(Received for publication, August 16, 1996, and in revised form, November 6, 1996)
From the Center for Molecular Biology, University of
Queensland, Brisbane 4072, Australia and the § Institute
of Molecular Biology, University of Oregon,
Eugene, Oregon, 97403-1229
Our understanding of lysosomal biogenesis and general vesicular transport in animal cells has been greatly enhanced by studies of vacuolar biogenesis in yeast. Genetic screens have identified a number of proteins that play direct roles in the proper sorting of vacuolar hydrolases. These include t-SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) proteins and Sec1p-like proteins, which have recently been implicated as key regulators of vesicle fusion. In this study we have extended these observations in yeast and have isolated and characterized a novel member of the Sec1p-like family of proteins from mammalian cells, mVps45. mVps45 shares a high level of identity with the Saccharomyces cerevisiae Sec1p-like protein Vps45p that is believed to function with the t-SNARE Pep12p in the fusion of Golgi-derived transport vesicles with a prevacuolar compartment. We found that mVps45 is a ubiquitously expressed peripheral membrane protein that localized to perinuclear Golgi-like and trans-Golgi network compartments in Chinese hamster overy cells. We found that mVps45 could bind specifically to yeast Pep12p and to the mammalian Pep12p-like protein, syntaxin 6, in vitro.
The interaction between a v-SNARE in a transport vesicle and a t-SNARE in the target membrane guides the specificity of distinct vesicular transport events in eukaryotic cells. Some of the key regulatory proteins that appear to modulate the interaction between v-SNARE proteins and t-SNARE proteins are members of the Sec1p-like family of proteins (1, 2). In vitro, Sec1p-like proteins bind to t-SNARE proteins and inhibit the binding of the t-SNARE with its cognate v-SNARE (3-6). Exactly how this interaction regulates fusion events in vivo is unknown; however, in vivo, loss of Sec1p-like function results in the accumulation of transport vesicles that are unable to properly fuse with their prospective acceptor membrane (7).
Within the family of Sec1p-like proteins, each member appears to interact with a specific subset of t-SNARE proteins (8-12). Several isoforms of the Sec1p-like protein Unc-18p from Caenorhabditis elegans (13) have been described in mammalian cells (Munc-18a, b, c) (3, 5, 14-17). Munc18a was first identified in brain, and biochemical studies revealed that Munc18a can bind to syntaxin 1A, a neuronal plasma membrane t-SNARE, and in so doing decrease the avidity of binding between syntaxin 1A and synaptobrevin, its cognate v-SNARE in vitro. Additional Munc18 homologs have been identified in mammalian cells, and they have a unique binding specificity with respect to different plasma membrane syntaxin homologs (10). Four Sec1p-like proteins have been found in yeast that function in diverse vesicle fusion events throughout the secretory pathway by interacting specifically with different t-SNARE proteins. These Sec1p-like proteins function in endoplasmic reticulum to Golgi transport (Sly1p) (9, 18), Golgi to plasma membrane transport (Sec1p) (19-21), and Golgi to vacuole transport (Vps45p and Vps33p) (22-26).
Of the known Sec1p-like proteins that function in Saccharomyces cerevisiae, Vps45p has been shown to participate in the delivery of hydrolases to the vacuole (22-24). Loss of function of Vps45p results in the accumulation of transport vesicles (~60 nm) that contain both Golgi proteins and vacuolar proteins, suggesting that Vps45p catalyzes the fusion of Golgi-derived vesicles with a post-Golgi/prevacuolar compartment (23).1 Vps45p is believed to interact with the yeast t-SNARE protein (syntaxin), Pep12p (28, 29). Mutations in PEP12 also result in the accumulation of transport vesicles, and recent fractionation experiments indicate that the bulk of Pep12p resides in a prevacuolar compartment (30).1
Recently, these observations have been extended to other cells types. A Pep12p homolog has been identified in Arabidopsis, which shows 32% identity and can complement pep12 mutations in yeast. Also, a t-SNARE related to Pep12p (syntaxin 6) has been identified by expressed sequence tag cloning and is localized to the Golgi apparatus of FAO cells (31). Given the close parallels of the secretory system between yeast and animal cells, we sought to identify proteins that may control intracellular trafficking events in mammalian cells using clues from work in yeast. In the present study we have identified a mammalian homolog of the Sec1p-like protein, Vps45p. We show that this mammalian Vps45p, mVps45, is localized in a Golgi-like pattern with a distribution similar to that of yeast Pep12p when expressed in a mammalian cell. Furthermore, mVps45 binds to Pep12p and syntaxin 6 in vitro but not to other t-SNARE proteins.
Enzymes used in DNA manipulations were from New England Biolabs (Beverly, MA), Boehringer Mannheim, Life Technologies, Inc., or U. S. Biochemicals. Radioactive nucleotides, nylon membranes (Hybond-N+), horseradish peroxidase-conjugated goat anti-rabbit IgG, and enhanced chemiluminescence (ECL) detection kits were from Amersham Corp. All tissue culture reagents and LipofectamineTM reagent were from Life Technologies, Inc. with the exception of fetal calf serum, which was from Commonwealth Serum Laboratories CSL (Sidney, Australia). BCA protein reagent was from Pierce. Glutathione-agarose beads were from Sigma. FITCconjugated2 sheep anti-rabbit secondary antibody was purchased from the Binding Site (Birmingham, UK). Fixed Staphylococcus aureus cells (IgGsorb) were obtained from the Enzyme Center (Malden, MA). [35S]Methionine label was from Amersham Corp. Oxalyticase was from Enzogenetics (Corvallis, OR).
Cloning of VPS45A BLAST search of the Washington
University/Merck express tagged sequence data base showed the presence
of sequences with a high degree of identity with yeast VPS45
within a human fetal brain cDNA library (32). These sequences
(GenBankTM accession numbers R10533[GenBank], R12336[GenBank], and R52780[GenBank]) were aligned,
and the resulting contiguous sequence was used to predict a series of
oligonucleotides that could be used to amplify the corresponding
cDNA fragment. The oligonucleotide sequences were:
CCAGGCCATGGTCCACGAACTACTAGGC (F2),
GAG
TCAGAGGTTGAGCAAGAACTGGCC (F3), and
CGC
CCAGGGTTTCCATGTAGGAAAAGGT (F8).
Amplification of a human cDNA library with oligonucleotides F2 and
F8 yielded a ~750-bp fragment from which a ~350-bp fragment could
be amplified with the oligonucleotides F3 and F8. To clone the
full-length mouse VPS45 homolog the F2/F8 fragment was
radiolabeled with random hexamer primers (Promega Corporation, WI) and
used to screen a random-primed mouse 3T3-L1 adipocyte cDNA library.
A total of 20,000 plaques were screened, and the four positive clones
that were isolated after sequential purifications were subcloned into pBluescript II SK, and the inserts were sequenced as described previously (16).
DNA
manipulations and DNA-mediated transformation were performed by routine
procedures (34). pM45-A, containing a full-length cDNA of+ mouse
VPS45 (mVPS45) in Bluescript was sequenced
(GenbankTM accession number U66865[GenBank]). This clone contained 680 bp of
the 5-untranslated region and 192 bp of 3
-untranslated region. pM45-B
was made by subcloning a PCR fragment encompassing the coding region of
mVPS45 but lacking the 680 bp of 5
-untranslated region
using the oligonucleotides TGC
CAATTCGCCACCATGAATGTGG and
CGACCG
TCATAGATCTGCTCTTCTGTTTGCTGACCTTGAGG.
The HA epitope was inserted into mVps45 by first placing a
BamHI site into the 3 end of the mVPS45 open
reading frame just prior to the stop codon by overlapping PCR
amplification using the oligonucleotides
AACAGAAG
ATGAGATGGCAGT in combination with the
downstream T7 primer and TCAT
CTTCTGTTTGCTGACCT in
combination with the upstream primer (M4)
GACAGGATCCAGAGGAGTAAAGGTGTTGCTGAGAAG. The resulting M4/T7 PCR fragment
was digested using EcoRV and subcloned into Bluescript
KS+ in which the BamHI site had been previously
destroyed with T4 DNA polymerase. The resulting plasmid, pRCP100, was
used to subclone a BglII linker (encoding three copies of
the HA epitope: YPYDVPDYA) into the BamHI site to yield
pRCP101. The EcoRV fragment of pRCP101 was subcloned into
the EcoRV sites pM45-A to yield a full-length cDNA clone
(pRCP102) containing the triple HA epitope that was subsequently
verified by sequencing of both strands. The
NotI/ApaI fragment from pRCP102 was then
subcloned into the NotI/ApaI sites of the
expression vector pRcCMV-Neo (Clonetech) to yield pRCP103.
To make pRCP104, a fragment encoding the open reading frame of S. cerevisiae PEP12 gene (12 to +879) was amplified with the oligonucleotides CACGAG
CCATAATTGTGTTGAGATGTCGGAAGA
(P12-5) and CAC
ATATTTGACGACGTGTGTTGGTTTGGTTTAC
(P12-3) and subcloned into the
HindIII/XbaI sites of pRcCMV-Neo. The predicted
sequence of the plasmid was verified by sequencing with the P12-5 and
P12-3 primers. A Pep12p/GST fusion protein plasmid was made by PCR
amplifying a fragment encoding amino acids 1-264 of PEP12
with the oligonucleotides ATGG
TACGTACGTTTTTGGTACCGCAT
and CACA
CCATGTCGGAAGACGAATTTTT and subcloning the
fragment into the BamHI/EcoRI sites of pGEX-3X to
yield the plasmid pRCP39. A human Vps45p/GST fusion protein expression
plasmid was made by subcloning the PCR fragment encoding the C terminus
of human Vps45p using the oligonucleotides F2 and F8 into the
BamHI site of pGEX-3X to yield pRCP105. A mouse Vps45p/GST fusion protein expression plasmid was made by subcloning a PCR fragment
encoding amino acids 276-437 into the
BamHI/EcoRI sites of pGEX-3X to yield pRCP106.
The PCR fragment used to generate pRCP106 was made using the
oligonucleotides M4 and M5 (GACGAATTCTTCTCGCCGCTCTCCACTGCCATC). A
GST-syntaxin 6 fusion protein plasmid (pGSTsynt6) was made by subcloning a PCR fragment encoding amino acids 1-234 of rat syntaxin 6 into the BamHI/EcoRI sites of pGEX-3X. The
syntaxin 6 PCR fragment was amplified with the oligonucleotides
CACA
AGATGTCCATGGAGGACCCCTTCTTTG and
CAT
TACTGGCGCCGATCACTGGTCATGTG. The
BamHI/EcoRI insert of pGSTsynt6 was verified by
sequencing.
A yeast 2-µm plasmid (pRCP107) for the expression of mVPS45 was made by inserting the open reading frame of the HA epitope-tagged allele of mVPS45 into the XbaI/XhoI sites of the ADH1 promoter containing vector pUT102. The Xba/XhoI fragment was generated by PCR using the oligonucleotides GTCTCTAGAAATGAATGTGGTCTTTGCTGTG and CAGACTCGAGTTCTCGCCGCTCTCCACTGCC. This plasmid was verified by sequence analysis.
Production of Antiserum to Mammalian Vps45pRabbits were immunized with a glutathione S-transferase fusion protein containing a C-terminal portion of the human Vps45p made from the expression plasmid pRCP105. Rabbits were subsequently boosted with another GST fusion protein produced from the expression plasmid pRCP106 containing amino acids 276-437 of mouse Vps45p. The bacterial fusion proteins were purified over GST-agarose and eluted with 25 mM glutathione as reported previously (35). Rabbit immunizations were performed as described previously (36). For affinity purification of anti-mVps45 antibodies, the fusion protein from pRCP106 was covalently attached to Affi-Gel 10 (Bio-Rad) and used as described previously (36).
The anti-Pep12p antibody was made by immunizing rabbits to the Pep12/GST fusion protein produced with pRCP39. The same Pep12p/GST fusion protein was used to affinity purify the resulting antiserum after attachment to Affi-Gel 15 (Bio-Rad).
RNA Blot AnalysisTotal RNA was isolated from rat tissues,
3T3-L1 fibroblasts, and adipocytes by the guanidine isothiocyanate
procedure (37). Northern blots were probed with a mouse
VPS45 DNA fragment labeled with [-32P]dCTP
by random priming. The hybridization conditions were: 50% formamide,
5 × SSPE, 5 × Denhardt's solution, 0.1% SDS, and 100 mg/ml denatured herring sperm DNA at 42 °C for 16 h. The blot was washed with 1 × SSC and 0.1% SDS at 50 °C. The RNA blots
were also probed with a
-actin cDNA probe as an internal control
(16).
Chinese hamster ovary cells (CHO-K1) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, 1% nonessential amino acids, 100 units/liter penicillin and 100 mg/liter steptomycin at 37°C in an atmosphere of 5% CO2. For concurrent fractionation and immunolocalization experiments, cells were grown in 150-mm tissue culture plates containing five No 1. coverslips. 3T3-L1 fibroblasts were grown to confluency and differentiated into adipocytes as described previously (23).
The expression plasmids (pRCP105 and pRCP106) harboring the HA epitope-tagged mVPS45 and yeast PEP12 were transiently transfected into subconfluent CHO cells using the LipofectamineTM reagent, according to the manufacturer's instructions. Cells were harvested after 40 h, scraped in HEPES buffer (20 mM, pH 7.4) containing 1 mM EDTA and 250 mM sucrose, lysed using a 1-ml syringe and 27-gauge needle and centrifuged at 220,000 × g for 30 min in a TLA-100.3 rotor (Beckman, Palo Alto, CA) to pellet total cellular membranes. Membrane pellets were resuspended in Laemmli sample buffer and subjected to SDS-PAGE and immunoblotted using the affinity purified polyclonal antiserum specific for mammalian Vps45p or a monoclonal antibody specific for the HA epitope (Babco, Berkeley, CA).
Subcellular membrane fractions were prepared from 3T3-L1 adipocytes as
described previously (38). Four separate membrane fractions were
obtained designated as plasma membranes, low density microsomes, high
density microsomes, and mitochondria/nuclei. The protein content of all
fractions was measured using the BCA reagent (Pierce). A whole cell
yeast extract was made by glass bead lysis of a mid log culture of the
wild type haploid S. cerevisiae strain SF838-9D as
described previously (23).
Wild type (RPY10) and vps45 cells (RPY14) were
transformed with pRCP107 using the lithium acetate procedure. Ura+
transformatnts were then measured for CPY secretion using a
[35S]methionine pulse/chase protocol as described
previously (23).
Cells were fixed for 30 min at 25 °C in PBS containing 2% paraformaldehyde. Cells were then washed in PBS and then incubated for 5 min in PBS containing 0.2% Triton X-100 and 50 mM glycine. The cells were then immunolabeled with affinity purified anti-Pep12p (1:20 dilution) or anti-HA rabbit polyclonal antibodies. Double labeling was achieved with combining one of the above rabbit polyclonal primary antibodies with a monoclonal antibody generated to the TGN P200 coat protein kindly provided by Dr. J. Stow (University of Queensland, Brisbane, Australia) (39). Primary antibodies were detected with FITC-conjugated sheep anti-rabbit secondary antibody (25 µg/ml) and Texas Red-conjugated goat anti-mouse secondary antibody. Coverslips were mounted on glass slides and visualized with a 63×/1.40 Zeiss oil immersion lens using a Zeiss Axioskop fluorescence microscope (Carl Zeiss) equipped with a Bio-Rad argon laser confocal apparatus. Digital TIFF files were pseudo-colored with Adobe PhotoshopTM.
In Vitro Binding AssaysThe mVPS45 cDNA
(pM45A and pM45B) and Munc-18b cDNA in pBluescript KG (16) were
transcribed and translated in vitro with T3 RNA polymerase
using a coupled reticulocyte lysate system (Promega) supplemented with
[35S]methionine (Amersham Corp.). The mVPS45
cDNA in pBluescript KG (pM45A) did not translate efficiently until
the 5-untranslated region was removed as per plasmid pM45B.
Reticulocyte lysate reactions containing
[35S]methionine-labeled mVps45 or Munc-18b protein were
incubated for 1 h at 4 °C with GST, GST-syntaxin-3, GST-Pep12p,
or GST-syntaxin-6 in PBS containing 0.5% bovine serum albumin. The
GST-Pep12p fusion protein was produced in bacteria using the plasmid
pRCP39. The GST-syntaxin 3 fusion protein was produced using an
expression plasmid kindly provided by Dr. R. Scheller (Stanford
University). The concentration of the fusion proteins used in these
reactions was 25 µg/ml. Glutathione-agarose beads (15 µl) were
added and incubated for a further 30 min at 4 °C. The beads were
washed three times by brief centrifugation in PBS at 4 °C. Washed
beads were then incubated for 5 min with Laemmli sample buffer (20 µl) and subjected to SDS-PAGE. The gel was fixed, incubated with 1 M sodium salicylate/0.5% glycerol, dried, and subjected to
autoradiography.
To identify a
mammalian homolog of the S. cerevisiae VPS45 gene, we
conducted a BLAST search against the Washington University/Merck express tagged sequence data base (32). Three overlapping sequences were found (GenBankTM accession numbers R10533[GenBank], R12336[GenBank], and R52780[GenBank])
that described a human cDNA that coded for a protein with a high
level of identity to Vps45p. This human cDNA fragment was PCR
amplified and used to probe a random primed mouse 3T3-L1 adipocyte
cDNA library. Four overlapping clones were isolated and sequenced
to reveal a full-length cDNA clone of 2.5 kilobases (GenBankTM
accession number U66865[GenBank]). This cDNA encoded a protein of 571 amino
acids with a predicted molecular mass of 65 kDa (Fig.
1). To confirm the open reading frame, the full-length
cDNA (pM45-A) was transcribed and translated in vitro. SDS-PAGE analysis showed the translation product migrated at an appropriate molecular mass (~72 kDa) (data not shown).
Sequence comparison of mVps45. A 2.5-kilobase cDNA fragment encoding the mouse Vps45p (GenBankTM accession number U66865[GenBank]) was cloned by hybridization screening using a probe corresponding to a human homolog of VPS45 that was identified by express tagged sequencing of human cDNAs. A, the predicted amino acid sequence of mVps45 is shown aligned with the S. cerevisiae Vps45p (GenBankTM accession number U07972[GenBank]) and Vps45p homologs from S. pombe (GenBankTM accession number Q09805[GenBank]) and C. elegans (GenBankTM accession number U41030[GenBank]). B, the phylogenetic relationship of mVps45 with other members of the Sec1p-like protein family is shown.
A search of the GenbankTM data base using the BLAST program (32) revealed that mVps45 shows identity with a family of proteins that are related to Sec1p (1) and is most like Vps45p from S. cerevisiae (Fig. 1). The Sec1p family includes the yeast proteins Sec1p (19, 21), Sly1p (9, 40), Vps33p/Slp1p (25, 26), the C. elegans protein Unc-18p (41), the Drosophila melanogaster ROP protein (42), the recently described mammalian Munc-18a (3, 5, 14), Munc-18b (15, 16, 17, 44), and Munc-18c (16). With the exception of the S. cerevisiae Vps45p, mVps45 shares limited identity (17-20%) with other members of the Sec1p-like family of proteins (Sec1p, Vps45p, Vps33p, Sly1p, and Munc-18a, b, c). This low level of overall identity is a common feature among members of this protein family (1). Other data base searches showed that VPS45 homologs were also identified in Schizosaccharomyces pombe (GenBankTM accession number Q09805[GenBank]) and C. elegans (GenBankTM accession number U41030[GenBank]) (45). mVps45 is most related to the Vps45p-like proteins of S. cerevisiae (38% identity), S. pombe (43% identity), and C. elegans (49% identity). This level of identity among the Vps45p homologs is similar to the level of identity between the endoplasmic reticulum to Golgi yeast Sec1p-like protein, Sly1p, and the recently described mammalian analog of Sly1p, m-Sly1p (40).
Northern blot analysis was performed on total RNA samples isolated from
a variety of rat tissues (Fig. 2). A single band of ~2.7 kilobases was found using the full-length mVPS45
cDNA clone as a probe. Levels were highest in testis and brain;
however, longer exposures showed that mVPS45 message was
expressed in all tissues examined.
mVps45 Is a Peripheral Membrane Protein Associated with Golgi/Endosomal Membranes
In S. cerevisiae, Vps45p
associates in a saturable manner with membranes enriched in Golgi and
endosomal markers (22, 23). Therefore, we sought to characterize the
biochemical properties of mVps45. CHO cell homogenates were centrifuged
at 220,000 × g to generate whole cell membrane and
cytosol fractions. These fractions were then subjected to immunoblot
analysis using affinity purified antibodies specific to amino acids
276-437 of mVps45. Fig. 3A shows that the
anti-mVps45 antibody recognizes a ~72-kDa protein. The apparent
molecular mass of this protein was identical to that produced from an
in vitro translation using the mVPS45 cDNA
(see below). This 72-kDa band was effectively competed with the
mVps45-GST fusion protein (Fig. 3A) but not GST alone (data not shown), thus showing the specificity of the anti-mVps45 antibody. All of the mVps45 was recovered with the whole cell membrane fraction consistent with the behavior of yeast Vps45p. mVps45 was found in the
supernatant fraction under alkaline conditions indicating that mVps45
is a peripheral membrane protein (data not shown).
We then pursued subcellular fractionation in mouse 3T3-L1 adipocytes to characterize the localization of mVps45. We have previously used 3T3-L1 adipocytes to produce four well characterized membrane fractions namely, plasma membranes, low density microsomes, high density microsomes, and mitochondria/nuclei (38). The plasma membrane fraction is enriched in plasma membrane markers; the mitochondria/nucleus fraction is enriched in markers for the mitochondria and nuclei; the high density microsomes are enriched in endoplasmic reticulum markers; and the low density microsomes are enriched in Golgi/endosomal markers (38). As shown in Fig. 3B, the 72-kDa mVps45 protein was highly enriched in the low density microsome fraction consistent with the fractionation profile of Vps45p in S. cerevisiae cells (22, 23). An additional band was also observed in the mitochondria/nuclei fraction; however, this band migrated at a much lower apparent molecular mass and did not compete with the GST-mVps45 fusion protein (data not shown). Thus, this likely corresponds to a nonspecific interaction of the antibody. Further attempts to affinity purify the antiserum did not result in complete loss of this immunoreactivity. Therefore, to facilitate subsequent immunolocalization experiments we constructed an epitope-tagged version of mVps45.
The epitope-tagged mVPS45 was made by inserting a triple concatamer of an epitope derived from the heamagglutinin spike glycoprotein (YPYDVPDYA) prior to the termination codon. This site was chosen for the insertion of the HA epitope because insertion of this epitope into the S. cerevisiae Vps45p does not disrupt function.3 A protein of ~75 kDa was generated when the HA epitope-tagged mVPS45 cDNA was transiently transfected into CHO cells in addition to the endogenous 72-kDa mVps45 band (Fig. 3A).
Immunoblotting these fractions with the anti-HA antibody confirmed that this additional ~75-kDa band corresponded to the HA epitope-tagged Vps45p. Based on the immunoreactivity of the anti-mVps45 antibody, the level of the HA-tagged Vps45p was ~4-5% of that of the endogenous mVps45. Correspondingly, immunofluorescence localization studies performed on cells from the same culture plate of transfectants showed that only 3-4% of the cells expressed mVps45-HA (see below). Based on these results, it is likely that mVps45-HA was expressed to levels comparable with those of the endogenous mVps45 in individual cells.
The fractionation experiments shown in Fig. 3A also show that all of the HA epitope-tagged mVps45 protein was associated with the whole cell membrane fraction, consistent with the behavior of the endogenous mVps45. These data together with the localization studies (see below) argue that the epitope-tagged mVps45 behaves as the wild type mVps45 protein.
Expression of mVps45 in Yeast Inhibits Proper CPY SortingWe
then sought to determine if the mammalian VPS45 homolog
would have complementary function in yeast. The mVPS45 open
reading frame containing the HA epitope was subcloned into a yeast
expression vector behind the ADH1 promoter and transformed
into wild type yeast cells (RPY10) and vps45 (RPY14)
cells (23). Immunoblot analysis of whole cell extracts with anti-HA
antibodies showed that mVps45-HAp was expressed. S. cerevisiae cells carrying a disruption of the VPS45
gene (vps45
) secrete ~70% of the vacuolar protease CPY
in contrast to wild type cells that secrete ~2-4% (23). Secondary
to this defect in vacuolar protein sorting, vps45
cells
also display a growth defect at 37 °C (23). We observed no
functional complementation by expression of mVps45-HA because no
decrease in the amount of CPY secretion was observed when mVps45-HAp
was present in vps45
cells, nor did we detect a
correction of the temperature-sensitive growth defect. As an alternate
analysis, we reasoned that if mVps45 could interact with a subset of
proteins that the endogenous yeast Vps45p interacts with, we might see
a dominant-negative phenotype when mVps45-HAp was expressed in wild
type cells. Fig. 4A shows that there was a
significant level (~12-15%) of CPY secretion from wild type cells
harboring the mVPS45-HA expression plasmid. Only 2-3%
secretion of CPY was observed in wild type cells carrying the
expression vector alone.
In Vitro Binding of mVps45 with Yeast Pep12p and Syntaxin 6
One of the important functions of Sec1-like proteins is to
interact with t-SNARE proteins (syntaxins) (2, 46). Previous in
vitro studies have shown that Sly1p binds to Sed5p (6), that
Munc-18a and Munc-18b bind to syntaxins 1A, 2 and 3 (3-6, 10, 44), and
that Munc-18c binds to syntaxins 2 and 4 (10). Because of the proposed
relationship between the syntaxin-like protein Pep12p and Vps45p in
yeast, we investigated whether mVps45 could bind to Pep12p in
vitro. In addition, we also investigated whether a recently
identified t-SNARE protein, syntaxin 6, that shares ~35% identity to
Pep12p could specifically bind mVps45 (31, 47). In an in
vitro binding assay similar to those described previously (5),
[35S]methionine-labeled mVps45 and Munc-18b were
generated in in vitro transcription/translation coupled
reactions and incubated with a panel of glutathione
S-transferase fusion proteins containing the cytoplasmic
tails of either syntaxin 3, S. cerevisiae Pep12p, or rat
syntaxin 6 (Fig. 4B). The linearized mVPS45
plasmid, pM45B, gave a translation product of ~72 kDa consistent with
our immunoblot analysis (Fig. 3A). The Munc-18b plasmid
produced the expected translation product of 67 kDa (Fig. 4). In
agreement with previous studies (10), there was significant binding of
Munc-18b to GST-syntaxin 3 with only background levels of Munc-18b
binding to GST-Pep12p, GST-syntaxin 6, or GST alone (Fig.
4B). Consistent with the proposed relationship between
Pep12p and Vps45p in S. cerevisiae and the homology between
syntaxin 6 and Pep12p, we found that the mouse Vps45 protein bound
specifically to the GST-Pep12p and the GST-syntaxin 6 fusion proteins
compared with GST alone or GST-syntaxin 3 (Fig. 4).
Colocalization of mVps45 and Yeast Pep12p with the TGN Marker P200
To investigate the site of action of mVps45, we immunolocalized HA epitope-tagged mVps45 in CHO transfectants. The mVps45-HA protein was found in a perinuclear pattern that was reminiscent of the TGN (Fig. 5A). The specificity of the immunolabeling procedure can be seen by the absence of labeling in the majority of the cells that were presumably not transfected. This specificity was confirmed in cells transfected with vector alone (data not shown).
The labeling pattern observed for mVps45-HA was also observed using antibodies specific for the p200 coat protein (Fig. 5B). p200 has been shown to be a constituent of vesicles derived from the TGN and has been used as a specific marker of the TGN (39). Both mVps45 and p200 showed extensive colocalization as shown in Fig. 3A and confirmed by merging the confocal images digitally (data not shown).
Based on the interaction between Vps45p and Pep12p in yeast as well as the interaction detected in vitro between mVps45 and Pep12p (Fig. 4), we also localized the yeast Pep12p expressed in CHO cells. Similar localization experiments have been performed with the yeast cis-Golgi syntaxin protein Sed5p to demonstrate conservation of function and localization (48). In these studies, Sed5p localized to the same subcellular compartment (the cis-Golgi reticulum) as its mammalian counterpart (syntaxin 5) when expressed in animal cells. Important to these studies, the mammalian Pep12p-related protein (syntaxin 6) has been localized to the Golgi apparatus in a pattern similar to what we observed for mVps45-HAp (31). These data support the functional localization of mVps45 (a putative binding partner for syntaxin 6 or mammalian Pep12p) to the Golgi-like distribution observed in Fig. 5. However, despite the binding activity we detected in vitro between mVps45 and syntaxin 6, the relationship between syntaxin 6 function and Pep12p has been made solely on comparison of predicted protein sequence. Therefore, to explore the relationship between syntaxin 6 and yeast Pep12p, we expressed the yeast protein in CHO cells and determined whether it localized to Golgi membranes. The coding region of PEP12 was subcloned behind the CMV promoter, and the resulting plasmid, pRCP104, was transfected into CHO cells. Immunoblot analysis of whole cell membrane and cytosol fractions from transfectants confirmed the production of the 32-kDa Pep12 protein (Fig. 3C). Pep12p produced in CHO cells migrated with the same mobility in SDS gels as the endogenous Pep12p from whole cell yeast extract. A low but significant level of immunoreactivity was detected for a band of ~30 kDa in extracts from untransfected CHO cells. This band may correspond to an endogenous mammalian Pep12p homolog; however, further studies will be required to address this possibility. Immunolocalization experiments were performed on the same batch of CHO cells transfected with the PEP12 expression plasmid as were subjected to immunoblot analysis. Double immunolocalization experiments using the affinity purified anti-Pep12p antibody and the monoclonal anti-p200 antibody showed that like mVps45-HA, Pep12p was localized to Golgi-like structures (Fig. 5C). Pep12p showed extensive colocalization with the p200 protein, indicating that Pep12p was colocalized to a significant degree with the TGN.
The sorting of vacuolar hydrolases in the yeast S. cerevisiae has become an important genetic model system partly due to the insight it has given us into lysosomal biogenesis in animal cells (49). Current data support a model in which soluble vacuolar hydrolases such as CPY are recognized in the late Golgi apparatus by a receptor, Vps10p. This complex departs the Golgi apparatus and is delivered to a prevacuolar compartment where CPY dissociates from the receptor and recycles back to the Golgi. Soluble hydrolases as well as endocytosed proteins that are delivered to the prevacuolar compartment can then move on to the vacuole (28, 29). The same model can be used to describe how soluble lysosomal hydrolases are delivered to the lysosome by the mannose-6-phosphate receptor. Here proteins recognized in the TGN or at the plasma membrane are delivered to an endosomal/prelysosomal compartment. Following receptor/ligand uncoupling, the receptor recycles back to the TGN, whereas lysosomal hydrolases move on to the lysosome (50). Many questions remain surrounding the exact itinerary of various TGN/endocytic proteins as well as what signals, processes, and specific protein interactions are required for proper sorting in both the mammalian and yeast cell system. Because homologous proteins are involved in both systems, it is possible to apply data directly from one system to the other. Exactly how well these related membrane trafficking pathways mirror each other must be determined by characterizing and manipulating the protein machinery that controls these processes.
We describe the cloning and characterization of a new Sec1p-like protein in mammalian cells. Mammalian Vps45p (mVps45) shares a high level of homology with Vps45p, a Sec1p-like protein found in S. cerevisiae that is required for proper delivery of proteins to the vacuole. Genome sequencing efforts in S. pombe and C. elegans have also described analogs of Vps45p on the basis of predicted protein sequences (45). Exactly what membrane trafficking steps the various Vps45 proteins catalyze and with what proteins they interact are central questions. Like Sec1p itself, rapid loss of Vps45p function in yeast results in the accumulation of transport vesicles (23). Mutations in several other Class D VPS genes also result in the accumulation of small vesicles within the cytoplasm (52). These genes include the syntaxin homolog Pep12p (30), the Rab5 homolog, Vps21p (53), and a newly described protein, Vps9p (27). Based on these data Vps45p has been proposed to work together with the syntaxin-like protein Pep12p to affect vesicle fusion (22, 23, 28-30).
We also obtained evidence for the association of Vps45p with Pep12p by demonstrating that mVps45 could bind to a Pep12p-GST fusion protein in vitro (Fig. 4). Significant levels of binding were obtained with mVps45 and the yeast Pep12p-GST fusion protein, whereas no binding was observed to syntaxin 3 or other mammalian syntaxins (1A, 1B, 2, and 3; data not shown). We also found that mVps45 could specifically bind to the cytoplasmic domain of a novel t-SNARE, syntaxin 6, which has been proposed to represent the mammalian homolog of Pep12p based on sequence homology (31). The interaction between mVps45 and syntaxin 6 was quantitatively less than that observed between other Munc-18/syntaxin isoforms. Although we cannot exclude the possibility that syntaxin 6 may not be a physiological binding partner for mVps45, this difference may reflect the need for additional regulatory proteins. Together these data indicate that the functional relationships of Vps45p have been conserved in eukaryotic cells; a view that is supported by our observation that expression of mVps45 in yeast cells interfered with the vacuolar protein sorting machinery, implying that the mVps45 protein could interact with a subset of proteins that the endogenous Vps45p does.
Further support for the interaction between Vps45p homologs with Pep12p homologs in vivo comes from the observation that mVps45 and yeast Pep12p are both localized to a significant degree to the Golgi and TGN in CHO cells. We also observe that the yeast Pep12p protein is localized to a significant degree to the TGN, an observation that again further supports a model in which Vps45p and Pep12p function in the same trafficking step and a high degree of functional conservation. Furthermore, the localization of mVps45 and Pep12p is remarkably similar to that observed for syntaxin 6 (31), as one might expect if syntaxin 6 is the functional t-SNARE for mVps45 in vivo. One interesting question that follows is to determine how either syntaxin 6 or Pep12p achieves this localization in the Golgi apparatus.
The larger issue remains, however, that the localization of mVps45 and syntaxin 6 to Golgi-like structures is in some respects at odds with its proposed function in yeast where functional studies as well as fractionation and immunofluorescence data are consistent with Vps45p and Pep12p functioning in prevacuolar/endosomal compartments (30).3 Although Pep12p is believed to function as the t-SNARE (syntaxin) for a prevacuolar/endosomal compartment, a definitive localization to such a compartment has yet to be performed, making it possible that Pep12p could reside, perhaps nonexclusively, in an analogous TGN-type compartment in yeast. Likewise, the function and exact location of mVps45 and syntaxin 6 must be elaborated by further studies. Another possibility is that the functional compartmentalization in yeast is somewhat different than in mammalian cells. Perhaps a much more intimate connection exists between the biosynthetic pathway and the endosomal pathway that could be uncovered by examining and manipulating the machinery that controls these compartments. Accurate analogies between membrane trafficking pathways in yeast and animal cells may rely on comparing the specific protein machinery that controls these steps. Alternatively, it is possible that mVps45 works throughout the TGN/endosomal system to catalyze many different events. Perhaps fusion with the TGN as well as fusion of TGN-derived vesicles with Rab5 enriched early endosomes (33). Such a pleiomorphic role for mVps45 might be expected within the confines of the SNARE hypothesis, which demands that many different Sec1p-like proteins must exist to control a diverse set of vesicle fusion events that occur along the secretory pathway (8). Far fewer Sec1p-like proteins can be found (four total) in the now completed sequence of the S. cerevisiae genome than can account for a unique Sec1p-like protein participating in every plausible vesicle fusion event. Use of a given Sec1p-like protein in a number of different trafficking steps may be occurring in a manner similar to what has been shown for the Rab protein Ypt1p, which acts in fusion events throughout the yeast Golgi (43).
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U66865[GenBank].
We thank Robert Parton, Liz Conibear, Jenny Stow, and Nia Bryant for discussions and critical reading of the manuscript; Colin MacQueen for assistance with confocal microscopy; Michael Marusich of the University of Oregon Monoclonal Antibody Facility for assistance preparing polyclonal antibodies; and G. Ficher von Mollard for communicating unpublished data.