The rat arylalkylamine N-acetyltransferase gene promoter. cAMP activation via a cAMP-responsive element-CCAAT complex.

A 10-100-fold rhythm in the activity of arylalkylamine N-acetyltransferase (AA-NAT; EC 2.3.1.87) controls the rhythm in melatonin synthesis in the pineal gland. In some mammals, including the rat, the high nocturnal level of AA-NAT activity is preceded by an ∼100-fold increase in AA-NAT mRNA. The increase in AA-NAT mRNA is generated by norepinephrine acting through a cAMP mechanism. Indirect evidence has suggested that cAMP enhances AA-NAT gene expression by stimulating phosphorylation of a DNA-binding protein (cAMP-responsive element (CRE)-binding protein) bound to a CRE. The nature of the sites involved in cAMP activation was investigated in this report by analyzing the AA-NAT promoter. An ∼3700-base pair fragment of the 5′-flanking region of the rat AA-NAT gene was isolated, and the major transcription start points were mapped. The results of deletion analysis and site-directed mutagenesis indicate that cAMP activation requires a CRE·CCAAT complex consisting of a near-perfect CRE and an inverted CCAAT box located within two helical turns.

The rhythmic nocturnal increase in plasma levels of melatonin (1) in mammals is due to norepinephrine (NE) 1 stimulation of melatonin production in the pineal gland. NE release is regulated by the endogenous circadian oscillator in the suprachiasmatic nucleus (2), which is connected to the pineal gland by a multisynaptic pathway. NE controls melatonin production by regulating the activity of the penultimate enzyme in the melatonin biosynthetic pathway, serotonin N-acetyltransferase (arylalkylamine N-acetyltransferase (AA-NAT), EC 2.3.1.87) (3). The second messengers involved are cAMP and Ca 2ϩ i (3)(4)(5)(6)(7). The importance of transcriptional events in the regulation of AA-NAT activity varies remarkably on a species-to-species basis (8). An absolute requirement for de novo transcription is most evident in the rat (8,9), where an ϳ100-fold increase in AA-NAT mRNA is required for the ϳ100-fold increase in AA-NAT activity to occur. The mechanism involved in turning on expression of the AA-NAT gene does not require de novo protein synthesis (9). Rather, it appears to be initiated by cAMPdependent phosphorylation of cAMP-responsive element (CRE)-binding protein (CREB) (10) or another member of this growing family (11,12).
The molecular basis of cAMP stimulation of expression of the rat AA-NAT gene has not yet been investigated at the level of the promoter, and it is not known whether cAMP acts directly or indirectly on the AA-NAT gene. Here we describe the isolation of the rat AA-NAT promoter region and report the results of a functional analysis focused on the question of NE-and cAMP-dependent activation. The results of this study indicate that cAMP can effect gene activation from the AA-NAT promoter. Furthermore, it appears that cAMP acts through a CRE⅐CCAAT complex, which consists of a near-perfect CRE located within two helical turns of an inverted CCAAT box.

EXPERIMENTAL PROCEDURES
Ligation-mediated PCR-AA-NAT promoter sequences were identified using the Promoter Finder™ DNA Walking kit (CLONTECH, Palo Alto, CA) in conjunction with the nested AA-NAT primers 652, 581, and 587 (see below) derived from the 5Ј-flanking regions of AA-NAT. The PCR conditions were recommended by the manufacturer. Genomic fragments were subcloned into pCR3 (Invitrogen, San Diego, CA) for sequencing by the dideoxynucleotide chain termination method (13). To generate promoter/reporter hybrid constructs, different PCR-generated fragments of the AA-NAT promoter region were introduced into the XbaI site in pCAT-Basic (Promega, Madison, WI) or into the NheI site in pGL3-BASIC (Promega). These vectors carry the bacterial chloramphenicol acetyltransferase (CAT) and firefly luciferase (LUC) reporter genes, respectively.
Total RNA Isolation and S1 Nuclease Analysis-For S1 nuclease protection analysis, a 289-nucleotide end-labeled single-stranded probe was synthesized that was complementary to positions Ϫ207 to ϩ82 in the AA-NAT gene (relative to the start of transcription); synthesis was by asymmetric PCR driven by primer 581 (see below). The probe was gel-purified and added (50,000 cpm) to 30 g of total pineal RNA in 20 l of S1 hybridization buffer (80% deionized formamide, 40 mM Pipes, pH 6.4, 400 mM NaCl, and 1 mM EDTA, pH 8). The mixture was then denatured (10 min, 65°C) and hybridized overnight at 44°C. S1 nuclease digestion was carried out by adding 300 l of S1 nuclease buffer (0.28 M NaCl, 50 mM sodium acetate, pH 4.5, 4.5 mM ZnSO 4 , 10 g of sheared salmon sperm DNA, and 240 units of S1 nuclease). After a 1-h incubation at 37°C, 80 l of stop buffer (4 M ammonium acetate, 20 mM EDTA, pH 8, and 40 g/ml tRNA) were added. After ethanol precipitation, samples were boiled and electrophoresed on a 6% sequencing gel. A dideoxynucleotide sequencing reaction of Ϫ267/CAT primed with primer 581 was run in parallel to locate the transcription start point.
Tissue Dissociation and Transfection-Dissociated rat pinealocytes (5 ϫ 10 6 cells) were prepared by trypsinization essentially as described (14) with minor modifications. Briefly, 50 pineal glands (Taconic Farms Inc., Germantown, NY) were separated from residual afferent nerve fibers and the surrounding leptomeninges, partially teased apart, and rinsed in 10 ml of BGJb (Life Technologies, Inc.). Glands were incubated (37°C, 95% O 2 and 5% CO 2 ) in 5 ml of DMEM containing 0.2% trypsin and 40 g/ml DNase I (Boehringer Mannheim) for 50 min. Trypsinization was ended by the addition of 10 ml of DMEM supplemented with 10% fetal calf serum, 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin (DMEMϩ). Partially dissociated glands were allowed to sediment and triturated in 3 ml of DMEMϩ supplemented with 70 g of DNase I. Pineal cells were passed through a small funnel-shaped mesh (125-M opening; Bellco Glass, Inc., Vineland, NJ) to a new tube, spun at 1000 ϫ g for 5 min, and plated in two wells of a six-well plate (Costar Corp., Cambridge, MA) in 2 ml of DMEMϩ. Twenty-four hours later, cells in suspension were harvested, collected, and resuspended in serum-free Opti-MEM (Life Technologies, Inc.; ϳ0.5 ϫ 10 6 cells/ml); and 0.4-ml samples were transferred to individual wells in a 24-well plate (Costar Corp.). Transfections were performed by overlaying the cells with a precipitate of 3 l of Lipo-fectAmine™ (Life Technologies, Inc.) and 2 g of DNA for 30 min before adding ϳ2 ϫ 10 8 adenovirus shuttle particles to enhance transfection efficiency (15). Eighteen hours later, 0.5 ml of DMEMϩ were added. Where indicated, dibutyryl cAMP (Bt 2 cAMP) was added to the cultures at this point. Cells were harvested 48 h later. CAT assays were performed as described (16). Luciferase activity was measured with the luciferase assay system (Promega) according to the manufacturer's recommendation. Transfection of primary rat pituitary cells or fibroblasts was done using the same procedure as described above. COS-7 and C6 cells (American Type Culture Collection, Rockville, MD) were transfected using LipofectAmine following the manufacturer's recommendations.
Oligonucleotides-The synthetic DNA oligonucleotides and primers referred to in this study were synthesized using an Applied Biosystems 381B DNA synthesizer. Nested primers for promoter cloning were as follows EMSA-Pineal glands were quick-frozen on dry ice. To prepare extracts, a 30-l sample of ice-cold buffer C (20 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 0.42 M NaCl, 0.2 mM EDTA, 1 g/ml aprotinin, 1 g/ml leupeptin, 1 mM sodium fluoride, 5 M sodium orthovanadate, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, and 25% glycerol) was added to two glands. The glands were homogenized on ice with 20 strokes of a tight-fitting 1.5-ml microtube pestle. Lysates were centrifuged at 12,000 rpm for 10 s, vortexed, and refrozen on dry ice for 5 min. They were next incubated in ice water for 15 min and centrifuged at 12,000 rpm for 15 min at 4°C. This procedure extracted 40 -50 g of soluble protein/pineal gland. EMSA was performed using a 32 P-radiolabeled double-stranded oligonucleotide probe containing the natCRE or natCCAAT sequence (see Fig. 1) and 3 l of whole pineal extract as described previously (17). Competition EMSA was performed as described above with a 200-fold molar excess of unlabeled double-stranded oligonucleotides added before the probe or 1 l of one of the following antisera (added 20 min after the probe): anti-c-Fos- Site-directed Mutagenesis-In vitro mutagenesis of the CRE and CCAAT cis-acting elements within the AA-NAT promoter was done with the QuickChange™ site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's recommendations. Initial screening of putative mutant clones was performed by BsaHI and XbaI restriction endonuclease digests, respectively. Mutations were then confirmed by dideoxynucleotide sequencing. Promoter fragments were subsequently excised with XbaI and inserted into an NheI-linearized pGL3-BASIC reporter vector.

RESULTS
Isolation of the Rat AA-NAT Promoter-Putative regulatory regions in the sequence immediately upstream of the AA-NAT coding sequence were isolated using a ligation-mediated PCRbased technique as described under "Experimental Procedures." This generated an ϳ3700-bp genomic fragment that contains the first intron (intron 1; Fig. 1A) located in the 5Ј-untranslated region after position ϩ114; the intron is 1628 bp long. 2 The 2160-bp putative promoter sequence was analyzed to determine if it contains known transcription factor-binding sites (Fig. 1A). This failed to identify a canonical TATA box; however, an A/T-rich sequence resembling a TATA box (ATTA-GAAT) is located 38 bp upstream of the start of transcription, as defined below. This putative TATA box is preceded by a GC-rich region (positions Ϫ47 to Ϫ65), which could function as an SP1-binding (18) and/or AP-2-binding (19) site. An inverted CCAAT box (natCCAAT) exists at position Ϫ120. In addition, we detected an alternate purine-pyrimidine repeat (20) between positions Ϫ1165 and Ϫ1104, an AP-1-like element (21) at position Ϫ32, and a CRE-related sequence at position Ϫ139 (natCRE). The natCRE sequence (TGACGCCA) closely resembles (seven out of eight) the perfect CRE consensus sequence (TGACGTCA) (22,23).
Determination of the Transcription Start Point-The transcription initiation site was established using S1 nuclease protection analysis of total RNA isolated from Bt 2 cAMP-treated pineal glands. Three transcription start points were detected (Fig. 1B), the strongest of which generates the longest leader sequence (173 nucleotides). The leader sequences generated by the weaker start points are 148 and 136 nucleotides long. The upstream initiation site is located within an apparent initiator consensus sequence, characterized by an initiating A with a 3Ј-pyrimidine-rich flanking sequence ( Fig. 1C) (24,25).
Functional Characterization of the AA-NAT Promoter-Regions necessary and sufficient for AA-NAT promoter activity were identified using partial promoter constructs. These contained decreasing portions of the longest 5Ј-flanking genomic fragment placed in front of the CAT reporter gene.
Primary pinealocytes were transiently transfected with this series of constructs and tested for basal and stimulated CAT activity (Fig. 2). CAT activity was low in untreated cells in all constructs. Bt 2 cAMP treatment produced comparable levels of CAT expression from a family of intronless promoter deletion constructs with 5Ј-boundaries ranging from positions Ϫ2160 to Ϫ267 and encompassing sequences up to position ϩ82. Similar results were obtained following treatment with NE or a selective ␤-adrenergic agonist (isoproterenol), but not following selective ␣-adrenergic treatment (phenylephrine ϩ propranolol; data not shown). Maximum Bt 2 cAMP induction was achieved using a promoter construct that begins at position Ϫ207, spans the entire length of intron 1, and ends just prior to the first codon (Ϫ207/iCAT). However, a similar construct containing Ͼ90% of intron 1 but no 5Ј-untranslated region (ϩ371/CAT) failed to drive detectable CAT enzyme activity, indicating that the key elements promoting Bt 2 cAMP responsiveness are not likely to reside in intron 1. Taken together, these results suggest that the region between positions Ϫ267 and ϩ82 is sufficient for conferring cAMP responsiveness. This region contains the natCRE and natCCAAT sequences described above.
The Ϫ267/CAT construct fragment was also able to drive cAMP-dependent expression of CAT when tested in non-pineal cells, including rat glioma C6, COS-7, and primary rat pituitary cell cultures (Table I). It follows that this fragment does not confer tissue specificity, nor does it require pineal glandspecific proteins to support cAMP responsiveness.
The capacity of natCRE to recruit specific DNA-binding proteins was tested using a 30-bp-long double-stranded oligonucleotide centered around position Ϫ139; this probe does not contain the natCCAAT box. EMSA analysis revealed that this natCRE oligonucleotide formed complexes with pineal proteins (Fig. 3A). The patterns of retarded species were not dramatically different if extracts were prepared from pineal glands harvested during the day or night (data not shown), indicating that similar or identical binding proteins are present at both times.
To determine if this EMSA pattern reflected specific interactions, excess unlabeled CRE-containing oligonucleotides (somatostatin or c-fos CRE) were added (Fig. 3A); this blocked the appearance of most retarded species. In contrast, excess unlabeled oligonucleotides containing either CCAAT-binding transcription factor-or AP-1-binding sites did not alter this pattern. Accordingly, formation of these complexes is CRE-dependent.
To test for the presence of specific transcription factors among the proteins associated with natCRE, selected antisera were used (Fig. 3A, lanes 6 -12). An anti-CREB antiserum modified the EMSA pattern by preventing the formation of two nucleoprotein complexes (asterisks). However, antisera directed against several bZIP-containing transcription factors, belonging to the Fos, Jun, and CREM families, did not. Taken together, these results indicate that natCRE can specifically interact with several pineal proteins, including a CREB-like protein(s).
The function of the natCRE sequence was tested by sitedirected mutagenesis in the context of the minimal promoter construct Ϫ267/CAT. The mutation involved three base Genomic fragments containing AA-NAT promoter sequences were isolated as described (see "Experimental Procedures") and sequenced. Shown is a partial result of a computer-assisted search for putative regulatory elements. The alternate purine-pyrimidine tract between positions Ϫ1165 and Ϫ1104 (boldface) and other putative regulatory sites within the proximal promoter region (boldface and underlined) are identified. Open inverted triangles point to the proximal GC-rich region, and the putative TATA box located immediately downstream is double-underlined. Closed inverted triangles mark the three transcription start points as identified in B. The first nucleotide of the longest leader is used throughout this study as the ϩ1 reference position. The location of the first intron within the 5Ј-untranslated region is indicated, as is the position of the first codon. B, shown is the S1 nuclease analysis of the transcription start point. An antisense single-stranded DNA probe was generated by asymmetric PCR, as indicated in the upper diagram, with the antisense radiolabeled primer 581 (positions ϩ82 to ϩ63 in A), located between the end of the available cDNA (9) and intron 1, and trace amounts of unlabeled primer 661 (positions Ϫ217 to Ϫ191). The probe was hybridized to transfer RNA (lane 1) or to total pineal RNA from pineal glands stimulated with Bt 2 cAMP (1 mM) for 0, 1.5, and 3 h (lanes 2-4, respectively) and then exposed to S1 nuclease digestion. Protected products were electrophoresed through an 8 M urea-polyacrylamide sequencing gel. A dideoxynucleotide sequencing reaction of construct Ϫ267/CAT primed with primer 581 was run in parallel and served as a molecular weight standard (ACGT). C, the rat AA-NAT gene contains a putative initiator sequence. Known initiator (Inr) elements and the hypothetical initiator are aligned around the AA-NAT transcription start point 1 (TSP 1). As true in most cases, an adenine initiates AA-NAT transcription, and the downstream sequence is rich in pyrimidines. TdT, terminal deoxynucleotidyltransferase; Ad, adenovirus; HIV-1, human immunodeficiency virus type 1; dhfr, dihydrofolate reductase. changes (TGACGCCA 3 TTAAACCA), including a G/T transversion at position Ϫ141, which disrupts CRE function (26). The triple mutation abolished binding to specific nuclear proteins (data not shown). In transfection studies using primary pinealocytes, the CRE mut /CAT reporter construct partially reduced the response to Bt 2 cAMP (60 -80% inhibition) (Fig. 3B) as compared with the response exhibited by the unmutated control. The possibility that this apparent inhibition was due to uncontrolled reporter gene-dependent artifacts was ruled out using equivalent firefly luciferase constructs, indicating that the triple mutation partially reduced effects of cAMP (Fig. 4B, wt/LUC versus CRE mut /LUC).
The above results indicate that natCRE plays a key role in translating the adrenergic stimulus into an increase in AA-NAT gene transcription, consistent with predictions regarding the role of CREB phosphorylation in activation of the rat AA-NAT gene (10). However, a discrepancy is apparent: the triple natCRE mutation completely abolished binding, but only partially abolished the effects of Bt 2 cAMP on promoter activity. This raised the possibility that cAMP-mediated gene activation requires another element(s) in the Ϫ267 construct. The basis of this residual responsiveness was examined below.
The Inverted CCAAT Box at Position Ϫ120 Is Required for Full cAMP Responsiveness-As indicated above, an inverted CCAAT box is located downstream and two helical turns away from natCRE. The CCAAT box is of special interest because of two observations: a CRE⅐CCAAT complex in the fibronectin gene promoter is necessary for full cAMP activation of gene expression (27)(28)(29), and an inverted CCAAT box mediates cAMP activation of expression of the apparently CRE-less tryptophan hydroxylase promoter in primary pinealocytes (30).
To test whether natCCAAT binds pineal proteins, we analyzed the region containing the inverted natCCAAT sequence by EMSA using whole pineal protein extracts. The doublestranded oligonucleotide used did not contain natCRE. A distinct nucleoprotein complex was generated using either day or night pineal extracts. Proteins in this complex are referred to here as CCAAT-binding proteins (CATBPs) (Fig. 4A). Binding specificity was demonstrated by successful competition with a 200-fold molar excess of the wild-type natCCAAT sequence (Fig. 4A, compare lanes 7 and 8). In contrast, binding was not inhibited by double-stranded oligonucleotides containing either of two mutated natCCAAT oligonucleotides (Fig. 4A, lanes 9 and 10) or by AP-2 or AP-1 sequences (lanes 11 and 12). These studies indicate that natCCAAT specifically binds to pineal CATBPs.
The contribution of the natCCAAT site to gene expression was determined by site-directed mutagenesis of position Ϫ120 within the minimal promoter (construct Ϫ267) using the firefly luciferase reporter system. Promoter activity of the resulting construct (CCAAT mut /LUC) was reduced by 60 -80% when compared with that obtained using wt/LUC and CRE mut /LUC (Fig. 4B).
These observations raise the possibility that both natCRE and natCCAAT are important for inducible promoter activity. To determine whether both sites are required to achieve full transactivation, we generated a double mutant without functional natCRE and natCCAAT sites ((CRECCAAT) mut /LUC). Whereas the single mutant constructs CRE mut /LUC and CCAAT mut /LUC were partially active, the double mutant (CRECCAAT) mut /LUC was essentially refractory to cAMP stimulation (Fig. 4B). This supports the interpretation that both sites participate in cAMP regulation of the AA-NAT gene.
When the natCRE and natCCAAT sites are considered within the context of coiled DNA, their centers are separated by approximately two turns of a helix and are therefore likely to be in phase. To address the question of whether this precise alignment is crucial for cAMP-inducible promoter activity, phasing was disrupted by inserting a 5-bp sequence between the two elements (Fig. 4B, CRECCAATϩ5/LUC). This did not affect promoter activity, indicating that precise phasing between bound factors does not appear to be important when the distance between these sites is not significantly affected. Nested fragments of the AA-NAT promoter were generated by PCR, placed in front of the bacterial CAT gene, and transfected into primary pinealocytes as described (see "Experimental Procedures"). Eighteen hours later, individual cultures were stimulated with Bt 2 cAMP (DBcAMP; 1 mM) or left untreated. Whole cell extracts were prepared 48 h later, and CAT enzyme activity was measured as described (16). Transfection efficiency was assessed by cotransfection with Rous sarcoma virus-␤-galactosidase. Results represent data collected from five independent experiments. APP, alternate purine-pyrimidine repeat.

DISCUSSION
Transcription plays a pivotal role in the nocturnal stimulation of AA-NAT in the rat. This is evident from the findings that NE and cAMP protagonists, including Bt 2 cAMP, increase AA-NAT mRNA levels ϳ100-fold (9) and that the transcription blocker actinomycin D blocks the NE 3 cAMP stimulation of AA-NAT mRNA and activity (3,9). We have suspected that cAMP acts to increase AA-NAT mRNA through cAMP-dependent phosphorylation of CREB resident at a CRE site. This was based on the evidence that NE acts through a cAMP mechanism to phosphorylate CREB in the pineal gland and that the protein kinase A antagonist (R p )-8-CPT-cAMP-S inhibits cAMPdependent induction of AA-NAT activity (10).
Turning on Expression of the AA-NAT Gene-The studies presented here appear to support this notion because a nearperfect CRE (natCRE) is found in a 2160-bp fragment of the rat AA-NAT promoter. Functional dissection by deletion analysis revealed that an natCRE-containing region from positions . Unlabeled oligonucleotide competitors (as indicated) were used on a pool of day and night pineal protein extracts as described for Fig. 3. The migration of the specific CATBP-containing complexes is indicated. B, effect of single versus double natCRE⅐natCCAAT mutagenesis on AA-NAT promoter cAMP responsiveness. The natCCAAT element was mutagenized in the wild-type (CCAAT mut /CAT) and natCRE mutant (CRECCAAT mut /CAT) vectors. An additional construct (CRECCAATϩ5/CAT) contained a 5-bp insertion between natCRE and natCCAAT. The mutated promoters were fully sequenced, spliced in front of the luciferase reporter gene (to yield CCAAT mut /LUC, (CRECCAAT) mut /LUC, and CRECCAATϩ5/LUC), and transfected into primary pinealocytes. Parallel transfections were carried out with wt/LUC and CRE mut /LUC. In selected experiments, cultures were cotransfected with a CAT reporter vector driven by the dynorphin promoter construct to provide an internal standard for transfection efficiency. Cultures were stimulated with Bt 2 cAMP (DBcAMP; 1 mM) or left untreated for 48 h before assessing luciferase and CAT activities. Similar results were obtained in three different experiments utilizing two independent isolates of mutant clones.
Ϫ267 to ϩ82 confers cAMP-dependent responsiveness to either CAT or luciferase reporter genes in primary pinealocytes. Furthermore, the natCRE core sequence binds a CREB-like moiety present in whole pineal protein extracts, and mutagenesis of the natCRE site reduces promoter activity and binding, indicating that it functions as a bona fide CRE. Accordingly, it appears very likely that natCRE mediates cAMP stimulation of expression of the rat AA-NAT gene, presumably in response to cAMP-dependent phosphorylation of resident CREB molecules.
In addition to this site, a neighboring sequence appears to be involved in cAMP responsiveness because disruption of nat-CRE in the Ϫ267/ϩ82 region of the promoter does not completely abolish the response to NE or Bt 2 cAMP. The most likely element involved in this residual cAMP responsiveness is the inverted CCAAT box (natCCAAT) at position Ϫ120. natCCAAT represents a perfect match in reverse orientation to a known regulatory site referred to as the CCAAT box. Such sites are found in many promoters and appear to control gene expression through interaction with members of a growing family of different CATBPs (31)(32)(33)(34)(35). The natCCAAT element appears to exert a positive effect on transcription because site-directed mutagenesis abolished binding activity and decreased the level of reporter gene activity under both basal and stimulated conditions (Fig. 4B). In addition, a double mutant with a disrupted CRE⅐CCAAT complex displayed only basal luciferase activity after Bt 2 cAMP stimulation. This finding, together with the partial decrease obtained after mutagenesis of either site alone, is consistent with the hypothesis that both elements are necessary to achieve full activation of the AA-NAT promoter by cAMP.
It should be noted that there are reports in the literature of CRE-CCAAT cooperation and of cAMP activation being mediated by a CCAAT box. As indicated above, a CRE⅐CCAAT complex is found in the fibronectin gene promoter (27)(28)(29), in which occupancy of the CCAAT box is facilitated by the CRE. In addition to this example of a CRE-CCAAT interaction, there also is evidence that the CCAAT box can mediate activation of gene expression in the absence of a CRE; this comes from the tryptophan hydroxylase gene (30). This points to the possibility that cAMP may act through CATBPs, perhaps via phosphorylation, as it does through CREB to control gene expression. It is of further interest to note that tryptophan hydroxylase is very strongly expressed in the pineal gland, suggesting that CAT-BPs may play a common role in this tissue in gene expression.
Turning Off Expression of the AA-NAT Gene-This investigation focused on the mechanisms through which cAMP activates expression of the rat AA-NAT gene and thereby generates the rhythmic ϳ100-fold increase in AA-NAT mRNA. Mention should be made here of some of the possible mechanisms that turn off expression of this gene, based on the structure of the promoter revealed in this study and current thinking about this issue. One mechanism that will significantly reduce expression of this gene involves a decrease in cAMP, which would terminate positive downstream effects of this second messenger. An additional theoretical mechanism involves cAMP-dependent coinduction of negative transcription factors such as Fra-2 (Fos-related antigen-2) (17) and the inducible cAMP early repressor (ICER) (36), which could repress transcription through binding to AP-1 and CRE sites, respectively.
Levels of mRNA encoding both factors increase in the pineal gland during the night period of a typical lighting cycle providing 10 -12 h of darkness (17,36). The increase in fra-2 mRNA drives a Ͼ100-fold rhythm in Fra-2 protein that is generally similar to the increase in AA-NAT activity. We suspect that Fra-2 protein binds to AP-1 sites in the AA-NAT promoter, negatively affecting transcription. One such site, identified above at position Ϫ32, is located in close proximity to the major transcription start point. Fra-2 strongly binds to AP-1 sites, but does not possess a strong transactivation domain (37). Binding of Fra-2-containing AP-1 complexes at this location could conceivably disrupt the assembly of the basic transcription machinery, inhibiting cAMP activation of AA-NAT expression. Inhibition of Fra-2 protein synthesis may explain why the cAMP-induced increase in AA-NAT mRNA is greater if protein synthesis is blocked (9).
Accordingly, it is possible to hypothesize that Fra-2 could be part of a complex AA-NAT gene regulatory mechanism controlled by cAMP, in which cAMP rapidly turns on expression of the AA-NAT gene through phosphorylation of CREB and activation of the natCRE⅐CCAAT complex and coincidently induces expression of the fra-2 gene and accumulation of Fra-2 protein, which then progressively turns off expression of the AA-NAT gene.
In contrast to the close association of the rhythm in fra-2 mRNA and Fra-2 protein, the rhythm in ICER mRNA is not associated with a remarkable increase in ICER protein; rather, under typical lighting schedules, ICER protein appears to be relatively constant and stable (38). Accordingly, it seems reasonable to suspect that ICER and CREB proteins are present at all times of the day and that they compete for CRE occupancy. As a result, the relative abundance of ICER and CREB pools at these sites becomes an important factor because it could influence the magnitude of the AA-NAT response. Although the relative amount of ICER and CREB might not change under typical fixed laboratory lighting schedules, their relative abundance appears to change in response to very long and very short nights; for example, ICER levels are highest if animals are maintained in lighting cycles with 20-h dark periods (38). This could gradually increase the ICER/CREB ratio and might influence the magnitude of the AA-NAT rhythm. Such a mechanism could underlie seasonal changes in melatonin production.
Potentially Important Features of the AA-NAT Promoter-Our analysis revealed two additional features of the AA-NAT promoter that deserve comment. First, in addition to the data obtained by S1 nuclease protection analysis, two observations are consistent with the identification of the major transcription start point. One is the location, 38 bp upstream, of a TATA-like element. The other is the location of the transcription start point within a putative initiator sequence. This sequence belongs to a family of RNA polymerase II start sites (24,25) and is present in other dynamically regulated genes, including homeotic genes and genes expressed during immunodifferentiation (for review, see Ref. 39). It seems possible that the dynamic regulation of AA-NAT and these genes may involve interactions of this element with the same or similar nuclear factors (40).
A second point relates to the issue of translational control and the finding that large changes in sheep AA-NAT activity and protein 3 can occur with little or no change in levels of AA-NAT mRNA. It is reasonable to suspect that this translational control might be a conserved feature of AA-NAT regulation. In this regard, some features of the rat AA-NAT 5Јflanking region are relevant. Translation is known to be influenced by long leaders with extensive secondary structure (41,42). Such extensive secondary structure is likely to form in the 173-bp leader sequence according to computer analysis of this region (43); this predicts the formation of an elongated stem-loop structure with a free energy of Ϫ54.1 kcal/mol (data not shown). It will be of interest to investigate the possible role of the leader sequence upon translational regulation of AA-NAT.
Summary-The promoter of the AA-NAT gene has been isolated and partially characterized. It appears that a CRE and an adjacent inverted CCAAT box function as a combined target site for the adrenergic signaling cascade that activates expression of this gene. The pattern of expression of the AA-NAT gene is similar to that of an immediate-early gene in that it is rapidly activated in the absence of de novo protein synthesis (9). Thus, rapid induction of AA-NAT gene transcription appears to rely on adrenergically induced post-translational modification and/or recruitment of pre-existing trans-acting factors belonging to the CREB and CATBP families.