|Home | About | Journals | Submit | Contact Us | Français|
Wnt proteins control multiple cell behaviors during development and tissue homeostasis. However, pathological activation of Wnt signaling is the underlying cause of various human diseases. The ubiquitin-proteasome system plays important regulatory functions within the Wnt pathway by regulating the activity of several of its core components. Hence, multiple E3 ubiquitin ligases have been implicated in its regulation. Less is known, however, about the role of ubiquitin-specific proteases in Wnt signaling. Analysis of purified axin-containing protein complexes by liquid chromatography-tandem mass spectrometry revealed the presence of the ubiquitin protease USP34. Our results indicate that USP34 functions downstream of the β-catenin destruction complex to control the stability of axin and opposes its tankyrase-dependent ubiquitination. Reflecting on the requirement for tight control of axin homeostasis during Wnt signaling, interfering with USP34 function by RNA interference leads to the degradation of axin and to the inhibition of β-catenin-mediated transcription. Given the numerous human diseases exhibiting spurious Wnt pathway activation, the development of USP34 inhibitors may offer a novel therapeutic opportunity.
During embryonic development and tissue homeostasis in adults, the Wnt family of secreted glycoproteins modulates several cell behaviors, including differentiation, proliferation, cell movement, and polarity (32, 37). Malfunctioning Wnt-activated signaling pathways are associated with multiple human diseases, including cancer (10, 38). The etiology of colon carcinoma is a particularly striking example that reflects the critical importance of the integrity of this signaling cascade during intestinal epithelium homeostasis (45). Approximately 80% of all colon cancers are molecularly rooted in mutations of Wnt pathway components. These primarily consist of inactivating mutations in the gene coding for the tumor suppressor adenomatous polyposis coli (APC) (44, 47, 51) but also of activating mutations in the transcription factor β-catenin (39) and loss-of-function mutations in the scaffolding axin protein (22).
APC and axin are the core components of a cellular machinery dubbed the “destruction complex” that promotes the phosphorylation of the cytoplasmic pool of β-catenin (24). Axin, through binding to the destruction complex kinases casein kinase 1α (CK1α) and glycogen synthase kinase 3 (GSK3), orchestrates β-catenin phosphorylation (31). Phospho-β-catenin is in turn recognized by the SCFβ-TrCP (Skp1-Cullin1-FBOX) E3 ubiquitin ligase that polyubiquitinates β-catenin and promotes its proteolysis by the proteasome (26, 59). The destruction complex thereby maintains low levels of cytosolic β-catenin in the absence of Wnt stimulation. The recognition of Wnt ligands by the cell surface receptor complex Frizzled-LRP5/6 leads to the activation of Dishevelled (Dsh) (62), which promotes the GSK3- and CK1γ-dependent phosphorylation of the LRP5/6 cytosolic domain (12, 63). The phosphorylated LRP5/6 cytosolic domain acts as a high-affinity binding site for axin (36, 53) that is suspected to inactivate the destruction complex and to lead to β-catenin accumulation. Stabilized β-catenin can then enter the nucleus and cooperate with LEF/TCF transcription factors to regulate Wnt-dependent transcriptional programs in a context-dependent fashion (50).
The ubiquitin-proteasome system (UPS) is emerging as master regulator of Wnt signaling, controlling the pathway at multiple levels. In addition to the well-characterized function of the SCFβ-TrCP E3 ligase for β-catenin ubiquitination in the absence of Wnt-driven signals (17, 26, 59), other proteins of the pathway are either targeted for degradation or regulated by the UPS. The ubiquitination of APC (9, 56) and Dishevelled (3, 54), for instance, leads to their proteasome-mediated degradation or to degradation-independent functional regulation. This dual regulation by the UPS depends on whether K48- or K63-linked ubiquitin chains are involved. Although the E3 ubiquitin ligase for APC has not been identified, this process is thought to involve axin, at least for the situation where APC is degraded (56). Another example is the posttranslational control of Dsh stability by the Cullin3-KLHL12 E3 ligase (3). Consistent with roles in both β-catenin-dependent and -independent Wnt pathways for Dsh, the activity of this E3 ligase was shown to impact both pathways in Xenopus and zebrafish embryos. Axin has also been postulated to be regulated through the modulation of its stability, which might be a necessary step for the activation of the β-catenin pathway (27, 58). The precise mechanisms regulating the degradation of axin are, however, not known at present, but its parsylation by tankyrase and its sumoylation have recently been shown to control its ubiquitin-dependent degradation (20, 23).
Due to the multiple roles of the UPS in Wnt signaling, it is likely that members of the ubiquitin-specific proteases (USPs; also termed deubiquitinating enzymes [DUBs]) regulate some of these events and could therefore have important functional roles in Wnt signaling. An estimated 79 USPs are present in humans that function to remove ubiquitin conjugates from target proteins (43). Supporting the possibility that USPs may regulate Wnt signaling, recent report have identified the ubiquitin protease Trabid (56) and USP4 (64) as novel regulators of this pathway. Trabid regulates APC function through the editing of its K63-conjugated chains, whereas USP4 regulates TCF4 (64).
A recurrent theme in Wnt signal transduction is the reutilization of Wnt pathway components in different subcellular compartments, often to perform alternate functions. For example, Dsh has been localized to punctate structures within the cytoplasm (7, 49) or to the plasma membrane upon Wnt activation of the Frizzled-LRP receptor complex (5, 62). However, other studies have shown that Dsh is also translocated to the nucleus, where it performs a required but ill-defined role during Wnt signaling (15, 21). β-Catenin-independent Wnt signaling also likely involves the relocalization of Dsh to additional subcellular structures in order to modulate cytoskeleton-associated processes (4). Likewise, GSK3 acts primarily as a negative regulator of Wnt signaling by promoting the phosphorylation of β-catenin. However, as mentioned above, GSK3 also plays a positive role, at the plasma membrane, via the phosphorylation of the LRP5/6 Wnt coreceptor (12, 63) and has also been found to have nuclear roles (8). Similarly, in addition to its task in the destruction complex, a nuclear role has been proposed for APC in Wnt signaling. Indeed, APC contains bipartite nuclear localization and nuclear export signals that promote its nuclear cytoplasmic shuttling (18, 40, 46). Nuclear APC antagonizes β-catenin-mediated transcription by either the modulation of β-catenin nuclear export (18), the sequestration of β-catenin away from an active transcription complex (41), or its association with transcriptional repressors (16). In contrast, a recent genetic screen in Drosophila uncovered a positive functional role for APC homologs in Wg signaling (52). It is therefore a common theme in Wnt signaling that its effectors are reutilized in a context-dependent manner.
Axin, normally associated with the destruction complex, does not escape this trend as it is recruited to the activated and phosphorylated LRP5/6 coreceptor (36, 53) at the plasma membrane. Moreover, axin is also known to shuttle between the nucleus and the cytoplasm (11, 57) and is greatly enriched in the nuclei of cells from diverse cancer cell lines and tissues (1, 29, 48, 60). However, the precise function of nuclear axin in Wnt signaling is not well understood.
Here, by using a proteomic approach, we show that axin associates with ubiquitin-specific protease 34 (USP34). Our results indicate that USP34 controls the levels of axin and positively modulate Wnt signaling by acting downstream of β-catenin stabilization through controlling the nuclear accumulation of axin.
Human AXIN1 and AXIN2 cDNAs were cloned by PCR from a human brain cDNA library into the pGLUE tandem-affinity purification (TAP) plasmid (3) that contains streptavidin (SBP)- and calmodulin-binding peptides (CBP) to generate pGLUE-hAXIN1 and pGLUE-hAXIN2. AXIN1 was also cloned downstream of a cDNA coding for the Venus fluorescent protein in the pIRES-puro vector to generate the pIRES-puro-Venus-hAXIN1 plasmid. The human point mutant β-catenin (pt.mutant-hβCATENIN-CBP-HA-SBP) (34) and human Dishevelled-2 (pGLUE-hDSH2) (3) were described previously. USP34 core (residues 1892 to 2241) was expressed and purified as a His-tagged protein from Escherichia coli. USP2 core was expressed and purified as previously described (42). All PCR-amplified regions were sequence validated. Detailed description of plasmid maps and sequences will be provided upon request and are posted on the lab web site (http://phm.utoronto.ca/angers/).
Human HEK293T, RKO colon carcinoma (ATCC CRL-2577), SW480 colorectal adenocarcinoma (CCL-228), HCT116 colorectal carcinoma (CCL-247), and mouse L cells (CRL-2647/CRL-2648) were grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum and penicillin-streptomycin (Sigma-Aldrich, St. Louis, MO) in a 37°C humidified incubator with 5% CO2. HEK293T stable cell lines were generated by transfection with calcium phosphate, followed by puromycin selection (2 μg/ml). Transient cDNA transfections were performed according to the manufacturer's recommendations using Lipofectamine 2000 (Invitrogen, Carlsbad, CA).
For small interfering RNA (siRNA) experiments, cells were transfected with 20 nM siRNA with the recommended amounts of Lipofectamine RNAiMax (Invitrogen). Previously validated siRNAs against β-catenin, AXIN1, and AXIN2 (34) and nontargeting control (Dharmacon, Lafayette, CO) were used, while a set of four siRNAs targeting USP34 was obtained from Dharmacon (catalog no. LQ-006082-00-0002) and tested by Western blotting. Within this set, the USP34 siRNA A was the most effective, and its target sequence was 5′-GCAGGGAAGUUCUGACGAA-3′. The target sequences of the other USP34 siRNAs were as follows: B, 5′-CAACAGAUCAGUAGUAAUU-3′; C, 5′-GCAGCUAUCCAGUAUAUUA-3′; and D, 5′-CCAUGUGACUGGAGAUUUA-3′.
For the epistasis experiments involving the expression of pt.mutant-βCATENIN or DSH2 with a given siRNA, the siRNAs were first reverse transfected at the time of seeding cells, followed by replacement of the medium 24 h after seeding and cDNA transfection with Lipofectamine 2000. Cells were then assayed 36 h after cDNA transfection using the TopFlash reporter assay. pGIPZ-based shRNAs for USP34 were obtained from Open Biosystems and screened for their efficiency by Western blotting. The target sequence of the most efficient USP34 shRNA was 5′-CCTATGATGGTTGTTCAAATT-3′.
Mouse L cells expressing Wnt3A (CRL-2647) were cultured until reaching 90% confluence, after which the medium was collected and refreshed every 2 days for a total of 6 days. Medium from different days was assayed by using TopFlash assays to determine the fractions with maximal activity and subsequently used for Wnt stimulation experiments. Conditioned medium from parental mouse L cells not producing Wnt3A (CRL-2648) was also collected to use as a control.
Protein lysates were resolved using SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Blots were stained with antibodies indicated in the figure legends and then incubated with horseradish peroxidase-conjugated secondary antibodies and detected by using chemiluminescence. The antibodies used included αβ-catenin (catalog no. 9587; Cell Signaling Technologies); rabbit monoclonal α-AXIN1 (catalog no. 2074; Cell Signaling Technologies); polyclonal α-AXIN1 (obtained from J. Woodgett, Mt. Sinai, Toronto, Ontario, Canada); p44/42 mitogen-activated protein kinase (ERK; catalog no. 9102; Cell Signaling Technologies); α-USP34 (A300-824A; Bethyl Labs); α-lamin-B (sc-6217; Santa Cruz Biotechnology); α-HA (MMS-101P; Covance); α-FLAG (F1804; Sigma); and peroxidase-conjugated secondary anti-goat, anti-rabbit, and anti-mouse antibodies (catalog numbers 705-035-147, 711-035-152, and 715-035-150; Jackson Immunoresearch Laboratories).
HEK293T cells (~2 × 108 cells) expressing SBP-HA-CBP-tagged AXIN1 or AXIN2 were used for the tandem-affinity purification procedure as previously described (3). Briefly, the cells were lysed with tandem-affinity purification lysis buffer (10% glycerol, 50 mM HEPES-KOH [pH 8.0], 100 mM KCl, 2 mM EDTA, 0.1% NP-40, 2 mM dithiothreitol [DTT], 10 mM NaF, 0.25 mM NaOVO3, and protease inhibitors [Sigma]). The lysates were cleared by centrifugation at 16,000 × g for 10 min and then incubated at 4°C with 100 μl of packed streptavidin resin (GE Healthcare). The beads were washed, and protein complexes were then eluted from the streptavidin resin in calmodulin binding buffer supplemented with 2 mM biotin. The second round of affinity purification was performed with 100 μl of calmodulin resin (GE Healthcare). After several washes, the protein complexes were eluted with two 100-μl elutions with calmodulin elution buffer (50 mM ammonium bicarbonate [pH 8.0], 10 mM EGTA) and directly digested with sequencing-grade trypsin (Promega). The resulting peptide mixture was then analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) using data-dependent acquisition on a LTQ-XL mass spectrometer (Thermo Scientific). Acquired spectra were searched against a FASTA file containing the human NCBI sequences by using a normalized implementation of SEQUEST. The resulting peptide identifications returned by SEQUEST were filtered and assembled into protein identifications using the transproteomic pipeline software running on a Sorcerer platform (SageNResearch).
Lentivirus containing the TopFlash β-catenin-dependent luciferase reporter (firefly luciferase) and Renilla luciferase were produced and used to establish stable HEK293T, RKO, SW480, and HCT116 Wnt-reporter lines. Cells were seeded on 24-well plates, followed by cDNA transfection with Lipofectamine 2000 and/or reverse transfection with Lipofectamine RNAiMax for siRNA experiments. For experiments involving Wnt stimulation, the medium was replaced with a 1:1 mix of fresh DMEM-Wnt3A or DMEM-control conditioned medium. The cells were then assayed 24 h after stimulation in accordance with the dual luciferase assay protocol (Promega) using an Envision multilabel plate reader (Perkin-Elmer).
For co-affinity purification of endogenous proteins, HEK293T cells (5 × 106) stably expressing pGLUE-HA-hAXIN1 or pGLUE-HA-RADIL were lysed in tandem-affinity purification lysis buffer (10% glycerol, 50 mM HEPES-KOH [pH 8.0], 100 mM KCl, 2 mM EDTA, 0.1% NP-40, 2 mM DTT, 10 mM NaF, 0.25 mM NaOVO3, and protease inhibitors [Sigma]). Lysates were cleared by centrifugation at 16,000 × g for 10 min, and affinity purification was performed using streptavidin resin. Purified protein complexes were then analyzed by Western blotting using the antibodies noted in the figure legends.
Next, 1 μg of purified K48 chains from Boston Biochem (UC-230) was incubated in USP assay buffer (20 mM Tris [pH 8.0], 2 mM CaCl2, 2 mM β-mercaptoethanol) with 20 nM USP2 core, 100 nM USP34 core, 1 μg of affinity-purified axin complex, or 1 μg of axin complex (USP34 shRNA). The samples were incubated at 37°C for 30 min, and the reaction was stopped by the addition of SDS sample buffer. The appearance of monoubiquitin was monitored by Western blotting with α-ubiquitin antibody (Sigma catalog no. U5379).
20 nM USP2 core, 20 nM USP34 core, or 1 μg of total protein from purified axin complexes was mixed with 30 nM Ub-PLA2 and 20 μM NBD C6HPC (phospholipase A2 [PLA2] substrate; Invitrogen) in a total volume of 100 μl/well in a black 96-well-plate (Greiner Bio-One). The data were collected 45 min after the addition of Ub-PLA2 and NBD C6HPC on a Perkin-Elmer Envision fluorescence plate reader with excitation and emission filters of 475 and 555 nm, respectively. The net relative fluorescence units (RFU) were then used to calculate the signal (isopeptidase or complexes plus reporter)-to-background (reporter) ratio. For UBL selectivity assays, the relative isopeptidase activity against various UBL-PLA2 fusions was determined by adding the USP34 core to a final concentration of 20 nM in combination with 20 μM NBD C6-HPC and 30 nM concentrations of the individual UBL-PLA2 reporter constructs and is expressed as a percentage of the control isopeptidase: USP2 core (Ub-PLA2), Senp1core (SUMO3-PLA2), Den1 (NEDD8-PLA2), or PLpro (ISG15-PLA2). The UBL-PLA2 assay reagents are available from LifeSensors, Inc., as CHOP reporter kits.
HEK293T cells stably expressing STREP-HA-AXIN1 were transfected with a plasmid coding for FLAG-ubiquitin. In parallel, two 100-mm petri dishes of HEK293T were transfected with pIRES-puro plasmids expressing STREP-HA-USP34 core domain (amino acids 1696 to 2400) or a catalytically inactive STREP-HA-USP34 core domain (C1903S). At 16 h before lysis, STREP-HA-AXIN1 cells were treated with 1 μM MG132. The cells were lysed separately in TAP lysis buffer supplemented with protease inhibitors in the absence (USP34) or the presence (axin) of 5 mM N-ethylmaleimide (NEM). The proteins were affinity purified using streptavidin beads. After extensive washes in TAP lysis buffer and two washes in DUB buffer (50 mM Tris [pH 8], 150 mM NaCl, 2 mM EDTA, 2 mM MgCl2, 2 mM DTT), an equivalent amount of AXIN1 was incubated for 1 h at 37°C with the USP34 core domains, as indicated in the figures. Proteins were resolved by SDS-PAGE and blotted with FLAG antibodies to detect ubiquitin conjugates and hemagglutinin (HA) to monitor axin and USP34 core domain expression.
Cells were seeded on poly-d-lysine-treated coverslips and, when indicated, reverse transfected with siRNA. At 48 h after transfection the cells were fixed with 4% paraformaldehyde/phosphate-buffered saline (PBS) for 20 min and then permeabilized and blocked with 0.2% Triton X-100 and 10% normal donkey serum-PBS for 20 min. Where indicated, cells were treated with 5 ng of leptomycin B (LMB; LC Laboratories)/ml for 3 h. The cells were then stained for indirect immunofluorescence using polyclonal α-AXIN1 antibodies (provided by J. R. Woodgett) and Alexa 488-conjugated anti-rabbit antibodies. Cells were mounted with Vectorshield (Vector) and examined by using a laser scanning confocal microscope (Zeiss LSM 510).
HEK293T cells expressing scramble or USP34 shRNA were seeded onto a 6-well plate. The cells were treated with 3 μM XAV939 for 16 h, washed twice with PBS, and treated with 10 μg of cycloheximide/ml for the indicated times. Then, the cells were lysed using TAP lysis buffer supplemented with protease inhibitors. Equivalent amounts of proteins were resolved by using SDS-PAGE, followed by Western blotting with AXIN1 or tubulin antibodies.
RKO cells were reverse transfected with siRNA. At 48 h after transfection, the cells were stimulated with control or Wnt3A conditioned medium treatment for different times (0.5 to 6 h). The cells were washed and lysed with radioimmunoprecipitation assay buffer (25 mM Tris-HCl [pH 7.6], 150 mM NaCl, 1% NP-40, 0.25% sodium deoxycholate, 0.1% SDS) for 15 min and then cleared by centrifugation at 16,000 × g for 10 min before being resuspended in SDS sample buffer and resolved by SDS-PAGE. β-Catenin accumulation was monitored by Western blotting.
HEK293T cells stably expressing human AXIN1 (pGLUE-AXIN1) were transfected with FLAG-ubiquitin using calcium phosphate. The cells were lysed 48 h after transfection using TAP lysis buffer supplemented or not with 20 mM NEM (Sigma) and then cleared by centrifugation at 16,000 × g for 10 min. Axin was purified by streptavidin affinity chromatography for 1 h. Resin beads were then washed three times with lysis buffer (also supplemented with NEM when indicated), and the protein complexes were eluted using 2× SDS sample buffer, followed by SDS-PAGE electrophoresis and Western blotting with FLAG antibodies to detect ubiquitin-conjugated axin proteins.
Total RNA from SW480 cells treated with control or USP34 siRNAs was purified by using Tri-Reagent (Sigma). After DNase I (Invitrogen) treatment, RNA was reverse transcribed into cDNA by using a high-capacity cDNA reverse transcription kit (Applied Biosystems). The primer sequences used were as follows: CYCLOPHILIN, 5′-GGAGATGGCACAGGAGGAA-3′ and 5′-GCCCGTAGTGCTTCAGTTT-3′; NKD1, 5′-TGAGAAGAAGATGGAGAGAGTGAGCGA-3′ and 5′-GGTGACCTTGCCGTTGTTGTCAAA-3′; and TNFRSF19, 5′-GGAGTTGTCTAAGGAATGTGG-3′ and 5′-GCTGAACAATTTGCCTTCTG-3′. Primer pair efficiencies were validated as previously described (6). Quantitative reverse transcription-PCR (RT-PCR) analysis was carried out in triplicate using an Applied Biosystems Prism 7900HT instrument. Each reaction contained 12.5 ng of cDNA, 150 nM concentrations of each primer, and a Power SYBR green PCR Master Mix (Applied Biosystems). Gene expression analysis was performed by using the comparative cycle threshold (CT) method, normalized to CYCLOPHILIN expression, and the fold changes were calculated relative to control siRNA-treated cells.
To better understand the regulation of axin and its mechanism of action, we isolated human AXIN1 and AXIN2 protein complexes and analyzed their compositions by using LC-MS/MS. We constructed two expression vectors, pGLUE-AXIN1 and pGLUE-AXIN2, and used them to derive HEK293T human cell lines stably expressing fusion proteins of AXIN1 or AXIN2 harboring streptavidin- and calmodulin-binding peptides, as well as the HA epitope (SBP-HA-CBP) in frame with their N termini. We recently optimized this system to rapidly and efficiently purify protein complexes from mammalian cells by using dual affinity tags for their analysis by a gel-free LC-MS/MS approach (2, 3). The detection of several proteins previously demonstrated to associate with axin, including APC, CK1α, β-catenin, PP2A, GSK3β, and GSK3α, demonstrates the efficiency of our approach (Fig. 1A, green circles, and Table 1). Remarkably, we found in both AXIN1 and AXIN2 protein complexes the previously uncharacterized protein USP34 (Fig. 1A, blue circle, and Table 1), which contains 3,546 amino acids and possesses a central ubiquitin hydrolase domain characteristic of DUBs (33).
The presence of endogenous USP34 in AXIN1 complexes was then confirmed in coimmunoprecipitation studies (Fig. 1B). Cell lysates from HEK293T cells stably expressing SBP-HA-CBP-AXIN1 were subjected to affinity purification using streptavidin affinity chromatography to isolate axin protein complexes and probed for endogenous USP34 using Western blotting with anti-USP34 polyclonal antibodies (Fig. 1B). Importantly, a cell line stably expressing a control protein (RADIL) identically tagged and expressed at similar levels did not coprecipitate with USP34 (Fig. 1B, lane 4). Probably reflecting the transient nature of the interaction, attempts to perform endogenous coimmunoprecipitation of AXIN1 and USP34 were challenging. By stabilizing ubiquitinated axin with MG132, we were, however, able to reproducibly detect small amounts of USP34 in AXIN1 immunoprecipitates (data not shown). We therefore conclude that axin and USP34 are present in the same protein complex.
Since USP34 belongs to the family of USPs, we next tested the prediction that USP34 confers ubiquitin-protease activity to the axin complex. To test this possibility, we performed ubiquitin-protease assays using purified axin protein complexes. Axin complexes were isolated from SBP-HA-CBP-AXIN1 expressing cells using a single streptavidin affinity chromatography step and were incubated with recombinant K48-linked polyubiquitin chains. The presence of USP activity in the axin complexes was revealed by the production of a band corresponding to cleaved monoubiquitin as detected by Western blotting (Fig. 2A, lane 3).
As an alternative approach to monitor USP activity, we used the newly developed UB-PLA2 assay (42) to quantify ubiquitin isopeptidase activity present in purified axin complexes. Briefly, this assay consists of a fusion protein containing UBIQUITIN fused to the N terminus of PLA2 used as a reporter enzyme. Since PLA2 requires a free N terminus to be catalytically active, the UB-PLA2 fusion is inactive, and its enzymatic activity is only restored upon cleavage of the α-peptide-linked ubiquitin moiety. Since most USPs to date can cleave the α- or ε-linkage with equal efficiency (25, 30), the UB-PLA2 assay can act as a sensitive and quantitative reporter of ubiquitin isopeptidase activity. Affinity-purified axin or RADIL (control) protein complexes were assayed using the UB-PLA2 assay as described in Materials and Methods. Robust isopeptidase activity could be detected in axin complexes compared to control complexes that exhibited background activity (Fig. 2B, compare lanes 3 and 5). We then tested whether the isopeptidase activity present in the axin complexes was attributable to USP34 in both of these assays by depleting USP34 levels in SBP-HA-CBP-AXIN1 cells with the stable expression of a USP34 shRNA that reduced its protein levels by 90% (Fig. 2C). We found that the affinity-purified fraction isolated from equivalent number of cells expressing the USP34 shRNA was largely devoid of ubiquitin isopeptidase activity (Fig. 2A, lane 4, and Fig. 2B, lane 4). We also expressed and purified the USP34 core region as a recombinant protein in E. coli and found that it exhibited robust ubiquitin isopeptidase activity similar to the USP2 core region, which is included as a positive control for the assay (Fig. 2A and B, compare lanes 1 and 2). Variants of this assay where the ubiquitin protein fused to the N terminus of PLA2 is replaced with other ubiquitin-like proteins such as SUMO, NEDD8, or ISG15 allow the determination of the cleavage specificity of the isopeptidase. Using these different reporters, we showed that the recombinant core USP34 enzyme exhibits specificity for ubiquitin cleavage (Fig. 2D).
These results led us to investigate whether ubiquitin proteases could control the ubiquitination of axin. We used SBP-HA-CBP-AXIN1 stable cells in which we transfected a FLAG-ubiquitin expression plasmid. Axin was then affinity purified by streptavidin affinity chromatography, and the ubiquitin-axin (UB-axin) conjugates were detected by Western blotting with FLAG antibodies. Under our normal protein isolation conditions (first incubated with 1% SDS to disrupt protein-protein interactions), only small amounts of ubiquitinated axin could be detected (Fig. 2E, lane 1). However, since USP34 is present in axin complexes, it may cleave axin-linked ubiquitin chains. We thus inhibited USP activity by incorporating the sulfhydryl alkylating agent NEM in the lysis buffer. NEM is known to react with the catalytic cysteine residue within USP core domains to irreversibly inhibit their protease activity. Under conditions where NEM is present, we detected robust polyubiquitination of axin (Fig. 2E, lane 2). To directly show that USP34 can cleave ubiquitin chains conjugated to axin, we performed an in vitro deubiquitination reaction. We purified UB-axin conjugates and the catalytic core domain of USP34 by affinity purification from HEK293 transfected cells. We then incubated equivalent amount of UB-axin with (Fig. 2F, lane 2) or without (Fig. 2F, lane 1) USP34 core proteins and showed that the core domains could efficiently cleave the ubiquitin chains associated with axin. As a control, we generated a catalytically inactive USP34 core domain (C1903S), performed the same experiment, and showed that it had no effect on UB-axin conjugates (Fig. 2F, lane 3).
Our results suggest that axin protein complexes exhibit ubiquitin-protease activity and that ubiquitin proteases regulate the steady-state ubiquitination of axin. We also conclude that USP34 can directly deubiquitinate axin.
Since axin's stability has been found to be controlled by the UPS (20, 23), we tested whether USP34 was involved in this process. We treated cells stably expressing Venus-AXIN1 with control or USP34 siRNAs and determined the impact on the steady-state levels of axin proteins by fluorescence microscopy and Western blotting. Consistent with USP34 regulating the ubiquitination status of axin, its knockdown led to a robust decrease in Venus-AXIN1 levels (Fig. 3A, compare lanes 1 and 2 on the Western blots). This could be efficiently rescued when the cells were treated with 5 μM MG132 for 10 h prior to assay (Fig. 3A, compare lanes 2 and 5). These results are consistent with the possibility that reducing USP34 function leads to an increase in axin ubiquitination, thereby targeting it for proteolysis by the 26S proteasome. Similar results were obtained when we studied the stability of endogenous AXIN1 (Fig. 3A, Western blot at bottom right). Importantly, two independent USP34 siRNAs (A and D) had the same effect on axin stability (data not shown). To quantify axin levels, we have also measured the pixel intensities in images taken for several independent cells in the different conditions (Fig. 3C).
Recently, the tankyrase-dependent parsylation of axin was demonstrated to be a prerequisite for its ubiquitination, and the small molecule tankyrase inhibitor XAV939 was shown to stabilize axin and to inhibit Wnt signaling (20). If the degradation of axin induced by the depletion of USP34 is mediated through the regulation of its ubiquitination, treating USP34-depleted cells with XAV939 should rescue the degradation of axin. To test this prediction, we incubated green fluorescent protein (GFP)-AXIN1 expressing cells with control or USP34 siRNAs for 48 h and added XAV939 for the last 12 h. We showed that XAV939 reversed the degradation of axin resulting from USP34 depletion (Fig. 3B, compare lanes 2 and 4 on Western blot). To determine whether USP34 controls the turnover of axin, we performed a cycloheximide chase analysis. To perform this experiment, we used HEK293 cells expressing control or USP34 shRNA and first stabilized the endogenous pool of axin by treating the cells with XAV939 for 16 h. We then washed the XAV939 to restore the ubiquitination of axin and incubated the cells for different times in the presence of the protein synthesis inhibitor cycloheximide. We observed that the knockdown of USP34 leads to a precocious turnover of axin proteins compared to control shRNA-expressing cells (Fig. 3D). We conclude that USP34 controls the levels of axin by opposing its tankyrase-dependent ubiquitination.
We next assessed the functional importance of USP34 for Wnt signal transduction using RNAi. The effectiveness of four independent siRNAs designed to target the USP34 mRNA was first established by immunoblotting. RKO cells were transfected with control or USP34 siRNAs for 48 h and endogenous USP34 protein levels were subsequently measured by using Western blot with anti-USP34 antibodies. All four USP34 siRNAs could block protein expression with various efficiencies (Fig. 4A, bottom panels), and similar results were obtained with these siRNAs in downstream experiments. This is especially relevant since we were unable to clone and express a full-length USP34 cDNA (10,638 bp) to perform rescue experiments. We also attempted to rescue the USP34 siRNA effect with a cDNA expressing only the core domain but observed no effect (data not shown). Given that the core domain was sufficient to deubiquitinate AXIN1 in vitro (Fig. 2F), this suggests that other domains of USP34 are required in vivo, possibly to control the subcellular localization of USP34 or to regulate its activity. Since it consistently yielded the best knockdown we therefore carried all subsequent experiments with siRNA A. HEK293T and RKO cell lines stably expressing a β-catenin luciferase reporter and a Renilla luciferase control protein were then transfected with control, β-CATENIN, or USP34 siRNAs. In the control-transfected HEK293T and RKO cells, addition of Wnt3A led to 29- and 28-fold activations of the reporter, respectively (Fig. 4B, lanes 4 and 10) compared to cells treated with control conditioned medium. USP34-depleted cells showed a reduction in Wnt3A-mediated activation to 5.9- and 10.9-fold in the HEK293T and RKO cells, respectively (Fig. 4B, lanes 6 and 12). The impact of USP34 depletion was comparable, albeit less dramatic, to the depletion of β-CATENIN (Fig. 4B, lanes 5 and 11). We therefore conclude that USP34 acts as a positive regulator of Wnt signaling.
To functionally position USP34 within the Wnt pathway we next performed epistasis experiments where we tested the ability of the USP34 siRNA to block pathway activation at different levels. Strikingly, USP34 depletion inhibited the β-catenin reporter activity driven by the ectopic expression of a degradation-resistant form of β-catenin (Fig. 4C, compare lanes 4 and 2), as well as by Dishevelled (data not shown), but not by the constitutively activated chimeric VP16-LEF1 protein, a fusion protein between the activation domain of VP16 and LEF1 (19) known to be insensitive to β-catenin transactivation properties (Fig. 4C, lanes 6 to 8). These results position the function of USP34 downstream of the β-catenin stabilization step and argue that USP34 activity is important for the full activation of target genes. If USP34 functions downstream of the destruction complex, its knockdown should not influence the stabilization of β-catenin in response to Wnt pathway activation. To test this prediction, we used RKO cells, which lack β-catenin at adherent junctions. Under resting conditions these cells have minimal amount of cytosolic β-catenin whose levels can be strongly induced by Wnt3A conditioned medium (Fig. 4D). We performed a time course experiment of Wnt3A-induced β-catenin stabilization in control or USP34 siRNA-treated cells and observed that both the kinetics and the magnitude of β-catenin stabilization were unchanged (Fig. 4D, middle panels). Importantly, we show that USP34 was efficiently knocked down in USP34 siRNA-treated cells (Fig. 4D, upper panels).
To further support the site of USP34 action downstream of the destruction complex, we tested the effect of USP34 depletion on the constitutive Wnt signaling observed in SW480 and HCT116 colon cancer cells. These two cell lines were chosen since they harbor inactivating APC and activating β-catenin mutations, respectively. To monitor β-catenin-dependent transcription, the cell lines were transduced with lentivirus coding for the β-catenin responsive luciferase reporter TopFlash and for Renilla luciferase under the control of the constitutive EF1α promoter as a normalization probe. Since the Wnt pathway is strongly and constitutively activated in these cells (39, 55), a high ratio of firefly/Renilla luciferase activity was predictably observed (Fig. 4E, lanes 1 and 4). The constitutive reporter activity was β-catenin dependent since β-catenin knockdown virtually eliminated the TopFlash signal (Fig. 4E, lanes 3 and 6). USP34 knockdown also inhibited β-catenin signaling, indicating that it is also required in this context (Fig. 4E, lanes 2 and 5). To confirm whether the results obtained using the synthetic TopFlash reporter could be applicable to bona fide Wnt target genes, we examined the impact of USP34 depletion on the transcript levels of NAKED1 and TNFRSF19, two genes strongly regulated by β-catenin in colon cancer cells (35). Knockdown of USP34 in SW480 cells reduced the steady-state levels of the NAKED1 and TNFRSF19 transcripts by 38 and 56%, respectively (Fig. 4F, compare lanes 2 and 5 versus lanes 1 and 4). We conclude that USP34 is required at a step subsequent to β-catenin stabilization.
Axin has previously been shown to localize to the nuclei of colon cancer cells (1), to undergo nucleocytoplasmic shuttling (11, 57), and to translocate to the nucleus after Wnt stimulation in normal cells (61). Furthermore, we noted in a recent whole-genome siRNA screen of the Wnt pathway that depletion of the two AXIN genes in colon cancer cells with stabilized β-catenin consistently led to the inhibition of β-catenin-mediated transcription (35). These results argue that axin, in addition to its well-described function as a negative regulator of the pathway, may be required downstream of the destruction complex to fulfill positive regulatory roles during pathway activation. We first confirmed these results by depleting AXIN1 and AXIN2 transcripts in SW480 and HCT116 by siRNA. As suggested by Major et al. (35), we found that independent knockdown of AXIN1 and AXIN2 using established siRNAs (34) inhibited Wnt reporter activity in both cell lines (data not shown). When transfected together, AXIN1 and AXIN2 siRNAs reduced the constitutive activity of the reporter to 64 and 30% of control siRNA-treated HCT116 and SW480 cells, respectively (Fig. 5A, lanes 2 and 5). This is in sharp contrast to what is usually observed in cells presumed to have intact Wnt pathway components, such as human HEK293 kidney cells or RKO colon carcinoma cells. In a resting state, these cells exhibit a low level of spontaneous activity of the β-catenin reporter that can be strongly induced by treatment with Wnt3A conditioned medium (Fig. 5B, lanes 2 and 5). In this context, however, and as expected from the known negative role of axin proteins in Wnt signal transduction, the reduction of function for AXIN1 and AXIN2 potentiated the Wnt3A-mediated activation (Fig. 5B, compare lanes 3 and 6 to lanes 2 and 5). However, when a degradation-resistant mutant of β-catenin is introduced in HEK293T cells to constitutively activate the pathway, a condition mimicking the mutated state of the signaling cascade in HCT116 colon cancer cells, the activity of the TopFlash reporter was antagonized by AXIN knockdown (Fig. 5C, compare lanes 4 and 2).
Since activation of the Wnt pathway in SW480 and HCT116 cells results from β-catenin escaping its normal regulation by the destruction complex, the results presented above suggest that axin performs a positive regulatory function downstream of β-catenin stabilization when the destruction complex is disassembled. This prompted us to examine the subcellular localization of axin in these cells with an anti-AXIN1 peptide antibody by indirect immunofluorescence. Consistent with previous reports (1), axin is localized mostly to the nuclei of these cells (Fig. 6A, left panels). Pretreatment with AXIN1 siRNA largely eliminated the observed immunoreactivity, confirming the specificity of the AXIN1 antibody (Fig. 6A, middle panels). Treatment of SW480 and HCT116 cells with USP34 siRNA established that the accumulation of axin in the nuclei of colon cancer cells depends on USP34 (Fig. 6A). Treatment of the cells with MG132 could rescue the nuclear localiza tion of axin in USP34 siRNA-treated cells. Together with the results described above, this supports a role for USP34 in controlling axin stability and exclude that the reduction of nuclear axin could be due to an effect on nuclear import and/or export.
At steady state, in cells with normal Wnt signaling, axin is mostly found in the cytoplasm but undergoes nucleocytoplasmic shuttling as it accumulates in the nucleus when cells are treated with the CRM1-dependent nuclear export inhibitor LMB (11, 57). We thus sought to determine whether the stabilization of axin by USP34 is required to observe the accumulation of axin in the nucleus after LMB treatment. As shown in Fig. 6B, whereas axin is predominantly nuclear in control-transfected HEK293T cells incubated with LMB, axin does not accumulate in the nuclei of USP34-depleted cells. These results indicate that the activity of USP34 is important for the nuclear accumulation of axin.
We conclude from these observations that in transformed colon cancer cells, where Wnt signaling is constitutively activated, axin plays a positive regulatory role. In two genetically different colon cancer cells exhibiting defective Wnt signaling and in LMB-treated HEK293T cells where the Wnt pathway is normal, axin exhibits a USP34-dependent nuclear accumulation. The simplest explanation for these results is that the inhibition of Wnt signaling resulting from USP34 depletion (Fig. 4) is a consequence of axin destabilization and impaired nuclear accumulation.
The ubiquitin-proteasome system controls multiple steps in Wnt signaling through the regulation of protein stability or function. Recently, axin has been shown to be parsylated by tankyrase, a step required for its ubiquitin-dependent degradation (20). Although the precise mechanisms controlling axin levels are still unknown, we highlighted here the importance of axin as a key regulatory node in Wnt signaling. Our findings add to the mechanism controlling axin levels by identifying USP34 as the ubiquitin-specific protease opposing the tankyrase-dependent ubiquitination of axin and show that this regulation is important for the nuclear accumulation of axin during Wnt signaling to positively influence β-catenin-mediated transcription.
Our data agree with those of other researchers to show that axin levels are dynamically regulated by the ubiquitin-proteasome system (20, 23). Recently, the tankyrase-mediated parsylation of axin was showed to be required for its ubiquitination and degradation. The E3 ubiquitin ligase catalyzing axin ubiquitination remains to be identified, but our findings suggest that USP34 counteracts this reaction. Recently, the SUMOylation of axin on a C-terminal domain and its phosphorylation by GSK3 were shown to protect axin from ubiquitination (23). Although the precise functional interplay between these processes and USP34 needs to be studied, one possibility is that these signals recruit USP34 to axin to promote its deubiquitination. Another important question to be addressed is the understanding of the cellular signals that control USP34 and ultimately axin ubiquitination and stability. For example more work is now needed to determine whether the catalytic activity of USP34 is regulated during Wnt signaling and/or whether axin is recruited to USP34 following the disassembly of the destruction complex.
Although the precise mechanistic details remain unclear, our study suggests that axin plays a positive role in the nucleus for the transmission of Wnt/β-catenin signaling. In cells with normal Wnt pathways, axin undergoes nuclear-cytoplasmic shuttling (11, 57), a process suggested to be important for the nuclear export of β-catenin in the absence of Wnt signaling. However, the precise function of the nuclear pool of axin and the regulatory mechanisms influencing this localization are still poorly defined. Our study contributes to the understanding of this process by unraveling a positive regulatory role of axin in the nucleus during signaling and identifying USP34 as a protein influencing its stability. The positive signaling role for nuclear axin has eluded the numerous screens and studies performed on the Wnt pathway. It is likely that, under normal circumstances, this positive function is masked by the strong negative regulatory task of axin within the destruction complex. Supporting this, the positive roles of GSK3 and axin at the plasma membrane when they are recruited to the Wnt coreceptor LRP6 have also escaped these screens. Similarly, the appreciation that APC has dual positive and negative regulatory functions for Wg signaling has only been uncovered recently using a repressor screen in Drosophila (52). Alternatively, it is possible that in lower organisms where the majority of the pioneer screens have been performed, axin does not perform nuclear functions. The use of two colon cancer cell lines where the destruction complex machinery is defective and where axin accumulates in the nucleus has allowed us to uncover this novel function. Further work is now needed to address how axin exert this positive role in the nucleus. Two possibilities are that axin serves as escort protein for β-catenin and influences its residency in the nucleus or that axin is an integral part of the β-catenin transcriptional machinery participating in the recognition of target genes as recently found for Dsh (15). In any case, axin is likely not absolutely required for β-catenin signaling but rather could determine the duration and the strength of signaling by regulating the availability of activated nuclear β-catenin. What emerges from our study, however, is that the tight regulation of the balance of axin ubiquitination/deubiquitination is likely an important control point during Wnt signaling.
Since the nuclear localization of axin and β-catenin are associated with Wnt pathway activation and are constitutively found in the nucleus in several human cancers, the control of their residency in the nucleus by modulating USP34 activity could represent a novel therapeutic approach for not only the treatment of cancers but of the numerous human diseases exhibiting spurious Wnt/β-catenin pathway activity. Current strategies for small molecule inhibitors of the Wnt pathway have relied, with limited success, on compounds inhibiting protein-protein interactions (14, 28). The protease activity of USPs, however, makes them highly amenable to inhibition by small molecules (13). That promoting axin stability or degradation both lead to inhibition of Wnt signaling suggests that the precise control of axin levels dictates the outcome of signaling but also indicates that the ubiquitin-proteasome system may be adjusting the availability of different pools of axin underlying its negative function within the destruction complex and its positive role in the nucleus.
We are grateful to Daniel Durocher, Ben Nicholson, Michael Mattern, and the members of the Angers lab for providing comments on the manuscript. We also thank Anne-Claude Gingras for her expertise and David Sterner and Xufan Tian for cloning, expressing, and purifying USP34.
S.A. holds the Canada Research Chair in Functional Architecture of Signal Transduction. This study was supported by a grant from the Canadian Institute for Health Research (CIHR 171619) to S.A. Work at Progenra, Inc., is supported by NIH grants HL083527, CA115205, DK071391, and CA132246.
Published ahead of print on 7 March 2011.