The Cell Surface Glycoprotein CUB Domain-containing Protein 1 (CDCP1) Contributes to Epidermal Growth Factor Receptor-mediated Cell Migration*

Background: Epidermal growth factor (EGF) activates EGF receptor (EGFR) to promote cell migration and cancer. Results: EGF/EGFR up-regulates the cell surface glycoprotein CDCP1, and blockade of CDCP1 reduces EGF/EGFR-induced migration of ovarian cancer cells lines. CDCP1 is expressed by ovarian tumors. Conclusion: CDCP1 contributes to EGF/EGFR-induced cell migration. Significance: Targeting of CDCP1 may be a rational approach to inhibit cancers mediated by EGFR. Epidermal growth factor (EGF) activation of the EGF receptor (EGFR) is an important mediator of cell migration, and aberrant signaling via this system promotes a number of malignancies including ovarian cancer. We have identified the cell surface glycoprotein CDCP1 as a key regulator of EGF/EGFR-induced cell migration. We show that signaling via EGF/EGFR induces migration of ovarian cancer Caov3 and OVCA420 cells with concomitant up-regulation of CDCP1 mRNA and protein. Consistent with a role in cell migration CDCP1 relocates from cell-cell junctions to punctate structures on filopodia after activation of EGFR. Significantly, disruption of CDCP1 either by silencing or the use of a function blocking antibody efficiently reduces EGF/EGFR-induced cell migration of Caov3 and OVCA420 cells. We also show that up-regulation of CDCP1 is inhibited by pharmacological agents blocking ERK but not Src signaling, indicating that the RAS/RAF/MEK/ERK pathway is required downstream of EGF/EGFR to induce increased expression of CDCP1. Our immunohistochemical analysis of benign, primary, and metastatic serous epithelial ovarian tumors demonstrates that CDCP1 is expressed during progression of this cancer. These data highlight a novel role for CDCP1 in EGF/EGFR-induced cell migration and indicate that targeting of CDCP1 may be a rational approach to inhibit progression of cancers driven by EGFR signaling including those resistant to anti-EGFR drugs because of activating mutations in the RAS/RAF/MEK/ERK pathway.

Epidermal growth factor receptor (EGFR) 3 is a receptor-tyrosine kinase consisting of an extracellular ligand-binding domain, a membrane spanning region, and a cytoplasmic tail containing a kinase domain and docking sites for signaling effectors and modulators (1,2). Ligand-mediated activation of EGFR requires binding by one of at least six growth factors, including epidermal growth factor (EGF), to the receptor extracellular domain (3,4), which initiates autophosphorylation of specific receptor cytoplasmic tyrosine residues, including tyrosine 1068 (5). This triggers further signaling events, including activation of RAS/RAF/MEK/ERK (6,7) and Src family kinase (SFK) (8 -13) pathways, which promote a range of cellular processes including migration (14 -16).
As with other mediators of migration, EGFR activation alters cellular plasticity to promote the transition from a morphology where cells are in intimate contact with neighbors and substratum to an elongated, spindle shape associated with increased ability to migrate and invade into surrounding matrix (17)(18)(19). In a range of cell lines EGF/EGFR-mediated migration is accompanied by a so-called epithelial-to-mesenchymal transition (EMT) involving loss of expression of proteins necessary for maintenance of epithelial phenotypes, such as E-cadherin (20,21), as well as augmented expression of proteins associated with mesenchymal phenotypes including N-cadherin (18,22) and vimentin (23).
In physiological settings EGFR-mediated cell migration is tightly regulated to mediate normal development and homeostasis (24). However, in disease states such as cancer, elevated EGFR signaling either via activating mutations or increased expression promotes aberrant cell migration. Interestingly, although this inappropriate activation of EGFR is a known promoter of malignancies of the lung, colon, pancreas, breast, head and neck, and ovary (25) and targeting this receptor has shown much promise in preclinical settings, anti-EGFR drugs have been largely ineffective for the treatment of each of these cancers except non-small cell lung cancer (25,26). The lack of efficacy of EGFR inhibition alone is due in part to drug resistance resulting from mutations in downstream signaling effectors such as RAS and RAF or activation of other receptors including IGF-1 receptor and Met (25,26).
Recently the cell surface glycoprotein CUB domain containing protein 1 (CDCP1, also named SIMA135, gp140, Trask, and CD318 (27)(28)(29)(30)) has emerged as a promoter of cell migration in vitro and cancer cell dissemination in animal models (31,32). For example, CDCP1 promotes in vitro migration and peritoneal dissemination of scirrhous gastric carcinoma cell lines (33) as well as in vitro migration of pancreatic cancer cells (34). In addition, antibody targeting of CDCP1 inhibits prostate cancer cell migration and invasion in vitro and metastasis in a mouse xenograft model (35). Antibody-based disruption of CDCP1 function has also been effective at blocking in vivo dissemination of a highly metastatic prostate cancer PC3 cell variant and HeLa and HEK293 cells ectopically expressing CDCP1 (36,37). The mechanisms regulating CDCP1 in cell migration have been largely unexplored, although recently this protein was shown to be regulated by hypoxia-inducible factor 1␣ and 2␣ (HIF 1␣ and 2␣) and to play a critical role in kidney cancer cell migration (38).
In this study we used EGF/EGFR-responsive epithelial ovarian cancer cell lines to explore the role of CDCP1 in EGFRinduced cell migration. We demonstrate that CDCP1 mRNA expression is up-regulated by EGF/EGFR signaling via a pathway that involves the activity of ERK but not Src. We also show that antibody and shRNA-mediated disruption of CDCP1 efficiently block EGF/EGFR-induced cell migration. Our immunohistochemical analysis demonstrates that CDCP1 is expressed during ovarian cancer progression. Targeting of CDCP1 may be a rational approach to inhibit malignancies, such as ovarian cancer, that are driven by EGF/EGFR.
Cell Lysis and Western Blot Analysis-Whole cell lysates were collected in a buffer containing Complete protease inhibitor mixture (1ϫ), 2 mM sodium vanadate, 10 mM sodium fluoride, 50 mM Tris-HCl (pH 7.4), NaCl (150 mM), and CHAPS (1%). Protein concentration was determined by a microbicinchoninic acid assay (Thermo Scientific). Cell lysates (20 g) were separated by SDS-PAGE under reducing conditions, transferred to nitrocellulose membranes, and blocked in Odyssey blocking buffer (LI-COR Biosciences). Membranes were incubated with primary antibodies diluted in blocking buffer overnight at 4°C, washed with Tris-buffered saline containing 0.05% Tween 20, and then incubated with secondary IRDye 680-or 800-conjugated mouse or rabbit IgG as appropriate. Images were generated, and densitometry analysis was per-formed using the Odyssey system and software (LI-COR Biosciences).
Phase Contrast and Time-lapse Images-Bright field images were captured using a phase contrast microscope and a digital camera (10ϫ objective, Nikon Eclipse TE2000-U, Nikon, Japan) and Vϩϩ software. Time lapse images were captured every 15 min up to 24 h using a Leica AF-1600 wide-field microscope and system software (Leica Microsystems, Sydney, Australia). Cell scattering was measured as the distance between cell nuclei in three independent experiments using InDesign software (Adobe, Adobe Systems Pty Ltd, Chatswood, Australia). For cells silenced for CDCP1, measurements were performed from images acquired from the Nikon Eclipse system on 200 pairs of cells from 10 randomly selected fields per treatment group. For cells treated with anti-CDCP1 antibody 10D7 or control IgG, measurements were performed from time lapse images acquired on the Leica AF-1600 system on at least 80 individual randomly selected cells per treatment group.
Confocal Microscopy Analysis-Cells were grown in serumcontaining media on sterile coverslips until 60% confluent. The cells were then grown in serum-free media for a further 24 h then either left untreated or treated with EGF (30 ng/ml) for 24 h. After washes with PBS, cells were fixed (4% (w/v) with paraformaldehyde in PBS), permeabilized (0.5% (v/v) Triton X-100 in PBS), and blocked (5% (w/v) BSA in PBS) then incubated with mouse monoclonal anti-CDCP1 antibody 10D7 (dilution 1:200 (v/v) in 1% BSA/PBS) at room temperature for 2 h. After washes, cells were incubated with a mouse Alexa Fluor 488-conjugated secondary antibody, Alexa Fluor 568 phalloidin, and DAPI. Images were acquired using a Leica-TCS SP5 confocal microscope (63 ϫ oil immersion objective lens).
Immunohistochemistry-Paraffin-embedded tissue sections (4 m) were from benign serous adenomas (n ϭ 3), primary serous tumors (n ϭ 3), and serous metastases (n ϭ 3). The source of these tissues was described previously (43), and each was used with institutional ethics approval (Queensland University of Technology certificate no. 080000213) and informed patient consent. Immunohistochemistry was performed as previously described (40). Briefly, sections (4 m) were deparaffinized in xylene and rehydrated followed by antigen retrieval with microwave heat treatment in 10 mM citric acid (pH 6.0). Sections were incubated overnight at 4°C with a rabbit anti-CDCP1-C-terminal antibody (1:100 in 1% (w/v) BSA in PBS) or rabbit IgG as the negative control (Dako, Campbellfield, Australia). Signal was detected using the EnVision TM peroxidase system (Dako), and sections were counterstained with Gill's hematoxylin. Staining was visualized using an Olympus BX41 microscope (Olympus, Japan) and photographed with a Qimaging digital camera (MicroPublisher 3.3RTV) and associated software (QCapture Pro 6.0, Burnaby BC, Canada). Images were processed using Adobe Photoshop CS3 and displayed using CorelDraw14 (Corel Pty Ltd, Sydney, Australia).
Statistical Analysis-One-tailed Student's t tests were performed for statistical analysis. p values with a 95% confidence interval were obtained from at least three independent experiments using GraphPad Prism (GraphPad Software, Inc, La Jolla, CA). A p value Ͻ0.05 was considered to be significant.

CDCP1 Is Up-regulated by EGF/EGFR Signaling Axis-To
identify cell lines suitable for examination of the role of CDCP1 in EGF/EGFR-mediated cell migration, we first analyzed the expression of CDCP1 in 10 epithelial ovarian cancer-derived cell lines. Anti-CDCP1 Western blot analysis was performed on lysates from cells cultured in serum-containing media. CDCP1 is produced as a 135-kDa cell surface protein (28) that is proteolytically processed in a range of cultured cell lines (29,44) as well as in in vivo settings (29,37) to a 70-kDa cell-retained form. As shown in Fig. 1A, 70-and 135-kDa CDCP1 species were most highly expressed by OVCA420 cells, with high relative expression of these species also apparent in OVCA432, PEO1, and PEO4 cells. Lower level expression of both bands was also apparent in OV90 and PEO14 cells, whereas the 135-kDa band was apparent in OAW42 and SKOV-3 cells (Fig. 1A). Caov3 cells expressed the 135-kDa band at high levels with much lower levels of 70-kDa CDCP1 generated by this cell line. OVMZ-6 cells were the only ovarian cancer line that did not express CDCP1.
A subset of the CDCP1-expressing cell lines (Caov3, OVCA420, OVCA432, PEO4) were next evaluated for the ability to respond to EGF stimulation. This was performed by Western blot analysis for phosphorylated EGFR-Tyr-1068, a site that is phosphorylated by Jak2 to provide a docking site for Grb2 and activation of MAP kinases (46). In this and all subsequent experiments the effects of EGF stimulation were maximized by growth of cell lines in serum-free media for 24 h before EGF stimulation. As shown in Fig. 1B, EGF induced rapid phosphorylation of EGFR in each cell line, and this was sustained for at least 1 h.
The effect of EGF signaling on CDCP1 expression was analyzed at mRNA and protein levels in two of the EGF-responsive cell lines, Caov3 and OVCA420. Quantitative real-time RT-PCR analysis was performed on total RNA isolated from cells treated with EGF for 0.2 and 24 h. As shown in Fig. 2A, although EGF had no effect on CDCP1 mRNA levels at 0.2 h, it induced a greater than 2-fold increase in expression in both cell lines at 24 h.
Consistent data showing EGF-induced up-regulation of CDCP1 were obtained at the protein level. In these experiments cells were serum-starved for 24 h then stimulated with EGF for 0.1, 0.2, 0.5, 1, 8, and 24 h. In control experiments cells were treated with the potent and selective EGFR inhibitor AG1478 (47). As shown in Fig. 2B, left, EGF induced an ϳ2.5-fold increase in expression of 135-kDa CDCP1 in Caov3 cells within 1 h, and this was sustained up to 24 h. AG1478 treatment blocked this induction, indicating that EGF-induced up-regulation of CDCP1 was mediated by EGFR. In contrast with OVCA420 cells grown in serum-containing media, which generated abundant levels of 70 kDa CDCP1 (Fig. 1A), these cells grown in serum-free media expressed very low levels of this species, and like the Caov3 line, 70-kDa CDCP1 was not induced by EGF (Fig. 2B). Although it is clear that CDCP1 can undergo proteolytic conversion from 135-to 70-kDa in a range of cultured cells (44), the identity of the protease responsible for cleavage of CDCP1 in these lines has not been determined. This is in contrast with proteolytic processing of CDCP1 that occurs during metastasis in vivo that has recently been shown to be mediated in mice by plasmin (37). We propose that because not all cell lines that express 135-kDa CDCP1 also generate 70-kDa CDCP1 when grown in serum-containing media (Fig. 1A; OAW42 and SKOV-3 cells), either CDCP1-cleaving cell lines produce a protease that activates a latent protease present in serum that cleaves CDCP1, or these cell lines produce a protease that directly cleaves CDCP1. Based on comparison of the data for Caov3 and OVCA420 cells in Fig. 1A (serum containing media) and Fig. 2B (serum-free media Ϯ EGF), we suggest that EGF does not induce cleavage of CDCP1 via induction of either a protease that directly cleaves CDCP1 or a protease that activates a CDCP1-cleaving protease present in serum that has remained associated with the cells in serum-free conditions. In addition, we also note from Fig. 2B that there was no change in expression of the EMT markers E-cadherin (21) and N-cadherin (22) in response to EGF.
ERK Signaling Is Required for EGF/EGFR-mediated Up-regulation of CDCP1-We also examined whether two known EGF/EGFR-activated pathways are required for the observed increased expression of CDCP1. We first assessed whether these pathways (ERK (15,16) and Src (8 -13)) are activated downstream of EGF/EGFR in the two lines used in this study, Caov3 and OVCA420 cells. Western blot analysis showed that ERK (phosphorylation of Thr-202/Tyr-204) is rapidly activated in both cell lines in response to EGF and phosphorylation of this protein reduced gradually 8 -24 h after initial stimulation (Fig.  3A). In contrast, activation of Src (phosphorylation of Tyr-416) paralleled the increases in CDCP1 expression seen in Fig. 2B with increasing pSrc levels apparent after 0.5 h of EGF treatment, and this was sustained up to 24 h (Fig. 3A). In these experiments AG1478 treatment blocked phosphorylation of ERK and Src, demonstrating that EGF-induced activation of these kinases was mediated by EGFR (Fig. 3A). These data indicate that both ovarian cancer cell lines signal similarly in response to EGF/EGFR via ERK and Src.
Using selective pharmacological agents, we examined if antagonism of these pathways affects EGF/EGFR-induced upregulation of CDCP1. Activation of ERK was inhibited with

U0126 (48) and SFK activation with SU6656 (49). Inhibition of EGF/EGFR-induced expression of CDCP1 was quantified by densitometric analysis of Western blots of three independent
experiments. This showed that the ϳ2-fold increase in CDCP1 expression induced by EGF in Caov3 and OVCA420 cells was reduced to background levels by U0126 inhibition of ERK (Fig.  3B). In contrast, the SFK inhibitor SU6656 caused a 100% increase in CDCP1 expression above the effect of EGF in Caov3 cells and had no impact on its expression in OVCA420 cells (Fig. 3B). EGF treatment in the presence of vehicle (0.1% DMSO) was used as a positive control for up-regulation of CDCP1 (Fig. 3B). These data indicate that EGF/EGFR up-regulation of CDCP1 requires signaling via the RAS/RAF/MEK/ ERK pathway but not via SFKs.
EGF/EGFR Signaling Induces Cell Migration and Relocalization of CDCP1-The effect of EGF/EGFR signaling on cell migration and localization of CDCP1 was examined by microscopy. As shown in Fig. 4A, a and b, unstimulated and vehicletreated (0.1% DMSO) Caov3 cells grow as defined colonies. In response to 24 h of EGF stimulation, these cells acquire a spindle-shaped morphology and migrate as evidenced by cell scattering (Fig 4Ac). Pretreatment of Caov3 cells with the EGFR inhibitor AG1478 blocked both the transition to a spindle-shaped morphology and cell scattering, demonstrating that EGF-induced changes were mediated by EGFR (Fig 4Ad). OVCA420 cells, which grow as colonies that are less clearly defined than Caov3 cell colonies, also acquired a spindleshaped morphology and underwent cell scattering in response to EGF, which was blocked by antagonism of EGFR (Fig 4A,  e-h).
To examine the effect of EGF on the cellular localization of CDCP1, we performed confocal microscopy on unstimulated and EGF-stimulated Caov3 and OVCA420 cells staining for CDCP1 (green), F-actin (red), and cell nuclei (blue). In unstimulated Caov3 and OVCA420 cells, CDCP1 was largely restricted to the cell membrane at cell-cell contacts where it co-localized with F-actin (Fig 4B, a, b, e, and f; yellow). However, 24 h after EGF treatment, intense punctuate CDCP1 staining was seen in the cytoplasm and also on the outside of the F-actin boundary of cells along filopodia that developed in response to EGF (Fig 4B, c, d, g, and h; green). The relocalization of CDCP1 from cell-cell junctions to filopodia is consistent with this protein having a function in EGF-induced cell migration.

Disruption of CDCP1 Expression and Function Reduces EGFinduced Cell Morphology Changes and Migration-Two
approaches were employed to directly address the role of  MARCH 23, 2012 • VOLUME 287 • NUMBER 13 JOURNAL OF BIOLOGICAL CHEMISTRY 9797 CDCP1 in EGF-induced cell migration. First, we interfered with the up-regulation of CDCP1 expression induced by EGF. Second, we disrupted CDCP1 function using a previously characterized monoclonal anti-CDCP1 antibody, 10D7, that blocks the ability of cancer cells to disseminate in chicken embryo and mouse models of metastasis (36,37).

CDCP1 in EGFR-induced Cell Migration
To interfere with EGF-induced up-regulation of CDCP1, we generated polyclonal pools of Caov3 and OVCA420 cells stably infected with either a CDCP1 shRNA construct or a scramble control construct. The CDCP1 shRNA construct abolished CDCP1 expression in both Caov3 and OVCA420 cells compared with scramble controls and blocked the ability of EGF to induce, compared with the scramble control, up-regulation of CDCP1 at both 0.2 and 24 h (Fig. 5A). Microscopy analysis indicated that loss of CDCP1 had little impact on Caov3 and OVCA420 cell morphology (Fig. 5B, ϪEGF, compare Scramble with shCDCP1). However, silencing of this protein reduced the ability of these cells to migrate and acquire an EGF-induced elongated spindle-shaped morphology including the acquisition of filopodia (Fig. 5B, ϩEGF, compare Scramble with shCDCP1). This qualitative analysis was confirmed by quantitative analysis of the distance between cells in three independent experiments which showed that silencing of CDCP1 blocked Caov3 and OCA420 cell migration induced by EGF (Fig. 5C). These data suggest that the role of CDCP1 in EGF/EGFR-induced ovarian cancer cell morphological changes and increased migration requires its up-regulation at the transcriptional level.
In experiments to assess the effect of blockade of CDCP1 function on EGF/EGFR-induced cell migration, Caov3 and OVCA420 cells in the presence and absence of the anti-CDCP1 antibody 10D7 were imaged by time-lapse microscopy, collecting images every 15 min up to 12 h for OVCA420 cells and 24 h for Caov3 cells. Consistent with data in Fig. 4A, images from time-lapse microscopy demonstrated that EGF treatment in the presence of control IgG induces a spindle-shaped morphology and cell scattering in both cell lines (Fig. 6, A and B, IgGϩEGF). These changes were inhibited by concomitant treatment with 10D7, consistent with a role for CDCP1 in EGF/EGFR-induced cell migration (Fig. 6, A and B). In Caov3 cells antibody blockade of cell morphological changes and migration was first apparent ϳ8 h after commencement of EGF treatment (Fig. 6A,  10D7ϩEGF, 8 h). Quantitative analysis of at least 80 cells per treatment group showed that at 24 h EGF in the presence of IgG induced an increase of ϳ15-fold in migration over IgG only, and this was decreased to an ϳ8-fold increase by 10D7 treatment (Fig. 6A, graph). In OVCA420 cells blockade was first apparent about 4 h after commencement of EGF treatment (Fig. 6B, 10D7ϩEGF, 4 h). Quantitative analysis of at least 80 cells per treatment group showed that at 12 h EGF treatment in the presence of IgG induced an increase of ϳ16fold in migration over IgG only, and this was decreased to an ϳ6-fold increase by 10D7 treatment (Fig. 6B, graph). These data support a functional role for CDCP1 in the morphological changes and increased migration of ovarian cancer Caov3 and OVCA420 cells induced by EGF/EGFR. CDCP1 Is Expressed by Primary and Metastatic Ovarian Tumors-To examine whether CDCP1 is expressed in ovarian cancer we performed immunohistochemical analysis on a series of ovarian cancers representing the progression from benign to malignant to metastatic disease including benign tumors from three patients, primary serous epithelial ovarian cancers from three patients, and metastases from three patients. Representative images of anti-CDCP1 antibody and control staining are shown in Fig. 7. Little expression of CDCP1 was seen in the three benign serous adenomas, whereas higher levels of CDCP1 staining were seen in primary (P) well differentiated (Grade 1) serous tumors (compare Fig. 7, A and B with  C and D). The most intense CDCP1 staining was seen in the three metastatic (M) ovarian tumors (Fig. 7, E-G), including intense plasma membrane and cytoplasmic staining apparent in two of these metastases (Fig. 7, E and F). No staining was seen in control slides in which the primary antibody was replaced with rabbit IgG (Fig. 7H). These data indicate that CDCP1 is expressed in epithelial ovarian tumors. A larger sample size is required to examine whether changes in expression of CDCP1 occur consistently during progression from adenoma to malignant ovarian cancer.   MARCH 23, 2012 • VOLUME 287 • NUMBER 13

DISCUSSION
The EGF/EGFR axis has well studied roles in cell migration and cancer progression (25). Our data, which are summarized in Fig. 8, define a new mechanism involving the cell surface glycoprotein CDCP1, which regulates this cell migration signaling axis. We have demonstrated that disruption of CDCP1 either by silencing or the use of a function blocking antibody substantially reduces the ability of EGF/EGFR to induce migration of ovarian cancer Caov3 and OVCA420 cells. Our data indicate that EGF/EGFR-induced cell migration is preceded by increased expression of CDCP1, and this increased expression is accompanied by relocalization of CDCP1 from cell-cell junctions to punctate structures on filopodia. Studies with pharmacological agents indicate that inhibition of ERK but not Src signaling blocks up-regulation of CDCP1, indicating that the RAS/RAF/MEK/ERK pathway is required downstream of EGF/ EGFR to induce increased expression of CDCP1. Our observation from immunohistochemical analysis that CDCP1 is expressed by epithelial ovarian tumors has potential implications for treatment of this and other malignancies driven by EGF/EGFR signaling. In particular, it is possible that disruption of CDCP1 function may be a useful approach to limit the dissemination of cancer cells in vivo, including ovarian, lung, and colorectal cancer cells that are resistant to anti-EGFR therapies due to activating mutations in downstream signaling effectors such as RAS and RAF (25,26,50). It will be important to perform further analyses of larger patient cohorts to examine how CDCP1 expression alters with changes in expression and/or mutation of EGFR pathway components and whether targeting of CDCP1 represents a rational approach to treat these cancers.
EGF/EGFR regulation of CDCP1 is the third reported CDCP1 regulating pathway. The other pathways are: (i) SFK phosphorylation and (ii) hypoxia. SFK phosphorylation of CDCP1 is important for cancer cell survival, migration, and invasion in vitro and in vivo (33,34,37,(51)(52)(53) and is activated in at least three cellular settings: cell de-adhesion, increased serine protease activity, and overexpression of CDCP1 (45). Mechanistically it involves SFK phosphorylation of CDCP1 at tyrosines 734, 734, and 762, coupling of PKC␦ at this last phosphotyrosine, and reciprocal activation of phosphorylation of SFKs at Tyr-416 (34). Although we saw that pSrc-Tyr-416 levels increased in parallel with the expression of CDCP1 induced by EGF/EGFR (compare Fig. 3, A and B, with Fig. 5, A and B), we did not observe any change in phosphorylation of CDCP1-Tyr-734 during EGF/EGFR-induced cell migration (data not shown). This indicates that mechanistically EGF/EGFR-mediated cell migration does not require phosphorylation of CDCP1 and suggests that in ovarian cancer cells and potentially other cancer lines, signaling via EGF/EGFR and SFKs represents distinct CDCP1 regulating pathways.
In the second reported CDCP1-regulating pathway, expression of this protein is induced in kidney cancer cell lines by hypoxic conditions via a mechanism requiring the transcription factors HIF-1␣ and -2␣ (38). As EGFR is upstream of HIF signaling in prostate (54) and kidney (55) cancer cells, it is possible that EGF/EGFR and HIF pathways may converge to regulate CDCP1 function. These reports on SFK and HIF regulation of CDCP1 and this work begin to shed light on mechanisms controlling CDCP1 function and highlight the potential complexity of these pathways.
In our in vitro cell assays we were able to isolate EGF/EGFRregulated effects from other cellular modulators, such as proteolysis, by growing cells in serum-free conditions. One of the interesting observations from these experiments was that there was no evidence of the 70-kDa CDCP1 species previously shown to be generated from the 135-kDa precursor protein by the in vitro action of serine proteases such as trypsin (31) and matriptase (32,47) and in vivo by urokinase (31) and plasmin (36). However, in more complex cellular settings, such as serum containing in vitro cell culture and in animal models, EGF/ EGFR regulation of CDCP1 will occur contemporaneously with other regulatory mechanisms such as proteolysis and hypoxia. To understand how CDCP1 functions mechanistically to support EGF/EGFR-induced cell migration, we are currently exploring whether this pathway functions synergistically or in competition with serine protease and hypoxic regulation of CDCP1.
EGF via its receptor activates downstream signaling pathways that modify the cytoskeleton and initiate transcriptional and post-transcriptional regulation leading to altered cell plasticity and increased cell migration (56). We have shown that this signaling pathway not only initiates up-regulation of CDCP1 mRNA and protein via activation of ERK, but it also results in relocalization of CDCP1 from cell-cell junctions to punctate structures along filopodia. We propose that it is this repositioning of CDCP1 that is essential for its role in EGF/ EGFR-induced cell migration. Our antibody-mediated blockade of CDCP1 prevented it from fulfilling this critical migration-promoting role on filopdia. It is worth noting that although this antibody, 10D7, blocked ovarian cancer cell FIGURE 8. Schematic representation of pathways required for the role of CDCP1 in EGF/EGFR-induced cell morphological changes and increased cell migration. CDCP1 mediates, at least in part, EGF/EGFR-induced cell morphological changes and increased migration. EGF signals via EGFR (as indicated through use of the EGFR antagonist AG1478) to induce signaling via ERK that leads to up-regulation of CDCP1 mRNA and protein. Targeting of CDCP1 either via silencing of its mRNA or through use of an anti-CDCP1 function blocking antibody is effective at reducing these EGF-induced effects. Approaches that disrupt CDCP1 function may be useful to inhibit progression of malignancies driven by EGFR signaling, such as ovarian cancer, and those resistant to anti-EGFR drugs because of activating mutations in the RAS/RAF/ MEK/ERK pathway. migration, it had no impact on cell survival (not shown), in contrast with data from in vivo systems where this antibody caused apoptosis of cells escaping from blood vessels (36,37). This would suggest that the ability of this antibody to induce cell death via blockade of CDCP1 function is context-dependent. In in vivo settings where CDCP1 promotes cell survival, antibody 10D7 blockade of this function induces cell death; however, in vitro this antibody blocks the role of CDCP1 in EGF/EGFR-induced cell migration without impacting on cell survival. This supports the proposal that the molecular mechanism regulating the role of CDCP1 in cell survival (SFK signaling) is distinct from the pathway promoting cell survival (EGF/ EGFR signaling).
Interestingly, although EGF/EGFR signaling clearly altered the plasticity of ovarian cancer Caov3 and OVCA420 cells, resulting in loss of cell contact with neighbors, elongation, and scattering (Fig. 4), this was not accompanied by changes in expression of markers associated with an EMT (Fig. 2B). This contrasts with two other ovarian cancer lines, OVCA433 and SKOV-3 cells, that undergo an EMT in response to EGF/EGFR activation as evidenced by changes in marker expression including decreased E-cadherin (21) and increased N-cadherin (22). EMT is thought to be required for the cell migration essential for a range of physiological processes as well as for progression of a number of cancers (57). However, the involvement of EMT in ovarian cancer is not clear because of the unique mixed epithelial and mesenchymal features of normal ovarian epithelium including low expression of the epithelial marker E-cadherin and expression of the mesenchymal markers keratin, vimentin, and N-cadherin (17). In addition, analysis of patient samples indicates that in contrast with many other cancers, ovarian cancers express high levels of E-cadherin mRNA and protein (18,58,59). This lack of clarity on the role of EMT in ovarian cancer is also highlighted by cell line studies. For example, although some E-cadherin-expressing ovarian cancer cells can be invasive in vitro and in vivo (60,61), loss of expression of this protein promotes invasiveness of other ovarian cancer cell lines (39). Our data indicate that disruption of CDCP1 blocks the EGF/EGFR-induced migration of two ovarian cancer cell lines that do not show altered expression of EMT markers. It will be interesting to examine whether blockade of this protein can also reduce the EGF/EGFR-induced migration of ovarian cancer cells that do undergo EMT.
In conclusion, these data highlight a novel role for the cell surface glycoprotein CDCP1 in EGF/EGFR-induced cell migration and indicate that targeting of CDCP1 may represent a rational approach to inhibit progression of cancers driven by this pathway.