Putative intestine-specific enhancers located in 5′ sequence of the CDX1 gene regulate CDX1 expression in the intestine

Erinn B. Rankin, Wei Xu, Debra G. Silberg, EunRan Suh

Abstract

CDX1 is a homeobox transcription factor that plays a critical role in intestinal epithelial cell growth and differentiation. CDX1 gene expression is tightly regulated in a temporal and cell-type specific manner. However, very little is known about the regulatory mechanisms that direct CDX1 gene expression in the intestine. To elucidate these mechanisms, we employed a series of transgenic mouse studies using the 5′ flanking sequences of the human CDX1 gene. Transgenic mice containing nucleotides between -5667 and +68 relative to the transcription start site of the CDX1 gene demonstrated ectopic expression of the transgene in the brain and gastric smooth muscle. However, transgenic expression of the nucleotides -15601 to +68 of the CDX1 gene was restricted to the intestinal epithelium, which was identical to endogenous CDX1 gene expression. Taken together, the upstream sequences between -15601 and -5667 contain regulatory elements that direct transgene expression specifically to the intestinal epithelium. Furthermore, DNase I hypersensitivity assays revealed two active chromatin regions in the CDX1 gene (hypertensive sites 1 and 2) located at approximately -5.8 and -6.8 kb upstream of the CDX1 gene, respectively, which may function as potential intestine-specific enhancers.

  • transgenic mice
  • homeobox protein

the caudal-related homeodomain transcription factors are members of the nonclustered, hexapeptide superclass of homeobox genes. Several lines of evidence suggest that Cdx1 is an important nuclear protein involved in the regulation of the cellular phenotype of the intestinal epithelium (13, 36, 37). In experiments with intestinal epithelial cell-6 line (IEC-6) cells, an undifferentiated rat intestinal epithelial cell line that does not express Cdx proteins, Cdx1 inhibits cellular growth and induces cellular differentiation characterized by multilayered structures with microvilli (24, 37). We have previously shown that Cdx1 inhibits cellular proliferation via G0/G1 cell cycle block through reduction of cyclin D1 in IEC-6 (24) and DLD1 colon cancer cells (23). Our studies and those of others (26, 35, 39) have demonstrated that CDX1 expression is markedly downregulated in both adenomas and carcinomas of the colon.

The Cdx1 gene is expressed with a complex pattern in two stages during mouse development. At the first stage, Cdx1 is expressed from 7.5 to 12.5 days postcoitum (pc) in the primitive ectoderm and the mesoderm and later concentrates in the neuroectoderm, somites, and developing limb buds, whereas there is no detectable expression in the endoderm (9, 34). At the second stage, Cdx1 expression is limited to the developing intestine, which is first observed at 12.5 days pc in the distal developing intestine and subsequently increases from 13.5 to 14.5 days pc during the time of transition from a stratified endoderm to a columnar epithelium (36). Cdx1 protein expression remains strong in the murine intestinal epithelium throughout life with a decline moving up to the surface epithelium in both small intestine and colon (35). These intricate spatial and temporal patterns of expression imply that regulatory mechanisms that modulate CDX1 gene expression are complex. It has been previously suggested (21) that the Wnt/β-catenin signaling pathway is involved in murine Cdx1 gene regulation in embryonic intestine through Wnt/β-catenin responsive element in the -0.7 kb Cdx1 promoter. However, recent transgenic mice studies (22) revealed that neither -3.6 nor -0.7 kb Cdx1 upstream regulatory sequences were sufficient to drive expression in the definitive endoderm at later stages of development, although they were able to direct the early phase of Cdx1 expression in the primitive streak. To date, very little is known about the molecular mechanisms that regulate the developmental and spatial patterns of CDX1 expression in normal intestine or what induces its downregulation in colonic adenomas and cancers.

In the present study, we focus on the initiation and maintenance of human CDX1 gene transcription in the intestine and analyze cis-acting regulatory elements through comprehensive transgenic mice studies. We observed that putative intestine-specific enhancers are located at the upstream of the CDX1 gene between nucleotides -15601 and -5667. Furthermore, we mapped two hypersensitive regions between -6.8 and -5.8 kb of the CDX1 gene using DNase I hypersensitive site assays, which may play roles as enhancers. Our results suggest that there are three distinct complexes involved in establishing an active configuration of chromatin around the CDX1 gene: 1) the minimal promoter (-327 to +68), 2) intestinal-specific enhancers (-6.8 to -5.8 kb), and 3) an insulator. Tissue-specific CDX1 enhancers may play a significant role in orchestrating the switch from the earlier stage expression of the CDX1 gene to later, persistent expression in the intestine.

MATERIALS AND METHODS

Plasmids. The CDX1-327+68LacZ transgene was constructed by subcloning a ApaI-BamHI NLSLacZ fragment digested from KSNLSLacZ (obtained from Dr. Klaus Kaestner) into ApaI-BglII digested -327CDX1 Luc (39). Subsequently, the SmaI-NarI fragment from -327CDX1LacZLuc was subcloned into EcoRV-AccI digested Bluescript KS to obtain KS-327CDX1LacZ. The CDX1-5.6k+68LacZ construct was generated by inserting a 5.7-kb EagI fragment containing -5667 + 68 of the CDX1 gene into EagI-digested KS-327CDX1LacZ. For CDX1-15k+68LacZ, a 13.6-kb NotI-SnaBI fragment from the 201E12 containing -15601 to -2302 was inserted to extend the upstream sequence of the CDX1 gene into NotI-SnaBI digested -5.6 kb CDX1LacZ, which replaces -5667 to -2302 of the CDX1 gene in the CDX1-5.6k+68LacZ. The HindIII fragment from the CDX1-5.6k+68LacZ construct was inserted into the HindIII site in pGL2 basic (Promega) to generate a -5667CDX1 Luc reporter. For the -10800CDX1Luc reporter, the MluI-SnaBI fragment from the 201E12 containing -10800 to -2302 was inserted into MluI-SnaBI digested -5667CDX1Luc to extend the upstream sequence of the CDX1 gene.

Generation of CDX1 transgenic mice. All animal experiments were performed under an animal use protocol approved by the University of Pennsylvania Office of Regulatory Affairs. Transgenic mice were produced by the Transgenic Core Facility at the University of Pennsylvania. The DNA construct was injected into the male pronuclei of fertilized eggs and implanted into pseudopregnant females by standard methodology. DNA from tail biopsy samples of the resulting mice was extracted with the QIAamp tissue kit (Qiagen, Valencia, CA). The presence of the transgene in mouse genomic DNA was determined by PCR and Southern blot analysis as described previously (27, 42). Transgene founders of the C57BL/6J strain were bred with normal CD1 mice (Charles River Laboratories, Wilmington, MA), and offspring were analyzed for the transgene by PCR.

RNase protection assay. RNA was extracted from multiple tissues using a CsCl gradient method as previously described (38). RNase protection assays were performed by using the RPA II kit (Ambion, Austin, TX) according to the manufacturer's recommendations. Riboprobes for the detection of mouse Cdx1 and transgene CDX1 mRNA were prepared as described previously (38) using HincII digested pCDX1–5 (5) and StuI-digested pCdx13′ (9), respectively.

In situ hybridization. In situ hybridization was performed as described previously (6, 35). Briefly, intestinal tissues or embryos were fixed in 4% paraformaldehyde and embedded in paraffin. Sense and antisense riboprobes specific for either human CDX1 or mouse Cdx1 were synthesized by transcribing BglI (sense) or HincII (antisense) digested pCDX1–5 or StuI-digested pCdx13′ (mouse Cdx1 antisense) in the presence of [33P]UTP. After hybridization and washes, slides were dehydrated through increasing concentrations of ethanol, air dried, and exposed on Biomax MR film (Eastman Kodak, Rochester, NY) for 24 h. Slides were then dipped in liquid emulsion (NTB2 diluted 1:1 in water; Kodak) and exposed in light-tight boxes at 4°C. The slides were developed after 2 wk of exposure and stained with hematoxylin and eosin.

β-Galactosidase detection in embryos and tissues. Animals were dissected, and tissues or embryos were harvested, rinsed in ice-cold PBS, and fixed in 0.2% glutaraldehyde, 2 mM MgCl2, 5 mM EGTA, and 0.02% Nonidet P-40 in PBS for 30 min (dissected tissues), 1 h (embryos), and 10 min (embryonic intestine). Embryos were genotyped by PCR using DNAs isolated from yolk sac with primers detecting the LacZ transgene. After fixation, tissues or embryos were washed three times for 10 min in PBS and incubated for 12–16 h in staining solution {5 mM K3[Fe(CN)]6, 5 mM K4[Fe(CN)]6, 2 mM MgCl2, 0.02% Nonidet P-40, 0.01% sodium deoxycholate, 1 mg/ml X-Gal (GIBCO-BRL) in PBS}. Subsequently, embryos were washed three times for 10 min in PBS and postfixed in 4% paraformaldehyde, pH 7.2, overnight at 4°C. Tissues or embryos were then washed in PBS, dehydrated, paraffin embedded, sectioned, and photographed.

Cell culture and transient transfection. Human cancer cell lines, HCT116, T84, and Hela were obtained from American Type Culture Collection (Manassas, VA). HCT116 cells were maintained in McCoy's 5A medium supplemented with 10% FBS, T84 cells in 50% DMEM, and 50% F-12 medium supplemented with 5% FBS, and Hela cells in DMEM with 10% FBS. Penicillin (100 U/ml) and streptomycin (0.1 mg/ml) were added to each medium. Transfections were performed with Lipofectamine (Invitrogen, Carlsbad, CA) according to the manufacturer's recommendations. Basically, cells at 40–60% confluence were incubated with a constant amount of 1.5 μg of total DNA and 2.5 μl of Lipofectamine/1 ml of OPTI-MEM for 18 h. Luciferase activity was determined 72 h after the transfection with the luciferase assay kit (Promega Biotech). The β-galactosidase expression vector was cotransfected in each experiment as a measure of transfection efficiency, and the results were reported as light units per unit of β-galactosidase.

DNase I hypersensitivity analysis. Nuclei were isolated from T84 or Hela cells as described previously (16). Aliquots of 1.5 × 107 nuclei were digested with increasing amounts of DNase I as described, (14) and genomic DNA was isolated. DNA was digested with specified restriction enzyme, separated on an agarose gel, transferred to a nylon membrane, and hybridized with either 0.5 kb StyI-NotI or 0.3 kb BamHI-XbaI fragment labeled with [32P]dCTP. Hypersensitive sites were mapped by comparison with labeled 1-kb ladder (Invitrogen).

RESULTS

A cosmid clone 201E12 contains the necessary and sufficient regulatory elements for correct spatiotemporal pattern of CDX1 gene expression in the intestine. To assess whether the CDX1 gene fragment in a cosmid (201E12: Fig. 1A) (5) contained all the necessary information to direct the correct pattern of CDX1 gene expression in an intestine-specific manner, a fragment containing the nucleotides between -15 and +24 kb of the CDX1 gene was purified to produce transgenic animals. PCR and Southern blot analysis were performed to screen for the human CDX1 transgene (data not shown). Three founders containing the CDX1 transgene were bred with normal CD-1 mice. The tissue expression pattern of the CDX1 transgene in the transgenic animals was studied by RNase protection assay with RNAs isolated from multiple tissues. RNAs isolated from the cecum of a nontransgenic mouse and a CDX1 expressing colorectal cancer cell line, T84 cells, were used as positive controls for endogenous mouse Cdx1 or human CDX1, respectively.

Fig. 1.

Transgene CDX1-15k+24k expression recapitulates the endogenous mouse Cdx1 mRNA expression in the intestine. A: transgene construct of the human CDX1 gene used in transgenic mice, CDX1-15k+24k, contains a 40-kb CDX1 gene fragment including 15 kb of the 5′ flanking sequences and all 3 exons. (solid boxes, exons; arrows, transcription start site). B: RNase protection assay of CDX1-15k+24k transgene expression. Total RNA isolated from intestinal and nonintestinal tissues of a 8-wk-old CDX1-15k+24k transgenic mouse was analyzed for human CDX1 (hCDX1) and mouse Cdx1 (mCdx1) transcripts. RNAs isolated from the cecum of a nontransgenic littermate and T84 cells, were used as positive controls for endogenous mouse Cdx1 or human CDX1, respectively. tRNA was used as a negative control. CD: transgene expression during development and along the vertical axis in the small intestine and colon were analyzed by in situ hybridization. E13.5 embryos (C) and intestinal tissues (D) were sectioned and hybridized with an antisense human CDX1 probe. Positive signals are shown in pink. Sense CDX1 and antisense mouse Cdx1 probes were used as negative and positive controls, respectively. Sections were counterstained with hematoxylin and eosin (H&E) when indicated. Original magnification: C, ×200; D, ×100.

As shown in Fig. 1B, expression of the human CDX1 transgene (CDX1-15k+24k) from founder 1 was limited to intestinal tissues, which was similar to that of endogenous mouse Cdx1. Quantitation obtained by Phosphorimager showed that the ratio of the transgene to endogenous mouse Cdx1 in the small intestine, cecum, proximal colon, or distal colon was approximately the same (data not shown). To examine the temporal and spatial expression of the transgene in the intestine, we performed in situ hybridization using the CDX1 riboprobe specific for the human CDX1 transcript. Results demonstrated that the human CDX1 transgene was first observed in the intestinal endoderm at 12.5 days pc. Expression of the CDX1 transgene increased from 13.5 to 14.5 days pc during the time of transition from a stratified endoderm to the columnar epithelium of the intestine (Fig. 1C). Expression of the transgene CDX1-15k+24k continued in the intestinal epithelium throughout life, predominantly in the crypt of the small intestine and the base of the crypt in the colon, which reflects the pattern of endogenous Cdx1 gene expression (Fig. 1D). Identical results were obtained from all three founders, suggesting that expression of the transgene CDX1-15k+24k was not dependent on the insertion site in the mouse genome. Results demonstrate that the sequence between -15 and +24 kb of the CDX1 gene contains all the necessary elements to regulate the correct pattern of CDX1 gene expression in the intestine.

Putative intestinal epithelial cell specific enhancers are located at outside of -5.6 kb of the 5flanking sequence of the CDX1 gene. We have previously demonstrated that most of the intestine-specific transcriptional activity resides between the nucleotides -327 and +68 of the CDX1 gene in vitro (39), suggesting that the CDX1 minimal promoter contains the necessary elements that direct transcription in intestinal cell lines. To localize intestine-specific regulatory elements in the upstream sequence of the CDX1 gene in vivo, we generated transgenic mice using a series of CDX1 gene constructs to drive the LacZ reporter. Two or more independent transgenic founder lines were generated for each construct. The CDX1 minimal promoter transgene construct (CDX1-327+68LacZ) was generated by linking the minimal promoter (-327 and +68) of CDX1 to the nuclear localizing β-galactosidase coding sequence (Fig. 2, top). Five founder mice had the intact transgene and were analyzed for expression of β-galactosidase by LacZ staining in mouse tissues. In the offspring from founder 532 (Fig. 2, AD), expression of the CDX1-327+68LacZ transgene was detected in the intestinal epithelium at the base of the crypt of the small intestine and scattered within the colon. LacZ staining of the offspring from founder 533 was detectable in granular cells and the Purkinje cells in the brain (Fig. 2, EF). However, three other founders (534, 535, and 536) showed no expression of the CDX1-327+68LacZ transgene. The data suggest that for the minimal promoter, transgene expression is dependent on the site of insertion. Therefore, we generated transgenic animals with an extended promoter region, containing nucleotides -5667 to +68 of the CDX1 gene (Fig. 3, top). Of three transgenic founders, LacZ activity was detected in only one line (founder 984). In offspring of founder 984, the transgene CDX1-5.6k+68LacZ was detected in the serosa of the small intestine and colon (mesothelium; Fig. 3, AC), submucosal muscle layers in the stomach (Fig. 3D), the squamous cells lining Bowman's capsule of the kidney (Fig. 3E), and granular cells in the cerebellum (Fig. 3F). The results suggest that expression of the transgene CDX1-5.6k+68LacZ is insertion-site dependent and that it does not contain an intestine-specific regulator within the nucleotides -5667 to +68 of the CDX1 gene.

Fig. 2.

Ectopic expression of the transgene CDX1-327+68LacZ. Top: schematic diagram of the transgene construct of the CDX1 minimal promoter linked to the LacZ reporter. The solid box represents a nuclear localization signal (NLS) located upstream of the LacZ reporter. Expression of the CDX1-327+68LacZ transgene was detected by LacZ staining in the intestinal epithelium of the 8-wk-old offspring from founder 532: proximal small intestine (A); distal small intestine (B); proximal colon (C); distal colon (D) and in the cerebellum of the brain from the offspring of founder 533 in the granular layer (E) and Purkinje cells (F). Selected positive signals in the nucleus are shown by arrowheads. Original magnification: A, C, E, and F, ×200; B and D, ×100. Normarski optics.

Fig. 3.

Ectopic expression of the transgene CDX1-5.6k+68LacZ. Top: schematic diagram of the transgene construct of the CDX1 gene linked to the LacZ reporter (CDX1-5.6k+68LacZ) (solid box, NLS). Expression of the CDX1-5.6k+68LacZ transgene was detected by LacZ staining in the 8-wk-old offspring of founder 984 in the serosa in the small intestine (A) and proximal (B) and distal colon (C), submucosa muscle layers in the stomach (D), kidney (E), and the cerebellum (F). RC, renal corpuscles in the kidney; GL, granular layer; ML, molecular layer. Original magnification: AD, F, ×100; E, ×200. Normarski optics.

To determine whether an intestine-specific region could be identified, we further extended upstream sequences of the CDX1 gene to generate the transgene CDX1-15k+68LacZ (Fig. 4; top). Two founders were identified that had an intact transgene, and tissues were analyzed for expression of β-galactosidase by LacZ staining. In the offspring from founder 1574, strong LacZ staining was detected only in intestinal epithelial cells of the ileum and the colon (Fig. 4, A and C), but not other tissue types (Fig. 4, EG). Identical results were obtained of the offspring from founder 1742 suggesting that expression of the transgene CDX1-15k+68 was insertion-independent in the mouse genome. Results indicate potential intestinal epithelial-specific enhancers residing between the -15601 and -5667 upstream sequences of the CDX1 gene.

Fig. 4.

Intestinal epithelial specific expression of the CDX1-15k+68LacZ transgene. Top: schematic diagram of the transgene construct of the CDX1 gene linked to the LacZ reporter (CDX1-15k+68LacZ) (solid box, NLS). Expression of the CDX1-15k+68LacZ transgene was detected by LacZ staining in the 8-wk-old offspring of founder 1574 in the epithelial cells of the small intestine (A) and colon (C), but not the kidney (E), stomach (F), or brain (G). Small intestine (B) and colon (D) from nontransgenic littermate shows no LacZ staining. Original magnification: A, B, EG, ×100; C and D, ×400. Normarski optics.

Transgene CDX1-15k+68LacZ expression during early development. To examine whether the transgene CDX1-15k+68 LacZ contains the essential elements that direct early expression of CDX1 during mouse development, the temporal expression patterns of the CDX1-15k+68LacZ transgene were analyzed at 10.5, 12.5, 13.5, and 14.5 days pc. At 10.5 days pc, transgene expression was detected in the somites, the neural tube, limb bud, and the tail bud with strong expression in caudal region (Fig. 5A). Transgene expression in the mesoderm and ectoderm decreased gradually after 12.5 days pc (Fig. 5, B and C). In the embryonic intestine, LacZ activity was detected initially at 10.5 days pc (data not shown), and transgene expression was increased in the intestinal epithelium at 13.5 days pc (Fig. 5, D and E) when the endoderm-to-intestinal epithelium transition occurs. To examine distribution of transgene expression along the anterior-posterior axis in the developing intestine, an entire embryonic digestive tract was isolated from a 14.5-day pc embryo and stained for β-galactosidase activity. As shown in Fig. 5F, intense LacZ staining was observed in the midgut, which forms jenunum and ileum, and weakened gradually in the foregut and hindgut, whereas the embryonic stomach was negative. These results demonstrate that the nucleotides between -15 and +68 kb of the CDX1 gene in CDX1-15k+68LacZ mice are able to switch transgene expression from mesoderm and ectoderm during early development to endoderm specific in later development, which is a similar pattern to endogenous Cdx1 gene expression.

Fig. 5.

Expression of the CDX1-15k+68LacZ transgene during early development. AC: LacZ staining of whole mouse embryos, E10.5 (A), E12.5 (B), and E13.5 (C). D: sagittal section of E13.5 embryo. E: expression of transgene was restricted in the endoderm of the midgut. Intestinal epithelium is shown with arrowhead. F: horizontal distribution of transgene expression in the isolated digestive tract (E14.5). Tb, tail bud; Lb, limb bud; St, stomach. Original magnifications: AD, F, ×20; E, ×200, Normarski optics.

Mapping of the intestine-specific enhancers of the CDX1 gene. To localize the intestine-specific enhancer that directs transgene expression in the intestinal epithelial cells in CDX1-15k+68LacZ mice, a DNase I hypersensitive assay was carried out on the chromatin structure from colonic-derived T84 cells and Hela cells, a cervical cancer cell line, as a nonintestinal control. After digestion with limited DNase I, genomic DNA was digested with either NotI or BamHI and analyzed by Southern blot analysis. DNase I hypersensitive sites were mapped by indirect end labeling using probe 1 or probe 2, respectively (Fig. 6). Analysis of the NotI fragment spanning sequences approximately -40 to +802 kb relative to the transcriptional start site of the CDX1 gene revealed two major DNase I hypersensitive site at -6 and -7 kb that are specific to T84 DNA (Fig. 6, A and B). These two hypersensitive sites were confirmed by hybridization of the BamHI digested T84 DNA on the filter with probe 2 (Fig. 6, C and D). Results demonstrate that there are two intestine-specific hypersensitive sites mapped at approximately -5.8 and -6.8 kb upstream of the CDX1 gene, designated as hypersensitive sites 1 and 2, respectively. We then examined the functional relevance of the enhancer sequences by reporter gene assays using the deletional constructs of the CDX1 gene linked to the luciferase reporter. The transfection experiments were performed in T84 and Hela cells. As shown in Fig. 6D, the reporter activity of the deletional construct containing the upstream sequences between -10800 and +68 of the CDX1 gene was fourfold higher than the reporter with further deletion to -5667 and -327, whereas Hela cells showed little changes. This result supports our notion there are intestine-specific enhancers as emerged from the transgenic mice studies and DNase I hypersensitive assays.

Fig. 6.

Identification and analysis of the putative intestine-specific enhancers. AC: DNase I-hypersensitive sites (HS) at the upstream of the CDX1 gene. Nuclei from T84 and Hela cells were treated with increasing amounts of DNase I and genomic DNA was isolated, restricted with NotI (A), or BamHI (B), and hybridized with probes 1 and 2, respectively, as indicated by black boxes (C). The autoradiographs show the result of hybridization. A: hypersensitive bands a and b are produced from hybridization with NotI digested DNA. B: two hypersensitive sites are colocalized with the BamHI-digested fragment (c and d). C: schematic diagram represents the CDX1 gene, showing the exon 1 as an open box. The restriction map includes HindIII (H), BamHI (B), and NotI (N) sites. Two hypersensitive sites within the CDX1 gene are present in T84 and are indicated by arrows (HS1 and HS2). These hypersensitive sites are absent in nuclear DNA from Hela cells. D: functional analysis of the deletional constructs of the CDX1 gene. Plasmid constructs containing various lengths of the 5′ flanking region of the CDX1 gene linked to the luciferase reporter gene (-327, -327CDX1Luc; -5667, -5667CDX1Luc; -10800, -10800CDX1Luc) were transfected into T84 and Hela cells. Values for luciferase activity were normalized to β-galactosidase activity. Values are expressed as means ± SD of 4 independent experiments.

Hypersensitive sites of the CDX1 gene are highly conserved between human and mouse. We then aligned the human and mouse genomes using the Vista program (http://www-gsd.lbl.gov/vista) (7, 8) to determine whether the hypersensitive sites 1 and 2 were evolutionarily conserved between the human CDX1 gene and corresponding sequences of the mouse Cdx1 gene. Results demonstrated that there were high similarities in sequences between nucleotides -6111 and -5852, and between nucleotides -6801 and -6631 relative to the human CDX1 transcription start site (Fig. 7). We then analyzed the hypersensitive sites 1 and 2 sequences for potential binding sites for transcription factors using the TESS program (http://www.cbil.upenn.edu/tess). As shown in Fig. 7, there are conserved elements including CAAT displacement protein (CDP) CR3/HD, CAC binding protein, nuclear factor-1, C/EBPα, HNF3, WT1 sites, and E box.

Fig. 7.

Sequence analysis of the HS1 and HS2 regions. HS1 (A) and HS2 (B) are located in highly conserved human and mouse sequences. Shaded areas represent regions of complete homology between the human and mouse genes. Putative binding sites for transcription factors are shown in solid lines.

DISCUSSION

Studies of homeobox genes can provide a model system to understand the mechanisms of transcriptional regulation during intestinal development. Complex patterns of gene expression are governed by multiple regulatory elements including promoters, enhancers, silencers, and insulators. Very little is known about the switch that occurs in CDX1 gene regulation from ectomesoderm specific to endoderm-intestine-specific expression. In the present study, we identify putative intestine-specific enhancers within the 5′ flanking sequences of the CDX1 gene by transgenic mouse reporter analysis and further explore its functional significance by DNase I hypersensitivity assays. Comparisons among the series of transgenic mouse experiments identified expression directed by the cis-regulatory elements of the CDX1 gene. This revealed complex mechanisms in the regulation of CDX1 gene expression.

Whereas the minimal promoter (-327 to +68) of the CDX1 gene contains the necessary cis-regulatory elements for intestine-specific gene transcription in vitro (39), the expression pattern of the transgene CDX1-327+68LacZ in mice was highly insertion dependent. One potential silencing mechanism of the transgene CDX1-327+68LacZ is the lack of an insulator. Several insulators have been described in Drosophila (4, 11, 12, 45) and vertebrates (17, 28). Insulators can either block communication between an enhancer and a promoter, if they are located between them or protect an integrated transgene from local position effects (28, 40, 44). Previously, we have shown that the CpG islands of the CDX1 gene promoter is not methylated in normal intestinal epithelium but is hypermethylated in colon cancer that accordingly silences endogenous CDX1 gene expression (39). It is interesting that the integrated transgene CDX1-327+68LacZ was highly methylated in the founders with no transgene expression (data not shown). This observation supports the notion that absence of insulators may protect the CpG islands of the CDX1 promoter from binding of repressor complexes, such as histone deacetylase, methyl-binding proteins, DNA methyl-transferases, and corepressors (28). Transgene silencing in two of three transgenic mice containing -5667 to +68 of CDX1 might also be explained by the absence of an insulator.

By contrast, localized expression of transgene CDX1-5.6kb+68 LacZ in a number of mesoderm and neuroectoderm tissues from one founder suggests that there may be a mesoectoderm specific enhancer within -5667 to +68 of the CDX1 gene. This result agrees with a previous report demonstrating that 3.6 kb of the mouse Cdx1 promoter region is sufficient to drive transgene expression in the primitive streak, but this sequence does not contain the cis-regulatory elements necessary to drive later transgene expression in the definitive endoderm (22). Gaunt et al. (10) reported studies with chick Cdx-A/LacZ reporters in transient transgenic mouse embryos, demonstrating evidence for an intron enhancer that includes two functional control elements: 1) a DR2-type retinoic acid response element and 2) a Tcf/β-catenin binding motif. Both studies identified elements necessary for the first phase of CDX1 expression in neuroectoderm and mesoderm, but not for the second phase of CDX1 expression in the intestinal epithelium. Importantly, transgene CDX1-15kb+68LacZ restricted transgene expression in the intestinal epithelium, suggesting that the 5′ flanking sequences between -15601 and +68 of the CDX1 gene contain both insulator sequences and intestine-specific enhancers. The distribution of LacZ staining of transgene CDX1-15kb+68LacZ in the intestinal epithelium was not stronger in the crypt than in the villus, compared with endogenous Cdx1 expression. It is possible that the 5′ flanking sequences between -15601 and +68 of the CDX1 gene lack regulatory elements for fine tuning of regulation of CDX1 protein synthesis in the crypt-villus axis. However, we wish to add a caveat, namely that differences in turnover of CDX1 and LacZ proteins between the villus and crypt compartments of the intestine may potentially play a role.

Our sequence analysis indicates that the newly identified putative intestine-specific enhancer region (hypersensitive site 1) contains binding sites for a number of transcription factors. Among them, CDP (also called Cux) is one of potent factors that may play a role in CDX1 gene regulation via the intestine-specific enhancer, because CDP is known to function as either a repressor or activator in cell proliferation and differentiation. CDP contains three cut repeats and a divergent homeodomain, each of which interacts with DNA in a complex manner (1, 6, 29, 41). CDP was originally cloned by an interaction with an element that is important for myeloid cell differentiation (30, 31). Several reports, mainly based on gene expression studies, suggest that CDP orthologs function as transcriptional repressors by direct competition with transcriptional activators (1, 15, 25, 43) or by active repression by interacting with the HDAC1 protein to promote histone deacetylation (20, 25). Other reports, by contrast, demonstrated that CDP acts as a transcriptional activator (41). Although CDP binds to the putative intestine-specific enhancer hypersensitive site 1 of CDX1 through the CR3/HD domain (data not shown), its functional role in the regulation of the transition between the first and second phases of CDX1 expression will be further studied. On the other hand, hypersensitive site 2 contains DNA binding sites with a consensus E-box (NCANNTGN) for the basic helix-loop-helix (HLH) proteins. The bHLH proteins are important regulatory factors in transcriptional networks of many developmental pathways including cell proliferation, cell lineage, neurogenesis, and myogenesis (2, 3, 18, 33), whereas homeobox proteins have been implicated in patterning, segmentation, and cellular differentiation. In pancreatic β-cells, for example, stable protein-protein interaction between bHLH factors (i.e., E47/Pan1) and Pdx1 can synergistically transactivate the insulin promoter through cooperative DNA binding (32). During neurogenesis, proneural bHLH factors such as Ngn2 and NeuroM interact with homeobox protein, Isl1 and Lhx3, to regulate motor neuron identity (19). The functional dependency of LIM-homeodomain (LIM-HD) factors on proneural bHLH proteins on the motor neuron enhancer establishes a regulatory link between these transcription factor pathways and cellular differentiation. With these examples of transcriptional synergy between bHLH proteins and homeobox proteins, future investigations will test the interplay of different combinations of distinct transcription factors on hypersensitive sites in detail.

In conclusion, we have identified for the first time the putative intestine-specific enhancer region that is sufficient to direct CDX1 expression in the intestinal epithelium. Studies on assembly of protein factors around putative tissue-specific enhancers of the CDX1 gene will provide further insights into mechanisms that orchestrate the switch of CDX1 gene expression from mesoderm and ectoderm-specific to an intestine-specific compartment during development.

Acknowledgments

We specially thank Dr. Peter G. Traber for generous support. We thank Dr. Anil K. Rustgi for review of the manuscript and Drs. Gary Swain and John P. Lynch for discussions. We also thank Dr. François Boudreau for the GST-CDP fusion proteins and Sandra Mancano for technical assistance.

GRANTS

This work was supported by the National Institutes of Health Grants R01-CA-81342 (to E. R. Suh), R01-DK-46704 (to E. R. Suh), R01-DK-59539 (to D. G. Silberg), and P30-DK-50306 (to the Morphology, Transgenic, Molecular Biology, and Cell Culture Cores of the Center for Molecular Studies in Digestive and Liver Diseases at the University of Pennsylvania).

Footnotes

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • * E. B. Rankin and W. Xu contributed equally to this work.

References

View Abstract