Introduction

Thyrotropin releasing hormone (TRH: pglu-his-proNH2), the first hypophysiotropic peptide discovered, was later found in various regions of the limbic system where it has a neuromodulatory function [1]. TRH is synthesized, in the rat, from a protein precursor containing five repeated sequences of gln-his-pro-gly that give rise to the active peptide [2]. TRH, released from neurons of the hypothalamic paraventricular nucleus (PVN) that project to the median eminence, controls the synthesis and release of the pituitary hormones thyrotropin and prolactin; these neurons behave as neuroendocrine transducers that play a critical role in maintaining energy homeostasis as their activity is modulated by multiple metabolic, hormonal, and neuronal signals [3, 4]. In response to a neuronal stimulus as cold exposure, TRH mRNA levels in the PVN increase very rapidly (30–60 min) and transiently as levels normalize by 2 h, despite leaving the rats in the cold for up to 6 h [5]; activation includes non-hypophysiotropic TRHergic neurons of the anterior PVN [6]. In this condition, serum corticosterone levels increase [5, 6].

In fetal hypothalamic cultures, norepinephrine, the neurotransmitter proposed to mediate the effects of cold on TRH release [7], increases TRH mRNA levels; this is mimicked by activation of protein kinase A (PKA) or C (PKC) [8–10]; levels are also enhanced by dexamethasone (dex), an analog of corticosterone that activates GR [9, 10]. The stimulatory effect of dex or 8Br-cAMP is inhibited when cells are co-incubated with both drugs. These effects occur at the transcriptional level since they are observed after transfecting a plasmid containing the minimal pro-TRH promoter bound to luciferase cDNA into either hypothalamic cells or the cell lines NIH-3T3 [10] or neuroblastoma SH-SY5Y (unpublished).

The TRH gene promoter contains response elements to various transcription factors activated by neuronal or hormonal effectors [11]. Using electrophoretic mobility shift assays (EMSA) we demonstrated that nuclear extracts from hypothalamic cultures incubated with norepinephrine or 8Br-cAMP bind to the CRE site of TRH gene promoter located at −101/−94 bp (CRE-2) [10] and not to site 4 (−59/−52) previously proposed as the CRE site [12]. We also recognized a half site GRE (GRE1/2) located at −210/−205 that is flanked by two AP-1 recognition sites forming a composite GRE (cGRE: −218/−198) [10]. GR monomer is able to bind to cGRE as an heterodimer with c-Jun or c-Fos [13, 14]. Binding of nuclear extracts from hypothalamic cells, incubated with 8Br-cAMP and dex, to either CRE or cGRE is diminished or avoided when cells are co-incubated with both compounds [10]. Involvement of pCREB, GR, or c-Jun is inferred by supershift [10] and chromatin immunoprecipitation assays (unpublished).

In contrast to the mutual antagonism between PKA signaling and glucocorticoids, dex and phorbol esters (TPA) have an additive effect on TRH mRNA levels of hypothalamic cells [9] as well as on the binding of their nuclear factors at the cGRE [10]. These results suggest that PKC stimulates AP-1 activity allowing formation of a stable GR:c-Jun or GR:c-Fos heterodimer [13–15]. Protein kinase C isoenzymes constitute a family of serine/threonine kinases that are classified into three subfamilies based on their ligand and cofactor requirements [16]. Depending on the cellular context, the isotype and its localization, PKC regulates a wide array of cellular responses including signaling of mitogen activated protein kinase (MAPK) [17]. MAPK affect the activity of transcription factors; for example, MEK1, 2 (through the extracellular regulated kinases 1 or 2 [ERK1/2]) and MEK5 phosphorylate CREB or GR, the p38 MAPK phosphorylates hGR, and JNK activates c-Jun [18–20].

Glucocorticoids in turn, through GR, enhance the expression of various components of the MAPK cascade (Ras, Raf-1, ERK1 and 2) [21, 22] but also, inhibit the activity of ERK1/2, p38, and JNK through dephosphorylation by the phosphatase MKP-1, whose expression is fastly increased by GR activation (MKP-1 mRNA at 5–15 min; protein levels at 30 min) [23]. Regulation of the MAPK pathways and MKP-1 constitute thus a negative feedback loop controlling glucocorticoid action.

The aim of this work is to study the involvement of PKC and MAPK on dex-induced TRH gene expression as both kinases constitute a convergence point with the GR mediated signaling pathway [24] and are common mediators in signal transduction pathways activated by growth factors, cytokines and stress [24–26]. The involvement of PKC and MAPK signaling pathways was evaluated on the effect of dexamethasone alone or combined with 8Br-cAMP on TRH mRNA levels in hypothalamic cells and, on the binding of activated nuclear factors at the CRE and cGRE sites of TRH promoter.

Experimental Procedure

Animals

Female Wistar rats (280–300 g) raised at the Institute's vivarium were mated on proestrous–oestrous stage of the cycle and kept in 12 h dark-light cycle, controlled temperature (23°C) and fed at libitum (Purina 2018S, Harlam). Care was taken to preserve adequate conditions following the guidelines of the Society for Neuroscience. Protocols were approved by the Institute’s Ethical commission.

Cell Culture

Primary cultures of hypothalamic cells were performed as described [9, 27]. Briefly, on the 17th day of gestation, dames were anesthesized with 0.45 ml Sedalphorte (63 mg sodium pentobarbital). The embryos were removed and the brain excised through an imaginary line between both eye and ear superior edge. The brain was kept in the skull to avoid tissue distortion, and the hypothalamus dissected by cutting the area limited by the optic chiasm and lateral sulcus including the mammillary bodies to a 2–3 mm depth that avoids the thalamic area [28]. 6 × 105 cells (yield: 2 × 106/hypothalamus) were plated in 15 mm dishes for those experiments where levels of TRH mRNA were evaluated and in 60 mm dishes for the preparation of nuclear extracts. Cultures were incubated for 14–18 days in vitro (DIV) at 90% humidity, 7% CO2, 93% air, 37°C. The effect of 10 nM TPA alone or together with 10 nM dex was tested on staurosporine-treated or PKC-depleted cells incubating the cells at 18 DIV for 3 h; cells were pretreated for 30 min with 1 μM staurosporine [29] and PKC depletion was achieved by incubating the cells at 17 DIV, for 18 h, with 5 μM TPA [30]. The following experiments were performed at 14 DIV; cells were incubated for 30 min with either 10 μM FPT III (inhibitor 3 of farnesyltransferase, enzyme that associates Ras to the membrane [31]), 50 μM PD98059 (MEK inhibitor [32]), 25 μM SB0203580 (SAPK/p38 inhibitor [33]), or 6 μM Ro-318220 (PKC inhibitor [30]) and then, for 1 h with 1 mM 8Br-cAMP, or 100 nM forskolin, 10 nM dex or combined. The dose of 10 nM dex is sub-maximal on TRH mRNA levels [9]. Cultures were observed under an inverted microscope (Zeiss Axioskop microscope); no damage was detected.

The neuroblastoma cell line SH-SY5Y (ATCC CRL-2266), kindly donated by Dra. Rosa Maria Pardós (Instituto de Fisiología, Benemérita Universidad Autónoma de Puebla, México), was grown as described [10]. Cells were differentiated at seeding with 1.6 nM TPA, cultured for 8 days with medium changes every 2 days. These cells do not express TRH mRNA (unpublished) but possess the transduction pathways involved in glucocorticoid, PKC, PKC and MAPK signals [34, 35]; nuclear factors of stimulated cells show the similar binding characteristics as those from hypothalamic cultures [10].

RT-PCR

Total RNA was purified as described [9] and analyzed in 1% agarose gels to verify its integrity. Samples showing a 260/280 nm ratio less than 1.8 were discarded. RT-PCR were carried out as described [9, 10] using 25 pM of the reported oligonucleotide primers for TRH and 50 pM for cyclophilin [10]: for TRH, sense: 5′ CCC TGG ATG GAG TCT GAT GT 3′, antisense: 5′ GAC AGC TAG TGA AGG GAA CAG G 3′ (356 bp product), 64°C, 29 cycles; for cyclophilin, sense: 5′ GGG GAG AAA GGA TTT GGC TA 3′, antisense: 5′ ACA TGC TTG CCA TCC AGC C 3′ (259 bp product), 64°C, 24 cycles. For some cultures, 50 pM of specific primers for rat G3PDH was used as control (sense primer, 5′-TGAAGGTCGGTGTCAACGGATTTGGC-3′; antisense primer, 5′-CATGTAGGCCATGAGGTCCACCAC-3′) with the amplification conditions previously described [10, 36].

Electrophoretic Mobility Shift Assay

The oligonucleotides described in Fig. 1 were synthesized at the Institute’s unit [37] and end-labeled with T4 polynucleotide kinase (Promega) and 30 μCi [γ-32P] ATP/100 ng oligonucleotide (NEN Life Science Products, Boston MA) as reported [10]. EMSA was performed as published [10, 38]; briefly, 10 μg of protein from nuclear extracts were incubated 30 min with one of the double stranded labeled oligonucleotides, and the mixture submitted to non-denaturing gel-electrophoresis (5% acrylamide:bisacrylamide 29:1) in TBE buffer (50 mM Tris HCl, 45 mM boric acid, 0.5 mM EDTA), and analyzed with a phosphorimager. The mutated sequences map-cGRE or mg-cGRE were analyzed with TRANSFAC 4.0 and Mat Inspector to verify that each would not recognize any known transcription factor.

Fig. 1
figure 1

Sequences of double stranded oligonucleotides used in EMSA. Letters in bold and italics mark the sequences on the TRH promoter, of the cAMP response element (CRE-2), the composite GRE (cGRE), with the GRE1/2 consensus sequence in the 5′-3′strand and the two AP-1 sites in the 3′-5′ strand; arrows indicate direction of the sequence. Mutations of the GRE1/2 (mg) or the AP-1 sites (map) of cGRE are marked with letters in minuscule

Full size image

Statistical Analysis

Results were calculated as % of controls of each culture; data were then calculated as the mean ± S.E.M. Data of the relative mRNA levels were analyzed by ANOVA, considered significant at P < 0.01, followed by Fisher’s PLSD test. Data from densitometric analyses were calculated taking the signal from control unstimulated cells as 100%.

Results

Regulation of TRH mRNA Levels in Hypothalamic Cells

As co-treatment with TPA and dex results in an additive increase in TRH mRNA levels [9], we evaluated the effect of a non specific inhibitor of PKC (staurosporine) to determine the dependency of these effects on PKC activity. Cells were treated for 3 h since the effect of 10−7 M TPA on TRH mRNA levels is observed only after 2 h probably due to the time required for adequate synthesis of c-Fos [8]. The hypothesis behind this experiment was that dex may stimulate TRH mRNA levels by GR heterodimerization with c-Jun or c-Fos [13, 14], and PKC inhibition affect phosphorylation of c-Fos or c-Jun as well as AP-1 formation (c-Fos:c-Jun heterodimer). As seen in Fig. 2, staurosporine not only inhibited the stimulatory effect of TPA but also that of dex + TPA (values were below controls). Likewise, PKC depletion, by long term TPA preincubation [30], blocked the effects of TPA or dex + TPA on TRH mRNA levels. These results suggest that PKC activation of c-Jun or c-Fos is required for an adequate response of the TRH promoter to dex.

Fig. 2
figure 2

Effect of PKC inhibition on TPA and dex effects on TRH mRNA expression. Hypothalamic cells were preincubated at 18 DIV for 30 min with 1 μM staurosporine or 18 h with 5 μM TPA; cells were then incubated for 3 h with 100 nM TPA or TPA + 10 nM dexamethasone. The relative levels of TRH mRNA were evaluated by RT-PCR and calculated as the ratio of TRH cDNA signal/G3PDH cDNA signal; values were expressed as % of controls. Insert depicts a sample of the RT-PCR products, with arrows signaling G3PDH and TRH cDNAs. *Significant difference against control, n = 6/group; letters: f, h, significant difference against F, G; e, h = between these two groups; P < 0.01 by Post-hoc Fisher analysis

Full size image

The extent and loci of GR phosphorylation determines its stability, nuclear translocation and, in a promoter-dependent manner, some of its genomic effects [18–20]. Since various MAPKs phosphorylate GR [22, 24] reducing its activity, we studied their involvement in the stimulatory effect of dex and its interaction with cAMP-stimulated factors. As previously reported [9, 10], hypothalamic cells cultured for 1 h with either 8Br-cAMP or dex increased TRH mRNA levels, but when both drugs were added simultaneously the stimulation was lower that caused by any of them (Fig. 3a). Preincubation with the MEK inhibitor PD98059 did not change the stimulatory effect of 8Br-cAMP but inhibited that of dex, allowing the full effect of 8Br-cAMP when both drugs were added to the culture (Fig. 3a) (ANOVA: F 7,86 = 13.9, P < 0.0001; n = 6/group). These results support the involvement of the ERK pathway in the effect of dex on TRH mRNA levels as well as on its interference with that of 8Br-cAMP.

Fig. 3
figure 3

Effect of inhibitors of the MAPK pathway on TRH mRNA expression. Primary cultures of hypothalamic cells were preincubated at 18 DIV for 30 min with one of the following inhibitors: PD98059 (PD) for MEK, FPT III for Ras, and SB203580 for p38. Cells were then incubated with 1 mM 8Br-cAMP (A), or 10 nM dexamethasone (D), or 8Br-cAMP + dexamethasone (AD). The relative levels of TRH mRNA were evaluated by RT-PCR and calculated as the ratio of TRH cDNA signal/cyclophilin cDNA signal; values were expressed as % of controls. Insert depicts a sample of the RT-PCR products, with arrow signaling TRH cDNA. *Significant difference against control; P < 0.01 by Post-hoc Fisher analysis. Equal letters represent significant difference between these groups; in B: f are significantly different against F (P < 0.01, n = 4)

Full size image

To evaluate different stages of activation of the ERK and p38 pathways, other inhibitors were used on cells incubated with dex with or without 8Br-cAMP present; the 8Br-cAMP group was not included to be able to have at least three dishes/group and evaluate the effect of these inhibitors within the same culture. Results are the mean of two independent cultures (F 11,82 = 7.4, P < 0.0001, n = 6–8/group). As shown in Fig. 3b, the Ras inhibitor FPT III caused a small but significant increase in the basal levels of TRH mRNA suggesting a basal inhibitory effect of Ras; addition of dex or dex + 8Br-cAMP caused no further increase. The inhibitor of p38 kinase (SB203580) did not alter the basal TRH mRNA levels but reduced dex stimulatory effect, albeit to a lower extent than that observed with the MEK inhibitor PD98059 (Fig. 3a, b). In SB203580-treated cultures, co-stimulation with 8Br-cAMP and dex increased TRH mRNA levels compared to cells treated with the inhibitor but non stimulated (Fig. 3b), suggesting that the p38 inhibitor impeded the interfering effect of dex on 8Br-cAMP stimulation; the stimulatory effect of dex did not achieve significance due to the slight enhancement in TRH mRNA levels of unstimulated cells (compared to the control group). The effect of dex to increase TRH mRNA levels, and on its interference with cAMP activated pathways on TRH mRNA levels, seems thus to depend on the activation of ERK (1, 2 or 5); this latter event can be influenced by p38.

Binding of Nuclear Extracts from Hypothalamic Cells to Response Elements of the TRH Promoter

To further characterize the role of PKC and MEK, the ability of nuclear extracts from hypothalamic cells to bind to CRE-2 or cGRE sites was studied by EMSA, using a more specific PKC inhibitor (Ro-318220) than staurosporine [30, 39], and the MEK inhibitor PD98059. Nuclear extracts from controls or 8Br-cAMP-treated cells show two bands bound to oligonucleotides containing the CRE site that can be displaced with an oligonucleotide containing the consensus CRE sequence, or with anti-pCREB or anti-c-Jun [10] suggesting that the proteins bound are either pCREB homodimers, pCREB heterodimers with c-Jun or other unidentified factors of the ATF family. Similar results were obtained in these new EMSA experiments, being the lower-mobility band (a) considerably more intense than band (b). 8Br-cAMP or dex treatment increased binding of nuclear extracts compared to controls (Fig. 4a). Pretreatment with the MEK inhibitor increased the signal of band (a) from non stimulated cells to a similar intensity as those treated with 8Br-cAMP. Compared to the binding of nuclear extracts from cells incubated with dex only, a significant decrease in binding, and blurring of the bands, was observed using nuclear extracts from dex treated cells in presence of PD98059; this blurring indicates loss of binding stability [40]. This latter effect was more pronounced with Ro-318220 (Fig. 4a) (ANOVA: F 6,18 = 3.7, P = 0.01, n = 4). Results suggest that dex effects on nuclear factors capable of binding to CRE-2 require the activity of ERK or PKC.

Fig. 4
figure 4

Effect of MEK or PKC inhibition on binding of hypothalamic nuclear extracts to CRE or cGRE. Primary cultures of hypothalamic cells, preincubated at 18 DIV for 30 min with 50 μM PD98059 (pd), or 6 μM Ro-318220 (Ro) and then incubated 1 h with 1 mM 8Br-cAMP (A), or 10 nM dexamethasone (D). Their nuclear extracts were incubated with double stranded oligonucleotides containing CRE-2 (A) or cGRE sequences (B) (see Fig. 1) for EMSA analysis. (C) Histograms show the mean ± S.E.M. of the densitometric values of band (A) of CRE-2 or band (C) of cGRE calculated as % of controls (=100%)

Full size image

A similar analysis was performed using the same nuclear extracts with the oligonucleotide containing cGRE (Fig. 1b). Our previous work demonstrated that the lower band (c) detected using this oligonucleotide is displaced with excess of unlabeled oligonucleotides containing the consensus GRE, while the upper bands, with oligonucleotides containing consensus CRE or AP-1 suggesting binding of GR and AP-1 respectively [10]. As previously observed with hypothalamic cells [10], the main band was that of higher mobility band (c) whose signal was increased using nuclear extracts from cells treated with 8Br-cAMP similarly if they were preincubated with the MEK inhibitor; increased signal was observed with nuclear extracts from dex incubated cells which was further increased by PD98059 preincubation (Fig. 4b) (ANOVA: F 6,20 = 4.3, P = 0.006, n = 4). These results suggest that binding to the GRE site was not diminished by MEK inhibitor. The signal of bands (a) and (b) were very faint and difficult to quantify; the intensity of band (a) was slightly increased only with nuclear extracts from 8Br-cAMP treated cells and from unstimulated cell pretreated with PD98059 (Fig. 4b).

Effect of Mutations of the Composite GRE

To evaluate the role of each site in cGRE, we mutated the GRE1/2 site (mg-GRE) or the 2 AP-1 sites (map-GRE) (Fig. 1). As hypothalamic cell cultures contain many cell populations including glia, we used the differentiated neuroblastoma cell line SH-SY5Y. This line does not express TRH mRNA (unpublished) but its nuclear extracts have similar binding specificities to CRE or cGRE containing oligonucleotides, as hypothalamic cell cultures [10]. An exception is that the lower mobility band (a) of cGRE is more intense due to higher concentrations of AP-1 factors as cells are differentiated with TPA treatment [10]; band (b) behaves as with hypothalamic cells, very faint and without changes upon cell-treatments probably due to non-specific binding. We first stimulated cells with forskolin (FK) or dex that induced increased binding, of similar intensity, to both bands using cGRE (Fig. 5). Mutation of the AP-1 sites (map-cGRE) diminished binding compared to cGRE although, nuclear extracts from FK stimulated cells still showed increased signal compared to binding of control cells. Mutation of the GRE1/2 site (mg-cGRE) favored binding to band (a) of nuclear extracts from FK stimulated cells (Fig. 5; ANOVA: F 17,99 = 6, P < 0.0001; n = 6, two independent experiments). These results suggest that when the AP-1 sites are mutated, binding of nuclear extracts is considerably diminished, even at the GRE1/2 site suggesting that GR binding as a monomer is not stable.

Fig. 5
figure 5

Binding of nuclear extracts of neuroblastoma SH-SY5Y cells to cGRE and mutated cGREs. Differentiated neuroblastoma cells were incubated for 1 h with forskolin (FK) or dexamethasone (D) and their nuclear extracts analyzed by EMSA using the double stranded oligonucleotides shown in Fig. 1. Results of densitometric analyses of bands (a) and (c) from two independent experiments are shown in the table, calculated as % of control values (=100%; SEM was less than 9%, not shown), n = 4–6, *P < 0.005, +P < 0.05 against controls; e: P < 0.05 between them

Full size image

Neuroblastoma cells were then treated as the hypothalamic cell cultures, with 8Br-cAMP, dex, +/− MEK inhibitor. Results were similar as binding to band (a) at the CRE-oligonucleotide increased when cells were incubated with 8Br-cAMP or dex. Treatment with MEK inhibitor slightly increased binding compared to nuclear extracts from unstimulated cells but impeded the effect of 8Br-cAMP or dex treatment (Fig. 6a).

Fig. 6
figure 6

Effect of MEK inhibition on binding of nuclear extracts of neuroblastoma SH-SY5Y cells to CRE or cGRE. Differentiated neuroblastoma cells were preincubated for 30 min with PD98059 (pd or vehicle), followed by 1 h incubation with 1 mM 8Br-cAMP (A), or 10 nM dexamethasone (D). Nuclear extracts were incubated with double stranded oligonucleotides with CRE, cGRE or the mutated AP-1 (map-cGRE) or GRE1/2 sites (mg-cGRE) (see Fig. 1) for EMSA analysis. Histograms show the mean ± S.E.M. of the densitometric values of the band indicated on the abscissa, n = 3

Full size image

The signal of band (a) was increased using cGRE and nuclear extracts from cells treated with 8Br-cAMP, dex, MEK inhibitor alone, or with 8Br-cAMP, and only slightly with dex + PD98059. In contrast, the intensity of band (c) was increased only with dex treatment in presence of PD98059 (Fig. 6b).

Using the mutated cGRE lacking the two AP-1 sites (map-cGRE) lowered the intensity of bands compared to those of cGRE, as shown above; increased signal was observed on band (a) with nuclear extracts from 8Br-cAMP or dex treated cells which was completely abolished (signal was below that of controls) with the MEK inhibitor (Fig. 6c). Although there were no differences in the intensity of band (c) upon 8Br-cAMP or dex incubation, when these compounds were added to cells pretreated with PD98059, signal diminished below that of non stimulated cells treated or not with the inhibitor (Fig. 6c).

The signal of the DNA-protein complex corresponding to band (a) of the mutated cGRE lacking the GRE1/2 site (mg-cGRE) was increased using nuclear extracts from 8Br-cAMP or dex treated cells and, considerably higher than with cGRE; the duplet of band (a) suggests a different composition of the nuclear proteins (Fig. 6d). Pre-treatment with MEK inhibitor increased binding to (a) compared to untreated cells, this signal slightly decreased when 8Br-cAMP or dex was added in presence of the inhibitor (Fig. 6d). The increased signal of the inferior band (c) of 8Br-cAMP or dex treated cells was not observed if cells were preincubated with PD98058. The augmented intensity of mg-cGRE compared to cGRE points for binding of AP-1 complex with higher affinity than the heterodimer c-Fos:GR or c-Jun:GR.

Discussion

Through the combined approach of measuring TRH mRNA levels in hypothalamic cultured cells, and binding of their nuclear extracts to the CRE-2 and cGRE sequences of TRH promoter, we previously demonstrated a negative interaction between GR and PKA signaling pathways while a positive, between GR and PKC [9, 10]. It is well recognized that the specificity of the response depends on the tissue and cell type, as well as on the characteristics of the promoter. As glucocorticoid–PKA negative interactions are reproduced in an heterologous cell line (NIH-3T3) transfected with TRH-luc [10] or neuroblastoma cells (unpublished), and their nuclear extracts present similar binding responses in EMSA, these approaches allow recognition of cross-talk of intracellular pathways interacting with TRH promoter. The mutual antagonism between PKA and GR pathways occurs at the transcriptional level, and on the binding to CRE-2 (−101/−94) or cGRE (−218/−198) sites (observed by EMSA [10) and recently, using DNase protection assays [unpublished]).

In this work we provided evidence for the requirement of PKC and ERK/MAPK activities on the stimulatory effect of dex, and on its interference with cAMP-induced increase of TRH mRNA levels in hypothalamic cultured cells. As the stimulatory effect of dex on TRH mRNA levels, and its antagonism of cAMP signaling-induced transcription probably involve ligand-activated GR [10, and unpublished], the simplest explanation of our data could be that PKC, ERK, and partially p38, are required for GR action. However, activation of MAPK pathways provokes GR dysregulation [41] so their inhibition should increase GR function.

Unliganded GR is bound in the cytoplasm to a multiprotein complex containing heat-shock proteins (hsp90) and several kinases of the MAPK signaling system, including Src [42, 43]. Upon ligand binding, GR is released, phosphorylated and translocated to the nucleus where it acts as a transcription factor [42–44]. The activity and stability of the receptor varies depending on the site and degree of phosphorylation. Ligand binding induces phosphorylation at ser-211 (human GR) and translocation to the nucleus (within 5 min) [42]; translocation and transactivation is further enhanced with inhibitors of JNK or ERK, but decreased with inhibitors of p38 [20, 42–44]. Our results on loss of dex effect with p38 inhibition can be explained by p38 phosphorylating specific coactivators required for GR action and, indirectly inhibiting the ligand binding domain [45]. In contrast to the reported enhanced transactivation produced by GR after ERK inhibition [46], dex stimulatory effect on TRH mRNA levels was lost with PD98059, a result coincident with other examples in which PKC and ERK-MAPK inhibitors attenuate the corticosterone-mediated gene transcription [47]. We thus propose that the effect of PKC and ERK inhibition on dex effects on TRH transcription might be occurring at different levels: (1) the stimulatory effect of dex being impeded by inhibition of AP-1 components that are required for the formation of a stable GR heterodimer, and the appropriate recruitment of coactivators; (2) the interfering effect of dex on cAMP signaling to be due to effects on GR translocation (see below).

We previously suggested that dexamethasone may induce TRH transcription, and binding of nuclear factors to CRE-2 [10, unpublished], by increasing CREB phosphorylation (deduced by CHIP, unpublished) through increased synthesis of serum and glucocorticoid inducible kinase (SgK), that is able to phosphorylate CREB [48, 49]. SgK activity is regulated not only by glucocorticoids but also by TPA (in 1 h) and ERK at the transcriptional and posttranscriptional levels [50]. Furthermore, corticosterone treatment of AtT20 cells increases, in 1h, the amount of Ras, Raf-1 and Erk1/2 protein [21] which could also increase pCREB. The decreased binding to CRE-2 by PKC and MEK inhibitors in dex-stimulated cells can be explained by inhibition of ERK and SgK activities, since it has recently been shown that PKC can activate SgK, which in turn phosphorylates CREB via an ERK mediated mechanism [51].

The TRH gene promoter has a half site GRE, so GR can only bind to the cGRE that contains two flanking AP-1 sites as a monomer to the GRE1/2, which is stabilized by its union to either c-Jun, c-Fos, or even c-Jun:c-Jun or c-Jun:c-Fos, [10]. Mutations on cGRE showed that the AP-1 sites are essential for the stability of GR binding. PD98059 treatment inhibited the binding of nuclear extracts from dex stimulated cells to either CRE or the mutated cGREs (lacking either the AP-1 sites or the GRE1/2 site). The interaction of GR with AP-1 components involves the three-dimensional structure of GR [20], that could be affected by altering its phosphorylation state with the MEK inhibitor. This inhibitor caused a strong decrease on binding to the AP-1 sites of cGRE and, of mg-cGRE, suggesting diminished levels of activated c-Jun or c-Fos. This is supported by the effect of ERK1/2 activation on c-Fos induced transcription (observed within minutes), and its ability to determine the pattern of expression of other components of AP-1 [52], as well as their binding [53]. The stimulatory or inhibitory effects of the heterodimer formed, whether c-Fos:GR or c-Jun:GR depends on the promoter and cellular context [12, 13, 20]. At the short times of incubation studied, the effect of dexamethasone would require activation of GR, and heterodimerization with existing components of the AP-1 family as the MAPK pathways could be partially stimulated in the conditions used: cells are grown in presence of 10 % fetal bovine serum (FBS) and with conditioned medium, which contains BDNF and other growth factors, that allow higher basal levels of TRH [54].

The second effect, the interference of dex with 8Br-cAMP stimulation of TRH mRNA levels was avoided with inhibitors of p38 (SB0203580) and MEK (PD98059). Although ERK1/2 increases pCREB [55], its inhibition did not alter 8Br-cAMP induced enhancement of TRH mRNA levels, neither the binding of nuclear extracts of stimulated cells to CRE-2, suggesting PKA activation was sufficient for pCREB to increase TRH transcription independent of MEK activity. We have proposed that the mutual antagonism, between dex and 8Br-cAMP signaling pathways, involves the activated GR interacting with either pCREB [56] or cPKA [57] before binding of these transcription factors to their response elements [10]. Activation of GR depends, as mentioned above, not only on ligand binding but also, on the phosphorylation state of the receptor which can accelerate its nuclear translocation or, its export. MEK inhibition restores the nuclear translocation of GR in some systems [58, 59]; however, PD98059 treatment also promotes nuclear export of ERK/MAPK + Sgk into the cytoplasm [60]. As the transrepressive effect of glucocorticoids occurs only if glucocorticoids are present during the phase of active transcription [48], and we have shown that for mutual interference to occur, dex and 8Br-cAMP (or noradrenaline) must be added simultaneously (if dex is added 10 min before PKA stimulation, the interference is no longer observed) [61], the present results suggest that inhibition of ERK or p38 affects the nucleocytoplasmic traffic of activated GR, eliminating the interaction with pCREB (or PKA) leaving pCREB able to induce TRH transcription.

The results presented herein support a cross talk between various pathways that, depending on the stimulus could differentially modulate TRH transcription. The ERK1/2 pathway can be stimulated by growth factors as BDNF [25], that increases TRH mRNA levels in immature hypothalamic cultures and this effect is antagonized by PD98059 [62, 63]. In vivo, stress increases BDNF expression in the PVN in a fast and transient manner [64], which could activate TRH neurons expressing TrkB receptor [63]. One could envisage situations where the TRHergic neuron receives diverse stimuli [65, 66], simultaneously or in succession, adjusting the levels of immediate early genes expression, differentially modulating transcription factors as CREB, GR of AP-1 affecting thus, the inhibitory effect of rising corticosterone levels and, depending on the type of stress or previous history of the animal, alter the response to an incoming stimulus.