159

CELL STRUCTURE AND FUNCTION 31: 159–172 (2006)

© 2006 by Japan Society for Cell Biology

AMSH, an ESCRT-III Associated Enzyme, Deubiquitinates Cargo on MVB/Late Endosomes

Masanao Kyuuma1,6, Kazu Kikuchi1,6, Katsuhiko Kojima2, Yuriko Sugawara1, Mariko Sato1,

Nariyasu Mano3, Junichi Goto3, Toshikazu Takeshita2, Akitsugu Yamamoto4, Kazuo Sugamura1, and

Nobuyuki Tanaka1,5∗

1Department of Microbiology and Immunology, Tohoku University Graduate School of Medicine, 2-1 Seiryo-

machi, Sendai, 980-8575, Japan, 2Department of Microbiology and Immunology, Shinshu University School

of Medicine, 3-1 Asahi, Matsumoto, 390-8621 Japan, 3Tohoku University Graduate School of Pharmaceutical

Sciences, Aobayama, Sendai, 980-8578, Japan, 4Faculty of Bioscience, Nagahama Institute of Bio-Science

and Technology, Nagahama, Shiga, 526-0829, Japan, 5Division of Immunology, Miyagi Cancer Center

Research Institute, Miyagi, 981-1293 Japan, and 6These two authors contributed equally to this work.

ABSTRACT. The appropriate sorting of vesicular cargo, including cell-surface proteins, is critical for many cel-

lular functions. Ubiquitinated cargo is targeted to endosomes and digested by lysosomal enzymes. We previously

identified AMSH, a deubiquitination enzyme (DUB), to be involved in vesicular transport. Here, we purified an

AMSH-binding protein, CHMP3, which is an ESCRT-III subunit. ESCRT-III functions on maturing endosomes,

indicating AMSH might also play a role in MVB/late endosomes. Expression of an AMSH mutant lacking

CHMP3-binding ability resulted in aberrant endosomes with accumulations of ubiquitinated cargo. Nevertheless,

CHMP3-binding capability was not essential for AMSH’s in vitro DUB activity or its endosomal localization, sug-

gesting that, in vivo, the deubiquitination of endosomal cargo is CHMP3-dependent. Ubiquitinated cargo also

accumulated on endosomes when catalytically inactive AMSH was expressed or AMSH was depleted. These

results suggest that both the DUB activity of AMSH and its CHMP3-binding ability are required to clear ubi-

quitinated cargo from endosomes.

Key words: deubiquitination/endosome/ESCRT-III/MVB/ubiquitin

Introduction

The endosomal system of mammalian cells forms a dynamic

membranous network by which membrane proteins are

transported and/or degraded. A variety of cell-surface pro-

teins are endocytosed and translocated to early endosomes

as cargo. Although some receptors are recycled back to the

cell surface, cargo destined to be degraded is sorted into the

multivesicular body (MVB) (Hicke, 2001; Katzmann et al.,

2002). Cargo proteins concentrated on the limiting mem-

branes of the MVB are internalized by invagination, and

the membrane is pinched off to form intraluminal vesicles

(Gruenberg and Stenmark, 2004). Thus, MVB sorting of pro-

teins exposes them to degradation by lysosomal enzymes.

Accumulating evidence suggests that cargo sorting and

MVB formation require at least 18 conserved molecules,

called the class E Vps proteins (Babst, 2005). Mutation of

any of the genes encoding these proteins in yeast results in

the formation of an aberrant endosome called the “class E

compartment.” Class E Vps proteins are highly conserved

between yeast and mammals, and their mutation or deletion

in mammals results in the formation of a similar class E

structure (Kanazawa et al., 2003; Kobayashi et al., 2005).

These Vps proteins are constituents of at least four separate

heteromeric protein complexes called ESCRT-0 (Endo-

somal Sorting Complex Required for Transport-0, also

known as the STAM-HRS complex), ESCRT-I, ESCRT-II,

*To whom correspondence should be addressed: Nobuyuki Tanaka,Department of Microbiology and Immunology, Tohoku University GraduateSchool of Medicine, 2-1 Seiryo-machi, Sendai, 980-8575 Japan.

Tel: +81–22–717–8096, Fax: +81–22–717–8097E-mail: n-tanaka@mail.tains.tohoku.ac.jp

Abbreviations: AMSH, associated molecule with the SH3 domain ofSTAM; AMSH-LP, AMSH-like protein; DUB, deubiquitinating enzyme;STAM1 and STAM2, signal transducing adaptor molecule 1 and 2; Hrs,hepatocyte growth factor-regulated tyrosine kinase substrate; MVB, multi-vesicular bodies; ESCRT, endosomal sorting complexes required for trans-port; EGF, epidermal growth factor; Vps, vacuolar protein sorting; shRNA,short hairpin RNA.

160

M. Kyuuma et al.

and ESCRT-III.

Cargo first captured by ESCRT-0 on early endosomes is

handed off to ESCRT-I and subsequently to ESCRT-II.

Close to the final step of sorting, the ESCRT-III complex,

which is composed of CHMP1, CHMP2, CHMP3, CHMP4,

CHMP5, and CHMP6, concentrates the cargos on the

maturing endosomes (Raiborg et al., 2003; Gruenberg and

Stenmark, 2004; Morita and Sundquist, 2004). Finally,

additional class E VPS proteins, Vps4A and Vps4B, disso-

ciate the ESCRT complex from the endosomal membrane in

an ATP-dependent manner, allowing the cargo to be invagi-

nated into the MVB (Babst et al., 1997).

The MVB sorting machinery is mediated by ubiquitina-

tion. Ubiquitin is a highly conserved protein of 76 amino

acids that can be covalently conjugated to specific protein

substrates by a cascade of ubiquitin-conjugating enzymes,

including catalytic proteins called E3. Ubiquitination was

first identified as polyubiquitin chains on target proteins,

which lead the proteins to be degraded by proteasomes

(Weissman, 2001; Glickman and Ciechanover, 2002). In

contrast, monoubiquitination of proteins was shown to tar-

get the proteins to the MVB pathway (Bonifacino and

Traub, 2003). The single ubiquitin moiety is recognized by

a variety of proteins involved in cargo sorting, through their

various ubiquitin-interacting domains. In the initial sorting

events after monoubiquitination, the ubiquitin moiety is first

recognized by the UIM and VHS domains of STAM and

Hrs (Asao et al., 1997; Mizuno et al., 2003). Likewise, the

UEV domain of the ESCRT-I subunit Tsg101 also functions

in cargo sorting (Katzmann et al., 2002). Thus, ubiquitina-

tion is very important for the accurate sorting of cargo.

In studies aimed at finding molecules associated with

STAM, we identified the AMSH deubiquitinating (DUB)

enzyme (Tanaka et al., 1999; McCullough et al., 2004).

AMSH possesses an isopeptidase domain characterized by

the JAB1/MPN/Mov34 metalloenzyme (JAMM) domain. In

an in vitro assay, recombinant AMSH removed ubiquitin

moieties from K63-linked but not K48-linked ubiquitins

(McCullough et al., 2004). Since K63 ubiquitin, like mono-

ubiquitin, is associated with cargo sorting, the K63-specific

DUB activity suggested that AMSH might function in MVB

cargo sorting. However, a genetic approach in yeast failed

to identify any AMSH orthologues in MVB sorting. Indeed,

there is no evidence from the database that yeast possesses

an AMSH orthologue. Instead, yeast has a DUB called

Doa4 that functions in MVB sorting proximal to the last steps

(Swaminathan et al., 1999; Amerik et al., 2000; Katzmann

et al., 2001; Babst et al., 2002; Luhtala and Odorizzi, 2004).

Thus, it has been unclear whether a DUB is involved in

MVB sorting in mammals.

To further characterize the function of AMSH in MVB

formation and/or protein sorting, we sought AMSH-binding

proteins. Here we provide evidence that AMSH is involved

in cargo sorting on MVB/late endosomes through its bind-

ing to the ESCRT-III complex.

Materials and Methods

Cell culture

HeLa and 293T cells were cultured in Dulbecco’s modified

Eagle’s medium (DMEM) supplemented with 10% fetal calf

serum (FCS) and antibiotics, under 7% CO2 in a humidified incu-

bator. FuGENETM6 transfection reagent (Roche) was used to trans-

fect the cells with plasmid DNAs, following the manufacturer’s

protocol. For cross-linking experiments, cells were suspended in

10 ml PBS containing 1 mM MgSO4 (pH 8.3) and 100 µg/ml DSP

(Pierce) on ice for 20 min, then washed extensively with PBS. Cell

lysates were prepared and used for further experiments, as

described below. For EGF stimulation experiments, cells were

incubated with 200 µM E-64-d and 20 µM pepstatin A (Peptide

Institute Inc.), along with cycloheximide (CHX), in starvation

medium (DMEM containing 0.1% BSA) for 3 hours, and then

stimulated with 100 ng/ml hEGF (PeproTech).

Plasmids

A wild-type AMSH expression vector, p3xFLAG-AMSH, was

generated by inserting the human AMSH cDNA into p3xFLAG-

CMV-10 (Sigma). Mutant AMSH expression vectors were gener-

ated by PCR-based procedures (Kobayashi et al., 2005). An

expression plasmid for a catalytically inactive AMSH, p3xFLAG-

AMSHE280A, was generated using the QuickChangeTM site-directed

mutagenesis kit (Stratagene) according to the manufacturer’s

instructions. For the EGFP-tagged AMSH expression, pEGFP-

AMSH was generated by inserting the PCR-amplified cDNA into

pEGFP-C3 (Clontech). For the expression of GST-fused recom-

binant AMSH and CHMP3, pGEX-6P-3-AMSH and pGEX-6P-

3-CHMP3 were generated using pGEX-6P-3 (GE Healthcare).

pMyc-CHMP3 is an expression vector for the Myc-tagged wild-

type CHMP3, constructed by inserting the human CHMP3 cDNA

into pcDNA-Myc (Invitrogen). Myc-tagged CHMP3 mutants were

generated by a PCR-based method. pDsRed-CHMP3 was con-

structed by inserting the CHMP3 cDNA into pDsRed-Monomer-

N1 (Clontech). pcDNA3.1-V5-CHMP2A and pcDNA3.1-V5-

CHMP2B are expression vectors for wild-type CHMP2A and 2B,

constructed using pcDNA3.1/V5-HisA (Invitrogen). p3xFLAG-

VPS4 and p3xFLAG-VPS4E228Q were generated by inserting the

PCR-amplified cDNAs into p3xFLAG-CMV-10. pcDNA3.1-V5-

EGFR was generated using pcDNA3.1-V5 (Invitrogen). pcDNA3.1-

Myc-Ubiquitin is an expression vector for wild-type Ubiquitin,

constructed by inserting the Ubiquitin cDNA into pcDNA3.1/

Myc-HisA (Invitrogen). HA-c-Cbl is an HA-tagged E3 ubiquitin

ligase that targets EGFR, and was a generous gift from Dr Yosef

Yarden (Weizmann Institute of Science). The DNA sequences

were verified using a DNA sequencer (ABI3100, Applied Bio-

science) and the BigDye Terminator kit (Applied Bioscience).

Immunoprecipitation and immunoblots

Immunoprecipitation and immunoblotting were carried out as

Roles of AMSH on MVB/Late Endosomes

161

described previously. In brief, cells were lysed in the NP-40 Lysis

buffer (1% Nonidet P-40, 40 mM Tris-HCl [pH 7.5], 150 mM

NaCl, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 20

µg/ml aprotinin). The cell lysates were pre-cleared of cellular

debris by centrifugation (10,000×g) for 30 min at 4°C, and then

subjected to immunoprecipitation with antibodies immobilized on

Protein A-Sepharose beads (GE Healthcare) at 4°C, overnight. For

this assay, the monoclonal antibodies (mAbs) anti-Myc (Santa

Cruz), anti-FLAG-M2 (Sigma), anti-V5 (Invitrogen), anti-LAMP1

(Santa Cruz), and anti-AMSH (Tanaka et al., 1999), and poly-

clonal anti-α-tubulin (Sigma) were used. Anti-CHMP3 antisera

were prepared by immunizing rabbits with recombinant full-length

human CHMP3 protein. The immunoprecipitates were separated

by SDS-PAGE and transferred onto PVDF membranes (Milli-

pore). After blocking with 5% nonfat milk in Tris-buffered saline

(TBS) containing 0.1% Tween 20, the membranes were probed

with the indicated primary antibodies. After three washes, the

membranes were probed with HRP-conjugated secondary anti-

bodies (Cell Signaling Technology). Signals were visualized with

a Super signal pico detection kit (Pierce), and digital images were

collected with a Lumi-Imager F1 (Roche).

Preparation of cytosol and membrane fractions

Cells were washed with PBS and lysed in homogenization buffer

(10 mM HEPES, 3 mM imidazole, 250 mM sucrose) by repeated

passage through a 22G needle at 4°C. A postnuclear fraction

(PNF) supernatant was obtained by centrifugation for 10 min at

3000×g. The PNF was subjected to ultracentrifugation for 30 min

at 100,000×g, and the cytosolic and membrane fractions were sep-

arated.

Preparation of bacterially expressed recombinant proteins and a GST pull-down assay

GST-CHMP3 (2 µg), purified from E. coli BL21, was incubated

with cell lysates prepared from 293T cells transfected with various

AMSH-expression vectors and further mixed with 10 µl of Glu-

tathione-Sepharose beads (GE Healthcare) for one hour, at 4°C.

The beads were extensively washed with the NP-40 Lysis buffer,

and proteins associated with the beads were subjected to immuno-

blotting with anti-GST antisera (MBL) or anti-FLAG-M2 mAb.

Likewise, GST-AMSH recombinant protein was used to detect the

association with various CHMP3 constructs expressed by the

transfected 293T cells. For controls, 2% of the transfected cell

lysate was used to monitor the protein.

In vitro deubiquitination assay

K48-linked or K63-linked polyubiquitin chains (500 ng; Boston

Biochem) were incubated with 5 µg of GST-fusion proteins at

37°C for 18 hours in DUB buffer (PBS [pH 7.0], 5 mM MgCl2, 2

mM DTT) (Mizuno et al., 2005). Tris-Tricine Sample Buffer (0.1

M Tris-HCl [pH 6.8], 24% Glycerol, 1% SDS, 2% β-mercapto-

ethanol, and 2% Coomassie G-250) was added, and the reaction

mixture was heated to 95°C for 5 minutes and resolved on 16.5%

Ready Gels J Peptide (BioRad). Ubiquitin was detected using an

anti-ubiquitin mAb, P4D1 (Santa Cruz).

Depletion of cellular AMSH by shRNA

A human AMSH-specific short hairpin RNA (shRNA), pSIREN-

RetroQ-AMSH was generated using pSIREN-RetroQ (BD Bio-

sciences). The target sequences within human AMSH cDNA were

nucleotide residues 651–669 (5'-GCAGCAATTGGAACAGGAA-

3') and 463–481 (5'-GTAAAGAAATTAAAGGAGA-3'). A con-

trol plasmid, pSIREN-RetroQ-pRL-Null, targeted the 413–434

(5'-GCAATAGTTCACGCTGAAAAG-3') sequence of the firefly

luciferase gene. For retrovirus-mediated transfer of these con-

structs, 293T cells in 10-cm dishes were transfected with 15 µg of

pSIREN-RetroQ-AMSH along with 10 µg of pMD.G (a gift from

Dr. I. M. Verma) and 5 µg of gagpol-IRES-bsr (a gift from Dr. T.

Kitamura), by the calcium phosphate precipitation method. At 12

hours post-transfection, the cell culture medium was replaced. At

48 hours post-transfection, the supernatants were collected, filtered

through a 0.45-µm syringe filter, and spun at 6,000×g for 20 hours

at 4°C. The virus obtained from the pellet was used for further

transduction experiments.

Immunofluorescence microscopy and immunolabeling

HeLa cells were grown in 35-mm glass-bottomed dishes

(Matsunami) at a density of 4×104 cells/dish and transfected with

pDsRed1-CHMP3 or pEGFP-AMSH. After 48 hours, the cells

were washed twice with phosphate-buffered saline (PBS), fixed

with 4% paraformaldehyde (PFA) for 15 min, washed in wash

buffer (0.1% TritonX-100/PBS), and then blocked in the same

buffer containing 10% FCS. For immunostaining, the fixed sam-

ples were incubated in wash buffer containing 5% FCS at 4°C

overnight with the indicated primary antibodies: anti-EEA1, anti-

LAMP2 antisera (Santa Cruz), and anti-LAMP1 (Santa Cruz),

anti-LBPA (a gift from Dr. T. Kobayashi) (Kobayashi et al.,

1998), and FK2 (Affiniti Research Products) mAbs. The anti-

AMSH antiserum was described previously (Tanaka et al., 1999).

After three washes, the samples were further incubated with the

secondary antibodies (anti-rabbit, -mouse, and -goat IgG anti-

bodies conjugated with Alexa488, Alexa594, and Alexa633,

respectively, Molecular Probes) at room temperature for one hour.

The confocal microscopic images were examined using the 510

META microscope with a 60/1.30–0.60 oil-immersion objective

(Carl Zeiss).

Immuno-electron microscopy

For immuno-electron microscopy, the pre-embedding silver-

enhanced immunogold method was performed as previously

described (Yoshimori et al., 2000). HeLa cells were cultured on

plastic coverslips (LF; Sumitomo Bakelite).

162

M. Kyuuma et al.

Results

AMSH associates with the ESCRT-III protein CHMP3

To explore the function of AMSH on endosomes, we used a

co-immunoprecipitation-based strategy to identify AMSH-

binding proteins. 3×FLAG-tagged AMSH was expressed in

293T cells, and the anti-FLAG mAb-directed immunopre-

cipitates were analyzed using MALDI-TOF MASS spec-

trometry. Among the immunoprecipitates was a 33-kDa

band corresponding to human CHMP3, an ESCRT-III sub-

unit (Supplementary Fig. S1). To verify the association

between AMSH and CHMP3, we performed co-immuno-

precipitation assays of the FLAG-tagged AMSH and myc-

tagged CHMP3. The association of these two molecules

was clearly observed in the immunoblotted immunoprecipi-

tates only when both proteins were transiently expressed

in the 293T cells (Fig. 1A). To confirm this association in

native cells, we performed a co-immunoprecipitation assay

using HeLa cells. Anti-AMSH mAb-directed immuno-

precipitates included CHMP3, although we could not detect

AMSH in immunoprecipitates brought down with anti-

CHMP3 antisera (Fig. 1B). Next, to clarify the CHMP3-

association site within AMSH, we performed GST pull-down

assays using the following AMSH mutants: AMSHdN1,

AMSHdN2, AMSHdN3, AMSHdCC, AMSHdSBM, and AMSHdBS3

(see Fig. 1C). GST-CHMP3 showed direct binding to wild-

type AMSH expressed in 293T cells as well as to AMSHdCC,

AMSHdSBM, and AMSHdBS3, whereas CHMP3 did not asso-

ciate with AMSHdN1, AMSHdN2, or AMSHdN3 (Fig. 1D). We

next examined which region within CHMP3 is required for

its association with AMSH using the CHMP3 mutants

depicted in Fig. 1E. As shown in Fig. 1F, pull-down experi-

ments revealed that the association of CHMP3 with AMSH

requires the carboxy-terminal region of CHMP3, including

the third coiled-coil region (CC3). Collectively, these

results indicate that AMSH binds CHMP3 in vivo through

its N-terminal region, and the CC3 region of CHMP3 is

required for the association.

AMSH co-localizes with CHMP3 on endosomes

CHMP3’s known identity as one of the ESCRT-III subunits

(Morita and Sundquist, 2004; Babst, 2005) led us to inves-

tigate whether AMSH co-localizes with CHMP3 on matur-

ing endosomes. To clarify the intracellular localization of

AMSH, we transfected HeLa cells with a CHMP3 expres-

sion vector and analyzed the localization of the exogenous

CHMP3 as well as the endogenous AMSH by confocal

microscopy. In accordance with previous findings (Itoh et

al., 2001; McCullough et al., 2004), we detected AMSH as

scattered puncta throughout the cytoplasm with moderate

expression in the nucleus (Fig. 2A). In addition to the

AMSH co-localization with early endosomes, we found

clear evidence that AMSH was located on the MVB and

late endosomes, as determined by co-staining with LBPA

and LAMP1, respectively. Interestingly, although AMSH

and CHMP3 were co-localized at the LBPA-positive and

the LAMP1-positive vesicles (Fig. 2A, middle and bottom

panels, arrowheads), they showed little co-localization with

early endosomes (Fig. 2A, top panels, arrowheads). These

results suggest that AMSH and CHMP3 co-localize on

maturing endosomes rather than early endosomes, although

AMSH without CHMP3 is present on early endosomes.

We also examined the localization of exogenous, GFP-

tagged AMSH and its mutants, to learn what role CHMP3

plays in the localization of AMSH. As previously reported,

GFP-tagged AMSH was seen co-localized with early endo-

somes (Fig. 2B, arrowheads), but it was associated with late

endosomes to a lesser extent (Fig. 2B, arrows). Next, we

examined whether an AMSH mutant devoid of the CHMP3-

binding capability still localized to endosomes. Unexpect-

edly, AMSHdN3 showed a similar localization pattern to

wild-type AMSH, in that AMSHdN3 was clearly localized to

both the early and late endosomes (Fig. 2B, arrowheads

and arrows, respectively). In addition, the introduction of

AMSHdN3 induced a moderate increase in aberrant AMSH/

LAMP1/EEA1-positive endosomes (Fig. 2B, bottom pan-

els, asterisks). We therefore concluded that the AMSH-

Supplementary Fig. S1. SDS-PAGE analysis of the AMSH interacting

proteins. 3×FLAG tagged AMSH (right lane) or a control vector (left lane)

was transiently transfected in 293T cells. Lysates were obtained from 100

dishes (14.5 cm diameter). One hundred microgram proteins from each

sample were separated by SDS-PAGE. Silver staining was shown. AMSH

and CHMP3 (arrowheads) were identified from a mass spectrometry. Anal-

yses for other proteins are underway.

Roles of AMSH on MVB/Late Endosomes

163

Fig. 1. Identification of CHMP3 as an AMSH-binding protein. (A) AMSH binds CHMP3 in 293T cells. Cells were transfected with pMyc-CHMP3,

p3xFLAG-AMSH. After 48 hours, the cell lysates were used for immunoprecipitation (IP) and immunoblotting (IB) with the indicated antibodies. (B)

AMSH binds endogenous CHMP3. HeLa cells in culture were subjected to chemical crosslinking by 100 µg/ml DSP. The cell lysates were used for

immunoprecipitation by anti-AMSH, control mAbs, anti-CHMP3, and pre-immune antisera, respectively. Immunoblotting was performed with anti-

CHMP3 antisera. Approximately 5% of the prepared lysates was used for the anti-CHMP3 IP and for the total cell lysate (Lysate) lanes. (C) Structures of

wild-type AMSH and its mutants; (dN1, AMSHdN1; dN2, AMSHdN2; dN3, AMSHdN3; dCC, AMSHdCC; dSBM, AMSHdSBM; dBS3, AMSHdBS3). CC, SBM,

and JAMM stand for the coiled-coil region, STAM-binding motif, and JAMM domain, respectively. (D) CHMP3 binds to the N-terminal region of AMSH

in vitro. (Top) Proteins pulled down by GST-CHMP3 were immunoblotted with the anti-FLAG mAb or anti-GST antisera, as indicated. (Middle) For a

control study, proteins pulled down by GST were immunoblotted with the anti-FLAG mAb or anti-GST antisera, as indicated. (Bottom) 2% of the input is

shown. (E) Structures of wild-type CHMP3 and its mutants; (dC1, CHMP3dC1; dC2, CHMP3dC2; dCC1, CHMP3dCC1; dCC2, CHMP3dCC2; dN, CHMP3dN;

dM, CHMP3dM; dCC3, CHMP3dCC3). (F) AMSH binds to the coiled-coil3 region of CHMP3 in vitro. (Top) Proteins pulled down by the GST-CHMP3 were

immunoblotted with the anti-Myc or anti-GST antisera. (Middle) For a control study, proteins pulled down by GST were immunoblotted with the anti-Myc

or anti-GST antisera, respectively. (Bottom) 2% of the input is shown.

164

M. Kyuuma et al.

Fig. 2. AMSH co-localizes with CHMP3 on MVB/late endosomes, but the CHMP3-binding region is not required for this localization. (A) Co-

localization of AMSH and CHMP3 on MVB/late endosomes. HeLa cells were transfected with pDsRed-Monomer-CHMP3. After 48 hours, cells were

double-labeled with the antibodies indicated on the figure. (Top) AMSH (green), DsRed-CHMP3 (red), and EEA1 (blue). (Middle) AMSH (green), DSRed-

CHMP3 (red), and LBPA (blue). (Bottom) AMSH (green), DS-redCHMP3 (red), and LAMP1 (blue). Green labeling is from Alexa 488-conjugated

secondary antibody; blue is from Alexa 633-conjugated secondary antibody. Insets show magnification of boxed areas. Arrowheads indicate colocalization

of AMSH with DsRed-CHMP3 (Top), DsRed-CHMP3 and LBPA (Middle), DsRed-CHMP3 and LAMP1 (Bottom), receptively. (B) Localization of wild-

type AMSH and its dN3 mutant. HeLa cells were transfected with pEGFP-AMSH or pEGFP-AMSHdN3. After 48 hours, the cells were double-labeled with

the indicated antibodies: LAMP1 (red) and EEA1 (blue). Insets show magnification of boxed areas. Arrowheads indicate colocalization of GFP-AMSH or

GFP-AMSHdN3 with EEA1. Arrows indicate colocalization of GFP-AMSH or GFP-AMSHdN3 with LAMP1. Asterisks indicate colocalization of GFP-

AMSHdN3 with EEA1 and LAMP1. Fluorescent labeling is as described for (A), except red is from Alexa 594. Bars indicate 10 µm.

Roles of AMSH on MVB/Late Endosomes

165

CHMP3 complex resides on MVB/late endosomes, but the

localization of AMSH does not significantly correlate with

its binding of CHMP3, even though aberrant endosomes are

induced upon AMSHdN3 expression.

Ubiquitinated cargo clearance requires both the CHMP3-binding capability and the DUB activity of AMSH

We next investigated whether the CHMP3-binding capabili-

ty of AMSH was important for AMSH’s roles in vesicular

traffic. For comparison, we prepared an enzymatically

inactive AMSH, AMSHE280A, basing our construct design

on a report in which replacement of a glutamic acid resi-

due (E20) by alanine within the JAMM motif of the

Archaeoglobus fulgidus AfJAMM protein completely abol-

ished the catalytic activity (Ambroggio et al., 2004). To

verify that the AMSHE280A mutant possessed no DUB activi-

ty, we performed an in vitro deubiquitination assay. As

previously reported, wild-type AMSH showed no DUB

activity for K48-linked ubiquitin chains (Fig. 3A, left pan-

el), while it clearly cleaved ubiquitins from the K63-linked

chains (Ub2–7) (Fig. 3A, middle panel). Recombinant

AMSHE280A produced no monomeric ubiquitins from the

K48- or K63-linked ubiquitin chains. We also examined

AMSH truncation mutants for their DUB activities. Two

N-terminal deletion mutants, AMSHdN2 and AMSHdN3,

showed catalytic activity similar to wild-type AMSH. These

results show that AMSH carries a K63-linked ubiquitin-

specific DUB activity, which is abrogated by the E280A

mutation but is not affected by the deletion of the CHMP3-

binding region.

To examine the effect of the CHMP3-binding capability

and DUB activity of AMSH on ubiquitinated cargos in vivo,

intracellular ubiquitin was immunostained following the

transient introduction of AMSH, AMSHE280A, or AMSHdN3.

For this experiment, we used the FK2 antibody, which

specifically detects ubiquitin moieties on cargo proteins

(Fujimuro et al., 1994). When wild-type AMSH was

expressed, only a marginal amount of ubiquitinated cargo

was observed (Fig. 3B, top panels, arrowhead). Surpris-

ingly, the transfection of AMSHE280A induced heavy staining

of ubiquitinated cargo proteins, which merged with LAMP2

staining (Fig. 3B, middle panels, arrowheads). Deposits of

FK2-positive cargo were also seen in EEA1-positive

puncta (data not shown). Unexpectedly, the introduction

of AMSHdN3 also resulted in the marked deposition of

ubiquitinated cargo on late endosomes, similar to that seen

with AMSHE280A (Fig. 3B, bottom panels, arrowheads).

These results suggest that not only the DUB activity of

AMSH but also its CHMP3-binding capability is required for

the normal clearance of ubiquitinated cargo from endosomes.

AMSH associates with the ESCRT-III complex, and its catalytic activity is not essential for the dissociation of the ESCRT-III complex

Next, we asked whether AMSH is included in the ESCRT-

III complex. Previous reports suggest that ESCRT-III is

further divided into two subcomplexes, and the one that

contains CHMP3 also includes CHMP2A or CHMP2B

(Bowers et al., 2004; Morita and Sundquist, 2004). By tran-

sient expression experiments using 293T cells, we found a

physical interaction between CHMP2A and CHMP3, as

well as CHMP2B and CHMP3 (Fig. 4A, lanes 7 and 8). The

co-introduction AMSH induced the formation of ternary

complexes, including AMSH, CHMP2A, and CHMP3, and

AMSH, CHMP2B, and CHMP3 (Fig. 4A, lanes 5 and 6).

We also observed that CHMP2A but not CHMP2B showed

a weak association with AMSH, even in the absence of

CHMP3 (Fig. 4A, lanes 3 and 4). These results suggest

that AMSH is included in ESCRT-III, at least in a CHMP3-

containing subcomplex.

Next, we set out to clarify whether AMSH is involved

in MVB sorting. A characteristic property of class E VPS

pathway is that they are often relocalized to aberrant endo-

somes induced by expression of catalytically inactive

VPS4 mutant. We co-expressed GFP-AMSH and DsRed-

Vps4E228Q and found that GFP-AMSH accumulated on

aberrant DsRed-Vps4E228Q-induced endosomes (Fig. 4B).

Therefore, it is likely that AMSH is related to some MVB

function.

We then wondered whether AMSH is essential for the

association-disassociation of ESCRT-III, since the ESCRT-

III complex in yeast is transiently formed on vesicular

membranes and then subjected to constant dissociation by

the ATPase activity of VPS4p (Babst et al., 1998). We frac-

tionated 293T cell lysates into their cytoplasmic and mem-

brane components. In mock-transfected 293T cells, native

CHMP3 localized to the cytoplasmic but not the membrane

fraction, which included the LAMP1-positive late endo-

somes (Fig. 4C). Transient introduction of FLAG-tagged

AMSH did not alter the localization of CHMP3. Likewise,

the cytoplasmic localization of CHMP3 was unchanged by

the introduction of AMSHE280A or VPS4. In sharp contrast,

VPS4E228Q, a dominant-negative mutant, induced CHMP3

localization to the membrane fraction. Because the overex-

pression of neither wild-type AMSH nor catalytically inac-

tive AMSH altered the location of ESCRT-III relative to the

membrane fraction, these results suggest that AMSH may

not be essential for the proper functioning of the ESCRT-III

association-dissociation machinery.

AMSH deubiquitinates cargo proteins before their lysosomal degradation and AMSH depletion induces endosomes stuffed with ubiquitinated cargo

To further elucidate the roles played by AMSH at the

166

M. Kyuuma et al.

Fig. 3. In vivo DUB activity of AMSH requires CHMP3-binding capability. (A) CHMP3-binding incapable AMSH mutants retained their deubiquitina-

tion activity for K63-linked ubiquitin chains. K48-linked (top) and K63-linked (middle) polyubiquitin chains (Ub2-7) were treated for 18 hours at 37°C with

the indicated recombinant protein purified from E. coli. Samples were immunoblotted with an anti-ubiquitin mAb, P4D1. Polyubiquitin (Ub3–7), diubiquitin

(Ub2), and monoubiquitin (Ub) are indicated. Recombinants were immunoblotted with the anti-GST antibody (right panel). (Asterisk, GST-AMSH and

GST-AMSHE280A; black arrowhead, GST-AMSHdN3; white arrowhead, AMSHdN2) (B) Accumulation of ubiquitinated protein on late endosomes following

the introduction of the mutant AMSH. HeLa cells were transfected with the pEGFP-AMSH, pEGFP-AMSHE280A, or pEGFP-AMSHdN3. After 48 hours, cells

were double-labeled with the indicated primary antibodies FK2 (red) and LAMP2 (blue) are shown. Insets show magnification of boxed areas. Arrows indi-

cate colocalization of LAMP2 with GFP-AMSH (Top), GFP-AMSHE280A and ubiquitinated proteins (Middle), GFP-AMSHdN3 and ubiquitinated proteins

(Bottom). Bars indicate 10 µm.

Roles of AMSH on MVB/Late Endosomes

167

endosome, we next investigated whether gene silencing of

AMSH by shRNA affected the intracellular status of ubi-

quitinated cargo. First, we showed that two sets of our

shRNA indeed depleted the cells of AMSH (Fig. 5A). Next,

we investigated the ubiquitinated cargo deposition on/

within endosomes by using confocal imaging. In the control

shRNA-transduced cells, although scattered expression of

the endogenous AMSH merged partially with LAMP2, we

detected little if any deposits of ubiquitinated cargo that

were localized to late endosomes (Fig. 5B, upper panels).

On the other hand, in AMSH-depleted cells, a significant

deposit of ubiquitinated cargo co-localized with LAMP2-

positive endosomes (Fig. 5B middle and lower panels). We

observed similar phenotype by using two independent

Fig. 4. AMSH is included in the ESCRT-III complex in vivo and its DUB activity is not essential for the ESCRT-III dissociation. (A) Efficient associa-

tion of AMSH with CHMP2 requires CHMP3. 293T cells were transfected with p3xFLAG-AMSH, pMyc-CHMP3, and pcDNA3.1-V5-CHMP2, or combi-

nations of them. At 48 hours post-transfection, cell lysates were prepared and used for immunoprecipitation (IP) and immunoblotting (IB) with the indicated

antibodies. (B) Co-localization of GFP-AMSH with dominant negative Vps4 in HeLa cells introduced with DsRed-Vps4E228Q. HeLa cells were transfected

with pEGFP-AMSH and DsRed-Vps4E228Q. Bars indicate 10 µm. (C) The catalytic activity of AMSH is not essential for the ESCRT-III dissociation. 293T

cells were transfected with p3xFLAG-AMSH, p3xFLAG-AMSHE280A, p3xFLAG-VPS4, p3xFLAG-VPS4E228Q or a mock plasmid. Forty-eight hours later,

the membrane and cytosolic fractions were prepared and used for immunoblotting with the indicated antibodies. C and M indicate the cytosolic and mem-

brane fraction, respectively. The anti-V5 antibody recognizes the V5 tag on the indicated proteins.

168

M. Kyuuma et al.

Fig. 5. AMSH is essential for the deubiquitination of cargo before its degradation. (A) Depletion of AMSH in HeLa cells. Cells were infected with retro-

virus bearing the AMSH-specific shRNA. Immunoblotting was performed with anti-AMSH (top) or anti-α-tubulin antibodies (bottom). (B) HeLa cells

depleted of AMSH manifested deposits of ubiquitinated cargo on late endosomes. Cells were fixed and triple-labeled with the indicated antibodies: AMSH

(green), FK2 (red), LAMP2 (blue). Insets show magnification of boxed areas. Bars indicate 10 µm. (C) AMSH is essential for the deubiquitination of

EGFR. 293T cells were infected with retrovirus bearing the shRNA for AMSH, and transfected with pcDNA3.1-V5-EGFR, pcDNA3.1-Myc-Ubiquitin, and

HA-c-Cbl, an E3 ligase that ubiquitinates EGFR. At 48 hours post-stimulation, membrane fractions were prepared and used for immunoprecipitation (IP)

and immunoblotting (IB) with the indicated antibodies. (D) Endosomes in AMSH-depleted HeLa cells contain ubiquitinated cargo. HeLa cells depleted of

AMSH by shRNA were observed by immuno-EM with the FK2 mAb. Bars indicate 500 nm.

Roles of AMSH on MVB/Late Endosomes

169

shRNA, suggesting that the observed ubiquitin deposition is

not an off-target effect, and hereafter we used shRNA1 for

further experiments. Notably, under the AMSH-depletion

condition, the number of LAMP2-positive aberrant endo-

somes containing ubiquitinated cargo was noticeably greater

than in the control (Fig. 5B). Collectively, these results sug-

gest that AMSH depletion induces the formation of deposits

of ubiquitinated cargo within the LAMP2-positive aberrant

endosomes.

To elucidate the role of AMSH in cargo deubiquitination,

we next examined whether AMSH affects EGF receptor

(EGFR) ubiquitination within the membrane fraction,

because EGFR is one of the most intensively studied cargo

proteins. To detect EGFR, which is sorted into late endo-

somes, two lysosome inhibitors, E-64-d and Pepstatin A,

were simultaneously added to 293T cells that had been

infected with retrovirus bearing the AMSH-specific or con-

trol shRNA. To clarify the role of AMSH in ubiquitinated

EGFR, we co-expressed c-Cbl in the 293T cells. The level

of the exogenously introduced EGFR in the membrane frac-

tion was monitored before and after stimulation with EGF.

In the absence of the lysosome inhibitors, robust ubiquitina-

tion of EGFR was clearly detected irrespective of AMSH

expression 1 hour after the stimulation (Fig. 5C, top panel).

At the 6-hour time point, however, ubiquitinated EGFR was

observed in the AMSH-depleted lane, but not in the control

lane. Under this condition, EGFR degradation was about the

same regardless of the presence of AMSH (Fig. 5C, lower

panel). In sharp contrast, in the presence of the lysosome

inhibitors, EGFR degradation was inhibited, irrespective

of AMSH, even at the 6-hour time point. Notably, in the

AMSH-depleted samples, EGFR ubiquitination at the 6-

hour time point was as strong as at the 1-hour time point.

These results suggest that AMSH is essential for the cargo

deubiquitination at some time well after EGF stimulation,

but may not be required for EGFR degradation.

To investigate the ubiquitinated cargo sorting, we trans-

duced HeLa cells with the AMSH-shRNA. Using samples

prepared simultaneously, we performed immuno-electro-

microscopy (EM) analysis using the FK2 antibody. In the

AMSH-depleted HeLa cells, MVB containing ubiquitinated

cargo were observed, while little or no FK2-positve cargo

was found in the control cells (Fig. 5D). We concluded that,

in the absence of AMSH, ubiquitinated cargo is internalized

by the MVB, by invagination, and sorted into the intralumi-

nal vesicles within the MVB without being deubiquitinated.

Collectively, these results suggest that ubiquitinated cargo,

including EGFR, is deubiquitinated by AMSH before the

cargo’s lysosome-dependent degradation.

Discussion

In this study, we provide evidence that AMSH binds

ESCRT-III. As was expected from its high homology to

yeast Vps24p, CHMP3 is now known to be a mammalian

ESCRT-III subunit (Howard et al., 2001). Our present

conclusion that AMSH binds CHMP3 is consistent with

reports by others (Agromayor and Martin-Serrano, 2006;

McCullough et al., 2006), and similar binding has been

identified between Drosophila orthologues of these mole-

cules (Giot et al., 2003). Although genetic evidence has

already proven that ESCRT-III functions downstream of

ESCRT-II in sorting and concentrating the ubiquitinated

cargo of the MVB/late endosome (Bowers et al., 2004), the

involvement of a DUB in mammalian MVB sorting had not

been clarified previously. Therefore, our present data, along

with recent report from others, provide important insights

that support the hypothesis that mammalian protein sorting

relies on DUB activity.

The association of AMSH with CHMP3 requires

AMSH’s amino-terminal region, spanning residues 1–82

(AMSH1–82, Fig. 1D). Because earlier reports indicated that

AMSH1–103, which includes the MIT-like domain (AMSH34–97),

is required for CHMP3 binding (McCullough et al., 2006;

Tsang et al., 2006), our present study has further refined the

responsible region. However, we found that the localization

of AMSH to endosomes does not depend on the CHMP3

association (Fig. 2B). Nonetheless, clathrin heavy chain

(CHC) is also known to bind AMSH (McCullough et al.,

2006; Nakamura et al., 2006), and therefore may play some

role in its localization to late endosomes, as well as its

reported localization to early endosomes. Furthermore,

because we repeatedly observed an increased recruitment of

AMSH to endosomes in cells expressing CHMP3 (data not

shown), we speculate that, although CHMP3 is not required,

it may have at least an auxiliary function to support the

recruitment of AMSH to endosomes (Fig. 2B). Our findings

are consistent with a recent report demonstrating the effi-

cient recruitment of AMSH to CHMP1/3-positive endo-

somes by the forced expression of an ESCRT-III subunit,

CHMP1/3 (Agromayor and Martin-Serrano, 2006; Tsang et

al., 2006). It is possible that a combination of multiple asso-

ciated proteins, including CHC and the CHMPs, may deter-

mine the recruitment of AMSH onto endosomes. In the

present study, we also found that AMSH is involved in the

CHMP3-CHMP2 complex, which may act as a subunit of

the ESCRT-III complex (Fig. 4A). Since AMSH binds

CHMP2 in the absence of exogenous CHMP3, AMSH

may bind CHMP2 with less affinity, although we cannot

fully rule out the possibility that endogenous CHMP3 medi-

ates indirect binding between AMSH and CHMP2. In line

with the former possibility, a recent report demonstrated

direct binding between CHMP2A and AMSH (Agromayor

and Martin-Serrano, 2006). In any case, our present data

indicate that CHMP3 enhances the association between

CHMP2 and AMSH.

We previously documented that AMSH binds STAM.

Therefore, it is interesting to speculate whether the ternary

complex, STAM-AMSH-CHMP3, exists in vivo. Indeed,

170

M. Kyuuma et al.

by co-immunoprecipitation experiments, we observed this

three-molecule complex (data not shown). However, under

confocal imaging, we repeatedly detected more frequent

colocalization of AMSH-STAM to early endosomes (data

not shown), whereas the AMSH-CHMP3 complex showed

localization to MVB/late endosomes (Fig. 2A). In addition,

we found that colocalizations of AMSH-STAM and

AMSH-CHMP3 exist in a mutually independent manner

(Supplementary Fig. S2). Although associations of AMSH

with CHMP1, 2, and 3 have been reported, the localization

of these complexes was unknown. Therefore, our present

data is the first to identify the late endosomal localization of

AMSH, supporting our hypothesis that AMSH functions

within the ESCRT-III complex. Further experiments should

reveal whether a multi-ESCRT complex including the

ESCRT-0 (STAM/Hrs), -I, -II, and -III exists. In any case, a

requirement for a binding partner for AMSH to carry out its

in vivo functions may explain the diverse roles of AMSH, as

well as its localization to early endosomes and the nucleus.

We found that AMSHdN3 expression in HeLa cells resulted

in a remarkable deposition of ubiquitinated proteins that

colocalized with aberrant LAMP1/EEA1-double positive

endosomes, despite AMSHdN3’s retaining its in vitro DUB

activity (Fig. 2 and Fig. 3). In AMSHdN3-introduced cells,

approximately 95% of EEA1-positive vesicles and 80% of

the LAMP2-positive vesicles contained ubiquitinated cargo.

We speculate that a certain percentage of the ubiquitinated

cargo accumulated in the EEA1/LAMP2-double positive

endosomes. The less frequent accumulation within the late

endosome may be an outcome of a possible degradation of

ubiquitinated cargo within the vesicle. Ubiquitin deposition

in the presence of AMSHdN3 was similar to that observed

when AMSHE280A was expressed, or the endogenous AMSH

was depleted (Fig. 3 and Fig. 5). Therefore, although we

cannot fully exclude the possibility that ESCRT-III proteins

other than CHMP3 may be involved in the ubiquitin deposi-

tion resulting from AMSHdN3 expression, we speculate that

substrate recognition by AMSH requires CHMP3 binding,

at least on late endosomes. In any case, the formation of a

complex between CHMP3 and AMSH seems to be essential

for the deubiquitination of endosome cargo proteins. In sup-

port of this assumption, other JAMM domain-containing

isopeptidases, such as Rpn11 and CSN5, also function

within complexes (Cope et al., 2002; Verma et al., 2002;

Yao and Cohen, 2002). Indeed, there may be a difference

among DUBs in terms of their mode of action: one group

may function by direct recognition of their substrates, while

the other group, including the JAMM domain-containing

isopeptidases, may function as a constituent of a complex

(Amerik and Hochstrasser, 2004). The unique structures of

the catalytic regions of the DUBs, as well as the presence of

specific sets of interaction domains in the different DUBs

may explain such differences (Ambroggio et al., 2004).

Interestingly, unlike the ESCRT-0, -I, and -II components,

ESCRT-III constituents seem to lack previously identified

ubiquitin-binding motifs. Including AMSH in the complex

could result in efficient cargo deubiquitination, since

ESCRT-III has access to the ubiquitin moieties on the cargo

proteins. Interestingly, a recent report demonstrated that the

in vitro DUB activity of AMSH is augmented when AMSH

is bound to STAM (McCullough et al., 2006). However,

binding CHMP3 did not strengthen AMSH’s DUB activity

(McCullough et al., 2006). In addition to cargo proteins, it

is also possible that ESCRT proteins themselves could be

included in the vesicle as ubiquitinated proteins, because

ubiquitination of Hrs is reported (Hoeller et al., 2006). It

will be intriguing to determine the content of vesicle to clar-

ify the property of the ubiquitinated proteins.

McCullough et al. (2004) previously documented that

AMSH deubiquitinates EGFR in vitro but is not essential

for its sorting/degradation in vivo. In this study, we found

that AMSH is responsible for the EGFR deubiquitination in

vivo at time points well after (6 hr) EGF stimulation (Fig.

5C). In our EM analysis, an accumulation of ubiquitinated

cargo was identified within MVB in AMSH-depleted cells

(Fig. 5D). This is consistent with previous findings in yeast

Supplementary Fig. S2. Co-localizations of AMSH-STAM and AMSH-CHMP3 exist in a mutually independent manner. HeLa cells were transfected

with pDsRed-Monomer-CHMP3 and STAM1-V5. After 48 hours, cells were double-labeled with the antibodies indicated on the figure. AMSH (green),

DsRed-CHMP3 (red), and STAM1-V5 (blue). Arrowheads indicate localization of AMSH in DsRed-CHMP3-positive aberrant endosomes. Arrows indicate

localization of AMSH in STAM1-V5-positive aberrant endosomes. Bars indicate 10 µm.

Roles of AMSH on MVB/Late Endosomes

171

that deubiquitination by Doa4 is not essential for cargo sort-

ing into vacuoles (Amerik et al., 2000). Instead, in Doa4

mutants, intracellular free ubiquitins are depleted by the

insufficiency of Doa4-function, which cleaves ubiquitin

chains to produce free ubiquitins (Swaminathan et al.,

1999). In our hands, however, we detected little if any

decrease in the intracellular free ubiquitin regardless of

AMSH expression (data not shown). One possible expla-

nation is that there may be a redundant DUB activity in

mammals. Recently, a DUB called UBPY was suspected as

a possible Doa4 orthologue in mammals, but the same

report showed that the amount of free ubiquitin in UBPY-

depleted cells was unaltered (Mizuno et al., 2005). In

addition, there is no evidence so far supporting UBPY’s

involvement in ESCRT-III. Therefore, we assume that an as

yet unidentified ESCRT-III-related DUB(s) is essential for

the ubiquitin turnover. One such candidate is AMSH-LP,

an AMSH-related DUB (Kikuchi et al., 2003). Although

AMSH-LP is located on early endosomes and catalytically

inactive AMSH-LP induces ubiquitinated cargo accumu-

lation (Nakamura et al., 2006), AMSH-LP seems not to

be included in the ESCRT-III complex (Agromayor and

Martin-Serrano, 2006), our unpublished data). Although

further experiments are needed to identify DUBs that

strictly meet the criteria for being a yeast DUB orthologue,

we tentatively speculate that AMSH functionally corre-

sponds to Doa4.

In conclusion, we demonstrated that not only the DUB

activity of AMSH but also its CHMP3-binding capability is

required for the in vivo function of AMSH. AMSH deple-

tion results in the formation of aberrant endosomes in cells.

Further in vivo analyses of DUBs should continue to pro-

vide new insights into their specific roles in mammals.

Acknowledgements. This work was supported in part by a Grant-in-Aid

for Scientific Research from the Japan Society for the Promotion of Sci-

ence (JSPS), a Grant-in-Aid for Scientific Research in Priority Areas and a

Grant-in-Aid for a Center of Excellence (COE) program from the Ministry

of Education, Science, Sports and Culture of the Japanese Government,

and a Grant-in-Aid from the Takeda Medical Research Foundation. M.K. is

supported by a Grant-in-Aid for JSPS Fellows.

References

Agromayor, M. and Martin-Serrano, J. 2006. Interaction of AMSH with

ESCRT-III and deubiquitination of endosomal cargo. J. Biol. Chem.,

281: 23083–23091.

Ambroggio, X.I., Rees, D.C., and Deshaies, R.J. 2004. JAMM: a metallo-

protease-like zinc site in the proteasome and signalosome. PLoS Biol., 2:

E2.

Amerik, A.Y. and Hochstrasser, M. 2004. Mechanism and function of

deubiquitinating enzymes. Biochim. Biophys. Acta, 1695: 189–207.

Amerik, A.Y., Nowak, J., Swaminathan, S., and Hochstrasser, M. 2000.

The Doa4 deubiquitinating enzyme is functionally linked to the vacuolar

protein-sorting and endocytic pathways. Mol. Biol. Cell, 11: 3365–3380.

Asao, H., Sasaki, Y., Arita, T., Tanaka, N., Endo, K., Kasai, H., Takeshita,

T., Endo, Y., Fujita, T., and Sugamura, K. 1997. Hrs is associated with

STAM, a signal-transducing adaptor molecule. Its suppressive effect on

cytokine-induced cell growth. J. Biol. Chem., 272: 32785–32791.

Babst, M. 2005. A protein's final ESCRT. Traffic, 6: 2–9.

Babst, M., Katzmann, D.J., Estepa-Sabal, E.J., Meerloo, T., and Emr, S.D.

2002. Escrt-III: an endosome-associated heterooligomeric protein com-

plex required for mvb sorting. Dev. Cell, 3: 271–282.

Babst, M., Sato, T.K., Banta, L.M., and Emr, S.D. 1997. Endosomal trans-

port function in yeast requires a novel AAA-type ATPase, Vps4p.

EMBO J., 16: 1820–1831.

Babst, M., Wendland, B., Estepa, E.J., and Emr, S.D. 1998. The Vps4p

AAA ATPase regulates membrane association of a Vps protein complex

required for normal endosome function. EMBO J., 17: 2982–2993.

Bonifacino, J.S. and Traub, L.M. 2003. Signals for sorting of transmem-

brane proteins to endosomes and lysosomes. Annu. Rev. Biochem., 72:

395–447.

Bowers, K., Lottridge, J., Helliwell, S.B., Goldthwaite, L.M., Luzio, J.P.,

and Stevens, T.H. 2004. Protein-protein interactions of ESCRT com-

plexes in the yeast Saccharomyces cerevisiae. Traffic, 5: 194–210.

Cope, G.A., Suh, G.S., Aravind, L., Schwarz, S.E., Zipursky, S.L., Koonin,

E.V., and Deshaies, R.J. 2002. Role of predicted metalloprotease motif

of Jab1/Csn5 in cleavage of Nedd8 from Cul1. Science, 298: 608–611.

Fujimuro, M., Sawada, H., and Yokosawa, H. 1994. Production and char-

acterization of monoclonal antibodies specific to multi-ubiquitin chains

of polyubiquitinated proteins. FEBS Lett., 349: 173–180.

Giot, L., Bader, J.S., Brouwer, C. et al. 2003. A protein interaction map of

Drosophila melanogaster. Science, 302: 1727–1736.

Glickman, M.H. and Ciechanover, A. 2002. The ubiquitin-proteasome

proteolytic pathway: destruction for the sake of construction. Physiol.

Rev., 82: 373–428.

Gruenberg, J. and Stenmark, H. 2004. The biogenesis of multivesicular

endosomes. Nat. Rev. Mol. Cell Biol., 5: 317–323.

Hicke, L. 2001. A new ticket for entry into budding vesicles-ubiquitin.

Cell, 106: 527–530.

Hoeller, D., Crosetto, N., Blagoev, B., Raiborg, C., Tikkanen, R., Wagner,

S., Kowanetz, K., Breitling, R., Mann, M., Stenmark, H., and Dikic, I.

2006. Regulation of ubiquitin-binding proteins by monoubiquitination.

Nat. Cell Biol., 8: 163–169.

Howard, T.L., Stauffer, D.R., Degnin, C.R., and Hollenberg, S.M. 2001.

CHMP1 functions as a member of a newly defined family of vesicle

trafficking proteins. J. Cell Sci., 114: 2395–2404.

Itoh, F., Asao, H., Sugamura, K., Heldin, C.H., ten Dijke, P., and Itoh, S.

2001. Promoting bone morphogenetic protein signaling through nega-

tive regulation of inhibitory Smads. EMBO J., 20: 4132–4142.

Kanazawa, C., Morita, E., Yamada, M., Ishii, N., Miura, S., Asao, H.,

Yoshimori, T., and Sugamura, K. 2003. Effects of deficiencies of

STAMs and Hrs, mammalian class E Vps proteins, on receptor down-

regulation. Biochem. Biophys. Res. Commun., 309: 848–856.

Katzmann, D.J., Babst, M., and Emr, S.D. 2001. Ubiquitin-dependent

sorting into the multivesicular body pathway requires the function of a

conserved endosomal protein sorting complex, ESCRT-I. Cell, 106:

145–155.

Katzmann, D.J., Odorizzi, G., and Emr, S.D. 2002. Receptor downregula-

tion and multivesicular-body sorting. Nat. Rev. Mol. Cell Biol., 3: 893–

905.

Kikuchi, K., Ishii, N., Asao, H., and Sugamura, K. 2003. Identification of

AMSH-LP containing a Jab1/MPN domain metalloenzyme motif.

Biochem. Biophys. Res. Commun., 306: 637–643.

Kobayashi, H., Tanaka, N., Asao, H., Miura, S., Kyuuma, M., Semura, K.,

Ishii, N., and Sugamura, K. 2005. Hrs, a mammalian master molecule in

vesicular transport and protein sorting, suppresses the degradation of

ESCRT proteins signal transducing adaptor molecule 1 and 2. J. Biol.

Chem., 280: 10468–10477.

Kobayashi, T., Stang, E., Fang, K.S., de Moerloose, P., Parton, R.G.,

and Gruenberg, J. 1998. A lipid associated with the antiphospholipid

172

M. Kyuuma et al.

syndrome regulates endosome structure and function. Nature, 392: 193–

197.

Luhtala, N. and Odorizzi, G. 2004. Bro1 coordinates deubiquitination in

the multivesicular body pathway by recruiting Doa4 to endosomes. J.

Cell Biol., 166: 717–729.

McCullough, J., Clague, M.J., and Urbe, S. 2004. AMSH is an endosome-

associated ubiquitin isopeptidase. J. Cell Biol., 166: 487–492.

McCullough, J., Row, P.E., Lorenzo, O., Doherty, M., Beynon, R., Clague,

M.J., and Urbe, S. 2006. Activation of the endosome-associated

ubiquitin isopeptidase AMSH by STAM, a component of the multi-

vesicular body-sorting machinery. Curr. Biol., 16: 160–165.

Mizuno, E., Iura, T., Mukai, A., Yoshimori, T., Kitamura, N., and Komada,

M. 2005. Regulation of epidermal growth factor receptor down-regula-

tion by UBPY-mediated deubiquitination at endosomes. Mol. Biol. Cell,

16: 5163–5174.

Mizuno, E., Kawahata, K., Kato, M., Kitamura, N., and Komada, M. 2003.

STAM proteins bind ubiquitinated proteins on the early endosome via

the VHS domain and ubiquitin-interacting motif. Mol. Biol. Cell, 14:

3675–3689.

Morita, E. and Sundquist, W.I. 2004. Retrovirus budding. Annu. Rev. Cell

Dev. Biol., 20: 395–425.

Nakamura, M., Tanaka, N., Kitamura, N., and Komada, M. 2006. Clathrin

anchors deubiquitinating enzymes, AMSH and AMSH-like protein, on

early endosomes. Genes Cells, 11: 593–606.

Raiborg, C., Rusten, T.E., and Stenmark, H. 2003. Protein sorting into

multivesicular endosomes. Curr. Opin. Cell Biol., 15: 446–455.

Swaminathan, S., Amerik, A.Y., and Hochstrasser, M. 1999. The Doa4

deubiquitinating enzyme is required for ubiquitin homeostasis in yeast.

Mol. Biol. Cell, 10: 2583–2594.

Tanaka, N., Kaneko, K., Asao, H., Kasai, H., Endo, Y., Fujita, T., Takeshita,

T., and Sugamura, K. 1999. Possible involvement of a novel STAM-

associated molecule “AMSH” in intracellular signal transduction

mediated by cytokines. J. Biol. Chem., 274: 19129–19135.

Tsang, H.T., Connell, J.W., Brown, S.E., Thompson, A., Reid, E., and

Sanderson, C.M. 2006. A systematic analysis of human CHMP protein

interactions: additional MIT domain-containing proteins bind to

multiple components of the human ESCRT III complex. Genomics, 88:

333–346.

Verma, R., Aravind, L., Oania, R., McDonald, W.H., Yates, J.R., 3rd,

Koonin, E.V., and Deshaies, R.J. 2002. Role of Rpn11 metalloprotease

in deubiquitination and degradation by the 26S proteasome. Science,

298: 611–615.

Weissman, A.M. 2001. Themes and variations on ubiquitylation. Nat. Rev.

Mol. Cell Biol., 2: 169–178.

Yao, T. and Cohen, R.E. 2002. A cryptic protease couples deubiquitina-

tion and degradation by the proteasome. Nature, 419: 403–407.

Yoshimori, T., Yamagata, F., Yamamoto, A., Mizushima, N., Kabeya, Y.,

Nara, A., Miwako, I., Ohashi, M., Ohsumi, M., and 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.

(Received for publication, August 23, 2006

and accepted, November 20, 2006)