Báo cáo khoa học: Vps4 regulates a subset of protein interactions at the multivesicular endosome doc

Vps4 regulates a subset of protein interactions at the
multivesicular endosome
Parimala R. Vajjhala
1
, Elizabeth Catchpoole
1
, Chau H. Nguyen
1
, Carol Kistler
1
and Alan L. Munn
1,2
1 Institute for Molecular Bioscience and ARC Special Research Centre for Functional and Applied Genomics, University of Queensland,
St Lucia, QLD, Australia
2 School of Biomedical Sciences, University of Queensland, St Lucia, QLD, Australia
During endocytic trafficking, some integral membrane
proteins are sorted into internal vesicles which form by
invagination of the endosome limiting membrane. This
process, referred to as multivesicular body (MVB) sort-
ing, is critical for a number of important biological
processes including receptor down-regulation, antigen
presentation and exosome-dependent intercellular sig-
nalling (reviewed in [1–3]). Interest in the mechanism
of MVB sorting has escalated since the discovery that
components of the MVB sorting machinery are also
utilized for virus budding, a process topologically sim-
ilar to MVB sorting (reviewed in [4,5]).
The process of MVB sorting was first examined in
mammalian cells [6,7], but the components of the
MVB sorting machinery were first characterized in
Saccharomyces cerevisiae [8–10]. The pathway is highly

conserved from yeast to mammalian cells, although the
number of components is expanded in mammalian
cells because of the multiplicity of isoforms [11].
Recognition of a cargo protein, usually by the presence
Keywords
endocytosis; lysosome; macromolecular
disassembly; membrane traffic; vacuole
Correspondence
A. L. Munn, Institute for Molecular
Bioscience, University of Queensland,
St Lucia, Brisbane, QLD 4072, Australia
Fax: +61 73346 2101
Tel: +61 73346 2017
E-mail:
(Received 17 December 2006, revised 6
February 2007, accepted 9 February 2007)
doi:10.1111/j.1742-4658.2007.05736.x
During endocytic transport, specific integral membrane proteins are sorted
into intraluminal vesicles that bud from the limiting membrane of the
endosome. This process, known as multivesicular body (MVB) sorting, is
important for several important biological processes. Moreover, compo-
nents of the MVB sorting machinery are implicated in virus budding. Dur-
ing MVB sorting, a cargo protein recruits components of the MVB sorting
machinery from cytoplasmic pools and these sequentially assemble on the
endosome. Disassembly of these proteins and recycling into the cytoplasm
is critical for MVB sorting. Vacuolar protein sorting 4 (Vps4) is an AAA
(ATPase associated with a variety of cellular activities) ATPase which has
been proposed to play a critical role in disassembly of the MVB sorting
machinery. However, the mechanism by which it disassembles the complex
is not clear. Vps4 contains an N-terminal microtubule interacting and traf-

ficking (MIT) domain, which has previously been shown to be required for
recruitment to endosomes, and a single AAA ATPase domain, the activity
of which is required for Vps4 function. In this study we have systematically
characterized the interaction of Vps4 with other components of the MVB
sorting machinery. We demonstrate that Vps4 interacts directly with Vps2
and Bro1. We also show that a subset of Vps4 interactions is regulated by
ATP hydrolysis, and one interaction is regulated by ATP binding. Finally,
we show that most proteins interact with the Vps4 MIT domain. Our stud-
ies indicate that the MIT domain has a dual role in substrate binding and
recruitment to endosomes and indicate that Vps4 disassembles the MVB
sorting machinery by direct effects on multiple proteins.
Abbreviations
GFP, green fluorescent protein; GST, glutathione S-transferase; HRP, horseradish peroxidase; MIT, microtubule interacting and trafficking;
MVB, multivesicular body; PVDF, poly(vinylidene difluoride); Vps, vacuolar protein sorting.
1894 FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS
of a ubiquitin molecule, is followed by sequential
recruitment of components of the MVB sorting
machinery. The order of recruitment of the different
components to the endosome membrane is starting to
emerge, and structural data have recently been
obtained for several components (reviewed in [12]). A
critical process during MVB sorting is the disassembly
of the MVB sorting machinery, which allows recycling
and new rounds of vesicle budding. However, the
molecular mechanisms that regulate cycling of the
MVB sorting machinery on and off endosomes is not
yet well understood.
Vps4p, also known as Csc1p, End13p, Grd13p,
Vpl4p, Vpt10p, or Did6p, is the only essential compo-
nent of the MVB sorting machinery with known enzy-

matic activity. It is a member of the AAA (ATPase
associated with a variety of cellular activities) family of
ATPases [13,14]. Mammalian cells express two iso-
forms of VPS4, VPS4A and VPS4B, and both proteins
function in endocytic trafficking [15–17]. All members
of the AAA superfamily contain one or two copies of
a conserved ATPase domain (AAA module). Although
not known for Vps4p, other AAA ATPases assemble
into oligomeric rings. Distinct members of the AAA
ATPase family function in diverse cellular processes,
but a common theme is protein unfolding and macro-
molecular disassembly (reviewed in [18,19]).
Loss of Vps4p function in yeast and mammalian
cells disrupts MVB sorting and results in the formation
of an aberrant multilamellar endosomal compartment
referred to as the class E compartment [8,14]. As both
endocytic and biosynthetic traffic to the lysosome ⁄
vacuole proceeds via the MVB, the class E compart-
ment accumulates endocytic and biosynthetic material
as well as late Golgi proteins because of defective traf-
ficking out of this compartment [9]. In yeast, defective
recycling of late Golgi proteins including the receptor
that transports soluble vacuolar proteins from the
Golgi to the MVB results in missorting and secretion
of soluble vacuolar proteins to the extracellular medium
[20,21].
Loss of Vps4p function is also accompanied by the
redistribution of several components of the MVB sort-
ing machinery from the cytoplasm to endosomes [22,23].
Therefore Vps4p-dependent ATPase activity has been

proposed to be important for disassembly of the MVB
sorting machinery and release into the cytoplasm. An
N-terminal microtubule interacting and trafficking
(MIT) domain is required for recruitment of Vps4p to
endosomes [22,24,25], but it is not clear precisely how
Vps4p disassembles the MVB sorting machinery. Previ-
ous work from our laboratory has shown that Vps20p, a
component of the MVB sorting machinery, interacts
with Vps4p and dissociates from Vps4p upon ATP
hydrolysis [26]. This was the first evidence that Vps4p
ATPase activity can break intermolecular interactions.
As Vps20p is a coiled-coil protein and interacts with the
N-terminal MIT domain of Vps4p, Vps4p ATPase
activity may break coiled-coil interactions and thereby
disassemble the MVB sorting machinery.
Several putative interactions have been reported
between Vps4p and components of the MVB sorting
machinery [27–30]. However, as there is a complex net-
work of interactions between the components of the
MVB sorting machinery, it was not clear which inter-
actions with Vps4p are direct and which are indirect.
In addition, it was not clear how many of these puta-
tive Vps4p interactions with the MVB sorting machin-
ery may be regulated by Vps4p ATPase activity. That
not all Vps4p interactions are regulated by ATPase
activity is supported by our previous studies showing
that the Vps4p interaction with Vta1p is not affected
by ATP hydrolysis [26].
Here, we demonstrate new direct interactions
between Vps4p and the MVB sorting machinery in

yeast. We present evidence that a subset of Vps4p
interactions is regulated by ATP hydrolysis, and that
one interaction is regulated by ATP binding to Vps4p.
Finally, we also show that most Vps4p interactors
interact with the MIT domain of Vps4p.
Results
Vps4p binds directly to multiple components
of the MVB sorting machinery
Vps4p has been shown to bind directly to a few MVB
sorting machinery components including Vps20p,
Vta1p and Did2p ⁄ Chm1p [26,31]. However, it is not
known whether the function of Vps4p in disassembly
of the MVB sorting machinery is mediated solely via
interactions with these proteins or whether Vps4p
binds directly to and acts via other components of the
MVB sorting machinery. We therefore tested candidate
Vps4p interactions with components of the MVB sort-
ing machinery. When binding of purified Vps4p to
MVB components was examined in vitro, Vps4p was
found to bind directly to multiple MVB sorting
machinery components, including Vps2p and Bro1p, in
addition to Vps20p, Vta1p and Did2p, but not to
Snf7p (Fig. 1). Our data also show that, when molar
equivalent amounts of Vps4p interactors are com-
pared, the amount of Vps4p that binds to Did2p is
greater than that bound by any of the other interac-
tors. These data are consistent with Vps4p having a
relatively high affinity for Did2p.
P. R. Vajjhala et al. Vps4 function at the multivesicular endosome
FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS 1895

ATP binding and ATP hydrolysis by Vps4p
regulates protein interactions at the MVB
Vps4p has been proposed to function as a protein
complex disaggregation machine on endosomes [22].
Consistent with this, the binding of Vps20p to Vps4p
is regulated by ATP hydrolysis [26]. It was likely that
other components of the MVB sorting machinery are
also substrates for disassembly by Vps4p. To test this,
we performed the in vitro binding assay in the presence
and absence of ATP. In the presence of ATP, purified
wild-type Vps4 is catalytically active and will hydrolyse
added ATP. Thus interactions that are regulated by
Vps4p ATPase activity are predicted to decrease under
these conditions. In contrast, interactions that are not
regulated by Vps4p ATPase activity are predicted to
be unaffected. The data obtained show that binding of
both Vps2p and Bro1p to Vps4p was decreased in the
presence of ATP compared with binding in the absence
of ATP. However, the binding of Did2p to Vps4p was
not affected by the presence of ATP (Fig. 2).
To determine whether the decreased binding in the
presence of ATP is due to ATP hydrolysis or ATP
binding, the effect of ATP on binding to a Vps4p
mutant (Vps4p–E233Q) that is defective in ATP
hydrolysis was also studied. Binding of Vps2p and
Vps20p to Vps4p–E233Q was increased in the presence
of ATP, but binding of Bro1p to Vps4p–E233Q was
decreased in the presence of ATP (Fig. 2).
We surmise that the decreased binding of Vps2p to
Vps4p in the presence of ATP is due to Vps4p-depend-

ent ATP hydrolysis. In contrast, the decreased binding
of Bro1p to wild-type Vps4p and to Vps4p–E233Q in
the presence of ATP may be due to competitive bind-
ing or an allosteric effect.
Several components of the MVB sorting
machinery interact with Vps4p via the N-terminal
MIT domain
To determine whether there is any correlation between
the binding sites on Vps4p and the response of the
interacting proteins to ATP binding and hydrolysis, we
determined the region of Vps4p that mediates inter-
action with Vps2p, Snf7p and Bro1p using the yeast
two-hybrid technique (Fig. 3). Both Vps2p and Snf7p
interact with Vps4p mainly via the Vps4p N-terminal
coiled-coil domain (Fig. 3C). We did not detect an
interaction between Bro1p and full-length Vps4p or
any of the Vps4p domains using the yeast two-hybrid
technique (not shown) consistent with a previous
report [27].
To more precisely map the interaction sites of
Vps2p, Snf7p, Vps20p, and Did2p within the Vps4p
N-terminal domain, we generated two Vps4p N-ter-
minal mutants (Fig. 4A,B) and tested the effect of
Fig. 1. All of the GST-tagged Vps4p-interacting proteins that func-
tion during MVB sorting except Snf7p can bind directly to Vps4p.
An equal amount of 6His-tagged Vps4p was incubated with gluta-
thione–agarose bearing GST–Snf7p, GST–Vps20p, GST–Bro1p,
GST–Did2p, GST–Vps2p, GST–Vta1p or GST alone. Unbound pro-
tein was recovered in the supernatants. Bound protein was
released with Laemmli sample buffer. The bound and unbound frac-

tions were subjected to SDS ⁄ PAGE and immunoblotting with a
polyclonal antibody to Vps4p. The shift in the position of Vps4p-
6His bound to GST–Vps20p is due to the presence of the GST–
Vps20p, which migrates very close to Vps4p-6His. The data shown
are representative of at least two independent experiments.
Fig. 2. ATP binding and ATP hydrolysis by Vps4p regulates interaction with Bro1p, Vps2p and Did2p. The 6His-tagged wild-type Vps4p or
Vps4p–E233Q was incubated with glutathione–agarose bearing GST–Vps2p, GST–Bro1p, GST–Did2p or GST–Vps20p in the presence or
absence of ATP. The bound protein was released with Laemmli sample buffer and subjected to Western blotting using a polyclonal antibody
to Vps4p. The data shown are representative of at least two independent experiments performed in duplicate.
Vps4 function at the multivesicular endosome P. R. Vajjhala et al.
1896 FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS
these mutations on the different Vps4p interactions. In
the Vps4p–YEE mutant, residues 26–29 in the second
helix of the MIT domain were deleted. These residues
are completely conserved between the yeast and human
VPS4 isoforms. In the Vps4p–IRA mutant, residues
56–71 in the third helix of the MIT domain were dele-
ted. These residues are also highly conserved. The
YEE and IRA mutants are named after the first three
amino acids that were deleted in each motif. We also
tested the effect of a previously described Vps4p-
coiled-coil (CC) mutation [22] in which residues 50–87,
which comprises most of the second and third helices
of the MIT domain, were deleted.
Yeast two-hybrid analysis (Fig. 4C) revealed that
each mutation diminished but did not abolish Did2p
interaction, whereas Vps2p interaction was abolished
by the Vps4p–IRA and Vps4p-CC mutations. All the
N-terminal mutations tested abolished Vps20p and
Snf7p interactions. As expected, the interaction of

Vta1p with the Vps4p C-terminal domain was not
diminished by any of the N-terminal mutants we
tested.
To confirm the yeast two-hybrid interactions des-
cribed above and to identify the domain of Vps4p to
which Bro1p binds, we performed in vitro protein-
binding assays (Fig. 4D). The data obtained indicate
that the Vps4p–YEE and Vps4p–IRA mutations
diminish direct binding of Vps2p, Vps20p and Did2p.
The Vps4p-CC mutation appeared to increase binding
to all interactors. However, the Vps4p-CC mutant pro-
tein also displayed an interaction with glutathione
S-transferase (GST) alone (Fig. 4D) and displayed an
increased interaction with the Vps4p antibody (not
shown). Thus, we cannot interpret the data obtained
for this Vps4p-CC mutant protein. Mutation of the
b domain abolished interaction with Vta1p as previ-
ously reported [32,33] and in addition increased bind-
ing to the N-terminal interactors, including Vps2p,
Vps20p and Did2p. None of the N-terminal MIT
domain mutations or the C-terminal b domain muta-
tions diminished binding of Vps4p to Bro1p.
In summary, our data indicate that most Vps4p-
interacting proteins (Vps2p, Vps20p, Did2p and Snf7p)
interact with the N-terminal MIT domain of Vps4p. In
addition, these data show that the Bro1p interaction
with Vps4p is unique as it is undiminished by muta-
tions in the N-terminal MIT domain or b domain.
Vps4p interactions with Did2p, Vps2p and
Vps20p are important for recruitment to

endosomes and for MVB sorting
To determine whether the Vps4p YEE and IRA motifs
play a role in Vps4p recruitment to endosomes, wild-
type and mutant Vps4p tagged with green fluorescent
protein (GFP) were expressed in cells in which the
chromosomal VPS4 gene has been deleted (vps4D).
Wild-type GFP-tagged Vps4p could be detected on
punctate structures (Fig. 5A) consistent with localiza-
tion to endosomes, as previously reported [22]. How-
ever, the GFP-tagged Vps4p–YEE and Vps4p–IRA
mutant proteins, like the GFP-tagged Vps4p-CC
mutant protein, exhibited severely reduced punctate
localization. We conclude that the Vps4p YEE and
IRA motifs are important for Vps4p recruitment to
endosomes.
To test whether the YEE and IRA motifs and the
region deleted in the Vps4p-CC mutant protein are
important for Vps4p function in vivo, we tested the
ability of the N-terminal mutant proteins to restore
A
B
C
Fig. 3. Vps2p and Snf7p interact with the N-terminal domain of
Vps4p. (A) Schematic representation of Vps4p with the domain
organization inferred from structural data from mammalian VPS4A
and VPS4B. (B) Constructs used for mapping the region of Vps4p
that mediates interaction. (C) Yeast two-hybrid interaction analysis
of Vps2p and Snf7p with full-length wild-type Vps4p (Vps4p-full),
the N-terminal region of Vps4p (Vps4p-N), the previously predicted
AAA domain (Vps4p-AAA), and the C-terminal region (Vps4p-C).

EGY48 carrying pLexA-based bait plasmids and pB42AD-based prey
plasmids as well as the p8OpLacZ reporter plasmid were spotted
on to medium containing X-gal. Plates were photographed after
overnight incubation, and two-hybrid interaction was assessed by
blue coloration. Three independent transformants are shown.
P. R. Vajjhala et al. Vps4 function at the multivesicular endosome
FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS 1897
MVB sorting and delivery of a soluble vacuolar pro-
tein to the vacuole (vacuolar protein sorting) in vps4D
yeast. As a marker for MVB sorting, we used a GFP-
and ubiquitin-tagged form of the iron transporter
homologue, Fth1p (GFP-Fth1p-Ub), which is known
to undergo MVB sorting into the vacuole lumen [34].
In vps4D yeast expressing wild-type Vps4p, the GFP-
Fth1p-Ub undergoes MVB sorting and is transported
to the vacuole lumen (Fig. 5B). However, in vps4D
yeast expressing the Vps4p–YEE, Vps4p–IRA or
Vps4p-CC mutant proteins, MVB sorting of GFP-
Fth1p-Ub was not significantly improved compared
with vps4D cells carrying empty vector. To assess the
ability of the Vps4p mutant proteins to restore trans-
port of a soluble vacuolar protein to the vacuole in
vps4D yeast, we tested their ability to correct the mis-
sorting and secretion of a soluble vacuolar protein,
carboxypeptidase Y. Expression of wild-type Vps4p,
but not Vps4p–YEE, Vps4p–IRA or Vps4p-CC mutant
proteins restored vacuolar transport of carboxy-
B
C
D

A
Fig. 4. Interaction with Vps2p, Snf7p, Vps20p and Did2p, but not with Bro1p, is diminished by mutation of conserved residues in the Vps4p
MIT domain. (A) Alignment of human VPS4A, VPS4B and S. cerevisiae (Sc) Vps4p sequences using
CLUSTAL W [50]. The conserved YEE and
IRA motifs that were deleted are shown in bold, and the previously described coiled-coil (CC) mutation is shown underlined. (B) Location in
the VPS4A MIT domain of the conserved YEE and IRA motifs and the region deleted in the previously described Vps4p-CC mutant. (C) Yeast
two-hybrid interaction analysis of wild-type (WT) Vps4p and Vps4p N-terminal mutants with Vps2p, Snf7p, Vps20p, Did2p, and Vta1p. Inter-
action analysis was performed as described in the legend to Fig. 3. Three independent transformants are shown. (D) In vitro binding of 6His-
tagged wild-type Vps4p and Vps4p N-terminal mutants to GST-tagged Vps2p, Vps20p, Did2p, Bro1p, and Vta1p, or to GST alone. Equal
amounts of full-length 6His-tagged proteins were incubated with glutathione–agarose beads bearing the different GST fusion proteins or GST
alone. Bound protein was released with Laemmli sample buffer and subjected to Western blotting using a polyclonal antibody to Vps4p. The
data shown are representative of two independent experiments. A Western blot of the different 6His-tagged proteins used for the in vitro
binding assay (5% input) is also shown.
Vps4 function at the multivesicular endosome P. R. Vajjhala et al.
1898 FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS
peptidase Y compared with vps4D cells transformed
with empty vector alone (Fig. 5C). We conclude that
the YEE and IRA motifs as well as the region deleted
in the Vps4p-CC mutant are important for Vps4p
function in MVB sorting and vacuolar protein sorting.
To test whether the phenotypes of vps4D cells
expressing the Vps4p N-terminal mutants were due to
lowered expression or degradation of the mutant pro-
teins, we tested the steady-state expression of the
mutant Vps4p proteins (Fig. 5D). Although both the
Vps4p–IRA and Vps4p–YEE mutants are expressed,
their steady-state connections are somewhat reduced
compared with that of wild-type Vps4p. However, such
a modest reduction in expression level is unlikely to
account for the inability of these mutant proteins to

restore MVB sorting and vacuolar protein sorting
in vps4D yeast. Surprisingly, the expression of the
Vps4p-CC mutant was significantly greater than that
C
AB
D
Fig. 5. An intact MIT domain is required for
Vps4p localization to endosomes and for
Vps4p in vivo function. (A) AMY245 vps4D
yeast cells expressing GFP-tagged wild-type
(WT) Vps4p, Vps4p-CC, Vps4p–YEE, and
Vps4p–IRA or carrying empty vector
(YCplac111) were grown in SD medium and
the GFP-tagged proteins visualized by fluores-
cence microscopy. The same fields of cells
are shown visualized by fluorescence (right)
and Nomarski (left) optics. Scale bar, 5 lm.
(B) Ubiquitin-dependent MVB sorting of
Fth1p-GFP-Ub in AMY245 (vps4D) yeast
cells carrying plasmids expressing wild-type
Vps4p or Vps4p mutant proteins or carrying
empty vector (YCplac111). Cells were incu-
bated in SD medium containing 100 l
M
bathophenanthrolinedisulfonic acid for 6 h to
chelate iron and induce Fth1p-GFP-Ub
expression. Cells were then washed with
buffer containing 1% sodium azide, 1%
sodium fluoride, and 100 m
M phosphate,

pH 8.0, to stop further transport. The same
fields of cells are shown visualized by fluor-
escence (right) and Nomarski (left) optics.
Scale bar, 5 lm. (C) Vacuolar protein sorting
in AMY245 (vps4D) yeast cells carrying plas-
mids expressing wild-type Vps4p or Vps4p
mutant proteins or carrying empty vector
(YCplac111) or no vector. Cells were grown
on YPUAD solid medium for 2 days at
24 °C in contact with a nitrocellulose filter.
RH1800 (wild-type) yeast cells without any
vector (boxed in both panels) was included
as a control. Cells were eluted from the
filter, and carboxypeptidase Y on the filter
was detected by immunoblotting with anti-
carboxypeptidase Y serum. To test for cell
lysis, the blot was stripped and re-probed
with an antibody to a cytoplasmic protein
(calmodulin). (D) Total cell lysates from
AMY245 vps4D yeast cells expressing
wild-type Vps4p, Vps4p-CC, Vps4p–YEE,
and Vps4p–IRA or carrying empty vector
(YCplac111) were subjected to Western
blotting using a polyclonal antibody to Vps4p
as well as an antibody to actin.
P. R. Vajjhala et al. Vps4 function at the multivesicular endosome
FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS 1899
of wild-type Vps4p. We conclude that the pheno-
types observed in vps4D cells expressing the Vps4p
N-terminal mutants are due to loss of function of

these mutant proteins.
We surmise that the interactions of the Vps4p MIT
domain with Did2p, Vps2p, Vps20p and Snf7p are
critical for Vps4p recruitment to endosomes and Vps4p
function in MVB sorting.
Discussion
Here, we show that Vps4p has the ability to interact
directly with multiple components of the MVB sorting
machinery. A number of these interactions are medi-
ated by the MIT domain of Vps4p, and a subset are
regulated by Vps4p-dependent ATP hydrolysis. Interes-
tingly, however, two interactions had unique features.
The interaction of Bro1 with Vps4p is regulated by
ATP binding rather than hydrolysis, and interaction
of Did2p with Vps4p is regulated by neither ATP
binding nor ATP hydrolysis. Our data highlight the
fact that the role of Vps4p in MVB sorting is more
complex than previously assumed. As there exists
mammalian orthologues of these MVB sorting pro-
teins, our findings are likely to have relevance to VPS4
function in mammalian cells.
An indication that Vps4p may interact with Vps2p
and Bro1p came from previous studies [28,35]. How-
ever, a network of interactions connects components
of the MVB sorting machinery, therefore it was not
clear whether these putative Vps4p interactions were
direct or indirect. Ours is the first study to show that
the interactions with Vps2p and Bro1p are direct. Sev-
eral lines of evidence suggest that the interactions we
have characterized are physiologically important.

Firstly, the Vps4p interactions studied here are all with
proteins known to function in MVB sorting. Secondly,
a subset of the interactions are regulated by ATP bind-
ing or ATP hydrolysis by Vps4p. Finally, direct in vivo
evidence that these interactions are important comes
from our phenotypic analysis of vps4 mutants in which
some of these interactions are abrogated. We did not
detect a direct interaction between Vps4p and Snf7p,
consistent with a previous study [36]. However, while
this paper was in preparation, a report [37] was pub-
lished demonstrating a direct interaction using differ-
ently tagged constructs.
Our findings in this study allow us to classify Vps4p
interactions into three types; those regulated by ATP
binding, those that are regulated by ATP hydrolysis,
and those that are not regulated by either. These dif-
ferent types of interaction may contribute to Vps4p
function in different ways. The interactors may be
important for recruitment of Vps4p to endosomes,
they may be substrates acted upon by Vps4p during
disassembly of the MVB sorting machinery, or their
function at the MVB may be regulated by Vps4p. For
example, the high affinity of Did2p for Vps4p coupled
with the fact that this interaction is not regulated by
ATP binding or hydrolysis supports a role for Did2p
in recruitment of Vps4p to endosome membranes. This
is consistent with previous studies showing that Vps4p
is efficiently recruited to endosomes in both its nucleo-
tide-bound [22] and nucleotide-free [38] states. In con-
trast, Vps2p and Vps20p are likely to be substrates of

Vps4p during disassembly of the MVB sorting machin-
ery. This process is known to require cycling of Vps4p
between an ATP-bound and nucleotide-free state.
Finally, although Bro1p does not appear to be a sub-
strate, it might be regulated by Vps4p, as it is dis-
placed by ATP binding to Vps4p.
The MIT domain of Vps4p appears to play a dual
role in endosome localization and substrate binding.
Here, we have shown that motifs in the MIT domain
that are highly conserved between yeast Vps4p and
mammalian VPS4 isoforms are required for endosome
localization. These data are consistent with a previous
study showing that deletion of a larger region, which
includes most of the second and third helices of the
MIT domain, prevents Vps4p recruitment to endo-
somes. Moreover, our interaction studies have shown
that these conserved motifs are also important for
interaction with Vps2p, Vps20p, Did2p and Snf7,
indicating that some or all of these interactions may
be required for efficient targeting of Vps4p to the
endosome. Each of these proteins has been proposed
in various studies to be important for Vps4p targeting
to endosomes [23,26,39,40]. Although the Vps4p
N-terminal mutant proteins all retain interaction with
Bro1p and Vta1p, these are clearly not sufficient for
Vps4p recruitment to endosomes. As the binding
of Vps2p and Vps20p, which both bind to the MIT
domain, is regulated by ATP hydrolysis, our data
suggest that the MIT domain may also function as a
substrate-binding site. Thus our in vivo studies with the

Vps4p mutant proteins suggest that loss of Vps4p
function in MVB sorting and vacuolar protein sorting
may be due to both inefficient recruitment to endo-
somes and interaction with substrates.
Our studies together with data from a previous
study [25] indicate that four MVB sorting proteins,
Vps2p, Vps20p, Snf7p and Did2p, all interact with the
MIT domain of Vps4p, suggesting that these proteins
may have a common motif or common fold. Intrigu-
ingly, when we searched for a common motif in these
Vps4p-interacting proteins, we identified a motif,
Vps4 function at the multivesicular endosome P. R. Vajjhala et al.
1900 FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS
VDELMD, in Vps20p that is highly conserved in
Did2p (VDELMS) and Snf7p (VDETMD) and might
be present in a degenerate form in Vps2p (ADEIVN).
Although the precise binding sites are not known,
these motifs fall within the identified regions of Did2p
and Vps20p that bind to Vps4p. In contrast to these
interactors, Bro1p does not contain this motif and
does not appear to bind to the Vps4p MIT domain or
b domain, as mutations in these domains did not
diminish interaction with Bro1p. Although this is
intriguing, the motif may be shared by these proteins
because it has some other function important for
MVB sorting.
The N-terminal and C-terminal Vps4p domains have
been thought to function independently in Vps4p
endosome recruitment and assembly. This is because
the N-terminal Vps4p-CC mutant, which is not recrui-

ted to endosomes, assembles into an oligomer with
ATPase activity [22]. Also, mutation of the b domain
of Vps4p abrogates Vps4p oligomer assembly, but does
not affect Vps4p recruitment to endosomes [33]. How-
ever, our data indicate that mutations in the Vps4p
MIT domain strengthen Vps4p interactions mediated
by the b domain. In addition, mutation of the b do-
main strengthens interactions with the MIT domain.
Thus our data offer the first evidence that there may
be functional interactions between the N-terminal and
C-terminal Vps4p domains.
Our finding that the interaction of Bro1p with
Vps4p is regulated by ATP binding alone is intriguing.
Ubiquitinated cargo proteins are sorted and then de-
ubiquinated before their incorporation into an intralu-
minal vesicle. This prevents degradation of ubiquitin
along with the cargo protein [41]. Bro1p recruits the
de-ubiquitinating enzyme Doa4p to the MVB [42]. Dis-
placement of Bro1p by ATP binding to Vps4p may
allow removal of the de-ubiquitinating machinery
immediately before closure of the intraluminal vesicle
and disassembly of the MVB sorting machinery. This
will be an interesting topic for a future investigation.
An unexpected observation made during the course
of our study indicated that Vps4p may be subjected to
ubiquitin-mediated degradation. Deletion of most of
helices 2 and 3 of the MIT domain led to a consider-
able increase in expression of the mutant protein com-
pared with deletion of smaller regions within or
adjacent to this large deletion. A comparison of the

sequences deleted in the various mutants suggests that
this increased expression may be correlated with loss
of the sequence SYEENAAKKS. This sequence bears
some resemblance to a sequence, SINNDAKSS, which
is present in the cytoplasmic tail of the yeast mating
factor receptor, Ste2p [43]. In the SINNDAKSS
sequence, the serine residues are phosphorylated by
casein kinase I homologues, and this in turn is
required for ubiquitination of the lysine residue by the
Nedd4-like ubiquitin ligase, Rsp5p [44,45]. Mono-ubiq-
uitination of the SINNDAKSS sequence serves as a
signal for endocytosis and subsequent MVB sorting
and degradation in the vacuole [43]. The similarity of
this sequence and that present in the MIT domain of
Vps4p suggests that Vps4p may be subject to ubiqu-
itin-dependent degradation. This will be interesting to
investigate in the future.
Our findings support a model for MVB sorting in
which nucleotide-free Vps4p is recruited to endosomes
via interactions that may involve Vps2p, Vps20p,
Snf7p and particularly Did2p, which binds Vps4p to a
greater extent than all other interactors (Fig. 6A).
Upon recruitment of Vps4p to endosomes, it can inter-
act with Bro1p initially in the absence of bound ATP.
Upon ATP binding to Vps4p (Fig. 6B), Bro1p is dis-
placed. ATP binding is predicted to contribute to
Vps4p assembly into an oligomeric ring [22], which is
aided by Vta1p [46]. ATPase activity of Vps4p is sti-
mulated by Vta1p [33,46] and even more by Vps20p
[33]. Our data are in agreement with the proposed role

of Vps4p in disassembly of the MVB sorting machin-
ery, and support a role for disassembly via effects on
Vps2p and Vps20p (Fig. 6C), as interactions with both
of these proteins are sensitive to ATP hydrolysis (this
study [26]). Current data suggest that Vps20p in partic-
ular has numerous interactions with components of the
MVB sorting machinery [27,30] and appears to play a
key role in holding together the complex. Thus unfold-
ing of Vps20p is predicted to destabilize the assembled
MVB sorting machinery.
In summary, we have performed the first systematic
study of interactions between Vps4p and the MVB sort-
ing machinery. Several of these interactions have dis-
tinct properties. Only a subset of these interactions is
regulated by Vps4p ATPase activity, and one interac-
tion is regulated by Vps4p ATP binding. Studying the
role of Vps4p in the regulation of protein–protein inter-
actions at the MVB will lead to a better understanding
of the mechanism of MVB sorting and virus budding.
Experimental procedures
Media, reagents, strains and plasmids
YPUAD and SD minimal media were as described previ-
ously [26]. Lucifer Yellow (LY) carbohydrazide dilithium
salt was from Fluka AG (Buchs, Switzerland). Horseradish
peroxidase (HRP)-conjugated goat anti-rabbit IgG, and
bathophenanthrolinedisulfonic acid were from Sigma
P. R. Vajjhala et al. Vps4 function at the multivesicular endosome
FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS 1901
(St Louis, MO, USA). HRP-conjugated anti-mouse IgG was
from Bio-Rad Laboratories (Hercules, CA, USA), and

HRP-conjugated anti-goat IgG was from Zymed (San Fran-
cisco, CA, USA). Ni ⁄ nitrilotriacetate–agarose and monoclo-
nal antibody to pentaHis were from Qiagen (Hilden,
Germany). Immobilized glutathione on agarose was from
Scientifix (Melbourne, Australia). Pre-stained protein
molecular mass marker was from Fermentas (Hanover, MD,
USA). poly(vinylidene difluoride) (PVDF) membrane was
from Millipore (Bedford, MA, USA). Polyclonal anti-
(carboxypeptidase Y) and anti-calmodulin sera were gifts from
H. Riezman (University of Geneva, Geneva, Switzerland),
polyclonal antibody to Vps4p was from Santa Cruz Biotech-
nology (Santa Cruz, CA, USA), and monoclonal antibody
to actin was from Chemicon (Temecula, CA, USA).
S. cerevisiae strains used in this study are listed in
Table 1. Transformation of yeast with plasmid DNA was
performed using a modified lithium acetate protocol [47].
PCR primers used for plasmid constructions were from
GeneWorks (Thebarton, Australia) and are listed in
Table 2. Plasmids used in this study are listed in Table 3.
Fig. 6. A model for Vps4p-mediated disassembly of the MVB sorting complex. (A) Vps4p is recruited to the endosome membrane through
interactions that may involve Did2p, Vps2p, Vps20p and Snf7p. Did2p is predicted to play an important role because of its high-affinity inter-
action that is insensitive to ATP binding or ATP hydrolysis. The Vps4p dimers shown at the endosome membrane may form in the cyto-
plasm and then be recruited to endosomes, or monomeric Vps4p may be recruited to the endosome membrane and then oligomerize on
the membrane. Bro1p can initially interact with Vps4p and recruits Doa4p, which de-ubiquitinates the cargo protein. (B) ATP binding to Vps4p
at the endosome membrane mediates its assembly into an oligomeric ring, which is promoted by Vta1p. Bro1p is displaced by ATP binding
to Vps4p. (C) Upon ATP hydrolysis by Vps4p, effects on Vps2p and Vps20p are predicted to break interactions with these proteins and thus
disassemble the MVB sorting complex. In addition, the Vps4p high-molecular-mass oligomer disassembles. (D) The soluble components of
the MVB sorting machinery are ready for another round of MVB sorting. The Vps4 interactors that we have studied are shown in pink, and
other components are shown in grey. Components of the ESCRT complexes (0–III) are circled in box A.
Table 1. Yeast strains used in this study.

Strain Genotype Source
EGY48 MATa his3 trp1 ura3 LexAop(· 6)-LEU2 Clontech
AMY245 MATa vps4-D::KanMx leu2 ura3 his4
lys2 bar1
[33]
RH1800 MATa his4 leu2 ura3 bar1 Riezman lab
strain
Vps4 function at the multivesicular endosome P. R. Vajjhala et al.
1902 FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS
Table 2. Primers used for mutagenesis.
Primer
Sequence
(5¢-to3¢)
Vps4 Upstr F CGCTGCAGTAAGAGCAGTAAACCCG
Vps4 SalIR GAGAATCAGTGTCGACTTCATCTATAAAAATAATAGAAGGTTTATT
Vps4 SalIF GCCCATATTCGTCGACGCGCTAACAGGTACCAGAGGAGAAGGAGAGAGCGAAGCAAGTAG
Vps4 Dstr R GGGCGGATCCTCTGCTTTTCTTTATC
YEE F CTGGACACAGCCACGCAGTATACAGCATACTATAACGG
YEE R CCGTTATAGTATGCTGTATACTGCGTGGCTGTGTCCAG
IRA F CCTAAGTCGAAGGATTTGAAGCACTTGGAAAGTGAAG
IRA R CTTCACTTTCCAAGTGCTTCAAATCCTTCGACTTAGG
Table 3. Plasmids used in this study.
Plasmid Description Source
YCplac111 CEN4 ARS1 LEU2 E. coli ⁄ yeast shuttle vector [51]
pGEX5X-1 GST fusion expression vector GE Healthcare
pGEX-4T GST fusion expression vector GE Healthcare
pET11d T7 RNA polymerase-based gene expression vector Novagen
p8op-lacZ Two-hybrid reporter plasmid Clontech
pLexA Two-hybrid bait vector Clontech
pB42AD Two-hybrid prey vector Clontech

pPL 1640 URA3 CEN plasmid expressing Fth1p-GFP-Ub [34]
pAM 349 Original library clone of VPS20 in pB42AD (encoding Vps20p 3–221 ⁄ end) [26]
pAM 377 pGEX5X-1 expressing Vps20p with an N-terminal GST tag [26]
pAM 378 pGEX5X-1 expressing Vta1p with an N-terminal GST tag [26]
pAM 398 Original library clone of VTA1 in pB42AD (encoding Vta1p 108–330 ⁄ end) [26]
pAM 451 pLexA expressing LexA fused to Vps4p [26]
pAM 452 pLexA expressing LexA fused to N-terminal domain of Vps4p (residues 1–128) [26]
pAM 453 pLexA expressing LexA fused to AAA domain of Vps4p (residues 129–350) [26]
pAM 454 pLexA expressing LexA fused to C-terminal domain of Vps4p (351–437 ⁄ end) [26]
pAM 482 pET11a E. coli expression vector expressing Vps4p with a C-terminal 6His tag [26]
pAM 496 Original library clone of DID2 ⁄ CHM1 in pB42AD (encoding Did2p ⁄ chmlp 41–204 ⁄ end) [33]
pAM 813 YCplac111 expressing Vps4p [33]
pAM 862 pET11d E. coli expression vector expressing Vps4p-GAI with a C-terminal 6His tag [33]
pAM 863 YCplac111 expressing Vps4p with a C-terminal yEGFP tag [33]
pAM 870 pB42AD expressing the activation domain fused to Vps4p [33]
pAM 922 YCplac111 expressing Vps4p–E233Q [33]
pAM 922 YCplac111 expressing Vps4p-CC This study
pAM 932 YCplac111 expressing Vps4p D31-87 (Vps4p-CC) with a C-terminal yEGFP tag [33]
pAM 934 pB42AD expressing the activation domain fused to Snf7p This study
pAM 969 pB42AD expressing the activation domain fused to Vps2p This study
pAM 977 pGEX-4T expressing Snf7p with an N-terminal GST tag This study
pAM 979 pLexA expressing LexA fused to Vps4p-CC This study
pAM 980 pLexA expressing LexA fused to Vps4p D26–29 (Vps4p–YEE) This study
pAM 981 pLexA expressing LexA fused to Vps4p D56–71 (Vps4p–IRA) This study
pAM 982 pB42AD expressing the activation domain fused to Bro1p This study
pAM 985 YCplac111 expressing Vps4p (Vps4p–YEE) with a C-terminal yEGFP tag This study
pAM 986 YCplac111 expressing Vps4p (Vps4p–IRA) with a C-terminal yEGFP tag This study
pAM 987 pGEX-4T expressing Vps2p with an N-terminal GST tag This study
pAM 988 pGEX-4T expressing Did2p with an N-terminal GST tag This study
pAM 989 pGEX-4T expressing Bro1p with an N-terminal GST tag This study

pAM 990 pET11a E. coli expression vector expressing Vps4p-CC with a C-terminal 6His tag This study
pAM 991 pET11a E. coli expression vector expressing Vps4p–YEE with a C-terminal 6His tag This study
pAM 992 pET11a E. coli expression vector expressing Vps4p–IRA with a C-terminal 6His tag This study
pAM 996 YCplac111 expressing Vps4p–YEE This study
pAM 997 YCplac111 expressing Vps4p–IRA This study
P. R. Vajjhala et al. Vps4 function at the multivesicular endosome
FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS 1903
The sequence of all constructs was confirmed by automated
DNA sequencing (Australian Genome Research Facility,
Brisbane, Australia).
Plasmid construction
Genomic DNA was prepared as described previously [48]
from S. cerevisiae RH449, and an S. cerevisiae cDNA lib-
rary was a gift from Michael White (University of Texas
South-western Medical School, Dallas, TX, USA). PCR
was carried out using the proofreading DNA polymerases
Pfu (Fermentas) or Phusion (Finnzymes, Espoo, Finland).
N-Terminal YEE and IRA mutations were generated by
site-directed mutagenesis using the oligonucleotides listed in
Table 2. In each case two separate PCRs were set up with
either the Vps4 Upstr F primer and a mutagenic reverse
primer or a mutagenic forward primer and a SalI R primer.
The two PCR products were combined for a third PCR
using the Vps4 Upstr F and SalI R primers. The resulting
PCR product encoding the N-terminal region of Vps4p,
with a mutation, was ligated with pAM813 that had been
digested with PstI and SalI to release the wild-type
sequence. To generate pLexA constructs with N-terminal
CC, YEE or IRA mutations, mutant VPS4 was amplified
without any upstream sequence and with suitable restriction

sites for cloning. To express Vps4p–YEE and IRA N-ter-
minal mutants with a C-terminal GFP tag, both constructs
were amplified without a stop codon and cloned in-frame
into a YCplac111-based plasmid encoding yEGFP, which
was cloned from pYM12 [49]. To express N-terminal
mutants with a C-terminal 6His tag in Escherichia coli,
DNA encoding the N-terminal mutants were amplified with
a primer-encoded C-terminal 6His tag and ligated down-
stream of the T7 promoter of pET11d (Novagen, Madison,
WI, USA).
The wild-type VPS2, SNF7 and BRO1 genes were ampli-
fied by PCR from genomic DNA and subcloned into
pB42AD in-frame with the activation domain. The SNF7
and BRO1 genes were subcloned from the pB42AD expres-
sion plasmids into pGEX-4T1 in-frame with GST. The
DID2 gene was amplified from genomic DNA and cloned
into pGEX-4T in-frame with GST. The VPS2 cDNA was
amplified from a cDNA library and ligated into pGEX-4T
in-frame with GST.
Yeast two-hybrid protein interaction analysis
Protein interactions were assayed using the Matchmaker
LexA yeast two-hybrid system from Clontech (Palo Alto,
CA, USA) as described previously [26]. Briefly, bait plas-
mids containing LexA fusion proteins were cotransformed
into the yeast strain EGY48 along with prey plasmids
encoding proteins fused to an activation domain and the
reporter plasmid p8op-LacZ. To test for interaction, trans-
formants were spotted on to synthetic galactose ⁄ raffinose
(SG) complete medium lacking Ura, Trp, and His and
containing X-gal. The strength of protein interactions was

assessed by blue coloration on this medium.
In vitro binding assays
In vitro binding assays were performed as previously
described [26]. The 6His-tagged wild-type Vps4p,
Vps4p–E233Q, Vps4p-CC, Vps4p–YEE, and Vps4p–IRA
as well as GST-tagged Vps2p, Snf7p, Vps20p, Bro1p,
Did2p and Vta1p, and GST alone were expressed in BL21-
CodonPlus
TM
(DE3)–RIL E. coli and affinity-purified on
Ni ⁄ nitrilotriacetate–agarose or glutathione–agarose, respect-
ively. The 6His-tagged proteins were eluted from the resin
using 250 mm imidazole and buffer exchanged into binding
buffer (20 mm Hepes, 200 mm sorbitol, 100 mm potassium
acetate, 1 mm EDTA, 1 mm dithiothreitol, 20 mm MgCl
2
,
0.1% Triton X-100). The Vps4p YEE and IRA mutant
proteins were poorly expressed in E. coli compared with
wild-type Vps4p. To test for direct in vitro binding, gluta-
thione–agarose bearing 25 lg GST–Vps2p, GST–Snf7p,
GST–Vps20p, GST–Did2p, or 50 lg GST-Bro1p or 500 lg
GST were incubated with 10 lg wild-type 6His-Vps4p in
1 mL binding buffer. To test for ATPase-sensitive or ATP-
sensitive binding, the amount of GST fusion proteins used
in the in vitro binding assay was decreased to allow detec-
tion of any difference in Vps4p binding in the presence of
ATP. Thus 45 lg GST–Bro1p, 28 lg GST–Vps20p, 13 lg
GST–Vps2p, or 3 lg GST–Did2p were incubated with
10 lg wild-type 6His-Vps4p or 6His-Vps4p–E233Q in

1 mL binding buffer in the presence or absence of 1 mm
ATP. To compare the binding of wild-type and Vps4p
N-terminal mutant proteins, 2.5 lg full-length 6His-tagged
wild-type Vps4p or Vps4p mutant protein was incubated
with glutathione–agarose, bearing GST alone (500 lg),
7 lg GST–Bro1p, 5 lg GST–Vps20p, 1 lg GST–Vta1p,
1 lg GST–Did2p, or 5 lg GST–Vps2p in 1 mL binding
buffer. In all in vitro binding assays, samples were incuba-
ted overnight at 4 °C. The resin was washed four times
with binding buffer, and the protein bound to the resin
was eluted by heating with Laemmli sample buffer. An ali-
quot of the supernatant containing unbound protein was
diluted 1 : 1 in Laemmli sample buffer. Bound and
unbound proteins were resolved by SDS ⁄ PAGE and trans-
ferred to a PVDF filter. Wild-type Vps4p and Vps4p
mutant proteins were detected with an antibody to Vps4p
and enhanced chemiluminescence.
Western blot analysis of total yeast cell lysates
For western blot analysis of total cell lysates, AMY245
(vps4D) yeast cells carrying expression plasmids were grown
at 24 °C for 24 h. Cells from 2 mL culture were pelleted
and resuspended in 240 lL lysis solution (1.85 m NaOH,
1.06 m 2-mercaptoethanol) followed by incubation for
Vps4 function at the multivesicular endosome P. R. Vajjhala et al.
1904 FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS
10 min on ice. Proteins were precipitated by the addition of
an equal volume of trichloroacetic acid, and the pellet was
washed with ice-cold acetone. The pellet was resuspended
in 50 lL resuspension solution (5% SDS, 0.5 m Tris) and
mixed with an equal volume of 75% glycerol ⁄ 0.12 m dithio-

threitol ⁄ 0.05% bromophenol blue and boiled for 5 min.
Samples were subjected to SDS ⁄ PAGE, and proteins were
transferred to a PVDF filter, which was then probed with a
goat polyclonal antibody to Vps4p. To assess sample load-
ing, the same samples were electrophoresed on a second
gel, transferred to PVDF, and probed with a monoclonal
antibody to actin.
Phenotypic assays
Assays for MVB sorting and carboxypeptidase Y secretion
were performed as described previously [33].
Microscopy
Microscopy was performed using an Olympus BX51
(Olympus Australia Pty, Ltd, Mount Waverly, Australia)
with a Nomarski filter for visualizing vacuoles and the
appropriate filter for viewing GFP fluorescence.
Bioinformatics
Alignment of yeast Vps4p and mammalian VPS4 sequences
to identify highly conserved sequence was carried out using
clustal w [50]. Identification of common motifs between
the Vps4p-interacting proteins was determined using the
glam (Gapped Local Alignment of Motifs) 2 program
(M. C. Frith and T. L. Bailey, personal communication).
Acknowledgements
This work was made possible by funding from the
National Health and Medical Research Council of
Australia Project Grant 298921 and from core support
from the Queensland State Government, all to A.L.M.
We thank M. C. Frith and T. L. Bailey for the use of
the glam2 program.
References

1 Katzmann DJ, Odorizzi G & Emr SD (2002) Receptor
downregulation and multivesicular-body sorting. Nat
Rev Mol Cell Biol 3, 893–905.
2 Trombetta ES & Mellman I (2005) Cell biology of anti-
gen processing in vitro and in vivo. Annu Rev Immunol
23, 975–1028.
3 Fevrier B & Raposo G (2004) Exosomes: endosomal-
derived vesicles shipping extracellular messages. Curr
Opin Cell Biol 16, 415–421.
4 Bieniasz PD (2006) Late budding domains and host pro-
teins in enveloped virus release. Virology 344, 55–63.
5 Morita E & Sundquist WI (2004) Retrovirus budding.
Annu Rev Cell Dev Biol 20, 395–425.
6 Palade GE (1955) A small particulate component of the
cytoplasm. J Biophys Biochem Cytol 1, 59–68.
7 Sotelo JR & Porter KR (1959) An electron microscope
study of the rat ovum. J Biophys Biochem Cytol 5, 327–
342.
8 Raymond CK, Howald-Stevenson I, Vater CA &
Stevens TH (1992) Morphological classification of the
yeast vacuolar protein sorting mutants: evidence for a
prevacuolar compartment in class E vps mutants.
Mol Biol Cell 3, 1389–1402.
9 Rieder SE, Banta LM, Kohrer K, McCaffery JM &
Emr SD (1996) Multilamellar endosome-like compart-
ment accumulates in the yeast vps28 vacuolar protein
sorting mutant. Mol Biol Cell 7, 985–999.
10 Odorizzi G, Babst M & Emr SD (1998) Fab1p PtdIns
(3) P 5-kinase function essential for protein sorting in
the multivesicular body. Cell 95, 847–858.

11 von Schwedler UK, Stuchell M, Muller B, Ward DM,
Chung HY, Morita E, Wang HE, Davis T, He GP,
Cimbora DM, Scott A, et al. (2003) The protein
network of HIV budding. Cell 114, 701–713.
12 Hurley JH & Emr SD (2006) The ESCRT complexes:
structure and mechanism of a membrane-trafficking
network. Annu Rev Biophys Biomol Struct 35, 277–298.
13 Riezman H, Munn A, Geli MI & Hicke L (1996)
Actin-, myosin- and ubiquitin-dependent endocytosis.
Experientia 52, 1033–1041.
14 Babst M, Sato TK, Banta LM & Emr SD (1997)
Endosomal transport function in yeast requires a novel
AAA-type ATPase, Vps4p. EMBO J 16, 1820–1831.
15 Yoshimori T, Yamagata F, Yamamoto A, Mizushima
N, Kabeya Y, Nara A, Miwako I, Ohashi M, Ohsumi
M & Ohsumi Y (2000) The mouse SKD1, a homologue
of yeast Vps4p, is required for normal endosomal
trafficking and morphology in mammalian cells.
Mol Biol Cell 11, 747–763.
16 Fujita H, Yamanaka M, Imamura K, Tanaka Y, Nara
A, Yoshimori T, Yokota S & Himeno M (2003) A
dominant negative form of the AAA ATPase
SKD1 ⁄ VPS4 impairs membrane trafficking out of endo-
somal ⁄ lysosomal compartments: class E vps phenotype
in mammalian cells. J Cell Sci 116, 401–414.
17 Sachse M, Strous GJ & Klumperman J (2004) ATPase-
deficient hVPS4 impairs formation of internal endoso-
mal vesicles and stabilizes bilayered clathrin coats on
endosomal vacuoles. J Cell Sci 117, 1699–1708.
18 Hanson PI & Whiteheart SW (2005) AAA+ proteins:

have engine, will work. Nat Rev Mol Cell Biol 6, 519–529.
19 Sauer RT, Bolon DN, Burton BM, Burton RE, Flynn
JM, Grant RA, Hersch GL, Joshi SA, Kenniston JA,
Levchenko I et al. (2004) Sculpting the proteome with
P. R. Vajjhala et al. Vps4 function at the multivesicular endosome
FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS 1905
AAA (+) proteases and disassembly machines. Cell
119, 9–18.
20 Cereghino JL, Marcusson EG & Emr SD (1995) The
cytoplasmic tail domain of the vacuolar protein sorting
receptor Vps10p and a subset of VPS gene products
regulate receptor stability, function, and localization.
Mol Biol Cell 6, 1089–1102.
21 Piper RC, Cooper AA, Yang H & Stevens TH (1995)
VPS27 controls vacuolar and endocytic traffic through a
prevacuolar compartment in Saccharomyces cerevisiae.
J Cell Biol 131, 603–617.
22 Babst M, Wendland B, Estepa EJ & Emr SD (1998)
The Vps4p AAA ATPase regulates membrane associa-
tion of a Vps protein complex required for normal
endosome function. EMBO J 17, 2982–2993.
23 Lin Y, Kimpler LA, Naismith TV, Lauer JM & Hanson
PI (2005) Interaction of the mammalian endosomal sort-
ing complex required for transport (ESCRT) III protein
hSnf7–1 with itself, membranes, and the AAA+ ATPase
SKD1. J Biol Chem 280, 12799–12809.
24 Takasu H, Jee JG, Ohno A, Goda N, Fujiwara K,
Tochio H, Shirakawa M & Hiroaki H (2005) Structural
characterization of the MIT domain from human
Vps4b. Biochem Biophys Res Commun 334, 460–465.

25 Scott A, Gaspar J, Stuchell-Brereton MD, Alam SL,
Skalicky JJ & Sundquist WI (2005) Structure and
ESCRT-III protein interactions of the MIT domain of
human VPS4A. Proc Natl Acad Sci USA 102, 13813–
13818.
26 Yeo SC, Xu L, Ren J, Boulton VJ, Wagle MD, Liu C,
Ren G, Wong P, Zahn R, Sasajala P, Yang H, Piper
RC & Munn AL (2003) Vps20p and Vta1p interact with
Vps4p and function in multivesicular body sorting and
endosomal transport in Saccharomyces cerevisiae J Cell
Sci 116, 3957–3970.
27 Bowers K, Lottridge J, Helliwell SB, Goldthwaite LM,
Luzio JP & Stevens TH (2004) Protein–protein interac-
tions of ESCRT complexes in the yeast Saccharomyces
cerevisiae. Traffic 5, 194–210.
28 Gavin AC, Bosche M, Krause R, Grandi P, Marzioch
M, Bauer A, Schultz J, Rick JM, Michon AM, Cruciat
CM et al. (2002) Functional organization of the yeast
proteome by systematic analysis of protein complexes.
Nature 415, 141–147.
29 Ho Y, Gruhler A, Heilbut A, Bader GD, Moore L,
Adams SL, Millar A, Taylor P, Bennett K, Boutilier K
et al. (2002) Systematic identification of protein
complexes in Saccharomyces cerevisiae by mass spectro-
metry. Nature 415, 180–183.
30 Ito T, Chiba T, Ozawa R, Yoshida M, Hattori M &
Sakaki Y (2001) A comprehensive two-hybrid analysis
to explore the yeast protein interactome. Proc Natl Acad
Sci USA 98, 4569–4574.
31 Howard TL, Stauffer DR, Degnin CR & Hollenberg

SM (2001) CHMP1 functions as a member of a newly
defined family of vesicle trafficking proteins. J Cell Sci
114, 2395–2404.
32 Scott A, Chung HY, Gonciarz-Swiatek M, Hill GC,
Whitby FG, Gaspar J, Holton JM, Viswanathan R,
Ghaffarian S, Hill CP, et al. (2005) Structural and
mechanistic studies of VPS4 proteins. EMBO J 24,
3658–3669.
33 Vajjhala PR, Wong JS, To HY & Munn AL (2006) The
beta domain is required for Vps4p oligomerization into
a functionally active ATPase. FEBS J 273, 2357–2373.
34 Urbanowski JL & Piper RC (2001) Ubiquitin sorts
proteins into the intralumenal degradative compartment
of the late-endosome ⁄ vacuole. Traffic 2, 622–630.
35 Fujita H, Umezuki Y, Imamura K, Ishikawa D,
Uchimura S, Nara A, Yoshimori T, Hayashizaki Y,
Kawai J, Ishidoh K, et al. (2004) Mammalian class E
Vps proteins, SBP1 and mVps2 ⁄ CHMP2A, interact with
and regulate the function of an AAA-ATPase
SKD1 ⁄ Vps4B. J Cell Sci 117, 2997–3009.
36 Peck JW, Bowden ET & Burbelo PD (2004) Structure
and function of human Vps20 and Snf7 proteins.
Biochem J 377, 693–700.
37 Nickerson DP, West M & Odorizzi G (2006) Did2
coordinates Vps4-mediated dissociation of ESCRT-III
from endosomes. J Cell Biol 175, 715–720.
38 Bishop N & Woodman P (2000) ATPase-defective
mammalian VPS4 localizes to aberrant endosomes and
impairs cholesterol trafficking. Mol Biol Cell 11,
227–239.

39 Babst M, Katzmann DJ, Estepa-Sabal EJ, Meerloo T &
Emr SD (2002) Escrt-III: an endosome-associated
heterooligomeric protein complex required for mvb
sorting. Dev Cell 3, 271–282.
40 Lottridge JM, Flannery AR, Vincelli JL & Stevens TH
(2006) Vta1p and Vps46p regulate the membrane
association and ATPase activity of Vps4p at the yeast
multivesicular body. Proc Natl Acad Sci USA 103,
6202–6207.
41 Amerik AY, Nowak J, Swaminathan S & Hochstrasser
M (2000) The Doa4 deubiquitinating enzyme is
functionally linked to the vacuolar protein-sorting and
endocytic pathways. Mol Biol Cell 11, 3365–3380.
42 Luhtala N & Odorizzi G (2004) Bro1 coordinates
deubiquitination in the multivesicular body pathway by
recruiting Doa4 to endosomes. J Cell Biol 166, 717–729.
43 Hicke L & Riezman H (1996) Ubiquitination of a yeast
plasma membrane receptor signals its ligand-stimulated
endocytosis. Cell 84, 277–287.
44 Hicke L, Zanolari B & Riezman H (1998) Cytoplasmic
tail phosphorylation of the alpha-factor receptor is
required for its ubiquitination and internalization. J Cell
Biol 141, 349–358.
45 Dunn R & Hicke L (2001) Multiple roles for
Rsp5p-dependent ubiquitination at the internalization
step of endocytosis. J Biol Chem 276, 25974–25981.
Vps4 function at the multivesicular endosome P. R. Vajjhala et al.
1906 FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS
46 Azmi I, Davies B, Dimaano C, Payne J, Eckert D,
Babst M & Katzmann DJ (2006) Recycling of ESCRTs

by the AAA-ATPase Vps4 is regulated by a conserved
VSL region in Vta1. J Cell Biol 172, 705–717.
47 Munn AL, Stevenson BJ, Geli MI & Riezman H (1995)
end5, end6, and end7: mutations that cause actin
delocalization and block the internalization step of
endocytosis in Saccharomyces cerevisiae. Mol Biol Cell
6, 1721–1742.
48 Adams A, Gottschling DE, Kaiser KA & Stearns T
(1997) Methods in Yeast Genetics. A Cold Spring Harbor
Laboratory Course Manual. Cold Spring Harbor
Laboratory Press, Cold Spring Harbor, NY.
49 Knop M, Siegers K, Pereira G, Zachariae W, Winsor B,
Nasmyth K & Schiebel E (1999) Epitope tagging of
yeast genes using a PCR-based strategy: more tags and
improved practical routines. Yeast 15, 963–972.
50 Thompson JD, Higgins DG & Gibson TJ (1994)
CLUSTAL W: improving the sensitivity of progressive
multiple sequence alignment through sequence
weighting, position-specific gap penalties and weight
matrix choice. Nucleic Acids Res 22, 4673–4680.
51 Gietz RD & Sugino A (1988) New yeast–Escherichia
coli shuttle vectors constructed with in vitro mutagen-
ized yeast genes lacking six-base pair restriction sites.
Gene 74, 527–534.
P. R. Vajjhala et al. Vps4 function at the multivesicular endosome
FEBS Journal 274 (2007) 1894–1907 ª 2007 The Authors Journal compilation ª 2007 FEBS 1907