Involvement of phosphatidylinositol 3-kinase and mitogen-activated protein kinases in glycine-extended gastrin-induced dissociation and migration of gastric epithelial cells.

The various molecular forms of gastrin can act as promoters of proliferation and differentiation in different regions of the gastrointestinal tract. We report a novel stimulatory effect of glycine-extended gastrin(17) only on cell/cell dissociation and cell migration in a non-tumorigenic mouse gastric epithelial cell line (IMGE-5). In contrast, both amidated and glycine-extended gastrin(17) stimulated proliferation of IMGE-5 cells via distinct receptors. Glycine-extended gastrin(17)-induced dissociation preceded migration and was blocked by selective inhibitors of phosphatidylinositol 3-kinase (PI3-kinase) but did not require mitogen-activated protein (MAP) kinase activation. Furthermore, glycine-extended gastrin(17) induced a PI3-kinase-mediated tyrosine phosphorylation of the adherens junction protein beta-catenin, partial dissociation of the complex between beta-catenin and the transmembrane protein E-cadherin, and delocalization of beta-catenin into the cytoplasm. Long lasting activation of MAP kinases by glycine-extended gastrin(17) was specifically required for the migratory response, in contrast to the involvement of a rapid and transient MAP kinase activation in the proliferative response to both amidated and glycine-extended gastrin(17). Therefore, the time course of MAP kinase activation appears to be a critical determinant of the biological effects mediated by this pathway. Together with the involvement of PI3-kinase in the dissociation of adherens junctions, long term activation of MAP kinases seems responsible for the selectivity of this novel effect of G(17)-Gly on the adhesion and migration of gastric epithelial cells.

ious events such as gastric acid secretion, endocrine secretion of histamine and somatostatin, expression of epidermal growth factors (2), activation of early genes (3), and proliferation (4). Compelling evidence also demonstrates that alternative forms of progastrin processing, such as glycine-extended gastrin 17 (G 17 -Gly) (1,5,6) and progastrin-(6 -80) (7), have a biological role in the stimulation of proliferation. Recently, G 17 -Gly has been shown to promote the invasiveness of the colon cancer cell line LoVo (8), and various reports showed that gastrin-related peptides are capable of activating a set of focal adhesion proteins such as p125 FAK , p130 Cas , and paxillin, which participate in the regulation of various cell functions such as preservation of morphology and migration (see Ref. 9 for review). However, direct effects of gastrins on epithelial cell adhesion have not yet been described.
Binding of G 17 -NH 2 to the gastrin/cholecystokinin B (gastrin/ CCK-B) receptor has been shown to activate various intracellular transduction pathways depending on the cell type. In the rat pancreatic cell line AR4-2J, gastrin-induced cell proliferation is thought to be mediated by activation of MAP kinases, leading to subsequent expression of immediate early genes like c-fos and c-jun (3,10), whereas gastrin-promoted cell growth in the rat pituitary adenoma cell line GH3 is supported by a Ca 2ϩ -dependent mechanism (11). Furthermore, in a rat intestinal epithelial cell line (IEC-6) (12) and in Chinese hamster ovary cells expressing transfected gastrin/CCK-B receptors (13), gastrin stimulates c-Src-like tyrosine kinases upstream of phosphatidylinositol 3-kinase (PI3-kinase) and MAP kinase.
To date, G 17 -Gly-induced transduction pathways have been studied mostly in tumor cell lines. In AR4-2J cells (3), as well as in the human colon cancer cell lines HT29 and LoVo (14), G 17 -Gly stimulates c-Jun amino-terminal kinase activation independently of the MAP kinase pathway. Binding studies strongly suggest that G 17 -Gly effects are mediated by a novel receptor that is insensitive to G 17 -NH 2 and classical gastrin/ CCK-B receptor antagonists (5,6,14), although a different receptor binding with a similar affinity to G 17 -Gly and G 17 -NH 2 was also identified on Swiss 3T3 fibroblasts (15). More studies are necessary in order to determine the signal transduction pathways activated by G 17 -Gly on other tumoral and nontumoral cell lines and to correlate these processes with the biological roles of the peptide.
For this work, we used a recently established gastric epithelial cell line (IMGE-5) (16) to compare the effects of amidated and glycine-extended gastrins on proliferation as well as cell adhesion and migration. The phenotype of these non-tumorigenic cells can be modulated in vitro by shifting them from a permissive temperature (33°C) in the presence of ␥-interferon to a non-permissive temperature (39°C), allowing their differentiation toward an epithelial phenotype (16). The gastric origin of IMGE-5 cells together with their differentiated phenotype make them particularly suitable to study the cellular effects of gastrin-related peptides.
In this paper, we report for the first time that, while both molecular forms of gastrin stimulate proliferation of IMGE-5 cells, dissociation and migration of these cells in a wound healing assay was induced only by G 17 -Gly. G 17 -Gly also induced tyrosine phosphorylation of the adherens junction protein ␤-catenin and dissociation of the complex between ␤-catenin and the transmembrane protein E-cadherin, followed by the partial disappearance of ␤-catenin from the cell membrane. We also show that a differential time course in the activation of MAP kinases by the two gastrin derivatives, as well as the involvement of PI3-kinase in the effect of G 17 -Gly on ␤-catenin, seem responsible for the selectivity of the effects of G 17 -Gly on the adhesion and migration of IMGE-5 cells.
The IMGE-5 cell line was established from the gastric mucosa of H-2 Kb-tsA58 transgenic mice as described previously (16). These mice are transgenic for a ␥-interferonand temperature-sensitive mutant of the SV40 large T antigen. The origins and detailed characteristics of this strain have been described previously (19,20). IMGE-5 cells were generally grown in DMEM ϩ 1 unit/ml ␥-interferon ϩ 5% FCS at 33°C (permissive conditions). For all experiments, they were transferred to 39°C in the same medium without ␥-interferon (non-permissive conditions), where they display differentiated characteristics such as expression of functional adherens and tight junctions. All experiments have been performed on cells between passages 15 and 35.
Bromodeoxyuridine (BrdUrd) Incorporation and Immunocytochemical Detection of ␤-Catenin-Immunocytochemistry and BrdUrd incorporation experiments were performed on cells grown under non-permissive conditions on 14-mm glass coverslips. Cells were treated with the agents to be tested for 18 h in DMEM containing 0.1% heat-inactivated FCS and 100 M BrdUrd (for BrdUrd staining) or in a similar medium without BrdUrd for the indicated times (for detection of membrane proteins). They were then fixed in ice-cold methanol for 3 min at 4°C (BrdUrd staining) or in 2% paraformaldehyde in PBS for 10 min at room temperature (for other immunocytochemistry). After three PBS washes, cells were incubated for 5 min in PBS ϩ 1.5 M HCl (for BrdUrd staining) or in PBS ϩ 0.2% Triton X-100 followed by PBS ϩ 0.2% gelatin for 10 min (for immunocytochemistry). Primary antibodies were then incubated for 2 h, and coverslips were washed three times in PBS, and the appropriate secondary antibody was incubated for 1 h. After two PBS washes and one rinse in water, coverslips were mounted on slides in Cytifluor (Oxford Instruments). Images were acquired using a Leica DC200 digital camera and DC viewer software (1280 ϫ 1024 pixels/image).
Detection of G-CCK-B Receptors by Reverse Transcriptase-PCR-Total RNA was prepared from confluent and non-confluent cells cultured in DMEM ϩ 5% FCS in permissive or non-permissive conditions according to Chomczynski and Sacchi (21). Reverse transcription experiments were performed with 10 g of total RNA using the Superscript II Reverse Transcriptase (Life Technologies, Inc.), according to the manufacturer's instructions. PCR was performed in a final volume of 25 l using 0.27 mM of each dNTP, 1 M of each G/CCK-B specific primer (sense, GCCCTTCACACTCCTGCCCAACC; antisense, GCGGAGCCC-TAGGTAGAGTTCGCGGG), 1.5 mM MgCl 2 , 1ϫ PCR buffer, 0.3 unit of Taq polymerase (Promega, Madison, WI), and 2 l of template cDNA. The standard PCR procedure involved denaturation of samples at 94°C, annealing at 65°C, and elongation of DNA strands at 72°C. 10 l of the samples were run on a 1.8% agarose gel containing ethidium bromide and photographed under UV light.
Western Blotting-Cells were grown in 100-mm Petri dishes under permissive conditions until they reached 90% confluency. They were then transferred to non-permissive conditions and serum-starved for 24 h, stimulated with the indicated concentrations of G 17 -NH 2 or G 17 -Gly for various times with or without 15 min of preincubation with either 10 M LY 294002 or 50 M PD 98059, and lysed using the standard procedure described previously (23). In the case of ␤-catenin/ E-cadherin association studies, 100 g of protein lysate per sample was immunoprecipitated in Tris/NaCl (pH 7.5) containing 1% Nonidet P-40, 100 M sodium orthovanadate, and 1 mM DTT (WLB buffer), using 1 g of anti-␤-catenin antibody for 2 h at 4°C, followed by 100 l of 20% protein A-Sepharose CL-4B (Amersham Pharmacia Biotech) overnight. Samples were washed three times in WLB buffer, and spun for 10 s at 10,000 ϫ g. The pellet was resuspended in loading buffer, denatured for 3 min at 95°C, and spun for 30 s at 10,000 ϫ g, and proteins in the supernatant were separated on an 8.5% SDS-polyacrylamide gel. For detection of MAP kinase and Akt activation as well as measurement of phosphorylation, cells were lysed with RIPA buffer (1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 0.2 mM sodium orthovanadate, 0.5 mM DTT, and protease inhibitors in 20 mM Tris, 150 mM NaCl (pH 7.5)). 20 g of total protein lysates were then mixed with loading buffer, denatured, and separated on a gel as described. Proteins were transferred onto a nitrocellulose membrane using a semi-dry blotting system (Bioblock, Nancy, France). Membranes were then incubated with the appropriate primary antibodies, and detection was performed with alkaline phosphatase-coupled anti-rabbit or anti-mouse IgG followed by incubation with a 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium solution, pH 9.2 (Sigma). Membranes were scanned using a Hewlett-Packard ScanJet 5200C, and densitometric analysis of protein bands was performed with a Fuji BAS software.
Migration Experiments-Wound healing experiments were performed in order to assess the effects of gastrins on cell migration. Cells were grown in 12-well plates under permissive conditions until they reached 80% confluence, then shifted to 39°C, and serum-starved for 24 h. The confluent monolayer was then wounded linearly using a pipette tip, washed three times with PBS, and treated with or without the agents to be tested for the indicated length of time in the presence of 0.1% FCS. Morphology and migration of cells was then observed and photographed at regular intervals for 24 h. When combined with immunofluorescence experiments, wound healing was performed similarly, except that cells were seeded onto 12-mm glass coverslips. At each time point, cells were photographed and then fixed in paraformaldehyde (2% in PBS) for 10 min at room temperature. Immunocytochemical localization of ␤-catenin was then performed as described earlier.

RESULTS
Dual Biological Effect of G 17 -Gly but Not G 17 -NH 2 on IMGE-5 Cells-The proliferative effects of G 17 -NH 2 and G 17 -Gly on IMGE-5 cells under non-permissive conditions were investigated. Both forms of the peptide were found to stimulate BrdUrd incorporation into IMGE-5 cell nuclei in a dose-dependent manner and with a similar amplitude (Fig. 1A). G 17 -Gly was found to be significantly more potent than G 17 -NH 2 as a stimulant of IMGE-5 proliferation (EC 50 ϭ 22.8 Ϯ 1.5 pM for G 17 -Gly and 84.3 Ϯ 2.9 pM for G 17 -NH 2 ).
We then investigated whether either form of gastrin was able to trigger changes in morphology and motility of IMGE-5 cells under non-permissive conditions. Morphology and migration were assessed using a wound healing model on a near-conflu-ent cell monolayer. In the presence of 0.1% serum, untreated cells slowly proliferated into the open wound but did not significantly change morphology or display any significant motility for up to 48 h (Fig. 1Ba, at t ϭ 0, 16 and 24 h). However, morphological changes were detected from 8 h after addition of 100 pM to 100 nM G 17 -Gly, with an elongation and spreading of cells. Cells started to migrate into the open wound about 12-15 h after treatment, and migration was maximal with 5-10 nM G 17 -Gly (Fig. 1, Bb and Bg). The wound was on average completely repaired by 24 -26 h in the presence of 5-50 nM G 17 -Gly ( Fig. 1, Be-Bg). On the contrary, no effect of G 17 -NH 2 was detected on morphology or migration for up to 48 h, at concentrations ranging from 0.1 nM to 10 M (data not shown).
We also assessed whether G 17 -Gly or G 17 -NH 2 (100 pM to 100 nM) had any effect on IMGE-5 cell differentiation under the same experimental conditions. However, the cells did not display any staining for markers of various specialized gastric cell types, such as H ϩ /K ϩ ATPase (parietal cells), mucin M1 peptide (mucus-secreting cells), histamine (ECL cells), or chromogranin A (endocrine cells) for up to 48 h after G 17 -Gly and G 17 -NH 2 treatment (data not shown).
G 17 -Gly Destabilizes Cell/Cell Contacts Prior to Cell Migration-In order to determine whether G 17 -Gly was also capable of inducing or stimulating the dissociation of epithelial cells at the front edge of the wound prior to migration, we performed further wound healing assays in the presence of 10 -100 nM G 17 -Gly using cells grown on glass coverslips, and we combined these assays with immunodetection of ␤-catenin. ␤-Catenin staining on IMGE-5 cells was typically strongly membrane associated, with a small proportion also present in the nucleus.
In untreated samples, the cells at the edge of the wound did not seem to dissociate from one another at any stage, and the ␤-catenin staining remained similar to that detected before or upon wounding for up to 24 h (Fig. 2, A1-A8). On the contrary, cells treated with G 17 -Gly significantly dissociated from each other at the front edge of the wound as early as 8 h after treatment (shown here for 10 nM G 17 -Gly, Fig. 2B6), and dissociation correlated with a strong shift of ␤-catenin localization from the membrane, mainly to the nucleus of these cells (Fig.  2B2). Interestingly, once the treated cells started to migrate into the wound, only about 30% consistently maintained a strong nuclear staining for ␤-catenin, whereas staining in the others was evenly decreased (Fig. 2B3). In all cases, upon closure of the wound, cells lost their spindle-like morphology and re-established contact with their neighbors, and ␤-catenin returned to a mostly membrane-bound localization (Fig. 2B4), around 24 -26 h after treatment with G 17 -Gly concentrations above 5 nM.
Different Receptors Are Responsible for Effects of G 17 -NH 2 and G 17 -Gly on IMGE-5 Cells-The only receptor selective for G 17 -NH 2 cloned to date is the gastrin/CCK-B receptor, expression of which in IMGE-5 cells was assessed by reverse transcriptase-PCR. IMGE-5 cells expressed gastrin/CCK-B receptor mRNA only under non-permissive conditions (Fig. 3A).
The presence of a selective receptor for G 17 -Gly was investigated by binding of 125 I-G 17 -Gly to adherent IMGE-5 cells. Although specific binding was detected under permissive conditions (data not shown), it was greatly increased under nonpermissive conditions. Binding was reduced in a dose-dependent manner by concomitant incubation of cells with increasing concentrations of unlabeled G 17 -Gly but not with G 17 -NH 2 or with the G/CCK-B receptor antagonist L365,260 (Fig. 3B).
G/CCK-B and CCK-A receptors are currently the best characterized receptors for gastrin-related peptides. By using reliable selective antagonists, the involvement of these subtypes in the proliferative effects of G 17 -NH 2 and G 17 -Gly on IMGE-5 cells was investigated. Interestingly, the stimulation induced by G 17 -NH 2 was dose-dependently reversed by the selective G/CCK-B antagonist L365,260 (18) (Fig. 3C), whereas L364,718, a selective CCK-A receptor antagonist (17), had a very weak effect, consistent with its interaction at high concentrations with G/CCK-B receptors (Fig. 3C). Neither antagonist had any effect on G 17 -Gly-induced proliferation, at any of the concentrations tested (Fig. 3C).
Phosphatidylinositol 3-Kinase and MAP Kinase Pathway Are Involved in the Biological Effects Induced by G 17 -Gly-The role of PI3-kinase in the mediation of the proliferative signal triggered by G 17 -Gly and G 17 -NH 2 was investigated. When IMGE-5 cells were preincubated with either of two PI3-kinase inhibitors (10 nM wortmannin or 10 M LY 294002), the proliferative effect of G 17 -Gly was almost abolished, whereas neither inhibitor significantly affected the stimulation induced by G 17 -NH 2 (Fig. 4A). In contrast, the stimulatory effects of both peptides were significantly decreased by the MEK1/2 inhibitor PD 98059 (50 M).
Furthermore, preincubation with 10 M LY 294002 (Fig.  4Be) prevented the stimulation of IMGE-5 cell migration induced by 10 -100 nM G 17 -Gly (Fig. 4Bd for effect of 100 nM), indicating that PI3-kinase is directly involved in this effect. Interestingly, although pretreatment with the MEK inhibitor PD 98059 abolished the motility response to G 17 -Gly, numerous cells were still able to dissociate from their neighbors at the edges of the wound (Fig. 4Bf).

Role of Phosphatidylinositol 3-Kinase and MAP Kinases in the Tyrosine Phosphorylation and Membrane Delocalization of
␤-Catenin-When confluent IMGE-5 cells were incubated for periods ranging from 30 min to 24 h with 100 pM to 100 nM G 17 -Gly under non-permissive conditions, a marked decrease was detected in the amount of ␤-catenin located at the plasma membrane from 1 h after treatment (shown for 100 nM), whereas cytoplasmic staining for the protein was greatly  increased (Fig. 5Ab). The enhanced cytoplasmic localization of ␤-catenin persisted for at least 4 h, but the protein returned to the plasma membrane within 12 h after G 17 -Gly treatment (Fig. 5Ab). This shift of ␤-catenin from the membrane to the cytoplasm was largely prevented by preincubating the cells with the PI3-kinase inhibitor LY 294002 (Fig.  5Ad) or with the tyrosine kinase inhibitor tyrphostin 25 (Fig.  5Ae), whereas the MEK1/2 inhibitor PD 98059 was found to have no effect on the G 17 -Gly-induced delocalization of ␤-catenin (Fig. 5Ac). Furthermore, G 17 -NH 2 did not display any effect on the membrane localization of ␤-catenin during 24 h of treatment (Fig. 5Af).
The effect of similar doses of G 17 -Gly on ␤-catenin in IMGE-5 cells was then studied by Western blotting. G 17 -Gly induced a rapid increase in the tyrosine phosphorylation of ␤-catenin (Fig.  5B) and a partial dissociation of ␤-catenin from its adherens junction partner E-cadherin (Fig. 5B). Increased tyrosine phosphorylation of ␤-catenin, as well as its dissociation from Ecadherin, was significant after 15 min and maximal 30 min after G 17 -Gly treatment. The association between ␤-catenin and Ecadherin returned to control values 4 h after treatment, while at that time the tyrosine phosphorylation levels of ␤-catenin were still slightly higher than those found in control cells (Fig. 5Be), perhaps because ␤-catenin can partly reassociate with E-cadherin before being completely dephosphorylated. Alternatively, a slight difference in sensitivity may exist between the detection of phosphorylated ␤-catenin and of the amount of Ecadherin co-immunoprecipitated with ␤-catenin. Tyrosine phosphorylation of ␤-catenin returned to control levels within 12 h after addition of G 17 -Gly (Fig. 5Bf). Thus, the time course of G 17 -Gly-induced changes in ␤-catenin phosphorylation and association with E-cadherin correlates well with the partial cytoplasmic relocalization of ␤-catenin detected by immunocytochemistry.
Both the G 17 -Gly-induced tyrosine phosphorylation of ␤-catenin (Fig. 5D) and its dissociation from E-cadherin (Fig. 5C) were abolished by preincubation with the PI3-kinase inhibitor LY 294002. In contrast, the apparent inability of the MEK1/2 inhibitor PD 98059 to block G 17 -Gly-induced delocalization of ␤-catenin and dissociation of IMGE-5 cells in the wound healing assay was further supported by a similar lack of effect of this inhibitor on the G 17 -Gly-induced stimulation of tyrosine phosphorylation of ␤-catenin (Fig. 5Dd) and of its dissociation from E-cadherin (Fig. 5Cd).
Finally, G 17 -NH 2 showed no effect on the phosphorylation level of ␤-catenin (Fig. 5De) or on its association with E-cadherin (Fig. 5Ce).

Differential Activation of the MAP Kinase Pathway by G 17 -NH 2 and G 17 -Gly Is Partly Responsible for the Difference in
Biological Effects-The time course of p42/p44 MAP kinase activation by G 17 -NH 2 and G 17 -Gly in IMGE-5 cells was then investigated. A very rapid but transient activation of p42/p44 phosphorylation was induced by G 17 -NH 2 (Fig. 6A, left panel); activation was found to be maximal within 1 min of stimulation, and the phosphorylation level returned to control levels between 15 and 30 min later (Fig. 6A, graph). On the contrary, the profile of activation by G 17 -Gly was quite different (Fig. 6A, right panel); p42/p44 phosphorylation was also detected from 1 min after stimulation, but the intensity of phosphorylation was found to increase continuously until 30 min after stimulation, to remain high for up to 3 h, and to return to control levels 6 h after stimulation (Fig. 6A, graph).
In order to determine whether the differential MAP kinase activation could explain, at least in part, the different biological effects displayed by both molecular forms of gastrin, activation of the MAP kinase pathway was blocked at different stages of the simulation induced by either form of gastrin, and the consequences on their respective biological activities were investigated. When the early (15 min) activation of the MAP kinase pathway was blocked by preincubating the cells with 50 M PD 98059 prior to G 17 -NH 2 or G 17 -Gly stimulation, the increase in proliferation was abolished (Fig. 6B). On the contrary, when the MAP kinase pathway inhibitor was added simultaneously with each one of the peptides, the early activation of p42/p44 MAP kinases was still detected, and the proliferative effect of G 17 -NH 2 or G 17 -Gly was no longer abolished (Fig. 6B). In contrast, the stimulation of migration by G 17 -Gly was always blocked whether the MAP kinase pathway was blocked 5, 15, or 30 min after addition of the peptide, whereas G 17 -Gly-induced cell dissociation was unaffected in these conditions (Fig. 6C).
G 17 -Gly Activates Akt/PKB Phosphorylation-Finally, we assessed whether G 17 -Gly and G 17 -NH 2 were capable of regulating a PI3-kinase-dependent pathway related to apoptosis, through control of Akt/PKB phosphorylation. Although G 17 -NH 2 had no effect (Fig. 7, upper panel), G 17 -Gly significantly increased Akt phosphorylation. Activation was detected about 15 min after addition of the peptide, was maximal after 30 min (Fig. 7, lower panel), and decreased regularly down to control values after 60 min. This activation was reversed by preincubation with the PI3-kinase inhibitor LY 294002.

DISCUSSION
The results presented in this paper demonstrate the existence of similarities but also major differences in the biological effects of G 17 -Gly and G 17 -NH 2 on a non-tumorigenic gastric epithelial cell line (IMGE-5). On the one hand, both G 17 -NH 2 and G 17 -Gly stimulated IMGE-5 proliferation, as has been described previously in several other models (1, 4 -6). On the other hand, a major difference between the two gastrins is the novel effect of G 17 -Gly only on the dissociation and migration of gastric epithelial cells. Our results underline the existence of a biological role for G 17 -Gly that seems specific to this peptide and is not exhibited by G 17 -NH 2 . This finding contrasts with previous work on G 17 -Gly, which described biological effects also displayed by G 17 -NH 2 , such as cell proliferation (1, 5, 6) or stimulation of gastric acid secretion (24). However, it is in agreement with the recent description of a stimulatory effect of G 17 -Gly on the invasiveness of the human colon cancer cell line LoVo (8). The enhancement of cell migration by G 17 -Gly re-ported herein may be profoundly important in the previously reported roles of progastrin-derived peptides during the differentiation of the gastrointestinal tract, as well as during cancer development and metastasis (see Refs. 1 and 9 for review).
Several reports have recently implicated gastrins in the activation of proteins involved in epithelial cell/matrix adhesion and morphology, such as p130 Cas and paxillin (9). Recent results showing the phosphorylation and activation of p125 FAK by G 17 -NH 2 via a Src-related pathway are also compatible with a role in cell motility (25), as the role of this non-receptor tyrosine kinase in cell migration is well documented (26). It is worth noting, however, that G 17 -NH 2 was shown to have an inhibitory effect on the spontaneous motility of glioblastoma cell lines (27). The results reported in this paper provide, to our knowledge, the first demonstration of a direct stimulatory effect for a non-amidated progastrin-derived peptide on cell/cell adhesion between epithelial cells.
This effect was detected at G 17 -Gly concentrations as low as 100 pM. Serum concentrations of G 17 -Gly are generally thought to be around 20 -50 pM in the fasting state. However, gastrin precursor production is significantly increased in various physiological conditions such as birth and weaning (28,29), as well as during pathological processes such as gastric adenocarcinoma (30) and gastrinoma (31,32). The migratory effect of G 17 -Gly was maximal around 5 nM, as compared with 1 nM for the maximal proliferative effect of the peptide. This slight difference in potency could be explained by differences in the sensitivity of the signaling pathways involved. Furthermore, previous in vitro investigations of G 17 -Gly effects on renal (33), pancreatic (3,5), colonic (6, 10), or gastric cells (34,35) showed a maximal effective dose for G 17 -Gly varying between 0.1 (14) and 10 nM (3), even when the apparent affinity of G 17 -Gly receptors seemed higher (see Refs. 33-35 and this work). This apparent discrepancy might indicate that the G 17 -Gly receptors could exist in various affinity states or that fractional occupancy only is necessary to activate the biological effects downstream of these receptors.
The signal transduction pathways activated by G 17 -NH 2 in IMGE-5 cells have been reported previously in some, but not all, cell types. The results obtained in this study indicate that MAP kinase activation is essential for the proliferative effect of G 17 -NH 2 on IMGE-5 cells, and Stepan et al. (10) recently showed that G 17 -NH 2 stimulated proliferation through a pathway involving MAP kinases in AR42J cells. In contrast, in GH3 cells, G 17 -NH 2 did not activate this pathway but induced proliferation in a Ca 2ϩ -dependent manner. In previous studies, activation of G/CCK-B receptors by G 17 -NH 2 has been shown to induce PI3-kinase activation in transfected Chinese hamster ovary cells (13), probably via the prior formation of a p60 Src -p125 FAK complex (25). Thus, there seems to be a certain degree of cell type specificity in the transduction pathways involved in the proliferative effects of G 17 -NH 2 .
Much less is known about the signal transduction pathways activated by G 17 -Gly. Reports by Todisco et al. (3) on pancreatic carcinoma cells and Stepan et al. (14) on colorectal carcinoma cells indicate that G 17 -Gly regulates the transcriptional activation of early genes, through the activation of enzymes such as c-Jun kinase. In these cell lines, the MAP kinase pathway is not involved in the proliferative effect of G 17 -Gly (3,13). On the contrary, our results show that p42/p44 MAP kinase activation by G 17 -Gly as well as G 17 -NH 2 was involved in their proliferative effect on IMGE-5 cells. However, PI3-kinase was involved selectively in the proliferative effect of G 17 -Gly only. Further studies on other cell lines will be necessary to assess whether there is also a cell type specificity in the transduction pathways triggered by G 17 -Gly, whether there are several subtypes of G 17 -Gly receptors, or whether the signal transduction coupled to these receptors is different in tumor cell lines.
The novel effect of G 17 -Gly on dissociation and migration of gastric epithelial cells involved both PI3-kinase and MAP kinase pathways. Activation of PI3-kinase was essential at least for cell dissociation, whereas the MAP kinase pathway seemed to be involved only in the motility response of IMGE-5 cells, without affecting their dissociation. This result is interesting, as the role of these two pathways in the dissociating effect of growth factors on other cell types is still unclear. PI3-kinase activation has been shown to participate in the ligand-induced migration of several cell types including renal epithelial cells (36) and vascular smooth muscle cells (37). However, the specific involvement of these pathways in the successive steps of a migratory response, i.e. cell dissociation and motility, has not been clearly defined. A report by Potempa and Ridley (38) showed that both PI3-kinase and MAP kinase activation (by Ras) seemed essential to hepatocyte growth factor-induced adherens junction disassembly in Madin-Darby canine kidney cells. However, in the same cell line, Royal et al. (36) reported that hepatocyte growth factor-induced motility required PI3kinase activation, whereas pathways downstream from Grb2 (including MAP kinases) were involved in branching tubulogenesis. Recent results on HepG2 human hepatoma cells also showed that PI3-kinase was involved in cell dissociation, whereas inhibition of MEK blocked the motility response to growth factors (39). Our results also directly implicate the PI3-kinase pathway in the G 17 -Gly-induced tyrosine phosphorylation of ␤-catenin, as well as its dissociation from E-cadherin and delocalization from the adherens junctions. In contrast to HepG2 cells, the MAP kinase pathway does not appear to be involved in this G 17 -Gly-induced event in IMGE-5 cells.
We also found in this study that the time course of MAP kinase activation by G 17 -Gly and G 17 -NH 2 was different, with the former triggering a rapid but short lived phosphorylation, whereas the latter induced a long term activation lasting for 3 h. Furthermore, we showed that the early activation of p42/44 MAP kinases is essential to the proliferative effect of G 17 -NH 2 and G 17 -Gly, whereas a longer term activation seems necessary for the migratory response to G 17 -Gly. A previous study (39) suggested a correlation between the lack of effect of epidermal growth factors on HepG2 cell motility and its ability to induce only a short term increase in the phosphorylation of p42/p44 MAP kinases, whereas factors inducing a scattering of these cells, like hepatocyte growth factor, stimulated MAP kinases for a longer time. However, to our knowledge, this is the first direct demonstration that long term activation of MAP kinases is essential to the migratory response to an exogenous factor.
Interestingly, the stimulation induced by G 17 -Gly triggered a delayed increase in the activation of Akt/PKB, which was not detected after stimulation with G 17 -NH 2 . This result contradicts recent data (40) showing the activation of an Akt-dependent anti-apoptotic pathway by the formation of E-cadherinmediated cell/cell contacts in Madin-Darby canine kidney cells. However, the existence of a similar mechanism has been recently demonstrated by Taupin et al. (41), who showed that the migratory effect of intestinal trefoil factor on intestinal cell lines is coupled to anti-apoptotic signals. It is possible that a balancing mechanism would induce activation of an anti-apoptotic pathway when cells are induced to migrate by an exogenous physiological activator. Such a mechanism would be necessary when cells need to migrate in vivo during the epithelial/mesenchymal transition or during processes leading to mucosal restitution and ulcer repair.
Our data showed that IMGE-5 cells are sensitive to both amidated and glycine-extended forms of gastrin. To date the only cell line responding to both molecular forms of gastrin is the rat pancreatic carcinoma cell line AR42J (5). The ligand selectivity of the G 17 -Gly receptor identified on IMGE-5 cells was similar to that described previously (6) both on AR42J and on YAMC cells. Furthermore, the expression of receptors for both G 17 -Gly and G 17 -NH 2 was correlated to the differentiation status of IMGE-5 cells. Although IMGE-5 cells do not express G/CCK-B receptors under permissive conditions, they still express binding sites for G 17 -Gly, albeit at a lower density. IMGE-5 could therefore represent a unique tool for the parallel and independent study of biological effects and signal transduction pathways associated with G 17 -Gly and G 17 -NH 2 activation.
Finally, our results demonstrating an effect of G 17 -Gly on the migration of gastric epithelial cells could reflect a potential physiological role for this peptide during ontogeny, in gastroduodenal ulcer disease, or during the progression of carcinomas. Although the available results are still scarce, there seems to be a general tendency toward the expression of partially processed rather than mature forms of gastrin in the early stages of development (at a stage when migration of cells to form the gastric pits is maximal), as well as during carcinoma development (42)(43)(44). The role of these partially processed forms in colonic proliferation is well established in vivo (1,9,45). Furthermore, a correlation could exist between serum concentrations of total progastrin products and the presence of liver metastasis in colorectal cancer (46) as well as in patients affected by the rare Zollinger-Ellison syndrome (32). Furthermore, antibodies neutralizing both amidated and glycineextended forms of gastrin have been shown to inhibit the spontaneous metastasis of a human colorectal tumor when injected into immunodeficient mice (47). Therefore, the potential role of progastrin-derived peptides on migration needs to be further investigated in other models, in order to assess to what extent the results presented in this paper extend beyond the gastric mucosa and represent a general regulatory mechanism in the gastrointestinal tract.