An Abd-B class HOX.PBX recognition sequence is required for expression from the mouse Ren-1c gene.

Expression from the mouse Ren-1(c) gene in As4.1 cells is dependent on a proximal promoter element (PPE) located at approximately -60 and a 241-base pair enhancer region located at -2625 relative to the transcription start site. The PPE (TAATAAATCAA) is identical to a consensus HOX.PBX binding sequence. Further, PBX1b has been shown to be a component of a PPE-specific binding complex present in nuclear extracts from As4.1 cells. The binding affinities of different paralog HOX members to the PPE were examined in the absence or presence of PBX1b. HOXB6, -B7, and -C8 failed to bind the PPE alone but showed weak affinity in the presence of PBX1b. In contrast, HOXD10 and to a lesser degree HOXB9 bound the PPE with high affinities regardless of whether PBX1b was present. Abd-B HOX members, including HOXD10, -A10, -A9, -B9, and -C9, are expressed in As4.1 cells. The ability of HOX and PBX1b to form a ternary complex with PREP1 on the PPE is also demonstrated both in vivo and in vitro. Point mutations in either the HOX or PBX half-site of the PPE disrupted the formation of the HOX.PBX complex and dramatically decreased transcriptional activity of the Ren-1(c) gene demonstrating that both the HOX and PBX half-sites are critical for mouse renin gene expression. These results strongly implicate Abd-B class Hox genes and their cofactors as major determinants of the sites of renin expression.

Renin is an aspartyl protease, which, as part of the reninangiotensin system, plays a critical role in the maintenance of blood pressure and electrolyte balance by converting angiotensinogen to angiotensin I (1). Additionally, the reninangiotensin system has been implicated in aspects of renal development by the observation that pharmacological and genetic disruptions of renin-angiotensin system function result in aberrant renal morphology (for reviews, see Refs. [2][3][4]. Renin gene expression is subject to complex developmental and tissue-specific regulation (for review, see Ref. 5). In murine fetuses, kidney renin transcripts can be detected as early as 14.5 days post coitum in the newly developing arteries (6). On development of the renal arterial tree, expression shifts to cells located in nascent portions of the growing arteries until expression is restricted to a small population of modified smooth muscle cells of the afferent arteriole proximal to the glomerulus called juxtaglomerular cells. In adult mice, renin is also expressed in adrenal gland, submandibular gland, gonads, and coagulating gland.
Some mouse strains have only a single renin gene (Ren-1 c ), whereas other strains have two copies, a Ren-1 locus with allele Ren-1 d and a duplicated locus with Ren-2 (7,8). Although these mouse renin genes are approximately equivalently expressed in the adult kidney, their expression patterns are different in some extra renal tissues (9).
Identification of cis-acting sequences in the mouse renin 5Ј-flanking region has been accelerated by the isolation of a kidney tumor-derived As4.1 cell line from transgenic mice containing the mouse Ren-2 5Ј-flanking sequence fused to SV40 T antigen (10). As4.1 cells are capable of expressing high levels of renin mRNA (10) and secreting active renin protein (11). By transiently transfecting these cells with wild-type or mutant Ren-1 c -chloramphenicol acetyltransferase (Ren-CAT) constructs, two regions in the Ren-1 c 5Ј-flanking sequence were found to be necessary for high level expression of mouse renin gene in addition to a TATA box, the PPE 1 located at Ϫ60 bp and a 241-bp enhancer located 2.6 kb upstream of the transcription start site (12). The PPE was shown to bind As4.1 cell nuclear proteins in electrophoretic mobility shift assays (EMSAs). Further competition assays indicated that the minimal sequence required for protein binding included N (1)(2)(3) TAATAAATCA. Mutation of the PPE in the Ren-CAT construct containing a 4.1-kb mouse renin sequence dramatically reduced the chloramphenicol acetyltransferase activity in transfection assays suggesting a critical role of this element in the regulation of mouse renin gene expression.
Hox genes are members of the homeobox family of transcription factors and control many aspects of morphogenesis and cell differentiation in animals (13). In vertebrates, there are 39 Hox genes organized in four clusters (A, B, C, and D) on separate chromosomes with members of each cluster classified into as many as 13 paralog groups based on sequence similarity (14). Hox gene products can bind DNA as monomers or heterodimers with three-amino acid loop extension (TALE) class homeodo-main proteins including PBX and MEIS on HOX⅐PBX or HOX⅐MEIS recognition sequences (15)(16)(17)(18)(19)(20)(21)(22). Interactions with PBX or MEIS proteins increase both DNA binding affinity and specificity for HOX proteins. Although the HOX⅐MEIS recognition sequences have not yet been identified in any natural genes, the HOX⅐PBX binding sequences have been found in genes such as Hox (23,24), ␣2(V) collagen (25), Eph receptor EphA2 (26), and fork head (fkh) (27). Moreover, MEIS or its homolog PREP1 interacts with PBX through its amino-terminally located HM domain (28 -30). This interaction is essential for PBX nuclear translocation. Several studies have also demonstrated that MEIS or PREP1 can form ternary complexes with HOX and PBX (31)(32)(33)(34)(35)(36), and this trimeric complex has been shown to play an important role in regulating several HOX-responsive genes (31, 34 -36).
In this report we show that the renin gene contains a functional HOX⅐PBX binding site in its promoter region. Homeodomain proteins PBX1b and HOX family members (preferentially HOX9 and -10) can bind this element in vitro and in As4.1 cells. Both PBX and HOX binding sites are necessary for the expression of the Ren-1 c gene. Moreover, we have demonstrated that PREP1 can form a ternary complex with HOX and PBX1b on this site. Plasmid Constructions-Plasmids Ϫ4.1R1, Ϫ4.1M, Ϫ4.1mh, Ϫ4.1mp, and Ϫ4.1mp2 were constructed by inserting the wild-type or mutant Ren-1 c sequences from Ϫ4100 to ϩ6 bp into pGL2-basic (Promega). The reporter construct 3XR1-TA was made by inserting three copies of Ren-1 c PPE into a pGL2-basic-derived plasmid containing the adenovirus E1b TATA box (37).

Analysis of Hox Gene Expression Using RT-PCR-Total
Full-length cDNA for HOXB6, -B7, -C8, -B9 or -D10 was isolated by RT-PCR from As4.1 cells and cloned into pcDNA3.1/myc-His(ϩ)A vector (Invitrogen), which contains a carboxyl-terminal Myc epitope. Fulllength cDNA for PREP1 was also cloned from As4.1 cells and inserted in pcDNA3.1/V5-His(ϩ)A (Invitrogen), which has a carboxyl-terminal V5 epitope. Full-length cDNA for PBX1b was cloned in the same vector without incorporating any epitope. These plasmids were used in in vitro transcription/translation and EMSAs. For expression in mammalian cells, a stop codon was added immediately before the Myc or V5 epitope sequence in the HOXD10 or PREP1 expression vector, respectively, so that the carboxyl-terminal Myc or V5 epitope was not translated. VP16-HOXD10 or VP16-PBX1b was constructed by inserting HOXD10 or PBX1b full-length cDNA, respectively, in pVP16 (CLONTECH) so that a nuclear localization signal and a VP16 activation domain were fused amino-terminally to HOXD10 or PBX1b. VP16-PREP1 and VP16-HM were similarly constructed. VP16-PREP1 contained a truncated PREP1 with residues 1-75 deleted, whereas VP16-HM contains the HM region of PREP1 from residues 75 to 557.
Cell Culture and Transient Transfections-As4.1 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and transfected using FuGENE 6 (Roche Molecular Biochemicals). For each transfection in a 35-mm culture dish, 2.2 g of DNA including 0.5 g of reporter plasmid, 0.5 g of each expression plasmid, nonspecific plasmid if necessary, and 0.2 g of plasmid containing the Rous sarcoma virus promoter driving ␤-galactosidase were mixed with 4.4 l of FuGENE reagent. Forty-eight hours after transfection cells were harvested and measured for luciferase and ␤-galactosidase activities using the Luciferase Assay System (Promega) and Galacto-Light Plus TM chemiluminescent reporter assay (Tropix), respectively. The luciferase activity is normalized with ␤-galactosidase activity to correct differences in transfection efficiency between experiments. All transfection results represent the average Ϯ S.D. of at least three separate experiments.
In Vitro Transcription and Translation-HOX, PBX1b, and PREP1 proteins were in vitro transcribed/translated by the TNT Coupled Wheat Germ System (Promega). Parallel reactions containing [ 35 S]methionine were performed to correct differences in translation efficiency.
EMSA-The EMSAs were performed as described previously (12). For each reaction (15 l), about 0.2 ng of labeled DNA probe (20,000 cpm) were mixed with As4.1 cell nuclear extract (3-6 g) or in vitro translated proteins (1-3 l for each protein) and 1 g of poly(dI-dC) in 10 mM Hepes, pH 7.9, 10 mM KCl, 50 mM NaCl, 2 mM MgCl 2, 1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10% glycerol. The reaction mixture was incubated on ice for 30 min and then run on a 6% polyacrylamide gel in 0.5ϫ Tris borate-EDTA buffer. In a competition or supershift assay, an excess amount of unlabeled DNA or 1 l of antibody was added to the reaction mixture 15 min or 1 h prior to the addition of labeled DNA probe, respectively. Antibodies against PBX1/2/3, PBX1, PBX2, PBX3, and PREP1 were purchased from Santa Cruz Biotechnology, Inc. The antibody against the Myc tag was purchased from Invitrogen.

RESULTS
The Ren-1 c PPE Is a HOX⅐PBX Protein Binding Site-The consensus sequence for HOX⅐PBX binding is NTGATTNATNN. Using in vitro site selection assays, the preferred binding site for PBX and HOX6 -10 paralog members was shown to be NTGATTTATNN (or inverting NЈNЈATAAATCANЈ) (18,19). The sequence is a precise match to the highly conserved PPE in mouse Ren-1 and Ren-2 and rat and human renin promoters (Fig. 1). To test whether the PPE is a HOX⅐PBX binding site, supershift EMSA was performed with antibodies against PBX1/2/3, which recognizes long isoforms of PBX proteins (PBX1a, PBX2, and PBX3a), PBX1, PBX2, and PBX3 (Fig. 2). Only PBX1 antiserum, which recognized both PBX1a and PBX1b, was able to supershift the R1⅐As4.1 nuclear protein complex. The results demonstrated that the PPE contains the PBX binding site and suggest that PBX1b is the major PBX species binding at the PPE.
To determine which HOX protein might bind to the PPE, the Hox gene complement in As4.1 cells was surveyed by amplifying the highly conserved homeodomain region using RT-PCR. A total of 45 clones was assessed, and the results from the analysis are shown overlaid on an organizational map of the mouse Hox gene clusters (Fig. 3). The results showed that predominantly (33 of 45) Abd-B classes 9 and 10 are represented, and specifically HOXD10, -A10, -A9, -B9, and -C9 are expressed in As4.1 cells.
To test whether HOX and PBX proteins can bind to the PPE, EMSA was performed with in vitro synthesized HOXB6, -B7, -C8, -B9, and -D10 proteins plus and minus in vitro synthesized PBX1b (Fig. 4). Each HOX protein contained a Myc carboxylterminal epitope so that the complex formed by the R1 probe and the HOX protein could be supershifted by antibody against the Myc epitope. PBX1b alone was not capable of binding the PPE (Fig. 4, lane 1). HOXB6 (Fig. 4, lane 2), -B7 (Fig. 4, lane 6), and -C8 (Fig. 4, lane 10) did not bind to R1 detectably, however, in the presence of PBX1b they were able to bind weakly (more apparent with longer exposure, Fig. 4, lanes 3, 7, and 11, and data not shown). However, HOXB9 and especially -D10 gave rise to prominent shifts with or without PBX1b (Fig. 4, lanes  14, 15, 18, and 19). The HOX and HOX⅐PBX1b complexes were able to be supershifted by Myc antiserum (Fig. 4, lanes 16 and  20), whereas the HOX⅐PBX1b complex was disrupted by the addition of antibody against PBX1 (Fig. 4, lanes 17 and 21).
PREP1 Can Form a Ternary Complex with HOX⅐PBX on Ren-1 c PPE-TALE class homeodomain proteins PREP1 and MEIS have been shown to form ternary complexes with HOX⅐PBX (31)(32)(33)(34)(35)(36). We examined whether PREP1 is a component of the As4.1 cell nuclear proteins that bind to the Ren-1 c PPE. In supershift assays, the antibody against PREP1 supershifted the complex formed by Ren-1 c PPE and As4.1 nuclear extracts (Fig. 5A). In vitro synthesized PREP1 was also tested for its ability to bind Ren-1 c PPE in the presence of HOXD10 and PBX1b (Fig. 5B). The results indicate that, although PREP1 could not bind to PPE alone or in combination with PBX1b (data not shown), it forms a ternary complex with HOXD10 and PBX1b (Fig. 5B, lane 3). The ternary complex could be supershifted or disrupted by the addition of antibody against Myc, PBX1, or PREP1 (Fig. 5B, lanes 4 -6). Moreover, the ternary complex formed by HOXD10⅐PBX1b⅐PREP1 has almost the same electrophoretic mobility as the complex formed by As4.1 cell nuclear proteins (Fig. 5B, lane 3 versus lane 7) further suggesting that the Ren-1 c PPE binds a ternary complex including HOX, PBX1b, and PREP1.
To further assess whether HOX is a member of As4.1 cell nuclear proteins binding to PPE, we tested whether in vitro translated HOXD10 can complex with PBX and PREP1 in As4.1 cell nuclear extract on the PPE. One microliter of 10-fold diluted HOXD10 protein synthesized in vitro was mixed with As4.1 cell nuclear extracts in EMSA, and the presence of HOXD10 in the complex was confirmed by supershift assays with Myc antiserum (Fig. 5C). The results showed that the complex formed by this amount of HOXD10 alone was barely visible (Fig. 5C, lane 1). However, the addition of the same amount of HOXD10 to As4.1 cell nuclear extract resulted in the formation of a complex that paralleled in size the complex formed by As4.1 cell nuclear extract alone and was more intense (Fig. 5C, lane 3 versus lane 2). The sharp increase in complex intensity by adding HOXD10 suggests the possibility that As4.1 cell nuclear extract can provide factors that improve the ability of HOXD10 to bind the R1 PPE, and these factors include PBX1 and PREP1. Myc antiserum was capable of supershifting almost all the complex formed (Fig. 5C, lane 4) suggesting that the exogenous HOXD10 was in excess of endogenous HOX proteins binding to the PPE. Antibodies against PBX1 and PREP1 also supershifted the complex (Fig. 5C, lanes  5 and 6) indicating that the formation of HOXD10⅐PBX⅐PREP1 ternary complex.
Both HOX and PBX Half-sites Are Crucial for Ren-1 c Gene Expression-Based on HOX⅐PBX crystal structure (38,39), crucial contacts are made between PBX and the adenine at the nucleotide 4 position of the heterodimer recognition sequence and between HOX and the nucleotide 8 position of the recognition sequence (see Fig. 6A). We examined the effects of point mutations at these positions on the binding of As4.1 cell nuclear extracts in EMSA (Fig. 6, A and B). A single mutation of nucleotide 4 (Fig. 6A, R3) or double mutations at nucleotides 3 and 4 (Fig. 6A, R4) resulted in the formation of complexes with high mobility equivalent to binding of the HOX monomer alone (Fig. 6B, lanes 2 and 3). The same results were obtained using in vitro translated HOXD10⅐PBX1b (Fig. 6C, lanes 4 and 5). In contrast, the single nucleotide mutation at nucleotide 8 (Fig.  6A, R5)  either As4.1 cell nuclear extracts (Fig. 6B, lane 4) or in vitro translated HOXD10⅐PBX1b (Fig. 6C, lane 6). In competition assays (Fig. 6B), the R1 oligonucleotide competed with itself in the formation of the R1⅐As4.1 cell nuclear protein complex very well at a 100-fold excess (Fig. 6B, lanes 5 and 6). However, oligonucleotides with PBX half-site mutations exhibited only partial competition (Fig. 6B, lanes 7 and 8), whereas the oligonucleotide with the nucleotide 8 mutation failed to compete (Fig. 6B, lane 9). These results are consistent with previous reports that HOX can bind its recognition site as monomer; however, PBX cannot bind to the HOX⅐PBX recognition sequence in isolation.
Effects of these point mutations on expression from the Ren-1 c promoter were also tested (Fig. 6D). The mutations were incorporated into plasmid Ϫ4.1R1. The nucleotide 8 mutation (Ϫ4.1mh) in the HOX protein binding site reduced the expression level of Ϫ4.1R1 by ϳ80%. The nucleotide 4 single mutation (Ϫ4.1mp) or nucleotides 3 and 4 double mutations (Ϫ4.1mp2) in the PBX half-site also resulted in an ϳ85% reduction in expression from the Ren-1 c promoter. These results suggest that the binding of PBX and HOX to PPE is crucial for mouse renin gene expression.
HOXD10, PBX1b, and PREP1 Can Bind the Ren-1 c PPE in As4.1 Cells-As4.1 cells were cotransfected with a reporter construct containing three copies of the Ren-1 c PPE upstream of an E1b TATA box and expression vectors for HOXD10, PBX1b, PREP1, and various combinations. Overexpression of these proteins had little effect on the transcription of the reporter gene (Fig. 7). However, fusion proteins VP16-HOXD10, VP16-PBX1b, and VP16-PREP1, which have a VP16 activation domain fused amino-terminally to HOXD10, PBX1b, and PREP1, respectively, were capable of activating reporter gene expression by 11-, 15-, and 30-fold. However, they did not activate a similar reporter gene with mutated Ren-1 c PPE (data not shown). These results suggest that HOXD10, PBX1b, and PREP1 are capable of binding to the PPE in As4.1 cells. Moreover, the fusion protein VP16-HM containing only the VP16 activation domain and HM domain, which is the region in PREP1 interacting with PBX, was still capable of activating the reporter gene by 30-fold. The results agree with previous reports that the homeodomain of PREP1 is not necessary in the ternary complex formation on a HOX⅐PBX site (31). DISCUSSION In this report, we have demonstrated that the PPE of the renin gene is a HOX⅐PBX heterodimer binding site. First, the nucleotide sequences of PPE match perfectly to those of the HOX⅐PBX heterodimer recognition site. Second, Abd-B HOX9 and -10 members and PBX1b were shown to bind this element with high affinity in vitro. Finally, point mutation of a critical nucleotide either in the HOX or PBX half-site showed the expected effects on HOX⅐PBX binding in EMSA. In addition, these mutations dramatically reduced the transcriptional ac-tivity of Ren-1 c gene suggesting that both PBX and HOX halfsites are necessary for renin gene expression.
The human renin PPE was previously identified as a Pit-1 site in GC cells (40). However, our results have shown that the PPE in the mouse renin promoter is not a Pit-1 binding site (12). The PPE in mouse is a poor consensus for Pit-1. Moreover, the As4.1 cell nuclear proteins were not capable of binding to a human Pit-1 consensus oligonucleotide. Furthermore, the major nuclear protein complex from As4.1 cells formed on mouse Ren-1 c PPE could not be competed by the Pit-1 oligonucleotide, although it could be competed by the oligonucleotide containing the human PPE. These observations suggest that in As4.1 cells the primary proteins binding to the PPE are HOX and PBX, not Pit-1. Whether HOX and PBX proteins bind to the human PPE in GC cells has not been examined.
A direct test of whether HOXD10 protein is involved in the formation of the ternary complex on the mouse PPE in supershift assays has not been possible because of the unavailability of specific antibody. However, other evidence suggests that HOXD10 is a major HOX protein binding to the PPE in As4.1 cells. HOXD10 is expressed in As4.1 cells. Moreover, in the presence of PBX1b, HOXD10 binds to the PPE in EMSA with higher affinity compared with other selected HOX paralog members, and the size of the complex mimics the endogenous complex formed in As4.1 cell nuclear extracts. Furthermore, a small amount of in vitro synthesized HOXD10 was capable of forming the HOXD10⅐PBX1b⅐PREP1 complex with endogenous PBX and PREP1. However, we cannot rule out a redundant role for other HOX family members such as HOX9 members, which are also expressed in As4.1 cells and recognize the same sequence motifs as HOXD10. Recent evidence suggests that although different HOX⅐PBX heterodimers have preferred sites in vitro, this binding specificity is not always utilized in vivo (41,42). It has been suggested that the specificity of a HOX⅐PBX site might be affected by cofactor binding sites. Thus, the study of renin gene expression in mice lacking different HOX proteins will be helpful in identifying the specific HOX family members necessary for renin expression.
We showed that overexpression of HOXD10, PBX1b, PREP1, or various combinations was not able to activate a promoter containing three copies of Ren-1 c PPE. These results are in agreement with previous results reported by Shen et al. (32) and Shanmugam et al. (33) that HOX⅐PBX⅐MEIS complex does not exhibit transcriptional activity on artificial promoters containing HOX⅐PBX binding sites. Thus, other cofactor binding sites may be necessary in the natural targets for the function of HOX⅐PBX⅐PREP1/MEIS. We also cotransfected Ϫ4.1R1 with HOXD10, PBX1b, PREP1, or various combinations into As4.1 cells. The results showed that the transcriptional activity of the reporter gene was not affected by co-expression of these proteins (data not shown). It is possible that a cofactor, which is necessary for HOX⅐PBX⅐PREP1-mediated expression from Ϫ4.1R1, is limiting. Therefore, overexpression of HOX⅐ PBX⅐PREP1 would not further result in an increase in transcriptional activity. When the VP16 activation domain is attached to either HOXD10, PBX1b, or PREP1, it can substitute for the limiting cofactor to activate expression from construct 3XR1-TA.
How the PPE bound HOX⅐PBX1b⅐PREP1 complex interacts with other required transcriptional factors to regulate the mouse renin gene expression remains to be elucidated. It is possible that the PPE-binding proteins directly interact with a factor binding to an element located within or immediately outside the renal enhancer region located at Ϫ2.6 kb to bring the enhancer closer to the TATA box by looping out. Recent findings by Ryoo 7. HOXD10 (D10), PBX1b, and PREP1 can bind to the Ren-1 c PPE in As4.1 cells. As4.1 cells were cotransfected with reporter 3XR1-TA and various expression vectors in combinations. The luciferase (Luc) activity is expressed relative to that of plasmid 3XR1-TA.