This Article Abstract Full Text (PDF) All Versions of this Article: 18/8/1941    most recent Author Manuscript (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager NURSA Molecule Pages Link Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Malo, M. S. Articles by Hodin, R. A. Search for Related Content PubMed PubMed Citation Articles by Malo, M. S. Articles by Hodin, R. A.

Thyroid Hormone Positively Regulates the Enterocyte Differentiation Marker Intestinal Alkaline Phosphatase Gene via an Atypical Response Element

Department of Surgery, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts 02114

Address all correspondence and requests for reprints to: Richard A. Hodin, M.D., Department of Surgery, Massachusetts General Hospital, Gray-Bigelow 504, 55 Fruit Street, Boston, Massachusetts 02114. E-mail: rhodin{at}partners.org.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
Thyroid hormone (T₃) is a critical regulator of intestinal epithelial development and homeostasis, but its mechanism of action within the gut is not well understood. We have examined the molecular mechanisms underlying the T₃ activation of the enterocyte differentiation marker intestinal alkaline phosphatase (IAP) gene. RT-PCR and Western blotting showed that thyroid hormone receptors TR1 and TRß1 were expressed in human colorectal adenocarcinoma Caco-2 cells. Northern blotting detected expression of two IAP transcripts, which were increased approximately 3-fold in response to T₃. Transient transfection studies with luciferase reporter plasmids carrying various internal and 5' deletion mutations of the IAP promoter localized a putative thyroid hormone response element (TRE) to a region approximately 620 nucleotides upstream (–620) of the ATG start codon. EMSAs using TR1-retinoid X receptor (RXR) on sequential 5' and 3' single nucleotide deletions defined the TRE between –632 and –612 (5'-TTGAACTCAgccTGAGGTTAC-3'). Compared with the consensus TRE, the IAP-TRE is novel in that it contains an everted repeat of two nonamers (not hexamers) separated by three nucleotides. Neither TR1 nor RXR binds to the IAP-TRE; however, TRß1 binds to this TRE with minimal affinity. In the presence of TR and RXR, only the TR-RXR heterodimer binds to the IAP-TRE. Mutagenesis of either nonamer abolishes the biological activity of IAP promoter. We have thus identified a novel response element that appears to mediate the T₃-induced activation of the enterocyte differentiation marker, intestinal alkaline phosphatase.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
THE HORMONE T₃ PLAYS a pivotal role in mammalian intestinal epithelial development and homeostasis, affecting the fundamental processes of growth and differentiation. Endogenous T₃ levels increase just before weaning, a period marked by a dramatic growth spurt within the gut, as well as a shift in the expression of a number of intestinal gene products, e.g. sucrase, maltase, and intestinal alkaline phosphatase (IAP) activities increase, whereas lactase and acid ß-galactosidase decrease (1). Hypothyroid animals display marked crypt-villus hypoplasia, and abnormally retain the suckling pattern of expression of the brush-border enzymes, e.g. high lactase and low IAP levels (2, 3, 4, 5, 6, 7). In addition to its effect on the developing gut, T₃ exerts dramatic effects in the adult intestine, including mucosal hyperplasia and changes in enterocyte phenotype, e.g. increase in IAP and decrease in lactase expression (2, 3, 4, 5, 6, 7, 8). T₃ mediates its effects through its nuclear receptors [thyroid hormone receptors (TRs), see below], and TR knockout mice have been examined to determine the effects of T₃ on gut development and differentiation. Although TRß knockout mice appear to have no obvious intestinal abnormalities, TR knockout mice display marked hypoplasia in both the crypts and villi, and also have decreased levels of various digestive enzymes (lactase, sucrase, aminopeptidase) and gut transcription factors (Cdx-1 and Cdx-2) (9, 10, 11, 12, 13, 14). Relatively little is known about the molecular mechanisms that govern T₃ action within the gut. As such, the present work was designed to elucidate the molecular mechanisms underlying the T₃-mediated regulation of one specific target gene, the enterocyte differentiation marker IAP gene.

IAP is a membrane-bound glycoprotein that hydrolyzes a wide variety of monophosphate esters at an alkaline pH optimum (15, 16, 17). IAP is exclusively expressed in villus enterocytes and hence serves as an excellent marker of crypt-villus differentiation (18). IAP comprises a major component of the surfactant-like particles that are seen after fat feeding (19). Although the IAP protein is thought to be involved in the process of fat digestion and absorption, a precise physiological role for IAP has not yet been determined. Recent work by Narisawa et al. (20) on IAP knockout mice has shown that IAP limits fat absorption during high-fat feeding. Previously, using both in vitro and in vivo models, we have demonstrated that the IAP gene is a direct target gene for thyroid hormone (21, 22).

T₃ action is generally mediated through its nuclear receptors (TR) (23, 24) that belong to a super family of transcription factors, which also includes receptors for retinoic acid, vitamin D, various steroids (androgen, estrogen, glucocorticoids, etc.), and others. Different isoforms of TRs have been isolated. The TR gene on human chromosome 17 generates TR1 and TR2 transcripts, which are the splice variants of a major transcript, and produce 410 and 490 amino acid proteins, respectively (23). In addition, other truncated TR proteins appear to exist and may be important in modulating T₃ action (10). The human TRß gene on chromosome 3 produces 461 and 476 amino acid proteins TRß1 and TRß2, respectively, from two splice variants of a single transcript (23). As bona fide receptors, TR1, TRß1, and TRß2 contain both DNA- and ligand (T₃)-binding domains. The TRs interact with specific DNA sequences termed thyroid hormone response elements (TREs), which generally contain a loosely conserved hexamer (AGGTCA) repeat intervened by four nonspecific nucleotides, also known as direct repeat 4 (DR4) (24, 25, 26, 27). The typical half-sites (AGGTCA) are rarely found to be 100% conserved in actual genes. TR can also bind to other forms of these two hexameric repeats, such as palindromic hexamers (28) and everted repeats (29). TRs require heterodimerization with retinoid X receptor (RXR) for their binding to DR4 and other TREs containing hexamer repeats (24, 30, 31, 32, 33). An octamer TRE (T(A/G)AGGTCA) has been shown to bind to a TR monomer, thus eliminating the necessity of TR-RXR heterodimerization for TR-mediated activity (34, 35).

In this study, we show that T₃ induces endogenous IAP gene expression in the human colorectal adenocarcinoma Caco-2 cell line. This IAP gene trans-activation appears to be mediated through an interaction between the TR-RXR complex and a unique TRE located between –632 and –612 in the IAP promoter consisting of an everted repeat of two nonamers spaced by three nonspecific nucleotides. These results provide a molecular description of how thyroid hormone activates a specific enterocyte target gene and will have implications for understanding the myriad effects of T₃ on gut development and homeostasis.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
TR1 and TRß1 Are Expressed in Caco-2 Cells
T₃ action is mediated through its receptors (TRs), and hence we determined the level of expression of different TR isoforms in Caco-2 cells using RT-PCR followed by nested PCR (Fig. 1A). An expected PCR product of 347 bp of TR1 cDNA was obtained using primers TRa1.1184F and TRa1.1530R (see Materials and Methods). Similarly, the primers TRb1.121F and TRb1.605R (see Materials and Methods) amplified the expected 485 bp fragment of the TRß1 cDNA. No PCR product was amplified for TR2 and TRß2. These results indicate that TR1 and TRß1 are expressed in Caco-2 cells, whereas TR2 and TRß2 are probably not expressed. The two extra 0.60- and 0.65-kb faint bands that are amplified in the case of TRß1 could represent other isoforms of TR, or artifacts of PCR.

View larger version (31K): [in this window] [in a new window]   Fig. 1. Activation of the IAP Gene by Endogenous TRs in the Caco-2 Cell Line A, Expression of TR mRNAs. RNA derived from Caco-2 cells was subjected to RT-PCR followed by nested PCR using isoform-specific primers (see Materials and Methods). PCR products were electrophoresed in a 2% agarose gel and stained with ethidium bromide. PCR was repeated more than three times, and the photograph of a representative gel is shown. B, Expression of TR1 and TRß1 proteins. Western blotting was performed on the lysate (UN) or nuclear extract (NE) from untransfected Caco-2 cells as well as on the lysate from Caco-2 cells transfected (TF) with an expression plasmid overexpressing TR1 or TRß1 (see Materials and Methods). The experiments were repeated more than three times, and the photograph of a representative autoradiograph is shown. C, Effects of T3 on the expression of IAP gene. Northern blots illustrating the time course of the T3 (100 nM) effects on the expression of IAP gene in Caco-2 cells. Twenty micrograms of total RNA were loaded in each lane. The 3.0- and 2.7-kb IAP mRNA bands are marked. The ß-actin control detecting equal loading is shown in the bottom panel. Northern analyses were repeated more than three times, and similar results were obtained. The figure shows the photographs of representative autoradiographs. D, Endogenous TR-mediated activation of the IAP-luciferase reporter gene. Transient transfection assays showing the time course of the T3 effects on the activation of the IAP-luciferase reporter gene in Caco-2 cells. The 2.5-kb 5' flanking region of the IAP gene proximal to the ATG start codon (between –49 and –2574) was cloned into the pGL3-Basic promoter-detection vector to construct the IAP-luciferase reporter plasmid pIAP-2574/-49 (see Materials and Methods). Cells were cotransfected with pIAP-2574/-49 and the control plasmid pRL-CMV, and treated with T3 (100 nM) for the specified period of time. Cells were then lysed, relative luciferase activity determined, and fold-activation (T3+/T3–) calculated. The results were obtained from six independent experiments, and the values are expressed as mean ± SD (P < 0.05).

T₃ Activates Endogenous IAP Gene Expression
Northern analysis shows that T₃ induces the expression of two IAP transcripts of 2.7 and 3.0 kb in Caco-2 cells (Fig. 1C). Interestingly, Henthorn et al. (36) identified only one IAP transcript in Caco-2 cells using S1 mapping. The T₃ effects are seen as early as 6 h, and maximal levels are seen at 24 h. Densitometric quantitation of bands of Northern autoradigraphs showed that each IAP transcript was increased approximately 3-fold after 24 h treatment with T₃ (100 nM). These in vitro results are consistent with our previous data from the intact rat intestine (18, 21) that also showed T₃-induced activation of the IAP gene. Hence, these in vitro and in vivo results clearly establish the IAP gene as a T₃-responsive target gene.

T₃ Activates the IAP-Luciferase Reporter Gene Expression
We constructed an IAP-luciferase reporter plasmid (pIAP-2574/-49) carrying the 2.5 kb 5' flanking region of the human IAP gene proximal to its ATG start codon (see Materials and Methods). To corroborate the Northern data on expression of the IAP gene (Fig. 1C), we determined the temporal effect of T₃-induced activation of the IAP-luciferase reporter gene in Caco-2 cells. We transiently transfected Caco-2 cells with pIAP-2574/-49, and determined luciferase activities after different time periods. After 24-h treatment of the cells with T₃ (100 nM), the IAP promoter in pIAP-2574/-49 plasmid showed about 3-fold activation by the endogenous TRs (Fig. 1D). Minimal effects of T₃ were seen as early as 6 h. These results of transfection studies are consistent with that observed by Northern analyses.

The TRE Is Located Approximately at –620 of the 5' Regulatory Region of the IAP Gene
We constructed various derivatives of pIAP-2574/-49 carrying sequential 5' deletions of the IAP promoter (see Materials and Methods). Luciferase activities in Caco-2 cells were determined by transiently transfecting the cells with these pIAP-luciferase reporter plasmids in the presence of T₃ (Fig. 2A). T₃ induces the IAP gene about 3-fold in the larger IAP reporter plasmids, but there is a major decrease in activity in the case of the pIAP-616/-49, in which the level of activity was equal to that seen in the case of the vector pGL3-Basic (carrying no insert). Shorter plasmids also showed only basal level of T₃ induction, thereby localizing the TRE in between –750 and –616, a 135-bp region (Fig. 2A). We then constructed several internal deletion mutants of the pIAP-2574/-49, focusing on this 135 bp region (Fig. 2B). Comparison of the activations of pIAP-2574m1870 and pIAP-2574m1920 indicates that the TRE is localized in the sequence between –661 and –618. Comparison of the sequence between –661 and –618 with the established consensus TRE sequence shows that the sequence around –620, between –631 and –613 (5'-TGAACTCAgccTGAGGTTA-3'), is the likely TRE. Compared with the consensus octamer TRE sequence (T(A/G)AGGTCA) (34, 35), this putative TRE sequence appears to consist of an everted repeat of two octamer half-sites that have approximately 88% homology with the consensus TRE sequence. Most of the TREs identified until now consist of a direct repeat of two hexamers and/or octamers separated by four nonspecific bases (DR4-TRE) (24, 25, 26, 27). Based on the model of DR4-TRE, the putative IAP-TRE could also be assigned as a direct repeat of two octamer half-sites that have approximately 80% homology with the consensus octamer TRE sequence. However, this sequence has only three nucleotides (instead of the expected four) separating its octamer half-sites, indicating that this may be an atypical TRE. We constructed an internal deletion mutant pIAP-2574TRE, specifically deleting the putative TRE from the plasmid pIAP-2574/-49 (Fig. 2B). This deletion of the sequence between –631 and –613 completely eliminated T₃-induced activation of the IAP-luciferase gene, indicating that this is, indeed, the only biologically active TRE in the human IAP gene within its 2.5-kb proximal promoter region (Fig. 2B).

View larger version (50K): [in this window] [in a new window]   Fig. 2. Endogenous TR-Mediated Regulation of the IAP-Luciferase Reporter Gene in Plasmids Carrying Different 5' and Internal Deletions of the IAP Promoter Luciferase reporter plasmids carrying various IAP promoter regions were used to transfect Caco-2 cells (see Materials and Methods). The control plasmid pRL-CMV was used to cotransfect the cells, which were then treated ± T3 (100 nM). The relative luciferase activity was determined after 24 h, and fold-activation (T3+/T3–) calculated. The results were obtained from six independent experiments, and the values are expressed as mean ± SD (P < 0.05). A, T3 regulation of the IAP-luciferase reporter gene in plasmids carrying different 5' deletions of the IAP promoter. Various 5' deletion mutants (pIAPs) of the pIAP-2574/-49 plasmid were constructed by deleting specific restriction fragments of the IAP promoter (see Materials and Methods). B, T3 regulation of the IAP-luciferase reporter gene in plasmids carrying internal deletions of the IAP promoter. Various pIAP plasmids carrying internal deletions of the IAP promoter were constructed by deleting specific restriction fragments as well as by in vitro site-directed mutagenesis of the pIAP-2574/-49 plasmid (see Materials and Methods). Sc, SacI; Sp, SphI; Xh, XhoI; Pv, PvuII; Pf, PflMI; Be, BstEII; Bu, Bsu36 I; Sm, SmaI; Ml, MluI; Bx, BstXI; Bt, BtrI; Ps, PstI.

View larger version (44K): [in this window] [in a new window]   Fig. 3. Binding of TR1 and RXR to the Putative IAP-TRE Analyzed by EMSA The double-stranded oligonucleotide TRE1G (Table 1) carrying the putative IAP-TRE was 5' end-labeled with 33P, and used as the probe in EMSA. TR1, TRß1, and RXR proteins were synthesized in vitro; and nuclear extract from Caco-2 cells was obtained using established protocols (see Materials and Methods). Related oligonucleotide sequences are shown in Table 1. Protein-DNA complexes were incubated at room temperature for 20 min. Fifty-fold molar excess of the unlabeled (cold) competitor double-stranded oligonucleotide was used in the related binding reaction. The protein-DNA complexes were separated by nondenaturing 5% PAGE at 4 C. EMSA for each experiment was repeated more than three times, and similar results were obtained. The figure shows the photographs of representative autoradiographs. A, Binding of TR1 and RXR to the IAP-TRE. B, Binding of TRß1 and RXR to the IAP-TRE. C, Binding of nuclear proteins of Caco-2 cells to the putative IAP-TRE. G, TRE1G; DR4, DR4 TRE; mTR, mutant IAP-TRE (mTRE, Table 1); 1, anti-TR1 antibody; ß1s, anti-TRß1 antibody (Santa Cruz); ß1u, anti-TRß1 antibody (Upstate); TnT lysate, lysate for transcription and translation in vitro.

Nuclear Extract from Caco-2 Cells Binds to the IAP-TRE
Nuclear extract from Caco-2 cells was incubated with the radiolabeled TRE1G probe, and EMSA was performed. The results are shown in Fig. 3C. Two shifted bands are observed, which proves that components of nuclear extract are capable of binding to the TRE1G oligonucleotide containing the putative TRE (lane 2). The major band was supershifted with the anti-TR1 antibody (lane 6) confirming that Caco-2 cells express endogenous TR1, which is capable of binding to the IAP-TRE and could be recognized by the anti-TR1 antibody. However, the major shifted bands could not be supershifted with anti-TRß1 antibodies (lanes 7 and 8), even though Caco-2 cells express TRß1 as evidenced by RT-PCR (Fig. 1A) and Western blotting (Fig. 1B). The minor shifted band could not be supershifted with the anti-TR1 or anti-TRß1 antibodies suggesting the nonspecific nature of this band. These data concur with the EMSA results obtained by using the in vitro-synthesized TR1 and RXR proteins (Fig. 3A) and suggest that TR1 is the major TR isoform in Caco-2 cells.

T₃ Inhibits the Binding of Anti-TR1 Antibody to the TRE-TR1-RXR Complex
Using EMSA, we determined the effect of T₃ (100 nM) on the binding of the TR1-RXR protein complex to the IAP-TRE. We also determined the effect of T₃ (100 nM) on the binding of the TR antibodies to the protein-DNA complex. The results (Fig. 4A) suggest that T₃ has no effect on the binding of the TR1-RXR protein complex to the IAP-TRE (lanes 2 and 10). In contrast, comparison between lane 6 and lane 14 shows that T₃ inhibits the binding of anti-TR1 antibody to the TRE-TR1-RXR complex. We also determined the effect of T₃ on the binding of the TR1-RXR protein complex to the IAP-TRE at different concentrations of the cold competitor (Fig. 4B). The result shows that the TR1-RXR protein complex has similar binding affinity in the presence or absence of T₃. Taken together, these results suggest that T₃ does not affect the binding affinity of the TR for the IAP-TRE, but likely causes a structural change that alters the ability of the anti-TR1 antibody to bind to TR1.

View larger version (38K): [in this window] [in a new window]   Fig. 4. Effect of T3 on the Binding of TR1-RXR Protein Complex to the IAP-TRE The double-stranded oligonucleotide TRE1G (Table 1) carrying the IAP-TRE was 5' end-labeled with 33P, and used as the probe in EMSA. TR1 and RXR proteins were synthesized in vitro (see Materials and Methods). Related oligonucleotide sequences are shown in Table 1. Fifty-fold molar excess of the unlabeled (cold) competitor double-stranded oligonucleotide was used in the related binding reaction. The protein-DNA complexes were separated by nondenaturing 5% polyacrylamide gel electrophoresis at 4 C. EMSA for each experiment was repeated more than three times, and similar results were obtained. The figure shows the photographs of representative autoradiographs. A, Effect of T3 on the binding of TR1-RXR protein complex to the IAP-TRE. The TRE1G probe was incubated with TR1 and RXR in the presence of T3 (100 nM) at room temperature for 20 min. B, Effect of T3 on the binding of TR1-RXR protein complex with different concentrations of the IAP-TRE. The probe TRE1G was incubated with different amounts (fold-excess) of unlabeled (cold) competitor TRE1G in the presence of T3 (100 nM) and TR1-RXR protein complex at room temperature for 20 min. G, TRE1G; DR4, DR4 TRE; mTR, mutant IAP-TRE (mTRE, Table 1); 1, anti-TR1 antibody; ß1s, anti-TRß1 antibody (Santa Cruz); ß1u, anti-TRß1 antibody (Upstate).

Substitution Mutations in Either Half-Site Abolish the Binding of the TR1-RXR Complex to the IAP-TRE

View larger version (26K): [in this window] [in a new window]   Fig. 5. Binding of TR1 and RXR to the IAP-TREs Carrying Substitution Mutations The double-stranded oligonucleotide TRE1G (Table 1) carrying the putative IAP-TRE was 5' end-labeled with 33P, and used as the probe in EMSA. TR1 and RXR proteins were synthesized in vitro. Sequences of the IAP-TREs carrying substituted nucleotides are shown in Table 1. Protein-DNA complexes were incubated at room temperature for 20 min. Fifty-fold molar excess of the unlabeled (cold) competitor double-stranded oligonucleotide was used in the related binding reaction. The protein-DNA complexes were separated by nondenaturing 5% PAGE at 4 C. EMSA for each experiment was repeated more than three times, and similar results were obtained. The figure shows the photograph of a representative autoradiograph. Legend: M1- M10, TRE1sm1-TRE1sm10 (Table 1); mTR, mutant IAP-TRE (mTRE, Table 1); 1, anti-TR1 antibody; ßs, anti-TRß1 antibody (Santa Cruz); ßu, anti-TRß1 antibody (Upstate).

An Individual Half-Site of the IAP-TRE Does Not Bind to the TR1-RXR Complex

View this table: [in this window] [in a new window]   Table 2. Binding of the TR1-RXR Complex to Oligonucleotides Carrying 5' and 3' Deletions of the Putative IAP-TRE

View larger version (44K): [in this window] [in a new window]   Fig. 6. Binding of TR1 and RXR to the IAP-TREs Carrying Sequential Deletions and Nucleotide-Specific Mutations The double-stranded oligonucleotides (Table 2) carrying the putative wild-type IAP-TRE or its mutant derivatives were 5' end-labeled with 33P, and used as probes in EMSA. TR1 and RXR proteins were synthesized in vitro. Protein-DNA complexes were incubated at room temperature for 20 min. Except where indicated, 50-fold molar excess of the unlabeled (cold) competitor double-stranded oligonucleotide was used in the related binding reaction. The protein-DNA complexes were separated by nondenaturing 5% PAGE at 4 C. EMSA for each experiment was repeated more than three times, and similar results were obtained. The figure shows the photographs of representative autoradiographs. A, Binding of TR1 and RXR to the IAP-TREs carrying sequential deletions. Sequences of the IAP-TREs carrying sequential 5' and 3' deletions are shown in Table 2. The double-stranded oligonucleotide TRE1G (Table 2) carrying the putative wild-type IAP-TRE was used as the probe. B, Binding of TR1 and RXR to the IAP-TREs carrying substitution mutations of the extra nucleotide of the nonamers. Sequences of the IAP-TREs carrying substitution mutations of the extra nucleotides of the nonamers are shown in Table 3. The double-stranded oligonucleotide TRE1G (Table 2) carrying the putative wild-type IAP-TRE was used as the probe. C, Binding of TR1 and RXR to the IAP-TREs carrying nucleotide-specific mutations. Sequences of the IAP-TREs carrying nucleotide-specific mutations are shown in Table 3. The double-stranded oligonucleotides (Table 3) carrying the wild-type or a mutant IAP-TRE were used as probes. D, Binding of TR1 and RXR to the putative IAP-TRE in the presence of an incremental molar excess of a cold competitor. The double-stranded oligonucleotide TRE1G (Table 2) carrying the putative wild-type IAP-TRE was used as the probe. B-Y, TRE1B-TRE1Y (Table 2); G, TRE1G (Table 1); DR4, DR4 TRE (Table 1); mTR, mutant IAP-TRE (mTRE, Table 1); E-Yt, TRE1E-TRE1Yt (Table 3); U1, TREMU1 (Table 3); D9, TREMD9 (Table 3).

The Extra Nucleotides in the Nonamer Half-Sites Are Specific
We investigated the specificity of the two extra nucleotides, one at each end of the core IAP-TRE. We performed EMSA (Fig. 6B) on oligonucleotides carrying all four different types of nucleotides (A, C, G, T) in the extra positions (Table 3). For the 5' end extra nucleotide, we found that the TR1-RXR complex strongly binds to a TRE carrying T or G (Fig. 6B, lanes 9 and 12, respectively; Table 3), whereas the protein complex does not bind to a TRE containing A or C (lanes 10 and 11, respectively). Similarly, for the 3' end extra nucleotide, we found that the TR1-RXR complex strongly binds to a TRE carrying C (lane 14), whereas the protein complex does not bind to a TRE carrying T (lane 17), and a TRE containing A or G shows weak binding (lanes 15 and 16, respectively). The observed specificity of the extra nucleotides strongly suggests the nonameric nature of the half-sites of the IAP-TRE in these truncated forms of oligonucleotides. We then decided to elucidate the specificity of these two nucleotides in the context of the larger oligo TRE1G using specific point mutations at these extra positions (Table 3). We synthesized two mutant derivatives of TRE1G (TREMU1 and TREMD9, Table 3), and used them in EMSA for determining their binding property to TR-RXR complex (Fig. 6C). The results indicate that TR1-RXR protein complex probably binds to these two mutants with less efficiency. However, to determine accurate binding affinity of the TR1-RXR protein complex to these mutant oligos, we compared their competing capability with that of wild-type TRE (TRE1G) for TR1-RXR using incremental doses of cold oligos (Fig. 6D, Table 3). Comparison of intensities of respective bands of wild-type and mutant TREs shows that mutations of the two extra nucleotides reduce the binding affinity of TR1-RXR to the IAP-TRE flanked by four to six nucleotides at both ends. These results define the end nucleotide of the core IAP-TRE and strongly suggest the nonameric nature of this TRE. Similar results were obtained from the binding of the TRß1-RXR heterodimer to the wild-type and these mutant TREs (data not shown).

View this table: [in this window] [in a new window]   Table 3. Binding of the TR1-RXR Complex to Oligonucleotides Carrying Substitution Mutations of the Extra Nucleotides in the Nonamer Half-Sites of the IAP-TRE

Substitution Mutations of Internal Nucleotides of a Nonamer Half-Site Reduce the Binding of TR1-RXR to the IAP-TRE

View larger version (32K): [in this window] [in a new window]   Fig. 7. Binding of TR1 and RXR to the IAP-TREs Carrying Nucleotide-Specific Mutations The double-stranded oligonucleotides carrying the putative wild-type IAP-TRE or its mutant derivatives (Table 3) were 5' end-labeled with 33P, and used as probes in EMSA. TR1 and RXR proteins were synthesized in vitro. Protein-DNA complexes were incubated at room temperature for 20 min. Except where indicated, 50-fold molar excess of the unlabeled (cold) competitor double-stranded oligonucleotide was used in the related binding reaction. The protein-DNA complexes were separated by nondenaturing 5% PAGE at 4 C. EMSA for each experiment was repeated more than three times, and similar results were obtained. The figure shows the photographs of representative autoradiographs. A, Binding of TR1 and RXR to the IAP-TREs carrying nucleotide-specific mutations in the nonamers. Sequences of the IAP-TREs carrying nucleotide-specific mutations are shown in Table 3. The double-stranded oligonucleotides (Table 3) carrying the wild-type or a mutant IAP-TRE were used as probes. B, Binding of TR1 and RXR to the putative IAP-TRE in the presence of an incremental molar excess of a cold competitor carrying nucleotide-specific mutations in the nonamers. The double-stranded oligonucleotide TRE1G (Table 3) carrying the putative wild-type IAP-TRE was used as the probe. C, Binding of TR1 and RXR to the mutant IAP-TRE carrying nucleotide-specific mutations in the spacer region between two nonamers. Sequences of the IAP-TREs carrying nucleotide-specific mutations are shown in Table 3. The double-stranded oligonucleotides (Table 3) carrying the wild-type or a mutant IAP-TRE were used as probes. D, Binding of TR1 and RXR to the putative IAP-TRE in the presence of an incremental molar excess of a cold competitor carrying nucleotide-specific mutations in the spacer region between two nonamers. The double-stranded oligonucleotide TRE1G (Table 3) carrying the putative wild-type IAP-TRE was used as the probe. G, TRE1G (Table 1); mTR, mutant IAP-TRE (mTRE, Table 1); U23–D78, TREMU23–TREMD78 (Table 3); SR123, TREMSR123 (Table 3).

Mutations of the Spacer Region between the Two Nonamer Half-Sites Do Not Affect the Binding of TR1-RXR to the IAP-TRE

Substitution Mutations in Either Nonamer Greatly Reduces Biological Activity of the IAP-TRE
To assess the biological importance of each nonamer we performed site-directed mutagenesis to construct pIAP-luciferase reporter plasmids carrying substitution mutations for each nonamer (Table 4). Caco-2 cells were transfected with the mutant plasmids, and luciferase activity was determined in the presence of T₃ (100 nM). The results are shown in Fig. 8A and tabulated in Table 4. All mutations rendered the IAP-TRE biologically inactive.

View larger version (16K): [in this window] [in a new window]   Fig. 8. Endogenous TR-Mediated Regulation of the IAPLuciferase Reporter Gene in Plasmids Carrying Nucleotide-Specific Mutations in the IAP-TRE IAP-luciferase reporter plasmids carrying nucleotidespecific mutations in the IAP-TRE were constructed by in vitro site-directed mutagenesis, and the corresponding mutations are shown in Table 4. A mutant plasmid was used to cotransfect Caco-2 cells along with the control plasmid pRL-CMV, and the cells were then treated ±T3 (100 nM) for 24 h. The relative luciferase activity was determined, and fold-activation (T3+/T3–) calculated (see Materials and Methods). The results were obtained from six independent experiments, and the values are expressed as mean ± SD (P < 0.05). A, Regulation of mutant IAP-TREs by endogenous TRs. B, Ligand-independent regulation of mutant IAP-TREs by endogenous TRs.

The IAP-TRE Is a Novel TRE Consisting of an Everted Repeat of Two Nonamers
We compared the IAP-TRE with the various positively regulated TREs that have been previously described for other genes (37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51) (Table 5). It is evident that TRs recognize DNA sequences with great diversity; however, all of the positively regulated TREs are composed of a repeat of a loosely conserved hexamer (AGGTCA) sequence. Katz and Koenig (35) have identified that the dinucleotide TG or TA at the 5' end of the above-mentioned hexamer constitutes a part of a TRE. Approximately 50% of the identified TREs contain an octamer half site (Table 5). Accordingly, to define the space between two half-sites of a TRE we counted the space between the downstream octamer/hexamer and upstream hexamer/octamer (Table 5). In Fig. 9A, we summarized the results of in vitro binding of the IAP-TRE to the TR-RXR complex (EMSA) and biological (in vivo) effects of mutations of the IAP-TRE in T₃-induced activation of the IAP gene in Caco-2 cells (Fig. 9A). The data show that nucleotides at or close to the ends of each nonamer are required for biological activity as well as for in vitro binding of the TR-RXR complex. Thus, the data clearly establish that the IAP-TRE consists of two nonamers separated by three nonspecific nucleotides. Comparison of the nucleotide sequence of a consensus octamer TRE [T(G/A)AGGTCA] with the corresponding nucleotides of a nonamer in either strand of the IAP-TRE shows that the upstream nonamer in the antisense strand and the downstream nonamer in the sense strand each has two bases mismatched. Similarly, the downstream nonamer in the antisense strand and the upstream nonamer in the sense strand each has one base mismatched. Hence, to obtain maximum homology of the IAP-TRE with the consensus octamer TRE, we assigned the IAP-TRE as an everted repeat of two nonamers (Fig. 9A). Until now, no TRE has been identified containing nonamer half-sites. The IAP-TRE does not bind to TR1 alone; however, it binds to TRß1 alone making it also an atypical TRE. Further, in the presence of RXR, each TR binds to the TRE only as a heterodimer of TR-RXR. In Fig. 9B, we present the sequence of the proximal 800 bases of the IAP promoter, showing the TRE and other transcription factor binding sites.

View larger version (46K): [in this window] [in a new window]   Fig. 9. Nucleotide Sequence Analyses of the Human IAP-TRE and the 5' Flanking Region of the IAP Gene Showing the TRE and Other cis Elements A, Composition of the IAP-TRE. The nonamer half-sites of the IAP-TRE are shown in capital letters, and the nucleotides differing from their respective nucleotides of the consensus octamer (bottom panel) are underlined. The nucleotides involved in binding to TR-RXR protein complex in vitro (EMSA) are shown in bold capital letters (see Table 3 and Figs. 6 and 7). The nucleotides required for wild-type biological (in vivo) activity of the TRE are shown in bold and larger-font capital letters (see Table 4 and Fig. 8). The arrows (not broken) indicate the nonamers showing maximum homology to the consensus octamer. The broken arrows indicate the nonamers with less homology to the consensus octamer. Note: Considering the homology of the IAP-TRE half-sites with that of consensus TRE, it is most probable that the IAP-TRE is an everted repeat of two nonamers separated by three nonspecific nucleotides. B, Sequence of the human IAP gene 5' flanking region showing the IAP-TRE and other cis elements. The sequence (antisense) is numbered relative to the ATG start codon (marked +1). The IAP-TRE is underlined, and the nonamer half-sites are marked bold. The putative TATA-box (pTATA) is marked bold. The Sp1/GKLF binding sites and the relevant restriction sites are marked and underlined. Transcription start-sites, major (>>>) and minor (>), are marked. The initiation codon ATG is marked bold.

View larger version (33K): [in this window] [in a new window]   Fig. 10. Biological Activities of the Wild-Type and Mutant IAP-TREs in Caco-2 Cells Overexpressing an Individual Isoform of TR Cells were transfected with the pGL3-Basic vector, pIAP-2574/-49, or pIAP-2574TRE. Cells were also cotransfected with the control plasmid pRL-CMV along with a plasmid overexpressing an individual isoform of TR (TR1, TRß1, or TRß2). The transfectants were treated ± T3 (100 nM) for 24 h. Cells were then lysed, relative luciferase activity determined, and fold activation (T3+/T3–) calculated. The results were obtained from six independent experiments, and the values are expressed as mean ± SD (P < 0.05).

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

Alkaline phosphatases are membrane-bound glycoproteins that hydrolyze a wide variety of monophosphate esters at high pH optima and are widely distributed in different tissues (36). Depending on their biochemical nature and tissue distribution, the alkaline phosphatases are classified as placental, placental-like, intestinal (fetal and adult), and liver, bone, kidney (L/B/K) alkaline phosphatases. The human intestinal alkaline phosphatase (IAP) gene maps to chromosome 2q34–37 and produces a 528-amino acid polypeptide. The IAP protein has been identified as a content of the surfactant-like particles, the unilamellar secreted membrane associated with the process of lipid absorption and isolated from the apical surface of enterocytes (19). The IAP protein has also been identified as a Hela cell tumor-associated antigen (53). A recent report by Narisawa et al. (20) on IAP knockout mice demonstrated that IAP limits fat absorption during high-fat feeding. The IAP gene is overexpressed in hyperthyroid intestine, especially in the duodenum, thus indicating the T₃-mediated activation of this gene (22). Hypothyroidism is associated with villus atrophy and a decrease in the expression of the 3.0-kb IAP mRNA species, further supporting the involvement of T₃ in the positive regulation of the IAP gene (54).

The molecular mechanisms by which T₃ exerts its effect on the gut are not well understood. There have been few in vitro studies on T₃ action in intestinal cells, one exception being a study by Giannella et al. (55), showing that T₃ induces Na,K-ATPase gene expression in the Caco-2 colon cancer cell line. However, Na,K-ATPase is a ubiquitous trans-membrane protein mostly expressed in kidney, intestine, skeletal muscle, and cardiac muscle. In addition to IAP, lactase, and Na,K-ATPase, it will be of interest to identify the full range of T₃-responsive genes in the gut, similar to what has been done in other tissues such as the pituitary (56). Wu et al. (38) have tabulated the T₃-responsive genes that have been studied at the molecular level identifying their respective TREs; however, the status of expression of these genes in the gut is not known.

Various forms of TR and TRß isoforms are expressed in the gut; however, using RT-PCR we found expression of only the TR1 and TRß1 receptors in the colorectal adenocarcinoma Caco-2 cell line (Fig. 1A). We observed two 0.60 and 0.65 kb extra bands in the case of TRß1, which were produced by nested PCR, indicating the possible existence of other TRß isoforms in these cells. Western blotting on whole-cell lysates, as well as nuclear extract, confirmed the production of TR1 and TRß1 receptors by Caco-2 cells, which corroborates with the RT-PCR data. Based upon the Northern blot analyses, the basal level of TRs in Caco-2 cells was capable of approximately 3-fold activation of the endogenous IAP gene after T₃ treatment (Fig. 1C). Using S1 mapping, Henthorn et al. (15) have shown only one IAP transcript in the Caco-2 cells; however, we observed two transcripts in the Northern blots. This apparent inconsistency could be due to subtype variations among the Caco-2 cells or the relatively lower sensitivity of S1 mapping. Similar to the activation of endogenous IAP mRNA expression, transient transfections in Caco-2 cells with IAP-luciferase reporter constructs also showed approximately 3-fold activation of the IAP gene by T₃ (Figs. 1D and 2A).

Previously identified naturally occurring TREs generally consist of two hexamer half-sites arranged as either DR4, palindrome, inverted palindrome, or everted palindrome (24, 25, 26, 27, 28, 29, 30, 31, 32, 33). The TREs with hexamer half-sites bind only to TR-RXR heterodimer. An octamer TRE [T(A/G)AGGTCA] has been shown to bind to a TR monomer, thus eliminating the necessity of TR-RXR heterodimerization for TR-mediated activity (34, 35). Most previous work on these octamer TREs has been performed in vitro with synthesized oligonucleotides elucidating the precise sequences, spacing, and orientation (34, 35). Recently, Toyoda et al. (40) have described the TRE1 from the human type 1 deiodinase gene (hdio1) as a naturally occurring, biologically functional TRE consisting of two relatively independent octamer sequences (DR10) that do not require the RXR family of proteins for function.

Using transient transfection assays with IAP-luciferase reporter constructs carrying various 5' and internal deletion mutants of the IAP promoter (Fig. 2), we were able to convincingly localize the TRE to a 135-bp region in the proximal IAP promoter around –620. We compared the sequence of this 135-bp region with consensus DR4 and octamer TREs (24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35). We identified a putative IAP-TRE (5'-TGAACTCAgccTGAGGTTA-3') consisting of two octamer half-sites separated by three nucleotides instead of the expected four nucleotides. With respect to the spacing of three nucleotides between octamer half-sites, the IAP-TRE resembles the rat prestin TRE (Table 5). The TRE for rat uncoupling protein 1 consists of a downstream hexamer and upstream octamer separated by three nucleotides (Table 5). In addition, the pTRE for human thyroid hormone receptor ß1 consists of a direct repeat of two hexamers separated by three nucleotides, and the dTRE consists of an everted repeat of a downstream octamer and an upstream hexamer (Table 5). By internal deletion mutation of the identified TRE, we showed that this TRE was, indeed, the only biologically active TRE in the 2.5-kb proximal promoter region of the IAP gene. However, our result does not exclude the possibility of another TRE outside the region of the IAP promoter examined in this work. Most in vivo work on T₃ regulation of the IAP gene has been performed in rodents. However, by sequence comparison, we could not identify a close homology (>70%) of the identified human IAP-TRE within the 2.5-kb proximal promoter region of the mouse IAP gene. It will be interesting to investigate the localization and structure of the mouse IAP-TRE and compare the underlying mechanisms with those found for the human IAP gene.

Toyoda et al. (40) have shown that the hDIO1-TRE, presumably consisting of two independent DR10 octamers (Table 5), preferentially binds to two monomers (double occupancy) of TR1 in the absence of T₃, whereas in the presence of T₃ (100 nM), TR double occupancy is impaired, and binding of a single monomer (single occupancy) is also observed. The authors (40) have also shown that RXR as well as the TR-RXR heterodimer does not bind to the hDIO1-TRE. In contrast to the binding of TR to hDIO1-TRE, our EMSA data (Fig. 3A) show that TR1 cannot independently bind to the IAP-TRE, and only the TR1-RXR complex can bind to the TRE efficiently. However, our EMSA data show that TRß1 alone can bind to the IAP-TRE although with less affinity (Fig. 3B), and such binding is abolished in the presence of RXR, when only the TRß1-RXR heterodimer binds to the IAP-TRE. We also observed that TRß1 alone binds to the downstream half-site of the IAP-TRE (data not shown), and it will be interesting to further investigate such binding of TRß1 that might reveal the nucleotide composition of any likely isoform-specific TRE. Like the hDIO-TRE, the IAP-TRE also does not bind to RXR. In contrast to the effect of T₃ on the hDIO1-TRE, our results (Fig. 4) show that T₃ has no effect on binding of the TR1-RXR complex to the IAP-TRE. However, we found that T₃ inhibits the binding of the anti-TR1 antibody to TRE-TR1-RXR complex. We presume that binding of T₃ to TR1 changes the conformation of TR1 thus making it inaccessible to the antibody.

Using EMSA, we compared the binding of in vitro-synthesized TR1-RXR and TRß1-RXR complexes to the IAP-TRE (Fig. 3, A and B) with the binding of nuclear extract from Caco-2 cells to the IAP-TRE (Fig. 3C). The TR1-RXR complex, as well as the nuclear extract, is capable of binding to the IAP-TRE and can be supershifted with anti-TR1 antibody (Fig. 3C). The polyclonal anti-TRß1 antibody from Upstate Biotechnology (Lake Placid, NY) was able to supershift TRE-TRß1-RXR complex, whereas monoclonal anti-TRß1 antibody from Santa Cruz Biotechnology (Santa Cruz, CA) was not able to bind to the TRE-TRß1-RXR complex (Fig. 3B). Although Caco-2 cells express TRß1 (Fig. 1, A and B) we found that anti-TRß1 antibodies do not generate any supershifted band (Fig. 3C). We believe that the amount of TRß1 expressed in Caco-2 cells was probably not sufficient to bind to the TRE in competition with TR1. We also found that T₃ has similar effects on binding of in vitro-synthesized protein complex as well as on binding of nuclear proteins to the IAP-TRE (data not shown).

EMSA results (Fig. 5) on the IAP-TRE containing substitution mutations demonstrated that each half-site of the IAP-TRE is at least an octamer or it could be larger than an octamer. Efficient binding of TR1 alone to an octamer (TAAGGTCA) has been described (34); however, neither octamer in the IAP-TRE can independently bind to the TR1-RXR complex (Fig. 6), thus proving the necessity of both the half-sites for efficient binding to this TRE. Based on the sequence of the antisense strand, the upstream octamer of the IAP-TRE (TGAACTCA) varies by three bases, whereas the downstream octamer (TGAGGTTA) varies by two bases from the above octamer (TAAGGTCA) sequence (Fig. 9A). Similarly, based on the sequence of the sense strand, the octamers of the IAP-TRE (TAACCTCA and TGAGTTCA) each varies by two bases from the above-mentioned octamer (TAAGGTCA) sequence (Fig. 9A). It appears that binding to a single octamer TRE probably requires very stringent homology to TAAGGTCA.

Using EMSA on sequential deletion mutations of a large TRE, we show that the half site of the IAP-TRE is a nonamer (Fig. 6A). We studied the specificity of the extra nucleotides within the nonamers of the IAP-TRE, and we indeed found the specific role of these nucleotides for the binding of the TR1-RXR complex to the IAP-TRE (Fig. 6, B–D). The specificity of the extra nucleotides strongly indicates that the IAP-TRE is a repeat of nonamers. It will be interesting to investigate the biological role of these extra nucleotides in regard to the T₃ activation of the IAP gene. To further determine the nonameric nature of half-sites of the IAP-TRE, we also pe