Regulation of the Rat Serotonin-1A Receptor Gene by Corticosteroids*

Dysregulation of the serotonergic system and abnormalities of the hypothalamic-pituitary-adrenal axis function have been implicated to be involved in neuropsychiatric disorders. Serotonin-1A receptors have been shown to be suppressed by corticosteroid hormones in a variety of animal studies. This effect may play a central role in the pathophysiology of depression. However, little is known about the molecular mechanism underlying this suppressive effect of corticosteroids. Here, we show by functional analysis of the promoter region of the rat serotonin-1A receptor gene that two NF-κB elements in the promoter contribute to induced transcription of the rat serotonin-1A receptor gene. Furthermore, we show that corticosteroids repress this NF-κB-mediated induction of transcription. Remarkably, we observed that only the glucocorticoid receptor and not the mineralocorticoid receptor was able to mediate this repressive effect of corticosteroids. We argue that negative cross-talk between the glucocorticoid receptor and NF-κB may provide a basis for the molecular mechanism underlying the negative action of corticosteroids on serotonin signaling in the brain.

The brain serotonin (5-hydroxytryptamine; 5-HT) 1 system has been strongly implicated in the control of behavioral processes, including feeding, aggression, and response to stress (1). Extensive evidence also supports the involvement of serotonin function in neuropsychiatric disorders, such as depression and anxiety (2,3). In addition, selective serotonergic compounds have been shown to be clinically effective antidepressants and anxiolytics (4,5).
The complex physiological actions of serotonin are mediated by a family of related receptors of which particular attention has focused on the 5-HT1A receptor (6). The 5-HT1A receptor belongs to the family of G protein-coupled receptors, and it negatively regulates adenylate cyclase (7). In the midbrain raphe neurons, the receptor is thought to play a role in the feedback regulation of the 5-HT system. mRNA transcripts were found to be most abundant in areas of the limbic system (i.e. hippocampus, septum, and thalamus), cerebral cortex, and raphe nucleus (8 -10).
In addition to dysregulation of the serotonin system, abnormalities of hypothalamic-pituitary-adrenal axis function also have been described to be involved in neuropsychiatric disorders. The hypothalamic-pituitary-adrenal axis regulates the synthesis and secretion of the adrenal corticosteroids, which influence numerous processes in the central nervous system modulating mood, behavior, and neuroendocrine function (11)(12)(13). Corticosteroids exert their action by binding to two types of intracellular corticosteroid receptors, the mineralocorticoid receptor (MR) and the glucocorticoid receptor (GR) (14,15). Upon hormone binding, the receptor translocates to the nucleus and can activate transcription by binding to specific glucocorticoid response elements (GREs) in the regulatory region of target genes. Because of the high degree of homology in the DNA binding domain, MR and GR can interact with the same target DNA sequence (16). Of the two receptors, MR exhibits the highest affinity for corticosteroids, whereas GR mediates a higher stimulation of transcription (17). Besides activation of transcription, GR has been described to repress transcription of genes either via negative GREs or via protein-protein interactions with transcription factors such as AP-1 and NF-B (18 -20).
Both autoradiographic and immunohistochemical studies indicate that the hippocampus contains particularly high concentrations of both MR and GR compared with other brain regions (21,22). Several studies with adrenalectomized rats have shown a suppressive effect of exogenous corticosteroids on 5-HT1A receptor mRNA expression in the hippocampus (23,24). However, little is known about the molecular mechanism underlying this effect.
To investigate the mechanism by which corticosteroids repress rat 5-HT1A receptor expression, we functionally analyzed the promoter region of the rat 5-HT1A receptor gene in COS-1 cells and P19 embryonal carcinoma (EC) cells. Evidence is provided that two NF-B elements in the rat 5-HT1A receptor promoter contribute to induced transcription of the rat 5-HT1A receptor gene. Furthermore, corticosteroids are shown to repress this NF-B-mediated induction of transcription. Remarkably, we observed that only GR and not MR is able to mediate this repressive effect of corticosteroids. Our results suggest that repression of NF-B activity by GR may provide a mechanism underlying the negative action of corticosteroids on 5-HT1A receptor activity. Type Culture Collection (Manassas, VA). Mouse EC cells, P19S1801A1 were cultured as described before (25). Both cell lines were cultured in a 1:1 mixture of DMEM and Ham's F-12 medium (Life Technologies, Inc.), buffered with bicarbonate, and supplemented with 7.5% FCS. CV1-b cells were grown in DMEM (Life Technologies, Inc.), buffered with bicarbonate, and supplemented with 5% FCS. Dextran-coated charcoal-FCS was prepared by treatment of FCS with dextran-coated charcoal to remove steroids, as described previously (26).
Plasmids-The construct Ϫ1588luc containing 1588 base pairs of the rat 5-HT1A receptor 5Ј-flanking region was described previously (38). This 1588-base pairs XbaI/BglII was cloned into pGL3 digested with NheI/BglII. Subsequent deletion constructs were obtained using restriction digestions at specific sites in the promoter; Ϫ1388luc was created by digestion of Ϫ1588luc with HpaI/BglII and ligation into pGL3 digested with SmaI/BglII; Ϫ901luc was created by partial digestion of Ϫ1588luc with StyI, filling-in and ligation into pGL3 digested with SmaI, redigestion with HindIII, and religation; Ϫ208luc was created by digestion of Ϫ1588luc with KpnI/SacII, filling-in, and religation; Ϫ122luc was created by digestion of Ϫ1588luc with BamHI/BglII and ligation into pGL3 digested with BglII; Ϫ81luc was created by digestion of Ϫ1588luc with StyI, filling-in, and digestion with BglII and ligation into pGL3 digested with SmaI/BglII; and Ϫ1588 ⌬901-81 luc was created by partial digestion of Ϫ1588luc with StyI and religation. Ϫ901 365Mluc and Ϫ901 64Mluc were constructed by introducing point mutations into the original promoter constructs by site-directed mutagenesis using the oligonucleotides 5Ј-gagccgaattctacagactaa-3Ј and 5Ј-aactgcaaggagatctacatcgcccctcg-3Ј, respectively. Ϫ901 365/64Mluc was created by digestion of Ϫ901 64Mluc with SacII/HindIII and ligation into Ϫ901 365Mluc digested with SacII/HindIII. The reporter plasmid 2xGREtkluc has been described elsewhere (28). The CMV4 expression vectors containing full-length cDNAs encoding human p65 (RelA), p50 (NF-B1), and GR have been described before (29). The CMV4 expression vector containing full-length cDNA encoding human MR was made by digesting pRShMR, a kind gift from Dr. Evans (San Diego CA), with AvaI, filling-in, digesting with AccI, ligated into AccI/EcoRV-digested pBluescript SK Ϫ . Additionally, SK Ϫ hMR was digested with KpnI/SmaI, and the fragment was ligated into CMV4 digested with KpnI/SmaI. Transient Transfections-For transient transfections, COS-1 cells and P19 cells were cultured in 24-well plates in DMEM and Ham's F-12 medium. DMEM and Ham's F-12 medium supplemented with 5% dextran-coated charcoal-FCS for COS-1 cells and 7.5% dextran-coated charcoal-FCS for P19 cells was used when MR function was studied. Cells were transfected using calcium-phosphate coprecipitation with 0.4 g of luciferase reporter, 0.6 g of PDMlacZ, and 0.2 g of the indicated expression plasmids. pBluescript SK Ϫ was added to obtain a total amount of 1.8 g of DNA/well. After 16 h, the medium was refreshed, and, when indicated, hormone was added. Cells were harvested 24 h later and assayed for luciferase activity using the Luclite luciferase reporter gene assay kit (Packard Instruments, Meriden, CT) according to the manufacturer's protocol and the Topcount liquid scintillation counter (Packard Instruments, CT). Values were corrected for transfection efficiency by measuring ␤-galactosidase activity (30).
Primer Extension Analysis-For primer extension, CV1-b cells were cultured in 6-well plates in DMEM and transiently transfected with 1 g of Ϫ1588CAT reporter in combination with 100 ng of empty expression vector or 100 ng of expression vector encoding p65. The total amount of DNA was adjusted to 2-3 g with pBluescript SK Ϫ . Total RNA was isolated using RNeasy (Qiagen). 36-mers were end-labeled with [␥-32 P]ATP using T4 kinase for 30 min at 37°C. Labeled primers were ethanol-precipitated three times and resuspended in 0.3 M NaAc. 5 ϫ 10 4 cpm of primer was annealed overnight to 8 g of RNA for unstimulated cells and 4 g of RNA for p65 stimulated cells in S1 hybridization buffer (80% deionized formamide, 40 mM PIPES, 400 mM NaCl, 1 mM EDTA). cDNA was generated from the RNA using 20 units of Avian myeloblastosis virus reverse transcriptase at 42°C for 90 min. RNA template was digested with RNase A, and the resulting products were precipitated and run on a 14% denaturing polyacrylamide gel at 300 V, alongside an end-labeled DNA ladder. Subsequently, the gel was dried and autoradiographed.

Activation of the 5-HT1A Receptor Promoter by NF-B-
Although previous studies have shown that addition of corticosteroids to adrenalectomized rats resulted in suppression of 5-HT1A receptor mRNA expression in the hippocampus (23,24), the mechanism underlying this effect remained largely unknown. To elucidate the molecular mechanism by which corticosteroids repress 5-HT1A receptor expression, we functionally analyzed the promoter region of the rat 5-HT1A receptor gene. Sequence analysis of the 1.6-kb promoter did not reveal any GREs. However, sequence analysis did reveal potential response elements for other known transcription factors, including two potential NF-B elements, located at Ϫ365 and Ϫ64 in the promoter. Activation of NF-B in the brain, in response to a variety of stimuli, has been reported (31), and we examined whether NF-B transcription factors were able to influence the transcriptional activity of the 5-HT1A receptor promoter.
To investigate the effect of NF-B, we transiently transfected COS-1 cells and P19 EC cells with several reporter constructs, containing a variety of 5-HT1A promoter deletions (Fig. 1A), in combination with expression vectors encoding the p50 and p65 subunits of NF-B. As shown in Fig. 1 (B and C), NF-B was clearly capable of inducing luciferase activity of the Ϫ1588 promoter construct in both cell lines. Furthermore, deletion of the region between Ϫ122 and Ϫ81 resulted in complete abolishment of the NF-B effect (Ϫ81 luc and Ϫ1588 ⌬901-81 luc). Deletion of the distal region of the promoter from Ϫ1588 to Ϫ122 resulted in promoter fragments, which possessed stronger NF-B-induced activity when transfected into COS-1 cells (Fig. 1B). However, these proximal promoter fragments showed less induction by NF-B when transfected into P19 EC cells (Fig. 1C). This result suggests that the upstream region contains cis-acting DNA sequences that specifically repress transcription depending on the cell line used.
The 5-HT1A receptor promoter is a TATA-less promoter, and multiple transcription initiation sites have been found in the mouse and human 5-HT1A receptor promoter (32). To determine transcription initiation sites in the rat promoter and to investigate whether increased reporter activity actually reflects increased 5-HT1A receptor promoter-directed transcription, primer extension analysis was carried out. A primer directed at the chloramphenicol acetyltransferase coding sequence, just downstream of the 5-HT1A receptor 5Ј-flanking region, was used with mRNA from unstimulated CV1-b cells or mRNA from CV1-b cells stimulated with expression vector encoding p65. In the unstimulated condition, two bands of equal intensity were found representing major transcription initiation sites around position Ϫ90 and Ϫ105 (Fig. 2, lane 1). After stimulation with p65, the lower band (Ϫ90) significantly increased in intensity suggesting recruitment of the transcription machinery specifically at this site (Fig. 2, lane 2). This site is located just upstream of the nearby NF-B binding site at position Ϫ64. Longer exposure of the film revealed multiple minor bands (not shown).
To further examine the importance of the NF-B elements in the regulation of the 5-HT1A receptor promoter, we specifically mutated both NF-B elements in the promoter. To investigate the effect, we transfected COS-1 cells and P19 EC cells with promoter constructs, containing either one or two mutated NF-B elements (Fig. 3A) in combination with expression vectors encoding the p50 and p65 subunits of NF-B. Mutation of either the distal NF-B element located at Ϫ365 (Ϫ901 365Mluc) or the proximal NF-B element located at Ϫ64 (Ϫ901 64Mluc) resulted in a 50% decrease in NF-B-induced promoter activity in both cell lines (Fig. 3, B and C). (II), seen as an additional complex with intermediate mobility, could also be observed. In contrast, the Ϫ64 B element showed only efficient binding of p50 homodimers (I), whereas weak binding of p50/p65 heterodimers (II) could be observed when cell extract containing p65 was used. This specific p50/p65 heterodimer is probably formed with endogenous p50 from COS-1 cells. No binding of p65 homodimers could be observed with this Ϫ64 B element. Both mutated B elements (Ϫ365M and Ϫ64M) as used earlier in the promoter studies were unable to bind any p50-or p65-containing complex (results not shown).
As a control for binding and mobility of the specific B complexes, EMSA was also performed with probes containing B elements from the HIV long terminal repeat and the intercellular adhesion molecule-1 promoter (Fig. 4B). As described before (33), the B element in the HIV long terminal repeat preferentially bound p50 homodimers (I) and p50/p65 heterodimers (II) and more weakly p65 homodimers (III), whereas the B element in the intercellular adhesion molecule-1 promoter preferentially bound p65 homodimers (III) and more weakly p50 homodimers (I) and p50/p65 heterodimers (II). Together these data indicate that specific NF-B complexes are able to bind to the B elements in the 5-HT1A receptor promoter.
Repression of NF-B-induced 5-HT1A Receptor Promoter Activity by Corticosteroids-Previous studies have shown that GR is able to repress transcription of target genes via proteinprotein interactions with other transcription factors, including NF-B. To examine the effect of corticosteroids on NF-Binduced activity of the 5-HT1A receptor promoter, we transfected COS-1 cells and P19 EC cells with three different reporter constructs, containing 5-HT1A promoter deletions (Ϫ1588 luc, Ϫ901 luc, and Ϫ208 luc; see Fig. 1A) in combination with expression vectors encoding the p50 and p65 subunits of NF-B and GR. As shown in Fig. 5A, the NF-B-induced activity of all three promoter constructs could be repressed by GR after treatment of COS-1 cells with dexamethasone. Comparison of the promoter constructs showed that the repression varied from 40% when Ϫ1588 luc and Ϫ901 luc were used to 65% when Ϫ208 luc was used. In P19 EC cells (Fig. 5B), the NF-B-induced activity of both Ϫ1588 luc and Ϫ901 luc was also repressed hormone-dependently by GR to approximately 55%, whereas the activity of the Ϫ208 luc construct was not affected by GR. This was due to a low level of NF-B-induced activity of this construct in P19 EC cells as compared with COS-1 cells (see Fig. 1, B and C).
Because corticosteroids are known to exert their action not only via GR but also via MR, we examined whether MR was also able to mediate this repressive effect of corticosteroids. We induced promoter activity, treatment of the cells with aldosterone or hydrocortisone in the case of MR had no effect on the promoter activity in both cell lines (Fig. 6, A and B). In a control experiment, MR was able to activate transcription from a GREcontaining reporter construct after treatment of the cells with ligand, although the transcriptional activity of MR was found to be less strong than that of GR (results not shown), a phenomenon that has been described earlier (17). Taken together, our results indicate that repression of NF-B activity by GR could be the mechanism by which corticosteroids suppress rat 5-HT1A receptor expression in the brain.

DISCUSSION
To unravel the mechanism underlying the negative action of corticosteroids on serotonin signaling, we investigated the promoter region of the rat 5-HT1A receptor gene. Previous studies in rats showed that the 5-HT1A receptor gene is negatively regulated by corticosteroids (23,24). Because the gene does not contain obvious negative GREs, this negative regulation may be the result of protein-protein interactions of the corticosteroid receptors with other transcription factors. This negative action of GR via direct protein-protein interaction with other transcription factors, such as NF-B and AP-1, is thought to play a pivotal role in the anti-inflammatory action of glucocorticoids (18 -20).
In the present study, we show that NF-B binding sites in the rat 5-HT1A receptor promoter region contribute to induced transcription of the gene, whereas corticosteroids can repress this NF-B-mediated induction of transcription via GR. Mutation analysis showed the importance of two NF-B binding sites in activation of the 5-HT1A receptor promoter. Mutation of either one of the two elements resulted in a 50% decrease in NF-B-induced promoter activity. These data indicate that there is no synergism between the two NF-B elements but just an additive effect. The fact that the B element at Ϫ64 is conserved between mouse and rat and the element at Ϫ365 is conserved between mouse, rat, and human (32) already suggests the importance of these elements in the regulation of the 5-HT1A receptor gene. Functional binding of NF-B subunits was determined by EMSA, and it was shown that the Ϫ64 B element preferentially bound p50 dimers, whereas p50/p65 heterodimers preferentially bound to the Ϫ365 B element. These differences in affinity for NF-B subunits were consistent with binding characteristics as predicted for these sequences, based on binding of NF-B subunits to artificial B oligonucleotides (34).
Promoter deletion analysis revealed the presence of several important regulatory regions in the rat 5-HT1A receptor promoter. The region between Ϫ122 and Ϫ81 was shown to be essential for NF-B-induced activation of transcription. Although the Ϫ81luc construct still contained an NF-B binding site, transcriptional activity was completely abolished. In both the human and mouse 5Ј-flanking sequence, RNA 5Ј end mapping experiments demonstrated the existence of numerous transcription start sites (32). Primer extension analysis of the rat 5-HT1A receptor promoter showed the presence of two major start sites that represent initiation around Ϫ90 and Ϫ105. Both sites were equally used for basal transcription. Stimulation with p65 specifically induced initiation around position Ϫ90. This site is located just upstream of the NF-B binding site located at Ϫ64 and absent from the inactive Ϫ81luc reporter construct, suggesting recruitment of the transcription machinery selectively at this site located around Ϫ90. Because this arrangement of start sites is conserved between mouse and human, it is also likely to have some functional significance. The presence of several transcription initiation sites may provide a mechanism for differential control of 5-HT1A receptor transcription in different cell types and brain regions. Furthermore, deletion of the distal part of the promoter from Ϫ1588 to Ϫ122 resulted in promoter constructs that were more active in COS-1 cells and less active in P19 EC cells. This could be due to the presence of cis-acting DNA sequences in the upstream region that repress transcription specifically in COS-1 cells, suggesting that a different complement of transcription factors is present in P19 EC cells as compared with COS-1 cells.
In both cell lines, GR was shown to repress NF-B-induced transcription of the 5-HT1A receptor after addition of hormone. Both the synthetic glucocorticoid, dexamethasone, and a naturally occurring corticosteroid, hydrocortisone, were able to mediate this effect of GR. In contrast, MR, the other corticosteroid receptor, important for activity in the brain, was unable to repress this NF-B-induced transcription after addition of the mineralocorticoid, aldosterone, or hydrocortisone. In previous studies, similar results have been obtained on repression of AP-1 activity by both receptors. Under conditions in which GR represses AP-1-stimulated transcription, MR was inactive (35,36). Although both receptors can bind to the same target DNA sequence (GRE), only GR has been found to negatively regulate transcription via protein-protein interactions with other transcription factors. These data suggest that negative cross-talk between GR and NF-B may provide a basis for the molecular mechanism underlying the negative action of corticosteroids in the brain.
Besides the 5-HT1A receptor and the corticosteroid receptors, GR and MR, NF-B has also been found to be expressed in the brain, particularly in cortex and hippocampus (37). As in the periphery, the target genes for NF-B in brain also encode proteins involved in inflammation. However, new evidence is accumulating that NF-B also plays a specific role in the brain in processes such as neuronal plasticity and neurodegeneration (31). In addition, it has been shown that expression of the NF-B target gene, interleukin-1␤, is increased during longterm potentiation of synaptic transmission, a process thought to underlie certain forms of learning and memory (27). Obviously, the exact role of NF-B in brain function related to serotonin needs to be further examined.
Because removal of circulating corticosteroids by adrenalectomy causes up-regulation of 5-HT1A receptor expression and addition of exogenous corticosteroids suppresses 5-HT1A receptor expression, it seems clear that changes in hypothalamicpituitary-adrenal axis function can result in dysregulation of the serotonergic system. Controlling hippocampal 5-HT1A receptor expression is one way by which corticosteroids may act to decrease sensitivity to 5-HT in the brain. Based on the role of 5-HT1A receptors in the feedback regulation of the 5-HT system, we hypothesize that 5-HT1A receptors may be downregulated in depression and therefore display an inadequate ability to convert binding of 5-HT to its receptors into an adequate physiological response.