|
|
||||||||
Departamento de Genética del Desarrollo y Fisiología Molecular, Instituto de Biotecnología, Universidad Nacional Autónoma de México, AP 510-3, Cuernavaca, Mor. 62250, México
(Requests for offprints should be addressed to P Joseph-Bravo; Email: joseph{at}ibt.unam.mx)
* (A Cote-Vélez and L Pérez-Marínez contributed equally to this work)
| Abstract |
|---|
|
|
|---|
| Introduction |
|---|
|
|
|---|
At the transcriptional level, the negative feedback of thyroid hormones is well characterized (Segerson et al. 1987b, Hollenberg et al. 1995, Satoh et al. 1996, Guissouma et al. 2000, Abel et al. 2001) and the consensus sequences for thyroid hormone receptor (TR) binding identified in the human (h) (Yamada et al. 1990), rat (r) (Lee 1988) and mouse (m) (Abel et al. 2001) TRH genes. The sequence between 547 to +84 of rTRH promoter confers almost full basal transcriptional activity in transfected cells (Balkan et al. 1998); within this sequence, response elements of the transcriptional factors: activator protein 1 (AP-1), cAMP response element (CRE)-binding protein (CREB), and glucocorticoid receptors (GR) are present in the three species. Two DNA elements similar to consensus CRE are localized in the rTRH gene ( 101/ 94 (CRE-2) and 59/ 52 (CRE-1)). CRE-1 (TGACCTCA) (also called site 4) binds CREB (Hollenberg et al. 1995, Wilber & Xu 1998, Harris et al. 2001) and overlaps a thyroid hormone response element (THRE (AGGTCA)), recognized as an important site for thyroid hormone negative feedback in the three species (Hollenberg et al. 1995, Satoh et al. 1996, 1999). The stimulatory role of CREB on TRH gene transcription has been evidenced in cells transfected with constructs including human or murine TRH promoters (Wilber & Xu 1998, Harris et al. 2001); 8-h incubation with
-melanocyte stimulating hormone (
MSH; which increases phosphorylated CREB (P-CREB) (Sakar et al. 2002)) enhances TRH transcription (Harris et al. 2001). The stimulatory effect of
MSH is reduced in the presence of 3,5,3'-triiodothyronine (T3) and transfected thyroid hormone receptor TRß2 (Harris et al. 2001).
The effects of glucocorticoids on TRH gene expression are less well understood. A stimulatory effect of long-term treatment (days) with dexamethasone (dex) on proTRH mRNA levels was first described in CA77 cells (Tavianini et al. 1989), and later in rat anterior pituitary or in diencephalic primary cultures (Bruhn et al. 1994, Luo et al. 1995). Primary cultures of hypothalamic cells have a time- and dose-dependent response to dex: a rapid increase of proTRH mRNA levels (13 h) occurs at 106 to 108 M while levels are diminished at 1010 M or after 24-h incubation at 108 M (Joseph-Bravo et al. 1998, Pérez-Martínez et al. 1998). In vivo, glucocorticoids have an inhibitory influence (Kakucska et al. 1995) suggested to be due to their inhibitory effect on CREB phosphorylation in the PVN (Legradi et al. 1997a); upregulation of P-CREB levels follows glucocorticoid depletion (Whitehead & Carter 1997a). Rat TRH promoter lacks a canonical glucocorticoid response element (GRE) but has several copies of the strongest half-site (TGTTCT) (Dahlman-Wright et al. 1991) at: 735, 560 or 210, and a near-consensus inverted site at 284 (GGTCCAcacTCT TGT), as well as two other inverted half-sites at 275 and 310 (Lee 1988). Site 210/ 205 binds GR (Lee et al. 1996). Hela cells transfected with the plasmid containing the 242/+84 sequence of rTRH promoter, or with deletions of a cis-acting element located between 242/ 200, proved that this region is important for basal transcription and glucocorticoid response (Lee et al. 1996). Pancreatic cells transfected with rTRH promoter fragments 554/+84 or 242/+84, treated for 48 h with dex, showed increased transcriptional activity (Fragner et al. 2001). (See Fig. 1
for the position of putative response elements in the rTRH promoter).
|
| Materials and methods |
|---|
|
|
|---|
Wistar rats raised in the local vivarium were kept in a ratio of 12 h light:12 h darkness and were allowed to feed ad libitum. Care was taken to preserve adequate conditions following the FRAME guidelines and as approved by the institutes ethical commission.
Reagents
Poly-D-lysine, dex, 8Br-cAMP, TPA, the PKA inhibitor H89 and its inactive form H85, dithiothreitol, glucose, glutamine, cytosine arabinofuranoside and DNAse were from Sigma. Poly (dI-dC), dNTPs and luciferase assay kit were from Roche; thyroxine (T4) polynucleotide kinase was from Promega. Dulbeccos modified Eagles medium (DMEM), Hanks medium, trypsin, fetal bovine serum, vitamins and antibioticantimycotic were from GIBCO; guanidinium thiocyanate was from Fluka Steinheim (Sigma) and [32P]ATP from NEN Life Science Products (Boston, MA, USA). Expression vectors for CREB (SV-CREB) (González et al. 1989), AP-1 (c-Jun (RSV-cJun), c-Fos (RSV-cFos)) (Hirai et al. 1990, Schontal et al. 1988) and GR (CEO-GR) (Meyer et al. 1989) were a kind gift from Dr Gustavo Pedraza-Alva, IBT-UNAM; TRH-Luc plasmid was a gift from Dr Wayne Balkan, University of Miami School of Medicine, Miami, FL, USA. Oligonucleotides with the consensus and mutant CRE and AP-1 sequences and antibodies used in supershift analysis were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Other oligonucleotides described in this work were synthesized at the institute.
Cell culture
Primary cultures of hypothalamic cells were performed as described previously (Joseph-Bravo et al. 2002) using embryos from anesthetized pregnant Wistar rats on the 17th day of gestation. Cells were plated on 1.5 µg/ml poly-D-lysine pre-coated 35 mm plates (2.7 x 106 cells for TRH release studies, electrophoretic mobility shift assays (EMSAs) and transient transfection assays); for proTRH mRNA quantification, 0.6 x 106 cells were plated in 16 mm plates. DMEM was supplemented with 10% fetal calf serum (FCS), 0.25% glucose, 2 mM glutamine, 3.3 µg/ml insulin, 1% antibioticantimycotic and 1% vitamin solution (S-DMEM). Cultures were maintained at 37 °C with 7% CO2, 93% air atmosphere and 90% humidity. On the fourth day in vitro (DIV), cytosine arabinofuranoside (105 M) was added and mantained for 48 h to inhibit cell proliferation; afterwards half the incubation medium was replaced every second day with fresh S-DMEM (Charli et al. 1995). When drugs were dissolved in dimethyl sulfoxide (DMSO) (dex or TPA), an equivalent DMSO concentration was added to controls.
NIH-3T3 cells were maintained in DMEM supplemented with 10% FCS at 37 °C with 7% CO2, 93% air atmosphere, and 90% humidity.
The SH-SY5Y (ATCC CRL-2266) neuroblastoma cell line was kindly donated by Dr Rosa Ma Pardós (Instituto de Fisiología, Benemérita Universidad Autónoma de Puebla, México). Cells were grown in 125 ml flasks to confluency in S-DMEM, then rinsed with Hanks medium, trypsinized (4 min, with 5 ml of trypsin), centrifuged (5 min) and the pellet was resuspended in S-DMEM. For experiments with undifferentiated cells, 7 x 106 cells were seeded in 60 mm plates and incubated for 24 h. Differentiated cells were obtained by seeding 5 x 106 cells in 60 mm plates, incubating them for 24 h and replacing the medium with S-DMEM containing 5% FCS and 1.6 x 109 M TPA; half the medium was replaced every 2 days for 8 days (Renauld & Spengler 2002).
Transient transfection and luciferase assays
Hypothalamic cultures from 12 DIV were transiently transfected using polyethylenimine (Guerra-Crespo et al. 2003). Transfections were carried out in 35 mm dishes with 2.7 x 106 hypothalamic cells, 5 µg of a reporter plasmid containing rat TRH promoter ( 776/+85) linked to the luciferase reporter (TRH-Luc) and 5 µg RSV-ß-gal (Gynheung et al. 1982) used as internal control of transfection efficiency. Cells were cultured for 48 h in S-DMEM and then harvested for luciferase activity (following kit instructions) and ß-galactosidase activity (Lucibello et al. 1990). Luciferase values were normalized to values obtained for ß-galactosidase activity and protein content; data are expressed relative to luciferase activity.
NIH-3T3 cells were plated in 35 mm dishes and grown for 2024 h to a confluency of around 60%. The cells were then transfected with polyethylenimine. Cotransfection was performed with 1 µg transcription factor expression vector(s) together with 5 µg reporter plasmid (TRH-Luc), 3 µg pRSV-ß-gal and PUC18 to a total of 10 µg plasmid DNA per dish. Cells were cultured for 48 h in DMEM with 10% FCS; then treated for 3 h with 8Br-cAMP, TPA or dex and harvested for luciferase activity. Data are expressed as relative luciferase activity with respect to cells transfected only with TRH-Luc. For EMSAs, NIH-3T3 cells were transfected with 1 µg of the expression vector (CREB, AP-1 or GR) together with 3 µg pRSV-ß-gal and PUC18 to a total of 10 µg plasmid DNA per dish.
Determination of TRH content and release by RIA
Hypothalamic primary cells (18 DIV) were incubated with drugs diluted in DMEM at concentrations and times stated in the figures. The medium was collected in a tube and immediately frozen; cells were stored at 20 °C. TRH cell content was extracted as described previously (Joseph-Bravo et al. 2002). It was not possible to measure TRH directly in the culture medium due to interference in the RIA; therefore, the medium was thawed and extracted with C-18 cartridges (Sep-pak, Millipore) activated according to the manufacturers specifications; TRH was eluted with 60% methanol after washing with 2 ml water. The methanol extract was evaporated and the residue was resuspended in 250 µl 0.05 M phosphate buffer, pH 7.5, 0.25% BSA and immunoreactive TRH measured in duplicate by a specific RIA (Charli et al. 1995); Sep-pak-extracted medium included in the standard curve did not interfere with the RIA; recovery of 100 pg TRH (Peninsula, Belmont, CA, USA) added before extraction was > 90% (not shown).
mRNA purification
Total RNA was extracted from frozen cells scraped with 4 M guanidinium thiocyanate, 25 mM sodium citrate, pH 7.0, 0.5% sarcosyl, 0.1 M ß-mercaptoethanol as described previously (Pérez-Martínez et al. 1998).
ProTRH mRNA semi-quantification by RT-PCR
Reaction conditions for proTRH mRNA quantification were as previously reported (Perez-Martinez et al. 1998) except that 11 pmol proTRH and 25 pmol glyceraldehyde-3-phosphate dehydrogenase (G3 PDH) primers were used for co-amplification (30 cycles). Gels were analysed by laser densitometry to calculate the relative amounts of proTRH vs G3 PDH cDNAs.
Nuclear extract preparation
Hypothalamic cells (14 DIV), transfected NIH-3T3 cells or SH-SY5Y neuroblastoma cells were washed with PBS before obtaining nuclear extracts, as described previously (Schreiber et al. 1989) but with the following modifications: cells were lysed by 5 min incubation at 4 °C in hypotonic buffer (10 mM TrisHCl, pH 7.6, 10 mM NaCl, 1.5 mM MgCl2, 0.5 mM EDTA, 1 mM dithiothreitol, 1 µg/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride and 0.6% NP-40). Intact nuclei were washed with lysis buffer and nuclear extracts were obtained by incubating the nuclei in extraction buffer (20 mM TrisHCl, pH 8.0, 450 mM KCl, 0.5 mM EDTA, 1 mM dithiothreitol, 1 µg/ml leupeptin, 5 mM spermidine and 25% glycerol) for 45 min under constant mild agitation at 4 °C. DNA was eliminated by centrifugation for 15 min at 13 000 g. Protein content was determined by the Bradford assay (Bradford 1976).
Electrophoretic mobility shift assay (EMSA)
The double-stranded rat TRH promoter oligonucleotides used in EMSAs were as follows. CRE-L, containing the two potential CRE elements of the TRH promoter ( 101 to 94 and 59 to 52): CRE-1, 59/ 52; CRE-2, 101/ 94. For GRE: GRE-L, 220/ 193; GRE-A, 215/ 193. The sequences of double-stranded oligonucleotides used in EMSAs are depicted in Fig. 1
. Other probes used were: CRE consensus, 5'-AGA GAT TGC CTG ACG TCA GAG AGC TAG-3'; CRE mutant, 5'-AGA GAT TGC CTG TGG TCA GAG AGC TAG-3'; GR consensus, 5'-GAC CCT AGA GGA TCT GTA CAG GAT GTT CTA GAT-3'; GR mutant, 5'- GAC CCT AGA GGA TCT CAA CAG GAT GTT CTA GAT-3'; STAT, 5'-CTC CTA TTG GCT TGA-3' (mutated bases in bold). EMSAs were conducted as described previously (Pedraza-Alva et al. 1994). Briefly, oligonucleotides were end-labeled with T4 polynucleotide kinase using 30 µCi [
-32P]ATP/100 ng of oligonucleotide. Nuclear extracts (10 µg) were incubated for 20 min at room temperature with the labeled oligonucleotide (1 x 105 c.p.m.) in band shift buffer (25 mM Hepes, pH 7.9, 40 mM KCl, 3 mM MgCl2, 0.1 mM EDTA, 1 mM dithiothreitol and 10% glycerol) containing 1 µg poly(dI-dC) as a non-specific competitor. DNAprotein complexes were resolved by electrophoresis on non-denaturing 6% polyacrylamide gels for 23 h at 150 V (50 mM TrisHCl, 45 mM boric acid, 0.5 mM EDTA). They were analyzed either directly with a phosphor-imager (Molecular Dynamics, Piscataway, NJ, USA) or by film autoradiography (Fluor-S MultiImager; BioRad, Hercules, CA, USA).
For competition experiments, 10- to 100-fold molar excess of unlabeled oligonucleotide was added 5 min before adding the labeled probe. For immune band shift assays, 1 µg antibody was incubated with nuclear extracts (5 h at 4 °C) and then with labeled oligonucleotide. The same amount of an irrelevant antibody (rabbit IgG) or normal rabbit serum (NRS) was used as control. DNAprotein complexes were resolved by electrophoresis on 5% polyacrylamide gels.
Statistical analysis
Results were calculated as percentages of controls of each culture; data were then calculated as the means± S.E.M. Data were analyzed by ANOVA, considered significant at P<0.001, followed by Fishers PLSD test (P values stated in each figure).
| Results |
|---|
|
|
|---|
To verify if the upregulation of proTRH mRNA levels by 8Br-cAMP (Pérez-Martínez et al. 1998) was due to PKA activation, we studied the effect of inhibiting PKA by preincubating (30 min) hypothalamic cells with 50 nM H89. As shown in Fig. 2
, 1
mM 8Br-cAMP increased proTRH mRNA levels and this effect was avoided with H89; the presence of H89 revealed a dex stimulatory effect. As previously shown (Pérez-Martínez et al. 1998), 10 nM dex reduced the stimulation caused by PKA activation (Fig. 2
).
|
|
Regulation of TRH release and content in hypothalamic cultures
Rapid effects of glucocorticoids can occur at the membrane level affecting intracellular calcium or protein kinases that can alter, for example, vasopressin release (Makara & Haller 2001, Chen & Qiu 1999). We studied if the release of TRH was regulated in hypothalamic cells treated for 12 h with dex or 8Br-cAMP, alone or combined. Increased release was detected after 1 h incubation with 8Br-cAMP and was still evident at 2 h but TRH cell content was augmented only after 2 h incubation (Fig. 4
). 8Br-cAMP produced a higher response than dex in TRH release and content, coincident with the faster and higher increase in proTRH mRNA levels (Uribe et al. 1995a, Perez-Martinez et al. 1998). Longer times were not studied since it would be difficult to discern a higher basal release due to higher peptide cell content, from a direct effect on regulated exocytosis (Luo et al. 1995, Nillni et al. 2000). TRH content was increased by 8Br-cAMP and dex but not with combined treatment (Fig. 4B
), as previously shown for proTRH mRNA levels (Pérez-Martínez et al. 1998). These results indicate a role for cAMP in modifying TRH release, besides its effect on biosynthesis. Changes in cell content reflected the effects found at transcription level: dex increased TRH cell content at 2 h and impeded the stimulatory effect of 8Br-cAMP. The lack of decreased content supported the assumption that, at the concentrations used, no cytotoxicity occurred.
|
Fast modulation of TRH gene expression in the PVN has been observed in vivo when rats were exposed to cold ambient temperature (Uribe et al. 1993, Sánchez et al. 2001), a condition in which NA is a neurotransmitter candidate in the afferent pathway (Arancibia et al. 1989, Lechan & Toni 1992). Since PKA activation is one of the main intracellular pathways activated by catecholamines, we tested the effect of NA on TRH content and release, and proTRH mRNA levels in hypothalamic cell cultures. A 1-h incubation with 10 nM NA increased TRH release without affecting cell content (Fig. 5A
). A fast and transient increase in proTRH mRNA levels was observed from 30 min, was highest at 60 min and decreased after 120 min incubation with 10 nM NA (Fig. 5B
). NA at 1 or 100 nM increased proTRH mRNA levels to 180 ± 15 or 191 ± 17% respectively (controls=100 ± 5%; n=6, P<0.01) after 1 h incubation. Pretreatment with H89 abrogated the stimulatory effect of NA while staurosporine did not (Fig. 5C
). As observed for a cAMP analog (Pérez-Martínez et al. 1998), the NA-induced increase in proTRH mRNA levels was higher than that caused by dex and co-incubation of NA with dex diminished the response to NA (Fig. 5D and E
). These results show that NA, a known physiological modulator, regulated TRH expression in a similar way to that observed with 8Br-cAMP in primary cultures of fetal hypothalamus.
|
Binding to CRE
To study if the interference of CREB with GR or AP-1 elements was due to their binding to DNA cognate sequences, we first analyzed if the proposed elements in the TRH promoter, involved in CREB or GR recognition, would bind nuclear proteins extracted from cells of fetal primary hypothalamic cultures incubated with dex or drugs that activate PKA or PKC. Cultures were treated at 14 DIV since at this time they have maximal TRH mRNA levels (Pérez-Martínez et al. 2001). Putative DNA binding sites for CREB, GR, and AP-1 in the rTRH promoter region are shown in Fig. 1
. Two sites have been described as possible CREs: CRE-2 at 101/ 94 and CRE-1 at 59/ 52; as mentioned, CRE-1 (site 4) has received exclusive attention as a target for CREB, AP-1 and TR. Since CREB phosphorylation increases DNA binding in canonical sequences and particularly in non-canonical sites (Benbrook & Jones 1994, Bullock & Habener 1998), we analyzed by EMSA whether nuclear extracts from hypothalamic cells treated with 8Br-cAMP had enhanced DNA-binding activity to the putative CRE elements located on the TRH promoter. Using CRE-1, two faint bands were observed with nuclear extracts from hypothalamic cells; they did not differ in intensity if the nuclear extracts were obtained from cells incubated (13 h) with 8Br-cAMP, TPA or dex (representative gel of some groups, in Fig. 7A
).
|
|
Binding to GRE
As mentioned, the GRE half-site located at 210/ 205 confers dex response in transfected cells (Lee et al. 1996). However, adjacent to this site in the complementary strand are two sequences similar to the AP-1 response element (Fig. 1
); this arrangement has been observed for corticotropin-releasing hormone (CRH) and other genes, and is referred to as composite GRE (Miner & Yamamoto 1992, Harrison et al. 1995, Malkoski & Dorin 1999). We tested the ability of nuclear extracts from dex-treated hypothalamic cells (3 h) to bind to the oligonucleotide sequence 220/ 193 (GRE-L); two bands were observed with the upper one (lower mobility shown by letter a in Fig. 8A
) increasing upon dex treatment (Fig. 8A
). Competition assays with up to 100-fold excess of GRE-L or a consensus GRE completely abolished DNA binding while no competition was observed when the same amount of mutated GRE or STAT oligonucleotide was added (Fig. 8A
). GR expressed in NIH-3T3 cells bound to GRE-L; no binding was detected with nuclear extracts from cells transfected with RSV-ß-gal (Fig. 8B
). Binding was increased with nuclear extracts from hypothalamic cells incubated with dex (13 h); TPA stimulation (3 h) increased the intensity of the faster mobility complex (indicated by letter b; Fig. 8C
). Co-incubation of dex with 8Br-cAMP diminished binding compared with nuclear extracts from cells incubated only with dex; a slight decrease was observed in the signal of nuclear extracts from cells co-incubated with dex+TPA (Fig. 8C
).
|
|
|
Competition and supershift analysis
Binding to CRE-2 with nuclear extracts from hypothalamic cells treated with 8Br-cAMP or NA was abolished with an excess of CRE-2 or consensus CRE; even though no increased binding to CRE-1 had been detected, this oligonucleotide was also able to displace binding of both complexes. GRE-A, which contains the perfect AP-1, displaced both bands but to a lesser extent, band a (Fig. 11A
). The same extracts tested against GRE-A, with consensus GRE, GRE-A, CRE-2 and CRE-1 as competitors, showed complete displacement of band b. Mutual displacement of CRE-2 for GRE-A binding, and vice versa, suggests recognition of the AP-1 site present in GRE-A by some of the protein complexes bound to CRE-2 (Habener 1990, Rutberg et al. 1999). One band of lower mobility (band a) was observed with nuclear extracts from NA-treated cells and GRE-A; the band was displaced with GRE-A and CRE-1 but only slightly with CRE-2 or consensus GRE (Fig. 11B
).
|
In an attempt to characterize the nature of the proteins binding to CRE-2 or GRE-A we performed supershift analyses using antibodies recognizing P-CREB, c-Jun or GR. We were unable to detect a shifted band whether the antibody was incubated before or after DNAprotein complex formation; this could be due to instability of protein complexes with the antibody due to the short length of the oligonucleotides used (16 bp). However, binding to CRE-2 with nuclear extracts from hypothalamic 8Br-cAMP-treated cells diminished with P-CREB antibody (46% vs 100% (signal in stimulated cells)) and with c-Jun antibody (52%); this also occurred in dex-treated cells but to a lesser magnitude (74%) (Fig. 12A
). In contrast, binding to GRE-A with nuclear extracts from 8Br-cAMP-treated cells decreased only with c-Jun antibody (24%), while from dex-treated cells binding decreased with c-Jun and GR antibodies (54%) (Fig. 12B
). The last lane of both gels corresponds to an equivalent amount of normal rabbit serum (NRS) and shows no decreased signal.
|
| Discussion |
|---|
|
|
|---|
The hypothalamic cell culture optimized in our laboratory (Charli et al. 1995, Pérez-Martínez et al. 2001, Joseph-Bravo et al. 2002) has proven to respond not only to 8Br-cAMP but also to NA stimulation by causing an increase in variables that act as indicators for the activation of TRH neurons: increase in peptide release, and its mRNA (Joseph-Bravo et al. 1998).
Release of glucocorticoids occurs in almost any stressful condition and even in events such as suckling (Herman et al. 1992, Uribe et al. 1993, Sánchez et al. 2001). The rise is immediate and therefore it is difficult to envisage situations that are free of variations in the levels of this hormone. How these variations modulate further responses is currently being investigated by many research groups. Positive or negative effects of glucocorticoids depend on the gene and the cellular make-up; their actions are not exclusive to an interaction with the mineralocorticoid or glucocorticoid receptors (with or without DNA binding) but also affect the transcriptional activity of other factors (Newton 2000, de Bosscher et al. 2001, Schaaf & Cidlowski 2003). It is now recognized that glucocorticoids can act at the membrane level activating second messenger pathways and release of various neuromodulators and hormones (Chen & Qiu 1999). We found no effect of dex on TRH release after 1 h incubation but the possibility remains that some of its effects on proTRH mRNA levels were initiated at the plasma membrane. We have previously reported that dex-mediated upregulation of proTRH mRNA levels (1 h) occurs independently of protein synthesis (Pérez-Martínez et al. 1998). The stimulatory effect of PKA activation or glucocorticoids on proTRH mRNA levels in hypothalamic cell cultures can occur at the level of gene transcription since in transfected hypothalamic cells, 8Br-cAMP or dex increased luciferase activity under the control of TRH promoter. The degree of transcriptional activation varied between these pathways depending on the cell system. Compared with the extent of stimulation obtained by PKA activation or dex treatment in primary cultures on proTRH mRNA levels (4.5-fold by 8Br-cAMP vs 3-fold by dex (Pérez-Martínez et al. 1998)), 8Br-cAMP-induced luciferase activity was higher than that induced by AP-1 or GR, despite the fact that there were no major differences in the expression levels of the transcription factors as determined by DNA-binding activity (transfected homologous system: primary hypothalamic cell cultures, 7-vs 2-fold for 8Br-cAMP vs dex stimulation respectively; heterologous system: 32- vs 2.7-fold respectively). This discrepancy could be explained by the difference in GR activity on a stably integrated template (and its effect on chromatin remodeling and recruitment of coactivators) compared with a naked transiently transfected promoter (McKeena et al. 1999).
The interference caused by dex (or GR) on 8Br-cAMP (or CREB) activation on proTRH mRNA levels also occurred at the transcriptional level as observed for CRH (Guardiola-Díaz et al. 1996, Malkoski et al. 1997, King et al. 2002), vasopressin (Kuwara et al. 2003) or the
-subunit gene of glycoprotein hormone (Stauber et al. 1992); while in some of these cases dex alone can cause inhibition or no effect, TRH regulation coincides with that of annexin A1 that is stimulated by dex or cAMP but when combined, inhibition is observed (Antonicelli et al. 2001).
Interaction between transcription factors occurs at different steps, some of which are independent of DNA binding, at least for one of them (Newton 2000, Schaaf & Cidlowski 2003). We studied whether the inhibitory effect of dex on 8Br-cAMP activation was related to altered DNA binding to either CRE or GRE sites. Activation of the PKA-dependent signal pathway led to the binding of P-CREB to a synthetic oligonucleotide containing in tandem both putative TRH CRE non-canonical sequences. The specific nature of P-CREB binding to the TRH CRE-L element was demonstrated by competition studies, by supershift assays using an antibody specific for P-CREB, and binding of transfected CREB. TPA-mediated TRH promoter activity in cell cultures correlated with the induction of protein complexes binding to the TRH CRE; these complexes probably included AP-1 since over-expression of c-Jun and c-Fos in NIH-3T3 fibroblasts produced a complex able to bind CRE-L. P-CREB and AP-1 could thus bind to overlapping regulatory elements of the rat TRH promoter. When both CRE elements were independently studied with short oligonucleotides (16 bp), CRE-2 showed increased binding with nuclear extracts from 8Br-cAMP-treated cells but not CRE-1, which presented a very faint non-inducible band. These two elements differ from the consensus CRE (TGACGTCA): CRE-2 (TGCCGTCA) and CRE-1 (TGACCTCA). Variations from the consensus sequence cause differences in binding affinity (Benbrook & Jones 1994) and phosphorylation of CREB influences DNA-binding affinities on sequences deviating from the canonical symmetrical site, and also that of the consensus sequence (Bullock & Habener 1998). CRE-1, within a longer oligonucleotide ( 83/ 36), is recognized by transfected CREB and only slightly by the endogenous form present in 293T cells (Harris et al. 2001). Although the sequence present in CRE-1 has a similar binding affinity to consensus CRE (Bullock & Habener 1998), it remains to be studied whether the adjacent contextual sequences are required for CRE-1 to function properly in the TRH promoter (Deutch et al. 1988). The rat serine dehydratase gene possesses two CRE sites: a CRE-1 identical to TRH CRE-1, and another CRE-like sequence that has the conserved critical internal CpG dinucleotide (TGCCGCAA) supposed to have an important role in adequate binding (Deutch et al. 1988, Haas & Pitot 1999); the second site was found to be the preferred binding site to CREB (Haas & Pitot 1999). CRE-2 in the TRH promoter also contains this central CpG dinucleotide; whether this explains the difference in binding of hypothalamic nuclear extracts with CRE-2 vs CRE-1 will have to be confirmed by foot-printing and deletion analysis (in future studies). It cannot be overlooked, however, that when the CRE-1 site is mutated to TAAAACT, basal and
MSH-stimulated transcription of 150/+5 hTRH transfected promoter is lost (Harris et al. 2001). This site is also the preferred binding site of TRs and cells transfected with 177/+83 TRH promoter have increased transcriptional activity if cotransfected with TRs in the absence of T3, and repressed by addition of the hormone (Feng et al. 1994, Satoh et al. 1996, 1999). This site is close to the initiation site and the negative effect of T3 has been proposed to alter TR conformation such that it interferes, through proteinprotein interactions, with basal transcriptional machinery (Feng et al. 1994, Satoh et al. 1996). Whether mutation of this site affects not only CREB binding, but also TR or other factors, remains to be studied. Furthermore, since these studies have been performed in non-neuronal cells, the possibility remains that in other cell types CREB-like proteins (or particular heterodimers) different from those of neurons are able to recognize CRE-1.
Two DNAprotein complexes are observed with CRE-2: band a is only displaced by CREs but not by GRE-A; in contrast, band b could contain AP-1-related factors, as judged by preferential displacement by GRE-A. When binding to CRE-2 with nuclear extracts from hypothalamic cells was compared with that of SH-SY5Y cells, band a coincided in both cultures while band b differed from the most intense band, highly increased in TPA-differentiated cells. The nature of these complexes remains to be elucidated since CRE is recognized by several protein complexes with different affinities: various CREBs, CRE modulators (CREMs), AP-1, CREB-activating transcription factor-1 (ATF1), c-Jun homodimers and heterodimers with ATF2, to name a few (Habener 1990, Masquillier & Sassone-Corsi 1992, Rutberg et al. 1999). Partial characterization of complexes by supershift assays suggests that CREB and Jun antibodies inhibit binding to CRE-2; the relative levels of these factors may thus influence transcriptional activity and affect the levels of proTRH mRNA without evident changes in the intensity of bound complexes. The increased binding to CRE-2 in nuclear extracts from hypothalamic cells stimulated after dex incubation for 1 h points to a membrane-mediated effect through kinase pathways; although this is partially supported by the small inhibitory effect of H89, staurosporine had no effect; other pathways remain to be studied.
Many of the effects of dex on gene expression are mediated by binding of the glucocorticoid receptor to its cognate DNA sequences on the target promoters (Newton 2000, Schaaf & Cidlowski 2003). Our results show that dex stimulation of hypothalamic cells induced GR binding to a TRH GRE. This binding activity was competed with an oligonucleotide containing a consensus GRE and not by a mutated GRE; transfected GR bound to GRE-L with similar mobility. Since TPA stimulation produced a second band, this oligonucleotide was shortened to avoid the second AP-1 site. Dex-treated hypothalamic cells showed higher binding to GRE-A than cells treated with 8Br-cAMP for 1 or 3 h; in contrast, neuroblastoma cells presented an equivalent or higher binding after 8Br-cAMP than after dex stimulation. PKA activation increases GR binding to GRE (Rangarajan et al. 1992); however, the presence in GRE-A of AP-1 consensus sequences that recognize nuclear proteins activated by TPA stimulation and even by CREB (Deutch et al. 1988, Habener 1990, Masquillier & Sassone-Corsi 1992, Hill & Treisman 1995, Malkoski & Dorin 1999, Rutberg et al. 1999) stresses the need for characterization of these complexes.
It has been recognized that there is no general competition model to explain interaction between transcription factors; effects depend on the cell type or the promoter, and can occur at different levels (Harrison et al. 1995, McKeena et al. 1999, Yamada et al. 1999, de Bosscher et al. 2001). A clear antagonism was observed between dex and 8Br-cAMP stimulation whether in mRNA levels, transcription activity or, in certain conditions, DNA binding. Several forms of interaction have been proposed between PKA and glucocorticoids. Those related to DNA binding include: CREBGR direct interactions avoiding DNA binding (Imai et al. 1993, Yamada et al. 1999) and sequestration of PKA signaling (Doucas et al. 2000). However, with these mechanisms, a mutual inhibition would be expected either on CRE or GRE sites but we observed an inhibition of protein binding to CRE only after 3 h incubation. The decreased binding to CRE (either CRE-L or CRE-2) could be due to glucocorticoids inhibiting CREB phosphorylation (Legradi et al. 1997b, Whitehead & Carter 1997). In contrast, binding to the composite GRE (GRE-A) was diminished at 1 and 3 h in dex+8Br-cAMP hypothalamic 14 DIV cells and in differentiated neuroblastoma cells. Since GR half-life is diminished to 3 h after hormone binding but is increased by cAMP to 10 h (Dong et al. 1999), decreased binding to GRE-A cannot be explained by shortening GR half-life after 1-h stimulation. In the CRH promoter, a consensus CRE sequence and a composite GRE are involved in its transcriptional regulation; the stimulatory effect of PKA activation is mediated by CRE (Guardiola-Díaz et al. 1996) but cGRE participates in glucocorticoid-dependent repression of PKA-induced transcription (Malkoski et al. 1997, Malkoski & Dorin 1999). Furthermore, the CRH gene can be induced by glucocorticoids in certain conditions through a different region in the promoter (King et al. 2002). The nature of the different nucleoproteins able to bind to a composite GRE determines the composition of coactivators or corepressors that ultimately define the positive or negative effects. In the case of the proTRH gene, a heterodimer GRJun (suggested to bind to composite GREs (Miner & Yamamoto 1992, Maroder et al. 1993)) may be formed, and the response could depend on the type of complex bound (c-Jun homodimer, CREB, GRJun, etc). Differential effects of c-Jun forming particular dimers could affect CREB or GR binding, and/or their transcriptional activity; thus, a working hypothesis of c-Jun as a central regulator of proTRH expression seems worth considering.
In conclusion, the ensemble of these results demonstrates rapid regulation of TRH gene transcription by 8Br-cAMP which can be repressed by dex in hypothalamic or in NIH-3T3 cells. In hypothalamic cells, NA reproduced the stimulatory effect of 8Br-cAMP on proTRH mRNA levels and interacted similarly to 8Br-cAMP co-stimulation with dex, supporting a physiological relevance of this interaction. We identified a CRE site located at 101/ 94, and a composite GRE ( 210/ 205), that may regulate the transcriptional effects of various information pathways in neuronal cells. We have restricted the analysis to the role of PKA but other pathways activated by cAMP remain to be studied; additional work such as DNA foot-printing and deletion analysis, as well as further characterization of the protein complexes observed in EMSA stud