Transactivation of the epidermal growth factor receptor in colonic epithelial cells by carbachol requires extracellular release of transforming growth factor-alpha.

We have shown previously that the muscarinic agonist, carbachol (CCh), transactivates the epidermal growth factor receptor (EGFr) via calmodulin, Pyk-2, and Src kinase activation. EGFr phosphorylation causes extracellular signal-regulated kinase (ERK) activation and inhibits CCh-stimulated chloride secretion across intestinal epithelial cells. Here we investigated whether CCh-stimulated EGFr transactivation involves EGFr ligand release. Pre-incubation of T(84) cell monolayers with a neutralizing antibody to the EGFr ligand binding domain decreased CCh-induced phosphorylation of EGFr and ERK. CCh-stimulated efflux of (86)Rb+ from T(84) cell monolayers, which parallels changes in chloride secretion, was potentiated by anti-EGFr pre-incubation. Anti-EGFr did not reduce CCh-stimulated Pyk-2 phosphorylation. Co-incubation with the Src kinase inhibitor PP2 and anti-EGFr had an additive inhibitory effect on CCh-induced ERK phosphorylation greater than either inhibitor alone. CCh caused the basolateral release of transforming growth factor alpha (TGF-alpha) into T(84) cell bathing media. A metalloproteinase inhibitor, WAY171318, reduced CCh-induced phosphorylation of ERK and completely blocked EGFr phosphorylation and TGF-alpha release. We conclude that CCh-stimulated EGFr transactivation and subsequent ERK activation, a pathway that limits CCh-induced chloride secretion, is mediated by metalloproteinase-dependent extracellular release of TGF-alpha and intracellular Src activation. These findings have important implications for our understanding of the role of growth factors in regulating epithelial ion secretion.

In the intestinal tract, chloride is the predominant anion secreted by epithelial cells. Elevations in chloride secretion occur when pro-secretory factors, such as those mentioned, bind to receptors located on the surface of epithelial cells and elevate intracellular second messengers such as cyclic nucleotides and calcium. Increases in the levels or activation states of second messengers results in stimulation of proteins that are the key regulators in the chloride secretory process, namely, apically situated chloride channels and basolateral potassium channels (1,2).
Disregulation of chloride secretion underlies several pathophysiological conditions. Excessive secretion of chloride is the driving force behind secretory diarrhea, which is prevalent in inflammatory bowel disease, as well as during infection with a variety of intestinal pathogens (3)(4)(5). On the other hand, deficient chloride secretion occurs in cystic fibrosis (6,7). Our laboratory has shown previously that, in addition to pro-secretory pathways, mechanisms exist within intestinal epithelial cells that serve to limit chloride secretion (2, 8 -10). In particular, we have established the role of the epidermal growth factor receptor (EGFr) 1 as a key negative regulator of epithelial chloride secretion. Pre-treatment of T 84 cells, a well defined line of human colonic epithelial cells with a crypt cell phenotype, with EGF attenuates secretory responses to the muscarinic M3 receptor agonist carbachol (CCh) (2,9). This effect of EGF is mediated by activation of the enzyme phosphatidylinositol 3-kinase, and studies measuring 86 Rb ϩ efflux have shown that this mechanism targets a basolateral potassium channel (8,11). The chloride secretory response to CCh is mediated by an increase in intracellular calcium and is transient in nature. The rapid decrease in the response is due, at least in part, to transactivation of the EGFr by the muscarinic M3 receptor and subsequent activation of extracellular signalregulated kinase (ERK) isoforms of mitogen-activated protein kinase, which apparently leads to an overall decrease in chloride secretion via effects on an apical chloride channel (2,12). However, CCh fails to recruit phosphatidylinositol 3-kinase in the inhibitory phase of secretion. The reason why activation of the EGFr by EGF, versus transactivation in response to carbachol, recruits different downstream signals and impinges on different eventual targets in the chloride secretory mechanism is unclear.
Transactivation of growth factor receptors by G protein-coupled receptors (GPCR) is a well established physiological occurrence (13)(14)(15)(16). However, mechanisms by which GPCRs transactivate growth factor receptors are heterogeneous and depend upon the particular complement of signaling molecules expressed within a given cell type (15). Previous work from our laboratory has shown that CCh transactivates the EGFr in T 84 cells by a mechanism involving elevation in intracellular calcium, activation of the soluble tyrosine kinase Pyk-2, recruitment of p60 src kinase, and subsequent phosphorylation and activation of the EGFr (12). However, studies in fibroblasts have shown that CCh causes the rapid release of an EGFr binding ligand, heparin-binding EGF (HB-EGF), and that transactivation of the EGFr by the CCh-activated muscarinic receptor is dependent upon HB-EGF binding to the EGFr (17). The aim of this study, therefore, was to investigate whether CCh transactivation of the EGFr in intestinal epithelial cells might also involve the release of an EGFr binding ligand. We hypothesized that this could account for divergent outcomes of EGF versus CCh-stimulated EGFr activation. We also sought to identify this ligand and elucidate its potential role and overall importance in the negative regulation of calcium-mediated intestinal epithelial chloride secretion.
Cell Culture-Methods for maintenance of T 84 cells in culture were described previously (18). Briefly, T 84 cells were grown in Dulbecco's modified Eagle's/F-12 medium (JRH, Lenexa, Kansas) supplemented with 5% newborn calf serum. For Ussing chamber/voltage clamp experiments, 5 ϫ 10 5 cells were seeded onto 12-mm Millicell-HA polycarbonate filters. For immunoprecipitation/Western blotting experiments, 10 6 cells were seeded onto 30-mm Millicell-HA polycarbonate filters. Cells seeded onto filters were cultured for 10 -15 days prior to use. When grown on filters, T 84 cells are known to acquire the polarized phenotype of native colonic epithelia. In accordance with the known distribution of their receptors on intestinal epithelia CCh, EGF, and the anti-EGFr antibody were added basolaterally in all experiments.
Ion Transport Studies-The opening of basolateral potassium channels was monitored by measuring efflux of the potassium tracer 86 Rb ϩ in response to different stimuli by a slight modification of a published method (1,19 2), and 10 D-glucose. The cells were loaded for 30 min with 86 Rb ϩ (1 Ci/ml, added apically and basolaterally) at 37°C. Simultaneously, anti-EGFr (10 g) was added basolaterally as indicated by the experimental design. Cells were then rinsed gently with HBSS three times to remove extracellular isotope. After the final rinse, fresh HBSS (300 l) was added to individual wells of a 24-well cell culture plate. The buffer was maintained at 37°C by placing the culture plate on a thermostatic heating block. The first three aliquots were used to establish a stable baseline in efflux buffer only. For the 86 Rb ϩ efflux assay, cells were placed in wells containing one of the following solutions: HBSS alone, CCh (100 M) alone, anti-EGFr (10 g) alone, or anti-EGFr and CCh in combination. The inserts were then transferred sequentially to different wells containing the respective treatment conditions at 2-min in-tervals. At the end of the experiment, all inserts were immersed in scintillation fluid, and all basolateral bathing solutions were placed in individual scintillation vials. All samples were then assessed for their content of 86 Rb ϩ using open channel readings from a liquid scintillation counter (Beckman LS3180). The fraction of intracellular 86 Rb ϩ remaining in the cell monolayer at each time point was calculated from the sample and final insert counts. Time-dependent rates of 86 Rb ϩ efflux were calculated as ln( 86 Rb ϩ t ϭ 1 / 86 Rb ϩ t ϭ 2 )/(t 1 Ϫ t 2 ), where 86 Rb ϩ is the percentage of intracellular 86 Rb ϩ at time t, and t 1 and t 2 are successive time points.
Immunoprecipitations and Western Blotting-T 84 cell monolayers were washed three times with Ringer's solution, allowed to equilibrate for 30 min at 37°C, and were then stimulated with agonists (Ϯ antagonists) as appropriate. The reaction was stopped by washing in ice-cold phosphate-buffered saline, and the cells were lysed in ice-cold lysis buffer (1% Triton X-100, 1 g/ml leupeptin, 1 g/ml pepstatin, 1 g/ml antipain, 100 g/ml phenylmethylsulfonyl fluoride, 1 mM Na ϩ -vanadate, 1 mM sodium fluoride, and 1 mM EDTA in phosphate-buffered saline) for 45 min. Cells were then scraped into microcentrifuge tubes and spun at 12,000 rpm for 10 min, and the pellet was discarded. Samples were assayed for protein content (Bio-Rad protein assay kit) and adjusted so that each sample contained an equal amount of protein.
For immunoprecipitation studies, lysates were incubated with immunoprecipitating antibody, as per the manufacturer's instructions, for 1 h at 4°C followed by another 1-h incubation at 4°C with protein A-Sepharose. Lysates were then centrifuged for 3 min at 15,000 rpm, and the supernatant was discarded. The pellets were washed in ice-cold phosphate-buffered saline three times and resuspended in 2ϫ gel loading buffer (50 mM Tris (pH 6.8), 2% SDS, 200 mM dithiothreitol, 40% glycerol, 0.2 bromphenol blue) and boiled for 2 min prior to separation by SDS-polyacrylamide gel electrophoresis. Resolved proteins were transferred onto polyvinylidene membranes (PerkinElmer Life Sciences). After transfer, the membrane was preblocked with a 1% solution of blocking buffer (Upstate Biotechnology, Inc.) for 30 min followed by a 1-h incubation with the appropriate concentration of primary antibody in 1% blocking buffer. After washing (5 ϫ 10 min) in TBST (Trisbuffered saline with 1% Tween), membranes were incubated for 30 min in horseradish peroxidase-conjugated secondary antibody (anti-mouse or anti-rabbit IgG; Transduction Laboratories, Lexington, KY) in 1% blocking buffer. After washing in TBST (5 ϫ 10 min), immunoreactive proteins were detected using an enhanced chemiluminescence detection kit (Roche Molecular Biochemicals). Densitometric analysis of Western blots was carried out using NIH Image software.
TGF-␣ ELISA-ELISA for TGF-␣ in apical and basolateral supernatants from T 84 cells grown on 30-mm Millicell transwell polycarbonate filters were performed using a commercially available kit, according to the manufacturer's instructions. In brief, the kit utilized rabbit polyclonal anti-TGF-␣ bound to microtiter plates as the capturing antibody and biotinylated polyclonal rabbit anti-TGF-␣ as a secondary antibody. This was then followed by the addition of streptavidin-horseradish peroxidase as the reporter enzyme. Bound horseradish peroxidase was visualized using O-phenylenediamine and measured colorimetrically at an absorbance of 490 nm using a microtiter plate reader. The detection limit of the assay is 6 pg/ml.
Amphiregulin Radioimmunoassay-A sensitive and specific radioimmunoassay was used to measure amphiregulin (AR) released into T 84 cell bathing media. The same samples were used to measure both TGF-␣ and amphiregulin levels. Recombinant human AR and a polyclonal goat IgG antibody to AR were purchased from R&D Systems (Minneapolis, MN). AR was iodinated by the chloramine-T method. The detection limit of the assay is 0.013 pmol/assay tube.
Statistical Analysis-All data are expressed as means Ϯ S.E. for a series of n experiments. Student's t tests or analysis of variance with the Student-Newman-Keuls post-test were used to compare mean values as appropriate. p values Ͻ0.05 were considered to represent significant differences.

A Neutralizing Antibody to the Ligand Binding Domain of the EGF Receptor Blocks G Protein-coupled Receptor-mediated
Transactivation of the EGFr in T 84 Cells-We first set out to investigate whether the muscarinic M3 receptor agonist CCh, which transactivates the EGFr in T 84 epithelial cells via a pathway involving increases in intracellular calcium and activation of Pyk-2 and Src kinases (12), can also transactivate the EGFr by causing the release of an EGFr ligand. Polarized T 84 cells, grown on permeable supports, were treated with CCh (100 M; 2 min) or EGF (100 ng/ml; 5 min) basolaterally. Western blot analysis of cell lysates immunoprecipitated with an anti-EGFr antibody and probed for phosphotyrosine showed that CCh and EGF induce EGFr phosphorylation, as reported previously (2,20). Pre-incubation of the basolateral domain of T 84 cells with a neutralizing antibody to the extracellular ligand binding domain of the EGFr (10 g/ml; 30 min; clone LA1) reduced CCh-stimulated EGFr phosphorylation by 83 Ϯ 7% (p Ͻ 0.01; n ϭ 13; see Fig. 1). EGF-induced receptor phosphorylation was likewise decreased by 82 Ϯ 6% (p Ͻ 0.001; n ϭ 13). These data suggest a prominent role for an EGFr ligand in mediating CCh-induced transactivation of the EGFr. The selectivity of the neutralizing antibody for the EGFr in preference to other members of the ErbB receptor family was established in experiments where T 84 cells were incubated basolaterally with the ErbB3 receptor ligand heregulin-␣ (100 ng/ml; 5 min), in the presence and absence of anti-EGFr pre-treatment. Western blot analysis of immunoprecipitated ErbB3 revealed that pre-incubation of cells with anti-EGFr had no effect on heregulin-␣ stimulation of ErbB3 tyrosine phosphorylation (data not shown).
A Neutralizing Antibody to the Ligand Binding Domain of the EGF Receptor Reduces CCh-stimulated ERK MAPK Phosphorylation in T 84 Cells-Previous studies from our laboratory have shown that CCh stimulates phosphorylation and activation of the ERK 1 and 2 MAPK isoforms by a pathway that depends, at least in part, on EGFr activation (2). To ascertain whether the EGFr-dependent fraction of CCh-stimulated ERK phosphorylation is mediated by release of an EGFr binding ligand, T 84 cells were pre-incubated, as previously, with a neutralizing antibody against the ligand binding domain of the EGFr. Anti-EGFr pre-treatment reduced CCh-stimulated ERK phosphorylation by 62 Ϯ 5% (p Ͻ 0.001; n ϭ 13; see Fig. 2). EGF-stimulated ERK phosphorylation was decreased by 89 Ϯ 3% (p Ͻ 0.001; n ϭ 13) following anti-EGFr pre-incubation. These findings suggest that CCh transactivation of the EGFr and downstream activation of ERK 1/2 involves the binding of an EGFr ligand to the extracellular ligand binding domain of the EGFr.
Blockade of the EGFr Ligand Binding Domain Potentiates CCh-induced 86 Rb ϩ Efflux-We next explored the significance of our data for EGFr regulation of Ca 2ϩ -dependent transepithelial chloride secretion. Because of the limited availability of anti-EGFr and the large volumes that would be required for a study of its effects on CCh-stimulated Cl Ϫ transport in Ussing chambers, we instead examined the effect of blockade of the EGFr ligand binding domain on K ϩ transport. K ϩ efflux is required to sustain chloride secretion across intestinal epithelial cells. Changes in the efflux of 86 Rb ϩ , a surrogate for K ϩ , parallel changes in chloride secretion (1). CCh (100 M) stimulates a rapid efflux of 86 Rb ϩ across polarized T 84 cells, reflecting opening of potassium channels (p Ͻ 0.05; n ϭ 4; see Fig. 3). In T 84 cells pre-incubated with anti-EGFr and loaded with 86 Rb ϩ , anti-EGFr significantly potentiated CCh-induced 86 Rb ϩ efflux (p Ͻ 0.05; n ϭ 4). These data indicate that inhibition of EGFr activation removes the "braking" effect exerted by EGFr transactivation on Ca 2ϩ -dependent ion transport, thus causing potentiation of CCh-stimulated 86 Rb ϩ efflux. This finding complements those in our previous studies, which showed that inhibition of the kinase activity of the EGFr potentiated CChstimulated Cl Ϫ secretion across T 84 cells (2).
Anti-EGFr Has No Effect on CCh-stimulated Association of Phosphorylated Pyk-2 with the EGFr-It has been shown previously that CCh-induced transactivation of the EGFr likely requires upstream phosphorylation of the calcium-dependent

FIG. 1. Blockade of the EGFr ligand binding domain attenuates CCh-induced phosphorylation of the EGFr.
Polarized T 84 cell monolayers were incubated basolaterally for 30 min with a neutralizing antibody against the EGFr (10 g/ml). Cells were then stimulated basolaterally with CCh (100 M; 2 min) or EGF (100 ng/ml; 5 min) in the presence or absence of the anti-EGFr antibody, and cell lysates were subsequently immunoprecipitated (IP) with anti-EGFr. A, representative Western blot showing that an anti-EGFr antibody decreased CCh-and EGF-induced EGFr tyrosine phosphorylation. p-EGFr, phospho-EGFr; Con, control. B, the same Western blot as in A, stripped and reprobed to show that equal levels of unphosphorylated EGFr protein were present in each sample. C, densitometric analysis shows anti-EGFr decreased CCh-(p Ͻ 0.01; n ϭ 13) and EGFinduced (p Ͻ 0.001; n ϭ 13) EGFr phosphorylation but had no effect on basal phosphorylation levels (n ϭ 3). Results are presented as mean Ϯ S.E. for increases in EGFr phosphorylation expressed in a.u. Asterisks represent significant differences from stimulus in the absence of anti-EGFr (**, p Ͻ 0.01; ***, p Ͻ 0.001).
soluble tyrosine kinase Pyk-2. Therefore, we investigated whether Pyk-2 phosphorylation lies upstream or downstream of the apparent release of an EGFr ligand by CCh. EGFr immunoprecipitates from T 84 cell lysates were probed with an antibody specific for Pyk-2 phosphorylated at the Tyr-881 residue. Western blot analysis showed that basolateral pre-treatment of T 84 cells with anti-EGFr (10 g/ml) did not attenuate CCh-induced association of phosphorylated Pyk-2 with the EGFr (n ϭ 4; see Fig. 4). This finding indicates that Pyk-2 activation occurs upstream of EGFr ligand binding to the EGFr and possibly upstream of CCh-stimulated EGFr ligand release.
CCh Causes ERK 1/2 Phosphorylation by EGFr Liganddependent and p60 src -dependent Pathways-p60 src kinase activation has been shown previously to be involved in CCh transactivation of the EGFr (12). These previous studies indicated that Src kinase is activated downstream of elevations in intracellular calcium, activation of calmodulin kinase II, and subsequent activation of the non-receptor tyrosine kinase Pyk-2. To determine whether a relationship exists between p60 src activation and the apparent release of an EGFr ligand by CCh, T 84 cells were pre-treated with the Src kinase inhibitor PP2, with anti-EGFr antibodies or with both inhibitors in combination. Co-incubation of PP2 (20 M; 30 min) added bilaterally, with anti-EGFr (10 g/ml) added basolaterally, had a similar inhibitory effect on CCh-induced EGFr phosphorylation in T 84 cells to that exerted by anti-EGFr alone (p Ͻ 0.01 versus CCh; n ϭ 5), suggesting that p60 src and the putative EGFr ligand are not acting via separate pathways to cause EGFr phosphorylation (Fig. 5). Although PP2 did not significantly reduce CCh-stimulated EGFr phosphorylation in the current set of experiments, there was a trend toward inhibition that might have achieved significance with additional replicates, consistent with previ-ous findings (12). Moreover, anti-EGFr and PP2 co-incubation did have an additive inhibitory effect (74 Ϯ 7%; n ϭ 4) on CCh-induced ERK phosphorylation, greater than either inhibitor alone (53 Ϯ 7%, p Ͻ 0.05 and 35 Ϯ 7%, p Ͻ 0.01 for anti-EGFr and PP2, respectively; see Fig. 6). Taken together, these data imply that CCh transactivates the EGFr by a pathway involving Pyk-2 and p60 src activation that appears to culminate in the binding of an EGFr ligand to the EGF receptor. EGFr activation leads to ERK activation, but CCh also appears to phosphorylate ERK by an EGFr-independent, p60 src kinasedependent pathway, because blockade of the EGFr ligand binding domain and inhibition of p60 src kinase had an additive inhibitory effect.
Carbachol-stimulated EGFr Transactivation and Downstream Activation of ERK Is Associated with the Basolateral Release of TGF-␣ from T 84 Cells-Because TGF-␣ is the most abundant member of the EGF family of ligands found in the intestine and is a known product of intestinal epithelial cells, we investigated whether CCh causes the release of TGF-␣ from T 84 cells (21,22). Cells were treated for 1, 2, 5, and 15 min with CCh (100 M). The apical and basolateral bathing solutions were collected and analyzed for TGF-␣ content using a commercially available TGF-␣ ELISA kit. Basolateral TGF-␣ release from T 84 cells was detected following 15 min of treatment with CCh (65 Ϯ 11 pg/ml; n ϭ 6). TGF-␣ was not detected at earlier time points or in supernatants from control (untreated) cells. In addition, no TGF-␣ was detected in the apical bathing solution of the same cells. In a separate series of experiments, detection of TGF-␣ in the basolateral bathing media of CChtreated T 84 cells was enhanced significantly by basolateral pre-incubation with anti-EGFr (Fig. 7A). Indeed, pre-incubation with anti-EGFr, which presumably blocks binding sites for show that equal levels of unphosphorylated ERK protein were present in each sample. C, densitometric analysis shows anti-EGFr decreased CCh-(p Ͻ 0.001; n ϭ 13) and EGF-induced (p Ͻ 0.001; n ϭ 13) ERK phosphorylation, but anti-EGFr alone had no effect on background ERK phosphorylation (n ϭ 3). Results are presented as mean Ϯ S.E. for increases in ERK phosphorylation expressed in a.u. Asterisks represent significant differences from stimulus in the absence of anti-EGFr (***, p Ͻ 0.001).
released TGF-␣, permitted detection of CCh-induced TGF-␣ release as early as 5 min after CCh stimulation. The EGFr ligand amphiregulin, which is also produced by colonic epithelial cells (23), was not detected in these samples when tested by radioimmunoassay (data not shown). In parallel studies, preincubation with anti-EGFr significantly reduced CCh-induced EGFr phosphorylation, indicating that blockade of the extracellular ligand binding domain of the EGFr with anti-EGFr could not only increase detection of CCh-induced TGF-␣ release but could also inhibit CCh-induced EGFr phosphorylation (Fig.  7B). These findings suggest that CCh transactivates the EGFr, at least in part, by causing the extracellular release of TGF-␣.
Capture of Released TGF-␣ Blocks CCh-induced EGFr Phosphorylation-Having shown that blockade of the EGFr ligand binding domain inhibits CCh-stimulated EGFr phosphorylation and that CCh stimulates TGF-␣ release, we next investigated whether CCh-stimulated EGFr phosphorylation was mediated by TGF-␣ release. T 84 cells were incubated basolaterally with an antibody against TGF-␣. Western blot analysis of EGFr immunoprecipitates, probed with an anti-phosphotyrosine antibody, showed that pre-treatment of T 84 cells with anti-TGF-␣ (5 g/ml; 30 min) inhibited CCh-induced EGFr phosphorylation (p Ͻ 0.001; n ϭ 4; see Fig. 8). These data indicate that CCh transactivates the EGFr in T 84 cells predominantly via the binding of TGF-␣ to the ligand binding domain of this receptor.
CCh-stimulated Release of TGF-␣ Is Matrix Metalloproteinasedependent but p60 src -and ERK-independent-Having determined that CCh-induced EGFr phosphorylation was at least partially mediated by the binding of TGF-␣ to the extracellular ligand binding domain of the EGFr, and that CCh causes release of TGF-␣ from T 84 cells, we further investigated the mechanism whereby CCh stimulated the release of TGF-␣. T 84 cells were pre-treated bilaterally with a broad spectrum matrix metalloproteinase inhibitor, WAY171318, which inhibits the activity of tumor necrosis factor ␣ converting enzyme/a disintegrin and metalloprotease (ADAM 17), which cleaves TGF-␣, amphiregulin, and HB-EGF (24,25). Western blot analysis of EGFr immunoprecipitates, probed with an anti-phosphotyrosine antibody, showed that pre-treatment of T 84 cells with WAY171318 (10 M; 30 min) completely blocked CCh-induced EGFr phosphorylation (p Ͻ 0.05; n ϭ 3; see Fig. 9). CChstimulated ERK phosphorylation was also decreased by 88 Ϯ 6% (p Ͻ 0.05; n ϭ 3; see Fig. 10). Importantly, WAY171318 had no effect on the intrinsic kinase activity of the EGFr itself as pre-incubation with WAY171318 did not diminish the capacity of EGF (100 ng/ml) to phosphorylate the EGFr (73 Ϯ 18 and 78 Ϯ 21 a.u. for EGF ϩ WAY171318 and EGF alone, respectively; n ϭ 5) or TGF-␣ (100 ng/ml) to phosphorylate the EGFr (80 Ϯ 10 and 57 Ϯ 5 a.u. for TGF-␣ ϩ WAY171318 and TGF-␣ alone, respectively; n ϭ 2). These data indicate that CCh transactivates the EGFr in T 84 cells via a matrix metalloproteinasedependent extracellular release of TGF-␣, which then binds to the ligand binding domain of the EGFr thereby initiating receptor activation.
To investigate the signaling events leading to release of TGF-␣ following CCh stimulation of T 84 cells, cells were preincubated with WAY171318, the Src kinase inhibitor, PP2, and the mitogen-activated protein kinase/extracellular signal-  4). In cells incubated with anti-EGFr and loaded with radioactive 86 Rb ϩ , anti-EGFr alone had no effect on baseline 86 Rb ϩ efflux (n ϭ 4) but significantly potentiated CCh-induced 86 Rb ϩ efflux. Asterisks represents significant differences from control, unstimulated cells (*, p Ͻ 0.05; **, p Ͻ 0.01). # denotes a significant effect of anti-EGFr on CCh-induced responses (#, p Ͻ 0.05).

FIG. 4. Anti-EGFr has no effect on CCh-induced Pyk-2 phosphorylation and association with EGFr.
Confluent T 84 cells grown on 30-mm inserts were stimulated with CCh (100 M) for 2 min, and cell lysates were immunoprecipitated (IP) with anti-EGFr and probed for phosphorylated Pyk-2. A, representative Western blot showing that pre-treatment of T 84 cells with anti-EGFr (10 g/ml) did not reduce the amount of phospho-Pyk-2 associated with the EGFr in response to CCh. Con, control. B, densitometric analysis of n ϭ 4 experiments showing that anti-EGFr had no effect on the association of phospho-Pyk-2 with the EGFr induced by CCh. Results are presented as mean Ϯ S.E. for EGFr-associated phospho-Pyk-2 expressed in a.u. Asterisks represent significant differences from control, unstimulated cells (**, p Ͻ 0.01). regulated kinase kinase 1 inhibitor, PD98059, which blocks ERK activation, prior to CCh stimulation for 15 min. ERK activation has been shown in other systems to participate in growth factor and GPCR-induced metalloproteinase-dependent EGFr ligand release (26,27). Bathing media from the basolateral domains of treated cells were collected and analyzed for TGF-␣ content. Phosphorylation of EGFr immunoprecipitated from cell lysates under the various different treatment conditions was also investigated. Fig. 11 is a composite graphic of TGF-␣ release and EGFr phosphorylation following CCh stimulation in the presence and absence of pre-treatment of T 84 cells with the various inhibitors. WAY171318 completely blocked CCh-induced TGF-␣ release and EGFr phosphorylation, indicating a likely direct association between TGF-␣ release and EGFr phosphorylation following CCh stimulation. PP2 and PD98059 had no significant inhibitory effects on EGFr phosphorylation or TGF-␣ release. WAY171318, PD98059, and PP2 all caused significant inhi-bition of CCh-stimulated ERK phosphorylation in these cell lysates (data not shown). DISCUSSION In the present study, we expand upon previous findings from our group to more fully elucidate the mechanism(s) by which the EGFr is involved in the regulation of calcium-dependent epithelial chloride secretion. Previous studies show that CChinduced transactivation of the EGFr in T 84 cells occurs via an intracellular signaling pathway involving the non-receptor tyrosine kinases Pyk-2 and Src (12). The current studies indicate an extracellular component to this transactivation mechanism involving ligand binding to the extracellular domain of the EGFr. We have found that blockade of the ligand binding domain of the EGFr using a region-specific neutralizing anti-EGFr antibody dramatically reduced CCh-stimulated EGFr and ERK activation. This extracellular ligand binding component to EGFr transactivation is also involved in regulation of ion transport responses in intestinal epithelial cells, because blockade of ligand binding potentiated CCh-stimulated potassium channel opening, which parallels, and is believed to reg- Results are presented as mean Ϯ S.E. for increases in EGFr phosphorylation expressed in a.u. Asterisks represent significant differences from control, untreated cells (**, p Ͻ 0.01; n ϭ 5). # represents a significant reduction in CCh-stimulated phosphorylation by inhibitor(s) (##, p Ͻ 0.01; ###, p Ͻ 0.001; n ϭ 5).

FIG. 6. Effect of co-incubation of anti-EGFr and the Src kinase inhibitor PP2 on ERK phosphorylation. T 84 cell lysates from cells treated with
CCh alone or CCh in the presence of PP2 (20 M), anti-EGFr (10 g/ml), or both inhibitors in combination were probed for phosphorylated ERK 1/2 (p-ERK 1,2). Con, control. A, representative Western blot showing that PP2 and anti-EGFr both reduced CChstimulated ERK phosphorylation. B, blots were subsequently stripped and re-probed with anti-ERK (Santa Cruz Biotechnology, Inc.) to show equal loading of ERK protein in each sample. C, densitometric analysis of n ϭ 4 experiments showing that anti-EGFr and PP2 co-incubation had an additive inhibitory effect (n ϭ 4) on CCh-induced ERK phosphorylation greater than either inhibitor alone (p Ͻ 0.05 and p Ͻ 0.01 for anti-EGFr and PP2, respectively). Results are presented as mean Ϯ S.E. for increases in ERK phosphorylation expressed in a.u. Asterisks represent significant differences from control, untreated cells (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; n ϭ 4). # represents a significant reduction in CCh-stimulated phosphorylation by inhibitor(s) (#, p Ͻ 0.05; ###, p Ͻ 0.001; n ϭ 4). ulate, chloride secretion (28). In vivo, the potentiation of chloride secretion would be expected to increase net fluid secretion into the intestinal lumen. Thus transactivation of the EGFr via its ligand binding domain indicates the importance of this site, and the ligand(s) that binds to it, in the negative regulation of intestinal ion transport and fluid secretion.
GPCR transactivation of growth factor receptors has been observed in many systems, and a number of different signaling mechanisms have been proposed to explain this phenomenon (12-15, 17, 29). Initial studies concluded that EGFr transactivation was mediated by an intracellular signaling pathway based on the kinetics of the response, the absence of detectable levels of EGFr ligands, and the suggested involvement of a multiprotein complex containing the EGFr, c-Src kinase, the Ca 2ϩ -dependent tyrosine kinase Pyk-2, and adapter proteins including Shc and Grb-2, which play a key role in recruitment of downstream signaling events (12, 13, 15, 30 -33). However, more recent studies in other systems have provided ample evidence that GPCR transactivation of growth factor receptors can be mediated by the extracellular release of growth factors (16,17,34,35). Studies by Prenzel et al. (17) led to the development of a "triple membrane passing signal" mechanism of EGFr transactivation. This model of receptor cross-talk involves three membrane events. First, GPCR activation leads to the activation of a matrix metalloproteinase, which then, proteolytically cleaves membrane-bound EGF-like growth factors causing their release into the extracellular milieu and subsequent binding of these growth factors to the ligand binding domain of the EGFr. EGFr activation then results in activation of the Ras/MAPK signaling pathway (36). In light of these variously postulated mechanisms of GPCR-induced transactivation of the EGFr, we attempted to provide a functional explanation for our data that indicated a role for extracellular binding of an EGF-like ligand to the EGFr in GPCR-induced transactivation, as well as previously published findings that demonstrated that the non-receptor tyrosine kinases Pyk-2 and p60 src are involved in EGFr transactivation by an apparently intracellular signaling pathway (12).
Pre-treatment with the neutralizing anti-EGFr antibody, which greatly reduced EGFr phosphorylation in T 84 cells, had no effect on the capacity of CCh to induce association of phosphorylated Pyk-2 with the EGFr. This finding, in combination with the level of inhibition of EGFr phosphorylation achieved with the anti-EGFr, suggests that association of phosphoryl-FIG. 7. Carbachol causes basolateral release of TGF-␣ from T 84 cells. Confluent T 84 cell monolayers were incubated basolaterally with anti-EGFr (10 g/ml) for 30 min and then treated basolaterally for 5 and 15 min with CCh (100 M). The basolateral bathing solutions were collected and analyzed for TGF-␣ content using an ELISA kit. TGF-␣ concentrations are expressed in pg/ml. A, basolateral TGF-␣ release from T 84 cells was detected following 15 min of treatment with CCh (p Ͻ 0.05 versus control (Con); n ϭ 3). CCh-induced basolateral TGF-␣ release was increased significantly at both 5 and 15 min by pre-incubation with anti-EGFr (n ϭ 3). Asterisks represent significant differences from control, unstimulated cells (*, p Ͻ 0.05; ***, p Ͻ 0.001). # represents significant effects of anti-EGFr on responses induced by CCh (#, p Ͻ 0.05; ##, p Ͻ 0.01). B, cell lysates were immunoprecipitated with anti-EGFr and probed for phosphotyrosine. Results are presented as mean Ϯ S.E. for increases in EGFr phosphorylation expressed in a.u. Anti-EGFr pre-incubation completely inhibited CCh-stimulated EGFr phosphorylation at both 5 (p Ͻ 0.01) and 15 (p Ͻ 0.001) min, respectively (n ϭ 3).

FIG. 8. Selective blockade of TGF-␣ binding to the EGFr inhibits CCh-stimulated EGFr phosphorylation.
Monolayers of T 84 cells were incubated basolaterally with an antibody against TGF-␣ (5 g/ml; 30 min). Cells were then stimulated basolaterally with CCh (100 M; 2 min) in the presence or absence of the anti-TGF-␣ antibody, and cell lysates were subsequently immunoprecipitated (IP) with anti-EGFr. A, representative Western blot showing that an anti-TGF-␣ antibody decreased CCh-induced EGFr tyrosine phosphorylation. p-EGFr, phospho-EGFr; Con, control. B, the same Western blot as in A, stripped and reprobed to show that equal levels of unphosphorylated EGFr protein were present in each sample. C, densitometric analysis shows that anti-TGF-␣ significantly decreased CCh-induced EGFr phosphorylation (p Ͻ 0.001; n ϭ 4). Results are presented as mean Ϯ S.E. for increases in EGFr phosphorylation expressed in a.u. Asterisks represent significant differences from control untreated cells (*, p Ͻ 0.05; ***, p Ͻ 0.001). # denotes a significant effect of anti-TGF-␣ on CCh-induced responses (###, p Ͻ 0.001).
ated Pyk-2 with the EGFr is not directly responsible for EGFr phosphorylation. Therefore, it appears that Pyk-2 activation occurs upstream of the event responsible for EGFr phosphorylation, namely, the binding of an EGFr ligand to the receptor. Because previous evidence suggests that Src activation occurs distal to Pyk-2 activation in the sequence of events between CCh-induced activation of the muscarinic M3 receptor and EGFr activation (12), we also investigated whether p60 src activation was involved in EGFr ligand binding. Previous data from our laboratory have shown that the Src kinase inhibitor, PP2, causes a partial inhibition of CCh-stimulated EGFr phosphorylation (12). In this study, PP2 inhibition of EGFr phosphorylation failed to reach significance, although with increased numbers of experiments significance would likely have been reached. The anti-EGFr antibody completely abrogated EGFr phosphorylation explaining why no additive inhibitory effect was seen when the two inhibitors were applied in combination. This result, combined with previous data, suggests that p60 src activity plays only a partial role in the chain of events leading to EGFr phosphorylation.
Interestingly, the anti-EGFr antibody did not completely block CCh-stimulated ERK phosphorylation, and when anti-EGFr and PP2 were used in combination, a significant additive inhibitory effect was observed. This observation, coupled with the EGFr phosphorylation data, suggests that multiple pathways are employed by CCh to activate the ERK MAPKs. The partial reductions in CCh-stimulated EGFr and ERK phosphorylation by PP2 indicate that a p60 src -mediated, EGFr-dependent pathway leading to ERK phosphorylation exists. However, the additive inhibitory effect on CCh-stimulated ERK phosphorylation observed when PP2 and anti-EGFr are used in combination suggests that a Src-mediated, EGFr kinase-independent route also exists and contributes to ERK phosphorylation. Although p60 src may act independently of EGFr kinase activation by this route, this does not preclude the involvement of the EGFr as a molecular scaffold for the formation of a p60 src signaling complex capable of activating the Ras signaling pathway. Evidence for the existence of Src-independent EGFr transactivation, along with Src-dependent ERK activation, has been obtained in other systems (13,37,38). It should be noted that even with the use of anti-EGFr and PP2, CCh-induced ERK phosphorylation was not inhibited completely even under conditions where EGFr phosphorylation was essentially abolished. This may indicate that a third pathway exists leading to ERK activation that acts independently of EGFr and p60 src activation.

FIG. 10. Inhibition of TGF-␣ release blocks CCh-induced ERK phosphorylation.
Lysates from CCh-treated cells with or without WAY171318 (10 M; 30 min) pre-treatment were probed for phosphorylated ERK 1/2 (p-ERK1,2). A, Western blot showing that bilateral pre-treatment of T 84 cells with the metalloproteinase inhibitor WAY171318 decreased CCh-stimulated ERK phosphorylation. Con, control. B, the same blot as in A was stripped and reprobed with anti-ERK to show equal loading of ERK protein in each sample. C, densitometric analysis showed that CCh-stimulated ERK phosphorylation was significantly reduced by WAY171318 pre-treatment (p Ͻ 0.05; n ϭ 3). Results are presented as mean Ϯ S.E. for increases in ERK phosphorylation expressed in a.u. The asterisk represents a significant effect of WAY171318 on CCh-induced responses (*, p Ͻ 0.05).
Having shown that an EGFr ligand is involved in GPCRinduced transactivation of the EGFr in T 84 cells, we next tried to identify the ligand responsible for this effect. TGF-␣ is expressed more broadly than EGF in the gastrointestinal tract (21,22,39,40) and is expressed highly in colonic mucosa whereas EGF is only expressed at low levels (21,41,42). Consequently, we initiated our search for the EGFr ligand responsible for CCh-induced transactivation by investigating whether TGF-␣ was involved. We found detectable levels of TGF-␣ only in the basolateral bathing media of cells treated for 15 min with CCh. This result showed that CCh could cause the extracellular release of TGF-␣ but at a time later than that when CCh induces EGFr phosphorylation (2,20). However, the absence of detectable levels of TGF-␣ at earlier time points was explained conceivably by rapid binding of TGF-␣ to the EGFr, supporting previous work that has established that TGF-␣ is a locally acting growth factor that is consumed rapidly by EGFr following ectodomain cleavage (43). This conjecture was supported by the fact that pre-incubation with the EGFr neutralizing antibody enhanced the detection of TGF-␣ released by CCh. In addition, this pre-incubation with anti-EGFr completely blocked CCh-stimulated EGFr phosphorylation. Therefore, these data support the hypothesis that TGF-␣ release occurs within the time frame of CCh-induced EGFr phosphorylation, and ligand binding to the EGFr appears largely to be responsible for EGFr transactivation. The conclusion that TGF-␣ is likely the active ligand involved in CCh transactivation of the EGFr is further strengthened by the capacity of an anti-TGF-␣ antibody to inhibit CCh-stimulated EGFr phosphorylation.
Having shown that CCh causes the extracellular release of TGF-␣ and that TGF-␣ is involved substantially in transactivation of the EGFr by CCh we sought to identify the mechanism responsible for cleavage of membrane-bound proTGF-␣. The enzymes involved in proteolytic cleavage of EGF-like molecules are transmembrane metalloproteases known as ADAMs (44 -46). The most highly characterized of the ADAMs is ADAM 17, also known as tumor necrosis factor ␣ converting enzyme, which is responsible for cleavage of membrane-bound proTGF-␣ in vivo causing release of mature TGF-␣ (24,25). We found that an inhibitor (WAY171318) that selectively blocks the proteolytic activity of the metalloproteinase responsible for proTGF-␣ cleavage completely blocked CCh-stimulated transactivation of the EGFr and dramatically reduced ERK activation. These data provide solid evidence that CCh-induced transactivation of the EGFr in T 84 intestinal epithelial cells occurs by matrix metalloproteinase-mediated proteolytic cleavage of membrane-bound proTGF-␣.
FIG. 12. CCh-stimulated transactivation of the EGFr: a negative feedback mechanism for the regulation of Ca 2؉ -dependent chloride secretion across colonic epithelial cells. CCh stimulates epithelial chloride secretion via a mechanism requiring elevations in intracellular calcium. However, the magnitude and duration of secretory responses to CCh are limited intrinsically by a signaling pathway incorporating CCh-stimulated transactivation of the EGFr and activation of the ERK mitogen-activated protein kinase isoforms. CChstimulated transactivation of the EGFr appears to involve calcium, calmodulin,Pyk-2,andp60 src activationandthematrixmetalloproteinasedependent proteolytic release of membrane-bound TGF-␣. The exact chain of events following CCh-induced elevations in intracellular calcium has not been elucidated fully, but data from the present study, combined with previous studies from our group, suggest that calcium activates calmodulin, which then activates the Ca 2ϩ -dependent tyrosine kinase Pyk-2. Pyk-2 then associates with phosphorylated p60 src . This complex has been shown to associate with the EGFr. Pyk-2 appears to lie upstream of the metalloproteinase-dependent release of TGF-␣ and thus may participate actively in TGF-␣ release. The binding of TGF-␣ to the EGFr appears to be the major mechanism by which the EGFr becomes activated following CCh stimulation. The exact role and importance of p60 src in CCh-stimulated EGFr transactivation is still unclear, however. p60 src activation does not appear to be a pre-requisite for CCh stimulation of TGF-␣ release. However, activation of p60 src appears to be of greater importance in ERK activation as, in addition to EGFr-dependent ERK activation, CCh appears to be able to activate ERK by a p60 src -dependent but EGFr-independent pathway. ERK activation causes an overall decrease in epithelial chloride secretion by as yet undetermined mechanisms. Please note that the signaling pathway downstream of ERK leading to an apical chloride channel is not necessarily intended to imply a direct effect of these mechanisms on a chloride channel protein but rather (for simplicity) an overall negative effect on chloride secretion. This could also involve effects on basolateral membrane transport pathways, such as potassium channels.
FIG. 11. CCh-stimulated TGF-␣ release is mediated by matrix metalloproteinase but not by Src or ERK activation. Polarized T 84 cell monolayers were pre-incubated bilaterally with one of the following inhibitors: WAY171318 (10 M; 30 min), PP2, or the mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1 inhibitor PD98059 (50 M; 30 min) prior to stimulation with CCh for 15 min. Basolateral bathing media were collected and analyzed for TGF-␣ content. Cell lysates were immunoprecipitated with anti-EGFr and probed for phosphotyrosine. Open bars represent EGFr phosphorylation (expressed in a.u. Hatched bars represent TGF-␣ release (expressed in pg/ml). Asterisks represent significant differences from cells treated with CCh alone. *, p Ͻ 0.05; **, p Ͻ 0.01. p-EGFr, phospho-EGFr; Con, control.
Previous studies from our group indicate a role for Src in CCh-stimulated EGFr transactivation (12). Src kinase activation has also been shown to be required for metalloproteinasedependent HB-EGF shedding during ␣2A-adrenergic receptor transactivation of the EGFr (16). Consequently, we investigated whether p60 src activation was required for TGF-␣ release. In addition, we also investigated whether ERK activation was involved in TGF-␣ release, because studies by Fan and Derynck (26) demonstrated that ERK activation contributes to growth factor-induced release of TGF-␣. In experiments where cells were stimulated for 15 min with CCh following pre-incubation with WAY171318, PP2, or PD98059, WAY171318 completely inhibited both EGFr phosphorylation and TGF-␣ release indicating the likely necessity for TGF-␣ release in CCh-stimulated EGFr transactivation. On the other hand, PP2 had no effect on TGF-␣ release, indicating that p60 src activation is not involved in CCh-induced release of TGF-␣. However, EGFr phosphorylation levels were partially reduced by PP2 suggesting that p60 src may act independently of CCh-stimulated ligand release to phosphorylate the EGFr. The exact role of p60 src in CCh-stimulated transactivation of the EGFr is unclear, and whether it has a direct stimulatory effect on the EGFr or acts in a facilitatory capacity has yet to be conclusively determined. In light of the potentiating effect of PP2 on ion transport responses to CCh in T 84 cells (12), and the apparent evidence for a p60 src -dependent, EGFr-independent route of ERK activation shown in the present study, it appears likely that activation of p60 src is of greater importance in mediating CCh activation of ERK than in transactivation of the EGFr. Finally, PD98059 did not significantly reduce EGFr phosphorylation or TGF-␣ release, suggesting that ERK activation is not involved in TGF-␣ release in the system studied here.
In conclusion, our data demonstrate that CCh-stimulated transactivation of the EGFr is mediated by activation of a matrix metalloproteinase that proteolytically cleaves membrane-bound proTGF-␣. This causes the release of TGF-␣ from the basolateral domain of the cell and binding of TGF-␣ to the EGFr resulting in receptor activation and recruitment of the Ras/MAPK signaling cascade. We cannot exclude that other metalloproteinases may be involved in a cascade-like protease action and that other EGFr ligands may contribute to this process. A schematic model of this ligand-dependent transactivation pathway and its role in the negative regulation of epithelial chloride secretion is shown in Fig. 12. The pathophysiological relevance of appropriate regulation of chloride secretory mechanisms in the colon leads us to believe that not only is there a fundamental role for the EGFr in acting as a central regulator for a variety of signaling pathways involved in fluid secretion but also that autocrine activation of the EGFr may play a critical role in limiting ion transport responses to secretory stimuli. A greater understanding of the role of growth factor receptors and their ligands in mediating anti-secretory signaling should facilitate identification of new targets for therapeutic intervention in intestinal disorders that are associated with epithelial ion transport deregulation.