Upstream Stimulatory Factors (USF-1/USF-2) Regulate Human cGMP-dependent Protein Kinase I Gene Expression in Vascular Smooth Muscle Cells*

Cyclic GMP-dependent protein kinase I plays a piv-otal role in regulating smooth muscle cell relaxation, growth, and differentiation. Expression of the enzyme varies greatly in smooth muscle and in other tissues and cell types, yet little is known regarding the mechanisms regulating cGMP-dependent protein kinase gene expression. The present work was undertaken to characterize the mechanisms controlling kinase gene expression in vascular smooth muscle cells. A 2-kb human cGMP-dependent protein kinase I 5 (cid:1) -noncoding promoter sequence was characterized by serial deletion, and functional studies demonstrated that a 591-bp 5 (cid:1) promoter construct possessed the highest activity compared with all other constructs generated from the larger promoter. Analysis of the sequence between (cid:1) 472 and (cid:1) 591 bp from the transcriptional start site revealed the existence of two E-like boxes known to bind upstream stimulatory factors. Electrophoretic mobility shift assays and functional studies using luciferase reporter gene assays identified upstream stimulatory factors as the transcription factors bound to the E-boxes in the 591-bp promoter. Site-directed mutagenesis of the E-boxes abolished the binding of upstream stimulatory factor proteins and decreased the activity of the cGMP-dependent protein kinase I 591-bp promoter, thus confirming the involvement of these transcription factors in mediating gene expression. Cotransfection experiments demonstrated that overexpression of upstream stimulatory factors 1 and 2 increased cGMP-dependent All mutations were confirmed by sequencing the constructs. were performed to make sure that the mutated oligonucleotides failed to bind to nuclear extracts from cells. These mutated constructs, as well as the wild type, were used for transfection of bovine aortic SMC and A7r5 cells, as described above.

The serine/threonine protein kinase, cGMP-dependent protein kinase (PKG), 1 belongs to the large protein kinase family and mediates the actions of nitric oxide (NO) and natriuretic peptides in target cells (1)(2)(3). The effects of PKG activation in target cells include smooth muscle relaxation, inhibition of platelet aggregation, calcium homeostasis, intestinal secretion, cell growth, differentiation, and gene regulation (4,5).
Two genes encoding mammalian PKG have been identified, type I and type II (6 -8). The type I PKG (PKG-I) is expressed as two isoforms created by alternative mRNA splicing of the two first coding exons, and these two isoforms are referred to as type I␣ and type I␤. Because these two isoforms differ only in the initial coding region, they contain an identical catalytic domain with identical substrate specificity. Although the two isoforms of PKG-I are rather widely expressed in mammalian tissues, three cell types, smooth muscle, cerebellum Purkinje cells, and platelets, contain the highest levels of the enzyme. The type II PKG is expressed most abundantly in intestinal epithelium, chondrocytes, and certain regions of the brain but is not expressed in smooth muscle (9 -12).
In vascular SMC, both PKG-I␣ and -I␤ isoforms are abundant, but expression is highly variable. For instance, upon subculturing vascular SMC derived from rat aorta, PKG-I expression decreases as the cells modulate to a more dedifferentiated, fibroproliferative phenotype. Furthermore, PKG-I expression is reduced in response to balloon catheter injury in rat carotid artery and porcine coronary artery (13,14) and in fetal pulmonary artery in response to hypoxic conditions (15). Recently, our laboratory has shown that inflammatory mediators such as interleukin-1␤ and tumor necrosis factor-␣ rapidly down-regulate PKG-I mRNA and protein expression in freshly isolated vascular SMC from bovine aorta (16). These results suggest that one component of the proliferative response to injury by vascular SMC is the suppression of PKG-I expression and the disruption of NO signaling in the cells. The mechanisms underlying the suppression of PKG expression are not known, and virtually nothing is known regarding the mechanisms regulating PKG-I gene expression at the transcriptional or translational levels.
The human gene for PKG-I, including the promoter sequence, was characterized in 1997 (17). The human PKG-I 5Ј-flanking region (600 bp) demonstrates about 80% homology with the same region in other species (i.e. rat and mouse) and has no typical TATA-box or CCAAT-box. Several previous observations indicate that genes with tissue-specific expression, but with TATA-less promoters, utilize GC-rich promoter sequences that bind Sp proteins to regulate transcription (18,19), and our laboratory found that Sp proteins might be important for basal PKG-I expression in SMC (20). Given the importance of the NO/cGMP pathway in regulating not only SMC contractility but also growth and phenotype of the SMC (21)(22)(23), it seemed important to characterize the mechanism(s) governing PKG-I gene expression to understand further the different physiological and pathophysiological roles of this kinase in the vasculature. Hence, the goal of the present study was to characterize and identify 5Ј-promoter transcriptional regulation involved in the expression of human PKG-I in vascular SMC. The current findings suggest an important role for upstream stimulatory factors (USF) that bind to tandem E-like boxes in the 5Ј-region of the proximal PKG-I promoter in human vascular SMC. 32 P]ATP (3000 Ci/mmol) was purchased from PerkinElmer Life Sciences and Analytical Sciences (Billerica, MA). Restriction and modifying enzymes, reporter gene vector (pGL3basic), dual luciferase system, transfection reagent Tfx-20, and other biological compounds were purchased from Promega (Madison, WI). USF-1 (psv-USF1, PN3) and USF-2 (psv-USF2, PN2) expression vectors were provided by Dr. M. Sawadogo (University of Texas, Houston). Empty vector pSG5 was from Stratagene (La Jolla, CA). Antibodies, anti-USF-1 (C-20), anti-USF-2 (N- 19), and anti-IgG, were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Synthetic oligonucleotides and constructs sequencing were provided from MWG Biotec (High Point, NC).

Chemicals and Reagents-[␥-
Smooth Muscle Cell Culture-Bovine aortic SMC in passage 2-8 were cultured as described previously (24). Embryonic rat aortic A7r5 cells were purchased from American Type Culture Collection (ATCC, Manassas, VA). Cells were maintained in Dulbecco's minimal essential medium containing 10% fetal bovine serum and 50 g/ml gentamicin. The routine subculturing procedure was performed to split the cells 1:4.
Cloning of the PKG-I Promoter and Construction of Different Plasmids-The 5Ј-flanking region in the PKG-I 2-kb promoter was generated by PCR using human genomic DNA from the lung (Promega, Madison, WI) as a template and primers designed from the published PKG sequence, GenBank TM accession number Z92867 (sense, 5Ј-gactgagcacccagcatgtcttttcta-3Ј; antisense, 5Ј-gctgaagctttttcactgagcccctccgcg-3Ј). The cloned genomic fragment was ligated into the SacI/HindIIIdigested pGL3basic vector containing luciferase as a reporter gene. The cloned PKG-I 2-kb promoter served as a template to generate serial deletions in the PKG-I 2-kb promoter. All PCR products (Ϫ472, Ϫ591, Ϫ800, Ϫ1000, Ϫ1200, Ϫ1500, and Ϫ1700 bp) were purified and subcloned in the pGL3-basic vectors. Plasmids p50, p70, and p120, corresponding to positions Ϫ472 to Ϫ521 bp, Ϫ522 to Ϫ591 bp, and Ϫ472 to Ϫ591 bp of the human PKG-I untranscribed region, respectively, were generated by PCR using the PKG-I 591-bp promoter fragment as the template and the primers reported in Table I. The PCR fragments were subcloned in pGL3basic vector. In some experiments, the 70-and 50-bp oligonucleotides were inserted upstream of the human PKG-I 472-bp promoter. The integrity and fidelity of all constructs were verified by DNA sequencing.
Transient Transfection and Reporter Gene Assays-Bovine aortic SMC or rat aortic A7r5 cells were seeded in 24-well plates at 60 -70% of confluence and were grown overnight. After 16 -20 h, cells were transfected for 1 h, in the absence of serum, with 250 ng of different PKG-I promoter constructs using 1 l of transfecting reagent Tfx-20. Forty eight hours later, cells were washed with PBS and lysed, and luciferase assays were performed. The activity of the cotransfected control reporter gene (pRL-null vector, 50 ng, Promega) provided an internal control. The experimental reporter genes were normalized to the activity of the internal control to minimize the variability caused by differences in cell death and transfection efficiency. Dual luciferase (Dual-Luciferase Reporter Assay System, Promega) was quantified using a luminometer (TD 20/20 Luminometer Turner Design). The data are expressed as relative luciferase, where firefly luciferase activity was normalized to Renilla luciferase activity generated by cotransfecting with pRL-null vector. In cotransfection experiments, USF-1 or USF-2 expression vectors were used at different concentrations as indicated in the figures. The total amount of DNA was kept constant by adding the empty vector (pGL3 basic or pSG5).
Nuclear Extract Preparation and Electrophoretic Mobility Shift Assays (EMSA)-Nuclear extracts from bovine aortic SMC were prepared using NE-PER nuclear and cytosolic extraction reagents following protocols provided by the manufacturer (Pierce). 70-and 50-bp DNA fragments were prepared by PCR using the PKG-I 591-bp promoter as a template with the primers described in Table I. E1-E2 oligonucleotides (see Table II) and E2-box binding sites in the PKG-I promoter were synthesized by MWG Biotec. The oligonucleotides were annealed, endlabeled with [␥-32 P]ATP and T4 polynucleotide kinase, and purified using microSpin G-25 columns (Amersham Biosciences). Nuclear extract proteins (10 g) from control bovine cells were incubated with the 32 P-labeled probes (50,000 cpm) in a buffer containing 10 g/ml bovine serum albumin, 10 mmol/liter Tris-HCl (pH 7.5), 50 mmol/liter NaCl, 1 mmol/liter dithiothreitol, 1 mmol/liter EDTA, and 5% glycerol (total volume 20 l). To minimize nonspecific binding, 2 g of poly(dI-dC) was added to the reaction. The binding reaction was carried out on ice for 30 min. The DNA-protein complexes were resolved by 5% nondenaturing PAGE at 12 V/cm in low ionic strength buffer (0.25ϫ Tris/borate/EDTA) at room temperature. For mobility supershift assays, antibodies toward USF-1, USF-2, or control IgG were added to the binding reactions and incubated at 4°C for 10 min before the addition of the radioactive probe. In experiments involving competitive EMSA, the unlabeled oligonucleotides (A-E, E-box1, E-box2, mE1-box, mE2-box, 50 and 70 bp, USF-1, GATA, SREBP-1, Oct-1, Sp1, E1-E2, mE1-E2, E1-mE2, or mE1-mE2) at 50-fold excess (Table II) were preincubated with nuclear extracts for 10 min on ice before the addition of the radioactive probe and then further incubated for an additional 20 min. Gels were then dried and exposed to autoradiographic film.
Site-directed Mutagenesis-The human PKG-I 591-bp fragment cloned into pGL3basic served as the template for site-directed mutagenesis, which was employed to change the minimal consensus sequences for the putative USF-1/USF-2-binding sites. Mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). The E1-box was mutated (underlined) to cttttg (wt, cacatg); the E2-box was mutated to cttttg (wt, caagtg), and the double mutated USF-binding sites consisted of mutated E1-box and E2-box sequences (mE1-box/mE2-box). All mutations were confirmed by sequencing the constructs. EMSAs were performed to make sure that the mutated oligonucleotides failed to bind to nuclear extracts from cells. These mutated constructs, as well as the wild type, were used for transfection of bovine aortic SMC and A7r5 cells, as described above.
Western Blot Analysis-Western blots were performed to verify the overexpression of USF-1 or USF-2 proteins in control or transfected cells. Total protein extract (40 g) was separated on 10% SDS-PAGE and transferred to nitrocellulose membranes. The blots were probed with anti-USF-1 or anti-USF-2 antibodies. The signal was detected by enhanced chemiluminescence (Pierce).

Serial Deletions of the Human PKG-I 2-kb Promoter
Reveal a 591-bp Active PKG-I Promoter-Previously, our laboratory demonstrated that Sp1/Sp3-binding sites located within 100 bp upstream of the ATG translation initiation codon were important for basal human PKG-I promoter activity in vascular SMC (20). This small region of the 5Ј-untranslated PKG-I gene was about 60% conserved when comparing human versus rodents (rat and mouse) by using ClustalW (1.82) multiple sequence alignment (www.ebi.uk). 600-bp upstream of this region, however, there is more conservation of sequences (about 80% when comparing human PKG-I gene versus rat and mouse). In the current study, we have expanded these initial findings to identify additional mechanisms that may regulate human proximal PKG-I promoter activity in SMC. Therefore, the PKG-I 5Јflanking region (2 kb) was cloned by PCR using human genomic gctagagctcggatcaagtggcataaa gcgcaagcttcttgaaatagcaac DNA as a template. Serial deletions were generated using the cloned PKG-I 2-kb promoter as a template by using the primers shown in Table II. Fig. 1 demonstrates that the highest PKG-I promoter activity was contained within a 591-bp region of the 2-kb promoter when studied using luciferase reporter gene assays. As shown in Fig. 2A, the DNA sequence of the 591-bp region reveals the existence of a number of putative transcription factor binding sites, including two E-boxes capable of binding USF-1/2 transcription factors, to be discussed later. In fact, the activity of PKG-I 591-bp was ϳ2.5-5-fold more active in driving luciferase expression compared with the 472-bp promoter studied previously by our laboratory in aortic SMC (20).

Identification of Transcription Factor(s) Mediating Human PKG-I 591-bp Promoter
Activity-In order to identify the potential transcription factors that might bind to the human PKG-I 591-bp promoter compared with the 472-bp promoter region, 70-and 50-bp oligonucleotides were generated corresponding in the region Ϫ472 to Ϫ591 bp. These oligonucleotides were radiolabeled, and EMSAs were performed using nuclear extracts from bovine aortic SMC. As shown in Fig. 3A, one major band was retarded on the gel with the 70-bp probe (lane 2). In order to identify the potential binding elements within the 120-bp fragment between Ϫ472 and Ϫ591 bp that may be responsible for the increase in the binding activity, this fragment was further dissected into five regions (A-E). These regions were selected based on the potential transcription factors that might bind to this 120-bp fragment ( Fig. 1 and Table II). The results in Fig. 3A show that the binding of the retarded band was decreased significantly with excess unlabeled oligonucleotide B (lane 4, **) and partially decreased with excess unlabeled oligonucleotide C (lane 5, *). In contrast, oligonucleo-tides A, D, and E did not reduce the intensity of binding to the retarded band (Fig. 3A, lanes 3, 6, and 7). As shown in Fig. 3B, three or possibly four bands were retarded when the 50-bp DNA fragment was used as a probe. In unlabeled probe competition assays, excess of oligonucleotides A-C eliminated the binding of band 1 and decreased the binding of band 4 ( Fig. 3B, lanes 4 -6), whereas excess of oligonucleotides D and E eliminated the radioactive binding in bands 2 and 3 (lanes 7 and 8).
Computer analysis of the sequences of the 70-and 50-bp region of the human PKG-I promoter region revealed that oligonucleotides B and C might compete with GATA-1/3-, USF-1/2-, Oct-1-, or SREBP-1-binding sites. Therefore, consensus sequences for each of these binding sites were generated (see Table II), and EMSA competition assays were performed to identify which oligonucleotide would diminish the binding of 70-and 50-bp probes. The results in Fig. 4A show that an excess of USF-1/2 consensus binding oligonucleotide abolished the binding of the retarded band when the 70-bp DNA fragment was used as a probe (lane 4, *), whereas other unlabeled consensus oligonucleotides decreased only slightly the binding or failed to abolish it altogether. When the 50-bp DNA fragment was used as a probe (Fig. 4B), only consensus USFbinding oligonucleotides decreased bands 1 and 4 (lane 4), whereas the SREBP-binding oligonucleotide reduced the intensity of band 1 and abolished band 4. Oct-1-and GATA-1/3binding oligonucleotides decreased only slightly the intensity of the band 4 ( Fig. 4B, lanes 6 and 7). Finally, an excess of unlabeled Sp1 consensus oligonucleotide abolished the binding of bands 2 and 3 (Fig. 4B, lane 8), indicating that Sp proteins bound to the 50-bp DNA fragment. These data demonstrate that USF-1/2 is the predominant transcription factor binding to  Table II. Bovine or A7r5 cells were transiently transfected with different constructs, and relative luciferase activities were expressed after normalization to an internal control. Data are representative of at least three independent experiments. *, p Ͻ 0.05.
70-bp probe, whereas both USF and Sp proteins bind to 50-bp probe.

USF-1 and USF-2 Transcription Factors Bind to E-box Motifs in the Human PKG-I 591-bp
Promoter-To confirm the binding of USF proteins to the 70-bp DNA promoter fragment, additional EMSAs were performed using competition with unlabeled oligonucleotides and specific antibodies directed against USF-1 and USF-2 transcription factors. The results shown in Fig. 5A demonstrate that only one band is retarded on the gel (lane 2) when the 70-bp DNA fragment was incubated with aortic SMC nuclear extract protein. A 50-fold excess of unlabeled E1-box or E2-box oligonucleotide abolished the binding at 70 bp (Fig. 5A, lanes 3 and 4), whereas an excess of unlabeled mutated E1-box (mE1-box, ccgccagatcttttggtaattgtt) or mutated E2-box (mE2-box, caaggatcttttggcataa) oligonucleotides reduced only slightly the binding of the probe (lanes 5 and 6). Preincubation of aortic SMC nuclear extract with anti-USF-1 antibody caused a supershift of two bands (Fig. 5B,  lanes 2 and 3). An anti-USF-2 antibody resulted in a marked reduction of the intensity of the retarded band (Fig. 5B, lanes 4 and 5) but did not result in a supershift of the band. As a control, the anti-IgG itself did not affect the intensity (binding) of the retarded band. These results suggest that USF-1, and most likely USF-2, were the major transcription factors binding to the 70-bp DNA fragment of the human PKG-I 591-bp promoter. The lack of supershift using anti-USF-2 may be due to a lower affinity of the antibody compared with anti-USF-1 or, alternatively, to the ability of the polyclonal antibody to interact with the DNA binding domain of USF-2.
Similarly, we performed EMSA competitions and immunoshifts to identify which, if any, transcription factors bind to the 50-bp DNA promoter fragment. As shown in Fig. 5C, several bands were retarded (lane 2). An excess of unlabeled 50-bp oligonucleotide abolished the binding (Fig. 5C, lane 3). Compe- tition with unlabeled E1-box or E2-box oligonucleotides almost abolished band 1 and reduced significantly band 4 (Fig. 5C,  lanes 4 and 5) as did USF consensus oligonucleotides. However, mutated E1-box or mutated E2-box unlabeled oligonucleotides had no effect on the binding to band 1 (lanes 6 and 7). When nuclear extracts from aortic SMC was preincubated with anti-USF-1 or anti-USF-2 antibodies, there was a reduction in the intensity of the retarded bands (Fig. 5D, lanes 5 and 8).
Because the sequence of the putative E2-box (CAAGTG) has not been reported to be a USF-binding sequence, additional EMSA experiments were performed to determine the nature of the binding activity to the E2-box of the 50-bp probe. As shown in Fig. 6A, several bands were retarded when the E2-box sequence shown in Table II was used as a probe. In competition assays, an excess of unlabeled E1-box or E2-box oligonucleotide abolished binding to the retarded bands 1 and 2 (Fig. 6A, lanes  3 and 4). However, competition with a mutated E1-box or a mutated E2-box oligonucleotide did not affect the binding (Fig.  6A, lanes 5 and 6). Similarly, excess of unlabeled SREBP-1, GATA-1/3, or Oct-1 consensus oligonucleotides did not reduce binding (Fig. 6A, lanes 7-9). On the other hand, anti-USF-1 antibody reduced the intensity of the retarded bands and induced a supershift of band 2 (Fig. 6B, lanes 2-4). Preincubation of nuclear extracts with anti-USF-2 antibody produced the reduction in intensity of binding to band 2 (Fig. 6B, lanes 6 and  7) but did not supershift the band, as observed earlier. These data suggest that USF-1, and likely USF-2, transcription factors are binding to the 50-bp DNA fragment from the human PKG-I 591-bp promoter.
Functional Analysis of the 70-and 50-bp DNA Sequences for Human PKG-I 591-bp Promoter Activity-From the binding data shown in Fig. 4, we anticipated that sequences contained within the 50-and 70-bp DNA fragments might be necessary for maximal promoter activity. To determine the role of these sequences of the human PKG-I 591-bp promoter to impart activity to the promoter, constructs were generated where both the 50 and the 70 bp were separately inserted upstream of human PKG-I 472-bp promoter, ligated in pGL3 vectors, and luciferase activities were determined following transfection of the constructs in the bovine aortic or A7r5 SMC. As shown in Fig. 7, luciferase activities produced in the cells following transfection of p50-472 or p70-472 were lower or similar to the activity of p472 alone. On the other hand, the activity of the p120 construct alone was similar to that of the p472 promoter but lower than the p591 PKG-I construct that contains not only the sequences of the 120-bp region upstream of the 472-bp region, but also contain the Sp-binding sites identified previously (20). Thus, these data suggest that both regions of the

FIG. 5. EMSA analysis of USF binding to70-and 50-bp probes.
For EMSA competition assays, nuclear extracts (NE) from bovine aortic SMC were preincubated with wild type E1-box, wild type E2-box, mutated E1-box, or mutated E2-box plasmid DNA before adding the radiolabeled 70-(A) or 50-bp (C) probes. For immunoshift, nuclear extracts were preincubated with anti-USF1, anti-USF2, or anti-IgG antibodies at the indicated concentrations 10 min before adding the 70-(B) or 50-bp (D) probes. The arrows represent the retarded bands, and the shifted bands are represented with SS. The asterisks represent the decrease or the abolition of the binding. Data are representative of three experiments.

FIG. 6. Competition and immunoshift assays demonstrating USF binding to the E-boxes within the 591-bp PKG-I promoter.
E2-box oligonucleotide contained in the 50-bp DNA sequence was synthesized and used in EMSAs. A, competition assays, nuclear extracts (NE) from bovine aortic SMC were preincubated with 50-fold excess of E1-box, E2-box, mE1-box, mE2-box, GATA, Oct1, or SREBP1 oligonucleotides before the addition of E2-box radiolabeled probe. B, immunoshift assay, nuclear extracts were preincubated with anti-USF1, anti-USF2, or anti-IgG antibodies at the indicated concentrations for 10 min before the addition of the radiolabeled probe (E2-box). The arrows indicate the major retarded bands, and the supershifted bands are represented by SS. The asterisks indicate decreases in binding. The EMSA are representative of at least three separate experiments. p50 and the p70 are necessary for maximal human PKG-I 591-bp promoter activity.
The 120-bp region contains the two E-boxes that binding studies suggest could contribute to the activity of the human PKG-I 591-bp promoter. Furthermore, the EMSA data indicated that USF proteins bind to the region defined between Ϫ591 and Ϫ472 bp of the promoter. By using a site-directed mutagenesis approach, mutants of 3 bp, important for the function of two E-boxes, were generated to disrupt USF binding, and luciferase activity was assessed. As shown in Fig. 8A, mutation of either the E1-box (mE1) or the E2-box (mE2) equally reduced luciferase activity compared with the control (wt), suggesting that the activation of the human PKG-I 591-bp promoter by USF transcription factors was E-box-dependent. Most unexpectedly, the double mutated construct (mE1/mE2) was able to drive the luciferase activity even more than the wild type construct containing both intact E-boxes. This was likely due, however, to the creation of potentially new binding sites by mutation for other factors that could alter transcriptional activity of the promoter. In any event, this would be a nonphysiologic event, and as shown below, the double mutated construct did not bind USF.
USF-1 or USF-2 Overexpression Induces Transactivation of the Human PKG-I 591-bp Promoter-Further evidence for the involvement of USF proteins in activating the human PKG-I 591-bp promoter activity was obtained following cotransfection of bovine aortic SMC with the PKG-I 591-bp/luciferase vector and USF-1 or USF-2 expression vectors. As shown in Fig. 8B, overexpression of USF-1 or USF-2 resulted in an increase in human PKG-I 591-bp promoter activity, with an approximate 8-fold induction of activity with the highest DNA concentration (100 ng) for expression of the transcription factors. The transactivation of the human PKG-I 591-bp promoter with the combination of USF-1 and USF-2 expression vectors was not additive (data not shown), suggesting that the USF proteins bind to the same sites on the promoter sequence. In contrast, higher concentrations of USF-1 or USF-2 expression vectors (Ͼ200 ng) induced a decrease in promoter activity (data not shown). In a control experiment, empty vector transfection (pSG5, 25-100 ng) did not affect the activity of human PKG-I 591-bp promoter (data not shown).
The importance of USF transcription factors in regulating the activity of the human PKG-I 591-bp promoter was further evaluated using mutated E1-box (pmE1), mutated E2-box (pmE2), or double mutated E1-box and E2-box (pmE1-mE2) constructs. The results in Fig. 9A demonstrate that whereas USF-1 or USF-2 activated the wild type promoter (as reported above in Fig. 8B), USF-1 or USF-2 overexpression failed to activate the double mutated construct pmE1/mE2 promoter. On the other hand, the single mutated pmE1 or pmE2 promoter constructs were partially activated by USF-1 or USF-2 expression compared with the wild type promoter. These results demonstrate that both the E1-box and the E2-box are required for full USF-dependent activation of human PKG-I 591-bp promoter.
Further evidence for the involvement of USF-1 or USF-2 to activate both E-box sequences was obtained using p50, p70, and p120 luciferase constructs. As shown in Fig. 9B, overexpression of USF-1 or USF-2 had minimal effects on driving luciferase activity when compared with the activity of the 591-bp promoter itself. The p120 luciferase plasmid containing both the E1-box and the E2-box, on the other hand, demonstrated a greater degree of transactivation by USF-1 or USF-2 when compared with the p50 or p70 plasmids (containing only FIG. 7. Functional analysis of E-box motifs using luciferase assays. The oligonucleotides for 50, 70, and 120 bp were generated by PCR using the human PKG-I 591-bp promoter as a template and were cloned upstream of pGL3 (basic vector, p0) or PKG-I 472-bp promoter construct. Bovine aortic SMC and A7r5 SMC were then transfected with 250 ng of each plasmid, and luciferase assays were determined 48 h later. Data are representative of at least three independent experiments performed in triplicate. one E-box). However, the activity of the p120 construct was still much lower than that obtained with human PKG-I 591 bp. As a control, the empty expression vectors (V0, pGS5) did not affect the activities of any these luciferase constructs. These experiments demonstrate that USF-1 and USF-2 transcription factors activate the human PKG-I 591-bp promoter through the recognition of both E-boxes.
USF Binding to E1-and E2-boxes-To confirm the role for E-boxes in the binding of USF-1/2 in the PKG-I 591-bp promoter, new oligonucleotides containing both intact E1-box and E2-box (E1-E2), mutated E1-box (mE1-E2), mE2-box (E1-mE2), or both mutated E1-box and E-2-box (mE1-mE2) (see Table II for the sequences) were synthesized for EMSA binding analysis. As shown in Fig. 10A, the wild type E1-E2 box oligonucleotide provided three distinct bands when incubated with bovine aortic SMC nuclear extract (lane 2). However, the double mutated mE1-mE2 did not cause any mobility shift (Fig.  10A, lane 5). Mutation of the E1-box alone (mE1-E2) reduced the binding of band 1 (Fig. 10A, lane 3), whereas the mutation of the E2-box (E1-mE2) reduced the binding of bands 1 and 3 (Fig. 10A, lane 4). These binding assays confirm that both E-boxes bind nuclear protein, presumably USF-1 and -2, and explain why the double mutated mE1/mE2 PKG-I 591-bp promoter could not be transactivated by overexpressed USF-1 or USF-2. In the competition assays shown in Fig. 10B, using E1-E2 as a probe, the double mutated mE1-mE2 did not affect the binding of the wild type probe (E1-E2). In the EMSA immunoshift assay (Fig. 10C), anti-USF-1 antibody induced a supershift of band 1 (lanes 2 and 3), whereas anti-USF-2 antibody reduced the intensity of the retarded band (lanes 4 and 5). Taken together, these results demonstrate that both the E1box and the E2-box bind USF proteins, and these DNA sequences are functionally necessary for mediating the full transcriptional activity of the human PKG-I promoter in vascular SMC.

DISCUSSION
The NO-cGMP signaling pathway plays an important role in vascular smooth muscle relaxation, growth, and differentiation, and PKG is a critical enzyme for mediating these effects. Although there are many studies that address the effects of PKG on protein phosphorylation in SMC and other cells, there are very few studies addressing the mechanisms relating to PKG expression itself.
There are two PKG genes in humans and other species, and the PKG-I gene is the form that is abundantly expressed in SMC. The PKG-I is also expressed as two isoforms, I␣ and I␤, that differ only in the first exon that encodes the N-terminal domain containing the leucine zipper motif and auto-inhibitory site. Vascular SMC appear to be particularly rich in the I␣ isoform compared with nonvascular SMC. PKG-I expression decreases in cultured vascular SMC derived from the rat, particularly as the cells modulate to a fibroproliferative phenotype from a differentiated, contractile phenotype. It has also been shown that restoration of PKG-I expression to these fibroproliferative cells through transfection or adenoviral gene transfer results in the phenotype returning to a more differentiated, contractile one (21,22). In contrast to rodent PKG expression, PKG-I expression in human or bovine vascular SMC is maintained as the cells are cultured in vitro (25). Because of these findings, it seems important to begin to characterize the mechanisms regulating PKG-I gene expression in vascular SMC.
The human PKG-I gene has been partially characterized by Orstavik et al. (17) and appears to be a large (ϳ1.3 Mb) gene located to chromosome region 10q11.2. There are 19 different exons in the gene with the first two exons (i.e. the 1␣ and 1␤) spliced to give rise to the I␣ and I␤ isoforms of PKG. The human PKG-I␣ isoform cDNA has also been isolated and shown, using 5Ј-rapid amplification of cDNA ends, to contain a very short (40 bp or less) 5Ј-untranslated region. Thus, it was suggested that sequences upstream from the first exon in the 5Ј-flanking region contribute to transcriptional initiation. Previously, we reported that a 100-bp sequence upstream from the translational start site for the I␣ exon contained tandem GCrich islands and was capable of serving as a promoter for luciferase expression in vascular SMC transfection assays (20). Because this region, which is 60% conserved across species, contains 40 bp of untranslated sequence of the 5Ј Ϫ1␣ exon, we propose that there was sufficient active transcriptional activity to serve as a basal PKG-I promoter in these cells. However, both the size and activity of this minimal promoter sequence, as well as the conservation of additional DNA sequence 500 bp or more 5Ј from the minimal promoter sequence, suggest the existence of additional regulatory elements within the PKG promoter that might be involved in PKG expression.
In the present study, we found that two conserved E-box motifs binding USF proteins located ϳ500 -600 bp upstream of the translation start site of the human PKG-I gene are responsible for high levels of PKG-I expression in vascular SMC. Starting with a 2-kb genomic DNA sequence upstream of the translation start site, we found that DNA sequences between Ϫ472 and Ϫ591 bp are required for maximal PKG-I gene expression. Further- more, the approximate 120-bp sequence located between Ϫ472 and Ϫ591 bp contains key transcriptional regulatory activity for PKG-I expression based on EMSA and reporter gene assays. When analyzed using the Transfac program to identify putative transcriptionally active sites, a number of potential transcription factors such as USF, CREB, AP-1, GATA-1/3, NKx, ADR, Sp, SREBP-1, and Oct-1 were highlighted.
The identification of binding sites for a number of transcription factors in the PKG-I promoter region coupled with the redundancy of the sequences for recognizing these transcription factors necessitated a thorough analysis of the promoter region using a combination of competition binding, immunoshift, and luciferase assays to determine which, if any, of these transcription factors plays a functional role in regulating PKG-I promoter activity. By using these approaches, we found that the USF-1 and USF-2 transcription factors are the major proteins that function to transactivate human PKG-I gene expression by activating the 5Ј-promoter.
USF proteins belong to the basic helix-loop-helix leucine zipper family of transcription factors, and they share similar binding specificity in vitro and in vivo (26). The major USF isoforms, USF-1 (43 kDa) and USF-2 (44 kDa), are ubiquitously expressed and encoded by two separate genes (26). USF proteins bind as either homodimers or heterodimers to a consensus DNA sequence, CACGTG, termed E-box (27). However, the DNA-binding sites for USF proteins in genes appear to be not so strictly conserved, so these proteins have been reported to bind to a variety of E-box motifs including CGCGTG, CACATG, CAGCTG, CACCTG, CATCTC, CCCGTG, and CAGGTG to regulate gene expression (28 -30). In this study, both USF-1 and USF-2 are present in bovine aortic and A7r5 SMC as analyzed by Western blotting. However, the concentrations appear to be somewhat limiting in SMC in that their overexpression increases PKG-I 591-bp promoter activity in the cells. These studies also suggest that the human PKG-I promoter is quite sensitive to USF. Because we were not able to see additive effects when cells were cotransfected with both USF-1 or USF-2 expression vectors, there appears to be no predominance of one factor over another or that there is a preference for one of the two E-boxes for the binding of USF-1 or USF-2 proteins.
The evidence that USF, by binding to E-box sites in the PKG-I promoter, is responsible for the high level of PKG-I gene expression in vascular SMC is compelling. First, only USF antibodies produce either a supershift or diminished band intensity in EMSA immunoshift assays. Furthermore, the binding site for the USF proteins was confirmed to be both E1-box and E2-box sites in the region Ϫ472 to Ϫ591 bp of the PKG-I promoter. Second, by using a competition assay approach, an excess of unlabeled E2-box (CAAGTG) was able to compete and abolish both the binding of E1-probe (CACATG) and the E2probe just as the consensus USF sequence did. Based on these data, the E-like box CAAGTG sequence identified in the PKG-I promoter region can be added to the list of sequences interacting with USF proteins in addition to NK-2 homeodomain proteins (31). However, we cannot completely rule out that the E2-box may bind to some unidentified transcription factor given the multiple retarded bands (see for example, Fig. 6). Third, site-directed mutagenesis directed to the consensus sequence elements in both the E1-box and the E2-box demonstrates that both regulatory elements are required for full human PKG-I 591-bp promoter activity. Moreover, the individual contribution of each E-box in the promoter appears to be equivalent because single mutation of either element abolished the promoter activity by more than half in each case. Finally, overexpression of USF proteins transactivate PKG-I promoter constructs in transfection assays. In both singly mutated E-box constructs, overexpression of USF-1 or USF-2 does not activate PKG-I promoter, further supporting the notion that both Eboxes are necessary for the activation of the PKG-I promoter.
Transfecting cells with p50, p70, or p120 constructs further demonstrated the essential role of the E-boxes in mediating transcription. Effectively, p50 and p70 constructs are able to drive low levels of luciferase expression in aortic SMC, indicating that these DNA fragments by binding USF proteins are functional. However, their respective activities are significantly less than that of the p120 construct, again confirming that both E-boxes are needed for full activity. Moreover, the p120 activity is similar in reporter gene assays to that of the PKG-I 472-bp construct studied previously (20) but is less than the full 591-bp promoter construct. These data suggest that the 120-bp DNA fragment is sufficient to direct transcription efficiently through the tandem E-boxes. It is likely that this arrangement serves as a core for recruiting or stabilizing transcriptional complexes because it has been shown that USF is able to interact with TFIID (32) and can also mediate transcription in the absence of the TATA-box (33).

FIG. 10. E-boxes bind USF proteins exclusively.
A, the oligonucleotides containing both intact E1-box-E2-box (E1-E2), mutated E1-box (mE1-E2), mutated E2-box (E1-mE2), or both mutated E1-box and E2-box (mE1-mE2) (see Table II for the sequences) were synthesized and radiolabeled, and EMSA binding assays were performed. B, competition assays were performed by incubating bovine aortic SMC nuclear extracts with 50-fold excess of cold probes before adding E1-E2 radiolabeled probe. C, bovine aortic SMC nuclear extracts were incubated with anti-USF1 or anti-USF2 antibodies before adding the E1-E2 probe. The data in the figure are representative of two independent experiments.
The E-box motifs identified in the human PKG-I gene are also present in the rat and mouse PKG-I 5Ј-upstream sequences (see Fig. 2B). It is therefore possible that USF proteins regulate rodent PKG-I promoters in an analogous fashion to the human PKG-I promoter. On the other hand, given the differences in regulation of PKG-I gene expression in human (and bovine) vascular SMC compared with rat vascular SMC (25), it is also possible that unidentified transcription factors or the involvement of a post-transcriptional mechanism might contribute to PKG-I expression. However, it is also likely that sequences within introns contribute to PKG-I gene regulation because the first intron sequence downstream of exon 1␣ contains several conserved regions when comparing human and rodents sequences. In addition, using the Transfac program, many potential transcription factor binding sites were detected within this region. Furthermore, we have found that a cloned 1-kb 5Ј-intronic region of exon 1␤ (which is part of intron 1) has high activity using luciferase activity assays in vascular SMC. 2 Although USF-1 and USF-2 were initially postulated to be involved solely in regulation of so-called "housekeeping" genes, emerging data indicate that USF proteins may be associated with the regulation of tissue-specific genes, including those expressed in SMC. For example, USF proteins are necessary for the transcription of the smooth muscle-specific gene product, APEG-1 (34). APEG-1, which stands for Aortic Preferentially Expressed Gene, is expressed only in highly differentiated contractile vascular SMC, and its expression decreases when aortic SMC are cultured or in arterial SMC in response to injury in vivo. Likewise, the smooth muscle-specific genes ␣-actin and SM22␣ also contain E-boxes as specific binding sites for USF in the 5Ј-promoter regions (35). The smooth muscle myosin heavy chain gene, which is considered to be a dependable contractile phenotype marker protein, demonstrates the existence of several Eboxes in the 5Ј-promoter region, and several reports suggest a possible role for E-boxes in the control of smooth muscle myosin heavy chain gene expression (36 -38). The topology of the human PKG-I 591-bp promoter is similar to that described for the ␣-actin promoter where tandem E-boxes (CAGTTG and CAGCTG) exist, and both USF-1 and USF-2 may be involved in promoter transactivation (35). In contrast, the APEG-1 promoter was reported to be differentially regulated by USF-1 and USF-2 proteins acting as activators and suppressors, respectively (34). One similarity between the human PKG-I and the SMC-specific genes described here lies in the fact that vascular injury induces a decrease in both PKG-I expression and the expression of these SMC-specific gene products (39,40). Whether USF proteins are involved in PKG-I down-regulation in vascular injury is not known. Because USF proteins can also be induced in response to vascular injury (41,42), there may be interactions of USF proteins with transcription factors induced by injury.
In summary, this is the first study that demonstrates the functional importance of E-boxes and USF transcription factors in the regulation of the human PKG-I promoter. Although the detailed molecular mechanisms governing human PKG-I transcription remains unclear, the present study contributes significantly to our understanding of PKG expression in vascular SMC. Considering the variable distribution of PKG-I in different cells, and the ubiquitous expression of USF proteins, more studies will be needed to identify additional transcription factors involved in mediating specific tissue expression of the PKG-I.