Identification of a retinoic acid response element in the human oxytocin promoter.

Retinoids are known to have profound effects on cellular differentiation and embryo pattern formation. In the adult organism, retinoid acid (RA) receptors are present in a large variety of tissues, including brain. However, little is known of the precise roles of RA at these different sites. In the present study we have identified a novel potential target of RA action by identifying an RA response element (RARE) in the human oxytocin (OT) gene promoter. We have used DNA-mediated gene transfer techniques to introduce various portions of the OT 5'-flanking sequences next to the chloramphenicol acetyltransferase (CAT) gene in neuroblastoma cells. RA elicited a marked stimulation of the transcriptional activity of the OT promoter in cells cotransfected with either the human RA receptor alpha, beta, or gamma. In cells cotransfected with the RA receptor alpha, the ED50 of this response was 5 x 10(-10) M. The RA response could also be conferred to a heterologous promoter independent of orientation. 5'-Deletions as well as site-directed mutations demonstrated that four TGACC motifs, located at -162, -156, -103, and -83 in the OT promoter, are necessary for optimal RA induction. Mutation or deletion of any of these elements reduces significantly the RA response. Interestingly, the first two TGACC motifs overlap with the estrogen response element that we have previously characterized in this gene. Furthermore, the TGACC motif located at -83 overlaps with the CCAAT box. We further demonstrate that in neuroblastoma cells transfected with an RAR alpha expression vector expression of the endogenous OT gene is stimulated greater than 4-fold in response to RA. Our studies constitute the first report of a RARE in a neuropeptide gene and define a mechanism by which OT gene expression can be modulated by retinoic acid.

Retinoids are known to have profound effects on cellular differentiation and embryo pattern formation. In the adult organism, retinoid acid (RA) receptors are present in a large variety of tissues, including brain. However, little is known of the precise roles of RA at these different sites. In the present study we have identified a novel potential target of RA action by identifying an RA response element (RARE) in the human oxytocin (OT) gene promoter. We have used DNA-mediated gene transfer techniques to introduce various portions of the OT 6'-flanking sequences next to the chloramphenicol acetyltransferase (CAT) gene in neuroblastoma cells. RA elicited a marked stimulation of the transcriptional activity of the OT promoter in cells cotransfected with either the human RA receptor a, B, or y. In cells cotransfected with the RA receptor a, the EDao of this response was 6 X 10"' M. The RA response could also be conferred to a heterologous promoter independent of orientation. 6'-Deletions as well as site-directed mutations demonstrated that four TGACC motifs, located at -162, -156, -103, and -83 in the OT promoter, are necessary for optimal RA induction. Mutation or deletion of any of these elements reduces significantly the RA response. Interestingly, the first two TGACC motifs overlap with the estrogen response element that we have previously characterized in this gene. Furthermore, the TGACC motif located at -83 overlaps with the CCAAT box. We further demonstrate that in neuroblastoma cells tranfected with an RARa expression vector expression of the endogenous OT gene is stimulated >I-fold in response to RA. Our studies constitute the first report of a RARE in a neuropeptide gene and define a mechanism by which OT gene expression can be modulated by retinoic acid.
The retinoids form a group of related compounds which exert profound effects on cell growth and differentiation. In the developing chicken limb, a concentration gradient of retinoic acid (RA)' conveys positional information to individual cells along the anterior-posterior axis (for review, see Summerbell and Maden, 1990;Brockes, 1990). In addition, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertkement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Transduction of the RA signal at the level of gene expression involves a growing family of nuclear receptors which are members of the steroid/thyroid hormone receptor superfamily. In mouse and human, three types of RA receptors (denoted RARa, -@, and -7) have been discovered (Petkovich et al., 1987;Giguere et al., 1987;Brand et al., 1988;Benbrook et al., 1988;Krust et al., 1989;Zelent et al., 1989;Kastner et al., 1990). Furthermore, an RARy-related RAR (termed RARG) has been identified in the urodel blastema (Ragsdale et al., 1989). A novel RAR with a substantially different ligand specificity (termed RXRa) has been isolated from a human liver cDNA library (Mangelsdorf et al., 1990). Each member of the RAR family displays a unique pattern of tissue-specific expression in the developing as well as in the adult organism (Ruberte et al., 1990;Kastner et al., 1990). The presence of RARs in various adult tissues is one of the indications that the role of RA extends beyond the developmental period. Specifically, the high amounts of RARa and @ expression in adult brain (Giguere et al., 1987;Benbrook et al., 1988;Zelent et al., 1989) may suggest hitherto unrecognized roles of RA in brain function.
Like other nuclear receptors of the steroid/thyroid hormone superfamily, RA receptors function as ligand-induced tramactivators and modulate the expression of specific genes by binding as a ligand/receptor complex to specific DNA sequences termed RA response elements (RARE). Despite the large spectrum of RA-mediated actions, naturally occurring RAREs have only been identified in a small number of genes. These include the RARp gene itself (de Th6 et al., 1990;SUCOV et al., 1990;Hoffmann et al., 1990), the laminin B1 gene (Vasios et al., 1989), the osteocalcin gene (Schule et al., 1990), and the gene for complement H factor (Munoz-Canoves et al., 1990). Furthermore, in two cases, a thyroid hormone response element was also shown to impart RA responsiveness (Umesono et al., 1988;Bedo et al., 1989). However, no unequivocal RARE consensus sequence could be derived so far. Whereas DNA elements mediating responses to estrogens and glucocorticoids are typically palindromic, several naturally occurring RAREs identified thus far contain two to four direct repeats of the pentamers TGACC, TCACC, or variations thereof, separated by 3-15 nucleotides. The pentamer motif itself is related to estrogen and thyroid hormone response elements, since TGACC is found in the right half of most estrogen or thyroid hormone response elements (Beato, 1989).
We have previously analyzed the steroid hormone regulation of the gene encoding the hypothalamic neuropeptide oxytocin (OT) and identified an imperfect palindrome starting at position -164 that mediates estrogen responsiveness (Richard and Zingg, 1990). In these studies, we also noted the presence of two direct TGACC repeats located 3' to the ERE, which are conserved among the human, rat, and bovine genes. Our studies also showed that these two repeats impart only negligible estrogen responsiveness to the OT gene promoter and do not synergize with the palindromic ERE at -164. We reasoned, therefore, that these two repeats may form part of another unidentified response element. The accumulating knowledge of RARE structures prompted us to test the hypothesis that these two repeats may form part of an RARE. In the present report we examined, therefore, whether OT gene promoter activity was modulated by RA. Our interest in these studies was heightened by the fact that RA receptors are expressed in brain, specifically in hippocampus and hypothalamus (Giguere et al., 1987;Benbrook et al., 1988), yet no RA-responsive neuropeptide genes have been identified thus far. Our studies indicate that the 5'-flanking region of the human OT gene indeed contains a potent and sensitive RARE encompassing four direct TGACC repeats.

MATERIALS AND METHODS
Plasmid Construction-The plasmid pCAT contains the coding sequences for chloramphenicol acetyltranferase (CAT) (Gorman et al., 1982) inserted into vector Bluescript KS' (Stratagene). The plasmid POT-381CAT consists of the BamHI/NcoI fragment (-381 to +36) of the human OT gene (Sausville et al., 1985) inserted with Hind11 linkers at the HindIII site of the CAT gene in pCAT. The plasmid ptkCAT was constructed by removing a 109-base pair segment containing the herpes simplex thymidine kinase (tk) promoter linked to the CAT gene from the plasmid pTEl (Edlund et al., 1985) and ligating it into the BamHI site of Bluescript KS+. The plasmid pOT-381/-49tkCAT consists of the BarnHIISacI fragment of the human OT genomic subclone inserted into the HindIII site of ptkCAT using HindIII linkers. The plasmid pOT-49/-381tkCAT is similar to the previous one, but with the insert in the opposite orientation. The plasmid TREpal-tkCAT was created using a self-complementary oligonucleotide with the sequence 5'-AGC TTC AGG TCA TGA CCT GA-3'. The oligonucleotide was designed such that following selfannealing the double-stranded oligonucleotide contained HindIIIcompatible ends on either side. Using these sites, the oligonucleotide was inserted into the HindIII site of ptkCAT. The plasmid used contained a single insertion of the oligonucleotide. The plasmid RARB-RAREtkCAT (a gift from Dr. P. Chambon, Unit6 184 de YINSERM, Strasbourg, France) contained the RARE of the RARO gene linked to the tk promoter and the CAT gene.
Plasmids pOT-l64CAT, pOT-l55CAT, POT-116CAT, and POT-49CAT are 5'-deletion mutants constructed from the parent plasmid pOT-381CAT, using convenient restriction sites (Richard and Zingg, 1990). The plasmid pOTA-l45/-50CAT corresponds to POT-49CAT with sequences (-164 to -146) of the human OT gene inserted at their original distance from the transcription initiation site. It corresponds, therefore, as its designation implies, to a mutation of POT-164CAT in which OT sequences -145 to -50 have been replaced by Bluescript vector sequences. The plasmid was produced by inserting, at the HindIII site of pCAT, a 209-bp fragment containing human OT gene sequences -164 to +146, followed by 95 base pairs of Bluescript vector sequences, followed by OT gene sequences -49 to +36. This plasmid was constructed as described previously using the polymerase. chain reaction (PCR) technique (Richard and Zingg, 1990).
The plasmids pOTA-103 CAT and pOTA-83 CAT were constructed from POT-164 CAT by oligonucleotide-directed site-specific mutagenesis using PCR, essentially as described by Higuchi (1990). All constructs were verified by dideoxynucleotide sequencing.
Transfections and Chloramphenicol Acetyltransferase Assays-Neuro-2a cells (ATCC CCL 131; American Type Culture Collection, Rockville, MD) were grown and maintained Earle's minimal essential medium supplemented with 10% fetal bovine serum. The cells were plated at a density of 106/10-cm plate the day before transfection. Each plate was transfected with a total of 19 pg of DNA using the calcium phosphate procedure (Graham and Van der Eb, 1973). Unless stated otherwise, cells were transfected with 7 pg of the appropriate CAT construct and 7 pg of an expression vector containing either the human retinoic acid receptor a, p, or y (obtained from the Dr. Pierre Chambon, Strasbourg, France) (Brand et al., 1988, Krust et al., 1989Petkovich et al., 1987) or with 7 pg of salmon sperm DNA. Each transfection mixture also contained 5 pg of the plasmid pCHllO (Pharmacia LKB Biotechnology Inc.). This plasmid contains the structural gene for @-galactosidase under the control of the SV40 early promoter and was used as an internal control for transfection efficiency. Four hours after DNA addition, the cells were subjected to a 3-min glycerol shock, washed with phosphate-buffered saline, and replenished with 10 ml of phenol red-free Earle's minimal essential medium supplemented with 10% dextran-coated charcoalstripped fetal bovine serum. Twenty-four hours after transfection, the medium was changed and vehicle (ethanol, 0.01% final concentration) or retinoic acid M (unless otherwise stated) was added to the cultures. The cells were harvested 40 h after the addition of the DNA, and cellular extracts were prepared as described (Richard and Zingg, 1990). CAT activity was quantitated by measuring the radioactivity of excised bands of the thin layer chromatogram containing either the acetylated or the nonacetylated forms of ["C] chloramphenicol. CAT activity was expressed as a percentage of [' ' Cc] chloramphenicol converted to the acetylated form. The -fold stimulation was calculated by dividing the percent acetylation observed in RA treated cultures by the percent acetylation observed in vehicle treated cultures. Comparisons were always made between plates belonging to the same experimental set. CAT assays were performed with 10 pg protein for 20 min at 37 "C. 0-Galactosidase assays were performed as described (Richard and Zingg, 1990), using 30 pg of protein for 60 min at 37 "C. Within a given experimental group, variation in P-galactosidase activity did not exceed a factor of 2.
Endogenous OT Gene Expression in Neuro-2a Cell.-RNA was extracted from Neuro-2a cells and from BALB/c mouse hypothalami and quantitatively and qualitatively assessed by spectrophotometry at Azso and agarose gel electrophoresis. One pg of total RNA was reverse transcribed in "PCR buffer" (50 mM KC1, 10 mM Tris-HC1 (pH 8.4), 1.5 mM MgCI,, 100 pg/ml gelatin) in the presence of 1 p1 RNasin (28 units/pl; Promega), 1 mM dNTPs, 0.5 pg oligo(dT) primers, and 10 units of reverse transcriptase (Life Sciences, St. Petersburg, FL). The first strand reaction proceeded at 23 "C for 10 min and then at 42 "C for 60 min in a total volume of 20 pl. The reaction was stopped by heating to 94 "C for 5 min. Specific amplification of OT cDNA was achieved by the use of two primers corresponding to parts of exon A and exon C of the mouse OT gene. The exon A primer (forward primer) corresponded to nucleotides 99-115 and exon C primer (reverse primer) corresponded to nucleotides 773-798 of the mouse OT gene (Hara et al., 1991). One hundred pmol of each primer was added to the reaction mixture along with 2.5 units of Taq polymerase (Amplitaq, Perkin-Elmer Cetus Instruments) and 80 p1 of PCR buffer. As an internal control for amplification efficiency and RNA recovery, an additional primer pair corresponding to exons 3 and 5 of the mouse @-actin gene was added to the reaction. The forward primer corresponded to nucleotides 1501-1521, and the reverse primer was complementary to nucleotides 2800-2820 of the pactin gene (Nude1 et al., 1983). Based on the distance between the priming sites on a p-actin cDNA, the predicted size of the amplified band was 766 bp. The reaction conditions for the PCR amplification were: 25 cycles, 94 "C, 45 s; 55 "C, 45 s; 72 "C, 45 s; final extension: 72 "C for 10 min. One-tenth of a PCR reaction was loaded on a 1.5% agarose gel and, following electrophoresis and denaturation by 0.5 N NaOH, transferred to a nylon membrane (Amersham Corp.) and hybridized to a OT-specific oligonucleotide probe corresponding to nucleotides 615-624 of the mouse OT gene (Hara et al., 1991). Hybridization occurred at 42 "C in 0.8 M NaCl and 50% formamide. The washing solution contained 12.5 M NaCl and 0.125% sodium dodecyl sulfate (3 X 10 min at 55 "C).

Co-transfection of RARa, @, or y Expression Vectors with OT PromoterlCAT Constructs
In order to determine whether the transcriptional activity of the human OT gene promoter can be modulated by RA, a segment containing 381 bp of the OT gene 5'-flanking region was inserted upstream of the chloramphenicol acetyltransferase gene (POT-381CAT). This construct was transiently expressed in mouse neuroblastoma cells, with or without cotransfection with the human RAR expression vectors RARa, RARP, or RARr (Petkovich et al., 1987;Giguere et al., 1987;Brand et al., 1988;Benbrook et al., 1988;Krust et al., 1989;Zelent et al., 1989;Kastner et al., 1990). As shown in Fig. lA, in the absence of any co-transfected RAR expression vectors, POT-381 CAT exhibited a readily detectable level of transcriptional activity, with no significant change after RA addition. By contrast, if either RARa, RARP, or RARr was cotransfected with POT-381 CAT, addition of M RA led to a strong and reproducible increase in CAT activity. The maximal increase (RA-treated RAR co-transfected cells uersus untreated, non-RAR co-transfected cells) was 40-fold, using RARa.
In all cases, co-transfection with any of the RAR expression vectors led to a consistent increase of CAT expression in the absence of added ligand. The same phenomenon was observed POT-381CAT (7 pg) was cotransfected with either carrier DNA or an expression vector for the human retinoic acid receptor a, @, or y (7 pg). Cell extracts from either control (-) or retinoic acid M) (+)-treated cultures were assayed for CAT activity. CAT activity is expressed as percent acetylation (means of three independent tranfections). Statistical analysis was by analysis of variance and the Duncan-Kramer multiple range test (Kramer, 1956). B, plots of CAT activity versus concentration of retinoic acid. POT-381 CAT waa cotransfected with the expression vector for the human retinoic acid receptor a (7 pg). Retinoic acid or vehicle (ethanol, 0.01% final concentration) was added 24 h after transfection. Each point represents the mean f S.E. of five independent tranfections. Statistical evaluation was by analysis of variance and the Duncan-Kramer multiple range test (Kramer, 1956). with some other RARE/CAT constructs (Vasios et al., 1989) and may be due to the presence of low levels of retinoids remaining in charcoal-stripped serum and/or to some ligandindependent transactivation activity of RA receptors. Although marked activation was observed with any of the three RA receptor types tested, the strongest effect was observed with RARa. Subsequent experiments were, therefore, only carried out with RARa.
In order to quantitate more precisely RA action, a doseresponse curve was established. As depicted in Fig. lB, the OT RARE was highly sensitive and responded to RA with an EDSO of approximately 5 X 10"' M. A dose as low as lo-" M was able to elicit a statistically significant response ( p < 0.01) and doses of M RA caused a maximal response. We next determined the dose-response relationship with respect to the amount of transfected RAR expression vector. Fig. 1C demonstrates that a maximal response was obtained with 15 pg of transfected expression plasmid, but as little as ZOO ng was sufficient to elicit a statistically significant response to M RA.
To determine whether OT gene 5'-flanking regions could confer RA responsiveness to a heterologous promoter, OT gene sequences spanning from -381 to -49 were inserted in front of the herpes simplex thymidine kinase (tk) promoter, linked to the CAT gene. Independent of orientation, the OT gene sequences were able to confer a 7-fold RA-dependent induction of CAT activity (Table I). This effect was specific and was not observed with the parent plasmid ptk-CAT, which was devoid of any OT gene sequences (Table I).

Comparison with Other RAREs
To judge the effectiveness of the OT gene RARE in comparison with other known RAREs, the RA induction of the tk promoter by the OT RARE was compared with RA induction of the same heterologous promoter by the RARE present in the RARB gene (de Th6 et al., 1990) and a synthetic RARE (TREpal) described by Umesono et al. (1988). As shown in Table I, the -fold induction of both the synthetic RARE as well as the RARO RARE was relatively weak in Neuro-Za

TABLE I The 5"flanking region of the human OT gene confers retinoic acid responsiveness to a heterologous promoter and comparison with known RAREs
Plasmid ptkCAT contains 109 bp of the 5"flanking region of the herpes simplex thymidine kinase (tk) promoter linked to the structural gene for CAT in Bluescript (KS'). pOT-381/-49tkCAT and pOT-49/-381tkCAT contain the human OT sequences from -381 to -49 in front of the t k promoter in forward or reverse orientation, respectively. TREpal-tkCAT contains the rat growth hormone gene thyroid hormone response element linked to tkCAT. This element constitutes the first sequence element shown to act as a RARE (Umesono et al., 1988). RARp-RAREtkCAT contains the RARE of the RARp gene linked to tkCAT. (For sequences, see Fig. 3B). Neuro-2a cells were co-transfected with the above constructs along with 7 pg of RARa. CAT activity is shown as percent acetylation in the presence (+RA) and in the absence (-RA) of lo-' M RA and as RAinduced-fold stimulation (means f S.E. of five independent transfections). Statistical analysis was as in Fig. 1B Delineation of Sequences Necessary for RA Induction 5"Deletion Mutants-To determine more precisely the regions mediating the observed RA induction, 5"deletion mutants were tested for their ability to respond to RA. Deletion of sequences upstream of position -164 (POT-164 CAT) did not significantly affect the RA response. However, deletions extending to -155 or -116 (POT-155CAT and POT-116CAT, respectively) significantly reduced, but did not abolish, the RA response (Fig. 2 A ) . With the latter two constructs, RA induced a low, but reproducible, 2-fold stimulation. Further removal of OT sequences down to position -49 (POT-49CAT) or complete removal of OT sequences (pCAT) led to a complete loss of RA responsiveness.
The region mediating RA responsiveness comprises four direct repeat of the pentamer TGACC (as indicated by black boxes in Fig. 2 A )  repeats (POT-164CAT) exhibited strong RA induction (5-6fold), but removal of the two upstream repeats (POT-155CAT) led to a marked reduction in RA responsiveness (2-fold). Although these results clearly indicated that the two upstream TGACC repeats were necessary, but not sufficient for full RA responsiveness, the importance of the two downstream pentamer repeats remained unclear. Therefore, in order to delineate more precisely the role of the two downstream repeats, we next analyzed the RA responsiveness of constructs containing the two upstream repeats but carrying specific point mutations in either of the two downstream repeats. Mutations in the -1451-50 Region-The experiments in this section were designed to test for each of the two downstream repeats their roles in the context of entire -164/+36 region. In plasmids pOTA-103CAT only two nucleotides of the third TGACC repeat (at -103) were mutated, resulting in the sequence aGAtC. Mutant pOTA-83CAT differs from the wild type with respect to three nucleotides in the fourth TGACC repeat (at -83), changing it into gagCC. In both cases, these point mutations led to a marked reduction in RA inducibility, i.e. from 5-fold to 1.5-and 2.5-fold, respectively. Mutation of the entire stretch from -145 to -50, leaving only the two upstream repeats at their original position, led equally to a drastic decrease of RA responsiveness (1.5-fold induction). Taken together, these data indicate that all four TGACC repeats are required for the full RA response exhibited by the wild type promoter.
Sequence Comparisons-If the corresponding sequences in the human, rat, and bovine genome are compared, it is readily apparent that all four TGACC motives are conserved (Fig.  3A). There is only one exception in the bovine gene, where  the first pentamer is TAACC instead of TGACC. The conservation of the four pentamer repeats is all the more striking since there is considerable sequence diversion in areas surrounding the pentamer repeats.

RA Regulation of the Endogenous OT Gene in Neuro-2a Cells
To assess further the physiological relevance of these findings in a biological context, we examined the endogenous expression of the OT gene by Neuro-2a cells and its regulation by RA. By PCR analysis we were able to demonstrate that Neuro-2a cells express the endogenous OT gene, albeit at a lower level than hypothalamic neurons in uiuo (Fig. 4B). In order to determine whether expression of the endogenous OT gene could also be stimulated by the action of ligand activated RAR, Neuro-2a cell were transfected with an expression vector for the RARa and treated with M RA. As shown in Figs. 4, A and B, this treatment led to a marked increase of endogenous OT gene expression without a notable effect on the @-actin gene.

DISCUSSION
We have delineated an RARE between nucleotides -164 and -49 of the 5"flanking region of the human OT gene. Specific point mutations and 5"deletions indicate that four direct repeats of the pentamer TGACC are necessary for mediating full RA induction. The strong conservation of these pentamer motifs in the human, rat, and bovine genome are a further indication that these elements are of biological importance. Our contention that these elements form a RARE is based on their functional capacity to confer RA responsiveness to the homologous as well as a heterologous promoter in a DNA transfection assay. In analogy to other hormone response elements, it is, therefore, likely that they serve as binding sites for the corresponding receptor/ligand complexes. However, direct demonstration of their capacity to bind RAR/RA complexes will require an in vitro DNA binding assay.
So far only a few other RAREs have been characterized in detail (Fig. 3B). Initial observations by Umesono et al. (1989) indicated that the RARa is able to bind to a synthetic thyroid hormone response element (TGACC GGTCA). Analysis of a naturally occurring RARE in the laminin B1 gene revealed the presence of three direct repeats of TGACC-related elements (TGACC, TAACC, and TCACC) (Vasios et al., 1989). It is of interest to note, that the second element (TAACC) corresponds to the version of the first repeat element present in the bovine OT gene (Fig. 3A). The RARE present in the RARD gene contains a direct repeat of the element GTTCA (corresponding to TGAAC on the opposite strand). Mutation studies demonstrated that both these elements are necessary for RA responsiveness (Sucov et al., 1990). The RARE of the osteocalcin gene contains three occurrences of the TCACC element (one on the sense strand and two on the antisense strand) (Schule et al., 1990). .As shown in the present study, the repeats mediating RA responsiveness in the OT gene consist uniformly of the sequences TGACC (with one exception [TAACC] in the bovine gene). From the information available so far, it can be concluded that positions 1, 3, and 5 of the repeat pentamers are consistently occupied by T, A, and C, respectively, whereas positions 2 and 4 are more variable.
In the previously identified RAREs, the space between the repeat elements varies from 3 to 15 nucleotides. In the present case, the region necessary for maximal responsiveness extends over 83 base pairs and can be considered as consisting of two RAREs, each containing two repeats. The distance between repeats 3 and 4 is 20 bp, which corresponds approximately to two turns of the double helix, assuming a B-DNA conformation. Therefore, the structures are likely to be located on the same face of the double helix. By contrast, repeats 1 and 2 are much more closely spaced and are assumed to lay on opposing faces of the DNA. To what extent each of these pentamer pairs serves as a binding site for a RAR dimer remains to be investigated. Since our deletion and mutation experiments indicated that all four pentamer repeats are necessary for RA action, the two pentamer pairs are likely to act in synergism over the 53-bp distance. Cooperativity of two cis-linked nuclear receptor binding sites has been shown to occur at the level of DNA binding in the case of two tandemly arranged progesterone response elements (Klein-Hitpass et al., 1990;Ponglikitmongkol et al., 1990) as well as at the level of transactivation of transcription in the case of two consecutive estrogen response elements (Ponglikitmongkol et al., 1990). It is tempting to speculate that a similar interaction may take place between the two pairs of pentamer repeats, at the level of receptor/DNA interaction, at the level of transactivation, or both.
In addition to the growing list of demonstrated interactions between transcription factors that bind to neighboring cislinked DNA elements (Schule et al., 1988), there are also several recent examples of direct overlaps of binding sites for distinct transcription factors. These include an overlap between a CAMP response element and a glucocorticoid response element in the pituitary glycoprotein hormone a-subunit gene (Ackerblom et al., 1988) and an overlap between an AP1 site and a RARE in the case of the osteocalcin gene (Schule et al., 1990). Interestingly, the RARE identified in this study represents an example of a dual overlap (Fig. 3A). First, the last two Cs of the TGACC repeat 4 coincide with the first two Cs of the CCAAT box located at -80 in the human OT gene. Being aware of this overlap and in order to avoid interference with the CCAAT box function, our mutation of repeat 4 (pOTA-83 CAT) was designed in such a way that it did not alter the CCAAT box consensus sequence (Lewin, 1983). Second, repeats 1 and 2 form part of the previously identified palindromic estrogen response element (Fig. 3A) (Richard and Zingg, 1990). These overlaps may provide further examples where different regulatory factors modulate expression by binding to common or overlapping sequences (Schule et al., 1990).
At a molecular level, the model system presented in this paper should serve as a useful tool for further investigations of the precise mechanisms of RA receptor/DNA interactions and the mechanisms linking this event to transcriptional activation. Moreover, the present data may serve as an impetus for widening our current concepts of RA physiology and point towards a possible role of RA in the modulation of neuropeptide gene expression and brain function.