To confirm the interaction between SNX9 and Itch, we repeated the pulldown experiment using GST–Itch–PWW and extracts of HEK293 cells expressing vesicular stomatitis virus-G epitope VSV-t
Trang 1through an unconventional substrate recognition domain Claudia Baumann, Cecilia K Lindholm*, Donata Rimoldi and Fre´de´ric Le´vy
Ludwig Institute for Cancer Research Ltd, University of Lausanne, Switzerland
Introduction
Ubiquitin (Ub) ligases play a crucial role in regulating
intracellular protein levels in eukaryotes The reaction
catalysed by Ub ligase activity, ubiquitylation, results
in a reversible covalent modification of substrate proteins and constitutes a signal for proteasomal and lysosomal degradation This is an important
mecha-Keywords
E3 ligase; Itch; protein–protein interaction;
sorting nexin 9; ubiquitin
Correspondence
C Baumann, Ludwig Institute for Cancer
Research Ltd, Chemin des Boveresses 155,
CH-1066 Epalinges, Switzerland
Fax: +41 21 692 59 95
Tel: +41 21 692 59 77
E-mail: Claudia.Baumann@licr.unil.ch
Present address
*Philip Morris International SA, Neuchaˆtel,
Switzerland
Debiopharm SA, Lausanne, Switzerland
(Received 9 January 2010, revised 20 April
2010, accepted 27 April 2010)
doi:10.1111/j.1742-4658.2010.07698.x
The level of intracellular proteins is mainly regulated through modifications
by ubiquitin ligases that target them for degradation Members of the NEDD4 family of E3 ubiquitin ligases, such as Itch (atrophin-1 interacting protein 4), possess up to four WW domains for specific association with
PY motif-containing substrates We have identified sorting nexin 9 (SNX9),
a protein involved in endocytic processes, as a new substrate of Itch Itch ubiquitylates SNX9 and regulates intracellular SNX9 levels Using trun-cated proteins, we found that the interaction with SNX9 is mediated by the proline-rich domain (PRD) of Itch, a domain distinct from the conven-tional WW recognition domain, and the SH3 domain of SNX9 Interaction with the PRD of Itch is essential for SNX9 ubiquitylation and degradation Furthermore, this effect is specific for Itch, as NEDD4, a related PRD-containing E3 ligase, does not bind SNX9 SNX18, a second member of the SNX family containing an SH3 domain, was also found to bind to Itch Our results indicate that the pool of substrates of NEDD4 family E3 ubiquitin ligases extends beyond proteins containing PY motifs
Structured digital abstract
l MINT-7889719 : SNX18 (uniprotkb: Q96RF0 ) physically interacts ( MI:0915 ) with ITCH (uni-protkb: Q96J02 ) by anti tag coimmunoprecipitation ( MI:0007 )
l MINT-7889571 , MINT-7889619 : ITCH (uniprotkb: Q96J02 ) physically interacts ( MI:0915 ) with SNX9 (uniprotkb: Q9Y5X1 ) by pull down ( MI:0096 )
l MINT-7889653 : Melan-A (uniprotkb: Q16655 ) physically interacts ( MI:0914 ) with NEDD4 (uniprotkb: P46934 ) and ITCH (uniprotkb: Q96J02 ) by pull down ( MI:0096 )
l MINT-7889591 , MINT-7889673 , MINT-7890033 : SNX9 (uniprotkb: Q9Y5X1 ) physically interacts ( MI:0915 ) with ITCH (uniprotkb: Q96J02 ) by anti tag coimmunoprecipitation ( MI:
0007 )
l MINT-7889689 : SNX9 (uniprotkb: Q9Y5X1 ) physically interacts ( MI:0914 ) with ITCH (uniprotkb:
Q96J02 ) and Ubiquitin (uniprotkb: P62988 ) by anti tag coimmunoprecipitation ( MI:0007 )
l MINT-7889928 : Ub (uniprotkb: P62988 ) physically interacts ( MI:0915 ) with SNX9 (uniprotkb:
Q9Y5X1 ) by anti tag coimmunoprecipitation ( MI:0007 )
l MINT-7889610 : SNX9 (uniprotkb: Q9Y5X1 ) physically interacts ( MI:0915 ) with ITCH (uni-protkb: Q96J02 ) by anti bait coimmunoprecipitation ( MI:0006 )
Abbreviations
GST, glutathione S-transferase; HA, hemagglutinin; PRD, proline-rich domain; SH3, Src homology 3 domain; SNX, sorting nexin;
Ub, ubiquitin.
Trang 2nism to terminate protein activity and thereby control
numerous cellular processes The specific selection of
substrates targeted for ubiquitylation is determined by
E3 Ub ligases (EC 6.3.2.19)
Itch, or atrophin-1 interacting protein 4 (AIP4,
here-after referred to as Itch) (UniProtKB accession
num-ber Q96J02), is a memnum-ber of the NEDD4 family of E3
Ub ligases, and Itch knockout mice display severe
immunological disorders All NEDD4 family members
have a C-terminal HECT (homologous to E6-AP
C-terminus) domain catalysing the direct transfer of
Ub onto substrates In addition, they have an N-terminal
C2 domain for phospholipid⁄ membrane association
and multiple WW domains for protein–protein
interac-tions The latter are members of the group I WW
domains with specificity for ligands containing a
Pro-Pro-X-Tyr consensus sequence, the so-called PY motif
[1] Many substrate proteins of Itch have been
identified (e.g c-Jun, JunB, ErbB4, CXC-chemokine
receptor 4) [2] We have previously shown that, in
melanoma cells, Itch is involved in lysosomal
degrada-tion of the melanosomal protein Melan-A (UniProtKB
accession number Q16655) [3] Substrates of Itch
generally contain PY motifs that associate with one of
the four WW domains of the ligase, and ubiquitylation
usually targets them for degradation Unlike other
NEDD4 family members, Itch and NEDD4
(Uni-ProtKB accession number P46934) also possess a short
proline-rich domain (PRD)
Sorting nexin 9 (SNX9) (UniProtKB accession
num-ber Q9Y5X1) belongs to a large family of proteins
involved in endocytosis and intracellular trafficking
[4,5] Sorting nexins are characterized by the presence
of a phosphoinositide-binding PX domain that mediates
interactions with cellular membranes SNX9, originally
identified through its association with
metalloproteas-es [6], has since been shown to participate in
clathrin-mediated endocytosis of cell-surface receptors such as
the transferrin receptor [7] In addition to the PX
mem-brane binding module, SNX9 has a C-terminal Bin⁄
amphiphysin⁄ Rvs (BAR) domain and an N-terminal
Src homology 3 (SH3) domain The BAR domain
facili-tates homodimerization of SNX9, sensing of membrane
curvature and tubulation of membranes [8] The SH3
domain mediates protein–protein interactions by
bind-ing to proline-rich regions present in interactbind-ing
proteins Many interaction partners of the SH3 domain
of SNX9 have been described [7] It has also been
shown that SNX9 undergoes tyrosine phosphorylation,
and that this modification can modulate binding of
SNX9 to other proteins [9–11] Due to its important
role in endocytosis, intracellular SNX9 levels are
expected to be tightly regulated
Here we identify SNX9 as a new substrate of the E3
Ub ligase Itch We show that Itch ubiquitylates and regulates the level of SNX9 Unlike most substrates studied previously, we found that binding of SNX9 to Itch occurs through the SH3 domain of SNX9 and the PRD of Itch We demonstrate that the PRD of Itch is essential for SNX9 binding, ubiquitylation and degra-dation Taken together, these results extend the pool
of substrates of NEDD4 family E3 ligases to proteins containing SH3 domains
Results Itch interacts with sorting nexin 9
We previously showed that, in melanoma cells, the Ub ligase Itch is involved in lysosomal degradation of the melanosomal protein Melan-A [3] To identify addi-tional substrates of Itch in melanoma cells, we per-formed pulldown experiments by incubating an immobilized GST–Itch fusion protein with cell extracts from pigmented SK-MEL-23 melanoma cells The recovered associated proteins were identified by mass spectrometry The Itch sequence used in this experiment (GST–Itch–PWW) contained the proline-rich domain and four WW protein–protein interaction domains, but lacked the phospholipid-interacting C2 domain and the catalytic HECT domain (Fig 1) SNX9 was identified among the proteins that were found to associate with GST–Itch–PWW but not with GST alone SNX9 is a
Fig 1 Schematic representations of the constructs used to identify proteins associated with Itch and to analyse interactions between Itch and SNX9 The various domains and their localization within the proteins are indicated BAR, Bin ⁄ amphiphysin ⁄ Rvs domain; C2,
Ca 2+ -dependent phospholipid binding domain; GST, glutathione S-transferase; HECT, homologous to E6-AP C-terminus domain; myc, myc tag; PRD, proline-rich domain; PX, phosphoinositide bind-ing domain; SH3, Src homology 3 domain; WW, WW domain.
Trang 3cytosolic protein that contains an SH3 domain, a PX
domain and a homodimerization BAR domain In mass
spectrometric analysis, seven unique peptides matching
the SNX9 protein sequence were detected, covering
about 17% of the entire SNX9 sequence To confirm
the interaction between SNX9 and Itch, we repeated
the pulldown experiment using GST–Itch–PWW and
extracts of HEK293 cells expressing vesicular stomatitis
virus-G epitope (VSV)-tagged SNX9 As shown in Fig
2A, a band of approximately 75 kDa reacting with
anti-body against the VSV tag was detectable in the lane
containing material immunoprecipitated with GST–Itch–
PWW In parallel, an aliquot of the same cell extract was
analysed directly with the antibody against the VSV tag
Having identified SNX9 in GST–Itch–PWW
pull-down experiments, we confirmed the interaction in
cells We first analysed this interaction in cells
exoge-nously expressing the proteins HEK293 cells were
chosen for this and other over-expression experiments
as SK-MEL-23 melanoma cells have a very low
trans-fection efficiency In HEK293 cells co-transfected with
VSV-tagged SNX9 and HA-tagged Itch, Itch was
found to be associated with SNX9 (Fig 2B) The
inter-action of Itch with SNX9 was further validated in
SK-MEL-23 melanoma cells by
co-immunoprecipita-tion of endogenous Itch and SNX9 proteins (Fig 2C)
Itch ubiquitylates and regulates the level of SNX9
Next we tested whether SNX9 was ubiquitylated by
Itch Plasmids encoding SNX9 and HA-tagged
ubiqu-itin (UbHA) were transiently transfected into HEK293 cells The total pool of ubiquitylated proteins was immunoprecipitated with an antibody against the HA tag and analysed by Western blot with antibody against SNX9 In cells expressing SNX9 and UbHA, several bands corresponding to ubiquitylated SNX9 were detected (Fig 3A) We confirmed that Itch directly ubiquitylated SNX9 by performing an in vitro ubiquitylation assay Immobilized GST–SNX9 fusion protein was incubated with biotinylated Ub and a pro-teasome-depleted lysate of HEK293 cells transfected either with a vector encoding full-length Itch or with
an empty vector GST–SNX9 was eluted and re-immu-noprecipitated with antibody against SNX9 Ubiquity-lated GST–SNX9 was detected by immunoblotting with horseradish peroxidase-conjugated streptavidin
In the presence of over-expressed Itch, a smear of high-molecular-mass proteins was observed, indicating ubiquitylated forms of SNX9 (Fig 3B) Prolonged exposure of the membrane revealed a similar smear in the sample containing a lysate of control-transfected cells, probably due to endogenous Itch protein (data not shown) No evidence of ubiquitylation was found with GST alone Note that analysis of SNX9 ubiquity-lation in cells showed a few discrete bands (Fig 3A), while a smear of ubiquitylated SNX9 was detected in the in vitro assay (Fig 3B) This may be due to the fact that, in the latter case, a vast excess of ubiquitin was present and proteasomes were absent The absence
of highly ubiquitylated SNX9 in cells may also be due
to factors regulating SNX9 ubiquitylation in a cellular
A
C
B
Fig 2 Itch interacts with sorting nexin 9 (SNX9) (A) Extracts from HEK293 cells transfected with VSV-tagged SNX9 were incubated with immobilized GST or GST–Itch–PWW Bound material was resolved by SDS ⁄ PAGE and analysed by immunoblotting using antibodies against the VSV tag Cell extracts were analysed in parallel and correspond to 10% of the eluted material (B) Digitonin-soluble extracts were prepared from HEK293 cells transfected with HA-tagged Itch, VSV-tagged SNX9 or both Extracts were immunoprecipitated (IP) using anti-bodies against the VSV tag, resolved by SDS ⁄ PAGE and subjected to immunoblotting using antibodies against the HA tag or against the VSV tag as indicated Cell extracts were analysed in parallel (C) Digitonin-soluble extracts from SK-MEL-23 melanoma cells were immuno-precipitated using polyclonal antibodies against SNX9 or control polyclonal antibodies against CD2AP (Ctrl Ab) Proteins were analysed by immunoblotting with monoclonal antibody against Itch or polyclonal antibodies against SNX9 Cell extracts were analysed in parallel.
Trang 4context, such as co-factors and deubiquitylating
enzymes
To investigate whether ubiquitylation of SNX9 led
to its degradation, we analysed the impact of
modulat-ing Itch levels on intracellular levels of SNX9 We
transduced pigmented SK-MEL-23 melanoma cells
with a recombinant lentivector delivering a short
hair-pin RNA sequence that efficiently silenced Itch
expres-sion Itch silencing caused an increase in the level of
endogenous SNX9 protein (Fig 3C) Quantitative
analysis indicated that the level of SNX9 increased
approximately twofold (2.1 ± 0.4, mean ± standard
deviation from three independent experiments)
com-pared to cells transduced with control lentivector
Consistent with this observation, over-expression of
Itch in HEK293 cells decreased the level of SNX9 (Fig 3D)
To demonstrate that Itch regulates the degradation
of SNX9, we performed pulse–chase experiments using HEK293 cells over-expressing VSV-tagged SNX9 alone
or in combination with Itch We analysed SNX9 degradation for 24 h In the absence of exogenous Itch, SNX9 appeared to be a very stable protein, with
a life of approximately 24 h (Fig 3E) The half-life of SNX9 was reduced in the presence of over-expressed Itch, such that the level of labelled SNX9 protein has decreased to approximately 20% of its initial amount after 24 h (Fig 3E)
Altogether, these results show that the Ub ligase Itch binds to and ubiquitylates SNX9, directly affecting
E
Fig 3 Itch ubiquitylates SNX9 and targets it for degradation (A) HEK293 cells transfected with plasmids encoding HA-tagged ubiquitin (Ub HA ), SNX9 or both were treated with lactacystin before lysis Proteins immunoprecipitated (IP) with monoclonal antibody against the HA tag and cell extracts were analysed by immunoblotting using polyclonal antibodies against SNX9 (B) In vitro ubiquitylation assay GST–SNX9
or GST immobilized on glutathione beads were incubated with cytosolic extracts from HEK293 cells, supplemented with 2 lg biotinylated ubiquitin (Ub-biotin) and ATP HEK293 cells were transfected with Itch (+) or empty vector ( )), as indicated After the indicated time, the reaction was stopped by boiling the beads for 5 min in 2% SDS SNX9 was immunoprecipitated and analysed by immunoblotting using horseradish peroxidase-conjugated streptavidin to reveal SNX9-bound Ub-biotin (C) Extracts from SK-MEL-23 cells transduced with lentivec-tors encoding short hairpin RNA specific for Itch (rec lv ⁄ siItch) or an irrelevant target gene (rec lv ⁄ siControl) were resolved by SDS ⁄ PAGE and analysed by immunoblotting with the indicated antibodies (D) Extracts from HEK293 cells expressing SNX9 alone or together with Itch were analysed by immunoblotting with antibodies against SNX9 and Itch Note that, in (C) and (D), the level of SNX9 is inversely proportional
to the level of Itch (E) HEK293 cells transfected with VSV-tagged SNX9 or a combination of VSV-tagged SNX9 and myc-tagged Itch were incubated for 30 min in medium containing 35 S-methionine ⁄ cysteine to label newly synthesized proteins (pulse) After labelling, cells were incubated in regular medium for the indicated times (chase) Extracts were prepared and immunoprecipitated using antibody against the VSV tag Eluted material was resolved by SDS ⁄ PAGE The left panel shows autoradiography of the gel Immunoprecipitated radioactivity was quantified using a phosphorimager and the radioactivity remaining was determined relative to time 0 (right panel).
Trang 5the intracellular level of the latter by promoting its
degradation
The SH3 domain of SNX9 binds to the
proline-rich domain of Itch
Given that substrates of Itch and NEDD4 family E3
ligases generally contain PY motifs, we screened the
sequence of SNX9 for such motif but failed to detect
one Previous studies have shown that Itch binds to the
SH3 domains of endophilin A1 and beta-p21-activated
kinase-interactive exchange factor (bPIX) via its PRD
[12,13] We therefore produced a GST fusion product
that contained only the SH3 domain of SNX9 (Fig 1)
and tested whether this truncated version of SNX9 was
able to interact with Itch As shown in Fig 4A,
immo-bilized GST–SNX9(SH3) pulled down Itch from
cellu-lar extracts of SK-MEL-23 These results indicate that
the SH3 domain of SNX9 mediates the interaction with
Itch It is interesting to note that SNX9(SH3) associated
with Itch but not with NEDD4 (Fig 4A) NEDD4 is a close functional homologue of Itch that shares over 45% amino acid identity but diverges in the sequence of its proline-rich segment Under the same experimental conditions, the cytoplasmic tail of the melanosomal protein Melan-A associated with both Itch and NEDD4, as previously shown [3]
Having found that the interaction was mediated by the SH3 domain of SNX9, we tested whether SNX9 binds to the PRD of Itch We co-transfected myc-tagged full-length Itch protein or Itch lacking the PRD (myc-tagged DPRD-Itch) together with VSV-tagged SNX9 into HEK293 cells (Fig 4B) Association between SNX9 and full-length Itch was readily observed, but the interaction between SNX9 and DPRD-Itch was drastically reduced, confirming that SNX9 binds to the PRD of Itch Next, we investigated whether ubiquitylation of SNX9 depends on its interac-tion with the PRD of Itch We therefore co-transfected either full-length Itch or DPRD-Itch protein together
C
D
Fig 4 SNX9 interacts via its SH3 domain with the proline-rich domain of Itch (A) Extracts from SK-MEL-23 cells were incubated with immobilized GST, GST–Itch–PWW, GST–SNX9(SH3) containing only the SH3 domain of SNX9 or GST–Melan-A protein The GST fusion con-structs used are shown in Fig 1 Proteins that bound to the beads were analysed by immunoblotting with monoclonal antibody against Itch
or polyclonal antibodies against NEDD4 Note that SNX9(SH3) does not pulldown NEDD4, a close functional homologue of Itch The known interaction of Melan-A with Itch and NEDD4 was used as a positive control, and the lack of interaction of GST–Itch–PWW with either protein
as a negative control (B) Extracts were prepared from HEK293 cells co-transfected with VSV-tagged SNX9, myc-tagged Itch and myc-tagged DPRD-Itch (lacking the proline-rich domain) in the indicated combinations Extracts were immunoprecipitated using antibody against the myc tag, resolved by SDS ⁄ PAGE and subjected to immunoblotting using antibody against the VSV tag or the myc tag as indicated Cell extracts were analysed in parallel (C) Extracts were prepared from HEK293 cells co-transfected with VSV-tagged SNX9, tagged Itch and myc-tagged DPRD-Itch (lacking the PRD) in the indicated combinations Extracts were immunoprecipitated using antibody against the VSV tag, resolved by SDS ⁄ PAGE and subjected to immunoblotting using antibodies against the myc tag, ubiquitin or the VSV tag as indicated Cell extracts were analysed in parallel (D) Extracts from HEK293 cells transfected with myc-tagged Itch or myc-tagged DPRD-Itch were analysed
by immunoblotting with antibodies against SNX9, actin or the myc tag as indicated.
Trang 6with VSV-tagged SNX9 into HEK293 cells SNX9 was
immunoprecipitated and analysed by Western blotting
using an antibody to ubiquitin Ubiquitylated forms of
VSV-tagged SNX9 were easily detected upon
co-trans-fection of SNX9 with full-length Itch, but not with the
deletion variant lacking the PRD (Fig 4C) In
agree-ment with this finding, over-expression of full-length
Itch in HEK293 cells decreased the level of endogenous
SNX9, but over-expressing Itch protein lacking the
PRD had no impact on SNX9 levels (Fig 4D) The
level of endogenous SNX9 in cells over-expressing
full-length Itch was about one-third of the amount in
non-transfected control cells
Note that, in the analysis of cellular extracts in
Fig 4B,C, co-expression of Itch and SNX9 did not
decrease the level of SNX9 compared to expression of
SNX9 alone, in contrast to results with endogenous
SNX9 shown in Figs 2B and 3D This apparent
discrepancy was due to variability in co-transfection
efficiency (not shown)
Together, our results show that binding of SNX9 to
Itch is mediated by the SH3 domain of SNX9 and the
PRD of Itch Furthermore, interaction with the PRD
of Itch is essential for ubiquitylation of SNX9 and its
intracellular regulation
Phosphorylated SNX9 binds Itch
It has been reported that SNX9 can be phosphorylated
at tyrosine residues by ACK2 (activated
Cdc42-associ-ated tyrosine kinase 2) during membrane recruitment
[9,14] Furthermore, in Drosophila, ACK was found to
phosphorylate Tyr56 within the SH3 domain of SNX9,
abrogating as a result the binding to proline-rich
sequences [10]
To examine whether tyrosine phosphorylation of the
human SNX9 protein had the same impact on
interac-tions with PRD-containing proteins as in Drosophila, we
investigated whether Itch interacts with
tyrosine-phos-phorylated SNX9 HEK293 cells were co-transfected
with plasmids encoding Itch and SNX9 Itch was
immu-noprecipitated from lysates, and SNX9 protein bound
to Itch was eluted and re-immunoprecipitated to analyse
its tyrosine phosphorylation status (Fig 5)
Phosphory-lated SNX9 protein was detected, demonstrating that
tyrosine phosphorylated SNX9 still bound to Itch As a
control, tyrosine phosphorylation of SNX9
immunopre-cipitated directly from cell extracts was confirmed
Itch interacts with SNX18
SNX18 (UniProtKB accession number Q96RF0) is the
closest relative of SNX9, with a very similar domain
structure, including an N-terminal SH3 domain Little
is known about the function of SNX18 It has recently been reported to induce membrane tubulation in AP-1-positive endosomal trafficking [15] In addition, like SNX9, the SH3 domain of SNX18 can bind to dynam-in-2 [15] Given the similarity between SNX9 and SNX18, we analysed the potential interaction of SNX18 with the Ub ligase Itch by co-transfection of myc-tagged SNX18 and HA-tagged Itch into HEK293 cells (Fig 6) After immunoprecipitation of myc-tagged SNX18, bound proteins were eluted and subjected to Western blotting The membrane was probed using an antibody against Itch SNX18 was found to interact with over-expressed as well as endogenous Itch pro-tein, suggesting that it may be regulated in a manner similar to SNX9
Discussion
In this study, we found that the sorting nexin SNX9 is regulated by the E3 Ub ligase Itch We showed that the interaction is mediated by the SH3 domain of SNX9 and the PRD of Itch, leading to ubiquitylation
Fig 5 Tyrosine-phosphorylated SNX9 binds to Itch Extracts were prepared from HEK293 cells transfected with VSV-tagged SNX9, myc-tagged Itch or both, using lysis buffer supplemented with phos-phatase inhibitors Extracts were immunoprecipitated using antibod-ies against the VSV tag or the myc tag Proteins eluted from beads coupled to antibody against the myc tag were re-immunoprecipitated using antibody against the VSV tag Eluted proteins were resolved by SDS ⁄ PAGE and subjected to immunoblotting with antibodies against phospho-tyrosine (4G10), the VSV tag or the myc tag as indicated.
A fraction (75%) of eluted material from immunoprecipitated anti-myc + anti-VSV was analysed using 4G10 antibody, and 10% of eluted material was used for detection with antibody against the VSV tag From material immunoprecipitated using antibody against the VSV tag, a fraction (50%) of eluted material was analysed with 4G10 antibody, and 10% of eluted material was used for detection with antibody against the VSV tag Cell extracts were analysed in parallel.
Trang 7and degradation of SNX9 Previous studies revealed
that the Itch substrate endophilin A1 also associates
via its SH3 domain with the PRD of the ligase [12]
Thus our study represents the second characterization
of an interaction between a substrate and a ubiquitin
ligase of the NEDD4 family that is mediated by a
sequence distinct from the conventional WW domains
Importantly, while the SH3⁄ PRD interaction was
found to facilitate endophilin ubiquitylation in the case
of endophilin A1 [12], the interaction mediated by the
PRD of Itch appears to be essential for ubiquitylation
and degradation of SNX9 The ability of the PRD to
function as a substrate recognition domain in Itch
indicates that the pool of Itch substrates is not limited
to PY motif-containing proteins
We showed that the interaction of SNX9 with Itch
is specific, as SNX9 does not bind to the PRD of the
related Ub ligase NEDD4 NEDD4 and Itch can
inter-act with the same PY motif-containing substrate
pro-tein, such as Melan-A [3], but the two ligases diverge
in the sequence of their proline-rich segments Two
types of proline-rich binding motifs for SH3 domains
have been described, so-called class I and class II sites,
with the consensus sequences Arg-X-X-Pro-X-X-Pro
and Pro-X-X-Pro-X-Arg (X being any amino acid),
respectively [16,17] It has been shown that the SH3
domain of SNX9 binds most strongly to the class I
ligand motif Arg-X-Ala⁄ Pro-Pro-X-X-Pro [15,18]
Class I binding motifs are present in the PRDs of both
Itch and NEDD4 However, the amino acid context of
their binding sites is different, and may affect binding
to SH3 domains Itch has a single discrete class I site
with the sequence Arg-Pro-Pro-Pro-Pro-Tyr-Pro,
con-forming to the preferred consensus SH3 ligand motif
of SNX9 previously reported [18] In contrast, the
PRD of NEDD4 contains two class I sites that overlap
and may thus form a structure that cannot interact
with the SH3 domain of SNX9 In support of this
hypothesis, other SH3 domain-binding partners of
SNX9 such as dynamin-2 also present discrete class I sites within their PRDs that mediate binding to SNX9 [15] In addition to endophilin A1, another SH3 domain-containing protein, bPIX, has been found
to interact with the PRD of Itch [12,13] The complex
of bPIX and Itch serves as a scaffold in G protein-coupled receptor signalling [13] Binding of these two proteins to the PRD of NEDD4 has not been investi-gated, and it is therefore not known whether the inter-action can be mediated by either class I motif present
in Itch and NEDD4 It is likely that there are many as yet unidentified proteins regulated by the NEDD4 family ligases Itch or NEDD4 through interactions with their PRDs
We found that SNX18, a close relative of SNX9 with
a similar domain structure, also interacts with Itch Little is known about the function of SNX18 [15,19] and its regulation SNX9 and SNX18 show more than 40% identity in the amino acid sequence of their SH3 domains, which share a preference for binding to class I ligand motifs and dynamin-2 as a major inter-acting protein [15] It is therefore conceivable that SNX18 is regulated through ubiquitylation by Itch in a manner similar to SNX9 SNX33 is the third and final member of the SNX family identified to date that possesses an SH3 domain It is closely related to SNX9 and SNX18, and has also been found to associate with dynamin [20] It will be interesting to determine whether SNX18 and SNX33 are also substrates of Itch The SH3 domain of SNX9 is essential for the func-tion of SNX9 in clathrin-mediated endocytosis This domain was shown to mediate the interaction of SNX9 with dynamin-2 and N-WASP, a homologue of the Wiskott–Aldrich syndrome protein (WASP) [21–24] Given that the Ub ligase Itch binds to the SH3 domain of SNX9, one could speculate that Itch competes for binding to the SH3 domain with other interaction partners Dynamin-2 and N-WASP are key components of the actin polymerization machinery [10,25,26] Interaction of SNX9 with dynamin-2 recruits dynamin-2 to the clathrin-coated pit and stim-ulates the GTPase activity required for vesicle scission [21,23,27,28] Interaction of SNX9 with N-WASP stim-ulates actin remodelling during endocytosis by enhanc-ing N-WASP activity [22,24,29,30] It is possible that functionally active SNX9 forming a complex with dynamin-2 or N-WASP at membranes cannot be targeted for degradation as the binding site for Itch is not accessible It may be speculated that, once the end-ocytic process is completed, Itch competes for binding
to the SH3 domain of membrane-bound SNX9, thereby destabilizing the complex of SNX9 with dyn-amin-2 and⁄ or N-WASP Itch binding could enable
Fig 6 SNX9 family member SNX18 also interacts with Itch Extracts
were prepared from HEK293 cells transfected with myc-tagged
SNX18, HA-tagged Itch or both Extracts were immunoprecipitated
using antibody against the myc tag, resolved by SDS⁄ PAGE, and
sub-jected to immunoblotting with antibodies against Itch or the myc tag
as indicated Cell extracts were analysed in parallel.
Trang 8the removal of SNX9 from the membrane, followed by
SNX9 ubiquitylation and degradation
Alternatively, Itch may only bind to the SH3 domain
of cytosolic SNX9 In the cytosol, SNX9 forms a
com-plex with aldolase and dynamin-2 [27] Thus, in this
scenario, Itch would have to compete with dynamin-2
for binding to the SH3 domain of SNX9
Ubiquityla-tion of cytosolic SNX9 by Itch could serve as a
mecha-nism to regulate the pool of SNX9 protein in the
cell Our results indicate that SNX9 is a very stable
protein with a half-life of approximately 24 h, at least
when over-expressed Although increased Itch levels
increased the turnover of SNX9, the half-life of the
latter was still relatively long (approximately 15 h)
Given that endocytosis is a fast process, the hypothesis
of Itch-mediated removal of membrane-bound SNX9
upon endocytic vesicle release seems unlikely The long
half-life of SNX9 instead supports the hypothesis that
Itch may regulate the steady-state levels of SNX9
We found that tyrosine-phosphorylated SNX9 can
bind to Itch At odds with our findings, tyrosine
phos-phorylation of the SH3 domain of Drosophila SNX9 by
ACK was found to reduce interactions with
PRD-con-taining proteins [10], hinting at differences between
human and Drosophila SNX9 It has also been reported
that tyrosine-phosphorylated human SNX9 can bind to
the PRD of ACK1 [11], again indicating different roles
of tyrosine phosphorylation in the human and
Droso-phila SNX9 proteins Of note, in the human SNX9
protein, the presence of a tyrosine phosphorylation site
within the SH3 domain has not yet been formally
demon-strated One could thus speculate that phosphorylation
of human SNX9 involves a tyrosine residue that is not
located within the SH3 domain, and may therefore not
affect the binding of SNX9 to PRD-containing proteins
Future studies should address these issues in detail
Phosphorylation of the ligase Itch might also affect
interaction with and ubiquitylation of SNX9 The
PRD of Itch was shown to be phosphorylated by the
serine⁄ threonine kinase JNK (c-Jun N-terminal kinase)
upon treatment with epidermal growth factor (EGF)
[31–33] This phosphorylation leads to an increase in the
ligase activity of Itch, resulting in increased
ubiquityla-tion and degradaubiquityla-tion of substrates such as
endophi-lin A1 and Casitas B-endophi-lineage lymphoma gene (Cbl)
[31] Both endophilin and Cbl are involved in EGF
receptor down-regulation, and their increased
degrada-tion after EGF treatment has been proposed to serve as
a negative feedback mechanism to terminate
internaliza-tion of the EGF receptor [31] Given that SNX9 is also
involved in down-regulation of the EGF receptor after
EGF treatment [9], one could speculate that, like
endo-philin and Cbl, the level of SNX9 ubiquitylation
increases upon EGF-induced phosphorylation of the PRD of Itch
In conclusion, we found that SNX9 is a new sub-strate of the E3 Ub ligase Itch that binds to the PRD,
an unconventional substrate recognition domain in Itch Association of the SH3 domain of SNX9 with the PRD of Itch mediates ubiquitylation and degra-dation of SNX9 Binding of Itch to members of the sorting nexin family with ensuing degradation is an additional level at which E3 Ub ligases can regulate endocytic processes In the future, more SH3 domain-containing substrates of the Ub ligases Itch and NEDD4 are likely to be identified
Experimental procedures Antibodies
The rabbit antiserum against SNX9 was a generous gift from
S Schmid (Scripps Research Institute, La Jolla, CA) Another polyclonal antibody against SNX9 was produced in rabbits by immunization with a full-length GST–SNX9 fusion protein (Eurogentec, Seraing, Belgium) The NEDD4 rabbit antiserum was provided by O Staub (Department of Pharmacology and Toxicology, University of Lausanne, Switzerland) Monoclonal antibody against the myc tag (clone 9E10) was a gift from R.D Iggo (Institut Bergonie´, Bordeaux, France) Commercial antibodies included: monoclonal antibody against Itch (BD Biosciences⁄ Pharmingen, San Diego, CA, USA), monoclonal antibody against the HA tag (clone 16B12) (Covance, Princeton, NJ, USA), polyclonal antibody against actin (Sigma-Aldrich,
St Louis, MO, USA), monoclonal antibody against the VSV tag (clone P5D4) (Sigma-Aldrich), polyclonal antibody against CD2AP (Santa Cruz Biotechnology, Santa Cruz, CA, USA), monoclonal antibody against ubiquitin (clone FK2) (Biomol, Plymouth Meeting, PA, USA) and monoclonal anti-body against phospho-tyrosine (clone 4G10) (Upstate Biotechnology, Charlottesville, VA, USA) Secondary anti-bodies were horseradish peroxidase-conjugated anti-mouse IgG and anti-rabbit IgG (Amersham Biosciences, Pisca-taway, NJ, USA) and horseradish peroxidase-conjugated streptavidin (Dako, Glostrup, Denmark)
Plasmids and lentivirus production
The plasmids pGEX⁄ Melan-A and pEGFP ⁄ Itch-HA have been described previously [3] The plasmid encoding UbHA was a gift from S Rothenberger (Institute of Microbiology, University Hospital Lausanne, Switzerland) The pGEX⁄ Itch-PWW vector was constructed by inserting a blunted XbaI–BglII fragment from the pEGFP⁄ Itch-HA vector [3] into the SmaI site of the pGEX-4T-3 vector (Amersham Biosciences) The full-length SNX9 coding sequence was
Trang 9cloned by RT-PCR from mRNA extracted from
SK-MEL-23 melanoma cells The amplified fragment was digested
with EcoRI and XhoI and inserted into the corresponding
sites of the pCR-3⁄ Met-VSV vector (gift from P Schneider,
Department of Biochemistry, University of Lausanne,
Swit-zerland) or the pGEX-4T-1 vector (Amersham Biosciences)
to generate the pCR3⁄ VSV-SNX9 and the pGEX ⁄ SNX9
plasmids, encoding SNX9 with an N-terminal VSV tag or
fused to GST, respectively pGEX⁄ SNX9(SH3) was
con-structed by PCR amplification of the SH3 domain of SNX9
from the same cDNA as above The PCR product was
inserted into pGEX-4T-1 vector The plasmids encoding
myc-tagged Itch and myc-tagged DPRD-Itch were a
gener-ous gift from T.P Sakmar (Laboratory of Molecular
Bio-logy and Biochemistry, Rockefeller University, New York,
NY, USA) The plasmid encoding myc-tagged SNX18 was a
kind gift from S.R Carlsson (Department of Medical
Bio-chemistry and Biophysics, Umea˚ University, Sweden) The
plasmid pSuper-Itch, encoding Itch siRNA, was produced
by annealing two complementary DNA oligonucleotides
tar-geting nucleotides 88-108 of the ORF encoding human Itch,
as described previously [34], and ligating the fragment into
the pSuper vector (a gift from R Agami, Nederlands
Kan-ker Instituut, Amsterdam, The Netherlands) Recombinant
lv⁄ siItch was constructed by excising this siItch sequence
together with the H1-RNA promoter from the pSuper
vec-tor, and inserting the fragment into the pAB286 lentivector
(gift from R.D Iggo, Institut Bergonie´, Bordeaux, France)
The same strategy was used to produce the control
lentivec-tor recombinant lv⁄ siLamin to silence lamins A and C The
plasmid pSuper⁄ siLamin was a gift from R.D Iggo (Institut
Bergonie´, Bordeaux, France) and has been described
previ-ously [35] All constructs were verified by sequencing
Lenti-virus production was performed as described elsewhere [3]
Cells and transfections
Human embryonic kidney HEK293 cells were maintained
in Dulbecco’s modified Eagle’s medium with 10% fetal
bovine serum and antibiotics Pigmented Melan-A+
SK-MEL-23 human melanoma cells were maintained in
RPMI-1640 medium, supplemented with 10% fetal bovine serum
and antibiotics Transfections were performed using Fugene
(Roche Diagnostics, Indianapolis, IN, USA) or
Lipofecta-mine (Invitrogen, Carlsbad, CA, USA), according to the
manufacturers’ instructions Transfected cells were
main-tained for 48 h at 37C
Silencing of Itch
Expression of Itch and lamins A and C (control) was
silenced in SK-MEL-23 melanoma cells by recombinant
lentivector transduction Transduced cells were selected
using puromycin (2.5 lgÆmL)1) on day 2 and analysed
within 1–4 weeks of culture
Ubiquitin assays
HEK293 cells were transfected with indicated plasmids Twenty-four hours after transfection, the cells were treated for 2 h with 10 lm lactacystin (Biomol), a proteasome inhibitor Ubiquitylated proteins were immunoprecipitated from cell lysates using monoclonal antibody against the
HA tag coupled to protein G–Sepharose Proteins were resolved by SDS⁄ PAGE and immunoblotted using anti-body against SNX9
For the in vitro ubiquitylation assay, bacterially produced GST–SNX9 or GST was adsorbed onto glutathione beads and incubated for 0 or 24 h with cytosolic extracts from
3· 106HEK293 cells per reaction, supplemented with 2 lg biotinylated ubiquitin (Biomol), 10 mm ATP, 5 mm Mg acetate and 0.2 mm DTT To avoid degradation, protea-somes were removed from cytosol prior to the reaction with GST–SNX9, as described previously [3] Where indicated, HEK293 cells were transfected with the plasmid encoding Itch 48 h prior to lysis After the ubiquitylation reaction,
10 mm EDTA was added and the immobilized material was washed and eluted as described previously [3] The superna-tant was then diluted tenfold in cold lysis buffer and re-im-munoprecipitated using antibody against SNX9 The immunoprecipitated material was resuspended in SDS sample buffer, resolved by SDS⁄ PAGE and transferred onto nitrocellulose The ubiquitylated material was revealed using horseradish peroxidase-conjugated streptavidin (1 : 5000, Dako)
Preparation of GST fusion proteins and pulldown assays
Bacteria cultures were grown to an attenuance at 600 nm
of 0.6–0.7 Protein expression was induced with 0.5 mm iso-propyl thio-b-d-galactoside for 2 h at 37C, except for GST–SNX9, which was induced for 4 h at 25C Fusion proteins were purified using the ProFound pulldown GST protein:protein interaction kit (Pierce, Rockford, IL, USA), according to the manufacturer’s instructions Glutathione beads (30 lL) were used for each purification and subse-quent pulldown experiments Pulldown experiments were performed as described previously [3]
Immunoprecipitations and Western blots
Mammalian cells were lysed with Triton X-100 as described previously [3], except for SNX9 immunoprecipitations, where cells were lysed in 0.5% digitonin For analysis of tyrosine phosphorylated proteins, lysis buffer was supple-mented with 50 mm sodium fluoride and 10 mm sodium ortho-vanadate to inhibit phosphatases The protein con-tent was determined using a BCA protein assay kit (Pierce) For immunoprecipitations, cleared cell extracts were incu-bated either with agarose-conjugated antibodies against the
Trang 10HA tag (clone 3F10) (Roche) or antibodies against the
VSV tag (clone P5D4) (Sigma-Aldrich),
Sepharose-conju-gated antibodies against the myc tag (clone 9E10), or
pro-tein G beads (Pierce) plus polyclonal antibodies against
SNX9 or CD2AP (Santa Cruz Biotechnology) After 1–2 h
of incubation at 4C, immunoprecipitated material was
extensively washed, eluted in SDS sample buffer and boiled
at 95C for 5 min Proteins were reduced, resolved on
SDS⁄ PAGE and transferred onto nitrocellulose
Mem-branes were subsequently incubated with blocking solution
(5% milk in NaCl⁄ Pi, except for 4G10 monoclonal
anti-body against where 1% gelatine in NaCl⁄ Pi⁄ Tween was
used) and primary antibody as indicated Immunoreactivity
was detected using horseradish peroxidase-conjugated
sec-ondary antibodies and ECL (Amersham Biosciences)
according to the manufacturer’s instructions Quantitative
analysis was performed using ImageJ quantification
soft-ware (National Institutes of Health, Bethesda, MD, USA)
Pulse–chase experiments
HEK293 cells were plated on 60 mm plates and transfected
with VSV-tagged SNX9 or a combination of VSV-tagged
SNX9 and myc-tagged Itch Twenty-four hours after
transfection, cells were starved for 30 min in Dulbecco’s
modified Eagle’s medium lacking cysteine and methionine
(MP Biomedicals, Solon, OH, USA) supplemented with
10% dialysed fetal bovine serum, then incubated with
Easy-Tag Express 35S Protein Labeling Mix (Perkin Elmer,
Waltham, MA), using 110 lCi per 2 mL per plate After
30 min (pulse), radioactive medium was removed and cells
were incubated in Dulbecco’s modified Eagle’s medium
sup-plemented with 10% fetal bovine serum, 10 mm HEPES
and antibiotics for 6 or 24 h (chase) Cells were lysed for
20 min on ice in a buffer containing 80 mm Tris pH 7.6,
150 mm NaCl, 2 mm EDTA, 1% Nonidet P-40, 1% sodium
deoxycholate and 0.05% SDS, supplemented with a
cock-tail of protease inhibitors Aliquots of lysates were used
to measure the incorporated radioactivity using a liquid
scintillation counter Lysates were subjected to
immunopre-cipitation as described above Eluted material was resolved
by 10% SDS⁄ PAGE The gel was then fixed in 40%
metha-nol⁄ 10% acetic acid for 15 min, dried, and exposed to an
imaging plate (Fuji Medical Systems, Stamford, CT, USA)
overnight The plate was scanned using an FLA-3000
phos-phorimager (Fuji Medical Systems) and BASReader 3.14
software (Raytest, Straubenhardt, Germany) aida image
analyzer software (Raytest) was used for quantitative
analy-sis Autoradiography was performed by exposing the dried
gel to X-ray film (Kodak, Rochester, NY, USA)
Mass spectrometry
Gel bands were excised from SDS⁄ PAGE and subjected to
in-gel proteolytic cleavage using sequencing-grade trypsin
(Promega, Madison, WI, USA) [36] Proteolytic peptides were analysed by LC-MS⁄ MS on a SCIEX QSTAR Pulsar hybrid quadrupole-time of flight instrument (Concord, Ontario, Canada) equipped with a nanoelectrospray source and interfaced to an Ultimate HPLC system (LC Packings, Amsterdam, Netherlands) Peptides were separated on a PepMap reverse-phase capillary C18 column (internal diam-eter 75 lm, length 6.15 cm) at a flow rate of 200 nLÆmin)1 using a 52 min gradient of acetonitrile (0–40%) The Ana-lyst software (Applied Biosystems, Concord, Ontario, Canada) was used for peak detection and to automatically select peptides for collision-induced fragmentation Non-interpreted peptide tandem mass spectra were used for direct interrogation of the uniprot (Swissprot + TrEMBL) database using mascot2.1 (http://www.matrixscience com), limited to the subset of sequences from Homo sapiens The database used contained 64 386 sequences after taxonomy filter Generally, only proteins matched by at least two peptides were accepted
Acknowledgements
We thank Manfredo Quadroni (Department of Bio-chemistry, University of Lausanne, Switzerland) and the Protein Analysis Facility (Center for Integrative Genomics, Faculty of Biology and Medicine, Univer-sity of Lausanne, Switzerland) for mass spectrometry analysis We gratefully acknowledge the excellent tech-nical assistance of Anne-Lise Peitrequin and Nicole Le´vy C.L was supported by a post-doctoral fellowship from the Wenner-Gren Foundation (Stockholm, Swe-den) D.R was supported in part by grants from the Swiss National Funds (grant number 31003A-108283) and the Faculty of Biology and Medicine of the Uni-versity of Lausanne (grant number 22172) F.L was supported in part by grants from the Swiss National Funds (grant number 310000-107686), the Cancer Research Institute (New York, NY, USA), the Leena-ards Foundation (grant number 2081) and the Faculty
of Biology and Medicine of the University of Lausanne (grant number 22172)
References
1 Sudol M & Hunter T (2000) NeW wrinkles for an old domain Cell 103, 1001–1004
2 Melino G, Gallagher E, Aqeilan RI, Knight R, Pesch-iaroli A, Rossi M, Scialpi F, Malatesta M, Zocchi L, Browne G et al (2008) Itch: a HECT-type E3 ligase regulating immunity, skin and cancer Cell Death Differ
15, 1103–1112
3 Levy F, Muehlethaler K, Salvi S, Peitrequin AL, Lindholm CK, Cerottini JC & Rimoldi D (2005) Ubiquitylation of a melanosomal protein by HECT-E3