CUB domain-containing protein 1 and the epidermal growth factor receptor cooperate to induce cell detachment

The CUB domain-containing protein 1 (CDCP1) [1‐3], has been implicated in tumor resistance to cytotoxic chemotherapy agents such as gemcitabine [4], and also allows cancer cells to resist cell death induced by targeted therapeutics such as next-generation BCR-ABL inhibitors [5], and the human epidermal growth factor receptor 2 (HER2)-targeted monoclonal antibody trastuzumab (Herceptin) [6]. CDCP1 is a single-pass transmembrane protein with three extracellular CUB domains and a short intracellular tail. Tyrosine phosphorylation of the intracellular domain of CDCP1 results in downstream signaling through Src-family kinases (SFKs), Akt, and PKCδ [7‐11]. The mechanisms that regulate CDCP1 tyrosine phosphorylation, however, are incompletely understood.

CDCP1 likely contributes to metastasis, in part, by allowing cancer cells to survive and metastasize in the absence of attachment. In the MDA-MB-468 breast cancer cell line, enforced CDCP1 expression induces cell detachment and growth in suspension even in the presence of a suitable adhesive substrate [12]. CDCP1-mediated cell detachment is not observed universally, and how CDCP1 causes suspension growth in specific circumstances is unknown. Clarification of specific mechanisms by which CDCP1 induces cell detachment could provide valuable insights into how CDCP1 promotes metastasis, highlighting the importance of CDCP1 as a therapeutic target.

This paper reports that CDCP1 forms a ternary complex with Src and the EGFR, and that this complex mediates Src activation and Src-dependent tyrosine phosphorylation of CDCP1 and EGFR (i.e., EGFR transactivation). Furthermore, enforced expression of CDCP1 and EGFR cooperate to induce cell detachment from the substratum, and this effect is enhanced by stimulation of the cells with EGF. Together the results suggest that a novel CDCP1/EGFR/Src ternary complex activates several signaling responses that contribute to metastasis. These mechanisms include Src activation, CDCP1 tyrosine phosphorylation, and EGFR transactivation. Importantly, studies carried out with a new class of anti-cancer agents (i.e., Disulfide bond Disrupting Agents [DDAs]), which target epidermal growth factor receptor (EGFR) and its family members HER2 and HER3 [13], show that DDAs disrupt CDCP1 ternary signaling complexes.

Analysis of CDCP1-containing complexes using proteomics techniques revealed that CDCP1 associates with proteins involved in cell-cell and cell-substratum adhesion. These studies identified Galectin-1 and matrix metalloproteinase 14 (MMP-14) among the repertoire of proteins that preferentially associate with the full length or cleaved forms of CDCP1, respectively. The results suggest that the CDCP1/Src/EGFR complex is a novel, druggable target and that DDAs may be useful in abrogating the pro-metastatic functions of this signaling platform. Results presented here, along with previously published studies [11, 14], reveal that CDCP1 functions as a protein-protein interaction hub that interfaces with the signaling proteins and structural elements that control cell-cell and cell-substratum adhesion in a manner that is regulated by CDCP1 proteolytic processing and tyrosine phosphorylation.

Enforced CDCP1 overexpression in the MDA-MB-468 breast cancer cell line induces cell detachment from the substratum, cell proliferation and survival in suspension [12]. Similarly, wild-type CDCP1 induces a detached growth pattern of MDA-MB-468 cells even when they are plated on a substrate suitable for attachment (Fig. 1a). CDCP1 overexpression in this background is also sufficient to drive proliferation and survival of MDA-MB-468 cells in soft agar, as well as collagen I gels (Fig. 1b). Phosphorylation-defective CDCP1 (Y734F) does not induce suspension growth at similar levels of expression (Fig. 1b, c).

This effect of CDCP1 overexpression is not universal, suggesting that MDA-MB-468 cells are particularly susceptible to CDCP1-mediated suspension growth. MDA-MB-468 cells express high EGFR levels in comparison with most other breast cancer cell lines. Thus, crosstalk may exist between EGFR- and CDCP1-dependent signaling. DDAs downregulate EGFR and its family members HER2 and HER3 and prior to EGFR downregulation, a shift in EGFR electrophoretic mobility is evident that parallels EGFR dephosphorylation on the Src kinase phosphorylation site Tyr⁸⁴⁵ [13, 25]. Immunoblot analysis demonstrates that treatment of MDA-MB-468 cells with the DDAs increased the mobility of both EGFR and the full-length form of CDCP1 (Fig. 1d). The CDCP1 and EGFR mobility shifts occurred over a range of concentrations (Fig. 1d, center panel), and were reversible over a similar time course (Fig. 1d, left panel). Further, the patterns of CDCP1 and EGFR mobility were identical across a series of structurally related compounds (Fig. 1d, right panel). Collectively, these results suggest a connection between EGFR and CDCP1 signaling that is suppressed by DDAs, prompting us to investigate potential mechanisms of crosstalk between EGFR and CDCP1.

EGFR co-immunoprecipitated with adenovirally expressed, FLAG-epitope-tagged full-length and cleaved CDCP1 in MDA-MB-468 cells, and the coprecipitated EGFR exhibited tyrosine phosphorylation (Fig. 2a). In contrast, when the HER2-positive BT474 breast cancer cell line was employed in the same experiment, EGFR was barely detectable in association with CDCP1, but HER2 robustly associated with a mimetic of the cleaved form of CDCP1 (ΔCDCP1). DDA NSC624203 disrupted complexes between CDCP1 and EGFR and also dissociated the previously identified CDCP1 binding partners Src [8, 26‐28] and PKCδ [7, 9, 26, 29] (Fig. 2b). DDA effects on CDCP1-EGFR complexes occurred in a concentration-dependent manner (Fig. 2c). Endogenous EGFR, but not c-Met, copurified with CDCP1. Since both EGFR and CDCP1 are known substrates for Src-family kinases (SFKs) [8, 26‐28, 30‐32], we examined whether CDCP1 forms ternary complexes with EGFR and Src. Sequential immunoprecipitations of CDCP1 and Src were performed, followed by immunoblot analysis for all three proteins (Fig. 2d). The results confirmed that these proteins are capable of forming a ternary complex, and that pretreatment of the cells with DDA NSC624203 reduced the association of EGFR, but not Src, with CDCP1. Since Fig. 2b indicated that PKC™ is present in CDCP1-containing complexes, we performed sequential FLAG (CDCP1) and Src immunoprecipitations to determine whether PKC™ is in ternary complexes containing CDCP1 and Src. These experiments demonstrated that in unstimulated cells PKC™ can participate in CDCP1/Src complexes (Fig. 2e). This association was not affected by EGF stimulation, but was reduced by treatment with DDA RBF3. Stimulation of the cells with the tyrosine phosphatase inhibitor pervanadate strongly increased PKC™ association with the complex. This correlated with a dramatic increase in the level of activated Src in the complex, and RBF3 partially overcame the effect of pervanadate. A recent report indicated that PKC™ phosphorylates E-cadherin and blocks E-cadherin mediation of cell-cell adhesion [33]. Interestingly, the presence of E-cadherin in the CDCP1/Src complexes was only apparent upon EGF stimulation, and was not observed with pervanadate stimulation. The observations in Fig. 2e are consistent with a model where CDCP1 blocks cell-cell attachment in part by nucleating complexes containing EGFR, Src, and PKC™ that facilitate PKC™ activation and colocalization with E-cadherin.

Pretreatment of MDA-MB-468 cells with the Src inhibitor dasatinib decreased the degree of EGFR phosphorylation on the Src phosphorylation site Tyr⁸⁴⁵, but did not alter the amount of EGFR that co-immunoprecipitated with CDCP1 (Fig. 2f). Pretreatment of the cells with the EGFR/HER2 inhibitor lapatinib reduced the amount of EGFR that co-immunoprecipitated with CDCP1. Similar results were evident in the presence or absence of cell pretreatment with EGF. Overall, the data in Fig. 2a–f suggest the possibility that CDCP1/EGFR/Src ternary complexes may facilitate CDCP1 and EGFR tyrosine phosphorylation by CDCP1-associated Src. To provide support for this model, CDCP1-containing complexes were immunoprecipitated from MDA-MB-468 cells, eluted with FLAG peptide and the immunoprecipitates split into separate aliquots and subjected to in vitro kinase assays performed in the presence or absence of ATP, the Src inhibitor dasatinib, or the EGFR/HER2 inhibitor lapatinib. CDCP1-associated Src became phosphorylated on activation loop site Tyr⁴¹⁶; activated Src, and phosphorylated EGFR on the transactivation site Tyr⁸⁴⁵ because both events were blocked by dasatinib, but unaffected by lapatinib (Fig. 2g). Hence, CDCP1 may play a role both in Src activation, as previously shown [8], and also in promoting EGFR transactivation by Src.

These data (Figs. 1 and 2) suggested that the reason that CDCP1 overexpression drives suspension growth of MDA-MB-468 cells is that these cells express high levels of EGFR. This hypothesis was tested by examining whether enforced coexpression of EGFR and CDCP1 in a cell line with low basal levels of EGFR and CDCP1 is sufficient to disrupt cell-cell and cell-substratum attachment. The development and characterization of T47D breast cancer cells engineered to stably express a green fluorescent protein (GFP)-E-cadherin fusion protein for monitoring E-cadherin intracellular localization was previously described [16]. Derivatives of this T47D.E-cad-GFP line were constructed with stable, forced expression of EGFR, CDCP1, or both proteins (Fig. 3a). EGF stimulation of the resulting lines induced EGFR phosphorylation on Tyr⁸⁴⁵ in the EGFR-overexpressing lines (Fig. 3b). Coexpression of wild-type CDCP1 enhanced Tyr⁸⁴⁵ phosphorylation of EGFR, but the CDCP1[Y734F] mutant lacking the Src binding site did not. Treatment of these cell lines with DDA NSC624205 induced an EGFR mobility shift (Fig. 3c) as observed in Fig. 2, and NSC624205 reduced EGF-induced EGFR phosphorylation on Tyr⁸⁴⁵ (Fig. 3d), but a CDCP1 mobility shift was not observed under these conditions. Examination of the T47D.E-cad-GFP lines growing on tissue culture plastic revealed a somewhat similar morphology in the absence of EGF, with the exception that the EGFR/CDCP1 coexpressing line tended to contain loosely attached clusters of cells (Fig. 3e). Treatment of the cell lines with EGF for 24 hours caused the EGFR-expressing line to grow as single attached cells rather than in colonies. Coexpression of CDCP1 with EGFR further enhanced the effects of EGF and resulted in cultures in which most of the cells were either loosely attached or grew in suspension. Gentle pipetting of the cultures and counting the cells in suspension and the cells remaining attached to the plates showed a statistically significant cooperativity between EGFR, CDCP1, and EGF treatment in inducing cell detachment (Fig. 3f). Cells grown on collagen gels rather than on tissue culture plastic exhibited similar trends toward a more detached growth pattern with the expression of EGFR and CDCP1, and with EGF stimulation (Fig. 3g).

An interesting observation from Fig. 3e and g was that cultures of cells overexpressing EGFR, and to a greater extent EGFR + CDCP1, contained numerous single cells growing either attached or in suspension, suggesting that in addition to suppressing cell-substratum adhesion, CDCP1 and EGFR also cooperate to disrupt cell-cell adhesion. Examining E-cadherin-GFP localization in each line treated with or without EGF showed that the vector control cell line exhibited sharp, well-defined E-cadherin localization at cell-cell junctions (Fig. 4a). E-cadherin was also localized to the cell periphery around the outer edges of each colony. EGFR overexpression, CDCP1 overexpression, and EGF treatment individually increased the diffuseness of E-cadherin localization at cell-cell junctions, caused more cytoplasmic E-cadherin localization, and induced loss of E-cadherin localization at some sites along colony peripheries. The most dramatic alteration in E-cadherin localization and function occurred in cells that overexpressed both EGFR and CDCP1 and were stimulated with EGF. Under these conditions, cultures consisted of single cells or small clusters of cells that were loosely attached to the substratum, and E-cadherin was largely internalized to the cytoplasm in many of these cells.

As biochemical confirmation of the changes in E-cadherin localization observed by fluorescence microscopy, we performed surface biotinylation experiments. E-cadherin labeling was not observed in the control cells and labeling was slightly increased by EGF stimulation (Fig. 4b). Lack of E-cadherin surface labeling while the protein is at cell-cell junctions was previously observed in Madin-Darby Canine Kidney (MDCK) monolayers [34], and it was subsequently shown that stimulation of MDCK cells with hepatocyte growth factor (HGF) increases E-cadherin accessibility [35]. Since T47D cells are relatively well differentiated and have served as a model system for studying tight junction proteins [36, 37], the simplest interpretation is that in T47D cells E-cadherin is at cell-cell junctions, but largely inaccessible from the apical surface.

Enforced expression of either EGFR or CDCP1 dramatically increased E-cadherin surface labeling, while EGF stimulation induced partial E-cadherin internalization in both cell lines. In T47D cells coexpressing EGFR and CDCP1, E-cadherin was almost completely external. EGF treatment induced a striking internalization of E-cadherin. Similar patterns of redistribution were observed with the tight junction protein Occludin, EGFR, and CDCP1. The T47D/EGFR + CDCP1 cells showed high levels of tyrosine phosphorylation of cell surface proteins under control conditions, while after EGF treatment, tyrosine phosphorylated proteins were observed primarily in the internal fraction. These effects are not due to global changes in EGF signaling since EGF stimulated Erk phosphorylation similarly in all four cell lines.

Alternate interpretations are that rather than inducing alterations in protein accessibility, these transmembrane proteins could relocalize to detergent-insoluble domains, or the observations could result from changes in the overall levels of these proteins. However, when whole cells are extracted using similar conditions (1 % Triton X100 whole cell lysates), no differences in E-cadherin levels were observed between cell lines with or without EGF treatment (Fig. 4c). Thus, the simplest interpretation that is consistent with both the E-cadherin fluorescence and surface biotinylation results is that the control T47D cells possess strong cell-cell junctions that limit the access of antibodies or surface biotinylation probes to E-cadherin. However, elevated expression of EGFR or CDCP1 increases E-cadherin accessibility to labeling, but does not induce E-cadherin internalization. EGF treatment in conjunction with EGFR and CDCP1 coexpression induce E-cadherin and Occludin internalization, which coincides with loss of cell-cell adhesion. Together, these observations suggest that changes in E-cadherin localization through cooperative effects between EGFR and CDCP1 may explain how these proteins contribute to decreased cell-cell and cell-substratum adhesion. This may in turn account for the relationship between CDCP1 or EGFR expression in cancer and increased invasiveness, elevated cancer dissemination, and poor patient outcome [38‐40].

The cleaved form of CDCP1 and a deletion mutant, ΔCDCP1, which mimics the cleaved form of CDCP1, associate with E-cadherin [16]. In order to further explore the composition of these complexes, affinity purification of E-cadherin/ΔCDCP1 complexes employed T47D and MDA-MB-231 breast cancer cells stably expressing an E-cadherin-glutathione S-transferase (GST) fusion protein and an adenoviral vector encoding FLAG-tagged ΔCDCP1. Sequential anti-FLAG agarose and glutathione-agarose purifications revealed that E-cadherin/ΔCDCP1 complexed with the endogenous E-cadherin binding partner β-catenin, but not γ-catenin or p120 catenin (Fig. 5a). Immunoprecipitation of endogenous CDCP1 from T47D cells also showed association of both E-cadherin and β-catenin (Fig. 5b). γ-catenin interacted with E-cadherin in T47D cells, but not in MDA-MB-231 cells, due to lower γ-catenin levels in the latter cell line [16]. One interpretation of this observation is that the apparent lack of γ-catenin in E-cadherin/ΔCDCP1 complexes may be due to γ-catenin levels below the detection limit due to the two-step rather than a single-step affinity purification. Alternately, γ-catenin-containing complexes may be less stable than the corresponding β-catenin complexes. To address this point, FLAG-tagged ΔCDCP1 was expressed in MDA-MB-231-derived cell lines that express an E-cadherin-GST fusion protein in the presence or absence of γ-catenin coexpression using an adenoviral vector. E-cadherin or ΔCDCP1 were isolated using glutathione-agarose or anti-FLAG agarose, respectively. Analysis of the affinity-purified complexes revealed that E-cadherin, β-catenin, and γ-catenin coprecipitate with ΔCDCP1 with or without γ-catenin overexpression (Fig. 5c). In contrast, γ-catenin was detected in E-cadherin-GST pulldowns only in the context of γ-catenin overexpression.

CDCP1 may exist as a dimer [28]. Results obtained using differentially epitope-tagged CDCP1 constructs confirm CDCP1 dimerization (Fig. 5d). Dimer formation was not reduced by DDAs, indicating that the disruption of CDCP1/EGFR/Src complexes observed in Fig. 2b was not due to blockade of CDCP1 dimerization. In transient transfection experiments, the most robust dimerization was observed with ΔCDCP1 as compared with CDCP1/CDCP1 homodimers or CDCP1/ΔCDCP1 heterodimers (Fig. 5e). E-cadherin was detected in association with CDCP1 dimers, and appeared to be present in all three classes of CDCP1 dimers despite previous findings that E-cadherin associates preferentially with cleaved CDCP1 or the cleaved CDCP1 mimetic, ΔCDCP1 [16]. One explanation for this discrepancy is that at least some processed CDCP1 is constitutively present due to cleavage of the full-length form. Consistent with this, the short form of CDCP1 is detectable in the sequentially affinity-purified samples obtained from cells transfected with full-length CDCP1. This effect may be amplified by the apparent enhanced ability of the short form of CDCP1 to dimerize. It was also unexpected that similar levels of E-cadherin would be copurified when very different levels of the various dimers were isolated. This could indicate that E-cadherin associates with CDCP1 dimers indirectly and is limited by the availability of an intermediary protein. Overall, the findings in Fig. 5 are consistent with a model in which CDCP1 exists in one or more protein complexes that contain E-cadherin, and that CDCP1 dimerization may be enhanced by its cleavage.

The ability of EGFR and CDCP1 to cooperate in Src activation resulting in cell-cell and cell-substratum detachment may reflect the ability of these proteins to induce E-cadherin relocalization from the plasma membrane to cytoplasmic vesicles. This hypothesis is consistent with previous reports demonstrating the ability of EGFR- and Src-dependent signaling to disrupt cadherin function [41‐45]. CDCP1 harbors no known catalytic activities and thus mediates its anti-adhesive actions via protein-protein interactions. Therefore, affinity purification of CDCP1 complexes was followed by analysis of the multiprotein assemblies using mass spectrometry to gain insight into how CDCP1 actions are carried out. Under conditions where no proteins were detected in control affinity purifications from cells transduced with an adenovirus encoding GFP, several proteins were identified which copurified with either CDCP1 or ΔCDCP1. A gel representative of multiple experiments shows that many of the same protein bands were observed in the CDCP1 and ΔCDCP1 affinity purifications (Fig. 6a). The proteins listed were identified in the indicated bands or gel slices by mass spectrometry (proteomics results are presented in more detail in Additional file 1: Table S1, Additional file 2: Table S2 and Additional file 3: Figure S1). One band was present selectively in ΔCDCP1 complexes and found to contain the adaptor protein p62/SQSTM1 that is involved in targeted autophagy and proteasomal degradation (reviewed in [46, 47]), and the matrix metalloproteinase MMP14 that promotes cancer metastasis by permitting cancer cell survival and proliferation through extracellular matrix degradation [48‐50]. Since CDCP1 is essential for MMP14 trafficking to invadopodia to degrade the extracellular matrix during cell invasion [14], adenoviruses encoding the cleaved and full-length forms of CDCP1 were used to show that MMP14 selectively associates with ΔCDCP1 but not the full-length protein (Fig. 6b). The small amounts of MMP14 observed in association with full-length CDCP1 were apparently a result of CDCP1 cleavage because coexpression of the serine protease inhibitor PAI-1 blocked CDCP1 processing and rendered MMP14 pulldown undetectable in cells expressing almost exclusively full-length CDCP1. Association of MMP14 was not affected by mutating ΔCDCP1 at the major Src binding site, Tyr⁷³⁴. E-cadherin [16] and β1-integrin [11] bind selectively with the short form of CDCP1, while Src associates with CDCP1 irrespective of proteolytic processing. As expected, the pattern of MMP14 association with CDCP1 matched that observed with β1-integrin, but not that seen with Src (Fig. 6c).

Efforts to identify proteins that associate with the full-length, but not cleaved CDCP1 revealed a low-molecular-weight protein with this selectivity (Fig. 6d, red arrow). Analysis of this band by mass spectrometry identified it as the lectin Galectin 1. Galectin 1 specificity for full-length CDCP1 was verified in experiments showing that transiently expressed CDCP1, but not ΔCDCP1, pulled down endogenous Galectin 1, while both forms of CDCP1 pulled down Src equivalently (Fig. 6e). However, the results in Fig. 6a indicate that most CDCP1-associated proteins bind full-length and cleaved forms similarly. Immunoblot analyses of CDCP1/ΔCDCP1 immunoprecipitates were performed to support the identifications made by mass spectrometry and to further characterize these complexes. Interestingly, not only did ΔCDCP1 associate with E-cadherin, β-catenin, γ-catenin, and desmoplakin in cancer cell lines as expected based on the results in Figs. 5 and 6 and a previous study [16], but ΔCDCP1 also associated with the desmosomal cadherin desmoglein 2 (Fig. 7a). Similar results were obtained using AsPC1 and PANC1 pancreatic cancer cell lines (Fig. 7b). The CDCP1-containing multiprotein complexes were evident after cell extraction with several detergents, including 1 % each of Triton X100, IGE-PAL, CHAPS, and sodium deoxycholate, but were less apparent when extractions were performed with 1 % Brij 35 or sodium dodecylsulfate (Fig. 7c). Together, these results suggest that CDCP1 associates either directly or indirectly with numerous proteins present in adherens junctions and desmosomal junctions, consistent with the findings in Figs. 1, 3, and 4, which show that CDCP1 overexpression can contribute to the disruption of cell-cell junctions and cell-substratum contacts.

The CDCP1-associated protein Enigma/PDLIM7 identified in our mass spectrometry studies was selected for further study for three reasons. First, the function of Enigma has not been investigated extensively in epithelial cell lines. Second, studies in the heart indicate Enigma plays a key role in regulating cell signaling and the actinomyosin cytoskeleton [51‐54]. Third, Enigma contains an N-terminal PDZ domain and three C-terminal LIM domains, and thus may function as a molecular scaffold to coordinate cell-cell and cell-substratum adhesion. Enigma associated with both full-length and cleaved CDCP1 and this association was not strongly modulated by tyrosine phosphorylation (Fig. 8a). Ectopically expressed E-cadherin, Enigma, and ΔCDCP1 mutually elevated their expression levels (Fig. 8b). Enigma increased the levels of β-catenin and ΔCDCP1 in E-cadherin complexes, while reducing the levels of p120 catenin in E-cadherin complexes. These effects are likely due to differences in the overall expression of the respective proteins. E-cadherin was coexpressed with its major binding partners α-catenin, β-catenin, and γ-catenin in various combinations. The resulting complexes with ΔCDCP1 were analyzed by immunoprecipitation followed by immunoblotting and confirmed that Enigma was also present in these complexes, but did not significantly alter the levels of any of the other proteins in the complex (Fig. 8c). This observation suggests that ΔCDCP1 interacts with Enigma and E-cadherin through distinct, non-overlapping sites, or that ΔCDCP1 associates with Enigma and E-cadherin through different protein complexes. Reciprocal immunoprecipitations with E-cadherin antibodies produced similar results (Fig. 8d), although coexpression of Enigma with E-cadherin increased the amounts of E-cadherin, ΔCDCP1, and β-catenin in the immunoprecipitates. Together, the findings in Figs. 7 and 8 show that CDCP1 associates with several proteins involved in cell-cell and cell-substratum adhesion, identify novel CDCP1 binding partners such as Enigma and Galectin 1, illustrate that CDCP1 associates with MMP14 and Galectin 1 in a manner dependent on the state of CDCP1 proteolytic processing, and identify Enigma as a protein present in CDCP1- and E-cadherin-containing protein complexes.

These results indicate a new function for CDCP1 in facilitating Src-dependent EGFR transactivation. Ternary complex formation among CDCP1, EGFR, and Src is paralleled by disruption of cell-cell and cell-substratum adhesion in human breast carcinoma cells, coincident with relocalization of E-cadherin from cell-cell junctions to cytoplasmic vesicles. The current findings suggest a mechanism whereby CDCP1, EGFR, and Src cooperate to induce cancer aggressiveness and dissemination, and may provide a rationale for therapeutic targeting of these proteins in combination. Alternatively, since DDAs disrupt the CDCP1/EGFR/Src ternary complex, in addition to effectively inhibiting the growth of HER2-positive malignancies [13], DDAs may be useful for the treatment of breast tumors that overexpress EGFR, or for blocking metastasis driven by signaling via the CDCP1/EGFR/Src ternary complex. These possibilities are currently under investigation.

CDCP1(Y734F), CDCP1 with a mutation of tyrosine 734 to phenylalanine; CDCP1, CUB domain-containing protein 1; DDAs, Disulfide bond Disrupting Agents; EGFR, epidermal growth factor receptor; GFP, green fluorescent protein; GST, glutathione S-transferase; HER2, human epidermal growth factor receptor 2; IDA, information-dependent acquisition; MMP14, matrix metalloproteinase 14; PAI-1, plasminogen activator inhibtor-1; SFKs, Src-family kinases; ΔCDCP1, mimetic of the cleaved form of CDCP1