Developmental Cell

Article

The CORVET Tethering Complex Interactswith the Yeast Rab5 Homolog Vps21and Is Involved in Endo-Lysosomal BiogenesisKarolina Peplowska,1 Daniel F. Markgraf,1,3 Clemens W. Ostrowicz,1,3 Gert Bange,2 and Christian Ungermann1,*1 University of Osnabruck, Department of Biology, Biochemistry Section, Barbarastrasse 13, 49076 Osnabruck, Germany2 Biochemie Zentrum der Universitat Heidelberg (BZH), Im Neuenheimer Feld 328, 69120 Heidelberg, Germany3 These authors contributed equally to this work.*Correspondence: christian.ungermann@biologie.uni-osnabrueck.de

DOI 10.1016/j.devcel.2007.03.006

SUMMARY

The dynamic equilibrium between vesicle fis-sion and fusion at Golgi, endosome, and vacu-ole/lysosome is critical for the maintenance oforganelle identity. It depends, among others,on Rab GTPases and tethering factors, whosefunction and regulation are still unclear. Wenow show that transport among Golgi, endo-some, and vacuole is controlled by two homol-ogous tethering complexes, the previouslyidentified HOPS complex at the vacuole anda novel endosomal tethering (CORVET) com-plex, which interacts with the Rab GTPaseVps21. Both complexes share the four classC Vps proteins: Vps11, Vps16, Vps18, andVps33. The HOPS complex, in addition, con-tains Vps41/Vam2 and Vam6, whereas theCORVET complex has the Vps41 homologVps8 and the (h)Vam6 homolog Vps3. Strikingly,the CORVET and HOPS complexe can intercon-vert; we identify two additional intermediatecomplexes, both consisting of the class Ccore bound to Vam6-Vps8 or Vps3-Vps41. Ourdata suggest that modular assembled tetheringcomplexes define organelle biogenesis in theendocytic pathway.

INTRODUCTION

Vesicle-mediated protein transport between organelles of

the endomembrane system depends on a dynamic equi-

librium between fission and fusion. Alterations of this bal-

ance lead to a loss of organelle identity and subsequently

to disease (Di Pietro and Dell’Angelica, 2005; Munro,

2004). Vesicle fusion requires Rab GTPase-dependent

tethering of the vesicle, followed by the fusion process,

which is driven by SNARE proteins residing on the vesicle

and organelle membrane. Rab GTPases have been impli-

cated as key regulators of fusion (Grosshans et al., 2006).

To be able to bind their effectors, Rab GTPases need to be

Deve

converted from the inactive GDP to the active GTP form.

The conversion depends on guanine nucleotide exchange

factors (GEFs), which vary in size and domain compo-

sition. Inactivation of Rab proteins depends on Rab

GTPase-activating proteins (GAPs), which trigger GTP

hydrolysis (Haas et al., 2005).

Tethering factors or large multimeric tethering com-

plexes (tethers) cooperate with Rab GTPases to capture

vesicles and trap them prior to the action of SNAREs (Beh-

nia and Munro, 2005; Grosshans et al., 2006). Tethers are

therefore thought to coordinate Rab and SNARE function

and provide an essential layer of specificity to fusion reac-

tions. Several large tethering complexes have been iden-

tified, including the exocyst at the plasma membrane,

the TRAPP and the COG complex at the Golgi, the

GARP complex at endosomes, and the HOPS complex

at the vacuole (Grosshans et al., 2006; Whyte and Munro,

2002). Interestingly, some tethers have been identified as

Rab GEFs, including the TRAPP complex and the Vam6

subunit of the HOPS complex (Wang et al., 2000; Wurmser

et al., 2000). A number of tethering complexes, which bind

to specific Rab GTPases, have been characterized to

date. However, since many of them harbor numerous

subunits and multiple domains, their precise function

remains widely elusive.

We are interested in tethering within the endolysosomal

system. In yeast (and later in mammalian cells), the multi-

subunit HOPS/class C Vps complex was identified as

a tethering complex, which is required for homotypic fu-

sion at the vacuole. It consists of six proteins: Vps41

(Vam2) and Vam6 (Vps39) and the class C subunits

Vps11, Vps16, Vps18, and Vps33 (Nakamura et al.,

1997; Price et al., 2000b; Rieder and Emr, 1997; Seals

et al., 2000; Wurmser et al., 2000). The complex interacts

with the GTP form of Ypt7 (yeast Rab7) and can bind to

SNAREs (Collins et al., 2005; Laage and Ungermann,

2001; Stroupe et al., 2006). It is therefore thought that

HOPS mediates the transition from tethering to trans-

SNARE pairing during fusion. The individual subunits

have strikingly different domains. Vps33 is homologous

to Sec1/Munc18 proteins, Vps11 and Vps18 have essen-

tial RING domains at their C termini (Rieder and Emr,

1997), and Vps16 has two conserved domains of unknown

function (Richardson et al., 2004) (http://www.pfam.org).

lopmental Cell 12, 739–750, May 2007 ª2007 Elsevier Inc. 739

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Figure 1. Characterization of the vps3D and vps8D Mutants

(A) Localization of vacuolar protein markers in vps3D cells. Wt and vps3D cells expressing GFP-tagged vacuolar proteins were grown logarithmically

in YPD medium, harvested, washed once with PBS buffer, and analyzed by fluorescence microscopy. Size bar = 5 mm.

(B) Double mutant of vps8 and vam3. Cells lacking Vps8, Vam3, or both were stained with FM4-64 and analyzed by fluorescence microscopy. Size

bar = 10 mm.

740 Developmental Cell 12, 739–750, May 2007 ª2007 Elsevier Inc.

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Vps41 (Vam2) seems to have two roles, one in tethering

(as part of the HOPS complex) (Price et al., 2000a; Price

et al., 2000b) and the other in the biogenesis of AP-3 ves-

icles, which transport selected vacuolar proteins from the

Golgi to the vacuole (Rehling et al., 1999). Finally, the

HOPS subunit Vam6 acts as a GEF on the Rab GTPase

Ypt7 (Wurmser et al., 2000). Overexpression of the human

Vam6 leads to clustering of lysosomes, suggesting a direct

role in tethering (Caplan et al., 2001). Interestingly, mam-

malian cells contain two homologs of Vam6, hVps39/

TLP (Caplan et al., 2001; Felici et al., 2003) and TRAP-1

(Charng et al., 1998). Both are expressed in all tissues

and have been implicated in selective Smad signaling at

the endosome.

Deletion of any class C gene leads to severe fragmenta-

tion of the vacuole (Raymond et al., 1992; Rieder and Emr,

1997). In fact, the class C proteins function not only at the

vacuole, but also at endosomes (Peterson and Emr, 2001;

Srivastava et al., 2000), where they have been shown to

bind Vps8 instead of Vps41 and Vam6 (Subramanian

et al., 2004). Vps8 is conserved across species and is crit-

ical for sorting of proteins to the endosome but dispens-

able for the AP-3 pathway (Chen and Stevens, 1996;

Horazdovsky et al., 1996). It belongs to the class D VPS

genes, whose deletion leads to an enlarged vacuole phe-

notype. Intriguingly, a number of proteins implicated in

vesicle fusion at the endosome, like Vps21 (yeast Rab5),

Vps9 (yeast Rabex5), the SNARE Pep12, Vac1 (homolo-

gous to EEA1), and Vps45 (Sec1-like protein), also belong

to this group (Bowers and Stevens, 2005). However, little

is known about Vps3, another class D gene. It was identi-

fied by Stevens and coworkers and implicated in Golgi

and endosome vesicle trafficking (Raymond et al., 1990).

We now show that Vps3 is homologous to (h)Vam6 and,

together with Vps8, is part of a novel tethering complex at

the endosome, which we term class C core vacuole/endo-

some tethering (CORVET) complex. We present evidence

that the CORVET and the HOPS complex can interconvert

by dynamic subunit exchange. Our data indicate a link

between modular assembled tethering complexes, Rab

GTPases, and the transition between endosomes and

lysosomes.

RESULTS

Organelle Identity of the Vacuole in Class D Mutants

We previously identified Vps3 as one of the proteins im-

paired in salt-induced vacuole fragmentation (Figure S1A;

LaGrassa and Ungermann, 2005). VPS3 belongs to the

class D genes, whose deletion results in a characteristic

phenotype of enlarged vacuoles (Figure S1B; Raymond

et al., 1990; Rothman et al., 1989). We wondered if defects

in vacuole fragmentation in class D mutants could be due

to the lack of vacuolar markers and loss of vacuolar

Deve

identity. Indeed, vacuoles in vps3D cells do not acidify

(Preston et al., 1989; Raymond et al., 1990; Rothman

et al., 1989) but show efficient localization of Yck3,

Vps41, Vam3, Vac8, or Pho8 (Figure 1A), suggesting that

the class D organelle has lysosome-like properties.

If the class D organelles resemble wild-type vacuoles,

they should require the vacuolar Q-SNARE Vam3 for in-

tegrity. However, we found that large class D organelles

of a vps8 mutant do not fragment, even if the vacuolar

SNARE VAM3 has been deleted (Figure 1B). This pheno-

type contrasts to the complete fragmentation of vacuoles

in the single vam3 deletion mutant (Nichols et al., 1997). It

is possible that the endosomal Q-SNARE Pep12 is taking

over the Vam3 function in the vam3D vps8D mutant (Gotte

and Gallwitz, 1997). In agreement with this observation,

we observed that the endosomal Rab5 homolog Vps21

and the vacuolar Rab Ypt7 colocalize in the vps8D mutant

(Figure 1C). Thus, class D vacuoles in vps8D or vps3D

mutants behave like endosome-vacuole hybrids, which

do not acidify, and have a deficiency in vacuole fragmen-

tation (Figure S1B) and vacuole inheritance (Figure 1D).

Vps3 Is Part of a Novel Endosomal

Tethering Complex

To understand the biogenesis of this putative hybrid or-

ganelle on the molecular level, we addressed the localiza-

tion and function of Vps3, the least characterized of all

class D genes. By subcellular localization, Vps3 was re-

covered in equal amounts in membrane and soluble frac-

tions, and its steady-state localization was not influenced

by the deletion of Vam6, Vps8, or Vps21 (data not shown).

In agreement with its proposed endosomal function, we

found GFP-tagged Vps3 in dot-like structures (Figure S2).

To identify potential interaction partners, we chromoso-

mally tagged Vps3 at the C terminus with the tandem affin-

ity purification (TAP) tag and isolated the protein from

yeast cells on IgG Sepharose and calmodulin (CaM) beads

(Figure 2A). Several bands were specifically retained with

Vps3, and their identity was determined by mass spec-

trometry. Surprisingly, we found all four class C proteins

(Vps33, Vps16, Vps11, Vps18) and Vps8. The association

of Vps8 with some class C proteins has been previously

identified and implicated in Golgi-to-endosome transport

(Subramanian et al., 2004). We therefore asked whether

Vps3 associates with Vps8 and the class C proteins indi-

vidually or forms a complex with these proteins. To distin-

guish between these possibilities, we generated yeast de-

tergent extracts to determine the size of Vps3 and Vps8.

Using glycerol gradient centrifugation, we detected Vps3

and Vps8 at a molecular weight of 700 and 120 kDa

(data not shown), supporting the idea of both proteins be-

ing part of the 700 kDa high molecular weight complex. To

rule out the possibility that the two proteins comigrate in

the glycerol gradient due to homo-oligomerization, we

(C) Colocalization of Vps21 and Ypt7. RFP-Ypt7 and GFP-Vps21 were expressed in wt and vps8D cells and visualized by fluorescence microscopy.

Images were processed by deconvolution using the Autoquant software. Size bar = 10 mm.

(D) Vacuole inheritance. Wt and vps3D cells were logarithmically grown in YPD medium, stained with FM4-64 for 30 min, reisolated, and grown in fresh

YPD medium for 3 hr before being analyzed by fluorescence microscopy. Quantification of at least 200 cells of each strain is shown.

lopmental Cell 12, 739–750, May 2007 ª2007 Elsevier Inc. 741

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Figure 2. Identification of the Vps3-

Vps8-Class C Complex

(A) Vps3-TAP purification. BJ3505 cells (wt and

VPS3-TAP) were lysed and the Vps3-TAP pro-

tein was purified according to the TAP protocol

(see Experimental Procedures). The eluate was

loaded onto a 4%–12% SDS-PAGE gradient

gel and candidate proteins were identified by

mass spectroscopy. Vps3-cbp (CaM-binding

peptide) is left after TEV cleavage.

(B–D) Sizing of Vps3-8-class C complex by gel

filtration (B). BY4741 cells carrying Vps3-TAP

were lysed and Vps3-TAP was captured on

IgG-Sepharose. After TEV cleavage, the eluate

was applied onto a Superose 6 column, and

proteins in fractions 6–19 were analyzed on

a 4%–12% SDS-PAGE gradient gel, followed

by Coomassie staining. Bands were identified

by mass spectrometry. The same purification

from BY4741 VPS8-TAP cells is shown in (C).

As a comparison, Vps41-TAP was purified by

the same approach (D). Models of the com-

plexes are shown. Class C = Vps11, Vps16,

Vps18, Vps33; 8 = Vps8; 3 = Vps3; 41 =

Vps41; V6 = Vam6.

decided to repeat the TAP purification and determine the

mobility of the Vps3 and the Vps8 complex using gel filtra-

tion. As shown in Figure 2B, Vps3 is found in a 700 kDa

complex together with Vps8, Vps11, Vps16, Vps18, and

Vps33. We also observe Vps3 in a second peak at around

120 kDa. To determine if Vps3 dissociated partially from

the Vps8-class C complex during the chromatography

step, we used Vps8 as the bait in our purification

(Figure 2C). Strikingly, the purification yielded a very sim-

ilar picture, with Vps3 being present in the Vps8-class C

complex and as a putative monomer. Vps3 must have

been in a complex with Vps8 initially to be present on

the gel filtration column and most likely dissociated par-

tially from the complex during the column run, probably

due to its distinct binding characteristics to the other sub-

units. This observation is underscored by the equimolar

amounts of Vps3 and the other subunits. We therefore

conclude that Vps3 is a subunit of a novel endosomal

complex. Given that the complex contains the class C

tethering proteins, functions at the endosome, and is re-

quired for transport between endosome and vacuole

(see below), we term it CORVET complex (class C core

vacuole/endosome tethering).

742 Developmental Cell 12, 739–750, May 2007 ª2007 Elsevie

CORVET Interacts with Vps21 (Yeast Rab5)

The CORVET complex exhibits a remarkable similarity to

the HOPS complex (depicted in the models in Figures

2B and 2D). Both complexes are composed of six sub-

units, four of which—the class C proteins Vps11, Vps16,

Vps18, and Vps33—are found in both. Whereas the

HOPS complex also contains Vps41 and Vam6, the novel

complex has Vps8 and Vps3 instead. Both complexes

have the same size (Figure 2D). Moreover, Vps8 function-

ally interacts with the endosomal Rab5 homolog Vps21

(Horazdovsky et al., 1996). Since the HOPS complex has

been shown to be an effector of the Rab Ypt7 (Seals

et al., 2000; unpublished data), we analyzed the interac-

tion of the CORVET complex with Vps21. As shown in

Figure 3A, purified CORVET binds exclusively to Vps21-

GTP by pull-down.

Intriguingly, Vps3 shows homology to the human and

yeast Vam6 protein but contains an additional N-terminal

domain, which is not found in any other Vam6 homolog

(Figure S3). When this domain was excluded from the

alignment, we could observe a clear homology to hVam6

(Figure S3). Since Vam6 interacts with Ypt7-GDP and

can promote nucleotide exchange (Wurmser et al.,

r Inc.

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Figure 3. Interaction of the Rab-GTPase

Vps21 with CORVET and Vps3

(A) Interaction of CORVET with Vps21. COR-

VET was purified from a strain overexpressing

Vps3, similar to Figure 2B, and the complex

fraction (11, 12) was applied to immobilized

GST-Vps21, Ypt1, and Ypt7, which were pre-

loaded with the indicated nucleotide (NF, nu-

cleotide-free). Bound protein was eluted by

EDTA/high salt, TCA precipitated, and ana-

lyzed by SDS-PAGE and western blotting using

Protein A-peroxidase. The Rab-GTPase was

eluted by boiling beads in sample buffer and

analyzed as above. Note that purified CORVET

complex lacks most Vps3 due to its dissocia-

tion during chromatography (Figure 2B).

(B and C) Interaction of Vps3 with Vps21. De-

tergent lysate prepared from 3 l of cells overex-

pressing Vps3 in the presence (B) or absence

(C) of Vps8 was applied to GST-Rabs contain-

ing the indicated nucleotide. Analysis was

done as in (A).

(D and E) Localization of GFP-Vps21. Cells

lacking or overexpressing Vps8, Vps3, or

Vam6 in the presence of GFP-Vps21 were

analyzed by fluorescence microscopy as in

Figure 1A. Size bar = 10 mm.

2000), we asked whether Vps3 can bind to Vps21-GDP.

When lysates from cells overexpressing Vps3 were ap-

plied to GST-Ypt7, Ypt1, or Vps21, Vps3 was recovered

with Vps21 but not Ypt7 or Ypt1 (Figure 3B). Surprisingly,

we did not observe a nucleotide-specific binding, sug-

gesting that we analyzed different Vps3 populations in

this assay (as a monomer and as part of the CORVET

complex). We therefore repeated the assay using lysates

from vps8D cells, which lack the CORVET complex,

and detected Vps3 preferentially in complex with the

nucleotide-free and GDP form of Vps21 (Figure 3C).

Likewise, Vps3 bound preferentially to the GDP/nucleo-

tide-free-mimicking S21N mutant of Vps21 in a co-

overexpression experiment (Figure S4). Thus, the Vps3

binds to Vps21-GDP, which mirrors the Vam6 binding to

Ypt7-GDP.

To test for a functional relationship of CORVET and

Vps21 in vivo, we followed GFP-Vps21 localization. Vps8

overexpression led to a striking dot-like accumulation of

Vps21 on membranes, whereas it appeared dispersed in

wild-type cells (Figures 3D and S5). This membrane accu-

mulation of GFP-Vps21 required Vps3 (Figure 3E), indicat-

ing that the CORVET subunits Vps8 and Vps3 cooperate in

Vps21 localization. Overexpression of Vps3 seemed to

have a similar effect, although the fragmentation of the

vacuoles (see below) interfered with a quantitative analy-

sis. In sum, the CORVET complex acts as an effector of

the Rab-Vps21, while the Vps3 protein has preference

for Vps21 in the nucleotide-free and GDP forms, which

is consistent with a GEF-like function.

Deve

Identification of Intermediate Tethering Complexes

It is striking that the CORVET and HOPS complex both

contain the class C core and have homologous subunits.

We therefore wondered whether intermediate complexes

might exist that could form during endosome-vacuole bio-

genesis. If Vps3 and Vam6 would exchange, two new

complexes with the class C core would be possible: one

containing Vam6-Vps8 and the other containing Vps3-

Vps41. To probe for their existence in wild-type cells, we

TAP-tagged Vam6 and Vps41 and looked for their associ-

ation with Vps3 and Vps8. Indeed, the Vam6-Vps8 com-

plex is present in wild-type cells (Figures 4A and 4B);

HA-tagged Vps8 copurified with Vam6-TAP and the class

C core on a gel filtration column. This association is not

apparent by Coomassie staining, since the HOPS com-

plex is the primary complex purified. Likewise, the Vps3-

Vps41 complex was detected in wild-type cells using

Vps41-TAP as bait, but not in vps33D cells (Figure 4C,

lane 1 versus 5; see below). Thus, the class C core is

a module found in four different complexes: (1) the

CORVET (Vps3-Vps8) complex, (2) together with Vam6

and Vps8 (i-HOPS, see below), (3) in the HOPS complex,

and (4) in combination with Vps3 and Vps41 (i-CORVET,

see below).

Dynamics of Tethering Complexes

Whereas the CORVET and HOPS complex appear to be

more abundant in the cell, the intermediate complexes

seem to be transient. We wondered whether we could

accumulate the intermediate complexes under certain

lopmental Cell 12, 739–750, May 2007 ª2007 Elsevier Inc. 743

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Figure 4. Identification of Two Interme-

diate Complexes

(A and B) Identification of the Vam6-Vps8-class

C intermediate. Cells containing TAP-tagged

Vam6 and HA-tagged Vps8 were processed

for HOPS purification as in Figure 2B. Proteins

in each fraction were resolved on gradient

SDS-PAGE gels, which were either stained

with Coomassie ([A], 80% of the sample) or

transferred to nitrocellulose ([B], 20%). West-

ern blots were first decorated with antibodies

to the HA-tag and a mouse secondary anti-

body coupled to AlexaFluor680 (Molecular

Probes; [B], bottom panel), then with anti-

bodies against Vps41 coupled to a goat anti-

rabbit secondary antibody coupled to IR-

Dye800 (LI-COR). To visualize both proteins,

the channels were merged (top panel). A model

of the Vps8-Vam6-class C (i-HOPS) complex is

shown.

(C and D) Identification and dynamics of Vps3-

Vps41 complexes. Wt or deletion strains con-

taining Vps41-TAP (C) or Vps3-TAP (D) were

lysed. Tagged proteins were purified on CaM

beads using 700 mg of total protein (see Exper-

imental Procedures). EGTA eluates were TCA

precipitated, and proteins were analyzed on

a 7.5% SDS-PAGE gel, followed by western

blotting with antibodies against the tag, Vps3

or Vps41, Vps33, and Vac8. Note that both

the Vps33 and the Vps41 antibodies cross-re-

act in the total lysate with unknown proteins at

their apparent molecular weight. The proteins

are, however, not detected upon complex pu-

rification in the vps33D or vps41D background,

confirming that they are not Vps33 and Vps41,

respectively.

(E) Purification and sizing of the Vps3-Vps41-class C complex. TAP-tagged Vps3 was purified from vps8 cells as in Figure 2B. The proteins were an-

alyzed as in (A). The identity of Vps41 was confirmed by western blotting and mass spectrometry (data not shown). The model depicts the Vps3-

Vps41-class C (i-CORVET) complex.

conditions. To this end, we concentrated on the Vps3-

Vps41-class C intermediate (i-CORVET) and followed

the complex in several mutant strains lacking endosomal

or vacuolar proteins. Vps41 was TAP-tagged in wild-

type cells and the vps3, vam6, vps33, and vps8 deletion

mutants, and complexes were purified using CaM beads

(Figure 4C). In wild-type cells, the class C core subunit

Vps33 is associated with Vps41 (lane 1). As discussed,

a small amount of Vps3 was also detected, which was ab-

sent in a vps3D mutant (lane 5). This picture changed dra-

matically when Vam6 was absent: the association of Vps3

with Vps41 increased several-fold (lane 4). A similar asso-

ciation of Vps3 with Vps41 was also observed in the ab-

sence of Vps8 (lane 3). Vac8, a vacuolar fusion factor,

was not recovered in any complexes. Similar results

were obtained when Vps3 was tagged in mutant strains;

the level of the Vps3-Vps41 association was increased

several-fold in the vps8 or the vam6 mutant (Figure 4D).

To demonstrate the presence of an intermediate complex

containing Vps3, Vps41, and the class C proteins, we pu-

rified Vps3-TAP from the vps8D mutant. As shown in

Figure 4E, the complex is comparable to the HOPS and

CORVET complex in size and abundance, indicative of

744 Developmental Cell 12, 739–750, May 2007 ª2007 Elsev

an intermediate endosome-vacuole tethering complex.

Besides the identification of the Vps41-Vps3-class C in-

termediate complex, our data also indicate that Vps41

can replace Vps8. It is therefore likely that the proteins

may perform similar functions in the complexes.

Endosomal Vps3 Can Affect HOPS Function

and Vacuole Morphology

To rule out the possibility that Vps3 replaces Vam6 only if

Vam6 is absent from cells and to test if Vps3 can also com-

pete with Vam6 for Vps41 in wild-type cells, we overex-

pressed Vps3 and followed vacuole morphology and the

composition of the complex associated with Vps41. While

deletion of vps3 leads to a single enlarged vacuole in the

cell, overexpression of Vps3 caused complete vacuole

fragmentation (Figure 5A). This phenotype is strikingly

similar to the morphology of vam6D vacuoles (Wada

et al., 1992). We used the Vps41 protein as bait to analyze

the composition of the HOPS complex under these condi-

tions. In wild-type cells or when Vps3 expression was re-

pressed, Vam6 was associated with Vps41 (Figure 5B,

lane 1). However, when Vps3 was overproduced (lane 3),

the Vam6 amount bound to Vps41 was reduced and

ier Inc.

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Figure 5. Effect of Vps3 and Vps8 Over-

production on Complex Composition

and Vacuole Morphology

(A) Overproduction of Vps3 leads to severe

vacuole fragmentation, whereas overproduc-

tion of Vps8 does not influence vacuole mor-

phology. BY VPS41-TAP cells with or without

VPS3 or VPS8 under the control of the GAL1

promoter were grown overnight in YPD (glu-

cose) or YPG (galactose) medium, then diluted

and grown to logarithmic phase. Cells were

stained with FM4-64 and analyzed by fluores-

cence microscopy (see Experimental Proce-

dures section for the details). Size bar = 10 mm.

(B) Overproduced Vps3 binds to Vps41. BY

VPS41-TAP cells with or without VPS3 under

the control of the GAL1 promoter were grown

overnight in YPD or YPG medium. In the morn-

ing, cells were lysed and Vps41-TAP protein

was purified as described in Figure 4C and

the Experimental Procedures section. Western

blots were decorated with antibodies to Vam6,

Vps33, Vps3, and Vac8 (as a negative control).

A model indicates the conversion of the HOPS

complex into the Vps3-Vps41-class C interme-

diate.

(C) Loss of the Vps3-Vps41-class C complex

by overproduction of Vps8. BY wild-type or

vam6D cells containing VPS41-TAP and VPS8

under the control of the GAL1 promoter were

grown overnight in YPD or YPG medium. In

the morning, cells were lysed and the Vps41-

TAP protein was purified according to the

protocol described in the Experimental Proce-

dures section and Figure 4. Proteins copurify-

ing with Vps8 were analyzed by SDS-PAGE

and western blotting using antibodies against

Vps33 and Vps3. A model indicates the con-

version triggered by Vps8 overproduction.

(D) Vacuole morphology in the presence of

overproduced Vps3. Diploid cells carrying

one allele of VPS3 under the control of the

GAL1 promoter were grown in YPD. Synthesis

of Vps3 was induced by exchanging the

medium for YPG. Cells were incubated with FM4-64 1 hr prior to each time point and incubated for 30 min in YPD (time points 0–1 hr) or YPG (all

other time points) medium (pulse), then washed and incubated for an additional 30 min in YPD or YPG (chase). Cells were analyzed by fluorescence

microscopy.

(E) Expression control of Vps3. At the same time point of microscopic analysis, an aliquot was processed for whole protein analysis. All samples were

analyzed by SDS-PAGE and western blotting using the Protein A-peroxidase (PAP) and the Vps3 antibody.

(F) Pho8 and Ape1 maturation in GAL1-VPS3 strains. Total cell extracts from pep4D, wt, and GAL1-VPS3 cells grown in galactose were prepared and

analyzed by SDS-PAGE and western blotting using antibodies to Ape1 and Pho8 (m = mature, pro = precursor).

was in part replaced by Vps3. Thus, Vps3 and Vam6 com-

pete for binding to Vps41. In agreement with this, Vam6

(Wurmser et al., 2000) and Vps3 (this study) both require

the class C subunit Vps11 for binding to Vps33 (Figure S6).

Moreover, overexpression of Vps3 leads to a processing

defect of Ape1 and Pho8, indicating defective vesicle fu-

sion at the vacuole due to loss of Vam6 from the HOPS

complex (Figure 5F). Preliminary data suggest that purified

Vps3 can displace Vam6 from the HOPS complex in vitro

(data not shown). Our data indicate that Vps3 and Vam6

occupy the same binding site on the class C complex,

which allows a shift from the HOPS complex to the

Vps3-Vps41-class C intermediate.

Deve

We subsequently asked if Vps8 would compete for the

Vps41-associated Vps3 (Figure 5C). To address this ques-

tion, we first accumulated the Vps3-Vps41-class C inter-

mediate by suppressing Vps8 expressing (lane 2) or by de-

leting VAM6 in addition (lane 2 versus 4). When Vps8 was

overproduced, the Vps3-Vps41 interaction was lost (lanes

3 and 5), most likely due to a shift of Vps3 toward the COR-

VET complex. Interestingly, overproduced Vps8 did not

affect vacuole morphology (Figure 5A), suggesting that it

cannot per se displace Vps41 from the HOPS complex,

even though it is likely that Vps8 and Vps41 occupy the

same binding site on the class C core complex. Hence,

our data suggest that the HOPS complex can convert

lopmental Cell 12, 739–750, May 2007 ª2007 Elsevier Inc. 745

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Figure 6. Directionality of Complex

Assembly

(A) Induction and repression of Vps8 expres-

sion. Cells with TAP-tagged VPS41 and HA-

tagged VPS8 under the control of the GAL1

promoter were grown in medium containing

raffinose to repress Vps8 expression, then ga-

lactose was added (lane 2) for the indicated

time. Afterwards, glucose was added and cells

were incubated for an additional 6–18 hr. At

each time point, an aliquot was removed from

the culture. Cells were lysed, proteins were

TCA precipitated, and analyzed by SDS-

PAGE and western blotting using antibodies

against HA and Vac8.

(B) Conversion from i-CORVET to CORVET is

unidirectional. The experiment was performed

as in (A). Glucose addition was for 6 hr. Cells

were lysed using detergent, and Vps41 was

purified using CaM beads (see Experimental

Procedures for details). A loading control was

removed from the extracts and TCA precipi-

tated. Vps41-TAP-associated proteins were

analyzed by SDS-PAGE and western blotting

using an antibody to Vps3. Vps3 and Vps41

bands were quantified by laser densitometry.

Interaction in lane 1 was set to 100%; interac-

tions in lanes 2 and 3 were 24% and 26%,

respectively.

(C) Models of tether dynamics between endo-

some and vacuole. For details see text.

into the CORVET complex via a Vps3-Vps41-class C

intermediate.

In our previous experiments, we started from Vps3-

depleted cells and followed the effect of Vps3 overpro-

duction, which resulted in vacuole fragmentation. We

wondered if overproduction of Vps3 would also affect

the vacuole morphology if it were carried out in the wild-

type background. When diploid cells containing one

copy of Vps3 under the control of the GAL1 promoter

were grown in glucose (which represses Vps3 overpro-

duction), their vacuoles appeared like wild-type vacuoles

(Figure 5D). We followed the fate of the vacuole by FM4-

64 staining over time after inducing Vps3 overproduction

by the addition of galactose to the growth medium. In-

triguingly, as soon as Vps3 overproduction was detect-

able (Figure 5E), vacuole integrity was lost (Figure 5D);

with increasing cellular Vps3 content, vacuoles became

multilobed and then fragmented completely. Since Vps3

overproduction leads to the formation of the Vps3-

Vps41-class C complex (Figure 5B), we suggest that vac-

uole fragmentation in the presence of elevated amounts of

Vps3 is due to the loss of the HOPS complex and an in-

crease of the Vps3-Vps41-class C intermediate. It is pos-

sible that tethering complex dynamics are directly linked

to endosome-lysosome morphology and identity in yeast.

Directionality of Tether Assembly

Our data indicate that the Vps3 overproduction can drive

the formation of the Vps3-Vps41-class C intermediate,

whereas Vps8 overproduction does not influence the

746 Developmental Cell 12, 739–750, May 2007 ª2007 Elsevie

HOPS complex and, thus, vacuole morphology. Our find-

ings are consistent with a linear transition from HOPS via

the Vps3-Vps41-class C intermediate to the CORVET

complex by (1) exchanging Vps3 for Vam6 (Figure 5B),

then (2) replacing Vps41 with Vps8 (Figure 5C).

To test if the transition between complexes is direc-

tional, we focused on the transition from the Vps3-

Vps41-class C to the endosomal CORVET complex. We

took advantage of the vam6D mutant containing the

VPS8 gene under the control of a galactose-inducible pro-

moter (Figures 5A and 5C). Vps8 expression was induced

by adding galactose and led to strong overproduction

(Figure 6A, lanes 1 and 2). When glucose was added to

the medium, Vps8 expression was repressed and the

protein disappeared within 6 hr (lanes 3 and 4). We then

followed the dynamics of the intermediate complex by

Vps41 pull-down. As shown before, vam6D mutants accu-

mulate the Vps3-Vps41-class C intermediate in the ab-

sence of Vps8 (Figure 6B, lane 1; 100%). Overexpression

of Vps8 leads to a loss of the intermediate complex (lane

2), presumably by replacing Vps8 with Vps41 and by gen-

erating additional CORVET complex. Subsequently, we

repressed Vps8 expression by the addition of glucose to

the medium. If the reverse reaction would be possible,

Vps41 should replace Vps8 again, thus reforming the

lost Vps3-Vps41 complex. Based on our hypothesis that

regeneration of the Vps3-Vps41-class C intermediate

can occur via the HOPS complex, this route would be

blocked in the absence of Vam6. Interestingly, we could

not recover the intermediate complex upon loss of Vps8

r Inc.

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

(Figure 6B, lane 3; 26%); the retrieved amount was as low

as during the overexpression of Vps8 (lane 2; 24%), indi-

cating that the i-CORVET was not reformed during our ob-

servation period. It is possible that the lack of reformation

is due to its assembly kinetics. Alternatively, the efficient

reformation might require the HOPS complex. In sum,

our data show that endolysosomal tethers can convert in

a directed manner, suggesting a link to organelle identity

and Rab GTPase switching (Figure 6C).

DISCUSSION

The data presented here shed light on the transition be-

tween endosome and lysosome in yeast and give valuable

insight into the maintenance of organelle identity in eu-

karyotic cells. We identify a novel endosomal tethering

complex containing Vps3 and Vps8 plus the class C pro-

teins (Vps11, Vps16, Vps18, and Vps33), which we name

the CORVET complex. This identification paved the

ground for several important findings. Our data reveal

that the CORVET and the HOPS complexes are homolo-

gous. They share the class C core as a common platform,

onto which the two additional subunits assemble: Vps41

and Vam6 for the HOPS complex, and Vps8 and Vps3

for the CORVET complex. In support of this, we show

that Vps3 is homologous to and shows similar binding

characteristics to the class C core as the HOPS-subunit

Vam6 (Figure S4). Our data indicate that CORVET is an

effector of the Rab5 homolog Vps21 and suggest that

Vps3 may act as a GEF for Vps21, since it binds Vps21

preferentially in its nucleotide-free and GDP form (Fig-

ure 3); Vps3 would therefore be equivalent to Vam6, which

has GEF activity for the yeast Rab7 homolog Ypt7

(Wurmser et al., 2000). Furthermore, as Vps8 seems to

be able to displace Vps41 from the i-CORVET (Figures 5

and 6), we suggest that the two proteins have similar

functions in each complex. Finally, we identify intermedi-

ate complexes with subunits from the CORVET and the

HOPS complex.

We present two models on how the complexes may

cooperate between endosome and lysosome. In the first

model, we assume that all four complexes can assemble

de novo and bind to different Rab-GTPases on endocytic

membranes (Figure 6C, model I). Indeed, homologs of

Vps21, called Ypt52 and Ypt53, exist and may bind to

the intermediate complexes (Singer-Kruger et al., 1994).

Each complex could promote tethering of vesicular inter-

mediates. Based on the previously identified GEF function

of Vam6 for the Rab Ypt7, we assume that the Vps8-Vam6

intermediate (intermediate toward HOPS) is required in the

direction from the endosome to the vacuole, while the

Vps41-Vps3 intermediate (i-CORVET) could be required

in retrieving material from the late endosome. In this

model, the tether composition observed during overex-

pression of Vps3 or deletion of VAM6 or VPS8 could be

the result of the de novo assembly pathway or a shift in

the balance. An alternative model (model II) links the ob-

served order of interconversion to changes in the Rab-

GTPases. Here, the CORVET complex would first convert

Deve

into the Vam6-Vps8-class C (i-HOPS) complex. By ex-

changing the Rab-GDP binding proteins Vps3 for Vam6,

the prevailing Rab GTPase could be switched from

Vps21 to Ypt7. The i-HOPS complex would in the second

stage convert into the HOPS complex by replacing Vps8

with Vps41. In a third step, the HOPS complex can change

to the Vps3-Vps41-class C intermediate (i-CORVET).

Replacing Vps3 for Vam6 leads again to a change in

Rab protein binding (Vps21 for Ypt7) due to the change

in GEF activity. To complete the postulated cycle, the

intermediate complex could convert into the CORVET

complex by replacing Vps41 with Vps8.

Our data are consistent with either model. All com-

plexes described in each model are observed in wild-

type cells, with the intermediate complexes being of low

abundance. Interestingly, the shift in complex composi-

tion is associated with morphology changes of the vacu-

ole. Only overproduction of Vps3 leads to a fragmentation

of the vacuole and an accumulation of the i-CORVET com-

plex, while Vps8 overproduction did not affect vacuole

morphology. We speculate that Vps41 and Vps8 recog-

nize Ypt7 and Vps21, respectively, in their active GTP

form. In support of this notion, Vps8 seems to interact

with Vps21 (Horazdovsky et al., 1996), in particular with

its GTP form (A. Merz, personal communication), and

binding of CORVET/Vps3 to Vps21-GTP was reduced in

the vps8D background (Figure 3C). Moreover, we show

that the CORVET complex binds Vps21-GTP and that

overexpression of the CORVET subunit Vps8 drives the

accumulation of Vps21 to dot-like structures adjacent to

the vacuole (Figure 3E). This would mean that each com-

plex consists of three parts, the class C core, a GEF (Vps3

or Vam6), and an effector protein (Vps41 and Vps8). Dur-

ing the conversion of tethers induced by overproduction,

the putative GEF is exchanged first, followed by the poten-

tial Rab effector. This order is appealing, as the GEF would

recruit the next Rab, which would bind the next effector

(Model II). On Golgi-derived vesicles, a cross-talk between

Rabs and a GEF has been observed (Ortiz et al., 2002);

here, Ypt31-GTP binds to the GEF Sec2, which then re-

cruits the Rab Sec4. In model I, each complex would exist

as a stable entity to mediate Rab binding and tethering.

Future experiments will need to address the relevance of

each model.

Our observations are consistent with the findings on the

endosome-lysosome transition in mammalian cells (Bright

et al., 2005; Luzio et al., 2000; Mullock et al., 1998; Rink

et al., 2005) and offer a molecular explanation for this pro-

cess. However, one issue is puzzling. Rink et al. (2005)

showed that hVam6/hVps39 was eluted from both

Rab5-GDP and Rab-GTP columns. In yeast, Vam6 does

not bind to the Rab5-homolog Vps21 in its GDP form

(Wurmser et al., 2000), but Vps3 does (our study). In

fact, two Vam6 homologs have been described in mam-

malian cells, TRAP-1 (Charng et al., 1998) and hVam6/

hVps39 (Caplan et al., 2001), which has also been termed

TLP (Felici et al., 2003). Both proteins seem to be ex-

pressed in all human tissues. Possibly, Vps3, which shows

higher similarity to the human Vam6 variants than yeast

lopmental Cell 12, 739–750, May 2007 ª2007 Elsevier Inc. 747

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Vam6, corresponds to the human hVps39 analyzed by

Zerial and colleagues, whereas yeast Vam6 could corre-

spond to TRAP-1. In fact, depletion of hVps39 led to the

generation of a swollen hybrid organelle containing Rab5

and Rab7 (Rink et al., 2005), similar to our observations

on the deletion of VPS3 that results in a class D vacuole

phenotype (Figure 1). Therefore, the transition of endo-

some to lysosome could be mediated by directed alter-

ation of tethers on endosomal membranes, followed by

homotypic fusion with lysosomes (Figure 6C).

Our results on the modular assembly of the C com-

plexes are not without precedence. The Rabex5-Rabap-

tin5 complex consists of a Rab5 GEF and a Rab5 effector

(Horiuchi et al., 1997). Both proteins need to act together

to promote endosome fusion (Lippe et al., 2001). The

TRAPP tethering complex has been identified in two com-

positions: TRAPPI is required for ER-Golgi transport and

TRAPP II at the trans Golgi-early endosome interface

(Cai et al., 2005; Sacher et al., 1998). TRAPP I contains

GEF activity for the Rab Ypt1 (Kim et al., 2006; Wang

et al., 2000). Recently, it has been shown that the addi-

tional subunits of TRAPP II, Trs120, and Trs130 confer

GEF activity for the endosomal Rabs Ypt31/32 (Morozova

et al., 2006). Thus, TRAPPI may convert into TRAPPII to

drive vesicular transport during secretion. It is possible

that this principle also applies to other intracellular tether-

ing complexes.

In sum, our data provide an important molecular

extension to our understanding of endosome-vacuole/

lysosome biogenesis. We suggest that tethering com-

plexes control Rab GTPase switching and stability. Future

studies need to focus on the assembly and disassembly

reaction, the regulation of each complex, and their con-

nection to cargo and Rabs.

EXPERIMENTAL PROCEDURES

Yeast Strains and Molecular Biology

Strains used are listed in Table S1. Deletions and tagging of genes

were done using homologous recombination in BY4741 (MATa

his3D1 leu2D0 met15D0 ura3D0). Details on strain construction and

plasmids are available in Supplemental Data.

Microscopy

Staining of cells with the lipophilic dye FM4-64 or CMAC was per-

formed as described (LaGrassa and Ungermann, 2005). For GFP

microscopy, cells were grown logarithmically in YPD or selective

medium, collected by centrifugation, washed once with 1 ml PBS

buffer and analyzed by fluorescence microscopy. Images were ac-

quired with a Zeiss Axiovert 35 microscope equipped with an AxioCam

and a 1003 objective using filter set 10 or phase contrast, or with a

Leica DM5500 microscope and a SPOT Pursuit camera using GFP,

FM4-64 by phase contrast or DIC filters. Pictures were processed

using Adobe Photoshop 7.0.

Yeast Cell Lysis

Cells were fractionated essentially as described (LaGrassa and Unger-

mann, 2005). For details see Supplemental Data.

CaM Pull-Down

A cleared detergent extract of total cell lysate was generated as above.

A fraction (5%) of the total protein amount was removed as a loading

748 Developmental Cell 12, 739–750, May 2007 ª2007 Elsevie

control (followed by TCA precipitation), and the rest was loaded onto

80 ml of prewashed CaM beads (GE Healthcare). Samples were incu-

bated for 2 hr at 4C in the presence of 2 mM CaCl2. Beads were

washed three times for 10 min with buffer (0.2 M sorbitol, 150 mM

KCl, 20 mM HEPES/KOH, [pH 6.8], 2 mM CaCl2) with decreasing Triton

X-100 concentrations (0.5%, 0.1%, 0.025%), and proteins were eluted

by addition of 2 mM EGTA followed by a 20 min incubation at 30C.

Beads were reisolated, the eluate was transferred to a fresh tube,

TCA precipitated, and analyzed by SDS-PAGE and western blotting.

Blots were analyzed using the primary antibody indicated, and sec-

ondary antibodies coupled to dyes or horseradish peroxidase. Detec-

tion was performed using a LICOR-ODYSSEY system or by standard

enhanced chemiluminescence.

TAP-Tag Protein Purification

TAP-tag protein purification was performed as described in Rigaut

et al. (1999) using the following buffer: 50 mM HEPES/KOH, (pH 7.4),

300 mM NaCl, 0.15% NP-40 (Igepal CA-630; Sigma-Aldrich), and

1.5 mM MgCl2.

Gel Filtration

Protein complexes were purified as described in TAP-tag protein puri-

fication protocol omitting the CaM-bead purification step. The TEV

eluate was centrifuged for 10 min at 20,000 3 g to pellet insoluble

material. The supernatant was applied to a Superose 6 10/300 column,

connected to an AKTA-FPLC-System (GE Healthcare), which had

been equilibrated with two column volumes of TAP-purification buffer.

The flow rate was set to 0.3 ml/min, and 24 1 ml fractions were col-

lected. For analysis, fractions 6–19 were TCA precipitated and loaded

onto a 4%–12% SDS-PAGE gradient gel (NuPAGE, Invitrogen).

GSH Pull-Down

GST fusion proteins (400 mg per sample) were bound to GSH beads

and washed three times with 500 ml 20 mM HEPES/KOH (pH 7.4),

100 mM NaCl, 10 mM EDTA, 0.1%TX100. Beads were resuspended

in 20 mM HEPES/KOH (pH 7.4), 100 mM NaCl, 1 mM MgCl2, 0.1%

TX100, and 0.5 mM GDP, GTPgS or no nucleotide, and incubated

for 1 hr at 4C. Purified CORVET (fractions 10 and 11) was prepared

from the GAL1-TAP-VPS3 overexpression strain as described above,

and 1 ml was added to each sample. Alternatively, lysates from the

indicated strains were prepared by glass bead lysis from 3 l of cells

in 20 mM HEPS/KOH (pH 7.4), 100 mM NaCl, 1 mM MgCl2, 0.1%

TX100, centrifuged (1 hr, 100,000 3 g, 4C), concentrated to 2 ml using

an Amicon Ultra Centrifugal Filter Device (MWCO 10,000), and added

to the prebound Rab GTPases. Beads were incubated for 1 hr at 4C

on a rotating wheel, washed four times with decreasing TX100 con-

centrations, and eluted by incubating beads in 20 mM HEPES/KOH

(pH 7.4), 1.5 M NaCl, 20 mM EDTA, 0.025% TX-100 for 20 min at

room temperature. Eluates were TCA precipitated and analyzed by

SDS-PAGE and western blotting.

Supplemental Data

Supplemental Data include supplemental Experimental Procedures

and references, six supplemental figures, and one supplemental table

and are available at http://www.developmentalcell.com/cgi/content/

full/12/5/739/DC1/.

ACKNOWLEDGMENTS

We thank Alexey Merz for communicating results prior to publication;

Tom Stevens for kindly sharing the Vps3 antibody with us; Francis

Barr, Scott Emr, Alexey Rak, Roger Goody, and Bruno Antonny for dis-

cussions; Michael Knop for plasmids; Christoph Meiringer for help with

microscopy; Angela Perz and Gabriela Muller for expert technical as-

sistance; and all members of the Ungermann group for support and

discussion. This work was supported by the DFG (UN111/3-1; Heisen-

berg program), the SFB 638 and 431, the EMBO Young Investigator

Program, and the Fonds der Chemischen Industrie. C.U. is supported

r Inc.

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

by the Hans-Muhlenhoff foundation. K.P. has been a recipient of a sti-

pend of the Landesgraduiertenforderung Baden-Wurttemberg. Part of

this work has been performed at the University of Heidelberg.

Received: September 25, 2006

Revised: December 27, 2006

Accepted: March 8, 2007

Published: May 7, 2007

REFERENCES

Behnia, R., and Munro, S. (2005). Organelle identity and the signposts

for membrane traffic. Nature 438, 597–604.

Bowers, K., and Stevens, T.H. (2005). Protein transport from the late

Golgi to the vacuole in the yeast Saccharomyces cerevisiae. Biochim.

Biophys. Acta 1744, 438–454.

Bright, N.A., Gratian, M.J., and Luzio, J.P. (2005). Endocytic delivery to

lysosomes mediated by concurrent fusion and kissing events in living

cells. Curr. Biol. 15, 360–365.

Cai, H., Zhang, Y., Pypaert, M., Walker, L., and Ferro-Novick, S. (2005).

Mutants in trs120 disrupt traffic from the early endosome to the late

Golgi. J. Cell Biol. 171, 823–833.

Caplan, S., Hartnell, L.M., Aguilar, R.C., Naslavsky, N., and Bonifacino,

J.S. (2001). Human Vam6p promotes lysosome clustering and fusion

in vivo. J. Cell Biol. 154, 109–122.

Charng, M.J., Zhang, D., Kinnunen, P., and Schneider, M.D. (1998). A

novel protein distinguishes between quiescent and activated forms of

the type I transforming growth factor beta receptor. J. Biol. Chem. 273,

9365–9368.

Chen, Y.J., and Stevens, T.H. (1996). The VPS8 gene is required for

localization and trafficking of the CPY sorting receptor in Saccharomy-

ces cerevisiae. Eur. J. Cell Biol. 70, 289–297.

Collins, K.M., Thorngren, N.L., Fratti, R.A., and Wickner, W.T. (2005).

Sec17p and HOPS, in distinct SNARE complexes, mediate SNARE

complex disruption or assembly for fusion. EMBO J. 24, 1775–1786.

Di Pietro, S.M., and Dell’Angelica, E.C. (2005). The cell biology of

Hermansky-Pudlak syndrome: Recent advances. Traffic 6, 525–533.

Felici, A., Wurthner, J.U., Parks, W.T., Giam, L.R., Reiss, M., Karpova,

T.S., McNally, J.G., and Roberts, A.B. (2003). TLP, a novel modulator

of TGF-beta signaling, has opposite effects on Smad2- and Smad3-

dependent signaling. EMBO J. 22, 4465–4477.

Gotte, M., and Gallwitz, D. (1997). High expression of the yeast syn-

taxin-related Vam3 protein suppresses the protein transport defects

of a pep12 null mutant. FEBS Lett. 411, 48–52.

Grosshans, B.L., Ortiz, D., and Novick, P. (2006). Rabs and their effec-

tors: Achieving specificity in membrane traffic. Proc. Natl. Acad. Sci.

USA 103, 11821–11827.

Haas, A.K., Fuchs, E., Kopajtich, R., and Barr, F.A. (2005). A GTPase-

activating protein controls Rab5 function in endocytic trafficking. Nat.

Cell Biol. 7, 887–893.

Horazdovsky, B.F., Cowles, C.R., Mustol, P., Holmes, M., and Emr,

S.D. (1996). A novel RING finger protein, Vps8p, functionally interacts

with the small GTPase, Vps21p, to facilitate soluble vacuolar protein

localization. J. Biol. Chem. 271, 33607–33615.

Horiuchi, H., Lippe, R., McBride, H.M., Rubino, M., Woodman, P.,

Stenmark, H., Rybin, V., Wilm, M., Ashman, K., Mann, M., and Zerial,

M. (1997). A novel Rab5 GDP/GTP exchange factor complexed to

Rabaptin-5 links nucleotide exchange to effector recruitment and

function. Cell 90, 1149–1159.

Kim, Y.G., Raunser, S., Munger, C., Wagner, J., Song, Y.L., Cygler, M.,

Walz, T., Oh, B.H., and Sacher, M. (2006). The architecture of the multi-

subunit TRAPP I complex suggests a model for vesicle tethering. Cell

127, 817–830.

Devel

Laage, R., and Ungermann, C. (2001). The N-terminal domain of the

t-SNARE Vam3p coordinates priming and docking in yeast vacuole

fusion. Mol. Biol. Cell 12, 3375–3385.

LaGrassa, T.J., and Ungermann, C. (2005). The vacuolar kinase Yck3

maintains organelle fragmentation by regulating the HOPS tethering

complex. J. Cell Biol. 168, 401–414.

Lippe, R., Miaczynska, M., Rybin, V., Runge, A., and Zerial, M. (2001).

Functional synergy between Rab5 effector Rabaptin-5 and exchange

factor Rabex-5 when physically associated in a complex. Mol. Biol.

Cell 12, 2219–2228.

Luzio, J.P., Rous, B.A., Bright, N.A., Pryor, P.R., Mullock, B.M., and

Piper, R.C. (2000). Lysosome-endosome fusion and lysosome biogen-

esis. J. Cell Sci. 113, 1515–1524.

Morozova, N., Liang, Y., Tokarev, A.A., Chen, S.H., Cox, R., Andrejic,

J., Lipatova, Z., Sciorra, V.A., Emr, S.D., and Segev, N. (2006). TRAPPII

subunits are required for the specificity switch of a Ypt-Rab GEF. Nat.

Cell Biol. 8, 1263–1269.

Mullock, B.M., Bright, N.A., Fearon, C.W., Gray, S.R., and Luzio, J.P.

(1998). Fusion of lysosomes with late endosomes produces a hybrid

organelle of intermediate density and is NSF dependent. J. Cell Biol.

140, 591–601.

Munro, S. (2004). Organelle identity and the organization of membrane

traffic. Nat. Cell Biol. 6, 469–472.

Nakamura, N., Hirata, A., Ohsumi, Y., and Wada, Y. (1997). Vam2/

Vps41p and Vam6/Vps39p are components of a protein complex on

the vacuolar membranes and involved in the vacuolar assembly in

the yeast Saccharomyces cerevisiae. J. Biol. Chem. 272, 11344–

11349.

Nichols, B.J., Ungermann, C., Pelham, H.R., Wickner, W.T., and Haas,

A. (1997). Homotypic vacuolar fusion mediated by t- and v-SNAREs.

Nature 387, 199–202.

Ortiz, D., Medkova, M., Walch-Solimena, C., and Novick, P. (2002).

Ypt32 recruits the Sec4p guanine nucleotide exchange factor,

Sec2p, to secretory vesicles; evidence for a Rab cascade in yeast.

J. Cell Biol. 157, 1005–1016.

Peterson, M.R., and Emr, S.D. (2001). The class c vps complex func-

tions at multiple stages of the vacuolar transport pathway. Traffic 2,

476–486.

Preston, R.A., Murphy, R.F., and Jones, E.W. (1989). Assay of vacuolar

(pH) in yeast and identification of acidification-defective mutants.

Proc. Natl. Acad. Sci. USA 86, 7027–7031.

Price, A., Seals, D., Wickner, W., and Ungermann, C. (2000a). The

docking stage of yeast vacuole fusion requires the transfer of proteins

from a cis-SNARE complex to a Rab/Ypt protein. J. Cell Biol. 148,

1231–1238.

Price, A., Wickner, W., and Ungermann, C. (2000b). Proteins needed

for vesicle budding from the Golgi complex are also required for the

docking step of homotypic vacuole fusion. J. Cell Biol. 148, 1223–

1230.

Raymond, C.K., Howald-Stevenson, I., Vater, C.A., and Stevens, T.H.

(1992). Morphological classification of the yeast vacuolar protein sort-

ing mutants: Evidence for a prevacuolar compartment in class E vps

mutants. Mol. Biol. Cell 3, 1389–1402.

Raymond, C.K., O’Hara, P.J., Eichinger, G., Rothman, J.H., and

Stevens, T.H. (1990). Molecular analysis of the yeast VPS3 gene and

the role of its product in vacuolar protein sorting and vacuolar segrega-

tion during the cell cycle. J. Cell Biol. 111, 877–892.

Rehling, P., Darsow, T., Katzmann, D.J., and Emr, S.D. (1999). Forma-

tion of AP-3 transport intermediates requires Vps41 function. Nat. Cell

Biol. 1, 346–353.

Richardson, S.C., Winistorfer, S.C., Poupon, V., Luzio, J.P., and Piper,

R.C. (2004). Mammalian late vacuole protein sorting orthologues par-

ticipate in early endosomal fusion and interact with the cytoskeleton.

Mol. Biol. Cell 15, 1197–1210.

opmental Cell 12, 739–750, May 2007 ª2007 Elsevier Inc. 749

Developmental Cell

Endolysosomal Biogenesis Linked to the CORVET Complex

Rieder, S.E., and Emr, S.D. (1997). A novel RING finger protein com-

plex essential for a late step in protein transport to the yeast vacuole.

Mol. Biol. Cell 8, 2307–2327.

Rink, J., Ghigo, E., Kalaidzidis, Y., and Zerial, M. (2005). Rab conver-

sion as a mechanism of progression from early to late endosomes.

Cell 122, 735–749.

Rothman, J.H., Yamashiro, C.T., Raymond, C.K., Kane, P.M., and

Stevens, T.H. (1989). Acidification of the lysosome-like vacuole and

the vacuolar H+-ATPase are deficient in two yeast mutants that fail

to sort vacuolar proteins. J. Cell Biol. 109, 93–100.

Sacher, M., Jiang, Y., Barrowman, J., Scarpa, A., Burston, J., Zhang,

L., Schieltz, D., Yates, J.R., 3rd, Abeliovich, H., and Ferro-Novick, S.

(1998). TRAPP, a highly conserved novel complex on the cis-Golgi

that mediates vesicle docking and fusion. EMBO J. 17, 2494–2503.

Seals, D.F., Eitzen, G., Margolis, N., Wickner, W.T., and Price, A.

(2000). A Ypt/Rab effector complex containing the Sec1 homolog

Vps33p is required for homotypic vacuole fusion. Proc. Natl. Acad.

Sci. USA 97, 9402–9407.

Singer-Kruger, B., Stenmark, H., Dusterhoft, A., Philippsen, P., Yoo,

J.S., Gallwitz, D., and Zerial, M. (1994). Role of three Rab5-like

GTPases, Ypt51p, Ypt52p, and Ypt53p, in the endocytic and vacuolar

protein sorting pathways of yeast. J. Cell Biol. 125, 283–298.

750 Developmental Cell 12, 739–750, May 2007 ª2007 Elsevier

Srivastava, A., Woolford, C.A., and Jones, E.W. (2000). Pep3p/Pep5p

complex: A putative docking factor at multiple steps of vesicular

transport to the vacuole of Saccharomyces cerevisiae. Genetics 156,

105–122.

Stroupe, C., Collins, K.M., Fratti, R.A., and Wickner, W. (2006). Purifi-

cation of active HOPS complex reveals its affinities for phosphoinosi-

tides and the SNARE Vam7p. EMBO J. 25, 1579–1589.

Subramanian, S., Woolford, C.A., and Jones, E.W. (2004). The Sec1/

Munc18 protein, Vps33p, functions at the endosome and the vacuole

of Saccharomyces cerevisiae. Mol. Biol. Cell 15, 2593–2605.

Wada, Y., Ohsumi, Y., and Anraku, Y. (1992). Genes for directing vac-

uolar morphogenesis in Saccharomyces cerevisiae. I. Isolation and

characterization of two classes of vam mutants. J. Biol. Chem. 267,

18665–18670.

Wang, W., Sacher, M., and Ferro-Novick, S. (2000). TRAPP stimulates

guanine nucleotide exchange on Ypt1p. J. Cell Biol. 151, 289–296.

Whyte, J.R., and Munro, S. (2002). Vesicle tethering complexes in

membrane traffic. J. Cell Sci. 115, 2627–2637.

Wurmser, A.E., Sato, T.K., and Emr, S.D. (2000). New component of

the vacuolar class C-Vps complex couples nucleotide exchange on

the Ypt7 GTPase to SNARE-dependent docking and fusion. J. Cell

Biol. 151, 551–562.

Inc.

1

Developmental Cell, Vol. 12

Supplemental Data

The CORVET Tethering Complex Interacts

with the Yeast Rab5 Homolog Vps21

and Is Involved in Endo­Lysosomal Biogenesis Karolina Peplowska, Daniel F. Markgraf, Clemens W. Ostrowicz, Gert Bange, and Christian Ungermann

Supplemental Experimental Procedures

Yeast strains and plasmids

To generate plasmids carrying GFP­tagged proteins, a BamH1­BglII fragment from pBS­eGFP

(provided by E. Hurt, Heidelberg, Germany) was inserted into a BamH1 site of the indicated plasmid.

Plasmids were either genomically integrated by cutting with Bsp119I (pRS406­NOP1pr­GFP­Vam3),

or maintained within cells by growth on selective medium (pRS415­NOP1pr­GFP­VPS41, pRS416­

NOP1pr­GFP­YCK3, pRS426­NOP1pr­VAC8­GFP). C­terminal tagging of VPS3 and VPS8 in the

indicated strains was done by integrating a PCR­amplified region coding for the TAP­tag (pYM13) or

the GFP­tag (pYM12) and the kanamycin marker (kindly provided by M. Knop, Heidelberg, Germany)

via homologous recombination. TAP­tagging of VPS41 or VPS3 was done similarly, using pBS1539 as

a template (Puig et al., 1998). VPS3, VPS8, VPS9, or the Rab GTPases VPS21, YPT7 and YPT1 were

placed under the control of the GAL1­promoter using PCR fragments containing flanking regions of

the respective genes amplified from pFA6a­HIS3MX6­GAL1pr (VPS3), pFA6a­HIS3MX6­GAL1pr­

3xHA (VPS8), pBS1761­TRP­GAL1pr­TAP (VPS3, VPS9), or pFA6­kanMX6­GAL1pr­GST (VPS21,

YPT7, YPT1) (Longtine et al., 1998). PHO8 and VPS21 were genomically tagged at the N­terminus

using a URA3­PHO5pr­GFP­Myc cassette, amplified from plasmid pGL (a gift from S. Munro, MRC,

Cambridge, UK; Levine and Munro, 2001). VPS21, YPT7, and YPT1 were cloned into pGEX4T3 or

pGEX2T (GE Healthscience) and purified according to the manufacturer.

2

Yeast cell lysis

After overnight growth in rich medium containing 2% glucose (YPD) or 2% galactose (YPG), cell

cultures were diluted to OD600=0.5 and incubated for 2 hours in 30°C. Cells (30 OD600 units) were

collected, washed once with DTT buffer (10 mM DTT, 0.1 M Tris/HCl pH 9.4), resuspended in 1 ml of

DTT buffer and incubated for 10 minutes in 30°C. They were then centrifuged (2 min at 4620g),

resuspended in 300 µl of spheroplasting buffer (0.16x YPD, 50 µM KPi buffer, pH 7.4, 0.6 M sorbitol),

and incubated for 20 min at 30°C in the presence of lyticase. Cells were centrifuged for 3 min at 1530g,

the pellet was resuspended in 1ml of lysis buffer (0.2 M sorbitol, 150 mM KCl, 20 mM HEPES/KOH,

pH 6.8, 1 mM DTT, 1 mM PMSF, 1xPIC) supplemented with 6 µl of 0.4 mg/ml DEAE­dextran, and

incubated for 5 minutes on ice. Samples were briefly heat­shocked (2 min/30°C), and unlysed cells

were removed by centrifugation at 300g for 3 min. The cell lysate was used in further experiments

(pull­downs, subcellular fractionations).

GST pull down

Cells were grown overnight in the presence of galactose to overproduce GST­Rab protein

(Vps21, Ypt7 or Ypt1) together with TAP­tagged Vps3 and Vps9. 200 ODs of cells were collected,

washed once with 1 ml of buffer A (20 mM Tris/HCl, pH 7.4, 150 mM NaCl, 5 mM MgCl2) and lysed

with glass beads in a presence of 300 µl of buffer A containing1xPIC and 1 mM PMSF. Lysis was

repeated twice and each time 300 µl of lysate were collected. The lysate (25 mg) was supplemented

with 0.5% Triton X­100, centrifuged (30 min, 100,000g, 4°C), and then loaded onto 50 µl equilibrated

glutathione (GSH) beads. An aliquot of the lysate (0.1%) was removed as a loading control. Beads

were incubated at 4°C for 1.5 hour and then washed extensively (2 x 15 min with buffer A + 0.1%

Triton X­100 and 2 x 15 min with buffer A + 0.025% Triton X­100). Proteins were eluted by

3

incubating the beads for 20 min at room temperature in 600 µl elution buffer (20 mM Tris/HCl, pH 7.4,

1.5 M NaCl, 20 mM EDTA), TCA­precipitated, and analyzed by using 7.5% SDS­PAGE gels and

Western blotting.

References:

Longtine, M. S., McKenzie, A., 3rd, Demarini, D. J., Shah, N. G., Wach, A., Brachat, A., Philippsen, P., and Pringle, J. R. (1998). Additional modules for versatile and economical PCR­based gene deletion and modification in Saccharomyces cerevisiae. Yeast 14, 953­961.

Puig, O., Rutz, B., Luukkonen, B. G., Kandels­Lewis, S., Bragado­Nilsson, E., and Seraphin, B. (1998). New constructs and strategies for efficient PCR­based gene manipulations in yeast. Yeast 14, 1139­1146.

4

Figures

Figure S1. Characterization of vps3∆ mutants.

(A) Osmotic stress response. Logarithmically grown wt and vps3∆ cells were stained with FM4­64,

reisolated, incubated in YPD medium containing 0.4 M NaCl, and analyzed by fluorescence

microscopy after 10 and 60 minutes. Control cells did not receive salt. Quantification of at least 200

cells per condition is shown.

(B) Vacuole morphology of vps3∆ cells. BY4741 cells (wt,vps3∆ or vps8∆) were grown

logarithmically in YPD medium. To visualize vacuoles, cells were incubated for 30 minutes (pulse)

with 25 µM CMAC dye (7­amino­4­chloromethylcoumarin) , washed, and incubated for another 30

minutes (chase) in fresh YPD before being analyzed by fluorescence microscopy.

Figure S2. Characterization of the Vps3 protein.

Localization of the Vps3 protein. BY4741 cells expressing Vps3­GFP were analyzed by fluorescence

microscopy.

Figure S3. Alignment of Vps3 and Vam6 proteins from yeast and human.

The alignment was done with Jalview (Clamp, M., Cuff, J., Searle, S. M. and Barton, G. J. (2004),

"The Jalview Java Alignment Editor", Bioinformatics, 12, 426­7.).

5

Figure S4: Co­overexpression of Vps3 and Vps21.

Cells co­overexpressing the indicated Rab­GTPase in the wt or GDP­locked form (S21N) and TAP­

tagged Vps3 (3) or Vps9 (9) were grown overnight and processed for the GSH pull­down as described

in Experimental Procedures. Specifically bound proteins were eluted and analyzed by SDS­PAGE and

Western blotting. The lower band in lane 3 is most likely a degradation product of Vps3 that seems to

bind efficiently to Vps21­S21N. Note that Vps3 was overproduced more strongly in the absence of any

GTPase (lane 6), similarly Vps9 showed stronger expression (lane 10). A lower exposure of the load is

shown in the bottom right panel.

(C) Rab­GTPase present on the remaining GSH beads. Beads were boiled in SDS­sample buffer and

analyzed as above. Note that Vps21­S21N is poorly recovered on GSH beads, potentially due to its

decreased stability caused by the mutation.

(D) Expression of the fusion proteins in vivo. Whole cell extracts were prepared from galactose­

induced cultures as described in Experimental Procedures, proteins were resolved by SDS­PAGE and

analyzed by Western blotting using anti­GST and ProteinA­peroxidase coupled antibodies.

Figure S5. Accumulation of GFP­Vps21 upon Vps8 overexpression.

The experiment was done as in Figure 3F, and shows representative fields. Size bar is 10 µm.

Figure S6. Vps3 interaction with the Class C complex.

Purification of Vps3­TAP from vps11∆ cells was done as described above. Eluted proteins were

analyzed by SDS­PAGE and Western blotting as described before.

6

7

8

9

10

11

12

Table S1. Strains used in this study

Strain Genotype Reference CUY1 BJ3505;MATa pep4::HIS3 prb1­∆1.6R HIS3 lys2­208 trp1∆101 ura3­52 gal2 can Haas et al., 1994 CUY476 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 Euroscarf library CUY473 BY4741;MATa his3∆ leu2∆ met15∆ ura3∆ vps8∆::kanMX Euroscarf library CUY765 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vam3∆::kanMX This study CUY865 CUY476; pRS426­NOP1pr­VAC8­GFP Subramanian et al., 2006 CUY887 BY4741;MATa pRS415­NOP1pr­GFP­VPS41 This study CUY959 CUY476; PHO8::HIS5­PHO5pr­GFP­MYC LaGrassa and Ungermann, 2005 CUY1014 BY4741;MATa his3∆ leu2∆ met15∆ ura3∆ vps3∆::kanMX Euroscarf library CUY1616 BJ3505; VPS3::TAP­kanMX This study CUY1792 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 VPS8::TAP­kanMX This study CUY1794 CUY476; VPS3::GFP­kanMX This study CUY1795 CUY476; VPS3::TAP­URA3 This study CUY1796 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vps41∆::kanMX VPS3::TAP­URA3 This study CUY1797 CUY473; VPS3::TAP­URA3 This study CUY1798 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vam6∆::kanMX VPS3::TAP­URA3 This study CUY1799 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vps33∆::kanMX VPS3::TAP­URA3 This study CUY1800 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 VPS41::TAP­URA3 This study CUY1801 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vam6∆::kanMX VPS41::TAP­URA3 This study CUY1802 CUY1014; VPS41::TAP­URA3 This study CUY1803 CUY473; VPS41::TAP­URA3 This study CUY1804 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vps33∆::kanMX VPS3::TAP­URA3 This study CUY1805 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 ypt7∆::kanMX VPS3::TAP­URA3 This study CUY1806 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vps21∆::kanMX VPS3::TAP­URA3 This study CUY1819 CUY473; VPS21::URA3­PHO5pr­GFP This study CUY1820 BY4741;MATa his3∆ leu2∆ met15∆ ura3∆ vam6∆::kanMX VPS21::GFP­URA3 This study CUY1826 CUY1014; VPS21:: URA3­PHO5pr GFP­MYC This study CUY1836 CUY1014; PHO8::URA3­PHO5pr­GFP­MYC This study CUY1837 CUY1014; pRS426­NOP1pr­GFP­VAC8 This study CUY1838 CUY476; URA3::pRS406­NOP1pr­GFP­VAM3 This study CUY1839 CUY1014; URA3::pRS406­NOP1pr­GFP­VAM3 This study CUY1841 CUY476; pRS416­NOP1pr­GFP­YCK3 (URA3) This study CUY1842 CUY1014; pRS416­NOP1pr­GFP­YCK3 (URA3) This study CUY1845 CUY1794; vps8∆::URA3 This study CUY1847 CUY1800; VPS3::HIS3­GAL1pr This study CUY1849 CUY1014; pRS415­NOP1pr­GFP­VPS41 (LEU2) This study CUY1850 BY4741;MATa his3∆ leu2∆ met15∆ ura3∆ vps8∆::kanMX VAM3::URA3 This study CUY1877 BY4741;MATa his3∆ leu2∆ met15∆ ura3∆ VPS8::TAP­kanMX vps33∆::URA3 This study CUY1878 BY4741;MATa his3∆ leu2∆ met15∆ ura3∆ VPS3::TAP­URA3 vps11∆::kanMX This study CUY1883 BY4741;MATa his3∆ leu2∆ met15∆ ura3∆ VPS41::TAP­kanMX VPS8::HIS3­GAL1pr­3HA This study CUY1895 BY4733;MATalpha; his3∆200 leu2∆0 met15∆0 trp1∆63 ura3∆0 VPS3::TRP1­GAL1pr­TAP This study CUY1897 BY4741;MATa his3∆ leu2∆ met15∆ ura3∆ vam6∆::kanMX VPS8::HIS3­GAL1pr­3HA

VPS41::TAP­URA3 This study

CUY1915 BY4741;MATa his3∆1 leu2∆0 met15∆0 ura3∆0 VPS3::kanMX­GAL1pr This study CUY1936 CUY1895; VPS21::kanMX­GAL1pr­GST This study CUY1949 CUY1895; VPS21 S21N::kanMX­GAL1pr­GST This study CUY1948 CUY1895; YPT1::kanMX­GAL1pr­GST This study CUY1952 BY4719;MATa trp1∆63 ura3∆0 VPSs8::kanMX VPS21::URA­PHO5pr­GFP pV2­RFP­YPT7 (TRP1) This study CUY1960 BY4719;MATa trp1∆63 ura3∆0 vps8∆::kanMX VPS21::URA­PHO5pr­GFP pV2­RFP­YTP7(TRP1) This study CUY1967 BY4719;MATa trp1∆63 ura3∆0 VPS21::URA3­PHO5pr­GFP This study CUY1969 BY4733;MATalpha; his3∆200 leu2∆0 met15∆0 trp1∆63 ura3∆0 VPS9::TRP1­GAL1­TAP

VPS21::kanMX­GAL1pr­GST This study

CUY1972 BY4719;MATa trp1∆63 ura3∆0 VPS21::URA3­PHO5pr­GFP pV2­RFP­YPT7 (TRP1) This study CUY2251 BY4733;MATalpha his3∆200 leu2∆0 met15∆0 trp1∆63 ura3∆0 VAM6::kanMX­GAL1pr

VPS21:: URA3­PHO5pr­GFP This study

CUY2252 CUY1915; VPS21::URA3­ PHO5pr­GFP This study CUY2253 BY4733;MATalpha his3∆200 leu2∆0 met15∆0 trp1∆63 ura3∆0 VPS8::HIS3­GAL1pr­3HA

VPS21:: URA3­PHO5pr­GFP This study

CUY2359 BY4719;MATa trp1∆63 ura3∆0 vps8∆::kanMX VPS3::TRP1­GAL1pr­TAP This study CUY2278 CUY1826; VPS8::HIS3­GAL1pr­3HA This study