We show that the b domain of the budding yeast Vps4p is not required for recruitment to endosomes, but is essential for all Vps4p endocytic func-tions in vivo.. The C-terminal region of
Trang 1a functionally active ATPase
Parimala R Vajjhala1, Julin S Wong1, Hui-Yi To1and Alan L Munn1,2,3,4
1 Institute for Molecular Bioscience and ARC Special Research Centre for Functional and Applied Genomics, University of Queensland,
St Lucia, Queensland, Australia
2 School of Biomedical Sciences, University of Queensland, St Lucia, Queensland, Australia
3 Laboratory of Yeast Cell Biology, Institute of Molecular and Cell Biology, A*STAR Biomedical Sciences Institutes, Singapore
4 Department of Biochemistry, Faculty of Medicine, National University of Singapore, Singapore
The multivesicular body (MVB) is a central sorting
station in the itinerary of proteins that traffic through
the endocytic pathway In the MVB, integral
mem-brane proteins that are destined for delivery to the
lysosome lumen undergo MVB sorting into internal
vesicles, which form by invagination of the limiting
membrane of the MVB This process sequesters the
cytoplasmic tails of endocytosed signalling receptors
and allows efficient silencing Mature MVBs then fuse
with the lysosome and transfer their contents,
inclu-ding internal vesicles, into the lysosome lumen In some cells, MVBs have been shown to fuse with the plasma membrane and have been suggested to function
in intercellular signalling [1,2] Moreover, the machin-ery that generates the internal vesicles of MVBs has recently been implicated in the budding of several clas-ses of enveloped viruclas-ses (reviewed in [3–5])
Both endocytic and biosynthetic traffic to the lyso-some proceed via the MVB [6,7], thus defects in the function of the MVB affect both endocytic transport
Keywords
class E vacuolar protein sorting; dopamine
responsive gene-1; LYST-interacting protein
5; suppressor of K + uptake growth defect 1
(SKD1); SKD1-binding protein 1
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: A.Munn@imb.uq.edu.au
(Received 16 February 2006, revised
14 March 2006, accepted 20 March 2006)
doi:10.1111/j.1742-4658.2006.05238.x
Endocytic and biosynthetic trafficking pathways to the lysosome⁄ vacuole converge at the prevacuolar endosomal compartment During transport through this compartment, integral membrane proteins that are destined for delivery to the lysosome⁄ vacuole lumen undergo multivesicular body (MVB) sorting into internal vesicles formed by invagination of the endo-somal limiting membrane Vps4 is an AAA family ATPase which plays a key role in MVB sorting and facilitates transport through endosomes It possesses an N-terminal microtubule interacting and trafficking domain required for recruitment to endosomes and an AAA domain with an ATPase catalytic site The recently solved 3D structure revealed a b domain, which protrudes from the AAA domain, and a final C-terminal a-helix However, the in vivo roles of these domains are not known In this study, we have identified motifs in these domains that are highly con-served between yeast and human Vps4 We have mutated these motifs and studied the effect on yeast Vps4p function in vivo and in vitro We show that the b domain of the budding yeast Vps4p is not required for recruitment to endosomes, but is essential for all Vps4p endocytic func-tions in vivo We also show that the b domain is required for Vps4p homotypic interaction and for full ATPase activity In addition, it is required for interaction with Vta1p, which works in concert with Vps4p
in vivo Our studies suggest that assembly of a Vps4p oligomeric complex with full ATPase activity that interacts with Vta1p is essential for normal endosome function
Abbreviations
CPY, carboxypeptidase Y; ESCRT, endosomal sorting complex required for transport; GFP, green fluorescent protein; GST, glutathione S-transferase; MIT, microtubule interacting and trafficking; MVB, multivesicular body; PVDF, poly(vinylidene difluoride); Vps, vacuolar protein sorting.
Trang 2and delivery of newly synthesized lysosomal⁄ vacuolar
proteins Insights into both processes have come from
studies of budding yeast vacuolar protein sorting (vps)
mutants which missort soluble vacuolar proteins into
the extracellular medium A subclass of vps mutants,
class E vps mutants, disrupt MVB sorting and form an
enlarged multilamellar endosome adjacent to the
vacu-ole termed the class E compartment [8–10] This
com-partment accumulates endocytic and biosynthetic
material as well as recycling late Golgi proteins [9,11–
13] resulting from defective transport out of this
aber-rant endosome There are 27 class E Vps proteins in
mammalian cells and 18 in yeast, which can be
grouped into complexes referred to as endosomal
sort-ing complexes required for transport (ESCRT I–III)
These three complexes act sequentially to sort and
deliver membrane proteins into forming intraluminal
vesicles [14,15]
Vps4p, also known as Csc1p⁄ End13p ⁄ Grd13p ⁄
Vpl4p⁄ Vpt10p ⁄ Did6p, is a class E Vps protein that
belongs to the AAA (ATPase associated with a variety
of cellular activities) family of ATPases [8,16] Vps4p
functions at multiple steps during endocytic transport
[8,17] and has recently been shown to function in
sterol metabolism [18] There are two isoforms of
VPS4 in human cells: VPS4A and VPS4B The
endocy-tic functions of yeast Vps4p are conserved in
mamma-lian VPS4B [19–21], and both human VPS4A and
VPS4B have been shown to be required for virus
bud-ding [22,23] Vps4p contains a single AAA domain,
and its ATPase activity is critical for function as vps4
mutants defective in ATP binding (K179A) or
hydroly-sis (E233Q), and a temperature-sensitive vps4 mutant
(M307T⁄ L327S) and the end13-1 mutant (S335F),
which also have mutations in the AAA domain, all
induce class E compartments and perturb endocytic
and biosynthetic traffic [8,17] Vps4p ATPase activity
is important for disassembly of ESCRTs to allow reuse
in multiple rounds of MVB sorting [24]
Whereas many AAA proteins form hexamers,
wild-type Vps4p has not been shown to form a higher-order
oligomer Wild-type Vps4p purified from bacteria
forms a dimer [24] However, a homotypic interaction
has not been demonstrated for wild-type Vps4p in vivo
using a yeast two-hybrid assay [25] Moreover,
endo-genous human VPS4B appears to exist as a monomer
[26] Thus it is not clear whether Vps4p exists as a
dimer in vivo or whether dimer formation is important
for Vps4p function in vivo, and the structural
determi-nants required for dimerization of Vps4p have not
been identified Although a homotypic interaction has
been described for the Vps4p-E233Q mutant in a yeast
two-hybrid assay [25], and this mutant forms a
10–12-mer in the presence of ATP [24,27], it is not clear whe-ther this is due to stabilization of a complex that is normally transiently formed by wild-type Vps4p or to formation of an aberrant complex Consistent with the ability of Vps4p to form oligomers, most vps4 mutants that have been described to date are dominant negative [17,24,28]
The N-terminal region of Vps4p contains a microtu-bule interacting and trafficking (MIT) domain [29] This region of Vps4p interacts with Vps20p⁄ Chm6p [30], a component of ESCRT III, and is required to recruit Vps4p to endosomes The C-terminal region of Vps4p binds to another class E Vps protein, Vta1p [30], and this interaction is conserved in mammalian cells [26] As the C-terminal region of Vps4p contains several motifs that are highly conserved between yeast Vps4p and both of the human VPS4 isoforms, we hypothesized that this region has an important role in Vps4p assembly or function In this study, we show that conserved motifs in the C-terminal region, which are mainly present in the recently identified b domain
of Vps4p, are essential for interaction with Vta1p In addition, we show that one of these motifs is required for a homotypic interaction and for formation of a highly active ATPase complex, but is not required for endosomal recruitment More importantly, the con-served motifs in the b domain are required for Vps4p functions in vivo
Results
Conserved motifs in the Vps4p C-terminal region are essential for function but not for protein stability
The Vps4p sequence located C-terminal to the previ-ously predicted AAA domain [8] contains several motifs that are highly conserved in human VPS4B (Fig 1A) as well as VPS4A (not shown) To examine the functional importance of these C-terminal motifs and the C-terminal region in general, we constructed various mutations in this domain and examined the effect of each on Vps4p stability and in vivo function
To assess the importance of the C-terminal region (res-idues 351–437) or half of it (res(res-idues 395–437), the corresponding C-terminal truncation mutants (Vps4p-Ter1 and Vps4p-Ter2, respectively) were generated To assess the importance of individual conserved motifs,
we chose three motifs to delete These deletion mutant proteins are named according to the first three amino acids of each motif deleted, i.e RKI, Vps4p-GAI and Vps4p-LTP In addition, we generated a mutant in which each charged residue in a conserved
Trang 3motif (RDE) at the extreme C-terminus was
substi-tuted with alanine (Vps4p-RDE)
The recent crystal structure of human VPS4B [27]
allowed us to map the positions of these motifs
(Fig 1A) The domain organization of yeast Vps4p
(Fig 1B) inferred from the elucidated 3D crystal
struc-ture of human VPS4B shows that the AAA domain is
unusual in that it contains a domain with three
antipar-allel b-sheets, referred to as the b domain within the
C-terminal subdomain Therefore the AAA domain
extends past the original predicted AAA domain Thus
Vps4p-Ter1 lacks the end of a-helix 8 in the AAA
domain, the entire b domain and both of the two
minal a-helices Vps4p-Ter2 lacks only the two
C-ter-minal a-helices The motif deleted in the Vps4p-RKI
mutant is positioned at the end of a-helix 8 of the AAA
domain and extends into the first b-sheet of the b
domain In comparison, the motifs deleted in the
Vps4p-GAI and Vps4p-LTP mutants are positioned adjacent
to and within the second b-sheet, and within the third
b-sheet, respectively, of the b domain The RDE motif
subjected to charged-to-alanine substitutions in the
Vps4p-RDE mutant is positioned at the end of the
C-terminal a-helix (Fig 1A) The positions of the highly
conserved residues that were mutated are shown on the
VPS4B structure in Fig 1C
To test whether the Vps4p mutant proteins are
sta-bly expressed in yeast, plasmids expressing the
wild-type and mutant Vps4p constructs were introduced
into vps4D cells, and Vps4p was detected in cell
extracts by immunoblotting The steady-state levels of
Vps4p-RKI, Vps4p-GAI, Vps4p-LTP and Vps4p-RDE
mutant proteins are comparable to wild-type Vps4p
(Fig 1D) In contrast, the level of Vps4p-Ter1 and Vps4p-Ter2 mutant proteins (38.3 and 43.9 kDa, respectively) was too low to detect (data not shown) Although the C-terminus itself may be required for protein stability, we conclude that the four conserved motifs that we mutated are not individually essential for protein stability
We next examined whether those Vps4p mutant con-structs that are stably expressed can substitute for wild-type Vps4p in its various functions Plasmids expressing wild-type Vps4p or the various Vps4p mutant proteins
or empty vector were introduced into vps4D yeast cells and the ability of the mutant proteins to rescue the
A
S.c Vps4 351 IRKIQSATHFKDV -STEDDE -TRKLTPCSPGD 380
.::.::::::: : : : :::::::: H.s Vps4B 353 VRKVQSATHFKKVRGPSRADPNHLVDDLLTPCSPGD 388
S.c Vps4 381 DGAIEMSWTDIEADELKEPDLTIKDFLKAIKSTRPT 416
:::::.: : : : :: : : :.:: H.s Vps4B 389 PGAIEMTWMDVPGDKLLEPVVSMSDMLRSLSNTKPT 424
S.c Vps4 417 VNEDDLLKQEQFTRDFGQEG 437 ::: :::: :: ::::::
444
α10
α8
α10
C
D
Vps4p
actin
B
AAA N-terminal subdomain
AAA C-terminal subdomain
β domain
C-terminal helix MIT domain
415 79
Fig 1 Construction of Vps4p C-terminal mutants (A) Alignment of
S cerevisiae (S.c) Vps4p and human (H.s) VPS4B sequences using
shown in bold, and residues that were substituted with alanine in
the RDE mutant are shown in bold italics The secondary structure
of the corresponding region of VPS4B is also shown (B) Schematic
representation of wild-type Vps4p with the domain organization
inferred from structural data of VPS4A and VPS4B [27,34] (C)
Locat-ion of highly conserved residues in the human VPS4B structure
that were mutated in yeast Vps4p The b domain is circled, and the
RKI, LTP and GAI motifs are shown in red, light blue and green,
respectively The charged residues in the RDE motif are shown in
dark blue The colour code for the nonmutated residues in the
dif-ferent domains is: b domain, orange; N-terminal AAA subdomain,
pink; C-terminal AAA subdomain, light brown; C-terminal a-helix,
brown (D) Total cell lysates from RH2906 (vps4D) yeast cells
carry-ing plasmids expresscarry-ing wild-type Vps4p, Vps4p-RDE, Vps4p-RKI,
Vps4p-LTP and Vps4p-GAI mutant proteins or carrying empty vector
(YCplac111) were subjected to western blotting using either a
poly-clonal antibody to Vps4p or a monopoly-clonal antibody to actin.
Trang 4phenotypes of vps4D cells was assessed Vacuolar
accu-mulation of a fluid-phase endocytic marker, MVB
sort-ing of a membrane protein to the vacuole lumen,
delivery of soluble vacuolar proteins to the vacuole and
growth at high temperature were examined
Endocytosis and subsequent vacuolar accumulation
of the fluid-phase marker, Lucifer Yellow, was restored
by wild-type Vps4p but not by RKI,
Vps4p-GAI or Vps4p-LTP mutant forms of Vps4p when
compared with empty vector alone (Fig 2A) In
con-trast, vacuolar accumulation of Lucifer Yellow was
efficiently restored by Vps4p-RDE To examine MVB
sorting, we used the iron transporter homologue,
Fth1p, fused to ubiquitin (Fth1p-Ub) as a marker
Fth1p normally resides on the limiting membrane of
the vacuole, but, when tagged with ubiquitin, it
under-goes ubiquitin-dependent MVB sorting and is delivered
to the vacuole lumen [31] Green fluorescent protein
(GFP)-tagged Fth1p-Ub was mainly observed in a
class E compartment adjacent to the vacuole in vps4D
cells expressing RKI, GAI or
Vps4p-LTP similar to vps4D cells carrying the empty vector
(Fig 2B) The small amount that reached the vacuole
was present on the limiting membrane In contrast,
expression of wild-type Vps4p or Vps4p-RDE resulted
in delivery of GFP-tagged Fth1p-Ub to the vacuole
lumen These data clearly demonstrate that the
con-served C-terminal motifs that are required for efficient
fluid-phase endocytosis are also critical for
ubiquitin-dependent MVB sorting
To assess the ability of the mutant Vps4p
con-structs to function in vacuolar protein sorting, we
assayed their ability to correct the missorting and
secretion of carboxypeptidase Y (CPY) in vps4D
cells CPY is a soluble resident vacuolar protein that
is translocated into the endoplasmic reticulum and then transported to the Golgi where a receptor, Vps10p, sorts it from secretory proteins destined for the cell surface into a pathway that takes it via endosomes to the vacuole In vps4D cells, CPY is missorted at the late Golgi into vesicles destined for the cell surface and secreted [8] Expression of wild-type Vps4p but not Vps4p-RKI, Vps4p-GAI or Vps4p-LTP mutant proteins restored vacuolar deliv-ery of CPY compared with empty vector alone (Fig 2C) In contrast, CPY sorting to the vacuole is restored by Vps4p-RDE
Finally, we examined the ability of the various mutant Vps4p constructs to function in cell growth at high temperature vps4D cells expressing wild-type or mutant Vps4p proteins or carrying empty vector were tested for growth on solid medium at high tempera-ture Expression of wild-type Vps4p but not Vps4p-RKI, Vps4p-GAI and Vps4p-LTP or empty vector rescued growth at 40C (Fig 2D) In contrast, Vps4p-RDE was able to significantly restore growth at
40C However, Vps4p-RDE was reproducibly less efficient than wild-type Vps4p in restoring growth at
40C to vps4D cells There was no obvious difference
in the growth rate at 24C between vps4D cells expres-sing wild-type Vps4p, any of the Vps4p mutants or those carrying empty vector
We conclude that the conserved motifs adjacent to and within the b domain that were deleted in the Vps4p-RKI, Vps4p-GAI and Vps4p-LTP mutants are essential for all Vps4p functions tested but not for pro-tein stability The charged residues in the RDE motif at the end of the C-terminal helix of Vps4p are not essen-tial for protein stability or for most functions, however, they are required for full cellular growth at 40C
Fig 2 Conserved motifs within and adjacent to the b domain of Vps4p are essential for several Vps4p in vivo functions (A) Lucifer Yellow uptake and accumulation in the vacuole was measured in RH2906 (vps4D) yeast cells carrying plasmids expressing wild-type Vps4p or Vps4p mutant proteins or carrying empty vector (YCplac111) The same fields of cells are shown visualized by fluorescence (left) and Nomar-ski (right) optics The vacuoles appear as indentations in the cell profile by NomarNomar-ski 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 incubated in YPUAD medium containing 100 l M bathophenanthroline disulfonic 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, 100 m M
phosphate, pH 8.0, to stop further transport The same fields of cells are shown visualized by fluorescence (left) and Nomarski (right) optics Scale bar, 5 lm (C) Vacuolar protein sorting in RH2906 (vps4D) yeast cells carrying plasmids 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 CPY on the filter was detected by immunoblotting with anti-CPY serum To test for cell lysis, the blot was stripped and reprobed with an antibody to a cytoplasmic protein (calmodulin) (D) Temperature-sensitive growth assay of RH2906 (vps4D) yeast cells carrying plasmids expressing wild-type Vps4p or Vps4p mutant proteins or carrying empty vector (YCplac111) Cells were serially diluted 10-fold, and 7-lL aliquots were spotted on to YPUAD solid medium and incubated at 24 C (left) or 40 C (right) Plates were photographed after 4 or 12 days, respectively.
Trang 5Conserved motifs adjacent to and within the
Vps4p b domain are required for interaction with
Vta1p, but not Vps20p or Did2p
A number of Vps4p-interacting proteins have
previ-ously been identified in a yeast two-hybrid screen
including the class E Vps proteins Vps20p and Vta1p [30] These interactions were confirmed by the demon-stration that recombinant glutathione S-transferase (GST)-Vps20p or GST-Vta1p bind GFP-tagged Vps4p present in yeast cell lysates as well as to recombinant His6-tagged Vps4p in vitro Using yeast
Vps4p-GAI
Vps4p-L TP
Vps4p-RDE
Vps4p-WT
empt y
vector
Vps4p-RKI
Vps4p-GAI
Vps4p-LTP
Vps4p-RDE
Vps4p-WT
empty vector
Vps4p-RKI
Fluorescence Nomarski
C
Vps4p-WT
Vps4p-GAI
Vps4p-RDE
no vector
empty vector Vps4p-RK I Vps4p- LT P wild-type
Vps4p-WT
Vps4p-GAI
Vps4p-RDE
no vector
empty vector Vps4p-RK I Vps4p- LT P wild-type
CPY blo t
Calmodulin blo t
D
Vps4p-WT empty vector
Vps4p-RK I
Vps4p-GAI Vps4p-L TP Vps4p-RDE
40 C O
24 C O
Trang 6two-hybrid assays, Vps20p was shown to interact
strongly with the N-terminal region of Vps4p and
weakly with both full-length Vps4p and a C-terminal
region containing residues 351–437 Vta1p interacts
strongly with full-length Vps4p and with the
C-ter-minal region, but does not interact with the
N-ter-minal region Did2p, also known as Chm1p⁄
Fti1p⁄ Vps46p, is also a class E Vps protein which
interacts with Vps4p [30,32], and this interaction is
also conserved in mammalian cells [33] We used a
yeast two-hybrid assay to determine the region of
Vps4p that interacts with Did2p (Fig 3A) Did2p binds strongly to full-length Vps4p and, like Vps20p, binds very strongly to the N-terminal region of Vps4p and only weakly to the C-terminal region This is consistent with a recent finding that the MIT domain
of VPS4A interacts with CHMP1B [34]
Loss of interaction with these proteins may be responsible for loss of Vps4p function caused by the C-terminal mutations We therefore tested whether the conserved motifs in the C-terminal region of Vps4p are required for two-hybrid interaction with each of these proteins (Fig 3B) Vta1p interacted with wild-type Vps4p but not with the Vps4p-RKI, Vps4p-GAI or Vps4p-LTP mutant proteins In contrast with the other mutant proteins, Vps4p-RDE retained the ability to interact with Vta1p As expected, all the Vps4p mutant proteins retained the ability to interact with Vps20p This interaction, which is mediated by the N-terminal region, should not be directly affected by C-terminal mutations Interestingly Vps4p-RKI, Vps4p-GAI and Vps4p-LTP exhibited an apparent increase in interac-tion with Vps20p The interacinterac-tion with Did2p was unaffected by any of the mutations
A
Vps4p
pLexA
Vps4p-RDE
Vps4p-RKI
Vps4p-GAI
Vps4p-LTP
pB42AD Did2p Vps20p Vta1p
B
pLexA
Vps4p
N-Vps4p
C-Vps4p
AAA-Vps4p
C
GST-Vps20p
GST-Vta1p
GST
-bound
unbound
bound
unbound
bound
unbound
-ATP
ATP
Vps4p-6His Vps4p-GAI-6His
Vps4p-6His Vps4p-GAI-6His
Vps4p-6His Vps4p-GAI-6His
Fig 3 Role of conserved Vps4p C-terminal motifs in interaction with Vta1p, Vps20p and Did2p (A) Yeast two-hybrid interaction anal-ysis of Did2p (residues 41–204 ⁄ end) with full-length wild-type Vps4p (residues 1–437 ⁄ end), the N-terminal region of Vps4p (N-Vps4p; residues 1–128), the previously predicted AAA domain (AAA-Vps4p; residues 129–350) [8] and C-terminal region (C-Vps4p; residues 351–437 ⁄ end) The interaction analyses with Vta1p (residues 108–333 ⁄ end) and Vps20p (residues 3–221 ⁄ end) were included as controls (B) Yeast two-hybrid interaction analysis of wild-type Vps4p and Vps4p C-terminal mutants with the same frag-ments of Vta1p, Vps20p and Did2p as in (A) In (A) and (B), EGY48 carrying pLexA-based bait plasmids and pB42AD-based prey plas-mids as well as p8op-LacZ reporter plasmid were spotted on to medium containing X-gal Plates were photographed after overnight incubation, and two-hybrid interaction was assessed by blue color-ation Two independent transformants are shown in (B) (C) In vitro binding of His 6 -tagged wild-type Vps4p and Vps4p-GAI to GST-Vps20p and GST-Vta1p Equal amounts of full-length His6-tagged proteins were incubated with glutathione ⁄ agarose bearing GST-Vta1p, GST-Vps20p or GST alone in the presence or absence of ATP The unbound protein was recovered in the supernatants Bound protein was released with Laemmli sample buffer The bound and unbound fractions were subjected to SDS ⁄ PAGE and immunoblotting with a polyclonal antibody to Vps4p The amount of wild-type Vps4p-His6or Vps4p-GAI-His6bound to GST-Vps20 was quantified by densitometry The shift in the relative positions of the wild-type Vps4p-His 6 and Vps4p-GAI-His 6 bound to GST-Vps20 is due to the presence of the GST-Vps20 protein, which migrates very close to the His 6 -tagged proteins.
Trang 7Because yeast two-hybrid is an indirect measure
of binding strength, we used in vitro protein-binding
assays to confirm the effect of the Vps4p-GAI
muta-tion on the interacmuta-tion with Vps20p and Vta1p Both
wild-type Vps4p and Vps4p-GAI were expressed with
His6 affinity tags in Escherichia coli and used in
binding assays with GST-Vta1p and GST-Vps20p
(Fig 3C) Consistent with the yeast two-hybrid data,
the Vps4p-GAI-His6 mutant protein did not interact
with Vta1p Also consistent with our yeast two-hybrid
results, binding to GST-Vps20p was increased
com-pared with wild-type Vps4p-His6 (both in the presence
or absence of ATP) In the presence of ATP, there was
a 35% decrease in the amount of Vps4p-GAI bound
to GST-Vps20p In contrast, there was an 87%
decrease in the amount of wild-type Vps4p bound to
GST-Vps20p Hence, the ATPase-dependent
dissoci-ation of GST-Vps20p is affected by deletion of the
conserved GAI motif in Vps4p
We conclude that the RKI, GAI and LTP
motifs within and adjacent to the Vps4p b domain
are essential for interaction with Vta1p The
increased interaction with Vps20p, when these motifs
are deleted may be due to defective
ATPase-depend-ent dissociation, which we showed directly for
Vps4p-GAI
The Vps4p GAI motif in the b domain is not
required for Vps4p recruitment to endosomes
The Vps4p N-terminal domain has been shown to play
a key role in recruitment of Vps4p to endosomes, but
whether recruitment also requires the C-terminal
domain was not tested [24] To test whether the
C-ter-minal GAI motif is important for Vps4p recruitment
to endosomes, we examined the subcellular localization
of GFP-tagged Vps4p-GAI and compared it with that
of GFP-tagged wild-type Vps4p (Fig 4A) In vps4D
yeast cells, the fluorescence distribution of wild-type
Vps4p-GFP and Vps4p-GAI was both diffuse and
localized to punctate structures that are likely to be
endosomes In contrast, the subcellular distribution of
GFP-tagged Vps4p-DCC, which lacks the N-terminal
MIT domain, was diffuse in the cytoplasm This is
consistent with a critical role for the N-terminal MIT
domain in recruitment of Vps4p to endosomal
mem-branes that has previously been described [24]
Fur-thermore, when the DCC mutation was introduced
into Vps4p-GAI, the subcellular distribution also
became diffuse in the cytoplasm We conclude that
Vps4p-GAI, but not Vps4p-DCC-GAI, is recruited to
endosomal membranes This suggests that the
C-ter-minal GAI motif is not essential for Vps4p recruitment
to endosomes
The Vps4p GAI motif is not essential for ATPase activity
To determine whether the GAI motif in the b domain is important for Vps4p ATPase activity, perhaps via con-formational effects on the AAA domain, we assayed the ATPase activity of His6-tagged Vps4p-GAI and com-pared it with that of His6-tagged wild-type Vps4p The affinity-purified Vps4p-GAI-His6 mutant protein was difficult to obtain as a full-length protein from bacteria and most preparations contained some degradation products In the best preparations, 30% of the protein was full-length as determined by densitometry When equivalent amounts of full-length protein were assayed
in the presence of 0.1 mm ATP, the Vps4p-GAI-His6 protein had 52% of wild-type activity (Fig 5) However, when equal amounts of full-length protein were assayed
in the presence of 1 mm ATP, the Vps4p-GAI-His6 pro-tein had only 14% of wild-type activity We conclude that the Vps4p GAI motif in the b domain is important, although not essential for Vps4p ATPase activity
Fluorescence Nomarski
vps4∆/
Vps4p-GFP
vps4∆/
Vps4p-GAI-GFP
vps4∆/
Vps4p-∆CC-GFP
vps4∆/
Vps4p-∆CC-GAI-GFP
Fig 4 The GAI motif in the b domain is not essential for localiz-ation of Vps4p to endosomes (A) RH2906 vps4D yeast cells expressing GFP-tagged wild-type Vps4p, Vps4p-GAI, Vps4p-DCC or Vps4p-DCC-GAI were grown in SD medium, and the GFP-tagged proteins visualized by fluorescence microscopy Scale bar, 5 lm.
Trang 8The phenotypes conferred by mutation of the
conserved motifs adjacent to and within the
Vps4p b domain are not dominant
To investigate whether mutation of the Vps4p RKI,
GAI and LTP motifs confers dominant phenotypes,
like most previously characterized vps4 mutants, we
expressed the mutant proteins in wild-type cells and
examined the effect on Vps4p function Wild-type
cells expressing the Vps4p-RKI, Vps4p-GAI and
Vps4p-LTP mutants did not missort and secrete
CPY, grew well at 40C, and transported
Fth1p-Ub-GFP to the vacuole lumen (data not shown) In
contrast, wild-type cells expressing the
dominant-neg-ative Vps4p-E233Q mutant [24] showed the full
range of dominant-negative phenotypes This
indi-cates that the defects conferred by mutation of the
Vps4p-RKI, Vps4p-GAI and Vps4p-LTP motifs are
not dominant
An intact b domain is essential for Vps4p
self-association
Although wild-type Vps4p expressed in bacteria is a
dimer [24], previous studies using the yeast two-hybrid
technique did not reveal a homotypic interaction in
wild-type Vps4p [25] We used a different yeast
two-hybrid system to determine whether we could detect
this homotypic interaction The data obtained show that wild-type Vps4p can interact with itself strongly (Fig 6) This is consistent with previous data showing that Vps4p forms a dimer
As the b-domain mutants are recessive, we consid-ered the possibility that these mutants may not be able
to interact with wild-type Vps4p To test this, we used the yeast two-hybrid assay to study the interaction between Vps4p-GAI and wild-type Vps4p However, the Vps4p-GAI mutant protein barely interacted with wild-type Vps4p and did not interact with itself (Fig 6) The Vps4p-RKI and Vps4p-LTP mutant pro-teins also did not interact with wild-type Vps4p (data not shown) We surmise that an intact b domain is required for Vps4p self-association
An intact b domain is required for the Vps4p-E233Q mutant to have a dominant-negative phenotype
The E233Q mutation in Vps4p confers a dominant-negative phenotype [8,28], and the equivalent muta-tion in mammalian VPS4 isoforms has been widely used to study the effect of VPS4 inactivation in mammalian cells [35,19] Vps4p-E233Q forms a 10– 12-mer [24,27], and it has been speculated that wild-type Vps4p may also form a high-molecular-mass oligomer that is transient in the presence of a func-tional ATPase domain The dominant-negative pheno-type conferred by Vps4p-E233Q is believed to be due
to interaction of Vps4p-E233Q with wild-type Vps4p
in vivo based on the ability of Vps4p-E233Q to
oligo-1.0 0.1
GAI WT GAI WT
[ATP] mM
0
5
10
15
20
25
Fig 5 The Vps4p-GAI mutant has a diminished ATPase activity.
Affinity-purified His6-tagged wild-type Vps4p (WT) and Vps4p-GAI
(GAI) were assayed in vitro for ATPase activity at 30 C The
amount of protein assayed was normalized based on the level of
full-length wild-type Vps4p and Vps4p-GAI ATPase activity was
assayed in the presence of 0.1 m M ATP or 1 m M ATP and is
expressed as nmol inorganic phosphate releasedÆh)1Æ(lg full-length
protein))1and shown graphically.
pLexA/ pB42AD pLexA Vps4p/ pB42AD Vps4p pLexA Vps4p/ pB42AD Vps4p-GAI pLexA Vps4p-GAI/ pB42AD Vps4p pLexA Vps4p-GAI/ pB42AD Vps4p-GAI
Fig 6 The conserved GAI motif in the b domain of Vps4p is required for homotypic interaction between wild-type Vps4p mole-cules The interaction between various combinations of wild-type Vps4p and Vps4p-GAI was assessed using the yeast two-hybrid technique EGY48 carrying a p8op reporter plasmid and pLexA-based bait plasmids and pB42AD-pLexA-based prey plasmids were spot-ted on to synthetic galactose medium containing X-gal Plates were photographed after 2 days, and two-hybrid interaction was assessed by blue coloration Three independent transformants are shown.
Trang 9merize with Vps4p-DCC [24] in vitro To demonstrate
a role for the b domain in Vps4p assembly in vivo,
we mutated the GAI motif in the dominant-negative
Vps4p-E233Q mutant Unlike Vps4p-E233Q, the
resulting double mutant (Vps4p-E233Q-GAI) did not
confer dominant-negative phenotypes (Fig 7A–C) As
the expression level of Vps4p-E233Q-GAI is similar
to that of Vps4p-E233Q (Fig 7D), loss of the
domin-ant-negative phenotype is not due to decreased
expression of the double mutant Our data strongly
suggest that the b domain is required for interaction
with wild-type Vps4p in vivo
Vps20p and Vta1p both stimulate Vps4p ATPase
activity, but Vps20p stimulates its activity to a
greater extent
As the b domain of Vps4p is required for full ATPase
activity as well as for interaction with Vta1p, it is
poss-ible that binding of Vta1p to Vps4p might have an
effect on ATPase activity To test this hypothesis, we
assayed the in vitro ATPase activity of wild-type His6 -tagged Vps4p in the presence of increasing concentra-tions of GST-Vta1p We also assayed the activity of Vps4p in the presence of increasing concentrations of GST and GST-Vps20p We included GST-Vps20p because Vps20p binds to Vps4p in an ATP-sensitive manner, and substrate binding is known to increase the activity of some AAA ATPases such as Hsp104 and Katanin [36,37] The data in Fig 8 show that both Vta1p and Vps20p have a stimulatory effect on Vps4p ATPase activity, but the effect of Vps20p is much greater It is interesting that, although the b domain is required for assembly of Vps4p into a complex with full ATPase activity, binding of Vta1p to the b domain does not inhibit Vps4p ATPase activity
Discussion Here we identify conserved motifs in the C-terminal region of Vps4p and provide evidence that those within the b domain are critical for all Vps4p in vivo
A
D
Vps4p-E233Q Vps4p-E233Q-GAI
Vps4p actin
Vps4p-E233Q
Vps4p- E233Q-GAI
Fluorescence Nomarski
C
24O
C Vps4p-E233Q
Vps4p-E233Q-GAI
B
Vps4p-E233Q Vps4p-E233Q-GAI
Vps4p-E233Q Vps4p-E233Q-GAI
CPY blot
calmodulin blot
Fig 7 The dominant-negative Vps4p-E233Q mutant becomes recessive upon mutation of the GAI motif (A) The Vps4p-E233Q-GAI mutant protein does not confer a dominant-negative MVB sorting defect RH1800 (wild-type) yeast cells expressing Fth1p-GFP-Ub and either E233Q or E233Q-GAI were grown in SD selective medium and assayed for MVB sorting as in Fig 2 Scale bar, 5 lm (B) The Vps4p-E233Q-GAI mutant protein does not confer a dominant-negative vacuolar protein sorting defect RH1800 (wild-type) yeast cells expressing Vps4p-E233Q or Vps4p-E233Q-GAI were grown on selective SD solid medium at 24 C in contact with a nitrocellulose filter CPY missorting was assayed as in Fig 2 (C) The Vps4p-E233Q-GAI mutant protein does not confer a dominant-negative growth defect Wild-type RH1800 yeast cells expressing Vps4p-E233Q or Vps4p-E233Q-GAI were assayed for growth at high temperature on solid SD selective medium as in Fig 2 (D) Total cell lysates from RH2906 (vps4D) yeast cells carrying plasmids expressing Vps4p-E233Q or Vps4p-E233Q-GAI were subjec-ted to western blotting as in Fig 1D.
Trang 10functions including fluid-phase endocytosis, MVB
sort-ing, vacuolar protein sorting and growth at high
tem-perature Two of these motifs, LTP and GAI, are in
the b domain and the third, RKI, is partly within the
b domain and partly within the AAA domain We
pro-vide epro-vidence that the b domain is important for full
ATPase activity of Vps4p We also show that the b
domain is required for two protein interactions The
first is a homotypic interaction that may be important
for assembly of a fully catalytically active oligomer,
and the second is with Vta1p Both of these
interac-tions are likely to be important for Vps4p function
in vivo We also show that the charged residues in an
RDE motif at the end of the final C-terminal a-helix
are not required for most Vps4p functions but are
required for full growth at high temperature
More-over, these charged residues are not required for Vta1p
interaction
Several lines of evidence suggest that mutation of
these motifs have specific effects on Vps4p in vivo
func-tion Deletion of the motifs did not compromise stable
expression, indicating that loss of in vivo function is
not merely due to lowered expression levels The
Vps4p proteins carrying mutations in the b domain
retained at least part of their native structure, as they
were able to interact with Vps20p and Did2p, which
interact with the N-terminal region of Vps4p ([30]; this
study) This was shown by yeast two-hybrid assay and
confirmed in one b domain mutant using an in vitro
binding assay In addition, the retention of native
structure in b domain mutant proteins is supported by
the ability of a b domain mutant protein to be
recrui-ted efficiently to endosomes in vivo as assessed visually
by microscopy Hence it is possible to mutate the b domain without grossly affecting Vps4p structure Moreover, based on the structure of the human VPS4B monomer, the b domain is an independent domain that is separated from the AAA domain by random coils (Fig 1E), thus mutations in this domain are unli-kely to perturb the structure of the AAA domain However, we cannot exclude the possibility that dele-tion of the conserved motifs within the b domain may perturb the local structure of the b domain
The requirement for an intact b domain for full ATPase activity is clearly demonstrated in the in vitro ATPase assays and further supported by interaction analysis of the b domain mutant with Vps20p The Vps4p–Vps20p interaction is known to be sensitive to ATP hydrolysis and is stabilized by the E233Q muta-tion that perturbs Vps4p ATPase activity [30] Consis-tent with a decreased ATPase activity, interaction of the Vps4p b domain mutant with Vps20p was enhanced in the in vitro binding assay and in the
in vivoyeast two-hybrid assay
Most AAA ATPases assemble into higher-order oligomeric rings Consistent with this, wild-type Vps4p purified from bacteria forms a dimer [24], and Vps4p forms a 10–12-mer when its ATPase activity
is compromised and it is locked in the ATP-bound conformation, as in the Vps4p-E233Q mutant [24,27] Arg352 in the AAA domain of Vps4p has recently been shown to be important for assembly of Vps4p-E233Q into oligomers but not for dimer for-mation [27] The motifs of Vps4p required for dime-rization have not previously been identified Previous studies using a yeast two-hybrid assay have
demon-1
0 2 4 6 8
ratio of GST alone or GST-fusion protein ( µg): Vps4p (µg)
GST alone+Vps4p GST alone GST-Vps20p+Vps4p GST-Vps20p GST-Vta1p+Vps4p GST-Vta1p
Fig 8 Vps20p strongly stimulates Vps4p ATPase activity, but binding of Vta1p to the b domain of Vps4p has only a marginal stimulatory effect The ATPase activity of recombinant Vps4p-His 6 was assayed at 30 C in the presence of 0.3 m M ATP and increasing amounts of recombinant GST-Vta1p, GST-Vps20p or GST alone ATPase activity is expressed as nmol inorganic phosphate releasedÆh)1Æ(lg Vps4p))1 The ATPase activities of GST alone and the GST fusion proteins without any Vps4p were also assayed and are shown.