Isolation of Human Type 2 Deiodinase Gene Promoter and Characterization of a Functional Cyclic Adenosine Monophosphate Response Element^*

Abstract

We analyzed the structure and function of the 5′ flanking region of the human type 2 deiodinase (hD2) gene. Two major transcription start sites were identified at −470/−474 from the ATG. The 5′ flanking region of hD2 gene efficiently directed transcription in transient transfection studies, using luciferase as reporter gene, in HEK 293 cells. Basal transcriptional activity was significantly reduced by deleting the region containing a canonical cAMP-responsive element (CRE) located −766/−759 from ATG. Forskolin treatment significantly increased luciferase activity in cells transfected with CRE-containing constructs. This effect was abolished in constructs that did not contain CRE or contained the mutagenized CRE. Northern blot analysis in JEG-3 cells revealed that the hD2 messenger RNA was markedly increased after stimulation with cAMP agonist. The electrophoretic mobility shift assay with hD2-CRE probe and HEK 293 nuclear extract showed the occurrence of a DNA-protein complex, which was competed by specific unlabeled oligonucleotides and supershifted by the anti-CREB and anti-CRE modulator-1 antibodies. A-CREB, a dominant negative inhibitor of CREB, completely inhibited forskolin induction of the hD2 promoter. CREB protein, once cotransfected with hD2 promoter construct and pKA in F9 teratocarcinoma cells, which are unresponsive to cAMP, was able to stimulate the hD2 gene transcription.

These results indicate the existence of a functional promoter within the 5′ flanking region of hD2 gene which is characterized by the presence of a CRE. The specific involvement of CREB in the cAMP-mediated hD2 gene promoter induction also has been demonstrated.

DEIODINASES ARE A family of enzymes that catalyze the intracellular deiodination of the prohormone T₄ to the active T₃ or to the inactive rT₃ (1, 2), indicating that these enzymes act as gate keepers to thyroid hormone action (3). Their tissue distribution and enzymatic activity are highly tissue-specific, and they contribute to maintain the thyroid homeostasis (4). In particular, type 2 deiodinase (5′-D2) catalyzes the activating outer ring T₄ deiodination and modulates the local availability of T₃ in selected tissues and, in humans, may significantly contribute to the circulating pool (4, 5). Its activity has been demonstrated in pituitary, placenta, brain, and brown adipose tissue of the rat (6). In humans 5′-D2 messenger RNA expression and/or activity seems more widespread and, in addition to the above mentioned tissues, its presence has been shown also in skeletal muscle, heart, thyroid, and skin (6, 7), suggesting a more prominent physiological role for this enzyme. The molecular characterization of human 5′-D2 (hD2) gene started from the isolation of a 1.9-kb brain complementary (cDNA) partial clone containing the open reading frame of the gene (7). The subsequent identification of expressed sequence tagged clones containing the 3′ untranslated (UT) region of human 5′-D2 cDNA supplied the whole cDNA sequence (5). A functional selenocysteine insertion sequence element in the 3′UT region has been identified, confirming hD2 as being a selenoprotein (5). We reported the chromosomal localization of this gene, as well as the genomic structure of the coding region (8), which encompasses two exons divided by a 7.4-kb intron.

The mechanisms that modulate 5′-D2 expression and activity are not entirely clarified. In rat, hypothyroidism is associated with marked elevations of 5′-D2 messenger RNA (mRNA) expression and activity (9, 10). However, no direct evidence of a transcriptional regulation of the 5′-D2 gene by thyroid hormone has been described so far, and the regulation seems to be exerted mainly at pre- or posttranslational levels, at least in rat tissues (6, 11). The modulatory action of the adrenergic system on 5′-D2 expression and activity has also been reported (3, 6, 12). In fact, 5′-D2 mRNA levels increase after the cold exposure and α1 and β adrenergic stimulation in rat brown adipose tissue (6), suggesting that 5′-D2 contributes to the control of thermogenesis in connection with the adrenergic receptor system (3). Furthermore, isoproterenol and (Bu)2AMP stimulation on 5′-D2 expression and activity was also reported in rat pineal gland (12), suggesting the involvement of the cAMP pathway. The evidence that forskolin and 8-bromo-cAMP may induce 5′-D2 mRNA expression in rat astroglial cells strengthens that hypothesis (13). In humans, Salvatore et al. (14) showed an up-regulation of 5′-D2 expression and/or activity in the thyroid of patients with hyperfunctioning thyroid diseases, suggesting that cAMP may regulate also the 5′-D2 mRNA levels in human thyroid gland. However, the promoter region of the human 5′-D2 gene has not been characterized so far, and thus the transcriptional mechanisms regulating this gene may not be properly studied. Here, we analyze the structure and function of the 5′ flanking region of the human 5′-D2 gene.

Materials and Methods

Isolation and computer analysis of the 5′ flanking region of hD2 gene

Human genomic clones have been obtained from a P1 clone library, as previously described (8). A 95-kb P1 clone containing the hD2 gene was digested with HindIII, and the fragments were subcloned in pBluescript KS II (Stratagene, La Jolla, CA). A 6-kb clone containing the 5′ flanking region of the hD2 gene was isolated by colony hybridization of recombinant colonies. A 345-bp PCR product, obtained from the exon 1 and part of the intron of hD2 gene, was labeled with [α-³²P] deoxycycidine triphosphate) by readyprime labeling protocol (Amersham Pharmacia Biotech, Arlington Heights, IL) and used as a probe. Washing conditions were: 2 × SSC, 0.1% SDS, 20 min at room temperature; 1 × SSC, 0.1% SDS, 15 min at 65 C. After washes, filters were exposed to autoradiography. A contiguous upstream 3.8-kb subclone was obtained by digesting P1 clone with BamHI and subcloning the fragments in pBluescript KS II. The recombinant colony was then isolated by colony hybridization using, as probe, an oligonucleotide labeled with [γ-³²P] deoxy-ATP by T₄ polinucleotide kinase (USB), complementary to the overlapping region between the HindIII and BamHI restriction sites. Filters were washed twice at room temperature for 15 min with 2 × SSC, 0.1% SDS and 10 min at 37 C with 2 × SSC, 0.1% SDS. Automatic sequence analysis was then performed on both strands. The sequence of the 5′ flanking region was submitted to the MatInspector software analysis (15) for identification of potential regulatory sequences involved in the hD2 gene transcriptional modulation.

5′ rapid amplification of cDNA ends (RACE)

Total human placenta RNA was isolated with the RNA STAT-60 kit (Tel-Test “B” Inc.), and first-strand tagged cDNA for 5′RACE was performed on 750 ng of RNA with a commercially available Kit (Life Technologies, Inc., Gaithersburg, MD). A 5′AACCAGCTAATCTAGTTTTCTTTCATCTCTTGCTG3′ sequence, localized on exon 2 of the human 5′-D2 gene, was used as gene-specific primer for the RT. The cDNA was then amplified using Advantage-GC Klen-Taq (CLONTECH Laboratories, Inc.) using the following primers: 5′GGCCACGCGTCGACTAGTAC3′ (universal primer) and 5′CTGGTTCCCCTTCACCCTC3′ (antisense gene specific primer, complementary to −54 to −35), designed on the basis of the previously published 1.9-kb brain hD2 cDNA sequence (A.N. Z44095) (7). The PCR conditions were as follows: initial denaturation (2 min, 94 C), followed by 35 cycles of amplification (30 sec at 55 C, 4 min at 72 C, 20 sec at 94 C), whereas, in the last cycle, the elongation time was 10 min. The PCR product was visualized on a 1% agarose gel, blotted, hybridized with an upstream ³²P end-labeled oligonucleotide probe (5′CTTTACCCTTTCATTGTCTCTATG3′, antisense, from −432 to −409) and washed twice at room temperature for 15 min with 2 × SSC, 0.1% SDS and 10 min at 37 C with 2 × SSC, 0.1% SDS. The specific band was gel-separated, electroeluted, and cloned in pCR 2.1, according to the manufacturer’s instructions (Invitrogen). Automatic sequence analysis was then performed on both strands.

Primer extension analysis

Total RNA from a primary culture of human trophoblastic cells was isolated by guanidinium thiocyanate and cesium chloride gradients. The 5′ end of hD2 gene transcript was determined by using a 20mer (5′CCTCTGTCAGAGTCCGTTAA3′) oligonucleotide primer complementary to− 344 to −325 of the hD2 gene and labeled with[γ -³²P] deoxy-ATP and T₄ nucleotide kinase (USB). The primer (1.2× 10⁶ cpm) was hybridized with total RNA (10μ g) at 50 C for 15 min and extended with Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.) for 1 h at 42 C. The resulting product was analyzed on a 6% polyacrylamide −8.3-m urea gel in parallel with a sequencing reaction generated with the extension primer.

Reporter gene constructs

Genomic sequences (from −1274 to +1, from −983 to +1, from− 744 to +1, from −560 to +1, numbered on the basis of their respective distance from the translational start site) were amplified by PCR with pBluescript containing the 6-kb fragment, together with forward and reverse primers which contained a 5′ HindIII adapter sequence. PCR products were cloned into HindIII site of plasmid pSVOALΔ5′, a luciferase expression vector (16). The luciferase constructs were designated as follows: hD2-luc (−1274), hD2-luc (−983), hD2-luc (−744), hD2-luc (−560). The recombinant constructs were sequenced to verify the absence of PCR-induced errors. The 5′ deletion mutant of rat somatostatin promoter Δ(−71), which contains a potent cAMP-responsive element (CRE) (17), was removed with HindIII from CAT reporter plasmid pSRIF Δ(−71)-CAT and cloned in the luciferase reporter vector pSVOALΔ5′. This construct, designated rSS Luc Δ(−71), was the positive control used in the experiments with stimulated cells. Mutagenesis of CRE was performed by digesting the hD2-luc (−1274) construct with Aat II, which cut the sequence GACGT/C. The 3′ protruding ends were removed by using the Klenow enzyme (Roche Molecular Biochemicals, Manza, Italy), and the resulting blunted ends were ligated. Four nucleotides (ACGT) were, therefore, deleted from the TGACGTCA sequence of CRE, and the construct was indicated as hD2-luc (−1270) mut.

Cell lines and transient transfections

Human embryonic kidney (HEK- 293) cells were maintained in DMEM (EuroClone Ltd.) supplemented with 10% FBS and 1% l-glutamine. The day before, transfection cells were seeded in 6-well plates and grown overnight to 50–60% confluence. For transfection, HEK 293 cells were incubated overnight in OPTIMEM (Life Technologies, Inc.) with 1 μg DNA, 5 μl lipofectamine/well (Life Technologies, Inc.). Cells were transfected with the promoter constructs or the promoterless pSVOALΔ5′ reporter plasmid, used to measure the background signal. To measure the basal promoter activity, the medium containing the DNA/liposomes complexes was replaced the day after transfections, and the cells were incubated for 4 h in DMEM containing 10% FBS. To measure the stimulated promoter activity, the day after transfection, cells were incubated for 4 h in DMEM containing 0.1% BSA and 10μ m forskolin. As forskolin was dissolved in DMSO, equimolar amounts of the vehicle were added to the nonstimulated cells for control purposes. To analyze the effect of a dominant negative inhibitor of cAMP responsive element binding protein (CREB), named A-CREB (18), on the forskolin-stimulation of the hD2 gene promoter activity, 1 μg of the hD2-luc (−1274) construct was cotransfected with 1 μg of the expression construct Zeo A-CREB or with an equal amount of the empty expression vector Rc/RSV (Invitrogen). Transfections and forskolin treatment were carried out as above.

F9 teratocarcinoma cells were maintained in DMEM supplemented with 10% FBS. The day before transfection, 200,000 cells/well were seeded in 6-well plates and grown overnight. For transfection, F9 cells were incubated overnight in OPTIMEM (Life Technologies, Inc.) with 3 μg DNA, 5 μl lipofectamine/well (Life Technologies, Inc.). One microgram of the promoter constructs hD2-luc (−1274) was cotransfected with 1 μg of the expression vector Rc/RSV pKA and/or with 1 μg of the expression vector Rc/RSV CREB. The empty expression vector Rc/RSV (Invitrogen) has been used to maintain a total of 3 μg of plasmid DNA. The day after transfections, the cells were incubated for 24 h in DMEM containing 10% FBS.

Plasmid pCMVβ-gal, which expresses β-galactosidase under the control of the CMV promoter, was used in all transfections to normalize the luciferase activity.

After transfection procedures, the cells were lysed in 0.5% Triton X-100, 0.25 m Tris (pH 8), and luciferase activity was measured in a Packard luminometer (Bioscan, Washington, DC). Results are expressed as the ratio of luciferase to β-galactosidase activity, relative to the promoterless or the full-length hD2 promoter constructs, and represent the mean ± sd of three independent experiments performed in triplicate.

JEG-3 cell culture, RNA isolation, and Northern blot analysis

Human choriocarcinoma JEG-3 cells were cultured in MEM supplemented with 10% FBS. When the cells reached 70–80% confluence, the medium was replaced with a serum-free, 0.1% BSA containing medium; and 10 μm forskolin was added for the times indicated. Total RNA was isolated with RNeasy mini Kit (QIAGEN), according to the manufacturer’s instructions; and 10 μg were loaded onto a 1.3% formaldehyde agarose gel, separated, and transferred to a nylon membrane (Gene Screne Plus; New Life Science Products, Boston, MA). The probe, a PCR-generated 663-bp hD2 cDNA fragment, spanning from codon 53 to 274, encompassing most of the coding region, was labeled with[α -³²P]deoxycycidine triphosphate using the random primer labeling kit (Amersham Pharmacia Biotech). After hybridization, the filter was washed twice with 2 × SSC, 0.5% SDS for 15 min at room temperature and twice with 0.2 × SSC, 0.5% SDS for 15 min at 50 C and exposed to x-ray film. The same filter was hybridized with human β-actin cDNA probe as a control.

Electrophoretic mobility shift assay (EMSA)

HEK 293 cell nuclear extracts were prepared as described by Granelli-Piperno et al. (19). Binding reactions mixture contained 0.6 ng ³²P-labeled oligonucleotide probe, 15 μg nuclear extract, 2 μg poly (dI-dC), 200 mm KCl, 75 mm HEPES (pH 7.9), 5 mm EDTA, 2.5 mm dithiothreitol, and 25% glycerol, in a total vol of 25 μl. To perform supershift experiments, 2 μg of the following antibodies were added to the binding reaction mixture: 1) an antibody, indicated asα DBDD, which recognizes an epitope within the specific DNA binding and dimerization domain, reactive with all three members of activating transcription factor (ATF)/CREB transcription factors family (Santa Cruz Biotechnology, Inc., Santa Cruz, CA); 2) the anti-CREB-1 (αCREB-1) (New England Biolabs, Inc., Beverly, MA), anti-CRE modulator-1 (αCREM-1), anti-ATF-1 (αATF-1) antibodies (Cruz Biotechnology, Inc.); 3) an antiestrogen receptor antibody used as nonspecific antibody (Cruz Biotechnology, Inc.). The effect of increasing amounts (4 and 6 μg) of anti-CREB-1 antibody has also been analyzed. In competition experiments, 100-fold molar excess of cold oligonucleotides was added to the binding reaction mixture. The oligonucleotide probe (hD2 CRE) used was: 5′-CTCTTTCTCAATGACGTCAAGA-TCTTTACCAAG-3′ (−777 to −745). The CRE consensus oligonucleotide was the same as that of Promega Corp. The mutated oligonucleotide (hD2 CRE mut) was 5′-CTCTTTCTCAATGCAAGATCTTTACCA-AG-3′ and contained the same mutation as the reporter construct used in the transfection experiments. Reaction mixtures were incubated for 30 min at room temperature, resolved on nondenaturing 4% polyacrylamide gel, dried, and exposed to autoradiography.

Results

Isolation and sequencing of the 5′ flanking region of hD2 gene

Subcloning of the human P1 clone, containing the type 2 5′ deiodinase gene, led to the isolation of a 6-kb clone that contained 1.3 kb, upstream from the translation start site of the gene. A further upstream 3.8-kb subclone, overlapping the previous one for only about 240 bp, was isolated and sequenced. Computerized sequence analysis revealed the presence of a canonical TGACGTCA CRE located from −766 to− 759 from ATG. Some activator protein 1 (AP1) binding sites, a ternary complex factor 1 (TCF-1) and a pituitary-specific transcription factor (Pit-1) site, were also detected. A TTTAAAA motif, usually described as a modified TATA box associated to functional CREs (19), has been also identified in position −489 to −483 from the ATG (Fig. 1).

Figure 1.

Sequence analysis of the 5′ flanking region of the human type 2 deiodinase gene. Numbers are relative to the adenosine of the ATG. The putative cis-acting regulatory elements are underlined. The TTTAAAA motif is indicated by an open box. The transcriptional start sites (see below) are indicated by arrows.

Genomic organization of 5′UT region and mapping of the transcription start site (TSS)

The 5′ RACE analysis has been used to analyze the structure of the 5′ UT region of the gene. Southern blot of 5′ RACE products showed the presence of a single band consisting of about 130 bp (Fig. 2A), which was cloned and sequenced. The result was confirmed by RT-PCR (not shown). The sequence analysis of the product, once compared with the genomic one, revealed the presence of a 274-bp intron in the 5′ UT region of the gene. That intron divided the described exon 1 (8) from an upstream small exon that we called 1a. (Fig. 2B). Therefore, the 5′ end of the RACE product corresponded to position −438 from the translation start site on the genomic sequence.

Figure 2.

5′ RACE. A, Total RNA obtained from human placenta was reverse transcribed and the cDNA was amplified using an universal and a gene-specific primer. The gel was blotted and hybridized with a specific radiolabeled primer. B, Genomic sequence corresponding to the 5′ UT region. Arrowhead indicates the 5′ end of the RACE product. Intron sequence is shown in lowercase. ATG is underlined.

To precisely locate the TSS, a radiolabeled primer, complementary to nucleotides −344 to −325 on exon 1a, was used for primer extension analysis. Two strong signals were detected, indicating the presence of two major transcriptional start sites located at −470 and −474 from the ATG (Fig. 3A). The genomic organization of the 5′ UT region of hD2 gene is depicted in Fig. 3B.

Figure 3.

Primer extension analysis. A, RNA derived from a primary culture of human trophoblastic cells was hybridized with a specific extension primer, and RT products were analyzed on a 6% sequencing gel along with a sequencing reaction. The positions of the transcriptional start sites are indicated by arrows. B, Organization of 5′ UT region of hD2 gene in placenta. Exons 1 and 1a are shown as boxes. Open boxes, 5′ UT region; filled box, coding region. The primer used for primer extension analysis (E.P.) is also indicated. C, Cytosine. T, Thymine. A, Adenine. G, Guanine.

Functional characterization of 5′ type 2 deiodinase gene promoter

To determine the ability of the full-length hD2-Luc (−1274) and its deletion mutant constructs to regulate transcriptional activity under basal conditions, they were transfected into HEK 293 cells. This cell line is highly transfectable and very responsive to cAMP agonist and thus routinely used in such studies. Compared with the promoterless control plasmid, a 35-fold higher luciferase activity was measured in cells transfected with hD2-Luc (−1274) construct, which implies that the DNA sequences upstream from the hD2 gene coding region may functionally direct transcription (Fig. 4). Deletion of 291-bp [hD2-luc (−983) construct] slightly increased luciferase activity, which was instead clearly reduced by deleting further 239 bp, as in hD2-luc (−744) construct, which did not contain the CRE. Luciferase activity was similarly reduced in cells transfected with the smaller construct[ hD2-luc (−560)], which, however, retained some significant promoter activity, compared with the promoterless vector. The effect of forskolin, a direct activator of adenylate cyclase, has been analyzed to examine cAMP responsiveness (Fig. 5). The 5′ deletion mutant of rat somatostatin promoter, designated rSS-lucΔ (−71) construct, which contains a potent CRE (17), has been used as reference for the forskolin effect. Forskolin treatment of cells transfected with rSS-luc Δ(− 71) construct significantly increased luciferase activity (3-fold), compared with the nonstimulated cells transfected with the same construct. In cells transfected with the hD2-Luc (−1274) and hD2-luc (−983) constructs, forskolin stimulated luciferase activity essentially to the same extent as in those transfected with rSS-luc Δ(− 71) construct. This effect was abolished in the deletion mutants hD2-luc (−744) and hD2-luc (−560), which did not include the CRE. The treatment with forskolin also failed to stimulate the cells transfected with hD2-luc (−1270) mut construct, which contained the mutagenized CRE (Fig. 5). The same results were obtained by transfecting the different constructs into JEG-3 cells (data not shown), a placenta-derived cell line in which hD2 gene is cAMP-inducible (see below).

Figure 4.

Promoter constructs and functional analysis of the hD2 promoter. Genomic sequences were amplified by PCR and cloned into HindIII site of plasmid pSVOALΔ5′. One microgram of each promoter construct and 0.1 μg of the pCMV β-gal plasmid were cotransfected overnight, in HEK 293 cells, by lipofection. The liposome-DNA complexes were then replaced with serum-containing medium. After 4 h, cells were lysed; luciferase and β-galactosidase activity was then measured. Results are expressed as light units, relative to the activity of promoterless vector, and are the mean± sd of at least three experiments, each performed in triplicate. LUC, Luciferase reporter gene; RLU, relative light units.

Figure 5.

Effect of forskolin treatment. The hD2 gene promoter constructs (deletion mutants identified by their relative distance from ATG and the full length containing the mutated CRE) and the rat somatostatin promoter construct, used as a positive control, were transfected into HEK 293 cells, as described above. The day after transfection, the liposome-DNA complexes were replaced with a 0.1% BSA-containing medium, and 10 μm forskolin were added. After 4 h, cells were lysed, and luciferase andβ -galactosidase activity was measured. Results are expressed as light units, relative to the activity of promoterless vector, and are the mean ± sd of at least three experiments, each performed in triplicate.

Northern blotting

To demonstrate the cAMP responsiveness of hD2 gene in a cell line physiologically responsive to cAMP, the expression of hD2 gene was analyzed in JEG-3 cells, a human choriocarcinoma cell line that possesses LH/HCG receptors (see Ref. 20). Human D2 gene expression was not detectable before stimulation with forskolin. A band of approximately 7.5 kb, corresponding to the full-length transcript previously described (5, 7), was detected after 1–4 h exposure to forskolin, with the maximum intensity at 2 h (Fig. 6). Similarly to results reported from Salvatore et al. (7), two additional bands, possibly representing smaller transcripts, were also seen.

Figure 6.

Northern analysis of the effect of forskolin on hD2 mRNA expression in JEG-3 cells. Total RNA (10 μg) from forskolin (10 μm)-treated (1, 2, and 4 h) or untreated (0 h) JEG-3 cells was run on a formaldehyde agarose gel and transferred to a nylon membrane. Blot was probed with ³²P-labeled 663-bp PCR-generated cDNA probe, encompassing most of the coding region of hD2 gene. The same filter was hybridized with a human β-actin probe as a control.

Mobility shift assay for analysis of CREB/DNA binding

To examine DNA-protein interaction, an EMSA was performed. After incubation with HEK 293 nuclear extract, the oligonucleotide was specifically shifted, indicating that interaction between the radiolabeled hD2 CRE probe and the nuclear extracts occurred (Fig. 7, lane 2). The pattern of migration of hD2 probe was identical to the one observed with the consensus CRE probe (not shown). The specific DNA-protein complex was partially supershifted by all amounts (2, 4, and 6 μg) of αCREB-1 used, therefore demonstrating that CREB actually binds DNA in vitro (Fig. 7, lanes 3–5). A partial supershift was also obtained by using the αCREM-1 (Fig. 7, lane 7), whereas the anti-ATF-1 (Fig. 7, lane 6), as well as the antiestrogen receptor antibody (Fig. 7, lane 9), did not modify the electrophoretic mobility of the DNA-protein complex. The antibody αDBDD, reactive with all three members of ATF/CREB transcription factors family, completely supershifted the electrophoretic band (Fig. 7, lane 8). The band corresponding to DNA-protein complex disappeared upon competition with a 100-fold molar excess of cold hD2 CRE or consensus CRE oligonucleotides (Fig. 7, lanes 10 and 11). In contrast, no competition occurred using the mutated oligonucleotide (hD2 CRE mut) under the same experimental conditions.

Figure 7.

EMSA. Binding reaction mixture contained 0.6 ng of consensus CRE or hD2 CRE-labeled probe in a total vol of 25 μl. HEK 293 cell nuclear extracts (15 μg) were added where indicated. Lane 1, no extracts; lane 2, nuclear extract with no addition. Two μg ofα CREB-1 (lane 3), αATF-1 (lane 6), αCREM-1 (lane 7), and αDBDD (lane 8) antibodies were added to the binding reaction mixture; the effect of increasing amounts (4 and 6 μg) of αCREB-1 antibody has been also analyzed (lanes 4 and 5). An antiestrogen receptor antibody has been used as nonspecific antibody (lane 9). In competition experiments, 100-fold molar excess cold oligonucleotides (consensus CRE, hD2 CRE, and hD2 CREmut) were added (lanes 10–12). Reaction mixtures were incubated 30 min at room temperature, resolved on nondenaturing 4% polyacrylamide gel, dried, and exposed to autoradiography. *, Nonspecific bands. αER, Antibody antiestrogen receptor.

Effect of a dominant negative inhibitor of CREB (A-CREB)

A-CREB is a novel dominant negative inhibitor of CREB in which the basic region of CREB was replaced by an acidic amphipatic extension (18). A-CREB prevents CREB homodimerization by interacting with CREB basic region and forming a coiled-coil extension of the leucine zipper (18). The resulting heterodimer CREB/A-CREB does not bind DNA and blocks the CREB-dependent transcriptional activation (18). To characterize the transcription factors that actually bind in vivo to the CRE found in the hD2 promoter, we cotransfected HEK 293 cells with hD2-luc (−1274) and Zeo A-CREB expression constructs. In control experiments, we cotransfected the hD2-luc (−1274) promoter construct with an empty expression vector. After transfection, cells were stimulated with forskolin for 4 h. As shown in Fig. 8, A-CREB completely inhibited forskolin induction of the hD2 luc (−1274) construct.

Figure 8.

Effect of dominant negative inhibitor of CREB (A-CREB). One microgram of the hD2-luc (−1274) construct was cotransfected in HEK 293 cells with 1 μg of the Zeo A-CREB plasmid or with an equal amount of the empty expression vector Rc/RSV (Invitrogen) and stimulated with forskolin, as described above. Results are expressed as light units, relative to the nonstimulated hD2-Luc (−1274) construct, and are the mean ± sd of at least three experiments, each performed in triplicate.

Effect of CREB and pKA cotransfection in F9 teratocarcinoma cells

To determine whether CREB is sufficient to stimulate hD2 gene promoter activity in response to cAMP, we transfected the hD2 luc (−1274) construct alone or with an expression vector of the C-subunit of pKA (Rc/RSV pKA) or/and an expression vector of CREB (Rc/RSV CREB) in undifferentiated F9 cells, which do not respond to cAMP (21). A 2- to 3-fold induction of promoter activity was obtained by individually cotransfecting pKA or CREB expression vector with hD2 promoter construct (Fig. 9). Cotransfection of both pKA and CREB expression plasmid caused a huge (43-fold) increase of hD2 luc (−1274) activity, indicating that phosphorylated CREB is able to stimulate hD2 gene transcription.

Figure 9.

Effect of CREB and pKA on hD2 luc (−1274) expression in undifferentiated F9 teratocarcinoma cells. Cells were transfected overnight with the hD2 luc (−1274) reporter gene plus pKA (C subunit), CREB, or both expression plasmids (pKA plus CREB) in F9 cells by lipofection. The liposome-DNA complexes were then replaced with serum-containing medium. After 24 h, cells were lysed; luciferase and β-galactosidase activity was then measured. Results are expressed as light units, relative to the activity of hD2 luc (−1274) construct alone, and are the mean ± sd of at least three experiments, each performed in triplicate.

Discussion

We report here the isolation and characterization of the 5′ flanking region of the human 5′-D2 gene and its functional properties, as well as the relevant role played by the CRE/CREB interaction and the cAMP pathway in the promoter activity.

We have previously described the coding region of human 5′-D2 gene encompassing two exons divided by a 7.4-kb intron (8). In this study on human placenta, the 5′ RACE analysis revealed the presence, in the 5′ UT region of this gene, of an additional small exon. We have evidence for two major TSSs located at −470/−474 from the translational start site, as indicated by primer extension analysis.

A canonical CRE, upstream from a TTTAAAA motif has been identified. It is worth noting that this atypical TATA consensus element has been previously described as being associated with a number of functional CRE (22) and characterizes the so-called Goldberg-Hogness promoters. However, this TTTAAAA motif is located closer than expected for a functional TATA box (22). The actual functionality of this element remains to be elucidated.

In this study, we demonstrated that the CRE located in the promoter region of hD2 gene is functional, based on the following: 1) the transient transfection studies under basal conditions revealed that the promoter constructs owned significant transcriptional activity if they contained CRE and a clearly reduced one if they were devoid of that element; 2) the effect of forskolin, which increases cAMP levels by adenylate cyclase stimulation, was completely abolished in cells transfected with hD2 promoter constructs where CRE was absent or specifically mutagenized; and 3) forskolin stimulation of HEK 293 cells transfected with the hD2 promoter constructs that contained the CRE was comparable with that obtained using the rat somatostatin promoter construct, which includes a well-characterized and functional CRE (17). This evidence also indicates that the CRE is critical to cAMP-induced hD2 gene expression (22). Several transcription factors interacting with CRE to couple the cAMP-mediated signal transduction with gene transcription have been described (see Ref. 23 for review). Some of them activate transcription (CREB, ATF-1, CREM-1), some other act as a repressor (CREM isoforms α,β,γ, CREB-2 gene members) (23). In this study, the results of the EMSA and supershift assay clearly indicated that CREB and CREM-1 proteins, as homodimers and/or heterodimers, are indeed bound to CRE in vitro. The effect of the dominant negative inhibitor of CREB (18), A-CREB, which abolished the forskolin-induced transcription of the reporter gene in the hD2-luc (−1274) construct, confirmed the functional role of CREB or its closely related family members in vivo (22, 23). The role of CREB was further confirmed in human undifferentiated F9 teratocarcinoma cells, which are not responsive to cAMP (21). In these cells, once cotransfected the full-length hD2 promoter construct with pKA plus CREB expression vectors, it was evident that phosphorylated CREB is able to stimulate hD2 gene transcription. The described mechanism of transcriptional activation has been proven to be effective on hD2 natural promoter in a placenta-derived cell line (JEG-3) responsive to cAMP agonists (20), thus indicating that, in this tissue, hormonal and nonhormonal signals, acting through the cAMP pathway, are able to modulate hD2 gene expression. These findings may represent the molecular basis of the adrenergic effects on 5′-D2 expression and activity (12, 13, 24) and their relationship with cAMP pathway described in rat (13, 25). In humans, Salvatore et al. (14) reported an increased hD2 mRNA expression and activity in hyperfunctioning thyroid tissues, suggesting a linkage with TSH receptor status. The stimulatory effect of TSH receptor on thyroid economy is mediated by the cAMP pathway through protein kinase A activation (26). The evidence that forskolin induced hD2 gene promoter via CREB is in keeping with these findings and may represent one of the missed links claimed in that previous report (14).

Other regulatory elements (e.g. thyroid response elements) may be involved in modulating the expression of hD2 gene, but evidence of a transcriptional regulation of 5′-D2 gene through these other pathways is, up to now, faint. In rat tissues, although 5′-D2 expression and activity are greatly affected by the iodothyronine levels, most of the effects seem, in fact, to be regulated via pretranslational and/or posttranslational mechanisms (4, 6, 11, 27). Furthermore, in our study, the computer analysis did not detect the presence of canonical thyroid response elements in the 5′ flanking region. These elements, described instead in the promoter region of human type 1 deiodinase (28, 29), are required to mediate the transcriptional effects of thyroid hormones. The hypothesis that iodothyronines predominantly regulate 5′-D2 activity at a posttranscriptional level is also in keeping with the presence of AU-rich motifs in the 3′ UT region of hD2 cDNA (5); these elements are usually characteristic of genes whose expression is regulated by shortening transcripts half-life through rapid mRNA deadenylation (30).

Finally, the different expression of 5′-D2 gene in human and in rat thyroid (14), and even in different human tissues (4, 6), indicates that this gene is regulated in a species- and tissue-specific fashion. This implies that some other transcription factor, beside the CREB, may be essential in the modulation of hD2 gene. In this view, this first characterization of its promoter region may allow us to further understand these issues.

In summary, a functional promoter has been described within the 5′ flanking region of human 5′-D2 gene which is characterized by the presence of a CRE. The specific involvement of CREB and/or one of its closely related family members in the cAMP-mediated hD2 gene promoter induction also has been demonstrated. These data provide new information to promote further understanding of the molecular mechanisms involved in 5′-D2 expression in humans.

Acknowledgments

We thank Dr. Charles Vinson for the gift of Zeo A-CREB expression construct and Dr. Richard Goodman for the gift of pSRIFΔ (−71)-CAT reporter plasmid, Rc/RSV pKA, and Rc/RSV CREB expression vectors.

^*

This work has been supported by a grant from Ministero dell’Uni-versità e della Ricerca Scientifica e Tecnologica of Italy and by the Institute of Experimental Medicine, CNR. The financial support of Telethon-Italy (Grant E.763) is also gratefully acknowledged.

^†

Dr. Canettieri and Dr. Celi equally contributed to this work.