Role of Specific CCAAT/Enhancer-binding Protein Isoforms in Intestinal Epithelial Cells*

Intestinal epithelial cells participate in the acute phase response in response to inflammation. We have shown that acute phase protein genes are induced during intestinal acute phase response, and that the CCAAT/enhancer binding protein family of transcription factors are involved. To address the role of specific C/EBP isoforms, we generated IEC-6 rat intestinal epithelial cell lines expressing different C/EBP isoforms, by retroviral infection. Overexpression of C/EBP (cid:1) p30 and C/EBP (cid:2) led to increases in C/EBP (cid:3) LAP and C/EBP (cid:3) LIP endogenous protein levels, as determined by electrophoretic mobility shift assays and Western blot. Inhibition of C/EBP activity with dominant negative C/EBPs (C/EBP (cid:3) LIP, 3hF, 4hF) decreased glucocorticoid-, cAMP- and IL-1 responsiveness of the endogenous haptoglobin gene, while overexpression of each C/EBP isoform increased the responsiveness to these regulators. In contrast, dominant negative C/EBPs or C/EBP isoforms did not alter the expression of (cid:1) -acid glycoprotein in response to dexamethasone and of C/EBP (cid:3) and C/EBP (cid:2) in response to various regulators as assessed by Northern blot. These data show that the three C/EBP isoforms are involved in the regulation of haptoglobin and that C/EBP (cid:3) , C/EBP (cid:2) , and (cid:1) -acid glycoprotein expression are not induced by C/EBP isoforms in contrast to other cell types. C/EBP (cid:3) LAP-expressing cells showed an inhibition of cell growth characterized by a delay in p27 Kip1 decrease in response to serum and a decrease in cyclin D isoforms and cyclin E protein levels. Finally, C/EBP isoforms interact with the E2F4 transcription factor. Thus, specific C/EBP isoforms are involved in the differential expression of acute phase protein genes in response to hormones and cytokines. Furthermore, C/EBP isoforms may play a role in the control of cell cycle progression.

Intestinal epithelial cells form a critical mucosal barrier between the host's internal milieu and the external environment and serve as an integral component of the mucosal immune system (1). The inflammatory response in the gastrointestinal tract is mediated by the concerted actions of cellular and humoral elements, including cytokines, reactive oxygen metabolites, and bacterial products (2). An increasing body of evidence has demonstrated that intestinal epithelial cells can actively participate in the generation of an immune response (3). In fact, intestinal epithelial cells participate in intestinal homeostasis and mucosal immunity by secreting and responding to cytokines (4).
The intestinal inflammatory response leads to the establishment of an acute phase response characterized by the local production of acute phase proteins (APPs) 1 in response to cytokines and hormones (5). For example, we have shown that the APP gene haptoglobin is induced in various rat models of intestinal inflammation (6) and in the rat intestinal epithelial cell line IEC-6 in response to cAMP and glucocorticoids (7,8). The CCAAT/enhancer-binding protein family of transcription factors is one class of transactivators involved in the regulation of APP expression in various tissues, including the intestine (9). These genes belong to a class of transcription factors with adjacent leucine zipper dimerization and basic DNA-binding domains that enable the various C/EBP isoforms (␣, ␤, ␦, ␥, ⑀, gadd153/CHOP) to form homo-or heterodimers (10). C/EBP isoforms show differential expression during adipocyte differentiation and liver regeneration, and during the acute phase response in multiple tissues, including the intestine (10). C/EBP␣ plays a key role in adipocyte differentiation (11) and is a regulator of homeostasis in mammals (12). C/EBP␣ inhibits cell proliferation in many cell lines (13) and may control the acute phase response in liver (14). In addition to a role in the early processes of adipocyte differentiation, C/EBP␤ and C/EBP␦ are direct mediators of the acute phase response in hepatocytes (9) and intestinal epithelial cells (8,15,16). Furthermore, C/EBP␤ is involved in the control of cell proliferation in liver (17). Lastly, transient transfection studies have shown that C/EBP isoforms control the levels of C/EBP␤ (18) and C/EBP␦ (19) mRNAs in many cell types by autoregulatory mechanisms.
C/EBP␤, and to a larger extent C/EBP␦, are induced in rodent small intestine after systemic inflammation induced by lipopolysaccharides, both at the mRNA (6,20) and the protein level (6,21). Moreover, we have previously shown that C/EBP isoforms are involved in the regulation of APP genes in intestinal epithelial cells (8,15,16). However, while the role of specific C/EBP isoforms, namely C/EBP␣, C/EBP␤, and C/EBP␦ as well as their differently translated isoforms (C/ EBP␣ p30, C/EBP␤ LIP) (12) is being unraveled in other tissues, their exact role in the control of intestinal APP gene expression or intestinal epithelial cell proliferation is not known. The goal of the present study was to increase the expression of specific C/EBP isoforms or to inhibit C/EBP activity in the rat intestinal epithelial cell line IEC-6 and to assess the phenotype of these cell lines according to the regulation of the acute phase response as well as of cell proliferation.

EXPERIMENTAL PROCEDURES
Animals and Dextran Sulfate Treatment-Adult rats from the Sprague-Dawley strain (Charles River, Saint-Constant, QC) were housed in a low-pathogen-breeding facility. Acute colitis was produced by adding 5% (w/v) dextran sulfate (M r 34,000 -45,000) (22) to the drinking water for 2 days before death. The control group was given normal drinking water. The animals were killed, and different portions of the small intestine (proximal, medial, distal) and colon (proximal, distal) were dissected, washed, and frozen in liquid nitrogen before RNA extraction. Animal experiments were approved by the institutional animal care committee.
Cell Culture-The rat intestinal epithelial cell line IEC-6 was provided by A. Quaroni (Cornell University, Ithaca, NY) (23). This cell line has features of undifferentiated small intestinal crypt cells and was derived from 18-to 24-day-old rat small intestine (24). Cells were grown in Dulbecco's modified Eagle medium (DMEM) containing 5% fetal bovine serum (FBS). Cells at 80% confluence were stimulated for 24 h without or with forskolin (1 M) (Calbiochem, La Jolla, CA), human recombinant IL-1␤ (10 ng/ml) (R & D Systems, Minneapolis, MN) or 10 Ϫ6 M dexamethasone (Sigma) as described before (7,15,16). To determine the status of p42/p44 MAP kinase phosphorylation, cyclin D1/2, cyclin E, and p27 kip1 -expression IEC-6 cells were growth-arrested for 48 h in DMEM without serum before stimulation with DMEM containing 5% FBS. The human enterocyte-like cell line Caco-2-15 used as a model to study intestinal cell differentiation and function was obtained from J.-F. Beaulieu (Department of Anatomy and Cell Biology, University of Sherbrooke). Caco-2 and 293T cells (obtained from A. Nepveu, McGill University) were grown in DMEM supplemented with 10% FBS.
Retroviral Constructs and Infection-The coding sequences for C/EBP␣ p42 and p30 (25), C/EBP␤ LAP, the dominant negative transrepressor C/EBP␤ LIP (26), and C/EBP␦ (27) were cloned in the EcoRI site of the retroviral vector pBabepuro (28). The dominant negative 3hF and 4hF sequences (29,30), which inhibit C/EBP DNA binding activity (29,30), were polymerase chain reaction-amplified and subcloned in the same retroviral vector. 293T cells were transfected overnight by lipofection (LipofectAMINE 2000, Life Technologies, Inc.) with 6 g of the retroviral plasmid vectors and 6 g of helper amphotropic DNA (obtained from A. Nepveu, McGill University). Two days after transfection, supernatants were filtered and stored frozen until use. IEC-6 cells at 40% confluence in 60-mm dishes were infected for 24 h with 2 ml of virus-containing supernatant in the presence of 8 g/ml Polybrene (Sigma). One day after the infection, the cells were split in selection medium containing 2 g/ml puromycin (Sigma). Empty, C/EBP␣ p30-, C/EBP␤ LAP-, C/EBP␤ LIP-, and C/EBP␦ IEC-6-infected clones or cell populations, and 3hF or 4hF cell populations were selected after 3 weeks.
Western Analysis-Nuclei from different infected cell lines were solubilized in sample buffer (62.5 mM Tris-HCl, pH 6.9, 2% SDS, 1% ␤-mercaptoethanol, 10% glycerol and 0.04% bromphenol blue) and sonicated on ice. To determine the expression of various regulators of cell cycle progression, total cellular extracts were isolated from empty or C/EBP␤ LAP1 vector-infected IEC-6 cells treated without serum or after serum induction. Protein concentrations were measured by the Bradford method (Bio-Rad Protein Assay kit, Bio-Rad Laboratories, Mississauga, ON). Proteins (50 g) were resolved on a 10% SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride membrane (Roche Molecular Biochemicals, Laval, QC). Incubations with affinity-purified rabbit polyclonal antibodies specific to C/EBP isoforms (Santa Cruz Biotechnology) were performed overnight at room temper-ature as described previously (33,34). Incubations with rabbit polyclonal antibodies against cyclin D1/2, cyclin E, p27 Kip1 (Santa Cruz Biotechnology), and the p42/p44 MAP kinases (New England Biolabs, Mississauga, ON) were performed as described (35,36). The immune complexes were detected with the Super Signal West Pico Substrate system (Pierce). Protein loading was verified by Coomassie Blue staining.
Immunoprecipitation-Electrophoretic Mobility Shift Assays-Coding sequences for C/EBP␣ p30 and C/EBP␤ LAP were polymerase chain reaction-amplified and cloned in frame in the green fluorescent protein expression vector EGFP (CLONTECH, Palo Alto, CA). Immunoprecipitation-electrophoretic mobility shift assays (IP-EMSA) were done as described previously (39). Briefly, lysates from stably infected IEC-6 C/EBP␣ p30 and C/EBP␤ LIP cell populations from proliferative Caco-2-15 cells or from transiently transfected 293T GFP-C/EBP␣ p30 and GFP-C/EBP␤ LAP-expressing cells were prepared in a buffer containing 50 mM HEPES (pH 7.4), 0.1% Nonidet P-40, 250 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 50 mM sodium fluoride, 20 mM ␤-glycerophosphate, 200 M sodium orthovanadate, and 5% glycerol. Total proteins were incubated overnight at 4°C with 1.5 g of primary antibodies, followed by incubation for 3 h with protein A agarose (Roche Diagnostics, Laval, QC) or with a GFP antibody-agarose conjugate (Santa Cruz Biotechnology) at 4°C. Protein-agarose complexes were washed three times in gelshift buffer, treated with 32 l of 0.8% deoxycholate for 15 min at 4°C and neutralized with 1% Nonidet P-40. 7 l of the supernatants were then used for electrophoretic mobility or supershift assays as described before.

RESULTS
Induction of Haptoglobin after Inflammation-We have previously shown that the acute phase protein gene haptoglobin was expressed and regulated in IEC-6 cells (7,8,16). We wanted to determine whether haptoglobin was induced in an in vivo model of intestinal inflammation. Dextran sulfate in drinking water was administered for 2 days to Sprague-Dawley rats. Dextran sulfate induces a colitis resembling human ulcerative colitis (22). As shown in Fig. 1, haptoglobin mRNAs were readily induced by a 2-day treatment with dextran sulfate in the distal small intestine and the proximal and distal colon as well.
Effect of C/EBP Isoform Expression on C/EBP Endogenous Protein Levels-To determine the role of specific C/EBP isoforms in intestinal epithelial cells, we isolated different cell clones expressing various C/EBP isoforms after retroviral infection and puromycin selection. 10 clones were isolated for each infection. Nuclear extracts were prepared and analyzed by electrophoretic mobility shift assays with a haptoA C/EBP DNA-binding site from the haptoglobin promoter (7,8). Representative results shown in Fig. 2A indicated an increase in DNA binding activity with nuclear extracts obtained from C/EBP␣ p30-, C/EBP␤ LAP-, C/EBP␤ LIP-, and C/EBP␦-expressing clones. We then determined the nature of C/EBP isoforms in these complexes by supershift assays. In the representative experiment shown in Fig. 2B, C/EBP␣ p30, C/EBP␤, and C/EBP␦ isoforms were increased in C/EBP␣ p30, C/EBP␤ LAP and LIP, and C/EBP␦ nuclear extracts, respectively when compared with empty vector nuclear extracts (puro1). The presence of these isoforms was detected both by a supershift and a decrease in the amount of the retarded C/EBP complex. Interestingly, endogenous C/EBP␤ isoforms increased in C/EBP␣ p30 and C/EBP␦ nuclear extracts. Western blot analysis confirmed this increase in both C/EBP␤ LAP and LIP endogenous protein levels in C/EBP␣ p30 and C/EBP␦ nuclear extracts (Fig. 3).
Effect of C/EBP Isoform Expression on the Regulation of Acute Phase Proteins and C/EBPs-We next assessed the effect of C/EBP isoforms on the expression of acute phase protein genes and C/EBPs and verified the degree of responsiveness of these genes to known regulators, namely glucocorticoids (dexamethasone), cAMP (forskolin), and IL-1␤. Our results show the importance of C/EBP isoform expression in the regulation of haptoglobin. Firstly, overexpression of C/EBP␣ p30, C/EBP␤ LAP, and C/EBP␦ increased the basal haptoglobin mRNA lev-els, as determined by Northern blot (Fig. 4, control). Secondly, the responsiveness of haptoglobin to dexamethasone, forskolin, and IL-1␤ was augmented in these C/EBP isoform overexpressing clones. Thirdly, overexpression of the dominant negative C/EBP␤ LIP inhibited the induction of haptoglobin (Fig. 4). This dominant negative C/EBP␤ isoform dimerizes with other C/EBP isoforms and binds DNA but is devoid of a transactivation domain (40). We also verified the expression of ␣-acid  4. Pattern of expression of haptoglobin, ␣-acid glycoprotein, and C/EBP isoforms in C/EBP overexpressing clones. Total RNAs were isolated from empty vector (puro1), C/EBP␣ p30, C/EBP␤ LAP, C/EBP␤ LIP, and C/EBP␦ selected clones treated for 24 h without (control) or with 10 Ϫ6 M dexamethasone, 1 M forskolin, or 10 ng/ml human recombinant IL-1. Equal amounts of RNA (30 g) were electrophoresed and analyzed sequentially by Northern blot with 32 P-labeled haptoglobin, C/EBP␣, C/EBP␤, ␣-acid glycoprotein, and ␣-tubulin probes.

FIG. 2. Composition of C/EBP DNA-binding complexes in IEC-6 cell lines overexpressing C/EBP isoforms.
A, nuclear extracts from representative empty vector (puro), C/EBP␣ p30, C/EBP␤ LAP, C/EBP␤ LIP, and C/EBP␦ selected clones were prepared and mixed with the haptoA C/EBP DNA-binding site-labeled probe. DNA-protein complexes were separated from the free probe on a native polyacrylamide gel for electrophoretic mobility shift assays. B, supershift analysis. Nuclear extracts were incubated with the haptoA DNA binding labeled probe without (Ϫ) or with antibodies against C/EBP␣, C/EBP␤, and C/EBP␦ to determine the composition of the C/EBP complex. Incubation with the HNF1 antibody was used as a specificity control.
glycoprotein. Basal levels of ␣-acid glycoprotein were not increased by C/EBP isoform overexpression. Furthermore, the responsiveness of ␣-acid glycoprotein to dexamethasone was not affected in C/EBP isoform-overexpressing clones, including the dominant negative C/EBP␤ LIP clones. This suggests that, in contrast to other cell types, ␣-acid glycoprotein expression is not regulated by C/EBP isoforms in intestinal epithelial cells.
It has been previously suggested that C/EBPs are involved in their own regulation by an autoregulatory mechanism. This has been proposed for C/EBP␤ and C/EBP␦ (18,19). We thus verified their expression levels in the above various clones. The data show that overexpression of the various C/EBP isoforms or of the dominant negative C/EBP␤ LIP did not alter mRNA levels of C/EBP␤ and C/EBP␦ in either non-induced or induced intestinal epithelial cells (Fig. 4).
To confirm the dominant effect of C/EBP␤ LIP, we established IEC-6 cell populations expressing the 3hF and 4hF dominant negative constructs by retroviral infection. These dominant negative mutants inhibit the DNA binding activity and the function of C/EBP isoforms (29). C/EBP DNA binding to the haptoA site was decreased by 3hF and 4hF overexpression in non-induced cells, as well as in dexamethasone-, IL-1-and forskolin-treated cells, as shown by electrophoretic mobility shift assays (Fig. 5A). The ability of the three C/EBP isoforms to bind to haptoA was reduced, as shown by supershift assays (Fig. 5B). While the dominant negative constructs decreased the responsiveness of the haptoglobin gene to dexamethasone, IL-1, and forskolin (Fig. 6), expression of C/EBP␤, C/EBP␦, and ␣-acid glycoprotein was not altered. This confirms that in intestinal epithelial cells, C/EBP isoforms are involved in the regulation of haptoglobin. However, ␣-acid glycoprotein induction by dexamethasone and C/EBP isoform expression is independent of C/EBP isoforms.
Effect of C/EBP Isoform Expression on Cell Growth-We further verified whether C/EBP expression affected cell proliferation of intestinal epithelial cells. As in other cell types (13), establishment of IEC-6 cell lines after infection with a C/EBP␣ p42-expressing vector was abrogated, suggesting that C/EBP␣ p42 overexpression arrested cell growth (data not shown). Doubling times of various C/EBP␣ p30, C/EBP␤ LIP, and C/EBP␦ clones were not significantly altered, as compared with empty vector-infected clones (Table I). However, doubling times of C/EBP␤ LAP clones did increase.
To get insights on how C/EBP␤ LAP overexpression reduced IEC-6 cell proliferation, we examined the activation of p42/p44 MAP kinases required to pass the G 1 restriction checkpoint in response to mitogens for progression into the cell cycle. As shown in Fig. 7A using a polyclonal antibody specific for the detection of the active phosphorylated p42/p44 isoforms, serum stimulation of p42/p44 MAP kinases in serum-deprived IEC-6 cells was rapid and maximal within 10 min and declined slowly thereafter. No significant differences in p42/p44 MAP kinase activation between the control and the C/EBP␤ LAP-overexpressing clones were observed.
To extend these findings, we performed a detailed comparison between levels of expression of various regulators of cell cycle progression by Western blotting and densitometric analyses. We determined the expression of cyclin D1, cyclin D2, cyclin E, and the cell cycle inhibitor p27 Kip1 as representative events for the early and late G 1 phases respectively (41). Expression of cyclin D isoforms and cyclin E was slightly induced by serum in serum-deprived puro1 and LAP1 clones (Fig. 7B). However, basal levels of these cyclins were significantly attenuated by 25-70% in LAP1 cells as compared with puro1 cells. In contrast, the protein levels of p27 Kip1 , a cell cycle inhibitor accumulating in G 0 /G 1 phase growth-arrested cells, were increased by 40 -75% in LAP1 cells as compared with puro1 cells. Interestingly, the decrease in p27 Kip1 expression observed after  Interaction of C/EBP Isoforms with E2F4 -In the course of our studies, we observed the formation of different complexes binding to a E2F DNA-binding site in C/EBP isoform-overexpressing cell lines. It has been recently proposed that C/EBP␣ inhibits cell growth via direct repression of E2F-mediated transcription (42). We hypothesized that a similar mechanism could be involved in the cell growth delay induced by C/EBP␤ LAP in intestinal epithelial cells. We thus verified by electrophoretic mobility shift and supershift assays the composition of E2F complexes in IEC-6 cell populations overexpressing C/EBP␣ p30 and C/EBP␤ LIP. We chose these populations to achieve the highest level of expression of C/EBP isoforms (Fig. 2, data not shown). Fig. 8A shows that both E2F1 and E2F4 were part of the E2F complexes, as determined by supershift assays with nuclear extracts from puro, C/EBP␣ p30, and C/EBP␤ LIP cell populations. The E2F4 antibody induced a supershift, while the E2F1 antibody inhibited complex formation. Supershift assays also indicated that C/EBP␣ was part of the E2F complex in C/EBP␣ p30-overexpressing cells, while C/EBP␤ was part of the E2F complex in C/EBP␤ LIP-overexpressing cells (Fig. 8B).
We therefore verified whether C/EBP isoforms physically interacted with E2F4, the major E2F transcription factor in proliferative crypt intestinal epithelial cells (43) since it has been shown that C/EBP isoforms do not bind directly to an E2F DNA-binding site (42). For this, we used a IP-EMSA assay (39). Total cell extracts from C/EBP␣ p30 and C/EBP␤ LIP cell populations were immunoprecipitated with the E2F4 antibody. The complexes were dissociated with deoxycholate and used for supershift assays with a haptoA C/EBP DNA-binding site. The IP-EMSA revealed a specific C/EBP-DNA complex in both cell extracts (Fig. 8C). Addition of a C/EBP␤ antibody attenuated and induced a supershift of this immunoprecipitated C/EBP complex from C/EBP␤ LIP-overexpressing cells. The same re-sult was obtained with the C/EBP␣ antibody with cell extracts from C/EBP␣ p30-overexpressing cells. To confirm the physical interaction between C/EBP isoforms ␣ and ␤ and E2F4, we used the IP-EMSA method, and immunoprecipitated cell extracts from the human intestinal epithelial cell line Caco-2 with the E2F4 antibody. Fig. 9A shows that the E2F4 antibody immunoprecipitated a haptoA C/EBP DNA binding complex. Both C/EBP␣ and C/EBP␤ were part of this complex, as determined by supershift assay. This interaction was specific since C/EBP-containing complexes were not immunoprecipitated with non-related antibodies (data not shown).
The interaction between C/EBP isoforms and E2F4 was also assessed with 293T cell extracts obtained from GFP-C/EBP␣ p30 and GFP-C/EBP␤ LAP transiently transfected cells. The fusion proteins were immunoprecipitated with a GFP antibody, and the haptoA C/EBP and E2F DNA binding capacity of the deoxycholate-eluted proteins was determined by electrophoretic mobility shift assays. The IP-EMSA showed specific complexes binding both to the haptoA C/EBP (Fig. 9B) and the E2F (Fig. 9C) sites in GFPp30 and GFPLAP extracts in contrast to GFP-expressing cell extracts. Supershift assays determined that the GFPp30 E2F complexes contained the C/EBP␣ isoform and E2F4 as well, while GFPLAP E2F complexes contained both C/EBP␤ and E2F4, as determined by the disruption of the complex with the C/EBP␤ antibody and by a supershift with the E2F4 antibody (Fig. 9D). Thus, these data indicate that C/EBP isoforms interact with the E2F4 transcription factor. DISCUSSION The data presented in this study show that C/EBP isoforms are involved in the regulation of the acute phase protein gene  8. E2F and C/EBP DNA binding to an E2F probe in IEC-6 cell populations overexpressing C/EBP isoforms. Nuclear extracts from representative empty vector (puro) and C/EBP␣ p30 and C/EBP␤ LIP cell populations were prepared and incubated with the E2F DNAbinding site-labeled probe without (control) or with antibodies against E2F1, E2F4, and E2F5 (A) or against C/EBP␣ and C/EBP␤ (B). DNAprotein complexes were separated from the free probe on a native polyacrylamide gel for electrophoretic mobility shift assays. C, total protein extracts from IEC-6 C/EBP␣ p30 and C/EBP␤ LIP cell populations were immunoprecipitated with an antibody against E2F4. The supernatants were incubated with the haptoA C/EBP DNA-binding site-labeled probe without (control) or with antibodies against C/EBP␣ or C/EBP␤. haptoglobin. First, inhibition of C/EBP activity by the dominant negative transactivation repressor C/EBP␤ LIP (40) or by dominant negative DNA binding inhibitors (3hF, 4hF) (29) leads to a decrease in haptoglobin expression in response to glucocorticoids, forskolin, and IL-1. This confirms the importance of C/EBP isoforms in the regulation of haptoglobin in intestinal epithelial cells. Second, overexpression of C/EBP␣ p30, C/EBP␤ LAP, and C/EBP␦ independently leads to increased basal levels of haptoglobin. Likewise, the level of haptoglobin mRNA induction in response to glucocorticoids, forskolin, and IL-1 is increased when C/EBP isoforms are overexpressed. This confirms our previous results showing that the C/EBP␤ and C/EBP␦ isoforms are the most potent activators of haptoglobin expression (7,8). Our results show that C/EBP␣ plays an important role in the regulation of haptoglobin in intestinal epithelial cells. It has been demonstrated by studying the neonatal acute-phase response to inflammation in C/EBP␣ knock-out mice that C/EBP␣ expression was required for haptoglobin expression in liver as opposed to lung (14). It is however surprising that C/EBP␣ p30 is so effective. Two isoforms of C/EBP␣ arise through the use of different translational initiation sites (25,44). The shorter C/EBP␣ p30, as opposed to the full-length C/EBP␣ p42 protein, has a decreased transactivation potential due to the absence of two of three transactivation domains. It remains to be determined whether the capacity of C/EBP␣ p30 to activate haptoglobin expression depends solely on its ability to increase the endogenous levels of C/EBP␤ in IEC-6 cells when overexpressed.
In contrast, C/EBP isoforms are not involved in the regulation of ␣-acid glycoprotein by dexamethasone in intestinal epithelial cells. Firstly, overexpression of dominant negative C/EBPs, affecting either the transactivation or the DNA binding potential, does not inhibit ␣-acid glycoprotein induction by dexamethasone. Secondly, overexpression of specific C/EBP isoforms does not increase basal ␣-acid glycoprotein mRNA levels. This indicates that the main regulator of ␣-acid glycoprotein expression is the glucocorticoid receptor, in contrast to ␣-acid glycoprotein regulation in rat liver or in other species. Indeed, glucocorticoids act through a minimal steroid responsive unit composed of a glucocorticoid receptor DNA-binding site overlapping C/EBP DNA-binding sites in the proximal ␣-acid glycoprotein promoter (45). In rat liver and in other species, a strong cooperation between the glucocorticoid receptor and C/EBPs is necessary for hormone induction. Our results suggest that this cooperation is not necessary for ␣-acid glycoprotein regulation by glucocorticoids in intestinal epithelial cells, at least in the IEC-6 cell model.
Transient transfection studies have shown that C/EBP isoforms may control the levels of C/EBP␤ (18) and C/EBP␦ (19) mRNAs by autoregulatory mechanisms in other cell types. Furthermore, the C/EBP␣ promoter is autoregulated directly by C/EBP␣ in murine (46) and human cells (47). A dominant negative construct stably expressed in hepatoma cells inhibits the expression of C/EBP␣ (48). In contrast to these studies, overexpression of dominant negative forms of C/EBPs or of specific C/EBP isoforms did not affect C/EBP isoform mRNA levels. Furthermore, C/EBP␣ protein levels were not affected in C/EBP isoform-overexpressing cell lines. Our results suggest that C/EBP isoforms are not subject to autoregulation in IEC-6 intestinal epithelial cells.
Thus, the intriguing increase in C/EBP␤ LAP and C/EBP␤ LIP endogenous protein levels following C/EBP␣ p30 or C/EBP␦ overexpression does not involve transcriptional autoregulatory mechanisms. Nothing is known of the mechanisms involved in dimerization of C/EBP isoforms in the cell. One may envision the possibility that dimers may be more stable than monomers. Increases in one isoform may alter the steady-state levels of dimers by increasing the probability of forming dimers, thereby increasing the stability of the C/EBP isoform pool. On the other hand, C/EBP isoforms may compete with other proteins known to interact with C/EBPs with less affinity. For example, C/EBP␣ interacts with the transcription factor NF-B (49). Increases in C/EBP␣ protein levels may favor the formation of C/EBP dimers as opposed to C/EBP␣-NF-B dimers, thus increasing overall C/EBP DNA binding activity.
We have observed that C/EBP␣ p42 overexpression leads to growth arrest in IEC-6 cells (data not shown), as in other cell types (13). C/EBP␣-dependent cell growth arrest depends on the transcriptional regulation of the cyclin-dependent kinase inhibitor p21 (50) or its stabilization at the protein level (51) in different cell lines. More recently, C/EBP␣ has been shown to inhibit cell growth by direct repression of E2F-mediated transcription (42) and may disrupt E2F-p107 complexes (52). The mechanisms involved in C/EBP␣-dependent growth arrest in IEC-6 cells do not implicate a transcriptional regulation of p21 (data not shown). Other C/EBP isoforms may be required to control intestinal epithelial cell growth. For example, overexpression of the gadd153/CHOP C/EBP isoform inhibits the growth of many cell types (10), including IEC-6 cells (data not shown). In addition, we have shown that C/EBP␤ LAP overexpression reduced IEC-6 cell growth, in contrast to C/EBP␣ p30, C/EBP␤ LIP, and C/EBP␦. C/EBP␤ LAP-expressing cells are FIG. 9. Interaction of E2F4 with C/EBP isoforms. A, total protein extracts from Caco-2 cells were immunoprecipitated with antibodies against E2F4. The supernatants were incubated with the haptoA C/EBP DNA-binding site-labeled probe without (control) or with antibodies against C/EBP␣ or C/EBP␤. DNA-protein complexes were separated from the free probe on a native polyacrylamide gel for electrophoretic mobility shift assays. B, total protein extracts obtained from transiently transfected 293T cells expressing GFP, GFP-C/EBP␣ p30, or GFP-C/EBP␤ LAP fusion proteins were immunoprecipitated with a GFP antibody. The supernatants were incubated with the haptoA C/EBP (B) or the E2F (C) DNA-binding site-labeled probes. D, the immunoprecipitates obtained with the GFP antibody in B and C were incubated with the E2F DNA-binding-site-labeled probe without (control) or with antibodies against C/EBP␣, C/EBP␤, or E2F4. characterized by a delayed down-regulation of p27 Kip1 and a delayed pRb hyperphosphorylation (data not shown) in response to serum and by decreased levels of cyclin D1/2 and cyclin E. Levels of the cyclin-dependent kinase inhibitors p21 Cip and p57 INK4B are not affected. C/EBP␤ has been shown to inhibit hepatoma cell proliferation (53). In contrast, C/EBP␤ is required for proliferation induced by TGF␣ in primary hepatocytes (54). Our data suggest that C/EBP␤ LAP overexpression leads to a delay in G 1 progression of intestinal epithelial cells.
The exact mechanism involved in the C/EBP␤-mediated growth delay remains to be determined. However, our results show that C/EBP isoforms may physically interact with the E2F4 transcription factor, leading to the recruitment of C/EBP isoforms to E2F DNA-binding sites. Firstly, C/EBP isoforms ␣ p30 and ␤ LIP are immunoprecipitated by the E2F4 antibody in IEC-6 cells overexpressing these isoforms, as determined by IP-EMSA. Secondly, the E2F4 antibody immunoprecipitates both C/EBP␣ and C/EBP␤ isoforms from human intestinal epithelial Caco-2 cells. Thirdly, immunoprecipitation of GFPC/ EBP␣ p30 and GFPC/EBP␤ LAP fusion proteins from 293Toverexpressing cells leads to the recovery of the E2F4 transcription factor, as determined by a supershift with a E2F DNA-binding site. These data also suggest that the minimal region involved in the interaction with E2F4 includes the C/EBP leucine zipper and that the transactivation domains of C/EBP␣ and C/EBP␤ are dispensable since C/EBP␤ LIP is devoid of most of the transactivation domains of C/EBP␤ and since C/EBP␣ p30 lacks two of three transactivation domains, including a putative E2F homology region (52). While all C/EBP isoforms bind to E2F4 at least by the C-terminal region, only C/EBP␣ p42 (13,42) and C/EBP␤ LAP (our results) induce growth arrest. One may speculate that the presence of the transactivation domains of C/EBP␣ p42 and C/EBP␤ LAP in E2F complexes negatively alters the transactivation potential of E2F complexes, leading to reduced expression of E2F target genes, such as cyclin D and cyclin E, and decreased cell proliferation. In addition, the presence of these domains may in some way stabilize the interaction of these complexes to the E2F DNA-binding site.
Thus, we have shown the presence of different regulatory mechanisms for C/EBP isoform as well as acute phase protein gene expression in intestinal epithelial cells. Specific C/EBP isoforms are involved in the differential expression of APP genes in response to hormones and cytokines. Furthermore, C/EBP isoforms, including C/EBP␤, may play an important role in the control of intestinal epithelial cell growth.