Retinoic acid-mediated activation of the mouse renin enhancer.

Previous studies demonstrate that the mouse renin gene is regulated by a complex enhancer of transcription located 2.6 kilobases upstream of the transcription start site which is under both positive and negative influence. We demonstrate herein that a positive regulatory element (Eb) is repeated 10 bp upstream (Ec), and both are required for baseline activity of the enhancer. The Eb and Ec core sequences are identical to the consensus sequence for the nuclear hormone receptor superfamily of transcription factors, and transcriptional activity of constructs containing the enhancer is increased after treatment with retinoic acid. Maximal induction requires both Eb and Ec. Expression of endogenous renin and a renin-promoter controlled transgene in As4.1 cells, and kidney renin mRNA in C57BL/6J mice was induced after retinoid treatment. Gel mobility supershift analysis revealed the binding of RARalpha and RXRalpha to oligonucleotides containing both Eb and Ec. Reverse transcriptase-polymerase chain reaction analysis revealed that As4.1 cells express both receptor isoforms, along with RARgamma, but do not express RARbeta, RXRbeta, or RXRgamma. Co-transfection of an expression vector encoding wild-type RARalpha increased enhancer activity, whereas a dominant negative mutant of RARalpha significantly attenuated retinoic acid-induced activity of the enhancer. These results demonstrate the importance of the Eb and Ec motifs in controlling baseline activity of the renin enhancer, and suggest the potential importance of retinoids in regulating renin expression.

The renin-angiotensin system is a critical regulator of arterial pressure and electrolyte homeostasis and is required for continued development of the kidney after birth. The cleavage of angiotensinogen by renin is thought to be the rate-limiting step in the biosynthesis of angiotensin II and is tightly regulated. Transcription of the renin gene, storage and processing of renin in juxtaglomerular cell secretory granules, and secretion of renin into the systemic circulation, each dictate the level of angiotensin II produced. Although the regulation of the renin gene has been studied for many years, the molecular mecha-nisms controlling its cell-specific expression and regulation in response to physiological cues remains incomplete.
Recent studies have identified an enhancer of transcription located upstream of the renin gene which can markedly induce transcription of renin promoter reporter constructs when transfected into As4.1 cells, a renin expressing tumor cell line isolated from the kidney thought to be derived from juxtaglomerular cells (1,2). This enhancer, located ϳ2.6 kb 1 upstream of the mouse renin gene is partially homologous to a sequence located ϳ12 kb upstream of the human renin gene (3). We previously used mouse/human chimeric enhancers spanning the conserved and nonconserved region to identify important sequences controlling expression (4). Those studies revealed that a 40-bp segment (m40) in the promoter proximal region of the mouse renin enhancer was required for maximal activity. The m40 segment contained two regulatory elements. The first sequence, element a (Ea), bound the factor NF-Y and acted as a transcriptional repressor because mutations abolishing binding of NF-Y significantly stimulated enhancer-mediated transcription. The second sequence, element b (Eb), was required for maximal activity of the enhancer, and its mutagenesis essentially abolished enhancer activity. Given that Ea and Eb overlapped, we hypothesized that NF-Y blocks enhancer activity by preventing the binding of transcription factors to Eb. This is supported by experiments in which the spacing between Ea and Eb is altered. 2 Based on the observation that the m40 sequence is insufficient to stimulate transcription on its own, we speculated that additional sequences further upstream of m40, but within the 242-bp enhancer are required for maximal induction. Herein we demonstrate that a third element, a direct repeat of Eb, termed Ec, lying upstream of m40 is also required for baseline enhancer activity. This sequence when multimerized can strongly stimulate renin promoter activity on its own. Moreover, the Ec-10 bp-Eb sequence matches the consensus binding site for members of the nuclear hormone receptor superfamily. This sequence can bind the RAR␣ and RXR␣ transcription factors and is required for induction of the renin promoter by retinoic acid. That retinoic acid can stimulate endogenous renin mRNA in As4.1 cells and mouse kidney suggests they may play a potentially important role in regulating renin expression.

MATERIALS AND METHODS
Plasmids-The luciferase (LUC) reporter vectors m4.1kLUC, mE2.6kLUC, mEa2.6kLUC, mEb2.6kLUC, mEba2.6kLUC, and mE117LUC were described previously (4) (Fig. 1A). mE represents the 242-bp mouse renin enhancer sequence. m4.1k represents a 4.1-kb 5Ј-flanking sequence of mouse renin (Ϫ4.1 kb to ϩ6) containing mE in its native position. m2.6k represents a 2.6-kb 5Ј-flanking sequence of mouse renin lacking mE. m117 is a minimal mouse renin promoter spanning from Ϫ117 bp to ϩ6 relative to the transcription start site. mE2.6k is a 2866-bp continuous sequence with a SphI site separating mE from m2.6k. Mutations in elements a and b previously identified are indicated by a and b. Oligonucleotide-mediated mutagenesis was performed using the GeneEditor in vitro Site-directed Mutagenesis System (Promega). The selection oligonucleotides came with the kit. The mutagenesis oligonucleotides were generated by Genosys Biotechnologies Inc. and were phosphorylated before use. The sequence of all mutants was confirmed by direct fluorescent DNA sequencing.
The c mutation in mEc2.6kLUC was generated using the oligonucleotide 5Ј-CTCAGAGGTCAGAGTACAGCCAGGAAACCATCTG-3Ј containing two substituted bases in Ec (underlined). The bc double mutant in mEcb2.6kLUC was generated with the oligonucleotide 5Ј-CTCAGAGGAAAGAGTACAGCCAGGAAACCATCTG-3Ј containing two substituted bases in both Ec and Eb. The triple cba mutant construct mEcba2.6kLUC was made by using standard DNA cloning taking advantage of an RsaI restriction digestion site present between Ec and Eb at the junction of Ϫ2662 and Ϫ2661. We excised the distal part containing c (from Ϫ2866 to Ϫ2662) from mEcb2.6kLUC using SmaI and RsaI and the proximal part containing ba (Ϫ2661 to Ϫ2625) from mEba2.6LUC using RsaI and SphI. The two fragments were then cloned into pGEM-7zf(Ϫ) to generate mEcba/pGEM-7 that was used to make the final mutant construct. The ca mutant construct, mEca2.6kLUC, was generated by mutagenesis of mEc2.6kLUC using the oligonucleotide 5Ј-TGTACTCTGACCTCTTCGCTGCTGGTT-GTG-3Ј containing four substituted bases in Ea (underlined). All of the m4.1k-based mutants were generated by recovering the BstXI to KpnI fragment (Ϫ2677 to Ϫ1217 bp) harboring the mE mutations from mE2.6k-based constructs, and ligating the fragments into a variant of the wild type m4.1kLUC construct containing a deletion of the corresponding BstXI to KpnI segment. mEa117LUC and mEcba117LUC were generated by swapping the DNA fragments between mE117LUC and mEa2.6kLUC or mEcba2.6kLUC, respectively. EcEb117LUC, 3EcEb117LUC, and 3 cb117LUC were generated by insertion of synthetic double-stranded oligonucleotides containing EcEb or cb in the forward orientation directly to the 5Ј-end of m117. SmaI to HindIII fragments containing EcEb117, 3EcEb117, and 3 cb117 were then excised from the corresponding subclones and inserted into pGL2-basic.
Cell Culture and Transient Transfection-Cell culture and transient transfection of the As4.1 cell line (American Type Culture Collection, CRL2193) was as described previously (4). In brief, As4.1 cells were cultured in reduced-serum Opti-MEM supplemented with 2% FBS, 1 mg/ml Albumax-II (Life Technologies, Inc.), penicillin (100 units/ml), and streptomycin (100 mg/ml) for 2 days before transient transfection. The conditioned cells were transfected with plasmid DNA by electroporation using equal molar amounts of each plasmid balanced with pUC19. 2.5 ϫ 10 7 cells were used in transfection for nondrug treatment studies. For ligand treatment studies, 10 8 cells were transfected and then split into four equal dishes containing Opti-MEM supplemented with 2% charcoal-treated FBS. A RSV-LUC vector was used as a standard positive control, and 0.1 g of cytomegalovirus-␤Gal was co-transfected as an internal control to monitor transfection efficiency. Cells were harvested and assayed for luciferase and ␤-galactosidase activity 48 h after initial transfection. Activity assays were performed as described previously (4,5). Luciferase activity was normalized to ␤-galactosidase activity from the same extract and presented as a percentage of luciferase activity of the Rous sarcoma virus promoter transfected separately in each experiment. All activity assays were performed in duplicate and the average of 2 readings was used as 1 data point.
Hormone Treatment-In our ligand treatment study, transfected As4.1 cells were cultured in the Opti-MEM supplemented with charcoal-treated fetal bovine serum (FBS) to minimize lipophilic hormones. To deplete small lipophilic compounds from FBS, 2 g of dextran-coated charcoal (C6197, Sigma-Aldrich Co.) was mixed with 100 ml of FBS. The mixture was gently rotated at 4°C for 16 h, and the charcoal was removed by centrifugation. The charcoal-treated FBS was sterilized by filtration and stored at Ϫ20°C. Hormone or vehicle was added to the media 24 h after transfection. The hormones used in the study were all-trans-retinoic acid (tRA), 9-cis-retinoic acid, 3,3Ј,5-triiodo-L-thyronine (T 3 ), and ␣1,25-dihydroxyvitamin D 3 (D 3 ) (Sigma-Aldrich Co.).
Three littermates were fed a normal diet as control. All animals had access to water ad libitum. tRA was administrated by subcutaneous injection (10 mg/kg). The test group received 5 injections, with 2 injections on each of the first 2 days (in the morning and evening) and 1 injection in the morning of the third day. The control group received vehicle. The animals were sacrificed 6 h after the last injection. The kidney was collected and frozen on dry ice immediately. Care and use of the mice met the standard procedures approved by the University Animal Care and Use Committee at the University of Iowa Electrophoretic Mobility Shift (EMSA) and Supershift Assay-Preparations of the nuclear extract from As4.1 cells and probes for EMSA were as described previously (4). To label probes, 5Ј-GATC overhangs at both ends of the annealed double-stranded oligonucleotides were filled with [␣-32 P]dATP (PerkinElmer Life Sciences) and 3 other cold nucleotides using Klenow DNA polymerase. The probes were purified through Sephadex G-50. Each binding reaction contained 0.02 pmol of labeled probe (about 60,000 dpm), 3 g of nuclear extract, 1 g of poly(dI-dC) (Roche Molecular Biochemicals), and binding buffer with the final concentration of (in mmol/liter): Tris-HCl (pH 7.5), 10, EDTA 1, dithiothreitol 1, MgCl 2 1, and KCl 60, as well as 5% glycerol in a total volume of 20 l. For competition assays, cold competitor oligos were preincubated with nuclear extract and binding buffer for 15 min on ice before the addition of probes. The binding reactions were then incubated on ice for another 15 min, and the products were resolved on a 5% nondenaturing polyacrylamide gel.
Rabbit polyclonal antibodies against human RAR subtypes: RAR␣, RAR␤, and RAR␥, and RXR subtypes: RXR␣, RXR␤, and RXR␥ were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). The human antisera cross-reacts with the specific RAR and RXR subtypes in the mouse. Universal monoclonal antibody against the mouse RXR family (including all three subtypes of ␣, ␤, and ␥) was the generous gift from Dr. Pierre Chambon (CNRS, INSERM, Universite Louis Pasteur, Strasbourg, France). 10 g of polyclonal antibodies (Santa Cruz) and the indicated amount monoclonal antibodies (Chambon) were added following the initial incubation of probe, nuclear extract, and binding buffer, and were left on ice for 60 min before electrophoresis.
RNA Isolation, RNase Protection Assay, and RT-PCR-Total cellular RNA was isolated from mouse As4.1 cells using TRI-REAGENT (Molecular Research Center, Inc., Cincinnati, OH) using the manufacturer's protocol. Total renal RNA was isolated from mouse kidneys using our standard procedure (6). T7 RNA polymerase was used to prepare antisense RNA transcripts as RNase protection assay probes. The fulllength and protected probe for mouse Ren-1 c mRNA was 235 and 175 nucleotides, respectively. The full-length and protected probe for mouse 18 S rRNA was 140 and 80 nucleotides, respectively. RNase protection was performed using the Hyb-speed kit (Ambion Inc., Austin, TX). The protected RNA probes were resolved on 6% polyacrylamide denaturing gel (containing 8 M urea) and quantified with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Construction of Wild-type and Dominant Negative RAR␣-The fulllength cDNA encoding mouse RAR␣1 was PCR amplified using the primers 5Ј-GACTTGCTAGCCTGTTTGCCTG-3Ј (NheI site underlined) and 5Ј-CTGAATTCCGTGTGTCGAGGTGGT-3Ј (EcoRI site underlined). This cDNA clone contains 1441 bp including the entire mRAR␣ coding sequence. The PCR product was digested with NheI and EcoRI to expose the restriction ends and was cloned into the mammalian expression vector, pCI (Promega, Madison, WI) under the control of the cytomegalovirus enhancer/promoter. To make the dominant negative mRAR␣403 expression vector, we first amplified a segment of cDNA, from ϩ634 to ϩ1209 from the full-length RAR␣1 using the primers 5Ј-CACTACGAACAACAGCTCAGAACA-3Ј and 5Ј-GGTCTAGACTAC-GGGATCTCCATCTTCAATG-3Ј (XbaI site as underlined). This segment of cDNA does not have the C-terminal AF-2 domain. This cDNA has an intrinsic EcoRV site in the center of the RAR coding region and a synthetic XbaI site at the 3Ј end. Therefore, the EcoRV to XbaI fragment was isolated and ligated into the pCI vector that contained all the N-terminal sequence upstream from the EcoRV site.
Statistical Analysis-All data are presented as mean Ϯ S.E. Multiple comparison of data was analyzed by one-way ANOVA using SigmaStat (SPSS Scientific). When the test for normalization failed, the analysis was performed nonparametrically. Single comparisons are performed using Student's t test.

RESULTS
Using chimeric enhancers derived from the divergent regions of the mouse and human renin enhancer we previously identified two regulatory elements in the promoter proximal portion of the mouse renin enhancer. These studies were performed using As4.1 cells, which express renin and are likely derived from juxtaglomerular cells. Ea acted as a negative regulatory element and bound the factor NF-Y, and Eb acted as a positive regulatory element (4). Mutation of Eb revealed that it is required for maximal enhancer activity, but that Eb and Ea alone were insufficient to stimulate renin promoter activity, suggesting the presence of other transcription factor-binding sites in the enhancer. The purpose of the current study is to further examine the requirements for enhancer-mediated transcriptional activation and identify the stimulatory factors.
Mutational analysis of Eb revealed that it has the core sequence TGACCT. Sequence analysis of the 242-bp mE revealed two other TGACCT motifs which lie upstream (more distal) of Eb. The first motif, termed Es is present at the 5Ј terminus of the enhancer at coordinates Ϫ2847 to Ϫ2852. The other TGACCT motif, termed Ec, is located 10 bp upstream of Eb from Ϫ2668 to Ϫ2673, and thus lies upstream of the original m40 sequence containing Eb and Ea (Fig. 1B). Since our previous study demonstrated that enhancer function requires Eb, we performed transient transfection analysis in mouse renin expressing As4.1 cells to test whether Es or Ec were also required for enhancer activity. Site-directed mutagenesis was performed to convert the GA in TGACCT to TT, which in our previous study caused a loss of function of Eb. Both mutations were individually generated in mE2.6kLUC, which contains the mouse enhancer fused upstream of a 2624-bp mouse renin promoter (Fig. 1A). Mutation of Ec significantly reduced transcriptional activity (Fig. 2), whereas mutation of Es had no effect (data not shown). Interestingly, mutation of Ec caused a significantly greater drop in enhancer activity than did mutation of Eb. Moreover, the increase in enhancer activity caused by mutation of the negative regulatory Ea required both Eb and Ec (Fig. 2). The importance of Ec and Eb was confirmed by mutagenesis of 4.1kLUC which contains the enhancer in its native position. Mutation of Eb and Ec lowered activity of the promoter to that of an enhancerless mutant (data not shown).
The mouse renin enhancer acts in a position-independent manner and can strongly stimulate (Ͼ100-fold) a minimal mouse renin promoter when placed directly upstream. To characterize whether Ec ϩ Eb has intrinsic enhancer activity on its own, we placed one or three copies of the Ec ϩ Eb sequence directly upstream of the 117-bp promoter. Although one copy of the Ec ϩ Eb sequence only slightly increased promoter activity (5.2-fold), three tandem copies of Ec ϩ Eb markedly increased promoter activity (110-fold) to nearly the same level as mE (Fig. 3). Mutation of Eb and Ec in the 3X(Ec ϩ Eb) construct abolished this induction.
We previously reported that Eb specifically interacted with unidentified nuclear proteins from As4.1 cells by EMSA. Since both Ec and Eb consisted of a TGACCT stretch we hypothesized that they possessed the same protein binding activity. We identified two major DNA-protein complexes (L and S) formed on Ec ϩ Eb (Fig. 4A). The two complexes were efficiently competed by Ec ϩ Eb as well as mutants lacking either Ec or Eb. On the contrary, a mutant lacking both Eb and Ec was not able to compete. The results suggest that both Ec and Eb have the ability to form both complexes. This was confirmed by demonstrating that mutants lacking either site, but not both, could still form complexes L and S when used as probes in EMSA (Fig. 4B). The two complexes formed on both Ecb and cEb were efficiently competed by competitor DNAs containing either one or two TGACCT stretches (data not shown).
To facilitate our identification of the Ec ϩ Eb-binding factors, we determined which bases in the TGACCT stretch were essential to its binding activity. To accomplish this we first demonstrated that complexes L and S could be efficiently competed by DNAs containing a single TGACCT motif from either Eb or Ec (Fig. 4C). We then examined the requirement of each base in the binding activity by using double stranded oligonucleotides mutated one base at a time as competitors in EMSA (Fig. 4C). Mutation of any base within the TGACCT motif resulted in loss of competition, whereas a base-change mutation outside of the TGACCT stretch did not affect competition.
A direct repeat of the TGACCT motif separated by a spacer of variable length is the consensus recognition sequence for transcription factors in the thyroid/retinoid superfamily of nuclear hormone receptors. As a candidate approach to identify which nuclear receptor could bind to Ec ϩ Eb, we transfected As4.1 cells with the 3X(Ec ϩ Eb)117LUC reporter vector, and then treated the cells with four different common nuclear receptor cognate ligands. To eliminate potential effect of lipophilic hormones, the transfected cells were grown in culture medium supplemented with charcoal-treated fetal bovine serum. Thyroid hormone did not affect promoter activity, while ␣1,25-dihydroxyvitamin D 3 modestly, but significantly reduced promoter activity (Fig. 5A). In contrast, promoter activity was significantly increased by 9-cis-retinoic acid and tRA, suggesting that Ec ϩ Eb may be a retinoic acid responsive element (RARE). The increase in transcription caused by tRA was dose responsive (Fig. 6).
To order to further confirm that the observed retinoic acid response was directly mediated by Ec ϩ Eb, we transiently transfected As4.1 cells with a vector containing 3 tandem repeats of a mutant Ec ϩ Eb sequence. Both baseline and tRAinduced transcription was abolished by mutation of Ec ϩ Eb (Fig. 6A). Similarly, transcriptional activity of the 4.1kbLUC construct (which contains the enhancer in its native position) was induced by retinoic acid, and required an intact Ec ϩ Eb sequence (Fig. 6B). Loss of the Eb and Ec sequence in 4.1kbLUC had the same effect as eliminating the enhancer. Given our observation that both Ec and Eb could bind the same nuclear factors we determined whether both sites were required for the tRA-induced response. Interestingly, although the induction was lower than wild-type mE, tRA treatment still induced (2-fold) the promoter activity of constructs containing a mutation of either Eb or Ec (Fig. 7). As above, loss of both Eb and Ec abolished tRA-mediated induction.
To evaluate the physiological relevance of our observation we examined the effect of tRA on the level of endogenous renin in As4.1 cells. Renin mRNA level in As4.1 was modestly increased after 24 h treatment with tRA (data not shown). tRA also modestly increased the level of the renin promoter SV40 T antigen transgene mRNA present in those cells (data not shown). We next examined the effect of tRA on the level of endogenous renin mRNA in mouse kidney. Mice received subcutaneous injections of tRA for 3 days after being fed a vitamin A-deficient diet for 2 months. The treatment significantly increased (3-fold) the level of endogenous renal renin mRNA (Fig. 8).
Based on these results we hypothesized that RAR bind to Ec ϩ Eb in mE to stimulate transcription from the renin promoter in As4.1 cells. We therefore performed supershift EMSA using antibodies specifically-targeting either retinoic acid receptors (RAR␣, -␤, or -␥), or RXR. The anti-RAR␣ antisera caused the appearance of a supershifted complex, whereas the antibodies against other subtypes of RAR did not shift the DNA-protein complex (Fig. 9). The anti-RXR monoclonal antibody also caused the appearance of a supershifted complex. Subtype-specific antisera for RXR revealed that RXR␣ bound to Ec ϩ Eb (data not shown). That both RAR␣ and RXR␣ antisera only partially supershifted the complex suggests that other proteins, perhaps other members of the nuclear hormone superfamily may also bind to mE. RT-PCR and DNA sequencing verified the expression of RAR␣, RAR␥, and RXR␣, but not RAR␤, RXR␤, and RXR␥ in As4.1 cells (data not shown).
Finally, we constructed expression vectors encoding mouse RAR␣1, and a dominant negative mutant of mRAR␣ (mRAR␣403) which lacked the C-terminal ligand-dependent AF-2 transactivation domain, to specifically test the role of RAR␣ in the Ec ϩ Eb-mediated retinoic acid-induced activation of the mouse renin promoter (8,9). Both cDNAs were placed under the control of the cytomegalovirus promoter/enhancer and were transiently cotransfected into As4.1 cells along with the 4.1kbLUC reporter vector. Co-transfection of the wild-type mRAR␣ expression vector significantly increased promoter activity of m4.1kLUC whether induced with tRA or left untreated (Fig. 10). On the contrary, co-transfection of the mRAR␣403 expression vector significantly attenuated the tRA-induced promoter activity of m4.1kLUC vector, but did not alter baseline expression. The stimulatory effect of the wild-type mRAR␣ was completely abolished when Ec and Eb were mutated (data not shown). Our results suggest an important role of the Ec and Eb sequence in controlling baseline activity of mE and demon-  8. Induction of endogenous REN by retinoids. C57BL/6J mice received subcutaneous injection of either tRA (n ϭ 5) or vehicle (n ϭ 6) as described under "Materials and Methods." Mouse kidneys were collected 54 h after the initial tRA treatment. RNA (10 g) isolated from right and left kidneys were independently subject to RNase Protection Assay using the mouse renin probe and 18 S rRNA probes. Gels were quantified using a PhosphorImager. Data from the right and left kidneys of one mouse were averaged as n ϭ 1. *, p Ͻ 0.05 versus vehicle. strate a potential role for retinoids in regulating expression of the renin gene.

DISCUSSION
The renin gene enhancer is a potent enhancer of transcription that works in a position-and orientation-independent manner in renin-expressing As4.1 cells (2). Its ability to stimulate up to a 100-fold increase in transcriptional activity of the renin promoter makes it an important candidate in renin gene regulation. We previously identified two transcription factorbinding sites which mechanistically oppose each other (4). The binding of NF-Y to Ea acts as a negative regulator because it blocks the binding of transcription factors to Eb. Studies where the spacing between Ea and Eb is altered strongly suggests that NF-Y sterically blocks the binding of factors to Eb. 2 The importance of Eb was shown by a loss of enhancer activity after mutagenesis. In the present study, we demonstrate that a second TGACCT motif located 10 bp upstream of Eb termed Ec is also required for maximal enhancer activity. Mutation of either Ec or Eb in mE essentially eliminates enhancer function. Mutation of a third TGACCT motif (Es) did not effect enhancer function.
The Eb and Ec sequences are clearly required for baseline activity of the enhancer. In addition, their importance is supported by the observation that they fit the consensus sequence for a member of the nuclear hormone receptor superfamily, are required to mediate induction of the renin promoter by retinoic acid, and bind RAR␣ and RXR␣. There was no induction when cells were treated with thyroid hormone or vitamin D 3 . Within the superfamily, selective recognition of different ligand/receptors is determined by the number of intervening base pairs between the two TGACCT motifs. In general, heterodimers of TR/RXR selectively bind to DR4 and heterodimers of VDR/RXR selectively bind to DR3. Interestingly, heterodimers of RAR/ RXR preferentially bind to DR2 or DR5 (10). There are 10 intervening base pairs between Ec and Eb in mE. Our EMSA and supershift studies clearly demonstrate the ability of the DR10 sequence to bind RAR␣ and RXR␣. Despite the preference for DR2 or DR5, the retinoic acid responsive ␥F-crystallin and medium chain acyl-coenzyme A dehydrogenase genes contain DR8, and the oxytocin and laminin B1 genes contain DR14 (11)(12)(13)(14).
It is interesting to note that constructs containing mutations in either Eb or Ec, but not both, still responded to tRA stimulation, but with reduced responsiveness. This is consistent with our EMSA results showing that probes containing one intact TGACCT motif were still capable of forming two complexes that had the same mobility as those with probe Ec ϩ Eb. Imperfect motifs for RAR/RXR exist in many genes, such as apolipoprotein A1, oxytocin, ␥F-crystallin, medium chain acylcoenzyme A dehydrogenase, phosphoenolpyruvate carboxykinase, and ␤-crystallin (11,(13)(14)(15)(16)(17). Each can still bind heterodimers of RAR/RXR and mediate retinoic acid-induced gene transcription. This may be an important consideration because the human renin enhancer (hE) has one perfect TGACCT motif (Ec) and one variant motif (TGGCCT, Eb). Baseline transcriptional activity of hE was considerably lower than that of mE (4), but nevertheless retained modest retinoic acid-induced transcription (data not shown).
When a heterodimer of RAR/RXR binds to a TGACCT direct repeat, RAR selectively binds to the upstream TGACCT while RXR selectively binds to the downstream TGACCT (18). Forcing RAR/RXR to bind RARE in the opposite direction abolished its transactivation activity. In our study, mutation of Ec always had a stronger effect than mutation of Eb, suggesting the possibility for asymmetric binding of RAR/RXR to the RARE. In a model described in Westin et al. (19), RAR plays a pivotal role to initiate the hierarchical assembly of transcriptional proteins on RARE. If this is true for RAR/RXR binding to Ec ϩ Eb in mE, RAR will bind to Ec and RXR will bind to Eb. Since RAR/RXR-mediated transactivation depends on coactivators, and the interaction between RAR/RXR and coactivators initiates from RAR, disturbance of RAR binding to Ec should cause a marked loss of the functional RAR/RXR-ligand-coactivator complex.
Despite the retinoid-mediated induction of the renin promoter (and endogenous renin gene) mediated by RAR, it is likely that other transcription factors bind to the Ec ϩ Eb sequence. EMSA revealed two major protein-DNA complexes, and supershift does not cause a reduction in either the L or S complex. Indeed, Eb and Ec are both required for the induction by retinoids and for baseline activity of the enhancer in the absence of ligand. These data suggest that other transcription factors, perhaps other members of the hormone receptor superfamily may also play a role in regulating renin gene expression. It is possible that the requirement for Ec ϩ Eb in mediating baseline transcriptional activity of mE may occur independently of RAR/RXR, while the retinoic acid induced activity requires RAR/RXR. We are currently attempting to identify other Ec ϩ Eb binding factors using yeast one-hybrid analysis and have preliminary data implicating an orphan nuclear receptor. Some orphan receptors have been reported to bind the TGACCT as a monomer to regulate RARE function (20).
In addition to Eb and Ec, Gross and colleagues 3 have identified a fourth transcription factor-binding site located 16 bp upstream of Ec termed Ed. This sequence is similar to the consensus binding site for members of the CREB/ATF-1 family of transcription factors. Interestingly, the transactivation function of both RAR/RXR and CREB/ATF-1 requires an interaction with coactivators, and both are able to interact with p300/ CBP (21,22). Therefore, it is possible that transcription factors binding to Ed and Ec may interact. Recall that the human renin enhancer retains an intact Ed and Ec, but lacks Eb. Mechanistically, the situation may be analogous to the pit-1 gene where RAR binds to a single core recognition motif to activate retinoic acid-dependent transcription (23). In the pit-1 gene, the RARE is immediately adjacent to a Pit-1-binding site. Retinoic acid-dependent transcription of pit-1 requires Pit-1, which like RAR and CREB/ATF-1 requires CBP as a coactivator (24).
In closing, our results pose the obvious question as to the relevance of renin gene regulation by retinoids in vivo. As this 3 T. A. Black and K. W. Gross, unpublished observation. is the first study to implicate this pathway, its importance in adults remains unclear. However, it is now clearly recognized that retinoids are critical signaling molecules during development. Vitamin A deficiency during development leads to fetal vitamin A deficiency syndrome, which includes abnormalities in the urogenital tract including the kidney (25). Moreover, retinoic acid is thought to be the active metabolite of vitamin A during development. Knockout mice deficient in specific subtypes of both RAR␣ and RAR␤ or RAR␣ and RAR␥ develop severe renal malformations generally characterized by renal agenesis and aplasia (26). Histological analysis revealed the defect to be a failure of the Wolffian or mesonephric duct to contact the metanephric blastema. This interaction is critical for the differentiation of the tubular and eventual development of the vascular system in the kidney. Interestingly, mice lacking genes in the renin-angiotensin system also develop severe renal abnormalities although at a much later stage in development (27). Renin expression is first visible in the metanephric kidney at 15.5 days postcoitum and is localized in the developing arterial tree (28,29). Expression of renin coincides with the growing branches of the arterial tree suggesting it may play an important developmental role. Therefore, it is possible that as retinoids are critical in very early events in renal development, so to are they needed to induce renin expression developmentally. The expression of RAR and RXR in As4.1 cells which are believed to be derived from juxtaglomerular cells is consistent with this notion. Clearly, additional studies examining the coexpression of RAR with renin during renal development would seem necessary.