Molecular Variation of the Human Angiotensinogen Core Promoter Element Located between the TATA Box and Transcription Initiation Site Affects Its Transcriptional Activity*

Recent genetic studies indicate that several molecular variants discovered in angiotensinogen (AG), the precursor of vasoactive octapeptide angiotensin II, could potentially be responsible for inherited predisposition to human blood pressure variation. We have previously shown that a ubiquitously expressed nuclear factor, AGCF1, bound to AGCE1 (AG core promoterelement 1 including the core nucleotides,CTCGTG, CTC-type) located between the TATA box and transcription initiation site (positions −25 to −1) is an authentic regulator of human AG transcription. In the present study, we showed that AGCF1 has biologically and immunologically similar properties to those of a helix-loop-helix nuclear factor USF1 and examined the effects of two other naturally occurring molecular variants (ATCGTG, ATC-type and ATTGTG, ATT-type) found in the AGCE1 position on the human AG transcriptional activity. Competitive gel-shift and transfection experiments demonstrated that the transcriptional activity for the CTC- and ATC-type promoters was 2.5 times higher than that for the ATT-type through the alteration of AGCF1-binding affinity. These results suggest the possible involvement of USF1 as a component in AGCF1 formation and the potential importance of AGCE1 variation in blood pressure regulation through human AG expression.

The renin-angiotensin system plays an important role in the regulation of blood pressure and electrolyte homeostasis. The reaction between renin and angiotensinogen (AG) 1 is the initial and rate-limiting step of this enzymatic cascade that generates the decapeptide angiotensin I, which is further processed to the functional octapeptide angiotensin II by angiotensin-converting enzyme (1)(2)(3). Because plasma AG concentration is close to the K m of the renin reaction, variation of plasma AG concen-tration can influence angiotensin II generation (4). Several observations indicate a direct relationship between plasma AG levels and blood pressure. First, plasma AG concentrations highly correlate with blood pressures in some patients (5), and associations between AG concentrations and hypertension have been demonstrated in families (6). Second, the overexpression of AG leads to elevated blood pressure in transgenic animals (7). Recently, Jeunemaitre et al. (8) showed that a common AG gene variant, M235T, was significantly linked to essential hypertension and was also associated with elevated plasma AG concentration. Whether M235T directly accounts for a physiological effect or acts as a marker for a causative mutation is as yet unresolved, they proposed that some other variants of the AG gene lead to a chronic increase in AG levels and thereby eventually to increased blood pressure.
AG is mainly synthesized in the liver and is secreted into the plasma through the constitutive pathway (9). Therefore, it is possible to suppose that the transcriptional regulation of the AG gene affects its plasma concentration. Kim et al. (10) have generated mice carrying two, three, or four functional copies of the murine wild-type AG gene at its normal chromosomal location and reported that plasma AG levels increased progressively with an increase in blood pressure. A recent study showed that the inhibition of AG transcription resulted in a reduction in plasma AG levels associated with a decrease in blood pressure of spontaneously hypertensive rats, by using synthetic double-stranded oligonucleotides as "decoy" cis-elements to block the binding of nuclear factors to the targeted promoter regions (11). These observations suggest an etiological importance of the transcriptional regulation of the AG gene.
We previously identified several regulatory elements of the human AG gene transcription including the upstream and downstream activating elements (12)(13)(14) and recently demonstrated that a ubiquitously expressed nuclear factor, AGCF1, bound to AGCE1 (AG core promoter element 1) including the core nucleotides (CTCGTG), an E box-like motif located between the TATA box and transcription initiation site, is an authentic regulator that mediates the responsiveness to multiple AG regulatory elements (15). Moreover, a recent genetic study found the three types of molecular variants, CTCGTG, ATCGTG, and ATTGTG, in the AGCE1 position of the human AG promoters (8). Therefore, we examined whether these variations affect the AGCF1-binding affinity and the human AG transcriptional activity. Here, we suggested the possible involvement of USF1, an E box-binding helix-loop-helix (HLH) transcription factor (16) as a component in AGCF1 formation and discussed the potentially important relationship between the naturally occurring AGCE1 variations and their transcriptional activity.
Transient Expression Assays-HepG2 cells were maintained in minimum essential medium containing 10% fetal bovine serum and nonessential amino acids. The cells were plated at a density of 5 ϫ 10 5 cells/60-mm dish and transfected 24 h later by calcium phosphate co-precipitation with reporter plasmids (3 g) and a ␤-galactosidase expression plasmid, pCMV-␤gal (1 g) to normalize transfection efficiency. In the co-transfection assay, 2 g of reporter plasmids were transfected with 1, 2, or 4 g of modulator plasmids and 1 g of pCMV-␤gal. Total amounts of DNA were adjusted 7 g by pcDNA3.
After 48 h of culture, ␤-galactosidase activities were measured (19), and cell extracts containing equivalent amounts of ␤-galactosidase activity were used for CAT assay (20). The extent of conversion of chloramphenicol to its acetylated form was measured with a Bio-imaging analyzer (Model BAS2000; Fujix, Tokyo, Japan).
Electrophoretic Mobility Shift Assays (EMSA)-Nuclear extracts from HepG2 cells were prepared using the protocol of Dignam et al. (21). Double-stranded DNA probe was end-labeled using [␥-32 P]ATP and T4 polynucleotide kinase. 5 g of nuclear extracts were incubated with 1 g of poly[d(I-C)] (Boehringer Mannheim) and end-labeled oligonucleotide (0.5 ng, approximately 15,000 cpm) at 20°C for 15 min in the presence or absence of the unlabeled oligonucleotides. The binding reaction was carried out in a solution containing 12 mM Hepes (pH 7.9), 60 mM KCl, 4 mM MgCl 2 , 1 mM EDTA, 12% glycerol, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride. In the supershift experiments, 2 l of USF1-specific antibody (Santa Cruz Biotechnology, CA) was added to the reaction mixture. The reaction mixtures were directly loaded onto 4.5% nondenaturing polyacrylamide gels containing 4% glycerol made in 1 ϫ TBE (90 mM Tris-HCl, pH 8.0, 89 mM boric acid, and 2 mM EDTA). After electrophoresis was performed at 150 V for 2.5 h at 4°C, the gels were dried and autoradiographed with an intensifying screen.

RESULTS
We previously showed that AGCF1 contained 31-, 33-, and 43-kDa proteins as major components by UV cross-linking (15). In the course of characterizing AGCF1, we noticed that its binding activity was heat-stable and affected by MgCl 2 (data not shown). These observations allowed us to speculate that the 43-kDa proteins identified would be USF1, which is classified as the members of basic HLH/leucine zipper family of transcription factors (16), although the nucleotide residue at position Ϫ18 in AGCE1 of the human AG core promoter was substituted to the thymidine (CTCGTG) from the adenine (CACGTG) in a consensus E box for USF. To examine whether USF1, a heat-stable MgCl 2 -sensitive 43-kDa transcription factor, is a component of AGCF1, we performed EMSA using the nuclear extracts prepared from HepG2 cells (Fig. 1). The DNAprotein complex formed by AGCF1 binding to AGCE1 was inhibited by molar excess of nonlabeled USF1 binding element of the adenovirus major late promoter (18) as well as AGCE1, although retinoic acid response element 1 of the cellular retinoic acid-binding protein II promoter (CRABPII-1) (22) and estrogen response element of vitellogenin promoter (vit-ERE) (23), which had partially related sequences to AGCE1 (Fig. 1A), did not compete for this binding at all (Fig. 1B, lanes 1-6). Similar results were obtained by using the adenovirus major late promoter as the probes (Fig. 1B, lanes 7-12). Furthermore, the addition of USF1-specific antibody to EMSA reactions generated a supershifted complex (Fig. 1B, compare lane 13 with lane 14), indicating that AGCF1 complex contains USF1.
The ability of USF1 to function as a transcriptional modulator of the human AG gene was analyzed by using a dominant negative form of USF1 protein (dnUSF) that lacked the NH 2terminal activation domain but possessed its DNA binding domain ( Fig. 2A) (24). The bacterially expressed GST-dnUSF fusion protein had the same binding specificity as compared with that of AGCF1 (data not shown). The expression of dnUSF dramatically reduced the transcriptional activity of the reconstituted human AG gene (13B2(3Ј)(ϩ)) but had little effect on that of the Am4 mutant (13Am4B2(3Ј)(ϩ)) that completely disrupted AGCF1 binding in vitro (15) in HepG2 cells (Fig. 2B). To further confirm the participation of USF1 in human AG transcription, an activator form of USF1 was co-transfected with the human AG reporters (Fig. 3A). In this assay, we used the minimal hAG core promoter (including only TATA box and AGCE1) as a reporter because of the saturated CAT activity in the reconstituted construct and to avoid the effects of hAG regulatory elements other than AGCE1. The expression of fulllength USF1 activated the human AG minimal promoter (DM12cat), but had little effect on that of the Am4 mutant (DM12Am4cat) in HepG2 cells (Fig. 3B). These results suggest that USF1 could regulate the human AG gene transcription in the AGCE1-dependent manner.
Interestingly, a previous genetic study identified the three types of molecular variants, CTCGTG, ATCGTC, and ATTGTG (8), localized to AGCE1 position of the human hAG promoters. In contrast, these E box-like motifs in AGCE1 were not conserved in the rodent angiotensinogen promoters (Fig. 4A) (25). These differences prompted us to examine the possibility that the AGCF1 binding affinity might be affected by the naturally occurring molecular variants by using competition assays. Although the DNA-protein complex formed by AGCF1 binding to the CTC-type AGCE1 was inhibited by a molar excess of the unlabeled CTC-type AGCE1, ATC-type AGCE1 partially prevented this complex formation. In particular, the ATT-type and rodent counterparts hardly competed for this binding (Fig. 4, B and C). These results indicate that the AGCF1 binding to AGCE1 is a species-specific interaction and that the molecular variation in AGCE1 alters the AGCF1 binding affinity.
AGCE1 plays an important role in mediating the responsiveness of multiple upstream and downstream cis-acting elements of the human AG gene to activate its promoter (15). As the disruption of AGCF1 binding, which resulted in the functional attenuation of the human AG regulatory elements, dramatically reduced the AG transcriptional activity (15), we examined the possibility that the alteration of AGCF1 binding affinity caused by the three types of naturally occurring mutations affected the AG transcription (Fig. 5). Although the transcriptional activity of the ATC-type in the reconstituted human AG gene (13(ATC)B2(3Ј)(ϩ)) was not significantly different from that of the CTC-type, the ATT-type represented about 40% of the transcriptional activity compared with that of the CTCtype. Taken together, these results demonstrate that molecular variation of the human AG core promoter, AGCE1, affects its transcriptional activity by alteration of the AGCF1-binding activity. DISCUSSION In the present study, we showed that AGCF1 has a DNA binding specificity similar to that of a HLH nuclear factor FIG. 2. Repression by the truncated USF1 protein of the reconstituted human AG gene. A, schematic maps of the truncated USF1 expression vector and the reconstituted human AG reporter plasmids. BR, HLH, and LZ represent a basic region, helix-loop-helix, and leucine zipper, respectively. Open boxes marked TATA box and white vertical lines represent TATA box and AGCE1 mutations, respectively. An Am4 mutation (from CGT to ATG at positions Ϫ17 to 15 in AGCE1) (13Am4B2(3Ј)(ϩ)) completely disrupted AGCF1 binding in vitro in HepG2 cells (15). B, HepG2 cells were transfected with 2 g of the CAT reporters, 1, 2, or 4 g of the modulators, and 1 g of the ␤-galactosidase expression plasmid (pCMV-␤gal) as an internal control for transfection efficiency. Total amounts of DNA were adjusted to 7 g by control vector pcDNA3. After a 48-h culture period, ␤-galactosidase activities were measured, and extracts containing equivalent amounts of ␤-galactosidase activities were used for CAT assays. The effect of dominant-negative USF1 on 13B2(3Ј)(ϩ) and 13Am4B2(3Ј)(ϩ) is indicated by Ⅺ and E, respectively. The CAT activity of 13B2(3Ј)(ϩ) is designated as 100, and each value of CAT activity represents the mean Ϯ S.E. for at least four independent experiments.

FIG. 3. Activation by an activator form USF1 of the human AG core promoter.
A, schematic maps of the full-length USF1 expression vector and reporter plasmids containing the human AG core promoter. AD, BR, HLH, and LZ represent activation domains, a basic region, helix-loop-helix, and leucine zipper, respectively. Open boxes marked TATA box and white vertical lines represent TATA box and AGCE1 mutation (Am4), respectively. B, transfection experiments and CAT assays were performed as described in Fig. 2B. The effect of USF1 on DM12cat and DM12Am4cat is indicated by Ⅺ and E, respectively. The CAT activity of DM12cat is designated as 1, and each value of CAT activity represents the mean Ϯ S.E. for at least four independent experiments. USF1 (26 -28) (Fig. 1). Furthermore, co-transfection experiments demonstrated that USF1 could regulate transcriptional activity of the human AG gene in the AGCE1-dependent manner (Figs. 2 and 3). Next, we examined the effects of naturally occurring molecular variants (CTCGTG, CTC-type; ATCGTG, ATC-type; and ATTGTG, ATT-type) localized to AGCE1 on the hAG transcriptional activity. Competitive EMSA and site-directed mutagenesis experiments demonstrated that the transcriptional activity for the CTC-and ATC-type mutations was significantly higher than that for the ATT-type by alteration of the AGCF1 binding affinity (Figs. 4 and 5).
USF was originally described as a transcription factor derived from HeLa nuclear extract that binds to an E box of the adenovirus major late promoter (16). This factor is also shown to be involved in the regulation of cellular genes, including the murine metallothionein I gene (29), the rat ␥-fibrinogen gene (30), the human growth hormone gene (31), the p53 gene (32), and the cardiac ventricular myosin light chain 2 gene (33). Furthermore, USF acts not only as a classical upstream activator, but also as a factor that interacts with initiator elements of a variety of core promoters, which can lead to markedly enhanced levels of basal transcription (34). We previously demonstrated that human AG promoter functioned without TATA box in the presence of AGCE1 (15). This initiator-like activity of AGCE1 may be explained by the presence of USF as a component of AGCF1 (Fig. 1), because USF could activate the basal level of transcription of the human AG core promoter in the AGCE1-dependent manner (Fig. 3).
Recently, Caulfied et al. (35) have shown significant linkage between hypertension and chromosomal regions including and close to the human AG gene, but they could not confirm association with the M235T mutation as a candidate marker for essential hypertension, probably due to ethnic differences in its allele frequency. As the functional effect of the M235T on essential hypertension was unclear, Lifton (36) pointed out the possible existence of functional variant(s) other than this mutation at AG locus. For example, we considered that the transcriptional regulation is one of the candidate control mechanism accounting for the variation of human angiotensinogen expression and provided the functional evidence that the transcriptional activities for the CTC-and ATC-type AG promoters were 2.5 times higher than that for the ATT-type by alteration of the AGCF1 binding affinity (Figs. 4 and 5). On the basis of the present results, it is suggested that the transcriptional activities of the CTC/CTC and ATC/ATC homozygotes or the CTC/ATC heterozygote are 2.5 times higher than that of the ATT/ATT homozygote and that the CTC/ATT and ATC/ATT heterozygotes is 1.75 times higher than that of the ATT/ATT homozygote.
A genetically chronic overactivity of the renin-angiotensin system could favor renal sodium reabsorption, vascular hypertrophy, and/or an increase in sympathetic nervous system activity, and predisposition to the development of common cardiovascular diseases. In this point, interestingly, associations between molecular variation of the AG gene and diseases including pre-eclampsia, coronary atherosclerosis, myocardial infarction, and nephropathy in insulin-dependent diabetes have FIG. 4. Effect of molecular variation of AGCE1 on the AGCF1 binding activities. A, alignment of the AGCE1 of the human, rat, and mouse AG genes. Optional alignments were generated using the HAR-PLT2 program (SDC-GENETICS). Dashes (-), indicating hypothetical deletions, were placed in the sequences to achieve maximum homology.
• indicates an identity among their nucleotides. The TATA box and E box-like sequences are denoted by boxes. An E box consensus motif was indicated by bold letters. The bases different from a consensus E box motif are indicated by underlines. Double-stranded versions of the indicated sequences were used in competition experiments. B, EMSA was performed as described in the legend to Fig. 1B. AGCF1 is indicated. C, linear plots of binding form are shown. The amount of AGCF1-DNA complex was quantified by imaging analyzer. q, Ⅺ, and f represent the effects of ATT-, ATC-, and CTC-type competitors (Fig. 4A), respectively, on AGCF1 binding activity to the CTC-type AGCE1. The complex formed in the absence of competitors is designated as 100, and each value of the complex represents the mean Ϯ S.E. for four independent experiments.  Fig. 4A. On the right, HepG2 cells were transfected with 3 g of the CAT reporters and 1 g of the ␤-galactosidase expression plasmid (pCMV-␤gal) as an internal control for transfection efficiency. CAT assays were performed as described in Fig. 2B. The CAT activity of 13B2(3Ј)(ϩ) is designated as 100, and each value of CAT activity represents the mean Ϯ S.E. for at least six independent experiments. been reported (37)(38)(39)(40)(41)(42). In addition to the systemic action of angiotensin II, the tissue function of this peptide is now considered to play an important role in local tissue regulation because components of the renin-angiotensin system have been demonstrated in a variety of tissues, such as adrenal glands, kidney, heart, and brain (43). Since AGCF1 is a ubiquitously expressed transcription factor, variants in AGCE1 may affect the levels of local AG synthesis, resulting in a change in the rate of angiotensin II formation.
Recently, Sato et al. (44) performed a case-control study in Japanese population to examine whether a genetic variant in AGCE1 is directly associated with increased risk of hypertension and suggested that a part of the previously reported genetic risk of hypertension associated with M235T might be explained by an increase in transcriptional regulation of AG induced by the AGCE1 polymorphisms. As discussed to date, the statistical significance that incriminates the AG gene locus is strongly associated with human hypertension (8,(35)(36)(37)(38)(39)(40)(41)(42)44). However, there was little direct evidence regarding the mechanism by which molecular variants of the AG gene affect the regulation of AG levels. Here we presented the first clue of human AG variations associated with the alteration of their transcription regulation, providing an experimental tool or a predictive marker to probe the predisposition of this disorder.