The Caudal-related Homeodomain Protein Cdx1 Inhibits Proliferation of Intestinal Epithelial Cells by Down-regulation of D-type Cyclins*

Cdx1 is a homeodomain transcription factor that regulates intestine-specific gene expression. Experimental evidence suggests that Cdx1 may be involved in cell cycle regulation, but its role is ill defined and the mechanisms have not been explored. We used stable transfection of inducible constructs and transient expression with a replication-deficient adenovirus to induce Cdx1 expression in rat IEC6 cells, a non-transformed intestinal epithelial cell line that does not express Cdx1 protein. Expression of Cdx1 markedly reduced proliferation of IEC6 cells with accumulation of cells in the G0/G1 phase of the cell cycle. Cell cycle arrest was accompanied by an increase in the hypophosphorylated forms of the retinoblastoma protein (pRb) and the pRb-related p130 protein. Protein levels of multiple cyclin-dependent kinase inhibitors were either unchanged (p16, p18, p21, p27, and p57) or were not detected (p15 and p19). Most significantly, levels of cyclins D1 and D2 were markedly diminished with Cdx1 expression, but not cyclins D3, E, or the G1 kinases. Additionally, cyclin-dependent kinase-4 activity was decreased in association with decreased cyclin D protein. We conclude that Cdx1 regulates intestinal epithelial cell proliferation by inhibiting progression through G0/G1, most likely via modulation of cyclin D1 and D2 protein levels.

Cdx1 is a homeodomain transcription factor that regulates intestine-specific gene expression. Experimental evidence suggests that Cdx1 may be involved in cell cycle regulation, but its role is ill defined and the mechanisms have not been explored. We used stable transfection of inducible constructs and transient expression with a replication-deficient adenovirus to induce Cdx1 expression in rat IEC6 cells, a non-transformed intestinal epithelial cell line that does not express Cdx1 protein. Expression of Cdx1 markedly reduced proliferation of IEC6 cells with accumulation of cells in the G 0 /G 1 phase of the cell cycle. Cell cycle arrest was accompanied by an increase in the hypophosphorylated forms of the retinoblastoma protein (pRb) and the pRb-related p130 protein. Protein levels of multiple cyclin-dependent kinase inhibitors were either unchanged (p16, p18, p21, p27, and p57) or were not detected (p15 and p19). Most significantly, levels of cyclins D1 and D2 were markedly diminished with Cdx1 expression, but not cyclins D3, E, or the G 1 kinases. Additionally, cyclin-dependent kinase-4 activity was decreased in association with decreased cyclin D protein. We conclude that Cdx1 regulates intestinal epithelial cell proliferation by inhibiting progression through G 0 /G 1 , most likely via modulation of cyclin D1 and D2 protein levels.
Homeobox genes encode for transcription factors that have been implicated as critical regulatory proteins for a variety of developmental and differentiation processes in organisms throughout the phylogenetic tree (reviewed in Refs. [1][2][3][4]. Moreover, there is evidence that members of this family of transcriptional proteins are involved in the control of cellular proliferation and oncogenesis (5). The Cdx 1 gene family includes mammalian homeobox genes that are related to the Drosophila melanogaster gene caudal (6). Three mouse Cdx genes have been identified, Cdx1 (7), Cdx2 (8 -10), and Cdx4 (11); two genes have been identified thus far in humans, CDX1 (12) and CDX2 (13). Cdx1 and Cdx2 are expressed in many tissues early in development, but are restricted to the intestinal and colonic epithelium later in fetal life and in the adult (7, 11, 14 -17).
Several lines of evidence suggest that Cdx1 and Cdx2 are important nuclear proteins involved in the regulation of the phenotype of intestinal epithelial cells (18 -22). Cdx1 (23) and Cdx2 (10,(23)(24)(25)(26)(27)(28) interact with DNA elements that have an AT-rich consensus sequence (TTTAT), similar to other homeodomain proteins. Binding of Cdx1 (23) and Cdx2 (10 ,23) to intestinal gene promoters leads to transcriptional activation that is mediated through an activation domain localized to the amino-terminal region of the protein (29,30). Most studies have shown that Cdx2 activates transcription of various genes (10,23,27,28,31), whereas fewer studies have examined Cdx1 as an activator of gene transcription (17,23). The role of these homeobox genes in regulation of proliferation is of particular interest, given the tightly controlled patterns of proliferation in the intestinal epithelium. Cdx2 is able to inhibit of cellular growth as shown by studies in transfected cell lines (18), but the effect of Cdx1 on cellular proliferation is not as well defined. The genes that are known to be regulated by either Cdx1 or Cdx2 are generally ones that define functional phenotype (for example, sucrase-isomaltase (Ref. 10), glucagon (Refs. 27 and 28), intestinal phospholipaseA/lysophospholipase (Ref. 23), and carbonic anhydrase (Ref. 32)) and are not known regulators of cellular proliferation.
Several studies have examined the role of Cdx1 on proliferation in intestinal cell lines. Cdx1 was shown to transform NIH-3T3 cells in a focus-forming assay in a study that tested many homeobox genes for transforming properties (5). In another study, a colon cancer cell line that expresses small amounts of endogenous CDX1 and CDX2 was transfected with CDX1 sense and antisense expression constructs (19). These investigators found that overexpression of CDX1 with the sense construct had no effect on growth, but that the CDX1 antisense-expressing cells had a significant reduction in their growth rate. Based on these findings, the authors hypothesized that CDX1 acted as a growth promoter, although CDX protein levels were not examined to determine whether there were indeed changes in CDX1 levels. Recently, another human colon cancer cell line, HT29, was transfected with CDX1, CDX2, or both CDX1 and CDX2 expression constructs (21). HT29 clones expressing CDX2 or both CDX1 and CDX2 had significantly reduced growth rates compared with controls. CDX1 expression alone had no effect on the growth of these cells, but neither CDX1 nor CDX2 protein levels were measured in this study.
Significantly, only CDX1 and CDX2 expressed together could reduce the tumor growth rate of the HT29 clones in nude mice. Two important weaknesses of these studies include the use of cells that constitutively express CDX genes and that levels of the expressed proteins were not measured. This experimental design might lead to a selection of cells that were able to grow despite CDX gene expression, or for cells that expressed very low levels of CDX1.
The expression of Cdx genes has been investigated in human colorectal tumors. Expression of CDX2 mRNA (33) and protein (13) is decreased in a large percentage of colorectal cancers. Expression of CDX1 mRNA (33) and protein (34) is also decreased in the majority of colorectal cancers. In studies using immunohistochemistry to detect CDX1, there appears to be a graded decrease in immunoreactivity between normal, adenomatous, and cancer cells (34). From these data, there appears to be a marked decrease in the expression of both CDX1 and CDX2 in colorectal cancers.
Therefore, it is currently unclear how Cdx1 is involved in the control of cellular proliferation. The purpose of our study was to examine the effects of Cdx1 on growth regulation of a nontransformed intestinal cell line. We found that Cdx1 inhibits cellular proliferation by blocking progression through the cell cycle in G 1 . This G 1 arrest was accompanied by decreased expression of the D-cyclins, inhibition of cyclin-dependent kinase activity, and accumulation of the hypophosphorylated forms of the retinoblastoma protein (pRb) 2 and the pRb-related p130. These findings have important implications for understanding the functional role of Cdx1 in the regulation of intestinal epithelial cell proliferation.

Construction of Expression
Vectors-An inducible Cdx1 expression vector, pMT-Cdx1, was constructed by subcloning the coding sequence of Cdx1, released by digestion of pEVRF1Cdx1 (16) with KpnI and XbaI, into pMTCB6 ϩ , which contains the promoter of the sheep metallothionein I gene (35).
Production and Characterization of Adenovirus Vectors-Mouse Cdx1 cDNA was subcloned into the adenoviral transfer vector, pAdC-MVlink, and used to produce the expression vector Adeno-Cdx1. pAd-CMVlink without a cDNA insert was used to generate the control Adeno-NULL virus. The linearized adenoviral transfer plasmid was cotransfected with adenoviral DNA into 293 cells to allow for homologous recombination. Viral plaques were then picked and expanded on 293 cells. Recombinent virus was expanded, purified by CsCl centrifugation, and the infectious titers determined by an adenoviral plaque assay.
Multiplicity of infection (m.o.i.) required for expression of Cdx1 in Ͼ90% of cell was determined for IEC6 cells subcultured at a density of 1 ϫ 10 4 cells/well in 24-well culture dishes. Cells were infected in 0.2 ml of infection medium (culture medium with only 2% fetal bovine serum) containing various dilutions of Adeno-Cdx1 to yield m.o.i. between 25 and 1000. Twenty-four hours after infection, the medium was changed to growth medium, and 24 h later cells were fixed and assayed for Cdx1 expression by immunohistochemistry using anti-Cdx1 antibody (1:200) (34) and a rhodamine-conjugated anti-rabbit secondary antibody. The optimal m.o.i. for expression of Cdx1 in Ͼ90% of cells was 50 for IEC6 cells.
Cell Culture, Transfection, and Infection-IEC6 cells were maintained as described previously (18) and transfected with pMT-Cdx1 by electroporation at 250 V and 960 microfarads with a Gene Pulser (Bio-Rad). NIH3T3 cells were grown in Dulbecco's modified Eagle's medium with 4.5 g/liter glucose and transfected by CaPO 4 precipitation, as described previously (10). Briefly, plasmids were incubated for 20 min in a solution containing 2ϫ BBS (50 mM BES, pH 6.95, 280 mM NaCl, and 1.5 M Na 2 HPO 4 ) and 0.25 M CaCl 2 and then added to NIH3T3 cells dropwise. The plates were incubated in 3% CO 2 for 24 h. The cells were washed with PBS, split 10:1, and incubated in 5% CO 2 incubator. Clones were selected in the presence of 0.6 mg of G418/ml, isolated, and screened for Cdx1 or Cdx2 expression by electrophoretic mobility shift assay (EMSA) and Northern blot analysis.
Adenovirus infections were performed in 24-well culture plates (cell count and BrdUrd assay) or 100-mm culture dishes (Western and Northern analysis). Twelve to 18 h before infection, exponentially growing IEC6 cells were subcultured at a density of 1 ϫ 10 4 cells/well or 1.4 ϫ 10 5 cells/100-mm culture dish. Cells were infected the following morning in 0.2 or 5.7 ml of infection medium (culture medium with only 2% fetal bovine serum) containing no virus (control) or enough Adeno-Cdx1 or Adeno-Null virus to yield an m.o.i. ϭ 50. Twenty-four hours after infection, the medium was changed to growth medium. At 48 h, the cells were processed for the various end points. For cell counts, the cells were trypsinized at several time points after infection, pelleted by centrifugation, and resuspended in PBS. Duplicate counts were performed for each well on a hemocytometer (Hausser Scientific).
EMSA-Nuclear proteins were isolated from cell cultures and EMSA was performed exactly as described previously (10).
Soft Agar and Focus Assay-10 4 or 5 ϫ 10 4 cells were suspended in 0.3% agar (Difco) solution in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and overlaid onto 0.5% agar solution in Dulbecco's modified Eagle's medium containing 10% fetal calf serum in 60-mm plates. Regular medium were added after top agar was solidified. For induction, ZnSO 4 were added to bottom, top agar, and top medium to final concentration of 50 M. Colonies were stained 2-4 weeks after plating with 50 g/ml p-iodanitrotetrazolium violet (Sigma) and scored 24 h later.
For the focus assay, 7 ϫ 10 3 cells from each stable lines (3T3MT1, 3T3Cdx1MT1, and 3T3Cdx2MT17) were mixed with 9.3 ϫ 10 4 NIH3T3 cells and plated in six-well plates. The culture medium was changed twice weekly for 4 weeks. At 2 and 4 weeks, cells were fixed with Bouin's solution for 1 h, rinsed with PBS, and stained with 4% Giemsa reagents for 24 h. Dishes were rinsed briefly with distilled water and then scored for the number of foci.
Cell Proliferation and Flow Cytometry Analysis-BrdUrd incorporation was measured using the 5-bromo-2Ј-deoxyuridine labeling and detection kit I (Roche Molecular Biochemicals). BrdUrd labeling of IEC6 cells was performed over 1 h, and then the cells were processed according to the instructions provided with the kit, except that in the final washing step, Hoechst dye (bisbenzimide; Sigma) was added to the wash reagent (final concentration ϭ 2 g/ml) to stain all nuclei. At least 200 cells from each well were counted using a fluorescent microscope, and those incorporating the BrdUrd label were expressed as a percentage of all cells counted. These percentages were transformed using the arcsine transformation prior to statistical analysis with analysis of variance and Tukey tests.
For flow cytometry, IEC6 cells were infected and cultured as described. DNA content was measured using a rapid propidium iodide staining method. Approximately 250,000 cells (one 100-mm dish) were trypsinized, washed once in PBS, and then resuspended in 0.35 ml of STAIN solution (3% polyethyleneglycol 8000, 50 g/ml propidium iodide, 360 units/ml RNase A, 0.1% Triton X-100, 3.6 mM citric acid buffer, pH 7.2). The cells were incubated for 30 min at 37°C, then 0.35 ml of SALT solution (3% polyethyleneglycol 8000, 50 g/ml propidium iodide, 0.1% Triton X-100, 0.4 M NaCl, pH 7.2) was added, and the solution incubated in the dark at 4°C for at least 1 h prior to flow cytometry and DNA quantitation using a FACScan (Becton Dickson).
Northern Blot and Ribonuclease Protection Analysis-Total RNA was isolated by the guanidinium thiocyanate/CsCl method from several 100-mm plates infected and cultured as described (18). 50 g of RNA was run on a 0.8% agarose and 6% formaldehyde gel with 1ϫ MOPS buffer. The RNA was transferred to a nitrocellulose membrane, the filter was air-dried, and the RNA was UV cross-linked to the membrane using a UV Stratalinker 1800 (Stratagene). The RNA was stained with methylene blue after blotting to evaluate for equal transfer.
Filters were prehybridized for several hours at 42°C in buffer containing 50% formamide, 5ϫ SSPE, 5ϫ Denhardt's solution, 100 g/ml sonicated salmon sperm DNA, and 0.1% SDS. A human cyclin D1 cDNA (kindly provided by Anil Rustgi, University of Pennsylvania) was restriction enzyme-digested and gel-purified. A random-primed, 32 P-labeled probe was prepared from 30 ng of the purified cDNA using the Random Primed DNA labeling kit (Roche Molecular Biochemicals) according to manufacturer's instructions. The probe was purified from unincorporated nucleotides using a G-50 Quick Spin Column (Roche Molecular Biochemicals). The membrane was hybridized with 1 ϫ 10 6 cpm of purified probe/ml of hybridization solution at 42°C for 24 h. The blot was the washed twice (10 min each, room temperature) in 250 ml of 2ϫ SSC and 0.1% SDS, with gentle agitation. The filter was then washed for 1 h at 55°C in 250 ml of 0.2ϫ SSC and 0.1% SDS, with the buffer changed after 30 min. The membrane was briefly air-dried and exposed to Kodak XAR film with an intensifying screen at Ϫ70°C for 48 h.
To reprobe blots for 36B4, the membrane was first stripped of the previous probe by washing in 1ϫ Blot Wash Buffer at 60°C for 1 h. Blot was and exposed to Kodak XAR film with an intensifying screen at Ϫ70°C for 48 h to confirm stripping of the previous probe, and then prehybridized and hybridized as before. The 36B4 probe used as a control is the mouse homologue of the human acidic ribosomal phosphoprotein P0 (GenBank accession no. M17885). It was generated by cloning and sequencing the RT-PCR products using the primers 5Јatgtgaagtcactgtgccag-3Ј and 5Ј-gtgtaatccgtctccacaga-3Ј.
Cyclin-dependent Kinase Activity Analysis-Several 100-mm plates of IEC6 were infected and cultured as described. At 48 h, they were briefly trypsinized, washed twice in PBS, then resuspended in IP buffer (50 mM Hepes (pH 7.5), 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 10% glycerol, 0.1% Tween 20, 1 mM NaF, 0.1 mM Na 3 VO 4 , 1 mM PMSF, 1 mM dithiothreitol, 25 g/ml leupeptin, 25 g/ml aprotinin, 150 g/ml benzamidine, and 10 g/ml trypsin inhibitor). Cells were briefly sonicated, then incubated in IP buffer at 4°C with rotation for 30 min. Cellular debris was pelleted by centrifugation at 14,000 rpm for 10 min at 4°C. The supernatant was saved and was further cleared by repeat centrifugation as before. The protein concentration of the supernatant was determined by the BCA protein assay (Pierce). Duplicate 100 g samples of protein extract was used in each kinase reaction. The lysates were first precleared with prewashed Protein G beads (Life Technologies, Inc.), and then with prewashed Protein A beads (Life Technologies, Inc.). These cleared lysates were then subject to immunoprecipitation with 1 g of either anti-Cdk4 (sc-260, Santa Cruz) or 1 g of rabbit pre-immune IgG (Santa Cruz) for 2 h at 4°C with rotation. Prewashed Protein A beads were added, and the incubation continued for an additional 1 h. The IP products were pelleted by centrifugation at 4000 rpm for 5 min at 4°C, and washed twice with IP buffer, then three times with kinase buffer (50 mM Hepes (pH 7.5), 10 mM MgCl 2 , 5 mM MnCl 2 , 1 mM NaF, 0.1 mM Na 3 VO 4 , 1 mM PMSF, 1 mM dithiothreitol, 1 g/ml leupeptin, 0.2 g/ml aprotinin). To initiate the reaction, 30 l of kinase buffer with 2.5 mM EGTA, 50 M cold ATP, 0.2 mg/ml BSA, 10 Ci of [␥-32 P]ATP, and 1 g of GST-Rb was then added to the pelleted beads, and the reaction incubated at 30°C for 30 min with frequent agitation. The reaction was halted with 2ϫ Laemmli buffer, and the products boiled 5 min and analyzed SDS-PAGE electrophoresis and autoradiography. The GST-Rb bacterial expression plasmid was a kind gift of Dr. Greg Enders (University of Pennsylvania). The fusion protein was expressed and purified as described (36).

Inhibition of Intestinal Cell Proliferation by Cdx1 in Stably
Transfected IEC6 Cells-IEC6 cells are non-transformed intestinal epithelial cells that were originally isolated from neonatal rat small intestine (37). They have some characteristics of intestinal crypt cells, but are morphologically undifferentiated and do not express genes that are associated with a differentiated phenotype (37). Since they lack expression of both Cdx1 and Cdx2, we have used IEC6 cells to evaluate the functional role of Cdx genes in intestinal epithelial differentiation (18). The pattern of IEC6 cell accumulation in tissue culture was markedly affected by the expression of Cdx2 (18). For 2-3 days following induced expression of Cdx2 in stably transfected IEC6 cells, there was a marked decrease in the accumulation of cells in culture (18). This period was followed by exponential growth of cells, reaching confluence at a higher density than cells that were not induced to express high levels of Cdx2 (18).
To determine the effect of Cdx1 on IEC6 cell proliferation, stable cell lines were made using an inducible expression construct for Cdx1. Immunoblots for Cdx1 proteins showed that stable cell lines containing the Cdx1 expression vector expressed low constitutive amounts of protein, which was induced following addition of zinc sulfate to the culture medium (Fig. 1A). The expressed Cdx1 protein was competent to bind to a known DNA binding element, the SIF1 element in the sucrase-isomaltase gene promoter (10) (Fig. 1B). Since the SIF1 element has been shown to have two adjacent Cdx binding sites (23), the gel shift has two specific complexes labeled A and B, which represent the monomer and dimer complexes, respectively. The cell lines transfected with the Cdx1 construct grew more slowly than cell lines transfected with the vector alone (Fig. 2). Induction of Cdx1 resulted in slower accumulation of cells in culture when compared with uninduced cells. However, the effect of induction was not great since the rate of growth of the uninduced cells was already diminished. It is possible that the decreased growth of uninduced cells is due to the small amount of expressed Cdx1 in the absence of inducing media (Fig. 1A).
Cdx1 Inhibits Proliferation of Non-intestinal Cells-To determine whether the effect of Cdx1 on proliferation is specific for intestinal IEC cells, we examined the effect on growth of the NIH3T3 fibroblast cell line. A stable cell line was isolated that expressed Cdx1 under the control of an inducible promoter (3T3 Cdx1 MT1), as demonstrated by immunoblot (Fig. 1A). The cell lines transfected with Cdx1 expression vector grew more slowly than those transfected with empty vector, growing to 2 ϫ 10 5 cells/100-mm plate at confluence versus 3-4 ϫ 10 5 cells for vector control cells (data not shown). Induction of Cdx1 resulted in an additional decrease in the number of cells at confluence to 1 ϫ 10 5 cells/100-mm plate (data not shown). Thus, the inhibitory effect on cellular growth by Cdx1 is not specific for intestinal cell lines.
There is a report in the literature asserting that expression of Cdx1 in NIH3T3 cells, as well as many other homeobox genes, leads to transformation as determined by in vitro assays (5). Our finding that Cdx1 inhibited growth was contrary to these published data. Thus, the growth characteristics of stable lines of NIH3T3 cells expressing either high or low levels of Cdx1 was examined using a soft agar assay. Comparison of colony formation in cells cultured in either control or induction media demonstrated no increase in the number of colonies as a result of high level expression of Cdx1 (data not shown). Therefore, our results suggest that Cdx1 is unable to transform NIH3T3 cells, consistent with the effects on proliferation.
Inhibition of Intestinal Cell Proliferation by Adenovirus Expressing Cdx1-The cell lines stably transfected with Cdx1 vector consistently grew to a lower density than the cells transfected with the vector alone. This finding suggested that even low level expression of Cdx1 in uninduced transfected cells may inhibit IEC6 cell growth. Therefore, we used an adenoviral vector to express Cdx1 in IEC6 cells. Adenovirus expressing Cdx1 was made as described under "Materials and Methods" and shown to express authentic Cdx1 protein by both Western blot (Fig. 1A) and EMSA (Fig. 1C). Comparison of Cdx1 protein levels on the immunoblot (Fig. 1A) shows that Cdx1 levels were much greater in the adenovirus-infected cells than in the stable lines.
The effects of Adeno-Cdx1 was compared with the same adenovirus without a cDNA insert, Adeno-NULL. The accumulation of IEC6 cells infected with Adeno-Cdx1 was markedly inhibited when compared with uninfected control cells or those infected with Adeno-Null (Fig. 3A). Additionally, there was a marked decrease in the percentage of cells incorporating BrdUrd into DNA (Fig. 3B). Thus, expression of Cdx1 inhibits proliferation of IEC6 cells, as suggested by the experiments with stably transfected cells. (Table I). Of note, the flow cytometry profiles did not show a subdiploid peak of DNA, suggesting that apoptosis did not result from expression of Cdx1 (data not shown). Therefore, Cdx1 expression in IEC6 cells inhibits proliferation by blocking cell cycle progression in G 1 .

Cdx1 Gene Expression Blocks Progression of the Cell Cycle in G 1 -Flow cytometry was used to evaluate cell cycle progression of IEC6 cells. Cells infected with Adeno-Cdx1 showed a marked increase in the percentage of cells in G 1 with a decrease in cells in S phase and G 2 in comparison to uninfected cells and cell infected with Adeno-Null
Cell Cycle Protein Changes Associated with Cdx1 Expression-To explore the mechanism of the G 1 block, we first examined the expression of the retinoblastoma family of proteins, which are key regulators of the G 1 /S transition. In cells expressing Cdx1, there was a shift of pRb to the hypophosphorylated form, there was a marked increase in hypophosphorylated p130, and a decrease in p107 (Fig. 4). These changes in Rb proteins are known to alter transcriptional events that inhibit the transition from G 1 to S phase.
We next examined cell cycle proteins that directly or indirectly affect the phosphorylation of Rb proteins. Cyclins D1 and D2 are both markedly decreased in cells expressing Cdx1 with only a slight decrease in cyclin D3 (Fig. 5). Cdks 4 and 6, which both associate with D cyclins, are only slightly decreased in comparison to cells infected with null virus. Cyclin A and B1 were both decreased, as occurs in any event that blocks the G 1 /S transition. There was no change in cyclin E. Since an important mechanism for inhibition of cyclin-dependent kinase activity is induction of inhibitor proteins, we examined the expression of multiple known cyclin-dependent kinase inhibitors. There was no change in expression of p16, p18, p21, p27, or p57 in cells expressing Cdx1 (Fig. 5). Immunoblots were also performed for p15 and p19, but there was no detectable protein (data not shown).
Cyclin-dependent Kinase Activity Is Decreased by Expression of Cdx1-The decrease in cyclin D1 and D2 protein levels represents a potential mechanism by which phosphorylation of Rb proteins is decreased, resulting in inhibition of the cell FIG. 1. Adeno-CDX1-directed CDX1 protein expression examined by Western blot and EMSA analysis. A, IEC6 and 3T3 Cell lines stably transfected with an inducible Cdx1 expression vector, or vector alone, were cultured with (ϩ) or without (Ϫ) 50 M ZnSO 4 for 48 h, and then whole cell extracts prepared. Additionally, IEC6 cells were sham-infected (Control), infected with Adeno-Null (Adeno-Null), and Adeno-CDX1 (Adeno-CDX1) as described under "Materials and Methods." Whole cell extracts were prepared at 48 h. 20 g of each extract was analyzed on a 10% SDS-PAGE for Cdx1 protein using a polyclonal antibody specific for Cdx1 (CPSP), as described previously (34). A is a composite figure; the 3T3 lanes were run on a separate gel in parallel. B, stably transfected IEC6 cells with the pMT-Cdx1 (IEC6Cdx1MT9) were cultured in the absence or presence of 50 M ZnSO 4 for 24 h. Nuclear proteins were then isolated and analyzed by EMSA to assess for nuclear binding protein to the SIF1 DNA element of SI promoter. The A and B DNA-protein complexes represent the interaction of one and two molecules of Cdx1 on the SIF1 DNA element, respectively. 10 ng of unlabeled SIF1 oligonucleotide was used for competition and anti-Cdx1 antibody was added for supershift analysis. DNA-protein complexes were separated in a native 4% polyacrylamide gel. The arrowhead marks the position of the complex that supershifted with addition of anti-Cdx1 antibody. FP, free probe. C, IEC6 cells were sham-infected (control) or infected with Adeno-Null and Adeno-CDX1 as before. Nuclear extracts were prepared 48 h after infection and 5 g of extract examined by EMSA using the CDX1 binding SIF1 element from the sucrase-isomaltase gene as probe. Lanes 1 and 10, probe only. Specific competition with unlabeled SIF1 (lanes 3, 5, and 7) and supershift using polyclonal antibodies to CDX1 (CPSP, lane 8) and CDX2 (CANL, lanes 9) were also performed. A, Cdx1 monomer binding; B, Cdx1 dimer binding complex. *, SS, supershift complex; NS, nonspecific complex; FP, free probe. cycle. However, the maintained cyclin D3 could potentially support enough cyclin D-dependent kinase activity to allow progression through G 1 and the G 1 /S transition. To test this question, we measured cyclin-dependent kinase 4 (Cdk4) activity in cellular extracts. Cdk4 kinase activity was decreased 60% of full activity in cells expressing Cdx1 (Fig. 6). Since Cdk4 protein is not significantly altered in cells expressing Cdx1 (Fig. 6), the decreased kinase activity is likely due to low levels of D cyclins. These data establish the link between decreased cyclin D protein levels and hypophosphorylation of Rb proteins.
Cdx1 Expression Decreases Cyclin D1 mRNA-The analysis of the expression of cell cycle proteins indicated that the likely cause of G 1 arrest was a decrease in the levels of D cyclins. Therefore, we examined the level of expression of cyclin D1 mRNA following induction of Cdx1. There was a dramatic reduction of cyclin D1 mRNA upon expression of Cdx1 in IEC6 cells with no change noted in control-infected cells (Fig. 7A). We next examined the expression of mRNAs simultaneously for cyclin D1, D2, and D3 using a ribonuclease protection assay (Fig. 7B). This experiment showed that the mRNA levels of the three D cyclins decreased, as would be predicted from the protein levels with cyclin D1 and D2 mRNAs being markedly reduced, whereas there was little change in mRNA for cyclin D3 (Fig. 7B). Therefore, expression of Cdx1 in IEC6 cells causes a reduction in cyclin D1 and D2 mRNA and protein, with little change in cyclin D3. On days 1-4, n ϭ 7, and day 5, n ϭ 4. Open square, control uninfected; closed diamond, Adeno-Null-infected; closed square, Adeno-CDX1-infected. B, IEC6 cells were infected and cultured as in A, but on days 1, 2, 3, and 4 after infection cells were incubated for 1 h with the nucleoside analogue BrdUrd. Cells were washed and fixed, and then probed with a monoclonal antibody for nuclear BrdUrd incorporation. Nuclei were counterstained with bisbenzamide, and BrdUrd (ϩ) nuclei were counted and expressed as a percentage of all nuclei counted. n ϭ 5 for each condition, each day. These values were transformed using an arcsin transformation and evaluated for statistical significance with analysis of variance and Tukey rank mean tests. *, significant difference from control sham-infected or Adeno-Null-infected cells, p Ͻ 0.01. Representative data from one of three experiments is presented. Black bars, control uninfected; dark gray bars, Adeno-Null-infected; light gray bars, Adeno-CDX1-infected. the cellular equilibrium in the normal epithelium and for adapting to pathophysiological stimuli. Taken together, the data presented in this report shows that Cdx1 is able to inhibit growth of intestinal epithelial cells by blocking progression of the cell cycle in G 1 . Moreover, inhibition of cell cycle progression appears to be due to decreased cyclin D1 and D2 and the decreased levels of protein correlate with decreased mRNA. A recent report came to a different conclusion regarding the effect of CDX1 on proliferation (21). These investigators constitutively expressed human CDX1 in human adenocarcinoma cell line HT-29 and examined the effect on proliferation and differentiation. They found that stable expression of CDX1 did not affect the proliferation of HT-29 cells, whereas co-expression of CDX1 with CDX2 augmented the growth inhibitory effect of CDX2. There are a number of explanations for the differences in our results in IEC6 cells. First, we used the mouse Cdx1 cDNA and these investigators used the human cDNA. It is doubtful that this is the reason for the differences since these two homologous proteins are highly similar. Second, the cell lines used are very different. Our studies were performed in a non-transformed rat intestinal cell line that does not form tumors in nude mice, whereas HT-29 cells are human adenocarcinoma cells. Cellular growth and signaling pathways may be very different between these two lines, which may affect the ability of Cdx1 to regulate proliferation. Third and most importantly, it is likely that there were very different levels of expression in the two studies. The expression of Cdx1 in the HT-29 cells was only evaluated using a luciferase reporter construct that is activated by Cdx proteins, and there was no direct evidence of protein expression. The luciferase reporter assay is very sensitive and suggests that the actual levels of Cdx1 protein may have been very low in the stably transfected lines. Thus, when the different experimental designs are considered, these two apparently contradictory results are potentially explained by differences in cell lines and levels of Cdx1 expression.
It is unclear whether the cell cycle inhibitory effect of Cdx1 is important for regulation of intestinal epithelial growth in vivo.
In the adult small intestinal epithelium, Cdx1 protein is found in greatest amount in the nuclei of crypt cells with decreasing levels in villus cells as they move toward the villus tip (34). Similarly, the highest levels of expression in the colonic epithelium are found in crypt cells with less in the surface epithelial cells (34). However, it is important to note that Cdx1 is expressed throughout the villus as well as the crypt, albeit at lower levels. In Cdx1 null mice, no gross abnormalities were found in the epithelium, but a careful histological analysis has not yet been reported (17). Since Cdx1 is also important for expression of enterocyte genes, there may be other changes in Cdx1 null mouse epithelium that alters proliferation in addition to loss of Cdx1 alone. Additionally, changes in Cdx2 expression in the Cdx1 null mice may mitigate the effect of loss of Cdx1. Thus, the available in vivo data are compatible with multiple possibilities including a pro-proliferative effect, an anti-proliferative effect, or no effect on proliferation. Our data suggest that Cdx1 may inhibit the cell cycle and thereby serve as a break on proliferation in crypt cells. In fact, the cell cycle inhibitory effect is evident over a wide range of levels of Cdx1 protein (compare IEC6 stable lines to adenovirus-infected cells; Fig. 1A). However, it is currently unclear how these protein levels in cell lines compare with the protein expression in colonocytes in the intact mucosa. Moreover, all conclusions of function in cell lines must be tempered by the fact that the microenvironment of the intact epithelium in which the enterocyte resides is very different from cultured cells. Cdx1 has also been suggested to play a role in neoplasia. A previous study suggested that Cdx1 was able to transform NIH-3T3 cells (5). We were unable to confirm these findings since we found no increase in transformed foci or soft agar colonies in 3T3 cells expressing Cdx1. In contrast to the previous study, we confirmed the expression of Cdx1 in the NIH-3T3 clones. Thus, we conclude that Cdx1 does not transform NIH-3T3 cells. The second association with neoplasia is a change in expression in human colonic adenocarcinomas. We (34) and others (33) have shown that the majority of colorectal cancers have decreased levels of Cdx1 in comparison to normal colonic mucosa. Preliminary data suggest that expression of Cdx1 in some colon cancer cell lines that do not express Cdx1 results in inhibition of cell growth (38). Since HT-29 cells are not affected by CDX1 expression (21), the role in growth control of cancer cells will require additional studies.
There are many targets for regulation of the mammalian cell cycle. The balance between activators and inhibitors of specific kinases determines progression through the G 1 phase and entry into S phase. The Rb proteins are important gatekeepers of the G 1 -S transition by regulating the activity of the E2F family of transcriptional activators. Our data are consistent with Rb family members as the ultimate targets of Cdx1 protein expression since there was an increase in the inactive hypophosphorylated form of RB and p130, and loss of p107. Phosphorylation of Rb proteins may be affected by alterations in many proteins and their multiprotein complexes. Expression of Cdx1 does not change levels of the cyclin-dependent kinases or their inhibitors. However, there is a marked decrease in expression of cyclins D1 and D2, which serve to activate the kinase activity of CDK4 and CDK6. As a result, Cdk4-dependent kinase activity is decreased in cells expressing Cdx1. Thus, it is likely that depletion of these D-type cyclins results in decreased Rb family phosphorylation by activated CDKs 4 and 6 and inhibition of progression of the cell cycle. Since cyclin D1 and D2 mRNAs are decreased in parallel with the proteins, it appears that Cdx1 affects either transcription of the cyclin D genes or degradation of the mRNAs, or both. Unraveling this mechanism will require FIG. 4. Western blot analysis for pRb and pRB family proteins in IEC6 cells expressing CDX1. IEC6 cells were cultured and infected as described previously. At 48 h, cells from each condition were assayed by flow cytometry to confirm the G 0 /G 1 block with CDX1 expression. Whole cell extracts were then prepared. 50 g of this whole cell extract from IEC6 cells infected with Adeno-CDX1 (CDX1), Adeno-Null (Null), or uninfected control (Con) cells were loaded per lane and subjected to 6% SDS-PAGE gel separation and Western blotting with antibodies specific for pRb, p130, and p107. Arrows indicate positions of differentially phosphorylated species. Control, Adeno-Null, and Adeno-CDX1 extracts were always prepared simultaneously for each experiment. At least three separate experiments were performed, and the representative results from one are presented. Protein loading controls are presented in other figures. additional studies.
Taken together, our results suggest a role for Cdx1 in the regulation of intestinal epithelial cell proliferation as an inhibitor of cell cycle progression through a decrease in D-type cyclins. We hypothesize that the potency of the Cdx1 inhibitory effect is dependent on other cellular signaling pathways and the state of cellular differentiation. An understanding of the mechanism by which Cdx1 affects the cell cycle will provide insight into maintenance of the delicate balance of proliferation and differentiation in the intestinal epithelium and the pathogenesis of colonic neoplasia. FIG. 5. Western blot analysis for cyclins, cyclin-dependent kinases, and cyclin-dependent kinase inhibitors in IEC6 cells expressing CDX1. Western immunoblotting of whole cell extracts was performed as described previously using the specific antibodies as noted. All blots were routinely stripped and reprobed with a monoclonal antibody to ␣-actin to confirm equal protein loading. A representative blot from one of three experiments is presented.
FIG. 6. Cdx1 expression reduces Cdk4-associated pRb kinase activity in IEC6 cells. A, IEC6 cells were sham-infected or infected with Adeno-Null or Adeno-CDX1 as before. At 48 h, whole-cell extracts were obtained, and 100 g of each extract was subject to immunoprecipitation with an antibody specific for Cdk4 or with a control rabbit IgG. The Cdk4-associated kinase activity present in the IP products was assayed using purified GST-Rb as substrate. The 32 P-labeled products were analyzed on a 12% SDS-PAGE gel. To confirm equivalent levels of Cdk4 in the tested extracts, an immunoblot for Cdk4 and ␣-actin was performed on 50 g of the whole cell extracts. Representative experiments are presented. B, the dried SDS-PAGE gels were exposed to a PhosphorImager (Molecular Dynamics) and the labeling present in the GST-Rb kinase products quantified. Averages and standard deviations were determined for each condition (n ϭ 4). The 32 P labeling activity present in the rabbit IgG IP products was considered to be background, and the average values subtracted from the respective experimental values. To determine the relative Cdk4 kinase activity, the average 32 P labeling of GST-Rb in the control uninfected IEC6 cells was considered to be 100%, and the labeling activity in the Adeno-Nulland Adeno-Cdx1-infected cells is expressed as a percentage of this control value. IEC6 cells were cultured and infected as described previously. At 48 h, cells from each condition were assayed by flow cytometry to confirm the G 0 /G 1 block with CDX1 expression. Total RNA was then isolated. A, 20 g of total RNA from IEC6 cells infected with Adeno-CDX1 (CDX1), Adeno-Null (Null), or uninfected control (Con) cells were loaded per lane and subjected to 0.8% agarose, 6% formaldehyde gel separation and blotting. Blots were probed with 32 P-labeled human cyclin D1 cDNA, washed, and exposed to film. All blots were then stripped of probe and rehybridized for the 36B4 mRNA, an acidic ribosomal phosphoprotein P0 used as a transfer control. Several separate experiments were performed, and the representative results from one are presented. B, 32 P-labeled riboprobes for rat cyclin D1, D2, D3, and actin were prepared as described and hybridized to 10 g of total RNA, and subject to ribonuclease protection analysis. The products were then analyzed on a 5% polyacrylamide, 8 M urea gel, dried down, and exposed to film.