol Biol Cell
7. VanRheenen SM, Cao X, Sapperstein SK, Chiang EC, Lupashin VV,
Barlowe C, Waters MG. Sec34p, a protein required for vesicle
tethering to the yeast Golgi apparatus, is in a complex with Sec35p.
J Cell Biol 1999;147:729–742.
8. Spelbrink RG, Nothwehr SF. The yeast GRD20 gene is required for
protein sorting in the trans-Golgi network/endosomal system and for
polarization of the actin cytoskeleton. Mol Biol Cell 1999;10:4263–
9. Robinson JS, Klionsky DJ, Banta LM, Emr SD. Protein sorting in Saccharomyces cerevisiae: isolation of mutants defective in the delivery
and processing of multiple vacuolar hydrolases. Mol Cell Biol
10. Gaynor EC, Emr SD. COPI-independent anterograde transport: cargoselective ER to Golgi protein transport in yeast COPI mutants. J Cell
11. Franzusoff A, Redding K, Crosby J, Fuller RS, Schekman R. Localization of components involved in protein transport and processing
through the yeast Golgi apparatus. J Cell Biol 1991;112:27–37.
12. Rossanese OW, Soderholm J, Bevis BJ, Sears IB, O’Connor J, Williamson EK, Glick BS. Golgi structure correlates with transitional endoplasmic reticulum organization in Pichia pastoris and Saccharomyces cerevisiae. J Cell Biol 1999;145:69–81.
13. Suvorova ES, Kurten RC, Lupashin VV. Identification of a human orthologue of Sec34p as a component of the cis-Golgi vesicle tethering
machinery. J Biol Chem 2001;276: 22810–22818.
14. Kang HY, Yeh S, Fujimoto N, Chang C. Cloning and characterization of
human prostate coactivator ARA54, a novel protein that associates
with the androgen receptor. J Biol Chem 1999;274:8570–8576.
15. Fields S, Song OK. A novel genetic system to detect protein–protein
interactions. Nature 1989;340:245–246.
16. Durfee T, Becherer K, Chen PL, Yeh SH, Yang Y, Kilburn AE, Lee WH,
Elledge SJ. The retinoblastoma protein associates with the protein
phosphatase type 1 catalytic subunit. Genes Dev 1993;7:555–569.
17. Longtine MS, McKenzie A, III, Demarini DJ, Shah NG, Wach A, Brachat A, Philippsen P, Pringle JR. Additional modules for versatile and
Kim et al.
economical PCR-based gene deletion and modification in Saccharomyces cerevisiae. Yeast 1998;14:953–961.
Wuestehube LJ, Duden R, Eun A, Hamamoto S, Korn P, Ram R,
Schekman R. New mutants of Saccharomyces cerevisiae affected in
the transport of proteins from the endoplasmic reticulum to the Golgi
complex. Genetics 1996;142:393–406.
VanRheenen SM, Cao X, Lupashin VV, Barlowe C, Waters MG.
Sec35p, a novel peripheral membrane protein, is required for ER to
Golgi vesicle docking. J Cell Biol 1998;141:1107–1119.
Stevens T, Esmon B, Schekman R. Early stages in the yeast secretory
pathway are required for transport of carboxypeptidase Y to the vacuole. Cell 1982;30:439–448.
Novick P, Field C, Schekman R. Identification of 23 complementation
groups required for post-translational events in the yeast secretory
pathway. Cell 1980;21:205–215.
Botstein D. Why study the cytoskeleton in yeast? Harvey Lect
Salminen A, Novick PJ. A ras-like protein is required for a post-Golgi
event in yeast secretion. Cell 1987;49:527–538.
Kaiser CA, Schekman R. Distinct sets of SEC genes govern transport
vesicle formation and fusion early in the secretory pathway. Cell
Letourneur F, Gaynor EC, Hennecke S, Démollière C, Duden R, Emr
SD, Riezman H, Cosson P. Coatomer is essential for retrieval of dilysine-tagged proteins to the endoplasmic reticulum. Cell 1994;79:
Gaynor EC, Graham TR, Emr SD. COPI in ER/Golgi and intra-Golgi
transport: do yeast COPI mutants point the way? Biochim Biophys
Uetz P, Glot L, Cagney G, Mansfield TA, Judson RS, Knight JR, Lockshon D, Narayan V, Srinivasan M, Pochart P, Qureshi-Emill A, Li Y,
Godwin B, Canover D, Kalbfleisch T et al. A comprehensive analysis
of protein–protein interactions in Saccharomyces cerevisiae. Nature
Ito T, Chiba T, Ozawa R, Yoshida M, Hattori M, Sakaki Y. A comprehensive two-hybrid analysis to explore the yeast protein interactome.
29. Pfeffer S. Transport-vesicle targeting: tethers before SNAREs. Nature
Cell Biol 1999;1:E17–E22.
30. Wang W, Sacher M, Ferro-Novick S. TRAPP stimulates guanine nucleotide exchange on Ypt1p. J Cell Biol 2000;151:289–295.
31. Cao X, Ballew N, Barlowe C. Initial docking of ER-derived vesicles
requires Uso1p and Ypt1p but is independent of SNARE proteins.
EMBO J 1998;17:2156–2165.
32. Yamakawa H, Seog DH, Yoda K, Yamasaki M, Wakabayashi T. Uso1
protein is a dimer with two globular heads and a long coiled-coil tail.
J Struct Biol 1996;116:356–365.
33. Allan BB, Moyer BD, Balch WE. Rab 1 recruitment of p115 into a
cis-SNARE complex: programming budding COPII vesicles for fusion.
34. Rossi G, Kolstad K, Stone S, Palluault F, Ferro-Novick S. BET3 encodes
a novel hydrophilic protein that acts in conjunction with yeast
SNAREs. Mol Biol Cell 1995;6:1769–1780.
35. Fischer von Mollard G, Northwehr S, Stevens T. The yeast v-SNARE
Vti1p mediates two vesicle transport pathways through interactions
with the t-SNAREs Sed5p and Pep12. J Cell Biol 1997;137:1511–
36. Lupashin VV, Pokrovskaya ID, McNew JA, Waters MG. Characterization of a novel yeast SNARE protein implicated in Golgi retrograde
traffic. Mol Biol Cell 1997;8:2659–2676.
37. Fischer von Mollard G, Stevens T. The Saccharomyces cerevisiae vSNARE Vti1p is required for multiple membrane transport pathways
to the vacuole. Mol Biol Cell 1999;10:1719–1732.
38. Finger FP, Novick P. Synthetic interactions of the post-Golgi sec mutants of Saccharomyces cerevisiae. Genetics 2000;156:943–951.
39. Harris SL, Waters MG. Localization of a yeast early Golgi mannosyltransferase, Och1p, involves retrograde transport. J Cell Biol
40. Shim J, Newman AP, Ferro-Novick S. The BOS1 gene encodes an
essential 27-kD putative membrane protein that is required for vesicular transport from the ER to the Golgi complex in yeast. J Cell Biol
Traffic 2001; 2: 820–830