Meprinα Transactivates the Epidermal Growth Factor Receptor (EGFR) via Ligand Shedding, thereby Enhancing Colorectal Cancer Cell Proliferation and Migration*

Background: EGFR signaling pathway activation is a crucial step in colorectal cancer tumor progression. Results: Meprinα sheds the epidermal growth factor ligands EGF and TGFα. Phosphorylation of EGFR and ERK1/2 is increased and cell proliferation and migration is enhanced after stimulation with meprinα. Conclusion: Meprinα transactivates the EGFR by proteolytic processing of TGFα and EGF. Significance: Meprinα may be a therapeutic target in colorectal cancer treatment. Meprinα, an astacin-type metalloprotease is overexpressed in colorectal cancer cells and is secreted in a non-polarized fashion, leading to the accumulation of meprinα in the tumor stroma. The transition from normal colonocytes to colorectal cancer correlates with increased meprinα activity at primary tumor sites. A role for meprinα in invasion and metastatic dissemination is supported by its pro-angiogenic and pro-migratory activity. In the present study, we provide evidence for a meprinα-mediated transactivation of the EGFR signaling pathway and suggest that this mechanism is involved in colorectal cancer progression. Using alkaline phosphatase-tagged EGFR ligands and an ELISA assay, we demonstrate that meprinα is capable of shedding epidermal growth factor (EGF) and transforming growth factor-α (TGFα) from the plasma membrane. Shedding was abrogated using actinonin, an inhibitor for meprinα. The physiological effects of meprinα-mediated shedding of EGF and TGFα were investigated with human colorectal adenocarcinoma cells (Caco-2). Proteolytically active meprinα leads to an increase in EGFR and ERK1/2 phosphorylation and subsequently enhances cell proliferation and migration. In conclusion, the implication of meprinα in the EGFR/MAPK signaling pathway indicates a role of meprinα in colorectal cancer progression.

The epidermal growth factor receptor (EGFR) 2 signaling pathway has critical functions in normal cellular processes such as differentiation, proliferation, migration, and the modulation of apoptosis, but it is also crucial in the pathophysiology of hyperproliferative diseases such as cancer (1). Colorectal cancer is the third most prevalent cancer and the second leading cause of cancer related deaths, worldwide (2). The analysis of tumor samples by immunohistochemistry has shown that the EGFR protein is overexpressed in 65-75% of colorectal tumors (3). The EGFR is a transmembrane receptor that is activated after binding of specific extracellular protein ligands, including epidermal growth factor (EGF) (4), heparin-binding EGF-like growth factor (HB-EGF) (5), transforming growth factor-␣ (TGF␣) (6), betacellulin (7), amphiregulin (8), epiregulin (9), and epigen (10). The ligands are structurally and functionally related type I trans-membrane proteins that are shed after their presentation on the cell surface by an extracellular metalloprotease (11,12). TACE/ADAM17 (a disintegrin and metalloprotease) has been identified as the main sheddase of TGF␣, HB-EGF, amphiregulin, epiregulin, and epigen (13)(14)(15)(16)(17), and ADAM10 as the main sheddase of EGF and betacellulin (16). Furthermore, ADAM8, -9, -12, and -19, have been reported to contribute to shedding of EGFR ligands when overexpressed or deregulated, which is highly relevant in inflammation and cancer (18). Shedding of EGFR ligands by ADAMs is associated with diseases such as cancer, neurological and cardiovascular diseases, asthma, infection, and inflammation (19 -21).
Recently, it has been shown that meprin␣ is involved in the activation of the EGFR, via release of TGF␣, in human bronchial epithelial cells, 16HBE14o (22). Meprins and ADAMs both belong to the M12 family of metalloproteases (MEROPS, proteinase database) (23). Meprins are members of the astacinfamily (M12A) and ADAMs of the adamalysin-family (M12B). Meprin was first discovered in proximal epithelial cells of mouse kidney in 1981 by Beynon et al. (24). In addition, two other groups discovered the same metalloprotease in the early 1980s: Sterchi et al., in 1982, described the enzyme PABA peptide hydrolase (PPH) in microvillar membranes of human small intestinal epithelial cells (25), and Kenny et al., in 1987, found the endopeptidase-2 in the microvillar membrane of proximal epithelial cells of rat kidney (26). This metalloprotease is now known as "meprin." There are two evolutionary related isoforms: Meprin␣ and meprin␤. Both are synthesized as type I transmembrane proteins in the endoplasmic reticulum (27,28). The membrane anchor of meprin␣ is removed intracellularly leading to the secretion of this isoform from cells, whereas meprin␤ remains an integral protein of the plasma membrane (29). However, meprin␣ may be retained at the plasma membrane via covalent interaction with the transmembrane meprin␤ (30 -32). Meprin␣ is expressed in epithelial cells of the healthy colon mucosa where it is secreted apically into the colon lumen (33). In colorectal cancer, meprin␣ is released in a non-polarized fashion, leading to its accumulation in the tumor stroma (34,35). This aberrant secretion of meprin␣ into the tumor stroma exposes matrix components and other stromal elements to an increased proteolytic potential (35). Once secreted, meprin␣ is activated in vitro and in the gut lumen by the removal of the pro-peptide through trypsin (28). An alternative activation mechanism has been suggested in colorectal cancer. In colon carcinoma cells (Caco-2), basolaterally secreted meprin␣ is activated by plasmin, which in turn, is activated by the fibroblast-derived urokinase-type plasminogen activator (36). Meprin␣ has been demonstrated to have promigratory and pro-angiogenic effects in colorectal cancer, and thus may be involved in the transition from benign growth (adenomas) to malignant primary tumors (37,38).
We investigated the molecular mechanisms by which meprin␣ may influence tumor progression. For the first time we demonstrate that meprin␣ is able to shed EGF from the plasma membrane, resulting in the transactivation of EGFR signaling pathway and enhancement of Caco-2 cell proliferation and migration. We also confirm the shedding of TGF␣ by meprin␣.

EXPERIMENTAL PROCEDURES
Antibodies and Recombinant Protein-Antibodies specific for total EGFR (monoclonal rabbit antibody) and phospho-EGFR Y1068 (monoclonal rabbit antibody) were purchased from Epitomics (Burlingame, CA); antibodies specific for total ERK1/2 (monoclonal mouse antibody) and phospho-ERK1/2 (polyclonal rabbit antibody) were from Santa Cruz Biotechnology (Heidelberg, Germany). Horseradish peroxidase-linked anti-rabbit and anti-mouse secondary antibodies were obtained from Dako Cytomation (Denmark). Recombinant active human meprin␣ and recombinant human pro-meprin␣ were generated using a baculovirus expression system in insect cells as previously described (39,40).
Reagents-Cell culture media and all supplements were purchased from Invitrogen (Basel, Switzerland). All reagents for gel electrophoresis were obtained from Bio-Rad (Reinach, Switzerland). Complete EDTA-free protease inhibitor mixture tablets, PhosStop phosphatase inhibitor mixture tablets, and NBT/ BCIP ready-to-use tablets were purchased from Roche Applied Sciences (Rotkreuz, Switzerland). MEK inhibitor U0126 was obtained from Promega (Dübendorf, Switzerland). EGF and TGF␣ neutralizing antibodies were purchased from R&D (Abingdon, UK). All other reagents were purchased from Sigma.
Transfection was performed using PEI (Chemie Brunschwig, Basel, Switzerland). 1.5 ϫ 10 5 cells per well were seeded in a 12-well plate, 24 h before transfection. Transfection mixture (100 l of 150 mM NaCl containing 4 l of PEI plus 100 l of 150 mM NaCl containing 1.5 g DNA) was incubated for 30 min at room temperature, and then added to the cells.
Conditioned Medium and Meprin␣ Activity Assay-Caco-2 cells, grown 7 days over confluency, were cultured in serum-free medium for 16 h. The medium was collected (referred as conditioned medium) and accumulated meprin␣ was activated using trypsin (20 g/ml) for 2 h at 37°C. Trypsin was inhibited using soybean trypsin inhibitor (50 g/ml). Active meprin␣ was inhibited using actinonin (100 nM, in excess), a meprin␣ inhibitor.
Meprin␣ activity in conditioned medium or in medium containing recombinant active meprin␣ was verified using the substrate N-benzoyl-L-tyrosyl-p-aminobenzoic acid (PABA peptide) as described previously (30).
Ectodomain Shedding Assay-24 h after transfection, MDCK/Caco-2 cells were stimulated with serum-free medium containing either 1 g/ml recombinant active meprin␣ or 1 g/ml recombinant meprin␣ inhibited with 100 nM actinonin. After stimulation, medium was collected and centrifuged for 30 min at 4°C at maximum speed. Cells were washed twice with phosphate-buffered saline (PBS) followed by lysis on ice in 0.5 ml lysis buffer (25 mM Tris-HCl pH 8, 50 mM sodium chloride, 1% IGEPAL, 1% sodium deoxycholate, with complete EDTA-free protease inhibitor mixture tablets) for 30 min. Cells were scraped off, and cell debris was removed by centrifugation. Supernatants and lysates were kept on ice at all the times. Each data point was generated from two consecutive AP activity measurements shed from a single transfected well (n ϭ 3 experiments).
Detection of Alkaline Phosphatase-For the spectrophotometric detection of alkaline phosphatase (AP), 100 l of collected medium or lysate were mixed with 100 l 4-nitro-phenyl phosphate (2 mg/ml) in AP buffer (100 mM Tris, 100 mM NaCl, 20 mM MgCl 2 , pH 9.5) in a 96 well plate. After incubation at 37°C absorbance was measured at 405 nm in an ELISA reader. Absorbance was measured at different time points within a linear range (OD Ͻ 0.8) up to a maximum incubation time of 5 h. The total amount of AP measured from a single well, was used to normalize the absorbance value obtained for the supernatant of a certain condition.
For in-gel detection, AP in cell culture supernatants was concentrated using ConA beads. After elution with 50 mM Tris, pH 8.0, 0.5 M ␣-D-methyl-mannopyranoside, the AP-tagged EGFR ligands were loaded on a SDS-polyacrylamide gel. The SDS-gel was incubated in 2.5% Triton X-100 followed by incubation in AP buffer. AP was visualized using NBT/BCIP as substrate.
Phosphorylation of EGFR and ERK1/2-Caco-2 cells, seeded at a density of 5 ϫ 10 5 cells per 6 cm dish, were stimulated for 0, 5, 15, 30, and 60 min with either control medium, 1 g/ml recombinant active meprin␣, 1 g/ml recombinant pro-meprin␣, or 100 ng/ml EGF (positive control). Phosphorylation induced by recombinant active meprin␣, recombinant pro-meprin␣, or EGF was inhibited with 2 g/ml neutralizing EGF and TGF␣ antibodies, 10 M EGFR inhibitor AG1478, or 10 M MEK inhibitor U0126. Cells were pretreated with the inhibitors 30 min before stimulation. After stimulation, cells were washed once with PBS followed by lysis on ice for 30 min in 1 ml of cell lysis buffer (Epitomics, Burlingame, CA) supplemented with protease and phosphatase inhibitors. Cell debris was removed by centrifugation and the protein content in the lysates was determined using the bicinchoninic acid (BCA) protein assay (Pierce).
Cell Treatment for Proliferation and Migration Experiments-For proliferation and migration experiments Caco-2 cells were stimulated with conditioned medium containing activated meprin␣, inhibited meprin␣, or 100 ng/ml EGF (positive control). Inhibitors were added to media containing active meprin␣ at the beginning of the treatment (2 g/ml neutralizing EGF and TGF␣ antibodies, 10 M EGFR inhibitor AG1478, or 10 M MEK inhibitor U0126). Cells were pretreated with neutralizing antibodies and EGFR inhibitor for 2 h.
Alamar Blue Cell Proliferation Assay-Alamar Blue uses the natural reducing power of living cells to convert resazurin to the fluorescent molecule, resorufin (42). Caco-2 cells were seeded at a density of 1000 cells per well in a 96-well plate. After 48 h of incubation, cells were washed twice using phenol redfree medium, followed by stimulation for 24 h as described above. For the last 3 h, Alamar Blue (43) solution was added to a final concentration of 10 g/ml. Fluorescence at 595 nm was measured directly (0 h) and 3 h after addition in a multilabel plate reader (2300 EnSpire multilabel reader; Perkin-Elmer, Turku, Finland). Values obtained at time point 0 h were subtracted from those obtained at time point 3 h (20 replicates/ condition, n ϭ 3 experiments).
Cell Titer Glo Cell Viability Assay-This assay is a method to determine the number of viable cells in culture based on quantification of the ATP present. Consequently, ATP levels represent the number of metabolically active cells (44). 1000 Caco-2 cells/well were seeded in a 96-well plate. After 48 h, cells were serum-starved overnight followed by 24 h treatment with the different stimuli. Subsequently, 100 l of Cell Titer Glo reagent were added to each well, cells were incubated for 10 min in the dark, and luminescence was measured in a 2300 EnSpire multilabel plate reader (3 replicates/condition, n ϭ 3 experiments).
BrdU Incorporation-Cell proliferation was also determined by bromodeoxyuridine (BrdU) incorporation analysis. We used the In situ cell proliferation kit (Roche Applied Sciences; Rotkreuz, Switzerland) according to the manufacturer's instructions. Briefly, Caco-2 cells were plated at a density of 2 ϫ 10 4 cells per well of an 8 chambers culture slide (Lab-Tek). After 48 h, cells were serum-starved overnight, followed by treatment for 24 h with the different stimuli. During the last 90 min of the treatment, BrdU at a final concentration of 10 M was added to the medium to allow BrdU incorporation. Cells were fixed in 70% ethanol for 45 min at room temperature, and incubated with anti-BrdU antibody in the presence of nuclease for DNA denaturation. Cells were counterstained with 5 g/ml DAPI for 5 min. BrdU incorporation into cellular DNA was visualized by fluorescence microscopy. In three independent experiments a total number of 12 high-power fields (40ϫ), and at least 800 cells per condition were analyzed. The proliferation rate was determined as a proportion of the total DAPI-positive nuclei. The value for untreated control cells was arbitrarily set to 0.
In Vitro Wound-healing Assay-2 ϫ 10 5 Caco-2 cells per well were seeded in a 12-well plate and grown to confluency. The cell monolayer was wounded by scratching, using a 200 l pipette tip. After washing with PBS the cells were incubated with the corresponding stimuli. At time points 0 h and 16 h the same positions along the scratch wound were photographed using an inverted-phase-contrast microscope (Nikon microscope TS100 fluorescence and video camera) and Adobe Photoshop was used for quantification of the scratch wound. Three measurements per scratch were performed (2 replicates/condition, n ϭ 3 experiments).
Transwell Migration Assay-5 ϫ 10 4 Caco-2 cells were seeded on top of transwell filters (polyethylene terephthalate (PET), 8 M pores, 24-well format) from BD Biosciences. Cells were allowed to grow for 48 h followed by serum starvation for 24 h in medium containing 1% FBS. Then, medium in the lower chamber was replaced by conditioned medium containing 20% FBS and the stimuli. Medium in filter inserts was replaced by serum-free conditioned medium containing the corresponding stimuli. Cells were treated for 36 h and at the end of the treatment cells were washed twice with PBS followed by fixation for 15 min using 4% paraformaldehyde. Cells on the upper side of the transwell filters were removed with a cotton swab and cells on the lower side were stained for 5 min with 5 g/ml DAPI. Pictures were taken (Nikon microscope TS100) and migrated cells were counted using Image J software (Wayne Rasband, NIH) (3 replicates/condition, n ϭ 3 experiments).
Statistical Analysis-Data were analyzed using PRISM 5.0 software package (GraphPad, San Diego, CA). Results are shown as the mean Ϯ S.E. Statistical differences between two groups were determined by unpaired Student's t test. As significant differences considered were p Ͻ 0.05.

RESULTS
EGF and TGF␣ Are Shed by Meprin␣-To investigate the role of meprin␣ in ectodomain shedding of EGF and TGF␣, a cell-based assay using AP-tagged EGFR ligands was used. Shedding of EGF and to a lesser extent, TGF␣ was significantly stim-ulated in MDCK cells (p Ͻ 0.001; p Ͻ 0.05; Fig. 1, A and B) as well as in Caco-2 cells (p Ͻ 0.001; p Ͻ 0.01; Fig. 1, D and E) by meprin␣. Addition of the meprin␣ inhibitor actinonin showed a significant decrease in ligand shedding in MDCK cells (p Ͻ 0.01; p Ͻ 0.05; Fig. 1, A and B) and in Caco-2 cells (p Ͻ 0.01; p Ͻ 0.01; Fig. 1, D and E). Actinonin did not influence constitutive shedding, suggesting that this is catalyzed by a different metalloprotease (data not shown). Results obtained with the spectrophotometric assay were confirmed by in-gel detection of APtagged ligands, using NBT/BCIP as a substrate for the alkaline phosphatase (Fig. 1, C and F).
Endogenous EGF released from Caco-2 cells into the medium was quantified by EGF ELISA (Fig. 1G). Stimulation of Caco-2 cells with recombinant active meprin␣ resulted in a significant increase of soluble EGF compared with control values (p Ͻ 0.001) and recombinant pro-meprin␣, the inactive form of meprin␣ (p Ͻ 0.001). Quantification of soluble TGF␣ upon meprin␣ activation by ELISA has been shown before (22). Together these data suggest that meprin␣ acts as a sheddase for EGF and TGF␣.

Meprin␣ Induces EGFR and ERK1/2 Phosphorylation-EGFR
ligands, once released from the plasma membrane bind to the EGFR, which in turn is phosphorylated at several amino acid positions (Tyr-992, Tyr-1068, Tyr-1086, Tyr-1148, and Tyr-1173) (45). To investigate the effect of meprin␣ on the phosphorylation of EGFR via EGF or TGF␣ shedding, Caco-2 cells were stimulated for various periods of time with either medium alone, recombinant active meprin␣, recombinant pro-meprin␣, or with EGF (positive control). Fig. 2A shows representative Western blots of the phosphorylation experiment, and Fig. 2B shows the densitometric analysis of four individual experiments. In the latter, control values were subtracted and the values were normalized against total EGFR. After 5 min of treatment with meprin␣ ( Fig. 2A, lane 2), an increase in phosphorylation was observed. A maximum activation of EGFR was achieved between 15 min and 30 min of treatment ( Fig. 2A,  lanes 3 and 4), followed by a decrease (Fig. 2A, lane 5). Nonactive pro-meprin␣ showed only a very slight increase in phosphorylation of the EGFR. Minor amounts of pro-meprin␣ might be activated over time, for instance by plasmin or kallikreins (KLKs) (40,46). Treatment with recombinant EGF resulted in a similar EGFR activation pattern to meprin␣ albeit with a stronger signal, which may have been due to the relatively high concentration of EGF used. These data indicate that the proteolytic activity of meprin␣ is required for EGFR phosphorylation.
EGFR phosphorylation leads to the activation of intracellular pathways, such as the mitogen-activated protein kinase (MAPK) pathway. The classical MAP kinases, extracellular-signal-regulated kinases 1 and 2 (ERK1/2), are intracellular signaling molecules that are preferentially activated in response to growth factors and phorbol esters (47). To determine whether ERK1/2 are transactivated upon treatment of Caco-2 cells with meprin␣, the lysates obtained from the EGFR phosphorylation experiment were also analyzed for phosphorylated ERK1/2 (Fig. 2C). Phosphorylation was calculated by densitometric measurements (Fig. 2D). Control values were subtracted and data were normalized against total ERK1/2. Similar to the treatment with EGF, stimulation with meprin␣ led to a peak in phosphorylation after 5 min (Fig. 2C, lane 2), which was attenuated over time. Pro-meprin␣ as well as the negative control showed an increase in phosphorylation at time point 5 min, although to a lesser degree (Fig. 2C, lane 2). We assume that this ERK1/2 phosphorylation after 5 min is a transient process that might be caused by the change of culture medium. Taken together, we conclude that active meprin␣ leads to the activation and phosphorylation of EGFR and consequently, transactivates the MAPK pathway, which culminates in ERK1/2 phosphorylation.
EGFR and ERK1/2 Phosphorylation Are Meprin␣dependent-To analyze whether the EGFR/MAPK signaling pathway is activated by meprin␣ via EGF and TGF␣ shedding, phosphorylation experiments using neutralizing EGF and TGF␣ antibodies were performed (Fig. 3A). Caco-2 cells were stimulated for 5 or 15 min with meprin␣, pro-meprin␣ or EGF in the absence or presence of the neutralizing antibodies. In the absence of EGF and TGF␣ neutralizing antibodies, EGFR and ERK1/2 were phosphorylated when stimulated with meprin␣ or EGF but not with pro-meprin␣. Cells treated with neutralizing EGF and TGF␣ antibodies showed EGFR and ERK1/2 phosphorylation reduced to control levels, after stimulation with meprin␣. After stimulation with EGF, EGFR and ERK1/2 remained phosphorylated to a certain extent in the presence of neutralizing antibodies. This may be the result of ligand excess compared with the amount of antibodies used. EGFR and ERK1/2 phosphorylation remained the same after stimulation with pro-meprin␣. Total EGFR and ERK1/2 were not affected by the neutralizing antibodies. We conclude, that EGFR transactivation by meprin␣ occurs via shedding of EGF and TGF␣ from the plasma membrane by meprin␣.
We wondered if meprin␣-induced ERK1/2 phosphorylation was entirely mediated by transactivation of EGFR. For this rea-son, we tested the effect of the EGFR inhibitor AG1478, (Fig.  3B). Phosphorylation of EGFR and ERK1/2 was detected after stimulation of Caco-2 cells with meprin␣ and EGF but not after stimulation with pro-meprin␣. Inhibition of EGFR led to less EGFR phosphorylation while ERK1/2 phosphorylation was completely abrogated. Hence, meprin␣ activates ERK1/2 mostly via EGFR. We further confirmed that the MEK inhibitor U0126, which was later used in our functional assays, is a potent agent that completely abrogates meprin␣-induced ERK1/2 phosphorylation (Fig. 3C).
Meprin␣ Enhances Caco-2 Cell Proliferation-Activation of the EGFR pathway plays an important role in the regulation of cellular processes such as cell proliferation and migration. The functional relevance of meprin␣-dependent EGFR/ERK1/2 signaling on cell proliferation was studied in three independent assays: Alamar Blue, Cell Titer Glo cell viability assay, and BrdU incorporation. Cells were stimulated with conditioned medium containing activated meprin␣, inhibited meprin␣, or EGF. In addition, cell proliferation, induced by meprin␣, was analyzed in the presence of neutralizing antibodies for EGF and TGF␣, or inhibitors against EGFR and MEK. The cell proliferation rate for Alamar Blue and Cell Titer Glo experiments is shown in Fig.  4, A and B. A significant increase in proliferation was detected when cells were treated with active meprin␣ (p Ͻ 0.001; p Ͻ 0.001). Compared with that, inhibited meprin␣ reduced cellproliferation to a level slightly above that of controls (p Ͻ 0.001; p Ͻ 0.01). This may be due to incomplete inhibition of meprin␣ by actinonin. 3 EGF stimulation that was used as a positive control exhibited a comparable cell proliferation rate to meprin␣. In both assays, the increase in cell proliferation monitored after stimulation with meprin␣ was significantly reduced in the presence of neutralizing EGF and TGF␣ antibodies (p Ͻ 0.05; p Ͻ 0.01), EGFR inhibitor (p Ͻ 0.01; p Ͻ 0.01) or MEK inhibitor (p Ͻ 0.01; p Ͻ 0.01). DNA synthesis was quantified by BrdU uptake in Caco-2 cells after treatment with the different stimuli. Representative photographs are shown in Fig. 4D and quantification of the results of three independent experiments are shown in Fig. 4C. Nuclear BrdU labeling was significantly increased after stimulation with active meprin␣ (p Ͻ 0.001) and significantly decreased when treated with inhibited meprin␣ (p Ͻ 0.001). BrdU uptake was significantly reduced in the presence of neutralizing EGF and TGF␣ antibodies (p Ͻ 0.01), EGFR inhibitor (p Ͻ 0.01), or MEK inhibitor (p Ͻ 0.01). Altogether, these data demonstrate that the proliferative effect of meprin␣ in Caco-2 cells is regulated through the EGFR/MAPK pathway.
Migration of Caco-2 Cells Is Enhanced by Meprin␣ Activity-The effect of meprin␣ on the migration behavior of Caco-2 cells was assessed using an in vitro wound-healing assay. A scratch was induced to confluent Caco-2 cells using a 200-l pipette tip. We compared migration of Caco-2 cells treated with conditioned medium containing activated meprin␣, inhibited meprin␣, or EGF (positive control). Furthermore, the effect of inhibitors (neutralizing EGF and TGF␣ antibodies, EGFR inhibitor, or MEK inhibitor) on meprin␣-induced migration was analyzed. Representative photographs, taken at time 3 E. E. Sterchi, unpublished data. Caco-2 cells were stimulated for 5 or 15 min with 1 g/ml recombinant active meprin␣, 1 g/ml recombinant pro-meprin␣, or 100 ng/ml EGF in the presence or absence of neutralizing EGF and TGF␣ antibodies (nAB EGF/nAB TGF␣) (A), EGFR inhibitor AG1478 (B), or MEK inhibitor U0126 (C). Cells were pretreated with the inhibitors for 30 min. Phosphorylation of EGFR and ERK1/2 were determined using antibodies detecting the phosphorylated form of EGFR and ERK1/2. Total EGFR and ERK1/2 were used as loading control and were not affected by neutralizing EGF and TGF␣ antibodies, or EGFR and ERK1/2 inhibitor n ϭ 3. point 0 h and 16 h of the identical location, are shown in Fig. 5A. Quantification of the results of six separate experiments is shown in Fig. 5B. Under all conditions a closing of the wound was observed. A significant enhancement in wound closure was detected in cells exposed to active meprin␣ compared with inhibited meprin␣ and control values (p Ͻ 0.001, p Ͻ 0.01). EGF and TGF␣ inhibition through neutralizing antibodies as well as EGFR inhibition revealed a significant reduction in meprin␣-induced wound closure (p Ͻ 0.05, p Ͻ 0.05), as did ERK1/2 inhibition (p Ͻ 0.001).
The in vitro wound-healing assay represents a combination of cell migration and cell proliferation. To avoid the proliferative effect, we also performed a transwell migration assay. Caco-2 cells were cultivated on 8 M pore size filters in a 24-well culture plate with the same conditions as used for the in vitro wound-healing assay. Migrated cells were found under all conditions, but a significant increase in migration was monitored after stimulation with meprin␣ compared with control values (p Ͻ 0.001). Inhibited meprin␣ showed a significant decrease (p Ͻ 0.01; Fig. 5C) compared with active meprin␣, and stimulation with EGF led to a slightly higher increase in migration compared with meprin␣. The increase in migrated cells monitored after stimulation with meprin␣ was significantly reduced in the presence of neutralizing EGF and TGF␣ antibodies (p Ͻ 0.01), and EGFR inhibitor (p Ͻ 0.01), Inhibition of ERK1/2 led to less migration than the controls resulting in a negative value in the diagram (Fig. 5C). ERK1/2 is a key enzyme of many signaling cascades. Therefore, inhibition of ERK1/2 interferes with meprin␣-induced migration and most likely with supplemental pathways triggering cell migration. In conclusion, we show that meprin␣ activity enhances migration of Caco-2 cells and this effect is dependent on the transactivation of EGFR/MAPK signaling.

DISCUSSION
In this study, we set out to identify EGFR ligands as substrates for meprin␣. Our data demonstrate that human meprin␣ is effectively capable of shedding EGF. Additionally, we could confirm the shedding of TGF␣ by meprin␣, which was shown previously in lung epithelial cells (22). We also demonstrate that active meprin␣ transactivates the EGFR and ERK1/2 and subsequently increases Caco-2 cell proliferation and migration.
Shedding of EGF and TGF␣ was enhanced in cells treated with active recombinant meprin␣, and inversely, was reduced after inhibition of meprin␣ by actinonin, indicating that the proteolytic activity is required for shedding of EGF and TGF␣. Two other groups have demonstrated TGF␣ shedding by meprin in two individual assays (22,48). Back in 1991, Choudry et al., have identified the growth factor TGF␣ as an in vitro substrate for endopeptidase-2 (now known as meprin) (48). Using recombinant human TGF␣ and purified endopeptidase-2 from rat kidney, they showed that TGF␣ was processed in a time-dependent manner. In the presence of actinonin no hydrolysis was observed. Recently, Bergin et al. have analyzed shedding of TGF␣ by meprin␣ in human bronchial epithelial cells (16HBE14o-cells) (22). Using an ELISA assay, they found elevated levels of TGF␣ in the medium of cells after treatment with recombinant meprin␣. This effect was also inhibited by the addition of actinonin. The authors suggested that meprin␣ is activated by neutrophil elastase and, via TGF␣ precursor processing, induces Il-8 expression (22).
With our experimental setup using AP-tagged constructs of EGFR ligands we confirm TGF␣ shedding by meprin␣ in MDCK and Caco-2 cells. Furthermore, we show that EGF, another EGFR ligand, is also shed by meprin␣. Accordingly, we found increased levels of soluble EGF in the media (ELISA). Compared with TGF␣, EGF was shed by meprin␣ to a higher extent, and cleavage of both ligands was abrogated when meprin␣ was inhibited by actinonin. Other EGFR ligands were also analyzed as potential substrates for meprin␣. Epigen and betacellulin were not shed by meprin␣, HB-EGF, amphiregulin, and epiregulin were shed but shedding was not inhibited by actinonin, indicating that another protease was involved (data not shown).
Differentiated Caco-2 cells express meprin␣ endogenously, which makes them a preferred cell culture system for the analysis of meprin␣ function (35). Therefore, we used Caco-2 cells in our studies to investigate the consequences of meprin␣ expression on cell behavior in the context of colorectal cancer. We carried out shedding experiments using MDCK cells to confirm our results acquired with recombinant meprin␣ in a second cell line. MDCK cells do not express endogenous meprin␣. Nevertheless, the WT form in combination with recombinant meprin and stably transfected cell lines are widely used and established in meprin research (29,33).
Meprin consists of two homologous isoforms, meprin␣ and meprin␤. We have previously shown that meprin␤ cleaves and releases E-cadherin, which is considered to act as a tumor suppressor (49). Thus, we also considered meprin␤ as a potential sheddase for EGFR ligands. In contrast to meprin␣, meprin␤ is not implicated in the shedding of EGFR ligands (data not shown). Although, both isoforms have related cleavage sites, different substrate specificities have been described (50, 51) (Jefferson et al.,62), which most likely is the reason for their different behavior toward EGFR ligands. Additionally, meprin␣ and ␤ exhibit remarkable differences in their activation (46) and regulation by inhibitors (52) (Jefferson et al. 62). This certainly contributes to the different functions in cell proliferation and migration as observed previously (40). In the large intestine, only minor amounts of meprin␤ are expressed. Hence, most meprin␣ is released into the gut lumen in vivo and may be rapidly diluted. However in colorectal cancer, meprin␣ accumulates in the tumor stroma and persists close to the cell plasma membrane (34,35).
Two membrane-anchored metalloproteases, ADAM10 and ADAM17, were found to have critical roles in the release of EGFR ligands (14 -17, 53). Most likely the basal amounts of FIGURE 5. Migration of Caco-2 cells is increased through meprin␣ activity. Cells were treated with conditioned media containing activated meprin␣ (in the presence or absence of neutralizing EGF and TGF␣ antibodies (nAB EGF/nAB TGF␣), EGFR inhibitor AG1478, or MEK inhibitor U0126), inhibited meprin␣, or EGF. A/B, in vitro wound-healing assay. One representative of three independent experiments is shown in A, and in B quantification of the results of three separate assays is shown. C, transwell migration assay of Caco-2 cells. Cells fixed and stained with DAPI were counted. In C, the percentage increase compared with control values is shown and the control values were set as 0. n ϭ 3; ϮS.E., Student's t test ***, p Յ 0.001; **, p Յ 0.01; *, p Յ 0.05.
shed EGF and TGF␣ that we found in the ectodomain shedding assay are due to these enzymes. We inhibited ADAMs to exclude that these ADAMs interfere with the shedding of TGF␣ and EGF (data not shown). The inhibition of ADAMs led to a minimization of the constitutive shedding without affecting the shedding by meprin␣. In addition, meprin␣ generates a slightly smaller TGF␣ fragment (Fig. 1, C and F) to that obtained with ADAM17, indicating that the cleavage site of meprin␣ differs from that of ADAM17.
Mice lacking ADAM17 expression die perinatally and have a similar phenotype to TGF␣Ϫ/Ϫ mice (54). This points to ADAM17 as the main sheddase for TGF␣. In mouse cells ADAM10 is responsible for EGF shedding. Mice lacking ADAM10 die very early during embryogenesis and hence, determination of the physiological contribution to EGF signaling in animals remains to be determined (55). In cell-based assays, ADAM10 has been identified as a sheddase that can release TGF␣ almost as efficiently as its primary sheddase, in ADAM17Ϫ/Ϫ cells (16,18). This implies that the function of ADAMs may be replaced by other metalloproteases in tissues where ADAMs are not expressed or are not stimulated. We therefore propose a physiological role for meprin␣ in the local transactivation of the EGFR pathway.
Aberrant expression and/or activities of EGF family members and their receptors have been reported in solid tumors including colorectal cancer (56 -59). We have previously analyzed meprin␣ mRNA levels, protein expression, and proteolytic activity in colonic adenomas, primary tumors and liver metastases from colorectal cancer patients. Varied levels of meprin␣ mRNA were detected in all specimens. While expression of meprin␣ protein was very weak in adenomas, it was detected in primary tumor tissue as well as in liver metastases. In advanced primary tumors (UICC stages III and IV), subpopulations of cells were detected with a strong expression of meprin␣ protein. The activity of meprin␣ correlated mostly with protein expression except for liver metastases where activity was as low as in adenomas. This implies that the spreading of cancer cells correlates with increased meprin␣ protein as well as meprin␣ activity (37).
To further investigate the mechanism that leads to the spreading of colorectal cancer in response to meprin␣, we analyzed the ability of meprin␣ to transactivate the EGFR. Stimulation of Caco-2 cells with active meprin␣ showed a significant increase in EGFR phosphorylation compared with pro-meprin␣ (Fig. 2, A and B). Ligand binding to the EGFR activates two main intracellular pathways known to play a role in colorectal cancer: the mitogen-activated protein kinase (MAPK) pathway and the phosphatidylinositol-3-kinase-(PI3K-) protein kinase (AKT) pathway (60). In mammals five distinct groups of MAPKs have been characterized, with the most prominent being ERK1/2. Our data using neutralizing EGF and TGF␣ antibodies, as well as EGFR and ERK1/2 inhibitors (Fig.  3), imply that EGFR phosphorylation occurs after shedding of EGF or TGF␣ by meprin␣ and that this transactivation of EGFR leads to the activation of the MAPK signaling cascade and consequently to the phosphorylation of ERK1/2 (Figs. 2C and 3). ERK1/2 are preferentially activated in response to growth factors and have been implicated in cell migration, invasion, proliferation, angiogenesis, cell differentiation, and cell survival (60).
Proliferation and migration experiments were performed using conditioned media from Caco-2 cells, which endogenously express meprin␣. Meprin␣ is secreted as a zymogen into culture media due to constitutive proteolytic removal of the C-terminal transmembrane and cytosolic domain (34). Trypsin activation on cells is difficult to achieve as meprin␣ is removed with each washing step. Therefore culture medium was collected and accumulated meprin␣ was activated (see "Experimental Procedures").
Cell proliferation is a cellular response known to be enhanced in colorectal cancer upon EGFR activation (61). In three independent assays we demonstrate that Caco-2 cell proliferation is significantly increased in response to meprin␣. Further we demonstrate that inhibition of EGF and TGF␣, EGFR or ERK1/2 leads to a significant reduction in meprin␣-induced proliferation. Therefore, we conclude that the increase in cell proliferation occurs via transactivation of the EGFR/MAPK signaling pathway by a meprin␣-dependent mechanism.
We have previously demonstrated a pro-migratory effect induced by meprin␣ in MDCK cells using videomicroscopy (37). In that study plasmin-activated meprin␣ was used and migration was induced by hepatocyte growth factor (HGF). In the present study, we demonstrate increased migration of colorectal cancer cells in response to meprin␣ in an in vitro scratch assay and in a transwell migration assay (Fig. 5). Caco-2 cells endogenously express meprin␣, and represent a more natural environment for meprin␣ function. Inhibition of EGF and TGF␣, EGFR, or ERK1/2 showed a significant reduction in meprin␣-induced migration in both assays. Altogether, both approaches highlight that meprin␣ enhances cell migration via EGFR/MAPK signaling pathway.
The identification of EGFR ligands as substrates for meprin␣ and the known pro-migratory, pro-proliferative, and pro-angiogenic effects of meprin␣ in vitro, lay the foundation for further analysis on the role of meprin␣ in colorectal cancer. Experiments, including the use of a transgenic meprin␣ KO mouse model, will be necessary to address the biological relevance of meprin␣ in shedding of endogenous forms of EGFR ligands in vivo. Further, in vitro experiments focusing on the role of meprin␣ in metastasis of colorectal cancer cells would be of great interest. The transactivation of the EGFR may be critical for the transition from benign growth to malignant primary tumors in colorectal cancer and thus meprin␣ may be an interesting target for the design of drugs that modulate its activity.