Abstract
Cross-talk between the Hippo and Wnt pathways has been implicated recently in breast cancer development, but key intersections have yet to be fully defined. Here we report that WBP2, a transcription coactivator that binds the Hippo pathway transcription factor YAP/TAZ, contributes to Wnt signaling and breast cancer pathogenesis. Clinically, overexpression of WBP2 in breast cancer specimens correlated with malignant progression and poor patient survival. In breast cancer cells, nuclear entry and interaction of WBP2 with β-catenin was stimulated by Wnt3A, thereby activating TCF-mediated transcription and driving malignant invasive character. Mechanistic investigations showed WBP2 levels were controlled by the E3 ligase ITCH, which bound and target WBP2 for ubiquitin-dependent proteasomal degradation. Accordingly, ITCH silencing could elevate WBP2 levels. Wnt signaling upregulated WBP2 by disrupting ITCH–WBP2 interactions via EGFR-mediated tyrosine phosphorylation of WBP2 and TAZ/YAP competitive binding. Conversely, ITCH-mediated downregulation of WBP2 inhibited TCF/β-catenin transcription, in vitro transformation, and in vivo tumorigenesis. We identified somatic mutations in ITCH, which impaired its ability to degrade WBP2 and to block its function in cancer, even while retaining binding capacity to WBP2. Thus, the Wnt pathway appeared to engage WBP2 primarily by affecting its protein stability. Our findings show how WBP2/ITCH signaling functions to link the intricate Wnt and Hippo signaling networks in breast cancer. Cancer Res; 76(21); 6278–89. ©2016 AACR.
Introduction
The canonical Wnt signaling pathway is important to embryonic development and tissue homeostasis (1). Hyperactivated Wnt signaling is common in colon cancer and contributes to tumorigenesis via overexpression of constitutively active β-catenin or loss of adenomatous polyposis coli function (2). Aberrant Wnt signaling is also involved in other cancers (1), for example, breast cancer with high prevalence in triple-negative breast cancer (TNBC; ref. 3). When Wnt ligand is absent, β-catenin is phosphorylated by GSK3, targeted for β-TrCP–dependent ubiquitination and destruction in the cytoplasm. Upon Wnt stimulation, the GSK3 destruction complex is destabilized, leading to β-catenin stabilization and translocation into nucleus, where it coactivates transcription of target genes by TCF/LEF (4).
The Hippo pathway controls organ size and tissue homeostasis (5). The core mammalian Hippo pathway comprises of MST and LATS serine/threonine kinases. Together with their regulatory proteins, SAV and MOB, the activity of MST/LATS kinases results in the inhibition of the YAP and TAZ oncogenes, which activate the TEAD family of transcription factors in the nucleus. Dysregulation of Hippo pathway, for example, overexpression of YAP/TAZ or inactivating mutation of Hippo components triggers transformation of mammalian cells and promotes tumorigenesis in transgenic mouse models (6).
Recent studies have generated new insights into canonical Wnt signaling. Cross-talks between the Hippo–Wnt signaling pathways have been demonstrated (7–13). Hippo and Wnt exert antagonistic effect on growth through the regulation of common components like YAP and TAZ. For example, phosphorylated YAP/TAZ (by LATS) binds/sequesters β-catenin in cytoplasm, whereas dephosphorylated YAP/TAZ promotes β-catenin nuclear translocation (7, 13). Recently, YAP/TAZ were found to be essential for the formation of the β-catenin destruction complex in the absence of Wnt ligand (10). Separately, YAP/TAZ were found to constitute an alternative Wnt signaling pathway that activates TEAD following Wnt ligand treatment (14). Other mechanisms of Wnt regulation include inhibition of DVL phosphorylation by binding of TAZ/YAP to DVL (9) and degradation of phosphorylated DVL via ITCH (15).
WW domain-binding protein 2 (WBP2) was cloned as a cognate ligand of WW domain of YAP (16). It was shown to bind to PAX 8 thyroid-specific transcription factor but had no effect on its activity (17). In contrast, WBP2's binding to E6AP was required for estrogen and progesterone receptor coactivation (18). WBP2 is a substrate of EGFR signaling and is found to be associated with breast cancer progression (19). WBP2 cooperated with Drosophila YAP (Yki) and TAZ to drive tissue growth in Drosophila (20) and breast cancer (21), respectively. Overexpression of WBP2 phosphomimetic mutant transformed normal breast mammary epithelial cells, conferred aggressive traits to breast cancer cells, and activated the TCF transcription pathway (22). Recently, WBP2 was shown to promote ER transcription by binding to the phosphorylated form of RNA polymerase II and the histone acetyl transferase p300 (23). These evidences highlight that the transcription activation properties of WBP2 are important to Wnt signaling, tissue growth, and cancer.
However, the role of WBP2 in clinical breast cancer is unknown. Furthermore, how Wnt signaling engages and activates WBP2 is not understood. In this study, we identified ITCH E3 ligase as a negative regulator of WBP2. The effects of Wnt signaling and YAP/TAZ on ITCH–WBP2 interaction were investigated. Wnt signaling, YAP, and TAZ were found to promote WBP2 function by stabilizing it against ITCH-mediated proteasomal degradation. Tyrosine phosphorylation of WBP2 is a key control in this process.
Materials and Methods
Reagents
In-house WBP2 polyclonal antibody was generated as described previously (22); WBP2 mAb (MABS441-clone 4C8H10) was from EMD Millipore. Anti-PY20-HRP, anti-ITCH, and anti-β-catenin mouse mAbs were from BD Biosciences. Sources for other reagents are mentioned in Supplementary Information. V5-WBP2 and its phospho-mutants were previously generated as described previously (22). All other mutations of WBP2 and ITCH were generated using the QuikChange Lightning Site-Directed Mutagenesis Kit (Stratagene, Agilent Technologies). Other plasmids constructed, obtained as gifts, or purchased from Addgene together with siRNAs/shRNA constructs are described in Supplementary Information.
Cell culture and ligand/drug treatments
Cell lines used and transfection/lentiviral transduction protocols are described in details in Supplementary Information. Unless otherwise stated, cell lines were obtained between years 2005 and 2011 from ATCC, which authenticates via short tandem repeat profiling. They are expanded upon receipt, frozen in multiple vials, and maintained by passaging not exceeding one month. For experiments involving WNT3A-CM or rWNT3A (R&D Systems), cells were starved in serum-free medium for 16–24 hours before stimulation. For WNT3A/gefitinib treatment, cells were pretreated with gefitinib for 1 hour before WNT3A stimulation.
Protein–protein interaction studies and validation
Yeast two-hybrid screening against human breast tumor epithelial cells cDNA library was performed by Hybrigenics Services, S.A.S. Detailed screening and ranking/scoring process are described in Supplementary Information. Expression, purification, use of GST-WBP2 for pulldown assays, sample preparation for iTRAQ labeling, and mass spectrometry are described in Supplementary Information. GST control and GST–WBP2 complexes were labeled with iTRAQ 114 and 115 tags, respectively.
Subcellular fractionation
Nuclear and cytoplasmic extracts were prepared using the NE-PER extraction kit (Pierce) or Nuclear Complex Co-IP Kit (Active Motif), as per manufacturer's instructions.
Dual luciferase reporter assay
The Dual-Luciferase reporter assay (Promega) was performed according to the manufacturer's instructions and quantified using Luminoskan Ascent Microplate Luminometer (Thermo Scientific). Firefly signals were normalized to Renilla signals.
Gene expression analysis
Total RNA was isolated using PureLink RNA Mini Kit (Thermo Scientific) as per manufacturer's instructions. Transcript abundance was assayed by qRT-PCR using QuantiFast Probe PCR Kit (Qiagen) as recommended. Predesigned primers/probes were obtained from IDT and were available upon request.
In vivo ubiquitination assay
The in vivo ubiquitination assays were carried out as described previously (24, 25) on cells harvested following 36 hours post-transfection and treatment with 20 μmol/L MG132.
In vitro cell-based assays
The anchorage-independent growth was assayed using either CytoSelect 96-Well Cell Transformation Colorimetric Assay Kit (Cell Biolabs), as per manufacturer's instructions or protocol as described previously (26). Cell invasion was assayed using CytoSelect 96-Well Cell Invasion (BME) Fluorometric Assay Kit, as per manufacturer's instructions.
Immunofluorescence, immunoprecipitation, and immunoblotting
Subcellular localization of WBP2 and other proteins in cells and tissues were analyzed by immunofluorescence and confocal microscopy as described previously (22) with some modifications provided in Supplementary Information.
For immunoprecipitation, 500–1,000 μg of lysates were incubated with 1–5 μg antibodies overnight at 4°C. Immune complexes were captured by 50-μL Dynabeads Protein-G (Invitrogen) for 15–20 minutes at room temperature. Washing and elution of immunoprecipitates were performed as described previously (22).
Cell lysis and Western blot analyses were done as described in Supplementary Information.
Immunohistochemistry
Approvals were granted by the Institutional Review Boards of National University Health System and Singapore General Hospital for use of individual and tissue microarray (TMA) blocks. Tissue microarrays from US Genomax were also used. The details of the samples used, sectioning, IHC protocol, scoring, and statistical correlation studies are described in Supplementary Information.
Xenograft studies
Five-week-old female nude mice (In Vivos) were inoculated subcutaneously into the flanks with HeLa (2 × 106 in 200 μL of DPBS and Matrigel 1:1) or orthotopically into the mammary fat pad with BT-549 [1 × 107 in 200 μL of DPBS and Cultrex Basement Membrane Extract (Trevigen) 1:1] or MDA-MB-231 (5 × 106 in 200 μL of DPBS and Matrigel 1:1). Tumor development was monitored and tumor volumes calculated as (width2 × length)/2. The data represent mean ± SEM. Statistical significance was determined by Mann–Whitney test.
Statistical analyses
Differences among groups and treatments for all in vitro experiments (e.g. luciferase assays, qPCR, cell-based assays) were determined by Student t test. The indicated significance values correspond to <0.05 (∗), <0.01 (∗∗), and <0.001 (∗∗∗).
Results
WBP2 is overexpressed in breast cancer
The expression of WBP2 in cancer cells has never been examined. Analysis of the isogenic model of breast cancer progression (27) showed that WBP2 expression was low in normal cells (10A1), moderate in preneoplastic (AT1k) and low-grade (CA1h), but elevated in high-grade (CA1a) breast cancer cells (Fig. 1A). WBP2 expression was also low/undetectable in normal but overexpressed in 11 of 14 breast cancer cell lines (Fig. 1B, i). Quantitative PCR revealed about 50% concordance between WBP2 transcript and protein levels, suggesting that WBP2 is regulated post-transcriptionally (Fig. 1B, ii).
Assessment of >400 clinical specimens revealed that the majority of normal (N), benign, hyperplasia, and ductal carcinoma in situ (DCIS) had low/undetectable nuclear WBP2, while 80% or more of the invasive ductal carcinoma (IDC) and metastatic (Mets) lesions had moderate to high nuclear WBP2 (Fig. 1C, i). Box plot analysis showed that WBP2 expression was significantly elevated, as early as in the DCIS preneoplastic stage, and increased further in IDC and Mets (Fig. 1C, ii). The IHC signals generated were specific to WBP2 (Supplementary Fig. S1A). Elevation of cytosolic WBP2 was also observed in cancer cells but less striking than nuclear WBP2 (Supplementary Fig. S1B). Representative IHC images of WBP2 expression in clinical breast tissues are shown in Fig. 1D, i. Confocal microscopy confirmed the increased nuclear WBP2 in invasive ductal carcinoma (IDC) lesion as compared with the matched normal tissues. As shown in Fig. 1D, ii and consistent with the IHC data, confocal microscopy demonstrated that IDC tissues had more nuclear WBP2 compared with matched normal tissues.
Analysis of WBP2 in the progressive lesions within the same individual revealed that majority of them had nuclear WBP2 expression trend where IDC = DCIS > N (75%) followed by IDC > DCIS = N (25%; Fig. 1E). Where there was no DCIS, 79% had IDC > N. Nuclear WBP2 expression correlated positively with tumor size and grade but inversely with overall and disease-free survival (Fig. 1F). Cytoplasmic WBP2 was less well correlated with tumor size (P = 0.052, n = 309), grade (P = 0.019, n = 311), overall (P = 0.009, n = 308), and disease-free survival (P = 0.437, n = 305). The data suggest that WBP2 is required for cancer initiation and nuclear WBP2 is associated with progression and poor clinical outcomes.
WBP2 plays a role in WNT3A-mediated breast cancer and signaling
Previous overexpression studies supported an oncogenic role of WBP2. To augment WBP2 as a bona fide oncogene, we performed xenograft studies using MDA-MB-231 stably knocked-down for WBP2. As shown in the Fig. 2A, i, the in vivo tumor growth of MDA-MB-231 was significantly diminished when WBP2 was silenced. Next, we investigated the role of WBP2 in breast cancer in the context of Wnt signaling using MDA-MB-231 as it is highly responsive to Wnt ligand. Compared with control, WBP2 knocked-down MDA-MB-231 displayed drastically decreased WNT3A-induced anchorage-independent growth (>75%), invasion (>70%; Fig. 2A, ii), TCF activity, and AXIN2 expression (Fig. 2B, i). The partial downregulation of AXIN2 expression upon WBP2 knockdown (KD) could be due to the fact that AXIN2 is regulated by other regulators, for example, E2F1 (28), Barx2, and Pax7 (29). The role of WBP2 in Wnt signaling was further supported when WBP2 re-expression rescued the observed phenotype in MDA-MB-468 cells (Fig. 2B, ii).
Does WBP2 cooperate with β-catenin to drive TCF activity? Overexpression of WBP2 and β-catenin resulted in synergistic activation of WNT3A-induced TCF and AXIN2-promoter reporter activities compared with overexpression of either alone (Fig. 2C, i). Furthermore, WNT3A induced nuclear colocalization of WBP2 with β-catenin in MCF7 cells (Fig. 2C, ii). Treatment of HeLa and MDA-MB-468 cells with WNT3A for 24 hours and EGF for 1 hour also resulted in nuclear entry of WBP2 (Supplementary Fig. S1C). WBP2 coimmunoprecipitated with β-catenin in both HeLa (Fig. 2C, iii) and MDA-MB-231 (Supplementary Fig. S1D). Thus, WNT3A enhanced TCF activity by promoting nuclear entry and association of WBP2 with β-catenin.
Coexpression studies of nuclear β-catenin (indicative of active Wnt signaling) and WBP2 in 57 IDC samples revealed that nuclear WBP2 expression was significantly associated with active Wnt status (P = 0.006) with a correlation coefficient of 0.36 (Fig. 2D, i and ii). Representative IHC images showing coexistence of nuclear β-catenin and WBP2 are provided (Fig. 2D, iii). This demonstrates that active Wnt/WBP2 signaling axis exists in a subset of clinical breast cancers.
ITCH targets WBP2 for proteasomal degradation
To identify novel regulators of WBP2 oncogene, yeast two-hybrid and pull-down assays were performed. ITCH was identified as the top hit in both methods (Fig. 3A, i and ii). Endogenous WBP2–ITCH interaction in HeLa and MDA-MB-231 as well as in HeLa cells expressing exogenous WBP2 and ITCH could be demonstrated (Fig. 3A, iii). As ITCH is an E3 ligase, we postulated that ITCH negatively regulates WBP2 expression. Indeed, endogenous and exogenous WBP2 expression decreased significantly in the presence of WT but not ligase-dead ITCH mutant (C830A; Fig. 3B). ITCH did not alter the expression of other E3 substrates such as YAP, TAZ, β-catenin, and Runx2, which, like WBP2, contains PPxY motif (16) demonstrating specificity (Supplementary Fig. S1E).
Treatment with proteasome, lysosomal, and protease inhibitors revealed that proteasome inhibitors significantly prevented ITCH-mediated WBP2 downregulation (Fig. 3C, i). To confirm that WBP2 is a substrate of ITCH, cells were first pretreated with MG132 to prevent WBP2 degradation before ubiquitinated proteins were immunoprecipitated and probed for WBP2. Increased polyubiquitinated WBP2 was observed in the presence of WT but not ligase-dead ITCH (Fig. 3C, ii).
UbPred software (30) predicted 5 lysine residues (K36, K97, K222, K258, K259) as potential ubiquitination sites on WBP2. As shown in Fig. 3C, iii, K222R mutation abolished ITCH-mediated degradation substantially albeit incompletely. We conclude K222 as a putative site of ubiquitination on WBP2 although other site/s might also be targeted. This is not inconceivable as multiple ubiquitination sites on target proteins have been reported (31, 32). Coexpression of WBP2 with either K48R or K63R ubiquitin mutant in the presence of ITCH-WT or -C830A mutant shows that K48R, but not K63R mutant, could rescue WBP2 from degradation by ITCH-WT to almost the same level of WBP2 in the presence of C830A ligase–dead mutant (Fig. 3C, iv). Polyubiquitination of WBP2 was also substantially impaired in the presence of K48R ubiquitin mutant. The collective data indicate that ITCH downregulates WBP2 via ubiquitination and proteasomal degradation.
WBP2 has three PPxY (PY) motifs, whereas ITCH has four WW domains. The WW domain and PPxY motifs are cognate elements of a canonical mode of protein-to-protein interaction (16). We hypothesize PY/WW domain interaction is required for WBP2 degradation by ITCH. Degradation of WBP2 by ITCH was completely abolished when both PY2 and PY3 motifs were mutated (Fig. 3D, i) and this was concomitant with reduced association of WBP2 with ITCH (Fig. 3D, ii). Mutation of ITCH WW1 or WW3 alone impaired the ability of ITCH to bind and degrade WBP2 but the effect was strongest with WW1/WW3 double mutant (Fig. 3D, iii), which failed to coimmunoprecipitate with WBP2 (Fig. 3D, iv).
To test the hypothesis that ITCH is a rheostat for endogenous WBP2, we examined the prevalence of proteasomal degradation and ITCH as determinants of WBP2 expression by analyzing WBP2 protein level upon MG132 treatment or ITCH KD in a panel of 15 breast cell lines and the MCF10AT model. A positive effect is defined by >50% increase in WBP2 expression following the treatments. Upregulation of WBP2 was observed in 40% (6/15) and 62% (8/13) of the breast cell lines upon MG132 and ITCH KD, respectively (Fig. 3E, i). Endogenous WBP2 level in four cell lines was regulated by both proteasome inhibition and ITCH KD (Fig. 3E, ii). Results for the rest of the cell lines are provided in Supplementary Fig. SF. For the MCF10AT model, only 10A1 showed significant stabilization of WBP2 upon MG132 treatment (Fig. 3F, i) while all but CA1a cell lines showed WBP2 stabilization upon ITCH KD (Fig. 3F, ii). Collectively, the data suggest an exciting possibility that the expression of WBP2 oncogene is normally tightly controlled to prevent aberrant growth. Although the MCF10AT model originated from MCF10A transfected with mutated H-RAS, some of the evidence argues against RAS having a role in WBP2 stabilization. Stabilization of WBP2 by ITCH silencing was observed in 10A (no exogenous H-RAS mutant), AT1K (expresses exogenous H-RAS mutant), and Ca1h (expresses exogenous H-RAS mutant), indicating that WBP2 stability is independent of the RAS mutation/activation status. Moreover, none of the breast cancer cell lines used possesses H-RAS mutation.
Collectively, the findings in Fig. 3E and F show that proteasomal degradation and especially ITCH are considerable factors that regulate WBP2 expression in normal and breast cancer cells. It is unclear why WBP2 in some cell lines was regulated upon ITCH KD but not proteasomal inhibition. According to Fig. 3C, i, proteasomal inhibitors did not fully rescue ITCH-mediated downregulation of WBP2. It is conceivable that WBP2 may be regulated by ITCH through other additional means besides proteasomal degradation.
WNT3A promotes stability of WBP2
As WBP2 positively contributes to Wnt signaling, we postulate that Wnt signaling disrupts the WBP2–ITCH interaction and negates the negative pressure exerted by ITCH. Indeed, WNT3A reduced endogenous ITCH–WBP2 interactions (Fig. 4A), demonstrating that WBP2–ITCH interaction is physiologically regulated and reversible. However, this corresponded to an increase in nuclear but not cytosolic WBP2 level. Repeating this experiment in three Wnt-responsive cell lines (HeLa, MDA-MB-231, BT549) confirmed that cytoplasmic WBP2 did not increase upon Wnt ligand treatment in the setting used (data not shown).
It is conceivable that the decrease in ITCH–WBP2 interaction did not lead to accumulation/increase of WBP2 in the cytoplasm under the influence of Wnt ligand because WBP2 is controlled concomitantly by other mechanisms, for example, steady state. Thus, we hypothesized that WNT3A and ITCH regulates the half-life of WBP2. As shown in the Fig. 4B, WBP2's protein half-life was significantly increased by WNT3A treatment as well as by ITCH KD. This indicates that ITCH-mediated negative regulation of WBP2 protein stability could be inhibited by WNT3A, possibly through decreased WBP2–ITCH interaction. We propose that Wnt3a-induced tyrosine phosphorylation of WBP2 coupled to consequent reduced WBP2–ITCH interaction in the cytoplasm contribute to an increase in the half-life of cytosolic WBP2, which is induced concomitantly to translocate into the nucleus by WNT3A as shown previously in Fig. 2C, ii and iii.
The phosphorylation sites of WBP2 have been mapped to Y192 and Y231 (22), which are in close proximity to the PY motif 2 (a.a.197–200) and 3 (a.a. 249–252) of WBP2 that are required for binding ITCH. It is conceivable that Wnt disrupts ITCH–WBP2 interaction by phosphorylating WBP2. As Wnt crosstalks with EGFR, we postulated that WNT3A destabilizes WBP2–ITCH interaction via EGFR-dependent tyrosine phosphorylation of WBP2. Indeed, WNT3A treatment, in the form of conditioned medium (Supplementary Fig. S1G) or recombinant protein (Fig. 4C, i) resulted in tyrosine phosphorylation of WBP2 that was abolished by gefitinib, a selective EGFR inhibitor. To investigate whether EGFR has a direct role in phosphorylating WBP2, we performed substrate trapping experiment instead of in vitro kinase assay, which is often promiscuous and nonspecific in nature. Figure 4C, ii shows that kinase-dead EGFR, but not the EGFR-WT, coimmunoprecipitated with WBP2, supporting the notion that WBP2 is a substrate for EGFR. We noted that overexpression of the kinase-dead EGFR resulted in destabilization of WBP2 protein compared with EGFR-WT. This further supports the notion that EGFR/tyrosine phosphorylation is required for WBP2 stability.
WBP2-WT was less susceptible to ITCH-mediated degradation in the presence of WNT3A compared with WBP2 phospho-defective mutant (Fig. 4D, i). Furthermore, WBP2-phosphomimic mutant displayed increased stability and reduced polyubiquitination in the presence of ITCH compared with WBP-WT (Fig. 4D, ii and iii). In both cases, the increased stability of WBP2 could be attributed to the diminished interaction between WBP2 and ITCH that appears to be negatively regulated by WBP2 phosphorylation (Fig. 4E, i–iii). Isothermal calorimetry (ITC) confirmed that ITCH binds WBP2 directly but showed no significant differences in the Kd (dissociation constant) between the phosphorylated and nonphosphorylated peptides of WBP2 with ITCH WW domain 3 (Supplementary Fig. S1H). While this implies that phosphorylation does not affect WW-PY binding between ITCH and WBP2, it does not rule out the potential effect of phosphorylation on the conformation of WBP2 that influences such an interaction.
As WBP2 binds to YAP and TAZ in a PY-WW–dependent manner and all three have been demonstrated to be positively engaged by Wnt signaling (28), we hypothesized that YAP and TAZ could compete with and relieve the inhibitory effect of ITCH on WBP2 thereby promoting TCF activity. This is supported by the demonstration that WNT3A induced nuclear colocalization of WBP2 with YAP and TAZ (Supplementary Fig. S1I) and the latter two are required for maximal WBP2-mediated TCF reporter activity (Supplementary Fig. S1J). Moreover, overexpression of YAP-WT or TAZ-WT rescued ITCH-mediated WBP2 degradation (Fig. 4F, i). The effect was more prominent with the constitutively-active YAP-S127A or TAZ-S89A but was abolished when WW domain mutants of YAP/TAZ were used. Loss of WBP2's interaction with ITCH was concomitant with a gain in association with YAP/TAZ (Fig. 4F, ii), supporting the protective effect of YAP/TAZ on WBP2. Quantitative mammalian-two-hybrid assay showed more association of WBP2 with YAP-S127A and TAZ-S89A compared with WT-TAZ/YAP (Fig. 4F, iii)–possibly explaining their stronger protective effects on WBP2 from ITCH. The stronger association of WBP2 with YAP than TAZ is likely due to two WW domains in YAP compared with 1 WW domain in TAZ.
ITCH somatic mutants are defective against the molecular and cancer cellular function of WBP2
Negative regulation of WBP2 by ITCH raised a testable hypothesis that ITCH inhibits Wnt signaling. Indeed, overexpression of WT, but not C830A-ITCH, significantly abolished WNT3A/WBP2-driven TCF activity (Fig. 5A, i). In contrast, KD of endogenous ITCH in MDA-MB-231 and HeLa increased WNT3A-induced TCF activity (Fig. 5A, ii and iii). This could be reversed by KD of endogenous WBP2, suggesting that ITCH regulates Wnt signaling via WBP2. Subcellular fractionation of ITCH-KD cells stabilized WBP2 in both cytoplasm and nucleus (Supplementary Fig. S1K). Hence, nuclear translocation and stabilization are molecular switches through which Wnt turns on WBP2.
According to COSMIC database, there are 9 ITCH somatic mutations in breast tumors. Two were found in 3.4% of HER2+ breast cancer and three in 7.9% of TNBC cases (Fig. 5B, i). Majority of them occur in the HECT domain, raising a possibility that these might be inactivating mutations. E184K, R833C, and especially E855K mutants had reduced ability to degrade WBP2 (Fig. 5B, ii). C830A and C830A/E855K mutants were capable of binding WBP2, indicating that mutations did not affect binding to WBP2 (Fig. 5B, iii). Polyubiquitination of WBP2 was reduced when ITCH-E855K mutant was coexpressed (Fig. 5B, iv). E855K mutant failed to abolish the WNT3A/WBP2-mediated TCF reporter activity (Fig. 5C).
Does ITCH-mediated regulation of WBP2 stability affects breast cancer? As expected, overexpression of WBP2-WT alone in HeLa [Fig. 5D, i, ii (top), iii, and iv] and BT549 (Fig. 5D, ii, bottom) promoted anchorage-independent colony and/or in vivo tumor growth while coexpression with ITCH-WT abolished these phenotypes. In contrast, silencing ITCH alone has a significant oncogenic effect on anchorage-independent growth that was reversed by WBP2 KD (Fig. 5E, i). Consistent with the 3D colony growth assay, ITCH KD promoted in vivo tumorigenesis to almost the same extent as WBP2 overexpression (Fig. 5E, ii and iii). Although not statistically significant, some of the tumors driven by ITCH KD and WBP2 overexpression are larger in size compared with tumors with ITCH KD or WBP2 overexpression alone (Fig. 5E, ii). It is possible that the oncogenic effect of ITCH KD or WBP2 overexpression may be saturated in the presence of each other. ITCH is largely tumor suppressing in the context of WBP2-driven cancer biology. Fig. 5F shows that the C830A ligase-dead and E855K somatic mutants, both incapable of downregulating WBP2, failed to reduce WBP2-driven transformation in vitro. This establishes that ITCH mutation has molecular and cellular functional phenotypes.
Discussion
Wnt pathway regulates WBP2 stability
WBP2 is a growth promoter and oncogene (20–22). Tyrosine phosphorylation drives WBP2 into the nucleus (22). Here, we show that WBP2 is overexpressed in breast cancer and nuclear WBP2 level correlates with tumor aggression and survival. Aberrant nuclear level of WBP2 was observed as early as the DCIS stage implicating WBP2 in cancer initiation. Wnt-induced nuclear localization of WBP2 was concomitant with enhanced TCF activity, cancer cell growth, and invasion. These evidences support the notion that nuclear localization of WBP2 contributes to its oncogenic property.
Overexpression of WBP2 in breast cancer may be due to gene amplification as it resides on chromosome 17q25, a region frequently rearranged in many breast cancer cell lines and tumors (33). Here, we show that WBP2 is additionally regulated at the post-translational level. ITCH E3 ligase appears to act like a “rheostat” for controlling the basal expression level of WBP2 via ubiquitination/proteasomal degradation. WNT3A, which engages WBP2 as a downstream activator of TCF/β-catenin transcription, prevents WBP2 degradation by inducing tyrosine phosphorylation of WBP2 and disrupting ITCH–WBP2 interaction. Our data established that ITCH/proteasome-mediated downregulation of WBP2 occurred in a considerable number of the cell lines tested. Thus, protein stability is a novel mode through which WBP2 is regulated by Wnt through coordinated actions of YAP, TAZ, and ITCH. This is consistent with a recent observation that mitotic Wnt signaling regulates cell size by promoting stabilization of key proteins including C-MYC (34). However, we do not rule out the possibility that the steady state of WBP2 could also be regulated by other mechanisms such as miRNAs as WBP2 level in a few cell lines did not change following ITCH KD or MG132 treatment. Future studies are expected to shed more light on the additional modes of regulation that are required for tight control of WBP2 oncogene expression.
ITCH—an oncogene or tumor suppressor?
ITCH has been reported to act as a tumor suppressor by negatively regulating Wnt signaling through binding and degrading phosphorylated Dishevelled thereby inhibiting TopFlash reporter activity (DVL; ref. 15). Our study also positions ITCH as a tumor suppressor that binds to and promotes WBP2 degradation, leading to inhibition of Wnt signaling. Yet, ITCH appears to have oncogenic property by inhibiting the Hippo signaling, by binding and degrading LATS, and enhancing breast cancer progression (35–37). While ITCH may be a “double-edged sword”, we argue that understanding ITCH function can be better attained under the context of signaling by agonists/antagonists as this constitutes a more physiologic approach in addition to the commonly used overexpression method. Our data suggest that WNT3A and EGF oncogenic signals shift the equilibrium in favor of oncogenic WBP2 by blocking ITCH–WBP2 interaction. Furthermore, ITCH E855K and R833C somatic mutations enriched in TNBC were deficient in degrading WBP2. Together, the data indicate ITCH as a tumor suppressor and its inactivating mutations are “oncogenic” in a subset of breast cancers.
Tyrosine phosphorylation is a key switch for WBP2
Through our previous and current studies, tyrosine phosphorylation of WBP2 is established as a major regulatory switch for WBP2 function. E2-induced tyrosine phosphorylation stimulated nuclear localization of WBP2 and promoted ER activity while Wnt-induced tyrosine phosphorylation of WBP2 disrupted ITCH/WBP2 binding, stabilizing WBP2 that subsequently cooperated with β-catenin to drive TCF-mediated transcription. Thus far, direct activation of EGFR by EGF, cross-talk with EGFR by estrogen, and Wnt ligands were responsible for tyrosine phosphorylation of WBP2. This implicates EGFR/WBP2 as a central integrator of a myriad of signaling inputs and functions.
In conclusion, a model on the reciprocal regulation of WBP2 by Wnt/Hippo pathways with EGFR as the central machinery for activating/phosphorylating WBP2 is proposed (Fig. 6).
Disclosure of Potential Conflicts of Interest
D.M. Virshup is a consultant/advisory board member for Experimental Therapeutics Centre, Singapore. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: S.K. Lim, S.Y. Lu, M. Sudol, W. Hong, Y.P. Lim
Development of methodology: S.K. Lim, S.Y. Lu, S.-A. Kang, J.S. Guan, Y.P. Lim
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.K. Lim, S.-A. Kang, H.J. Tan, Z. Li, Z.N.A. Wee, A.A. Thike, Y.P. Lim
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.K. Lim, S.Y. Lu, S.-A. Kang, H.J. Tan, Z.N.A. Wee, M. Sudol, D.M. Virshup, S.W. Chan, W. Hong, Y.P. Lim
Writing, review, and/or revision of the manuscript: S.K. Lim, S.Y. Lu, S.-A. Kang, M. Sudol, D.M. Virshup, S.W. Chan, Y.P. Lim
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.K. Lim, S.-A. Kang, Z.N.A. Wee, J.S. Guan, S.W. Chan, W. Hong, Y.P. Lim
Study supervision: Y.P. Lim
Other (isothermal titration calorimetry experiments and analysis): V.P.R. Chichili, J. Sivaraman
Other (provision and review of pathology materials): P.H. Tan
Grant Support
This study was supported by funding from the Biomedical Research Council (BMRC), Agency for Science Technology and Research (A*STAR) to Y.P. Lim.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.