Research ArticleCancer

Convergence of Wnt, growth factor, and heterotrimeric G protein signals on the guanine nucleotide exchange factor Daple

See allHide authors and affiliations

Science Signaling  27 Feb 2018:
Vol. 11, Issue 519, eaao4220
DOI: 10.1126/scisignal.aao4220

Growth factor and Wnt pathways cross-talk on Daple

Many proteins that maintain tissue homeostasis are conversely implicated in tumor progression. What triggers this switch? The guanine nucleotide exchange factor Daple, which coordinates Wnt and G protein signals, acts as a tumor suppressor in the normal epithelium and early-stage tumors but facilitates metastatic progression in advanced tumors. Aznar et al. found that growth factor receptor activation, frequently observed in many cancers, phosphorylated a critical protein interaction motif in Daple that enhanced its binding to G proteins rather than to a Wnt receptor inhibitor, thereby stimulating ligand-independent Wnt signaling. Supported by protein signatures in colorectal tumors from patients, these findings suggest that concurrent activation of Wnt and growth factor receptor pathways fuels a Daple-mediated switch to cancer progression.

Abstract

Cellular proliferation, differentiation, and morphogenesis are shaped by multiple signaling cascades, and their dysregulation plays an integral role in cancer progression. Three cascades that contribute to oncogenic potential are those mediated by Wnt proteins and the receptor Frizzled (FZD), growth factor receptor tyrosine kinases (RTKs), and heterotrimeric G proteins and associated GPCRs. Daple is a guanine nucleotide exchange factor (GEF) for the G protein Gαi. Daple also binds to FZD and the Wnt/FZD mediator Dishevelled (Dvl), and it enhances β-catenin–independent Wnt signaling in response to Wnt5a-FZD7 signaling. We identified Daple as a substrate of multiple RTKs and non-RTKs and, hence, as a point of convergence for the three cascades. We found that phosphorylation near the Dvl-binding motif in Daple by both RTKs and non-RTKs caused Daple/Dvl complex dissociation and augmented the ability of Daple to bind to and activate Gαi, which potentiated β-catenin–independent Wnt signals and stimulated epithelial-mesenchymal transition (EMT) similarly to Wnt5a/FZD7 signaling. Although Daple acts as a tumor suppressor in the healthy colon, the concurrent increased abundance of Daple and epidermal growth factor receptor (EGFR) in colorectal tumors was associated with poor patient prognosis. Thus, the Daple-dependent activation of Gαi and the Daple-dependent enhancement of β-catenin–independent Wnt signals are not only stimulated by Wnt5a/FZD7 to suppress tumorigenesis but also hijacked by growth factor–activated RTKs to enhance tumor progression. These findings identify a cross-talk paradigm among growth factor RTKs, heterotrimeric G proteins, and the Wnt/FZD pathway in cancer.

INTRODUCTION

Molecular characterization of tumors has revealed that multiple signaling pathways are often simultaneously dysregulated in cancer cells. Although each of these pathways is often conceptualized as an independent entity, their complex cross-talk shapes proliferation, invasion, immune evasion, chemoresistance, and stemness (17), which promote tumor development and progression. Among these cross-talking pathways include Wnt/Frizzled (FZD) signaling, heterotrimeric G proteins (heterotrimeric GTP-binding proteins) and G protein–coupled receptor (GPCR) signaling, and growth factor receptor tyrosine kinase (RTK) signaling cascades. For example, aberrant activation of β-catenin–dependent Wnt signals [the so-called canonical β-catenin–T cell factor/lymphoid enhancer factor (TCF/LEF) transcriptional program] secondary to gain-of-function mutations in genes encoding adenomatous polyposis coli, axin, and β-catenin is associated with the development of colon cancer, desmoid tumors, gastric cancer, hepatocellular carcinoma, medulloblastoma, melanoma, ovarian cancer, pancreatic cancer, and prostate cancer [reviewed in (1)]. However, these mutations alone do not account for the observed β-catenin hyperactivity; instead, it is the cross-talk between the growth factor RTK and the β-catenin–dependent Wnt/FZD pathways that synergistically potentiate the β-catenin–dependent transcriptional program through distinct mechanisms [reviewed in (2, 814)]. These mechanisms underscore the importance of concurrent aberrant signaling triggered by sequential genetic or epigenetic “hits”; in solid tumors, aberrations in as few as three driver genes or pathways appear to suffice for a cell to evolve into an advanced cancer (15).

Although the elaborate cross-talk between growth factors and the Wnt/β-catenin–dependent signaling pathway is well documented, little is known about how growth factors affect β-catenin–independent (also referred to as “noncanonical”) Wnt signaling. The β-catenin–independent Wnt pathway behaves as a double-edged sword; it not only suppresses tumorigenesis in normal epithelium and in early tumors but also serves as a critical driver of epithelial-mesenchymal transition (EMT) and cancer invasion (1626). We recently identified a Wnt signaling paradigm in which FZD receptors activate G proteins and stimulate β-catenin–independent Wnt signaling through the guanine nucleotide exchange factor (GEF) Daple (encoded by CCDC88C), which is a Dishevelled (Dvl)–binding protein (27). Daple directly binds Wnt5a-activated FZD receptors and activates the G protein Gαi (Fig. 1A). Upon ligand stimulation, Daple dissociates from Dvl, binds to FZD receptors, displaces Dvl (Fig. 1B), and assembles Daple/Gαi complexes (Fig. 1C). Disassembly of Daple/Dvl complexes and formation of FZD/Daple/Gαi complexes facilitate the activation of trimeric Gαi near ligand-activated FZD receptors. Activation of Gαi by Daple suppresses adenosine 3′,5′-monophosphate (cAMP), whereas released “free” Gβγ heterodimers enhance Rac1 and PI3K (phosphatidylinositol 3-kinase)–AKT [protein kinase B (also known as PKB)] signaling (Fig. 1A). We and others have shown that Daple-dependent enhancement of noncanonical Wnt signals not only can suppress tumor growth (27) but also can fuel EMT, facilitate cell migration and invasion (27, 28), and promote metastasis (17). Furthermore, increased abundance of Daple in circulating tumor cells is predictive of a poor outcome (29). Thus, Daple behaves like a double-edged sword—it is a tumor suppressor in normal tissue during the early stages of oncogenesis, but it promotes metastatic invasion in later stages.

Fig. 1 The GBA motif and PBM in Daple allosterically inhibit each other’s functions.

(A) Schematic summarizing how Daple enhances noncanonical Wnt signaling downstream of Frizzled (FZD receptors). (B and C) Schematic showing two key intermolecular interplays encountered during noncanonical Wnt signaling via Daple, as shown in (27); that is, (i) FZD7’s internal PDZ-binding motif (PBM) and Daple’s C-terminal PBM compete for binding to the PDZ-domain of Dvl (B), and (ii) Dvl and Gαi compete for two distinct modules on Daple (C). (D and E) Immunoprecipitation assay assessing the binding of Gαi (D) or Dvl (E) to full-length Daple wild-type (WT) or mutants [F1675A (GBA-deficient, FA), ΔPBM, or the double mutant (FA + ΔPBM; also called 2M)]. Bound complexes were analyzed for Daple (myc) and Gαi3 [hemagglutinin (HA)]. Gβ was monitored as a positive control for Gαi3-bound proteins. Quantification of blots (n = 3) is shown in fig. S1 (A and B). IB, immunoblotting.

We investigated the mechanism that triggers the dissociation of Daple/Dvl complexes and the assembly of Daple/Gαi complexes. We found that growth factor–activated RTKs induce both events and hijack the Daple-Gαi cascade to potentiate noncanonical Wnt signals that drive EMT during cancer progression.

RESULTS

G protein regulatory function of Daple is regulated allosterically by its C-terminal PDZ-binding motif

We previously reported that upon Wnt5a stimulation, cytosolic Daple/Dvl complexes dissociate but Daple/Gαi3 complexes assemble (27). In vitro protein-protein interaction assays with purified recombinant proteins had revealed that Daple’s ability to bind Dvl or Gαi is mutually exclusive; binding of the PDZ domain of Dvl to the C-terminal PDZ-binding motif (PBM) in Daple competes with the binding of Gαi to the Gαi-binding and activating (GBA) module of Daple (Fig. 1C), and an increasing concentration of Gαi displaces Daple from Dvl (27). Such competition had been unexpected because the GBA (amino acids 1665 to 1685) and the PBM (amino acids 2025 to 2028) modules in Daple are separated by ~350 amino acids and are within a predicted disordered stretch of the molecule with no semblance to known three-dimensional (3D) structure modules, suggesting that the observed competition between Dvl/Daple-PBM and Gαi/Daple-GBA interactions may be allosteric. Here, when we analyzed Dvl/Daple and Gαi/Daple interactions in Cos7 cells using co-immunoprecipitation assays, we found that binding of Daple to Gαi was consistently higher (increased ~2-fold) when the PBM module of Daple was deleted [Daple-ΔPBM, a mutant that cannot bind Dvl (30)] (Fig. 1D and fig. S1A). Conversely, binding of Daple to Dvl was consistently higher (increased ~2-fold) when the GBA module of Daple was disabled by a single point mutation [Daple-F1675A, a mutant that cannot bind Gαi (27)] (Fig. 1E and fig. S1B). These findings suggest that the PBM and GBA modules of Daple allosterically inhibit each other. Because the increase in binding of Daple-ΔPBM to Gαi3 was also seen in in vitro pulldown assays in which the G protein was bacterially expressed (fig. S1, C and D), we conclude that allosteric inhibition of Daple/Gαi interaction by Daple’s PBM may be due to change in the properties of Daple, not Gαi. Although the structural basis for such inhibition remains unknown, on the basis of these findings and our previous results (27), we conclude that the Dvl-PDZ/Daple-PBM and Gαi/Daple-GBA interactions antagonistically inhibit each other. Binding of Dvl to Daple may therefore suppress Daple/Gαi-dependent enhancement of FZD/noncanonical Wnt signaling. These findings agree with the observations of others that overexpression of Dvl can indeed suppress FZD-dependent G protein activation within the noncanonical Wnt pathway (31). Release of such suppression, which appears to be allosteric, may initiate or enhance Daple-dependent G protein signaling in cells.

Multiple tyrosine kinases phosphorylate Daple’s PBM

We hypothesized that posttranslational modifications at the Dvl-PDZ/Daple-PBM interface may disrupt the interface and initiate Daple/Gαi-dependent noncanonical Wnt signaling in cells. We noted that Daple’s C-terminal PBM [Y2023EY2025GCV2028-COOH] has two evolutionarily conserved tyrosines (Fig. 2A). Multiple kinase prediction programs (Scansite Motif Scan, Massachusetts Institute of Technology; NetPhos 2.0, Denmark; KinasePhos, Taiwan; Phospho-Motif Finder, Human Protein Reference Database) indicated that RTKs and non-RTKs may phosphorylate Daple at those tyrosines. We performed in vitro kinase assays using various recombinant tyrosine kinases and purified, His-tagged, wild-type, or nonphosphorylatable mutants of Daple’s C terminus (Daple-CT; amino acids 1650 to 2028), in which Tyr2025 is mutated to Phe, either alone (YF) or along with a Tyr-to-Phe mutation at Tyr2023 (Y2F). We found that although Tyr2025 was phosphorylated by all RTKs tested [namely, EGFR (epidermal growth factor receptor), PDGFR (platelet-derived growth factor receptor), and InsR (insulin receptor)] and by the non-RTK Src, Tyr2023 was phosphorylated exclusively by Src (Fig. 2, B and C). Matrix-assisted laser desorption/ionization tandem mass spectrometry (MS/MS) spectra of in vitro phosphorylated peptides (in a segment of Daple comprising amino acids 1650 to 2028, which contains five Tyr residues) accounted for four of the five sites; RTKs like EGFR phosphorylated Daple at tyrosines 1750 and 2025, whereas Src phosphorylated Daple at tyrosines 1655, 2023, and 2025 (fig. S2B). The fifth tyrosine, Tyr1699, had been previously identified as a phosphosite in cells using high-throughput phosphoproteomics (www.Phosphosite.org). Tyr2023 and Tyr2025 were phosphorylated at ~75% efficacy. Using a wild-type and a nonphosphorylatable mutant of Daple (YF) and an antibody to pan-pTyr for immunoblotting, we confirmed that Tyr2025 is indeed phosphorylated in cells responding to epidermal growth factor (EGF) (Fig. 2D), as well as in cells expressing constitutively active Src (Src-CA; Fig. 2E). Using a combination of wild-type Daple (Daple-WT), a mutant of Daple that lacks Tyr2025 (ΔPBM), and two nonphosphorylatable mutants of Daple (YF and Y2F), we further confirmed that Src indeed phosphorylated Daple at both Tyr2025 and Tyr2023 (Fig. 2, B and C, and fig. S2E). Finally, we generated an antibody against pTyr2023 and pTyr2025 and confirmed that it can detect Daple exclusively upon phosphorylation in vitro (fig. S2, E and F) and in Cos7 cells (fig. S2, G and H). Using this antibody, we further confirmed that endogenous Daple is phosphorylated at Tyr2023 and Tyr2025 in HeLa cells responding to EGF; phosphorylation of Daple coincides with the onset of AKT phosphorylation and peaks at 15 min after stimulation with EGF (Fig. 2F). From these data, we conclude that multiple RTKs and the non-RTK Src phosphorylate Daple at Tyr2025 within the PBM and that Src can also phosphorylate Daple at Tyr2023. RTK-like orphan receptors ROR1 and ROR2, which are also activated by Wnt5a and potentiate β-catenin–independent Wnt signaling (3234), did not phosphorylate Daple on Tyr2023/2025 either in vitro (using recombinant, commercially obtained kinase or tagged overexpressed ROR1/2 immunoprecipitated from Wnt5a-activated human embryonic kidney cells) or in cells (using overexpressed ROR2) under the conditions tested here.

Fig. 2 Multiple Tyr kinases phosphorylate the C-terminal PBM module of Daple.

(A) (Top) Schematic showing the domain arrangement of Daple. (Bottom) Alignment of Daple’s C terminus in various species displaying an evolutionary conserved PBM (red box). The two phosphotyrosines defined in this work are highlighted in yellow. (B) In vitro kinase assays using several recombinant kinases and wild-type (WT) or nonphosphorylatable mutant [Y2025F (YF) and Y2023/2025F (Y2F)] His-Daple-CT proteins as substrate. Reactions were analyzed for phospho-Tyr-Daple with antibody to pTyr. (C) Schematic summarizing the findings in (B). (D) Immunoprecipitation assays assessing the phosphorylation of Daple after epidermal growth factor (EGF) stimulation in Cos7 cells co-expressing epidermal growth factor receptor (EGFR) and WT or YF-mutant myc-Daple. Immune complexes were analyzed for phosphorylated (pan-pTyr) and total (myc) Daple. (E) Immunoprecipitation assay assessing the phosphorylation of Daple by Src in Cos7 cells co-expressing a constitutively active (CA) mutant of Src (Y527F) and WT or Y2F-mutant myc-Daple. Immunoblotting was performed as in (D). (F) Tyr phosphorylation of endogenous Daple in HeLa cells in response to EGF. Whole-cell lysates were analyzed for pTyr-Daple with anti-pYDaple (pTyr2023 and pTyr2025; antibody validation in fig. S2) and phosphorylation of AKT (pSer473). Blots are representative of three to five experiments. t, total; p, phosphorylated.

Daple is required for activation of Gαi and enhancement of AKT and Rac1 signals downstream of EGFR

Next, we asked how the newly discovered phosphoevent on Daple may affect EGF signaling. We hypothesized that if tyrosine phosphorylation triggers a switch in the composition of Daple-bound complexes from Dvl to G protein, some of the G protein regulatory functions of Daple we reported previously, such as activation of Gαi and enhancement of AKT and Rac1 signals (27), must be affected in cells responding to growth factors. To answer that, first we asked whether Daple and its ability to bind and activate Gαi3 were essential for the cellular response to EGF, the ligand for the prototype RTK, EGFR. We found that Daple was essential for the EGF response in a manner similar to that we reported previously with regard to Wnt5a/FZD7 (27). First, using a monoclonal antibody (mAb) to Gαi-GTP that specifically recognizes Gαi1–3 in a GTP-bound active conformation (35) and previously validated to study Daple’s GEF activity (27), we found that, compared to controls, Daple-depleted HeLa cells failed to enhance Gαi3 activity in response to EGF (Fig. 3A and fig. S3), indicating that Daple facilitates such activation. Second, Daple-depleted HeLa cells also failed to enhance AKT phosphorylation (Fig. 3, B and C) and Rac1 activity (Fig. 3, D and E), the two downstream pathways previously shown to be dependent on the release of free Gβγ from Giαβγ heterotrimers by Daple’s GEF activity (27).

Fig. 3 Daple is required for activation of Gαi3, Rac1, and AKT signals and in the antagonistic inhibition of β-catenin–dependent Wnt signals downstream of EGF/EGFR.

(A) Immunoprecipitation assay assessing Gαi3 activation after EGF stimulation in control (shLuc) and Daple-depleted (shDaple) HeLa cells. Western blot for active Gαi3 (Gαi3-GTP) and total Gαi3. Quantification of blots is shown in fig. S3. (B) Whole-cell lysates of EGF-stimulated HeLa cells in (A) were analyzed for phosphorylated AKT (pSer473), Daple, β-catenin, and actin (loading control). (C) Bar graphs display quantification of phosphorylated/total AKT. Data are means ± SD; n = 3 experiments. (D) Rac1 activity assay in lysates of HeLa cells described in (A) upon EGF stimulation. Data are means ± SD; n = 3. (E and F) Whole-cell lysates of EGF-stimulated HeLa cells expressing Daple-WT or Daple-F1675A (FA) mutant were analyzed for AKT phosphorylation, as described in (B). (G) Bar graphs display quantification as a ratio of phosphorylated to total AKT. Data are means ± SD; n = 3. (H and I) Rac1 activity upon EGF stimulation, assessed in lysates from HeLa cells described in (F). Data are means ± SD; n = 3. (J and K) Quantitative polymerase chain reaction (qPCR) analysis assessing EPCAM (J) and other Wnt target genes (K) induced by EGF in HeLa cells described in (A). Data are means ± SD, fold change in RNA after EGF stimulation; n = 3 experiments. n.s., not significant.

More specifically, we explored whether Daple’s ability to activate Gαi was necessary for the enhancement of the AKT and Rac1 pathways downstream of EGFR. For this, we used previously characterized (27) HeLa cells that are depleted of endogenous Daple and stably express either Daple-WT or a GEF-deficient mutant (F1675A; Daple-FA) at near-endogenous levels. Upon EGF stimulation, Daple-WT but not Daple-FA cells enhanced the activation of AKT (Fig. 3, F and G) and Rac1 (Fig. 3, H and I). When it came to markers of EMT, we previously showed that compared to Daple-WT cells, Daple-FA cells have significantly less basal expression of genes encoding vimentin (VIM), lysyl oxidase-like 3 (LOXL3), and the EMT-associated transcriptional repressor ZEB1 (27). Here, we observed a decrease in the expression of the gene encoding the adhesion molecule EPCAM in Daple-FA cells (Fig. 3J); high expression of EPCAM reportedly facilitates metastasis (36). Because none of these markers showed ligand (EGF)–dependent changes above the baseline changes, these findings suggest that reversible regulation of Daple’s PBM may not have long-term effects on the transcription of the EMT target genes above and beyond the maximal steady-state effects already attributable to the presence or absence of Daple.

Finally, we analyzed the effect of Daple on EGF-stimulated changes in canonical Wnt→β-catenin–TCF/LEF target genes. We also found that the expression of several β-catenin–TCF/LEF target genes were enhanced in Daple-depleted cells (Fig. 3K), indicating that Daple antagonizes the β-catenin–dependent Wnt pathway in a manner similar to what we observed previously downstream of Wnt5a. Together, these studies indicate that, much like the Wnt5a/FZD pathway (27), Daple and its G protein modulatory function are also important for the activity of several components of signaling downstream in the EGF/EGFR pathway. We hypothesized that Tyr phosphorylation of Daple’s PBM may serve as an acute trigger for the Daple→Gαi axis by simply disrupting the Daple/Dvl interaction and releasing the allosteric inhibition imposed by Dvl.

Tyrosine phosphorylation abolishes Daple’s ability to bind Dvl

We next investigated whether the binding specificity between Dvl’s PDZ motif and Daple’s PBM is controlled by these identified phosphoevents within or flanking Daple’s PBM. We found that binding of His-tagged Daple-CT to a glutathione S-transferase (GST)–tagged peptide of the PDZ from Dvl2 (GST-Dvl2-PDZ) was reduced when the Daple-CT peptide was phosphorylated in vitro by EGFR (Fig. 4A and fig. S4A). By contrast, binding of EGFR-phosphorylated His-Daple-CT to GST-Gαi3 remained unchanged (Fig. 4B and fig. S4B). When full-length myc-Daple was phosphorylated in cells by co-expressing the catalytically active wild-type Src (Src-WT) kinase, binding of Src-phosphorylated full-length Daple to GST-Dvl-PDZ was also reduced, whereas binding to GST-Gαi3 was increased (Fig. 4, C to E, and fig. S4C). Together, these findings indicate that although Tyr phosphorylation of Daple’s PBM may not improve binding to Gαi3 directly (based on in vitro assay with two recombinant proteins, Daple and Gαi3), such binding may be enhanced in cells by triggering dissociation of PBM-bound PDZ proteins.

Fig. 4 Tyr phosphorylation inhibits Daple binding to Dvl but favors its binding to and activation of Gαi3.

(A and B) His-Daple-CT glutathione S-transferase (GST) pulldown assays to investigate the impact of such phosphorylation on binding to Dvl [GST-Dvl2-PDZ (A)] and Gαi3 [GST-Gαi3 (B)]. (Top) Bound total and Tyr-phosphorylated Daple (“pYDaple,” as determined using an antibody to pTyr2023/20–25 Daple) were analyzed by immunoblotting (quantification shown in fig. S4, A and B; n = 3 experiments). (Bottom) Confirmation of phosphorylation of His-Daple-CT by immunoblotting. (C to E) Lysates of Cos7 cells co-transfected with Src-HA and myc-Daple were used as source of Daple for GST pulldown assays to assess the impact of Src-dependent phosphorylation of myc-Daple on its ability to bind Dvl [GST-Dvl2-PDZ (C)] and Gαi3 [GST-Gαi3 (D)]. Bound total (myc) and phospho-Daple (pYDaple) were analyzed by immunoblotting as in (A) and (B) (quantification shown in fig. S4C; n = 3). KD, kinase-deficient. (F) GST pulldown assays investigating the impact of phosphomimicking (YE) and nonphosphorylatable (YF) mutants of Daple on binding to Dvl (GST-Dvl2-PDZ). Bound Daple (myc) was analyzed by immunoblotting (quantification shown in fig. S4E; n = 3). (G) Homology-based model of the PBM in Daple (black ribbon and sticks) bound to the PDZ domain of Dvl2 (gray surface), generated using solved crystal structures of Dvl2-PDZ bound to multiple peptides (see Materials and Methods). Each Tyr is predicted to form H-bonds (yellow dotted lines) with each other and with Dvl2 (Asp331 and Ser286). (H and I) Immunoprecipitation assays assessing the impact of EGF stimulation on the abundance of Daple/Dvl (H) or Daple/Gαi3 (I) complexes in Cos7 cells. Dvl- or Gαi3-bound immune complexes were analyzed for Daple (myc) (fold changes in binding are quantified in fig S4, F and G). (J) Immunoprecipitation assays investigating the impact of phosphomimicking (YE) and nonphosphorylatable (YF) mutants of Daple on binding to Gαi3. Gαi3-bound immune complexes were analyzed for Daple as in (H) (quantification shown in fig. S4H). (K) Immunoblotting for Daple, Gαi3, and actin in whole-cell lysates from control (shControl) or Daple-depleted (shDaple) HeLa cells stably expressing WT or mutant (YE) Daple. (L) HeLa cells in K were analyzed for activation of Gαi3 as in Fig. 3A (quantified in fig. S4K). (M) HeLa cells expressing Daple-WT or mutants were pretreated with forskolin and isobutylmethylxanthine (see Materials and Methods) and treated (+) or not (−) with 50 nM EGF for 30 min before analyzing adenosine 3′,5′-monophosphate (cAMP) levels by radioimmunoassay. Bar graphs display cAMP concentrations. Data are means ± SD; n = 3. (N) Schematic summarizing our conclusions from (A) to (M), that is, that phosphorylation of Daple’s PBM by protein tyrosine kinases (PTKs) triggers dissociation of Daple/Dvl complexes and enables Daple to bind and activate Gαi.

To study the effect of phosphorylation of Daple-PBM in cells, we generated phosphomimicking and nonphosphorylatable mutants of full-length myc-Daple in which Tyr2025, the multi-RTK substrate site, was replaced with either glutamate (YE) or phenylalanine (YF), respectively. Additional mutants were created in which both Y2025 and Y2023 were substituted (Y2E or Y2F) to study the effect of Src-dependent phosphorylation on those residues. We found that both phosphomimicking myc-Daple mutants, YE and Y2E, failed to bind GST-Dvl-PDZ (Fig. 4F and fig. S4D). Unexpectedly, we observed that the nonphosphorylatable YF mutant of Daple also bound poorly to Dvl, albeit ~60% less than wild type (Fig. 4F and fig. S4E), suggesting that the residue Y2025 may play an important role in the interface between Daple and Dvl that cannot be fulfilled by Phe [although both Tyr and Phe feature a six-carbon aromatic ring, the latter lacks a hydroxyl (−OH) group, which imparts its polar characteristics and enables participation in hydrogen (H) bonding and improved protein stability (37)].

To gain structural insights into these findings, we generated a 3D homology-based model of Daple’s PBM bound to Dvl’s PDZ domain using previously solved structures of Dvl2-PDZ domain co-crystallized with various peptides from the Pocketome (38). The model showed that although both tyrosines are predicted to interact directly with Dvl by forming H-bonds (Tyr2025 with Asp331 in Dvl and Tyr2023 with Ser286 in Dvl), the two tyrosines also stabilize each other with a third H-bond (Fig. 4G). These findings explain why the negative charge generated by phosphorylation at Tyr2025 (or by the Y2025E mutation) alone is sufficient to abolish Daple/Dvl interaction, likely by repelling Asp331 on Dvl. Because the interface is enriched in Tyr-OH–mediated H-bonds, replacement of Tyr with Phe is indeed predicted to destabilize or weaken the Daple-PBM/Dvl-PDZ interface, consistent with findings in our binding assays (Fig. 4, A to F). These results are consistent with a previous report of carefully conducted (Y→F) mutagenesis experiments, which have documented the importance of Y-OH–mediated interactions to the energy of protein-protein interactions (39).

To determine whether these findings with in vitro–phosphorylated Daple or phosphomimicking YE mutants hold true in cells, we analyzed the abundance of Dvl/Daple complexes in cells before and after stimulation of the prototype RTK EGFR with EGF. Co-immunoprecipitation studies confirmed that the amount of Daple complexed with Dvl dropped significantly (by ~75 to 80%) within 15 min after EGF stimulation (Fig. 4H and fig. S4F), similar to that previously observed in response to Wnt5a ligand (27). Together, these findings demonstrate that growth factors can indeed disrupt the Daple/Dvl interaction and that Tyr phosphorylation of Daple’s PBM by multiple RTKs and non-RTKs could serve as one such trigger and a mechanism for convergent signaling.

Tyrosine phosphorylation of Daple favors binding to and activation of Gαi downstream of EGF

Next, we investigated whether the loss of Dvl/Daple interaction upon EGF stimulation enhances Daple’s ability to bind and activate Gαi. Consistent with our previous findings that Tyr phosphorylation of Daple favors interaction with G protein (Fig. 4D), binding of Daple to G protein was increased ~2.5-fold after EGF stimulation (Fig. 4I and fig. S4G), binding of YE was enhanced ~1.5-fold, and binding of the YF mutant was abolished (Fig. 4J and fig. S4H). As for binding to FZD7, it appeared that Tyr phosphorylation may have no effect, because both Daple-WT and Daple-YE (the phosphomimic mutant) bound to a similar extent (fig. S4, I to J). It is noteworthy that the nonphosphorylatable Daple-YF mutant, which lacks the tyrosines and hence cannot form H-bonds to stabilize protein-protein interfaces (Fig. 4G), bound less (fig. S4, I to J), indicating that the tyrosines in Daple’s PBM somehow stabilize the Daple-FZD interaction we reported previously (27). Finally, we confirmed that increased binding of Gαi to Daple-YE (Fig. 4J) indeed translated to increased G protein activity in cells because compared to cells expressing Daple-WT, activation of Gαi was higher in cells expressing the phosphomimicking Daple-YE mutant (Fig. 4, K and L, and fig. S4K). Thus, the phosphomimicking YE mutant did not bind Dvl but bound and activated Gαi more efficiently and had no statistically significant effect on binding to FZD7. In contrast, the nonphosphorylatable YF mutant bound Dvl and FZD7, albeit with reduced affinity but did not bind Gαi.

We previously showed that activation of Gαi by Daple downstream in the Wnt5a/FZD7 pathway reduced the cellular concentration of the second messenger cAMP. Thus, we explored whether activation of Gαi that is initiated by Tyr phosphorylation of Daple translates into inhibition of cAMP. We found that in control cells, EGF stimulation induced a small but significant increase (~25 to 30%) in cAMP, whereas in cells in which Daple was depleted by short hairpin RNA (shRNA), the increase was 500% (Fig. 4M), indicating that Daple is important for the suppression of cAMP after EGF stimulation. This phenotype was reversed in Daple-depleted cells stably expressing Daple-WT. Basal levels of cAMP in cells expressing the YE or YF Daple mutants were consistent with their Gαi-binding properties (Fig. 4J), in that, compared to cells expressing the Daple-WT, cells expressing the YE mutant had reduced cAMP and cells expressing the YF mutant had increased cAMP (Fig. 4M). In addition, cells expressing Daple-YE were unresponsive to EGF, in that, it showed no ligand-dependent increase in cAMP.

Together, these findings indicate that the two tyrosines in Daple’s PBM have a direct impact on binding to Dvl and affect two other key interactions mediated by distinct modules in the C terminus of Daple that are located upstream of the PBM, namely, the FZD-binding domain and the GBA motif (fig. S2A). Although the mechanisms of such allosteric effect remain unclear, phosphorylation of the two tyrosines in Daple-PBM by growth factor RTKs markedly abrogated Daple’s ability to bind Dvl but increased its ability to bind and activate Gαi and inhibit cellular cAMP (Fig. 4N).

Tyrosine phosphorylation of Daple’s PBM enhances noncanonical Wnt signals and triggers EMT

To study the impact of tyrosine phosphorylation of Daple on pathways previously attributed to Daple (namely, Rac1 and PI3K/AKT), we generated HeLa cell lines stably depleted of endogenous Daple and rescued by expressing Daple-WT or various YE/YF mutants at close to endogenous levels (fig. S5A). Because Tyr2025 is a common target of RTKs and non-RTKs, and because substitution of this residue with Phe is sufficient both for maximally dissociating Daple/Dvl complexes and for binding and activating Gαi, we proceeded with an in-depth characterization of the cells expressing Daple-WT or Daple-YE. Because the nonphosphorylatable YF mutations continued to bind Dvl (Fig. 4F) and FZD7 (fig. S4, D and E), albeit more weakly than Daple-WT, but abolished its interaction with Gαi3 (Fig. 4J), we cautiously analyzed this mutant as a nonphosphorylatable control that is likely to be unresponsive to EGF stimulation.

Hyperactivation of Gαi in Daple-YE–transfected cells was associated with the enhancement of all previously published functions of Daple (27). For example, compared to Daple-WT–transfected cells, Daple-YE cells displayed a greater suppression of β-catenin protein abundance (Fig. 5, A and B, and fig. S5, B and C) and expression of its transcriptional targets (MYC, CCND1, and SFRP1; Fig. 5C, left). Unexpectedly, the expression of two other targets of β-catenin, the genes encoding osteopontin [OPN, which is known as a master regulator of EMT (40) through its ability to stabilize VIM (41)] and Axin2 [which enhances EMT via induction of the transcription factor Snail (42)], was not suppressed but instead was increased in Daple-YE cells (Fig. 5C, right). Thus, constitutive phosphoactivation of Daple’s PBM mimicked in Daple-YE cells was associated with suppression of β-catenin/TCF/LEF transcriptional targets except for those that trigger EMT. Suppression of β-catenin–dependent Wnt signaling in Daple-YE cells was also accompanied by reduced growth of these cells under both anchorage-dependent and anchorage-independent conditions (Fig. 5, D and E, and fig. S5, D to G). Compared to Daple-WT–transfected cells, Daple-YE cells displayed greater Rac1 activity (Fig. 5F) and AKT phosphorylation (Fig. 5G), enhanced expression of markers of EMT (VIM and LOXL3; Fig. 5, H and I), and increased migration (as determined by Transwell chemotaxis assays; Fig. 5J). Daple-Y2E cells resembled the phenotypes observed in Daple-YE cells, enhanced to a greater degree (fig. S5, B to I). Cells expressing Daple-Y2F (a mutant that cannot bind Gαi and binds poorly to Dvl and FZD) displayed an opposite phenotype; these cells showed increased colony growth and reduced migration (fig. S5, B to J) and, in doing so, resembled the previously characterized GEF-deficient Daple-F1675A mutant that cannot bind or activate Gαi (27). These findings are consistent with other instances in which imbalances in PDZ/PBM interactions perturb cellular homoeostasis and contribute to tumor cell phenotypes that ultimately fuel cancer progression [reviewed in (43)].

Fig. 5 Tyr-phosphorylated Daple enhances noncanonical Wnt signals and triggers EMT and cell migration.

(A and B) Daple-depleted HeLa cells stably expressing Daple-WT or Daple-Y2025E (YE) were analyzed for expression of β-catenin by immunoblotting and quantified. Blot (A) is representative; data (B) are means ± SD; n = 3. (C) Canonical Wnt responsive β-catenin/TCF/LEF target genes were analyzed by qPCR in HeLa cells described in (A). Bar graphs display the fold change in each RNA (y axis). Error bars are means ± SD; n = 3. (D and E) Anchorage-independent (D) and anchorage-dependent (E) colony growth assays from HeLa cells described in (A) (additional representative data are shown in fig. S5, D to G). Bar graphs display the number of colonies (y axis) seen in each cell line. Data are means ± SD; n = 3. (F) Rac1 activity assay analyzed in HeLa cells described in (A). (G) Western blot analysis for phosphorylated (pSer473) AKT and phosphorylated (pS552) β-catenin from HeLa cell lines used in (A). (H and I) Expression of epithelial-mesenchymal transition (EMT) markers vimentin and LOXL3 was analyzed by qPCR in HeLa cell lines described in (A). Bar graphs display the fold change in each mRNA (y axis) normalized to the expression in cells expressing vector control. Data are means ± SD; n = 3. (J) Chemotaxis of HeLa cell lines described in (A) was analyzed by Transwell assays along a serum gradient (0 to 10% FBS). Graphs display the number of cells per high-power field (HPF). Data are means ± SD; n = 3 experiments; representative images are shown in fig. S5H.

Selective enhancement of some EMT-associated β-catenin/TCF/LEF–dependent target genes (OPN and AXIN2) and suppression of other proliferation-associated transcriptional targets in Daple-YE cells (Fig. 5C), despite the relatively lower levels of β-catenin in these cells (Fig. 5, A and B), were an unexpected observation. Such differential response of β-catenin/TCF/LEF target genes in Daple-YE cells responding to EGF could be due to differences in either the distribution or the activity of the remaining pool of β-catenin. We found that enhanced AKT phosphorylation in Daple-YE cells is also accompanied by enhanced phosphorylation of β-catenin at Ser552 (Fig. 5G) and its nuclear localization (fig. S6, A and B); phosphorylation by AKT at that site is known to enhance the transcriptional activity of β-catenin and promote tumor cell invasion (44). It is possible that greater activity of a smaller pool of β-catenin in Daple-YE cells is sufficient for targets like AXIN2 and OPN, which are ultra-responsive to β-catenin/TCF/LEF because of the large number of TCF sites in their promoters, but is not sufficient for other targets. Additional factors (such as additional posttranslational modifications, distribution, proteasomal degradation, and protein-protein interactions) that affect the functions of β-catenin may coexist and, therefore, cannot be ruled out.

Concurrent up-regulation of Daple and EGFR in colorectal tumors predicts poor prognosis

To determine the impact of cross-talk between growth factors and Daple-GEF on clinical outcome, we assessed the correlation of mRNA expression levels with disease-free survival (DFS) in a data set from 466 patients with colorectal cancer (for further details on the cohort of patients, see Materials and Methods). Patients were stratified into “low” and “high” subgroups with regard to Daple (CCDC88C) and EGFR expression with the use of the StepMiner algorithm, implemented with the Hegemon software that performs hierarchical exploration of gene expression microarrays online (Fig. 6A) (45). Kaplan-Meier analyses of DFS over time showed that among patients who had colorectal tumors with high expression of EGFR, high expression of Daple was associated with a significantly poorer prognosis compared to those with low expression of Daple (Fig. 6B). Among patients who had tumors with low expression of EGFR, high expression of Daple was associated with a protective effect, although the trend was not statistically significant (Fig. 6C). Conversely, among patients with high expression of Daple, DFS was significantly different between those with high and those with low expression of EGFR (Fig. 6D); no such trend was noted among patients with low expression of Daple (Fig. 6E). Overall, the high Daple/high EGFR signature carried a poorer prognosis compared to all other patients combined (fig. S7). Together, these results indicate that concurrent up-regulation of the growth factor and noncanonical Wnt pathways spurred by Daple is associated with poorer clinical outcomes. This clinical correlation in colorectal cancer, supported by the fact that the pathways investigated here are frequently up-regulated during tumor development, progression, and dissemination (27, 46), suggests that the mechanism(s) of pathway cross-talk we report here (summarized in Fig. 6F) may one day benefit patients.

Fig. 6 Concurrently high expression of Daple (CCDC88C) and EGFR in colon cancer is associated with reduced disease-free survival.

(A) Graph displaying individual arrays according to the expression levels of EGFR and Daple (CCDC88C) in a data set containing 466 patients with colon cancer. (B and C) Survival analysis using Kaplan-Meier curves according to Daple expression in tumors with high (B) versus low (C) EGFR expression. (D and E) Survival analysis using Kaplan-Meier curves according to EGFR expression in tumors with high (D) versus low (E) Daple expression. (F) Schematic of working model. (Left) In the absence of growth factors or Src activation, Dvl remains complexed to Daple, and Gαi is largely inactive. Stimulation with growth factors like EGF triggers phosphorylation of Daple’s PBM, either directly by EGFR or indirectly via activation of the non-RTK Src, which triggers dissociation of Dvl/Daple complexes and favors the assembly of Daple/Gαi complexes and subsequent activation of Gαi. (Right) Schematic summarizing the key events triggered by the ligand Wnt5a, in the context of our previous report (27).

DISCUSSION

Previously, we reported that noncanonical Wnt signals initiated upon binding of Wnt5a to FZD receptors are enhanced and propagated by the multimodular nonreceptor GEF, Daple, through activation of Gαi proteins (27). The major discovery in the current work is an alternate path for activation of Daple-dependent noncanonical Wnt signaling, one that is triggered by growth factors. In both instances, a key mechanistic step is that ligand stimulation [either Wnt5a (27) or EGF (this study)] triggered the assembly of Daple/Gαi complexes to the detriment of Daple/Dvl complexes (Fig. 6F); such a switch in the composition of Daple-bound complexes is a prerequisite for activation of G protein and enhancement of noncanonical Wnt signals through G protein intermediates (specifically cAMP and free Gβγ). Together, these results place Daple at the point of convergence between Wnt/FZD, growth factor RTKs, and G proteins/GPCRs, three major signaling mechanisms that are conventionally thought to operate independently. In doing so, these findings define a new paradigm in signal transduction at the crossroads of noncanonical Wnt and growth factor signaling.

Mechanistically, the convergence between the three major pathways, the so-called cross-talk, plays out at the level of Daple, Gαi, and Dvl. Our experimental evidence indicates that Dvl and Gαi compete for binding Daple; such competition appears to be allosteric. By triggering tyrosine phosphorylation of Daple’s PBM, growth factors working via activation of multiple tyrosine kinases facilitate the dissociation of Daple/Dvl complexes and favor the assembly of Daple/Gαi complexes. This phenomenon has two clear implications: First, it marks another point of convergent signaling in which multiple RTKs and non-RTKs are able to phosphorylate Daple’s PBM to initiate the Daple→Gαi cascade. It is noteworthy that convergent signaling is a common phenomenon within this family of proteins; Girdin/GIV, the paralog of Daple and the prototypical member of this family of GEFs, is also phosphorylated by multiple RTKs and non-RTKs at two tyrosines within its C terminus, resulting in a common outcome—direct binding and activation of class I PI3Ks. Second, Dvl and perhaps other PDZ proteins that may also bind Daple’s PBM serve as negative allosteric modulators (NAMs) of the Daple→Gαi cascade. These findings do not contradict the well-accepted notion that Dvl is required for noncanonical Wnt signaling or that the Dvl/Daple interaction is required for Daple-dependent Wnt signaling (28). Because Dvl/Daple and Daple/Gαi interactions antagonize each other, but both are essential for Daple-dependent noncanonical Wnt signaling cascade, we conclude that (i) Dvl-PDZ may serve as a NAM for Daple and (ii) reversible dynamic phosphoregulation of the Dvl/Daple interaction may be critical for initiating (when phosphorylated) and terminating (when dephosphorylated) signaling by the GEF Daple.

Our results concur with the notion that many PDZ/PBM interactions between various signaling molecules are regulated by phosphoevents [reviewed in (43)]. For example, multiple Ser/Thr kinases [like breakpoint cluster region (BCR) protein (47), Ca2+/calmodulin-dependent protein kinase II (48), and cAMP-dependent protein kinase (49)] phosphorylate either PDZs or PBMs. There is an exception: Although Tyr phosphorylation of the PBM has been reported for syndecan-1 (SDC1), this phosphorylation did not affect binding to its PDZ partner Tiam1 (50); however, a key distinction of SDC1 from Daple that may affect the functional outcome of this event is the position of the target Tyr in its PBM (in SDC1, the phosphorylated Tyr is the next-to-last amino acid). To the best of our knowledge, this disruption of Daple/Dvl binding by phosphoevents is the first example of Tyr kinase–based signaling regulating a PDZ/PBM interaction that affects G protein signaling. These insights pinpoint a clear “event” that involves players from all three signaling pathways, that is, Tyr kinases within the growth factor signaling pathways, Wnt/FZD and Dvl within the Wnt pathways, and heterotrimeric G protein pathways. Findings also add to the cross-talk between heterotrimeric G proteins and Dvl that have been reported previously, namely, that Gβγ subunits released from Gαi subunits can bind Dvl (5153) and target it for degradation (54). Further studies are warranted to assess the ramifications of this cross-talk.

Our findings using phosphomimicking (YE) and nonphosphorylatable (YF) Tyr mutants of Daple in cell-based assays and hierarchical exploration of gene expression microarrays on patient-derived colorectal tumors shed light onto how concurrent deregulation in growth factor and noncanonical Wnt signaling pathways may affect cancer progression. Although the major physiologic function of noncanonical Wnt signaling is in the establishment of planar cell polarity, tissue morphogenesis, and suppression of tumors, noncanonical Wnt signaling is also known to promote the invasiveness and malignant progression of cancers (55); the latter traits are also fueled by growth factors (56). Overexpression of Wnt5a has been found to be associated with aggressive tumor biology and poor prognosis (5759). Deregulated growth factor signaling (for example, copy number variations or activating mutations in RTKs and increased growth factor production/concentration) is also often encountered in advanced tumors (56). We previously showed (27) that Daple enhances noncanonical Wnt signaling via its ability to activate G proteins downstream of Wnt5a/FZD7; it serves as a tumor suppressor in the normal epithelium and in early tumors (27) but aids metastatic progression in advanced tumors and in circulating tumor cells (17, 27, 29). What triggers such role reversal was unknown. We propose that when Wnt and growth factor pathways are deregulated concurrently during cancer progression (likely due to sequential mutations within each pathway, first in Wnt followed by growth factors), growth factor RTKs phosphorylate Daple’s PBM and augment the prometastatic Daple-dependent noncanonical Wnt signals, thereby fueling cancer dissemination.

In conclusion, our data suggest that the multimodular signal transducer Daple provides a platform for complex cross-talk between noncanonical Wnt, growth factor RTKs, and G protein signaling cascades in multireceptor-driven diseases such as cancer. The findings illuminate how such cross-talk coordinately shapes cellular phenotypes, sometimes even by reprogramming tumor-suppressive pathways to facilitate tumor cell dissemination.

MATERIALS AND METHODS

Reagents and antibodies

Unless otherwise indicated, all reagents were of analytical grade and obtained from Sigma-Aldrich. Cell culture media were purchased from Invitrogen. All restriction endonucleases and Escherichia coli strain DH5α were purchased from New England Biolabs. E. coli strain BL21 (DE3) and phalloidin-Texas Red were purchased from Invitrogen. GeneJuice transfection reagent was from Novagen. PfuUltra DNA polymerase was purchased from Stratagene. Goat anti-rabbit and goat anti-mouse Alexa Fluor 680 or IRDye 800 F(ab′)2 used for immunoblotting were from LI-COR Biosciences. Mouse anti-His, anti–α-tubulin, and anti-actin were obtained from Sigma-Aldrich; anti-Myc and anti-hemagglutinin (HA) were obtained from Cell Signaling Technology and Covance, respectively. Mouse mAb against phosphotyrosine (pTyr, Cat #610000) was obtained from BD Transduction Laboratories. Rabbit anti–pan-Gβ (M-14), anti-Gαi3, anti-Dvl, and anti–β-catenin were obtained from Santa Cruz Biotechnology; anti-phospho-AKT (S473) and anti-pan AKT were obtained from Cell Signaling Technology; anti-Rac1 was obtained from BD Transduction Laboratories. Rabbit polyclonal anti-Daple (total) antibodies were generated in collaboration with Millipore using the C terminus of Daple (amino acids 1660 to 2028) as an immunogen and validated previously (27).

An affinity-purified rabbit polyclonal phosphospecific anti-Daple (pY2023 and pY2025) antibody was generated using a phosphopeptide corresponding to the PBM motif in human Daple as an immunogen by 21st Century Biochemicals. Briefly, rabbits were immunized with a 1:1 mix of two phosphopeptides—Ahx-PQTVWYE[pY]GCV-OH and Ahx-PQTVW[pY]E[pY]GCV-OH—and purified by adsorbing antibodies against the nonphosphorylated sequence using an Ahx-PQTVWYEYGCV-OH peptide.

Plasmid constructs and mutagenesis

Cloning of N-terminally tagged myc-Daple was carried out as described previously (27). All subsequent site-directed mutagenesis and truncated constructs [myc-Daple full-length F1675A (FA), myc-Daple deleted from amino acids 2025 to 2028 (ΔPBM), myc-Daple FA + ΔPBM (2M), myc-Daple Y2025E (YE), Y2023-2025E (Y2E), Y2025F (YF), and Y2023-2025F (Y2F)] were carried out using myc-Daple-WT as template using QuikChange as per the manufacturer’s protocol. The His-Daple-CT WT, YF, and Y2F constructs (amino acids 1650 to 2028) used for in vitro protein-protein interaction assays were cloned from myc-Daple pcDNA 3.1 and inserted within the pGEX-4T vector, respectively, between Nde I/Eco RI restriction sites.

Cloning of rat Gα proteins into pGEX-4T-1 (GST-Gαi3 and Gαi3-HA) have been described previously (6064). C-terminal HA-tagged c-Src for mammalian expression was generated by cloning the entire coding sequence into pcDNA 3.1 between Xho I and Eco RI. HA-Src Y527F (constitutively active) and HA-Src K295R (kinase dead) were generated by site-directed mutagenesis using a QuikChange kit (Stratagene) as per the manufacturer’s protocol. All HA-Src constructs have been previously validated by us (65). Several constructs used in this work were gifts from other investigators: Untagged EGFR was from M. G. Farquhar [University of California, San Diego (UCSD), La Jolla, CA] (62); mouse Dvl1 was from M. V. Semenov (Harvard Medical School, Boston, MA); GST-Dvl2-PDZ was from R. Habas (Temple University, Philadelphia, PA); GST-tagged FZD7-CT construct (66) was from R. Yao (Japanese Foundation of Cancer Research Institute, Japan); GST-PBD (PDZ-binding domain) was from G. Bokoch (The Scripps Research Institute, La Jolla, CA). All these constructs have been previously used and validated by us in our recently published work on Daple (27). Daple shRNA constructs have been validated and are described in detail elsewhere (27).

Protein expression and purification

GST and His-tagged recombinant proteins were expressed in E. coli strain BL21 (DE3) (Invitrogen) and purified as described previously (62, 63, 67). Briefly, bacterial cultures were induced overnight at 25°C with 1 mM isopropyl-β-d-1-thiogalactopyranoside. Pelleted bacteria from 1 liter of culture were resuspended in 20 ml of GST-lysis buffer [25 mM tris-HCl (pH 7.5), 20 mM NaCl, 1 mM EDTA, 20% (v/v) glycerol, 1% (v/v) Triton X-100, and 2× protease inhibitor mixture (Complete EDTA-free; Roche Diagnostics)] or in 20 ml of His-lysis buffer [50 mM NaH2PO4 (pH 7.4), 300 mM NaCl, 10 mM imidazole, 1% (v/v) Triton X-100, and 2× protease inhibitor mixture (Complete EDTA-free; Roche Diagnostics)] for GST- or His-fused proteins, respectively. After sonication (three cycles, with pulses lasting 30 s per cycle, and with 2-min intervals between cycles to prevent heating), lysates were centrifuged at 12,000g 4°C for 20 min. Except for GST-FZD7-CT and GST-PBD constructs (see the “In vitro GST pulldown and immunoprecipitation assays” section), solubilized proteins were affinity-purified on glutathione-Sepharose 4B beads (GE Healthcare), dialyzed overnight against phosphate-buffered saline (PBS), and stored at−80°C.

Generation of stable transfected cell lines

Transfection was carried out using either PEI, as previously described (68), or GeneJuice from Novagen for DNA plasmids following the manufacturer’s protocols. HeLa cell lines stably expressing Daple constructs were selected after transfection and in the presence of Dulbecco’s modified Eagle’s medium (DMEM) supplemented with G418 (800 μg/ml) for 6 weeks. The resultant multiclonal pool was subsequently maintained in the presence of G418 (500 μg/ml). Unless otherwise indicated, for assays involving serum starvation, serum concentration was reduced to 0.2% fetal bovine serum (FBS) overnight for HeLa cells and 0% FBS for Cos7.

Cell lysis

Whole-cell lysates were prepared after washing cells with PBS before resuspending and boiling them in sample buffer. Cold PBS (4°C) was used whenever ligand stimulation was carried out for the designated time periods. Lysates used as a source of proteins in immunoprecipitation or pulldown assays were prepared by resuspending cells in Tx-100 lysis buffer [20 mM Hepes (pH 7.2), 5 mM Mg-acetate, 125 mM K-acetate, 0.4% Triton X-100, and 1 mM dithiothreitol (DTT), supplemented with sodium orthovanadate (500 μM), phosphatase from Sigma-Aldrich, and protease from Roche Life Science inhibitor cocktails], after which they were passed through a 28-gauge needle at 4°C and cleared by centrifugation (10,000g for 10 min) before use in subsequent experiments.

Quantitative immunoblotting

For immunoblotting, protein samples were separated by SDS–polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes (Millipore). Membranes were blocked with PBS supplemented with 5% nonfat milk (or with 5% bovine serum albumin when probing for phosphorylated proteins) before incubation with primary antibodies. Infrared imaging with two-color detection and band densitometry quantifications were performed using a LI-COR Odyssey imaging system exactly as done previously (27). All Odyssey images were processed using ImageJ software (National Institutes of Health) and assembled into figure panels using Photoshop and Illustrator software (Adobe).

In vitro GST pulldown and immunoprecipitation assays

Purified GST-Gαi3, GST-DVL-PDZ, or GST alone (5 μg) was immobilized on glutathione-Sepharose beads and incubated with binding buffer [50 mM tris-HCl (pH 7.4), 100 mM NaCl, 0.4% (v/v) Nonidet P-40, 10 mM MgCl2, 5 mM EDTA, 30 μM GDP, 2 mM DTT, and protease inhibitor mixture] for 90 min at room temperature as described before (62, 63, 65, 67). Lysates (~250 μg) of Cos7 cells expressing appropriate myc-Daple constructs or purified His-Daple-CT (amino acids 1650 to 2028) protein (3 μg) were added to each tube, and binding reactions were carried out for 4 hours at 4°C with constant tumbling in binding buffer [50 mM tris-HCl (pH 7.4), 100 mM NaCl, 0.4% (v/v) Nonidet P-40, 10 mM MgCl2, 5 mM EDTA, 30 μM GDP, and 2 mM DTT]. Beads were washed (four times) with 1 ml of wash buffer [4.3 mM Na2HPO4, 1.4 mM KH2PO4 (pH 7.4), 137 mM NaCl, 2.7 mM KCl, 0.1% (v/v) Tween 20, 10 mM MgCl2, 5 mM EDTA, 30 μM GDP, and 2 mM DTT] and boiled in Laemmli’s sample buffer. Immunoblot quantification was performed by infrared imaging following the manufacturer’s protocols using an Odyssey imaging system (LI-COR Biosciences).

GST-FZD7-CT and GST-PBD constructs were immobilized on glutathione-Sepharose beads directly from bacterial lysates by overnight incubation at 4°C with constant tumbling. The next morning, GST-FZD7-CT or GST-PBD immobilized on glutathione beads was washed and subsequently incubated with cell lysates at 4°C with constant tumbling. Washes and immunoblotting were performed as previously described (27).

For immunoprecipitation, cell lysates (~1 to 2 mg of protein) were incubated for 4 hours at 4°C with 2 μg of appropriate antibody, anti-HA mAb (Covance) for HA-Gαi3, anti-Myc (from Cell Signaling) mAb for myc-Daple, anti-Dvl mAb (from Santa Cruz Biotechnology), or their respective pre-immune control immunoglobulin Gs (IgGs). Protein G (for all mAbs)–Sepharose beads (GE Healthcare) were added and incubated at 4°C for an additional 60 min. Beads were washed in PBS-T buffer [4.3 mM Na2HPO4, 1.4 mM KH2PO4 (pH 7.4), 137 mM NaCl, 2.7 mM KCl, 0.1% (v/v) Tween 20, 10 mM MgCl2, 5 mM EDTA, 2 mM DTT, and 0.5 mM sodium orthovanadate], and bound proteins were eluted by boiling in Laemmli’s sample buffer.

In vitro and in cellulo kinase assays

In vitro kinase assays were performed using bacterially expressed His [6× His (hexahistidine)]–tagged Daple-CT (His-Daple-CT, amino acids 1650 to 2028) proteins (~5 μg per reaction) and ~50-ng recombinant kinases, which were obtained commercially (SignalChem). Reactions were started by adding ~1000 μM adenosine 5′-triphosphate and carried out at 25°C for 30 min in tyrosine kinase buffer [60 mM Hepes (pH 7.5), 5 mM MgCl2, 5 mM MnCl2, and 3 μM sodium orthovanadate]. Reactions were stopped by the addition of Laemmli’s sample buffer and boiling at 100°C.

For cellular phosphorylation assays using overexpressed Daple, Myc-tagged Daple was co-expressed with untagged EGFR or HA-tagged Src-WT, HA-tagged Src–kinase dead, or HA-tagged Src–constitutively active. To determine whether EGFR phosphorylates Daple, cells were serum-starved for 12 to 16 hours and stimulated with EGF (50 nM) for 10 min. Cells were preincubated for 1 hour with an Src inhibitor, PP2 (Calbiochem), as done previously (65). Reactions were stopped using PBS chilled at 4°C, supplemented with 200 μM sodium orthovanadate, and immediately scraped and lysed for immunoprecipitation. Phosphoproteins in both in vitro and in cellulo assays were visualized by immunoblotting with either a pan-pTyr antibody (BD Biosciences) or a phosphospecific anti-pTyr 2023-2025-Daple rabbit polyclonal antibody (21st Century Biochemicals).

Mass spectrometry

These studies were carried out as previously described (69, 70). The key steps are summarized here. For the sample preparation, His-Daple-CT proteins used in various in vitro kinase assays were diluted in TNE [50 mM tris (pH 8.0), 100 mM NaCl, and 1 mM EDTA] buffer. RapiGest SF reagent (Waters Corp.) was added to the mix to a final concentration of 0.1%, and samples were boiled for 5 min. TCEP [tris(2-carboxyethyl)phosphine] was added to a final concentration of 1 mM, and the samples were incubated at 37°C for 30 min. Subsequently, the samples were carboxymethylated with iodoacetamide (0.5 mg/ml) for 30 min at 37°C followed by neutralization with 2 mM TCEP (final concentration). Protein samples prepared as above were digested with trypsin (trypsin/protein ratio, 1:50) overnight at 37°C. RapiGest was degraded and removed by treating the samples with 250 mM HCl at 37°C for 1 hour followed by centrifugation at 14,000 rpm for 30 min at 4°C. The soluble fraction was then added to a new tube, and the peptides were extracted and desalted using C18 desalting columns (Thermo Scientific, PI-87782).

For ultra performance liquid chromatography (UPLC) coupled with tandem mass spectroscopy (LC-MS/MS), trypsin-digested peptides were analyzed by ultrahigh-pressure LC-MS/MS using nano-spray ionization. The nano-spray ionization experiments were performed using a TripleTOF 5600 hybrid mass spectrometer (ABSCIEX) interfaced with nano-scale reversed-phase UPLC (Waters Corporation nano ACQUITY) using a 20-cm, 75-μm-ID (internal diameter) glass capillary packed with 2.5-μm C18 (130) CSH beads (Waters Corporation). Peptides were eluted from the C18 column into the mass spectrometer using a linear gradient (5 to 80%) of ACN (acetonitrile) at a flow rate of 250 μl/min for 1 hour. The buffers used to create the ACN gradient were as follows: buffer A [98% H2O, 2% ACN, 0.1% formic acid, and 0.005% trifluoroacetic acid (TFA)] and buffer B (100% ACN, 0.1% formic acid, and 0.005% TFA). MS/MS data were acquired in a data-dependent manner: The MS1 data were acquired for 250 ms at a mass/charge ratio (m/z) of 400 to 1250 Da, and the MS/MS data were acquired at an m/z of 50 to 2000 Da. The independent data acquisition parameters were as follows: MS1-TOF acquisition time of 250 ms, followed by 50 MS2 events of 48-ms acquisition time for each event. The threshold to trigger an MS2 event was set to 150 counts when the ion had charge states of +2, +3, and +4. The ion exclusion time was set to 4 s. Finally, the collected data were analyzed using ProteinPilot 4.5 (ABSCIEX) for peptide identifications.

Gαi activity as determined by a conformational mAb to GαiGTP

These assays were carried out exactly as done previously (27, 71). Cells were maintained overnight at steady state in media containing 0.2% FBS (if an EGF stimulation was performed) or maintained at steady state in media containing 10% FBS before lysis. For immunoprecipitation of active Gαi3, freshly prepared cell lysates (2 to 4 mg) were incubated for 30 min at 4°C with the conformational Gαi/GTP mouse antibody (1 μg) (35) or with control mouse IgG. Protein G–Sepharose beads from GE Healthcare were added and incubated at 4°C for an additional 30 min (total duration of assay is 1 hour). Beads were immediately washed three times using 1 ml of lysis buffer (composition exactly as above; no nucleotides were added), and immune complexes were eluted by boiling in SDS, as previously described (27, 71).

Measurement of cAMP

HeLa cells were serum-starved (0.2% FBS, 16 hours) and incubated with forskolin and isobutylmethylxanthine (200 μM, 20 min) followed by EGF (30 min). Reactions were terminated by aspiration of media and addition of 150 μl of ice-cold trichloroacetic acid (TCA) 7.5% (w/v). cAMP content in TCA extracts was determined by radioimmunoassay and normalized to protein [determined using a dye binding protein assay (Bio-Rad)] exactly as we have done previously (27). Data are expressed as femtomoles of cAMP per microgram of total protein.

Measurement of Rac1 activity

These assays were carried out exactly as done previously (27) with slight modifications in FBS concentration. Briefly, to analyze the role of phosphotyrosine Daple in the regulation of Rac1 activity, we used Daple-depleted HeLa cell lines stably expressing Daple-WT or tyrosine mutants. Cells were maintained overnight at steady state in media containing 10% FBS before lysis. Lysis was carried out first in radioimmunoprecipitation assay buffer [20 mM Hepes (pH 7.4), 180 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.5% sodium deoxycholate, and 0.1% SDS, supplemented with 1 mM DTT, 500 μM sodium orthovanadate, phosphatase (Sigma-Aldrich), and protease (Roche) inhibitor mixtures] for 15 min on ice and then for an additional 15 min after the addition of an equal volume of Triton X-100 lysis buffer [20 mM Hepes (pH 7.2), 5 mM Mg-acetate, 125 mM K-acetate, 0.4% Triton X-100, and 1 mM DTT, supplemented with sodium orthovanadate (500 μM), phosphatase (Sigma-Aldrich), and protease (Roche) inhibitor mixtures]. During the second 15 min of incubation, cells were broken by passing through a 28-gauge needle at 4°C, and lysates were subsequently cleared (10,000g for 10 min) before use. Equal aliquots of lysates were incubated with bead-bound GST-PBD for 1 hour at 4°C with constant tumbling. Beads were washed in PBS-T buffer [4.3 mM Na2HPO4, 1.4 mM KH2PO4 (pH 7.4), 137 mM NaCl, 2.7 mM KCl, 0.1% (v/v) Tween 20, 10 mM MgCl2, 5 mM EDTA, 2 mM DTT, and 0.5 mM sodium orthovanadate], and bound proteins were eluted by boiling in Laemmli’s sample buffer.

Transwell cell migration

Chemotactic cell migration assays were performed using Corning Transwell plates according to the manufacturer’s protocol exactly as done previously (27) with slight modifications in FBS concentration. HeLa cells were trypsinized, counted, and placed in a Transwell with media containing no FBS (75,000 cells per well). Media in the bottom chamber of each well were supplemented with 10% FBS to trigger chemotactic migration. Cells were allowed to migrate for 24 hours and fixed before staining. Cells that had successfully migrated to the side of the permeable membrane facing the bottom chamber were visualized by staining the membrane with crystal violet. Cell migration (expressed as number of cells per high-power field) was quantified by analyzing 15 to 20 random fields per membrane insert per condition for the number of Giemsa-stained cells. Each experiment was repeated three times (biological repeats); three technical repeats were included during each biological repeat.

3D modeling of Dvl2-PDZ domain bound to the PBM of Daple

The approximate model of the complex was built using the Internal Coordinate Mechanics (ICM) homology modeling platform (72). Structures of Dvl2-PDZ domain co-crystallized with various peptides were collected from the Pocketome (38). A structural alignment of the bound peptides was built, and the PDM of Daple was aligned onto it, suggesting the highest sequence homology with the C1 peptide [Protein Data Bank (PDB) 3cbx (73)]. The initial model of the DVL2-PDZ/Daple-PBM complex was built by assigning the backbone coordinates of both target molecules to their counterparts in the template (PDB 3cbx). The model was further refined via conformational sampling of peptide and PDB domain residue side chains in internal coordinates, followed by full-atom local backbone minimization in the presence of harmonic distance restraints maintaining the secondary structure of the complex.

Anchorage-dependent tumor growth assay

Anchorage-dependent growth was monitored on a solid (plastic) surface as we have previously performed (27), with slight modifications in FBS concentration. Briefly, ~1000 HeLa cells stably expressing various Daple constructs were plated in six-well plates and incubated in 5% CO2 at 37°C for ~2 weeks in 10% FBS growth media. Colonies were then stained with 0.005% crystal violet for 1 hour. Each experiment was repeated three times (biological repeats); three technical repeats were included during each biological repeat.

Anchorage-independent tumor growth assay

Anchorage-independent growth of HeLa cells was analyzed in agar, as we have previously performed (27), with slight modifications in FBS concentration. Briefly, petri plates (60 mm) were prelayered with 3 ml of 1% Bacto agar (Life Technologies) in DMEM containing 10% FBS. About 5000 HeLa cells stably expressing various Daple constructs were then plated on top in 3 ml of 0.3% agar-DMEM with 10% FBS. All assays were carried out using three replicate plates at a seeding density of ~5000 cells per plate. After overnight incubation in a 5% CO2 incubator, 1 ml of DMEM supplemented with 10% FBS was added to maintain hydration. After 2 weeks of growth, colonies were stained with 0.005% crystal violet/methanol for 1 hour and subsequently photographed by light microscopy. The number of colonies in ~15 to 20 randomly selected fields was counted under ×10 magnification. Each experiment was repeated three times (biological repeats); three technical repeats were included during each biological repeat.

RNA isolation and quantitative polymerase chain reaction

These assays were carried out as described previously (27). Total RNA was isolated using an RNeasy kit (QIAGEN) as per the manufacturer’s protocol. First-strand cDNA was synthesized using Superscript II reverse transcriptase (Invitrogen), followed by ribonuclease H treatment (Invitrogen) before performing quantitative real-time PCR. Reactions omitting reverse transcriptase were performed in each experiment as negative controls. Reactions were then run on a real-time PCR system (ABI StepOnePlus; Applied Biosystems). Gene expression was detected with SYBR green (Invitrogen), and relative gene expression was determined by normalizing to GAPDH using the ΔCt method. Primer sequences are available upon request.

Stratification of colon cancer patients in distinct gene expression subgroups and comparative analysis of their survival outcomes

The association between the levels of Daple (CCDC88C) and EGFR mRNA expression and patient survival was tested in a cohort of 466 patients, where each tumor had been annotated with the DFS information of the corresponding patient. This cohort included gene expression data from four publicly available National Center for Biotechnology Information–Gene Expression Omnibus (NCBI-GEO) data series (GSE14333, GSE17538, GSE31595, and GSE37892) (7477) and contained information on 466 unique primary colon carcinoma samples, collected from patients at various clinical stages (AJCC Stages I to IV/Duke’s Stages A to D) by five independent institutions: (i) the H. Lee Moffitt Cancer Center (Tampa, FL, USA), (ii) the Vanderbilt University Medical Center (Nashville, TN, USA), (iii) the Royal Melbourne Hospital (Melbourne, Australia), (iv) the Institut Paoli-Calmettes (Marseille, France), and (v) the Roskilde Hospital (Copenhagen, Denmark). All 466 samples contained in this subset were cross-checked to exclude the presence of redundancies/duplicates. A complete list of all Gene Expression Omnibus (GEO) Sample IDs (GSMIDs) of the experiments contained within the NCBI-GEO discovery data set has been published previously (78). To investigate the relationship between the mRNA expression levels of selected genes (that is, CCDCDDC, WNT5A, EGFR, and FZD7) and the clinical outcomes of the 466 colon cancer patients represented within the NCBI-GEO discovery data set, we applied the Hegemon (“hierarchical exploration of gene expression microarrays on-line”) tool (45). The Hegemon software is an upgrade of the BooleanNet software (79), where individual gene expression arrays, after having been plotted on a two-axis chart based on the expression levels of any two given genes, can be stratified using the StepMiner algorithm and automatically compared for survival outcomes using Kaplan-Meier curves and log-rank tests. Because all 466 samples contained in the data set had been analyzed using the Affymetrix HG-U133 Plus 2.0 platform (GPL570), the threshold gene expression levels for Daple/CCDC88C and EGFR were calculated using the StepMiner algorithm based on the expression distribution of the 25,955 experiments performed on the Affymetrix HG-U133 Plus 2.0 platform. We stratified the patient population of the NCBI-GEO discovery data set in different gene expression subgroups, based on either the mRNA expression levels of Daple/CCDC88C alone (that is, CCDC88C neg versus CCDC88C pos), EGFR alone (that is, EGFR neg versus EGFR pos), or a combination of both (that is, CCDC88C neg/EGFR pos versus CCDC88C pos/EGFR pos versus CCDC88C pos/EGFR neg versus CCDC88C pos/EGFR pos). Once grouped based on their gene expression levels, patient subsets were compared for survival outcomes using both Kaplan-Meier survival curves and multivariate analysis based on the Cox proportional hazards method.

Statistical analysis

Each experiment presented in the figures is representative of at least three independent experiments. Displayed images and immunoblots are representative of the biological repeats. Statistical significance between the differences of means was calculated by an unpaired Student’s t test or alternatively by one-way analysis of variance (ANOVA) wherever more than two groups were compared. A two-tailed P value of <0.05 at 95% confidence interval was considered statistically significant. All graphical data presented were prepared using GraphPad or Matlab.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/519/eaao4220/DC1

Fig. S1. The GBA and PBM in Daple allosterically inhibit each other.

Fig. S2. RTKs and non-RTKs target distinct sets of tyrosines within the C terminus of Daple.

Fig. S3. Daple mediates EGF-EGFR signaling–induced activation of Gαi3.

Fig. S4. Phosphorylation or phosphomimic Y2025E mutation within the PBM of Daple inhibits its binding to Dvl, enhances its binding to Gαi, but has no impact on binding to FZD7R.

Fig. S5. Tyr phosphorylation of Daple’s PBM inhibits colony growth and enhances cell migration.

Fig. S6. Tyr phosphorylation of Daple affects the localization and transcriptional activity of β-catenin.

Fig. S7. Concurrent high Daple and EGFR expression in colon cancers is associated with poor prognosis.

REFERENCES AND NOTES

Acknowledgments: We thank G. N. Gill and M. G. Farquhar (UCSD) and D. Bhandari (California State University, Long Beach) for critical input during the preparation of the manuscript. We thank M. Farquhar (UCSD), M. V. Semenov (Harvard Medical School, Boston, MA), R. Habas (Temple University, Philadelphia, PA), R. Yao (Japanese Foundation of Cancer Research Institute, Japan), and the late G. Bokoch (The Scripps Research Institute, La Jolla, CA) for their gifts of reagents and constructs. Funding: This work was supported by NIH grants [CA100768, CA160911, and DK099226 (to P.G.); T32CA067754 (to J.E.); GM071872, AI118985, and GM117424 (to I.K.); R00CA151673 (to D.S.); and F31 CA206426 (to N.A.K.)]; by the American Cancer Society (ACS-IRG 70–002 to P.G.); by a Padres Pedal the Cause/C3 (#PTC 2017) pilot grant award and Moores Cancer Center intramural funding (to P.G. and D.S.); by a fellowship from the American Heart Association (AHA #14POST20050025 to I.L-S.); by the State Scholarship Fund of China Scholarship Council (no. 201208510048 to F.H.); by a fund from West China Hospital, Sichuan University, PR China (to F.H. during a tenure as a visiting professor to UCSD); and by the Bladder Cancer Advocacy Network (BCAN 286601 to D.S.). Author contributions: N.A., J.E., and P.G. designed, performed, and analyzed most of the experiments in this work. Y.D. cloned and generated all Daple mutants used in this work. N.S., F.H., and K.S. assisted with protein expression, purification, and protein-protein interaction assays. I.L.-S. carried out in vitro kinase assays. N.A.K. performed in vitro kinase assays. M.G. performed the MS analyses. D.S. and P.G. analyzed the patient data sets using the Hegemon software. I.K. generated the homology model for Daple-bound Dvl and provided structure-based guidance to study Daple mutants. N.A. and P.G. conceived the project and wrote the manuscript. P.G. supervised and funded the project. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The raw mass spectrometry data are deposited to the ProteomeXchange Consortium PRIDE (www.ebi.ac.uk/pride), accession ID PXD008671.
View Abstract

Navigate This Article