Intestinal epithelial cell differentiation involves activation of p38 mitogen-activated protein kinase that regulates the homeobox transcription factor CDX2.

The intracellular signaling pathways responsible for cell cycle arrest and differentiation along the crypt-villus axis of the human small intestine remain largely unknown. p38 mitogen-activated protein kinases (MAPKs) have recently emerged as key modulators of various vertebrate cell differentiation processes. In order to elucidate further the mechanism(s) responsible for the loss of proliferative potential once committed intestinal cells begin to differentiate, the role and regulation of p38 MAPK with regard to differentiation were analyzed in both intact epithelium as well as in well established intestinal cell models recapitulating the crypt-villus axis in vitro. Results show that phosphorylated and active forms of p38 were detected primarily in the nuclei of differentiated villus cells. Inhibition of p38 MAPK signaling by 2-20 microm SB203580 did not affect E2F-dependent transcriptional activity in subconfluent Caco-2/15 or HIEC cells. p38 MAPK activity dramatically increased as soon as Caco-2/15 cells reached confluence, whereas addition of SB203580 during differentiation of Caco-2/15 cells strongly attenuated sucrase-isomaltase gene and protein expression as well as protein expression of villin and alkaline phosphatase. The binding of CDX2 to the sucrase-isomaltase promoter and its transcriptional activity were significantly reduced by SB203580. Pull-down glutathione S-transferase and immunoprecipitation experiments demonstrated a direct interaction of CDX3 with p38. Finally, p38-dependent phosphorylation of CDX3 was observed in differentiating Caco-2/15 cells. Taken together, our results indicate that p38 MAPK may be involved in the regulation of CDX2/3 function and intestinal cell differentiation.

The epithelium of the small intestine is a highly dynamic system continuously renewed by a process involving cell generation and migration from the stem cell population located at the bottom of the crypt to the extrusion of the terminally differentiated cells at the tip of the villus (1,2). The crypt-villus functional axis unit, which develops relatively early during human ontogeny (being established by mid-pregnancy), can be defined by typical morphological and functional properties displayed by the mature villus enterocytes that distinguish them from crypt cells (1)(2)(3). Indeed, the villi are mainly lined by functional absorptive, goblet, and endocrine cells, whereas the crypts contain stem cells, proliferative and poorly differentiated cells, as well as a subset of differentiated secretory cells, namely Paneth cells (3). The differentiation of each cell type takes place as the cells move either upward toward the villus (absorptive, mucus and endocrine cells) or downward to concentrate at the bottom of the crypt (Paneth cells) (2). The basic mechanisms responsible for induction of cell differentiation are little understood. The decision to differentiate is taken by the committed crypt cells abruptly, while in their most rapid state of proliferation (4). The newly differentiated cells acquire their distinctive ultrastructural features and cell surface markers after leaving the proliferative cell cycle, at the top of the crypts or the base of the villi (2,(5)(6)(7). It is noteworthy that in all species studied, the crypt-villus axis junction represents a physical limit from which enterocytes acquire their final functional characteristics (1)(2)(3)(4)(5)(6)(7).
The process of cell differentiation in the intestinal epithelium has been the subject of extensive studies, for which the morphological and functional characteristics of the intestinal mucosa (8,9), the growth kinetics of the epithelial cells (10), and the chronological changes that affect brush border enzyme activities during pre-and postnatal development (11,12) have all been well documented. However, the basic mechanisms involved in the induction and the modulation of cell differentiation in the upper portion of the crypts, and the cellular interactions responsible for the orderly arrangement of the relative numbers of proliferative, maturing, and functional epithelial cells are still largely unknown. Hormones, such as glucocorticoids, and growth factors, such as epidermal growth factor (EGF), 1 have been implicated in the regulation of intestinal growth and development (12,13). However, little is known about the molecular signals responsible for the ontogenic changes in intestinal gene expression.
Several lines of evidence suggest that the intestinal specific, caudal-related cdx1 and cdx2/3 homeobox genes encode nu-clear transcription factors that play critical roles in intestinal cell proliferation and differentiation. CDX1 is mainly expressed in the crypt compartment although not restricted to proliferative cells (14), and its inhibition by antisense RNA reduces cell proliferation in vitro (15). The CDX2/3 homeoproteins (the protein designated CDX3 in the hamster and CDX2 in the mouse and humans) are mainly expressed in differentiating enterocytes (16), triggering growth retardation and cell differentiation by overexpression in several intestinal lines in vitro (15,17,18). Furthermore, genes regulated by either CDX1 or CDX2/3 generally define a functional differentiated phenotype (for example, sucrase-isomaltase (18), glucagon (19), intestinal phospholipase A/lysophospholipase (20), carbonic anhydrase (21), and lactase (22)). However, little is known about the intracellular signaling pathways that positively regulate the activities of CDX transcription factors, especially those involved in receiving and transducing extracellular cues.
In eukaryotic cells, the mitogen-activated protein kinase (MAPK) family has been shown to play various important roles in regulating gene expression via transcription factor phosphorylation (23)(24)(25)(26). In mammals, two distinct classes have been identified to date as follows: p42-p44 (extracellular signalregulated kinase) MAPKs inducible by growth factors, and SAPKs (stress-activated protein kinases), which include p38 MAPKs and p46-p54 JNKs, inducible by cytokines and cellular stress (27). Unique structural features, specific activation pathways, and varying substrate specificities support the contention that different MAPKs are independently regulated and control different cellular responses to extracellular stimuli (28,29). We (30) and others (31) recently analyzed the role and regulation of p42/p44 MAPKs in the process of proliferation and differentiation of human intestinal cells. Our results demonstrated that elevated p42/p44 MAPK activities stimulated cell cycle progression of intestinal epithelial cells, whereas low sustained levels were correlated with G 1 arrest and differentiation. However, the intracellular pathways responsible for establishment of differentiated cells occupying specific positions along the gut axis still remain largely unknown.
Several recent studies have demonstrated that p38 MAPK is involved in various vertebrate cell differentiation processes, namely adipocytic (32) and myogenic differentiation (33). The role of p38 MAPK in intestinal cell differentiation is, however, not known. In the present work, the role and regulation of p38 MAPK were analyzed in relation to human intestinal cell proliferation and differentiation in the intact epithelium as well as in well established intestinal cell models that allow the recapitulation of the crypt-villus axis in vitro as follows: Caco-2/15 cells, which have the ability to differentiate into fully functional villus-like enterocytes (34 -37); normal crypt-like HIEC cells, which are proliferative and undifferentiated (38); and finally PCDE cells, which are primary cultures of differentiated and non-proliferative villus enterocytes (39). By using a combination of different approaches, p38 MAPK was found to be activated rapidly in intestinal cells induced to differentiate. Specific inhibition of p38 significantly reduced the expression of several differentiation markers including sucrase-isomaltase, alkaline phosphatase, lactase, and villin. Finally, p38 exerted its stimulatory effect on intestinal differentiation by directly interacting with CDX2/3 and enhancing its transcriptional activity.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP and the enhanced chemiluminescence (ECL) immunodetection system were obtained from Amersham Pharmacia Biotech. Antiserum that specifically recognizes p38␣ on Western blots (40) was a kind gift from Dr. J. Landry (Laval University, Québec, Canada). Rabbit polyclonal antibodies against phosphorylated and active forms of p38 MAPK were from New England Biolabs (Mississauga, Ontario, Canada). Mouse monoclonal antibody against pRb (14001A) was purchased from PharMingen (Mississauga, Ontario, Canada). Monoclonal antibody HSI-14 (41) against sucrase-isomaltase was kindly provided by Dr. A. Quaroni (Cornell University, Ithaca, NY). Monoclonal antibody CII10 recognizing the 89-kDa apoptotic fragment and the 113-kDa non-cleaved fragment of poly(ADP-ribose) polymerase (PARP) was a kind gift from Dr. G. G. Poirier (Laval University, Québec, Canada). Polyclonal antibodies against CDX2/3 protein were provided by Dr. D. J. Drucker (University of Toronto, Ontario, Canada) (42). The monoclonal HA antibody raised against a peptide from influenza hemagglutinin HA1 protein was purchased from Santa Cruz Biotechnologies (Santa Cruz, CA). Goat anti-rabbit IgG-fluorescein isothiocyanate (FITC) and goat anti-mouse IgG-fluorescein isothiocyanate were from Roche Molecular Biochemicals. The specific inhibitors of MEK1/2 (PD98059) and of p38␣/␤ (SB203580) were purchased from Calbiochem. EGF was obtained from Collaborative Biomedicals (Bedford, MA), and insulin was from Connaught Novo Laboratories (Willowdale, Ontario, Canada). All other materials were obtained from Sigma-Aldrich unless stated otherwise.
Specimens and Indirect Immunofluorescence-Tissues from five fetuses of 20 weeks of gestation (post-fertilization fetal ages were estimated according to Streeter (43)) were obtained from normal elective pregnancy terminations. No tissue was collected from cases associated with known fetal abnormalities or fetal death. All studies were approved by the Institutional Human Subject Review Board. Segments of fetal small intestine were rinsed with 0.15 M NaCl, sectioned into small fragments, embedded in optimum cutting temperature compound, and quickly frozen in liquid nitrogen (36). Frozen sections 2-3 m thick were spread on silane-coated glass slides and air-dried for 1 h at room temperature before storage at Ϫ80°C. For indirect immunofluorescence, sections were fixed with 2% formaldehyde in phosphate-buffered saline (pH 7.4; 45 min, 4°C), before immunostaining as described previously (37). Negative controls (no primary antibody) were included in all experiments. Nuclei were stained with propidium iodide as per instructions of the manufacturer (Molecular Probes, Eugene, OR).
Cell Culture-The Caco-2/15 cell line was obtained from A. Quaroni (Cornell University, Ithaca, NY). This clone of the parent Caco-2 cell line (HTB 37; American Type Culture Collection, Manassas, VA) has been characterized extensively elsewhere (30,34,36,37) and was selected originally as expressing the highest level of sucrase-isomaltase among 16 clones obtained by random cloning. This cell line was cultured in plastic dishes in Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Inc.) containing 10% fetal bovine serum (FBS), as described previously. Caco-2/15 cells were used between passages 53 and 78. Studies were performed on cultures at subconfluence (50 -70% confluence), confluence, and between 2 and 40 days post-confluence. Human intestinal epithelial cells (HIEC) were cultured as described (38) in DMEM supplemented with 4 mM glutamine, 20 mM HEPES, 50 units/ml penicillin, 50 g/ml streptomycin, 5 ng/ml recombinant human epidermal growth factor, 0.2 IU/ml insulin, and 5% FBS. Primary cultures of human differentiated enterocytes (PCDE) prepared from specimens of small intestine from fetuses ranging from 18 to 20 weeks of age, were cultured in supplemented DMEM as described above for HIEC (39). When tested after 5-7 days, these primary cultures of differentiated enterocytes were well preserved; both goblet and absorptive cells exhibited the main characteristics of intact villus intestinal cells (39).
Protein Expression and Immunoblotting-Cells were lysed in SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, 2.3% SDS, 10% glycerol, 5% ␤-mercaptoethanol, 0.005% bromphenol blue, 1 mM phenylmethylsulfonyl fluoride (PMSF)). Proteins (40 g) from whole cell lysates were separated by SDS-polyacrylamide gel electrophoresis (PAGE) in 7.5 or 10% gels. Proteins were detected immunologically following electrotransfer onto nitrocellulose membranes (Amersham Pharmacia Biotech). Protein and molecular weight markers (Bio-Rad) were localized by staining with Ponceau Red. Membranes were blocked for 3 h at 25°C in phosphate-buffered saline containing 10% powdered milk. Membranes were then incubated overnight with primary antibodies in blocking solution and with horseradish peroxidase-conjugated goat antimouse or anti-rabbit (1:1000) IgG for 1 h. The blots were visualized by the Amersham Pharmacia Biotech ECL system. Protein concentrations were measured using a modified Lowry procedure with bovine serum albumin as standard (44).
p38 MAPK Assay-The cells were lysed for 10 min on ice with 1 ml/dish of lysis buffer (150 mM NaCl, 1 mM EDTA, 40 mM Tris, pH 7.6, 1% Triton X-100) supplemented with protease inhibitors (0.1 mM PMSF, 10 g/ml leupeptin, 1 g/ml pepstatin, 10 g/ml aprotinin) and phosphatase inhibitors (0.1 mM orthovanadate, 20 mM para-nitrophenyl phosphate, 40 mM ␤-glycerophosphate). Lysates (400 g) cleared by centrifugation (10 000 ϫ g, 10 min) were incubated for 2 h at 4°C with protein A-Sepharose (Amersham Pharmacia Biotech) that had been preincubated for 1 h with anti-p38␣. Immunocomplexes were then washed four times with ice-cold lysis buffer and three times with ice-cold kinase buffer (20 mM para-nitrophenyl phosphate, 10 mM MgCl 2 , 1 mM dithiothreitol, 30 mM HEPES, pH 7.4) before performing the kinase assay. The kinase reaction was initiated by incubating the immunocomplexes at 30°C in the presence of the substrate myelin basic protein (MBP) and [␥-32 P]ATP at 20 -100 M, 1-5 Ci/assay. After 30 min, the reaction was stopped by addition of Laemmli's buffer. Radiolabeled substrates were separated from immunocomplexes by SDS-PAGE and autoradiographed. Incorporation of 32 P by MBP was linear over the course of the kinase assay.
Expression Vectors and Reporter Constructs-The sucrase-isomaltase reporter construct used for luciferase assays contained the human sucrase-isomaltase promoter from residues Ϫ183 to ϩ54 cloned upstream of the luciferase gene of the pGL2 reporter construct as described previously (Dr. P. G. Traber, University of Pennsylvania, Philadelphia) (45). Plasmid E2F SV40-luc, which contains a high affinity E2F-binding site from the dihydrofolate reductase (DHFR) promoter coupled to a luciferase gene (46,47), was a kind gift of Dr. P. Farnham (University of Wisconsin). The expression vectors for wild-type p38␣ and the dominant-negative mutant p38␣ (kindly provided by Dr. J. Pouysségur, Université de Nice, Nice, France) were previously cloned into pECE vector. The hamster CDX3 expression vector was a gift from Dr. W. J. Rutter (University of California, San Francisco) (48).
Northern Blot Analysis-Total cellular RNAs were prepared from Caco-2/15 cells at subconfluence, confluence, and 3 and 6 days postconfluence by the guanidinium isothiocyanate/phenol method (TRIZOL, Life Technologies, Inc.) as described before (49). RNAs were subjected to agarose gel electrophoresis with formaldehyde and transferred to nylon membranes (Nytran, Schleicher & Schuell). Equal RNA loading was confirmed by hybridization to an ␣-tubulin probe. Hybridizations were performed with a random-primed 32 P-labeled probe (Amersham Pharmacia Biotech) of a PCR-amplified human CDX2 fragment from nucleotides 1102 to 1706.
Transient Transfections and Luciferase Assays-First, subconfluent Caco-2/15 cells were seeded in 24-well plates and transfected by lipofection (Lipofectin, Life Technologies, Inc.) as described before (30) with 0.1 g of E2F-SV40-luciferase reporter per well. One day after transfection, cells were exposed to 2-20 M SB203580 or 20 M PD98059 for 24 h, and luciferase activity was measured. The increase in luciferase activity was calculated relative to the basal level of E2F-SV40-luciferase set at 1 and corrected for the empty vector effects. Second, 1 day post-confluent Caco-2/15 cells were seeded in 24-well plates and cotransfected by lipofection (LipofectAMINE 2000, Life Technologies, Inc.) as described previously (30) with 0.1 g of SI-luciferase reporter and 0.1 g of the relevant expression vector (pECE) containing wildtype or dominant-negative mutant of p38␣ per well. In some experiments, 0.05 g of wild-type CDX3 expression vector was co-transfected. One day after transfection, cells were treated with or without 2-20 M of SB203580 for 24 h, and luciferase activity was measured. The pRL-SV40 Renilla luciferase vector (Promega, Madison, WI) was used as a control for transfection efficiency. Two days after transfection, luciferase activity was measured according to the Promega protocol.
Gal4 Assays-The 540-base pair sequence encoding the 180-amino acid transactivation domain of CDX3 (42) was PCR-amplified and cloned in frame with the DNA-binding domain of GAL4 in the mammalian expression vector pM2 (50). The expression vector was transfected by lipofection (Lipofectin, Life Technologies, Inc.) in Caco-2/15 cells with the pFR-luciferase reporter vector containing five tandem repeats of the GAL4-binding element upstream of a basic promoter element (Stratagene, La Jolla, CA). Luciferase activity was assessed after a 24-h treatment with Me 2 SO, 20 M SB203580, or 20 M PD98059.
Determination of Brush Border Enzyme Activity-Caco-2/15 cells, treated with or without 20 M SB203580, were harvested in water at confluence (day 0) and at 3, 6, and 9 days post-confluence, and sonicated. The disaccharidases sucrase-isomaltase and lactase-phlorizin were assayed using the method of Dahlqvist as modified by Ménard and Arsenault (51). Alkaline phosphatase was assayed by the method of Eichholz (52). Dipeptidyl peptidase IV (DPPIV) activity was assayed according to the method of Roncari and Zuber (53), with glycyl-Lproline-p-nitroanilide as substrate. Total homogenate protein content was determined using a modified Lowry procedure with bovine serum albumin as standard (44). Data were expressed in international units (micromoles of substrate hydrolyzed per min) per g of protein.
GST Fusion Protein Purification-Hamster CDX3 was ligated downstream of the glutathione S-transferase sequence in a pGEX plasmid (Amersham Pharmacia Biotech). The recombinant plasmid was introduced into Escherichia coli BL21 DE3, and the fusion protein was produced by growing 50 ml of a bacterial culture to an optical density between 0.9 and 1.1 and then treating the cultures with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside for 1.5-3 h. Cells were recovered and resuspended in 1.5 ml of SB buffer (16 mM sodium phosphate, pH 7.4, 150 mM NaCl, 15% glycerol, 0.02% Triton X-100, 1 mM dithiothreitol, 15 g/ml leupeptin, 5 g/ml aprotinin, 1 g/ml pepstatin A) and sonicated. Triton X-100 was added to the lysates to a final concentration of 1%. The bacterial lysates were incubated on ice for 15 min and centrifuged at 12,000 rpm for 15 min. The supernatants were recovered and mixed with 55 l of a 1:1 suspension of glutathione-Sepharose 4B beads, and the mixture was rotated at 4°C for 1 h. The beads were then washed extensively in SB buffer and used for in vitro binding assays, as described (57).
GST Pull-down Assays-Caco-2/15 cells were grown in 60-mm dishes to confluence in DMEM supplemented with 10% FBS. The cells were lysed in 700 l of lysis buffer (150 mM NaCl, 1 mM EDTA, 40 mM Tris, pH 7.6, 1% Triton X-100) supplemented with protease inhibitors (0.1 mM PMSF, 10 g/ml leupeptin, 1 g/ml pepstatin, 10 g/ml aprotinin) and phosphatase inhibitors (0.1 mM orthovanadate, 20 mM para-nitrophenyl phosphate, 40 mM ␤-glycerophosphate). Lysates were centrifuged at 15,000 ϫ g for 10 min. 5 g of GST-CDX3 or GST proteins were coupled to 50 l of glutathione-Sepharose (Amersham Pharmacia Biotech). The Caco-2/15 lysates were incubated for 2 h at 4°C with the immobilized fusion proteins by end-over-end rotation. The beads were washed four times with the lysis buffer. Laemmli buffer was added to the beads, and the mixture was boiled for 5 min. Bound proteins were visualized by SDS-PAGE (9% acrylamide gels) and immunoblotting as described above. In other experiments, the beads were washed four times with lysis buffer followed by three times with ice-cold kinase buffer before performing the kinase assay in the presence or absence of 0.1-20 M SB203580.
Data Presentation and Statistical Analysis-Luciferase assays were performed in triplicate, and results were analyzed by the Student's t test and were considered significantly different at p Ͻ 0.05. Typical Western blots, representative of two or three independent experiments, are shown. Densitometric analyses were carried out for each Western blot.

Activity of p38␣ MAPK in the Human Fetal Intestinal
Epithelium-Phosphorylation and activity of p38 MAPKs were investigated in intact fetal intestinal epithelium (20 weeks of gestation). It is generally agreed that by 16 -18 weeks of gestation, the overall morphological appearance of the small intestine and the expression of most of the functional markers including sucrase-isomaltase are comparable to those in adult intestine (3,58). However, lactase represents an exception, with its activity increasing between 36 and 40 weeks of gestation (58). The use of a specific antibody against p38, phosphorylated on the TGY motif, revealed that phosphorylated and active p38 MAPKs were mostly localized in the nuclei of all villus cells (Fig. 1A, see arrows), whereas the intensity of staining was significantly decreased in the crypt.
Activation of p38 MAPK during Differentiation of Intestinal Cells-Caco-2/15 cells that differentiate spontaneously to a small bowel phenotype after confluence (30, 34 -37) were harvested at 70 (day Ϫ2) and 100% confluence (day 0), and 3, 6, 10, 16, 25, and 31 days post-confluence, and analyzed by Western blot to confirm timing of induction of sucrase-isomaltase pro-p38 MAPK and Intestinal Epithelial Cell Differentiation tein expression. Consistent with previous observations (30,34,37), sucrase-isomaltase protein levels significantly increased at 3 days post-confluence ( Fig. 2A). Expression and kinase activity of p38␣ MAPK were also analyzed by immunoprecipitation. As shown in Fig. 2B, p38␣ abundance did not change with differentiation of Caco-2/15 cells. However, differential regulation of p38␣ kinase activity was observed during the differentiation of Caco-2/15 cells. As demonstrated in Fig. 2B, immunoprecipitated p38␣ exhibited very low basal activity in phosphorylating MBP in subconfluent growing Caco-2/15 cells, in contrast to a dramatic induction of p38␣ activity when cells reached confluence (day 0). This activation persisted during cell differentiation. Furthermore, Western blot analysis with an antibody recognizing the biphosphorylated and active p38 MAPK isoforms revealed that p38␣ phosphorylation significantly increased as soon as Caco-2/15 cells reached confluence (Fig. 2C). These results imply that p38␣ activation precedes the induction of sucrase-isomaltase, a differentiation marker. Of note, p38␤ protein was never detected in Caco-2/15 cells by Western blotting, and very low levels of RNA were detected by reverse transcriptase-PCR analysis. 2 Thus, p38␣ MAPK activation may be functionally linked to intestinal differentiation.
Because Caco-2/15 cells are derived from a human colonic adenocarcinoma (59), we wanted to support our results in normal human small intestine-derived cells. We then analyzed the phosphorylation of p38 MAPK in normal human intestinal cell models as follows: the crypt-like HIEC cells that are proliferative and undifferentiated (38), and the PCDE cells that are primary cultures of differentiated and non-proliferative villus enterocytes (39). Cell lysates were prepared from subconfluent growing HIEC cells and from PCDE cells. As shown in Fig. 2D, phosphorylation of p38 was significantly lower in subconfluent growing HIEC cells compared with that found in PCDE cells. Hence, p38 MAPK exhibits similar patterns of activity if we compare pre-and post-confluent Caco-2/15 cells versus normal HIEC and PCDE cells that together allow the in vitro repro-duction of the normal crypt-villus axis (9).
p38 MAPK Is Not Involved in Cell Cycle Progression of Intestinal Epithelial Cells-An important early event in the terminal differentiation of cells, especially in tissues exhibiting a rapid turnover such as the intestinal epithelium, is their withdrawal from the cell cycle (60,61). To evaluate the role of p38 MAPKs in intestinal cell cycle progression, SB203580 compound, a specific inhibitor of p38␣/␤ MAPKs (62), was tested on dihydrofolate reductase (DHFR) expression in subconfluent Caco-2/15 and HIEC cells. The DHFR gene, which is required for DNA synthesis and is transcribed at the G 1 /S transition, contains E2F-dependent binding sites in the promoter. In addition, microinjection of E2F into quiescent fibroblasts provokes S phase re-entry, underscoring the importance of E2F in cell growth control (47). Therefore, the plasmid construction containing the E2F-responsive DHFR promoter linked to a luciferase reporter gene represents a sensitive reporter of cell cycle progression and S phase entry (46). When the p38␣/␤ MAPKs were blocked with the SB203580 compound (2-20 M), E2F-dependent luciferase expression was not significantly affected in either cell line (Fig. 3A). In contrast, the MEK1 inhibitor PD98059 significantly reduced E2F-regulated reporter gene expression by 50% in Caco/15 cells and by 81% in HIEC compared with control untreated cells. These results confirm our previous observations (30)  We recently demonstrated that Caco-2/15 cells slowed their cell cycle at confluence to become almost completely arrested in the G 1 phase by day 6 post-confluence. Indeed, decreased phosphorylation of retinoblastoma proteins and reduced Cdk2 activity correlated with the induction of differentiation markers, namely sucrase-isomaltase and villin (61). To determine whether p38 MAPK activation plays a significant role in cell cycle arrest of confluent Caco-2/15 cells, the consequences of blocking p38 MAPK with SB203580 were examined on pRb phosphorylation by using specific antibodies detecting the active hypophosphorylated form of p105Rb protein (lower band) as well as the inactive hyperphosphorylated forms of the protein (upper bands). As shown in Fig. 3B, addition of SB203580 at days 0 -6 post-confluence had no significant effect on the decrease in pRb phosphorylation observed at days 3 and 6 post-confluence suggesting that p38 MAPK is not involved in the loss of proliferative potential as committed intestinal cells begin to differentiate.
p38 MAPKs appear to play important roles in the regulation of cell survival in other cell types (63,64). To verify the potential effect of the inhibition of p38 MAPKs on Caco-2/15 cell survival, expression of PARP, a well known substrate for caspase-3 (65), was measured in cells treated with SB203580. As shown in Fig. 3B, treatment of confluent Caco-2/15 cells with SB203580 had no effect on PARP cleavage, suggesting that persistent inhibition of p38 MAPK did not affect Caco-2/15 cell survival.
Inhibition of p38 MAPK Activity Prevents Enterocyte Differentiation-The dramatic induction of p38 MAPK activity led us to verify whether this MAPK was associated with differentiation of Caco-2/15 cells. Daily addition of SB203580 at confluence repressed sucrase-isomaltase protein expression in confluent Caco-2/15 cells (Fig. 4A). In addition, induction of villin expression was also decreased by 2-3-fold at days 3, 7, and 12 post-confluence. Equal protein loading was confirmed by using an anti-actin antibody (Fig. 4A). Enzymatic assays were also performed in order to verify the induction pattern of other differentiation markers, namely lactase, DPPIV, and alkaline phosphatase, in control and in SB203580-treated cells. As shown in Table I, treatment of confluent Caco-2/15 cells with 20 M SB203580 did not significantly affect the induction of DPPIV at post-confluence, whereas induction of lactase and alkaline phosphatase was significantly attenuated ( Table I).
The effect of p38 on sucrase-isomaltase gene expression was further analyzed by transiently transfecting newly confluent Caco-2/15 cells with the luciferase gene driven by the human sucrase-isomaltase promoter (45). As shown in Fig. 4B, sucrase-isomaltase gene expression was inhibited in a dose-dependent manner by the p38 inhibitor, SB203580, with maximal effect observed at 20 M (72% inhibition). Furthermore, in contrast to wild-type p38␣, ectopic expression of the dominantnegative mutant of p38␣ significantly reduced sucrase-isomaltase gene expression by 61%. Collectively, these results indicate that p38 activation is an early and necessary event for activation of the intestinal differentiation program, preceding the induction of various differentiation markers.
Of note, addition of SB203580 to differentiating Caco-2/15 cells at days 6 -12 post-confluence still reduced sucraseisomaltase expression by 47%, as observed at day 12 postconfluence (Fig. 4C). In addition, treatment of primary cultures of normal differentiated enterocytes (see "Experimental Procedures") with 20 M SB203580 also significantly reduced the expression of sucrase-isomaltase by 22 and 55% and villin by 38 and 57% after 2 and 4 days of treatment, respectively (Fig. 4D). Equal protein loading was confirmed by using an antikeratin-18 antibody. These data suggest that p38 MAPK activity is required for maximal expression of sucrase-isomaltase in differentiating Caco-2/15 cells and differentiated normal enterocytes.
Transactivation Activity of CDX2/3 Is Down-regulated by Inhibition of the p38 Pathway-The dramatic effect of the p38 inhibitor on the expression of sucrase-isomaltase prompted us to investigate whether p38 also affected the activity of the transcription factor CDX2/3 (the protein designated CDX3 in the hamster and CDX2 in the mouse and humans (66)), a key activator of sucrase-isomaltase transcription (45) and an inducer of intestinal epithelial cell differentiation (18). We first examined whether inhibition of p38 in confluent Caco-2/15 cells affected the expression of CDX2. Northern blot analysis demonstrated that addition of SB203580 at days 0 -6 post-confluence had no effect on CDX2 mRNA expression in confluent Caco-2/15 cells (Fig. 5A). Lysates from COS cells were used as negative control.

p38 MAPK and Intestinal Epithelial Cell Differentiation
The effect of p38 on CDX2 activity was then studied by transiently transfecting Caco-2/15 cells with a CDX3 expression vector. As reported previously (17) with CDX2, ectopic expression of CDX3 resulted in a strong increase (13-fold induction) in sucrase-isomaltase promoter activity. Addition of SB203580 reduced in a dose-dependent manner the inducing effect of CDX3 with maximal inhibition observed at 20 M (70% inhibition) (Fig. 5B, upper panel).
Previous studies revealed that p38 increases the transcriptional activity of various transcription factors by phosphorylation of the transactivation domain (67). To determine whether p38 had a similar effect on CDX2/3, fusion proteins were used containing the transactivation domain of CDX3 (amino acids 1-180) fused to Gal4(DBD) (see "Experimental Procedures"). To assess the transcriptional activity of Gal4-CDX3 proteins, Caco-2/15 cells were co-transfected with a luciferase reporter gene containing five copies of a Gal4 DNA-binding site upstream of a minimal promoter and Gal4-CDX3 expression plasmids. As demonstrated above, inhibition of the p38 pathway by SB203580 significantly reduced CDX3-dependent reporter gene expression (Fig. 5B, lower panel) suggesting that p38 modulates the transcriptional activity of CDX3. In contrast, inhibition of the p42/p44 MAPK pathway with the PD98059 inhibitor had no effect on CDX3-dependent reporter gene expression (Fig. 5B, lower panel).

Effect of SB203580 on the DNA-binding Capacity of Transcription Factors Involved in Sucrase-Isomaltase Expression-
There are three positive regulatory elements for transcription within the sucrase-isomaltase promoter region in intestinal epithelial cells known as sucrase-isomaltase footprint (SIF)-1, SIF-2, and SIF-3. SIF-1 binds the intestine-specific homeodomain transcription factor CDX2/3 (17), whereas SIF-2 and SIF-3 bind the transcription factors HNF-1␣ and HNF1-␤ (68). In addition, a negative cis-acting element SIF-4 may be a binding site for the E4BP4 transcriptional repressor protein (11). Electrophoretic mobility shift experiments were performed to determine whether the DNA-binding capacity of these transcription factors was affected by SB203580. As shown in Fig. 6, binding of nuclear proteins to SIF-1 was not affected by using extracts prepared from SB203580-treated cells except for increased binding observed at day 6 post-confluence, which was reproducibly blocked by SB203580. Furthermore, binding of nuclear proteins to the SIF-3 and SIF-4 oligonucleotides was unchanged with extracts prepared from SB203580-treated cultures.
CDX3 Specifically Interacts with p38 MAPK-To determine whether CDX2/3 can directly associate with p38 MAPK, pulldown assays were performed using the GST-CDX3 fusion protein to absorb naive newly confluent Caco-2/15 cell lysates. The absorbed material was analyzed by Western blot with the p38 antibody. Immunoprecipitated p38 from newly confluent Caco-2/15 cells was used as positive control. As shown in Fig. 7A, a significant amount of p38 MAPK bound to the GST-CDX3 fusion protein was detected. GST protein alone did not pull down the p38 protein (data not shown). Interestingly, the GST-CDX3 fusion did not pull down other MAPKs such as p42 MAPK or JNK1 (data not shown). To determine whether CDX3-p38 association may have some functional relevance, the capacity of pulled down p38 to phosphorylate the GST-CDX3 fusion protein was evaluated in a kinase assay. As shown in Fig. 7B, pulled down p38 efficiently phosphorylated the GST-CDX3 protein. More importantly, this phosphorylation was inhibited in a dose-dependent fashion by the addition of low concentrations of the specific p38 inhibitor SB203580 (50% inhibition with 0.5 M of SB203580). In this regard, an amino acid sequence analysis revealed that CDX2/3 contains putative phosphorylation sites for p38 MAPK (17,67). An in vitro kinase assay using bacterially expressed GST-CDX3 protein or MBP as substrates and p38 immunoprecipitated from newly confluent Caco-2/15 cells revealed that p38 MAPK was able to potently phosphorylate MBP and, more importantly, phosphoryl-

FIG. 4. Modulation of intestinal cell differentiation by SB203580.
A, confluent Caco-2/15 cells (day 0) were treated with Me 2 SO or 20 M SB203580 and were harvested at 3, 7, and 12 days post-confluence. 50 g of cell extracts were separated by 10% SDS-PAGE, and proteins were analyzed by Western blotting to determine the levels of expression of sucrase-isomaltase, villin, and actin. B, 1 day post-confluent Caco-2/15 cells were transfected with 0.1 g of sucraseisomaltase-luciferase reporter vector and 0.1 g of the expression vector pECE alone or containing epitope-tagged wild-type p38␣ (wt) or dominant-negative mutant of p38␣ (Thr 180 -Gly-Tyr 182 /Ala 180 -Gly-Phe 182 ) (DN). One day after transfection, cells were exposed to 5-50 M SB203580 for 24 h, and luciferase activity was measured. The increase in luciferase activity was calculated relative to the Me 2 SO level (0) of sucrase-isomaltase-luciferase, which was set at 1. Results are the mean Ϯ S.E. of at least three separate experiments. *, significantly different from control at p Ͻ 0.05 (Student's t test). C, 6 days postconfluent Caco-2/15 cells were exposed to 20 M SB203580 for 6 days and lysed at day 12, and proteins were separated by 10% SDS-PAGE. Sucrase-isomaltase protein expression was analyzed by Western blotting. D, primary cultures of human differentiated enterocytes (PCDE) were isolated, and after 2 days cells were exposed to 20 M SB203580 for 2 and 4 days and lysed, and proteins were separated by SDS-PAGE. Sucrase-isomaltase and keratin-18 protein expression were analyzed by Western blotting. Similar results were obtained in two different experiments.

p38 MAPK and Intestinal Epithelial Cell Differentiation
ate GST-CDX3 to a significant level (Fig. 7C). This suggests that CDX2/3 may indeed be a specific target for p38 MAPK.
We further verified whether CDX3-p38 association could be detected in vivo. CDX3 was co-transfected into 293T cells with a plasmid encoding wild-type HA-p38␣, and co-immunoprecipitations were performed on total cell lysates. As shown in Fig.  7D (lane 2), CDX3-p38␣ association was easily detected upon immunoprecipitation of HA-p38␣ in 293T cells. Parallel control experiments using non-transfected cells did not precipitate the CDX3 protein under similar conditions (Fig. 7D, lane 1).
EGF Represses p38 MAPK Activity and Sucrase-Isomaltase Expression in Differentiating Caco-2/15 Cells-Sucrase-isomaltase expression has been reported to be down-regulated by growth factors such as keratinocyte growth factor (69) and EGF (70). However, the mechanism involved in the repression of sucrase-isomaltase expression by growth factors is unknown. The effect of EGF was therefore examined on both sucraseisomaltase protein expression and p38 activity. As shown in Fig. 8, chronic treatment (from days 0 to 15) of confluent Caco-2/15 cells with 100 ng/ml EGF repressed sucrase-isomaltase protein expression compared with untreated cells (over 95% inhibition). Of interest, treatment with EGF significantly and persistently down-regulated p38 MAPK activity with maximal effect observed after 2 and 4 h of treatment (Fig. 8). Finally, the inhibitory action of EGF on sucrase-isomaltase expression was not attributable to the re-activation of the p42/ p44 MAPK pathway since the specific MEK inhibitor, PD98059, blocked p42/p44 MAPK activation but did not interfere with the repressive effect of EGF on sucrase-isomaltase expression.

DISCUSSION
The molecular mechanisms orchestrating cellular transitions and changes in gene expression during intestinal epithelial differentiation are largely unknown. In this report, we suggest for the first time the possible involvement of p38 MAPK in intestinal cell differentiation. We show that the activity of p38 was induced as soon as Caco-2/15 cells reached confluence and began differentiating into a small bowel-like phenotype with microvilli formation and expression of disaccharidases. Furthermore, the nuclear localization of p38 MAPK activity in the villi, which is indicative of its functional role in transcription, reflects the distribution of differentiated cells. Inhibition of p38 activity by the specific inhibitor SB203580 did not interfere with cell cycle progression of committed cells but inhibited intestinal cell differentiation and expression of various differentiation markers (namely sucrase-isomaltase, alkaline phosphatase, villin, and lactase). The effects of p38 MAPK on sucrase-isomaltase transcription revealed a regulation of sucrase-isomaltase expression, an effect mediated by CDX2/3. The latter was demonstrated previously to be an important modulator of enterocyte differentiation (17-18, 68).
A key issue in intestinal development is what triggers the differentiation process. Members of the CDX family have been shown to be involved in enterocyte lineage specification (18,71). Once specified, intestinal cells continue to proliferate until they receive a differentiation signal that has yet to be identified. However, in vitro cell culture experiments have shown that cell-cell contact can trigger differentiation and therefore substitute the in vivo signal. In Caco-2 cells, the establishment of cell-cell contact is a critical step initiating both cell cycle exit and induction of the differentiation process (30, 31, 34 -37, 59, 61, 72). Indeed, as with various clones of Caco-2 cell line (35,59,72), the Caco-2/15 clone has been extensively characterized for its ability to differentiate gradually between days 0 and 20 of post-confluence (34,36,37,41). For instance, sucrase-isomaltase transcription increases as soon as Caco-2/15 cells reach confluency (73). In this regard, junctional cell interactions play an important role in the control of cell differentiation during intestinal ontogeny and the continuous cell renewal of the mature organ (74,75). Our results illustrate a pathway by which cell-cell contacts can modulate enterocyte differentiation through activation of a distinct MAPK, the p38 MAPK. With respect to this, it has been observed that the p38 pathway is more efficiently activated in confluent muscle cells than in subconfluent myocytes cultured under the same conditions (33). Furthermore, we have recently shown that in contrast to p38 MAPK, p42/p44 MAPK (30) and p46/p54 JNK 3 activities dramatically decreased as soon as Caco-2/15 cells reached confluence and began to differentiate. Hence, persistent activation of p38 in differentiating Caco-2/15 cells, in the absence of a parallel JNK activation, distinguishes this pathway from those activated in response to stress or cytokines (24,(27)(28)(29)76).
Similar observations were recently reported in differentiating muscle cells (33) and in PC12 (77). Recent data have shown that assembly of E-cadherin-mediated adherens junctions is sufficient to trigger the activation of the PI3-kinase/Akt (78) and p42/p44 MAPK cascades in renal epithelial cells (79). However, the exact mechanisms through which cell-cell contacts activate the p38 MAPK pathway in intestinal epithelial cells remain to be determined. Cell-cell contacts may stimulate the p38 pathway by silencing the activity of a mitogen-dependent factor (e.g. a phosphatase). All together, these observations underscore the importance of cell density in the activation of p38 during cell differentiation. Members of the CDX family act within a regulatory network that establishes the differentiated phenotype of intestinal epithelial cells. Indeed, these homeobox proteins activate the expression of many intestine-specific genes (68). p38 MAPK appears to be a potential activator of CDX2/3 and could control intestinal differentiation. The fact that CDX2/3 plays a role in mediating p38 function in the activation of sucrase-isomaltase transcription and enterocyte differentiation is suggested by the 3 M. Houde, P. Laprise, and N. Rivard, unpublished data.  (48). Four isoforms of p38 are known (␣, ␤, ␥, and ␦) (76). We suggest that p38␣ is the major isoform involved in differentiation of intestinal cells because 1) the inhibitor SB203580, specific to the p38 ␣ and ␤ isoforms, inhibited the differentiation of Caco-2/15 cells; 2) specific immunoprecipitation of the p38␣ isoform from Caco-2/15 cell extracts significantly phosphorylated GST-CDX3; and 3) the p38␤ protein was not detected in differentiating Caco-2/15 cells and primary cultures of differentiated enterocytes (data not shown). Interestingly, the p38␤ isoform was mostly expressed in undifferentiated human intestinal crypt cells (data not shown). Further studies are needed to elucidate the role of the p38␤ isoform in intestinal cells.
Recently, we reported that p42/p44 MAPK activity was repressed during the differentiation of Caco-2/15 cells (30). How- FIG. 5. Transactivation activity of CDX2/3 is down-regulated by inhibition of the p38 pathway. A, expression of CDX2. Total RNA was extracted from COS cells (as negative control) and from Caco-2/15 cells at different periods of confluence (Ϫ2, 0, 3, and 6 days postconfluence), treated with or without 20 M SB203580, and analyzed by Northern hybridization using a 32 P-labeled human CDX2 probe. A murine tubulin probe was used to evaluate the relative amounts of mRNA transferred to the membrane. B, modulation of the transcriptional activity of CDX2 by SB203580. Upper panel, 1 day post-confluent Caco-2/15 cells were transfected with 0.1 g of the sucrase-isomaltaseluciferase reporter vector and 0.05 g of the CDX3 expression vector or the pBAT vector (EV). One day after transfection, cells were exposed to 2.5-20 M SB203580 for 24 h, and luciferase activity was measured. Lower panel, 1 day post-confluent Caco-2/15 cells were transfected with 0.1 g of the pFR-luciferase reporter vector and 0.05 g of the pM2 expression vector encoding the transactivation domain of CDX3 (amino acids 1-180) fused to Gal4(DBD). One day after transfection, cells were exposed to 20 M SB203580 or 20 M PD98059 for 24 h, and luciferase activity was measured. The increase in luciferase activity was calculated relative to the empty vector level of sucrase-isomaltase-luciferase, which was set at 1. Results are the mean Ϯ S.E. of at least three separate experiments. *, significantly different from control at p Ͻ 0.05 (Student's t test).
FIG. 6. Electrophoretic mobility shift assay of SIF1, SIF3, and SIF4 DNA-binding proteins in extracts prepared from control and SB203580-treated Caco-2/15 cells. Binding of nuclear proteins to the SIF1 (CDX2), SIF3 (HNF1), and SIF4 (E4BP4) elements was assessed in Caco-2/15 cells that were incubated in either medium alone (Ϫ) or medium containing 20 M SB203580 since day 0. Cells were harvested at subconfluence (Ϫ2), confluence (day 0), or 2, 6, 9 and 12 days post-confluence. Nuclear extracts were prepared and mixed with 32 P-labeled double-stranded oligonucleotides. DNA-protein complexes were separated from the free probe on a native polyacrylamide gel. The results are representative of two independent experiments. p38 MAPK and Intestinal Epithelial Cell Differentiation ever, significant levels of activated MAPK were detected in differentiated Caco-2/15 cells, predominantly p42 MAPK. We demonstrated that inhibition of MEK activation during differentiation interfered with sustained activation of p42 MAPK and sucrase-isomaltase protein expression, consistent with the conclusion that p42 MAPK is involved in the regulation of sucrase-isomaltase expression in Caco-2/15 cells. However, our data suggest that the p42/p44 and p38 MAPK pathways also exhibit distinct activities. First, p42/p44 MAPK and p38 activ-ities are differentially modulated; p38 is induced whereas p42/ p44 MAPK are strongly reduced in differentiating intestinal cells. Second, inhibition of p42/p44 MAPK with PD98059 only partially prevented sucrase-isomaltase protein expression (30) and did not inhibit the expression of alkaline phosphatase and villin, whereas the inhibition of p38 with SB203580 severely attenuated the expression of various intestinal specific markers. Third, the transcriptional activity of Gal4-CDX3 fusion protein was reduced by SB203580 but not by PD98059. These differences suggest that these two pathways perform distinct functions and that their combined activity may be required for the complete differentiation process, as recently demonstrated in muscle cells (33).
The marked difference observed in sucrase-isomaltase and alkaline phosphatase activities but not in lactase and DPPIV activities in SB203580-treated cells confirms that expression of these brush border enzymes in Caco-2 cells is regulated in different ways (34,70). Furthermore, these results suggest that p38 MAPK targets specific cellular functions rather than the overall program of cell differentiation. Current experiments are in progress to identify other signaling pathways activated early during intestinal differentiation.
EGF plays a major role in intestinal epithelial cell proliferation and maturation (13). Caco-2/15 cells grown in the presence of high concentrations of EGF exhibited increased DNA synthesis and proliferation and formed poorly differentiated multilayers (70). We have shown that EGF inhibits p38 activity as well as sucrase-isomaltase expression. Hence, the p38 pathway may well be an important target for inhibition of the intestinal differentiation program by growth factors in proliferating cells.
An important role for p38␣ MAPK in various mammal cell differentiation processes also has been proposed recently. Adipocytic differentiation of 3T3-L1 fibroblasts induced by insulin was blocked by expression of dominant-negative p38␣ or incubation of the cells with SB203580 (32,80). The transcription factors CCAAT/enhancer-binding protein ␤ (C/EBP␤) and peroxisome proliferator-activated receptor ␥ (PPAR␥) may be p38 targets during adipogenesis (80). In addition, differentiation of C2C12 and L8 myoblasts to myotubes was also mediated by p38␣ activation (81,82). The stimulation of muscle-specific gene expression by p38 was apparently mediated by the myocyte enhancer factor-2C (MEF2C) transcription factor, a p38 substrate known to be essential in myogenesis (33,56). Thus, the function of p38 in cell differentiation correlates with the regulation of the activity (usually associated with phosphorylation) of different transcription factors, for example C/EBP, FIG. 8. EGF represses p38 MAPK activity and sucrase-isomaltase expression in differentiating Caco-2/15 cells. Confluent Caco-2/15 cells were exposed to 100 ng/ml EGF for 15 days, in the presence (ϩ) or absence (Ϫ) of 20 M PD98059, and lysed. Sucraseisomaltase protein expression and p42/p44 MAPK phosphorylation were analyzed by Western blotting. p38 MAPK activity was assessed as described under "Experimental Procedures." Similar results were obtained in three different experiments.

FIG. 7. Specific interaction of CDX2/3 proteins with p38 MAPK.
A, CDX3 associates with p38 kinase in vitro. Lysates from newly confluent Caco-2/15 cells were prepared and incubated with 4 l of p38 antiserum bound to protein A-Sepharose (Ip ␣-p38) or 5 g of GST alone (not shown) or with purified GST-CDX3 bound to glutathione-Sepharose (pull-down GST-CDX3). The beads were washed and resuspended in SDS sample buffer, and the bound material was transferred onto nitrocellulose after SDS-PAGE. Immunological detections were performed using antibodies recognizing p38␣ MAPK. B, p38 phosphorylates CDX3 in a pull-down assay. Lysates from newly confluent Caco-2/15 cells were prepared and incubated with 5 g of GST alone (not shown) or with purified GST-CDX3 bound to glutathione-Sepharose. The beads were washed four times with lysis buffer followed by three times with ice-cold kinase buffer before performing the kinase assay in the presence or in the absence of 0.1-20 M SB203580. The kinase activity is demonstrated by the phosphorylation of GST-CDX3. Similar results were obtained in three different experiments. C, phosphorylation of CDX3 by immunoprecipitated p38 MAPK␣ in an in vitro kinase assay. Kinase assays were performed for 30 min at 30°C with 1 or 5 g of GST-CDX3 or 2 g of MBP, as described under "Experimental Procedures." Similar results were obtained in three different experiments. D, co-immunoprecipitation of CDX3 with HA-p38. Co-transfection of CDX3 and HA-p38␣ expression vectors was performed in 293T cells. 24 h after transfection, cells were lysed. Immunoprecipitations (Ip) of HA-p38␣ were performed using antibody against HA and were analyzed by SDS-PAGE followed by electrotransfer onto nitrocellulose. Immunological detections were performed with an antibody against CDX3 or an antibody against HA. Immunoblots of total lysates show the levels of HA-p38 and CDX3 expression. Similar results were obtained in three different experiments.

p38 MAPK and Intestinal Epithelial Cell Differentiation
MEF2, and CDX2/3 in adipocyte, muscle, and intestinal cell precursors. Although further studies are needed to pinpoint the upstream pathways activating p38 in committed cells induced to differentiate, our study provides novel fundamental insights into the function of p38 in the early events of intestinal epithelial differentiation.