Complex regulation of human neuronal nitric-oxide synthase exon 1c gene transcription. Essential role of Sp and ZNF family members of transcription factors.

Neuronal nitric-oxide synthase (nNOS) is expressed in a variety of human tissues and shows a complex transcriptional regulation with the presence of nine alternative first exons (1a-1i) resulting in nNOS transcripts with differing 5'-untranslated regions. We previously demonstrated that nNOS exon 1c, one of the predominant transcripts in the human gastrointestinal tract, is driven by a separate promoter (Saur, D., Paehge, H., Schusdziarra, V., and Allescher, H. D. (2000) Gastroenterology 118, 849-858). The present study focused on the quantitative expression of nNOS first exon variants in different human tissues and the characterization of the basal nNOS exon 1c promoter. In human brain, skeletal muscle, colon, and TGW-nu-I neuroblastoma cells, first exon expression patterns were analyzed by quantitative real-time reverse transcription-PCR. In these tissues/cells exon 1c was one of the most abundant first exons of nNOS. By transient transfections of TGW-nu-I and HeLa cells with reporter plasmids containing a series of 5' and 3' deletions in the exon 1c regulatory region, the minimal TATA-less promoter was localized within 44 base pairs. Gel mobility shift assays of this cis-regulatory region revealed a high complexity of the basal promoter with a cooperative binding of several transcription factors, like Sp and ZNF family members. When the Sp binding site of the minimal promoter construct was mutated, promoter activity was completely abolished in both cell lines, whereas mutation of the common binding site of ZNF76 and ZNF143 resulted in a decrease of 53% in TGW-nu-I and 37% in HeLa cells. In Drosophila Schneider cells expression of Sp1, the long Sp3 isoform, ZNF76 and ZNF143 potently transactivated the nNOS exon 1c promoter. These results identify the critical regulatory region for the nNOS exon 1c basal promoter and stress the functional importance of multiple protein complexes involving Sp and ZNF families of transcription factors in regulating nNOS exon 1c transcription.

Nitric oxide (NO), 1 a ubiquitous multifunctional mediator, is synthesized by nitric-oxide synthases (NOS) during the oxida-tion of L-arginine to L-citrulline. In the central and peripheral nervous system, skeletal muscle, the macula densa of the kidney, testis, and neutrophils, neuronal NOS (nNOS) is the predominant enzyme for the generation of nitric oxide. NO, synthesized by nNOS, acts as neurotransmitter, neuromodulator, or intracellular signaling molecule. It is involved in synaptic plasticity, regulation of gene expression, development, differentiation, and regeneration and plays an important role in neurodegenerative disorders and stroke as a mediator of neurotoxicity (for review see Refs. [1][2][3][4]. In the gastrointestinal tract NO generated by nNOS acts as an important mediator of the non-adrenergic non-cholinergic inhibitory innervation of intestinal smooth muscle (5) and as a neuromodulator within the enteric nervous system (6).
Although the transcriptional regulation of the other two NOS enzymes, the calcium-dependent endothelial NOS and the calcium-independent inducible NOS, are extensively studied (for review see Refs. 3,7,8), little is known about the transcriptional regulation of the nNOS gene (3,4,9,10), which is considered to be responsible for the largest proportion of NO in the body (1). Although usually named constitutive, recent observations suggest a tightly regulated gene expression of nNOS in response to different physiological and pathophysiological stimuli, resulting in an up-or down-regulation of nNOS mRNA (3,9).
Recently nine distinct first exons, called exons 1a-1i, of nNOS mRNA have been identified, leading to nNOS mRNA variants with different 5Ј-untranslated regions and translational efficiencies (11). The nNOS gene is therefore believed to be one of the most complex genes known in terms of first exon usage and alternative splicing (11)(12)(13). It has been shown that nNOS exons 1c (12) and 1f and 1g (13) (former called exons 1 5Ј3 , 1 5Ј2 , and 1 5Ј1 , respectively), which show high abundant expression in the human gastrointestinal tract (12), are driven by separate promoters in HeLa cells. The use of multiple alternative promoters allows a cell-, tissue-, and site-specific transcriptional regulation of nNOS in different physiological and pathophysiological stages.
An altered expression or biological activity of nNOS has been linked to several physiological conditions, like aging and pregnancy, as well as different pathophysiological conditions and diseases such as ischemia/hypoxia and injuries of the central nervous system, inherited diabetes insipidus, heart failure, arteriosclerosis, achalasia, diabetic gastroparesis, and hypertrophic pyloric stenosis (1-4, 7, 14 -17). nNOS␣ mutant mice, generated by targeted disruption of the nNOS gene by homologous recombination, showed a gastrointestinal phenotype resembling hypertrophic pyloric stenosis with delayed gastric emptying of solids and fluids (18,19). In addition, these mice have a hypertensive lower esophageal sphincter with impaired relaxation (20). A recent study comprising 27 families with inherited infantile pyloric stenosis identified nNOS as a susceptibility gene for this disorder (15), and expression of nNOS exon 1c mRNA is significantly reduced in the pyloric sphincter of patients with infantile hypertrophic pyloric stenosis. 2 Therefore it is of physiological and pathophysiological interest to investigate the molecular basis of nNOS exon 1c transcription. In addition, the distribution and quantitative expression of alternative first exon nNOS variants in different human tissues was determined.
Here we present evidence for a tissue-specific expression of nNOS first exon variants, with exon 1c being one of the pre-dominant forms in human brain, skeletal muscle, gastrointestinal tissue, and neuroblastoma cells. We furthermore characterize the basal promoter of exon 1c, showing a high complexity with a cooperative binding of several Sp and ZNF family members of transcription factors.
RNA Isolation-Liquid nitrogen-frozen rectum muscle layer preparations and fresh TGW-nu-I, ME-180, and HeLa cells were homogenized with a Polytron homogenizer (Kinematica, Switzerland), and total RNA was isolated using the guanidine isothiocyanate/phenol/chloroform extraction method as previously described (12,31). Pooled total RNA from human adult brain and skeletal muscle was purchased from CLONTECH (Heidelberg, Germany).
Quantitative Real-time RT-PCR-Real-time quantitative RT-PCR analyses for the alternative first exons 1a-1i and exon 6/7 of human nNOS mRNA were performed using an ABI Prism 7700 Sequence Detection System instrument (Applied Biosystems). cDNAs for quantitative analysis were generated with the TaqMan reverse transcription reagents (Applied Biosystems) as recommended by the manufacturer, using murine leukemia virus reverse transcriptase, random hexamer primers, and 5 g of total RNA from human rectum and from TGWnu-I, ME-180, and HeLa cells. In addition, total RNA obtained from CLONTECH and isolated from pooled human brain and skeletal muscle was used. nNOS transcripts were amplified with intron-spanning, isoform-specific primers and probes complementary to the alternative first exons 1a-1i (forward primers) and the common exon 2 (reverse primer, probe). As a parameter for total nNOS mRNA expression, an intronspanning pair of exon 6-and exon 7-specific primers present in all known human nNOS cDNAs was used with an exon 7-specific internal probe. Primers and TaqMan probes were designed to meet specific criteria by using the Primer Express software (Applied Biosystems). The 5Ј-end of the probe was labeled with a fluorescent reporter (6carboxy-fluorescein (FAM)) and the 3Ј-end with a quencher dye (6carboxyltetramethylrhodamine). Sequences of primers and fluorescent probes used in the study are shown in Table I. GAPDH, HPRT, and TF2D primers and probes were purchased from Applied Biosystems.
The principle of real-time RT-PCR has been described in detail elsewhere (32,33). Briefly, real-time PCR is based on a sequencespecific probe labeled with a 5Ј reporter and 3Ј quencher dye. When the probe is intact, reporter dye emission is quenched, but during the extension phase of the PCR, the nucleolytic activity of the Taq-DNA polymerase cleaves the hybridized probe, and due to the separation of reporter and quencher dye a fluorescence signal is released that is monitored by the sequence detector. A computer algorithm normalizes the signal to an internal reference (⌬Rn) and calculates the threshold cycle number (C t ), when ⌬Rn becomes equal to 10 standard deviations of the baseline. C t is used for quantification of the input mRNA number. For each amplicon, the amount of target and endogenous reference (glyceraldehyde-3-phosphate dehydrogenase, HPRT, TF2D) was determined from a standard curve generated by serial 5-fold dilutions (25,000 to 8 copies) of plasmids containing the respective target sequence. The standard curve was amplified in triplicate during every experiment, and the amount of target gene was normalized by the endogenous reference. Quantitative PCR was performed using the Taq-Man Universal PCR Master Mix (Applied Biosystems), with cDNAs corresponding to 100 ng of total RNA, 200 nM probe, and 900 nM primers in a 25-l final reaction mix (1 PCR cycle 50°C, 2 min; 95°C, 10 min; 50 PCR cycles 60°C, 1 min; 95°C, 15 s). Signals were analyzed by the ABI Prism Sequence Detection System software version 1.7 (Applied Biosystems).
Cloning of the 5Ј-Flanking Region of Exon 1c-The 5Ј-flanking region of human nNOS exon 1c was determined by rapid amplification of genomic ends (RAGE) using the human GenomeWalker kit (CLONTECH) as previously described (12). Digested and adapter-ligated genomic DNA fragments obtained from CLONTECH were amplified in a first round of PCR (seven cycles 94°C, 25 s; 72°C, 7 min and 32 cycles 94°C, 25 s; 67°C, 7 min adding 5 s each cycle) using the antisense gene-specific primer GSP1-AS/ex 1c (10 pmol) (for all primers see Table I) and the sense generic primer AP1 from CLONTECH specific for the ligated adapters. The second round of PCR was performed with nested antisense genespecific primer GSP2-AS/ex 1c (10 pmol) and sense generic primer AP2 (5 cycles 94°C, 25 s; 72°C, 7 min; 20 cycles 94°C, 25 s; 67°C 7 min adding 5 s each cycle). PCR products were cloned into pCRII plasmid and subjected to commercial sequence analysis (GATC). The BLASTn program was employed to search for sequence identity. Potential cis-acting DNA sequences and putative consensus elements were identified by analysis with the MatInspector professional software (Genomatix Software GmbH, Munich, Germany) using the Transfac data base.
Cell Culture, Transient Expression, and Reporter Gene Assays-HeLa cells were cultured and transiently cotransfected with the different nNOS exon 1c-pGL3 promoter gene constructs and the herpes simplex virus thymidine kinase promoter-driven Renilla luciferase expression vector pRL-TK (Promega) to normalize for transfection efficiency and cell number essentially as described previously (12). TGWnu-I cells were cultured in minimum Eagle's medium containing 10% fetal bovine serum (FBS), 25 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin sulfate (37°C/5%CO 2 ). 50 -60% confluent cells were transiently cotransfected by lipid-mediated transfer using 0.95 g of plasmid DNA of each pGL3 construct and 0.05 g of pRL-TK DNA with 6 l of Plus reagent and 4 l of LipofectAMINE (Invitrogen) per 3.5-cm dish. Cells were incubated in FBS and antibiotic-free minimum Eagle's medium containing the lipid-coated DNA complexes for 3 h and in complete medium for an additional 48 h. ME-180 cells were cultured in McCoyЈs 5A medium as described before (22). About 60%  11)). Nine distinct first exons (exon 1a-1i) driven by separate promoters and spliced to the common exon 2 of nNOS are shown. Arrows indicate the positions of sense and antisense strand primers used in quantitative real-time RT-PCR (Fig. 2). The position of the internal 6-carboxy-fluorescein (FAM)-labeled TaqMan Probe is marked above exon 2 by a line, and the translational initiation codon is noted by a vertical arrow. B, genomic organization of the alternative first exons 1a, 1b, and 1c of the nNOS gene. Exons 1a and 1b are located in the 5Ј-upstream genomic DNA region of exon 1c. The arrows below the nucleotide sequence depict the 5Ј deletion sites of the different nNOS pGL3 promoter constructs used in reporter gene assays. The nucleotide sequence reported in this study has been deposited in the EMBL/GenBank TM data base with accession number AJ308545. C, nucleotide sequence of human nNOS exon 1c (in boldface letters) and in part the 5Ј-flanking region including the minimal promoter. The transcription start site is indicated as ϩ1 and underlined. Cis-acting elements as determined by gel shift assays (Fig. 4, A-F) are indicated by lines above the sequence. Arrows below the nucleotide sequence depict the 5Ј deletion sites of the different nNOS pGL3 promoter constructs used in reporter gene assays. confluent cells were transfected with 0.38 g of each pGL3 test plasmid DNA and 0.02 g of pRL-TK DNA using 4 l of Effectene and 3.2 l of Enhancer (Qiagen) per 3.5-cm dish.
Approximately 2 ϫ 10 6 Drosophila Schneider cells (SL2 cells) were plated the day prior to transfection onto six-well plates and cultured at 25°C in DES Expression Media containing 10% FBS. After washing and addition of fresh medium without FBS, cells were transfected using 9 l of Cellfectin (Invitrogen) per well along with 1 g of each pGL3 construct, 0.5 g of the ␤-galactosidase expression plasmid p97b for normalization of transfection efficiencies, and variable amounts of the expression plasmids pPac-Sp1, pPac-Sp2, pPac-USp3, pPac-Sp3, pPac-Sp4, pPac-ZNF42, pPac-ZNF76, and pPac-ZNF143. Variations in the amount of the expression plasmids were compensated with the empty plasmid pPac0 to adjust the total DNA content of the transfection mix to 2 g per well. After 24 h of transfection, medium was removed and cells were incubated for an additional 24 h in complete medium containing FBS. SV40 promoter/enhancer-directed pGL3-control and promoter-/enhancerless pGL3-basic plasmids (both Promega) were used as positive and negative controls in all experiments, respectively. 48 h after transfection, cells were harvested by treatment with lysis buffer (Promega). Total cellular protein was determined by Bio-Rad II assay and ␤-galactosidase activity was assayed using the ␤-Galactosidase Enzyme Assay System (Promega) essentially as recommended by the manufacturer. Firefly and Renilla luciferase activities were measured in a luminometer (EG&G Berthold, Bad Wildbad, Germany) by using the Dual Luciferase Reporter Assay System (Promega) as previously described (12). All values were determined from three independent transfection experiments, each done in triplicate, and are expressed as mean values Ϯ S.D. Data of TGW-nu-I and HeLa cell transfections are presented as relative luciferase activity of firefly/Renilla luciferase. Values for SL2 cells are expressed as -fold induction of normalized firefly luciferase activity relative to that obtained following cotransfection of the pGL3 reporter plasmids with empty pPac0, which does not express Sp/ZNF proteins. In SL2 cells firefly luciferase activity was normalized in all samples for total protein content, because the transcription of the ␤-galactosidase expression vector p97b, intended as internal control, was highly inducible by cotransfection with pPac-ZNF76 and pPac-ZNF143 (but not with pPac-Sp1-4 and pPac-ZNF42).
Western Blot Analysis-Nuclear extracts (60 g of protein per lane) from transfected SL2 cells and untransfected TGW-nu-I and HeLa cells were separated by 7.5% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Bio-Rad) as previously described (12,31). Blots were probed with Sp1, Sp2, and Sp3 antibodies diluted 1:1000. Signal detection of the immunoreactive bands was facilitated by enhanced chemiluminescence using the ECL system (Amersham Biosciences, Inc.).
Data Analysis-Unless otherwise indicated, all data were determined from three independent experiments, each done in triplicate, and are expressed as mean values Ϯ S.D. Comparisons among data sets were made with analysis of variance, followed by Student's t test. Values of p Ͻ 0.05 or less were considered to be statistically significant.

Expression Pattern of nNOS First Exon mRNA Variants-
We analyzed mRNA expression of the nine alternative first exon variants of human nNOS (Fig. 1A) by real-time quantitative RT-PCR (5Ј nuclease assay) in human brain, skeletal muscle, rectum, TGW-nu-I neuroblastoma cells, ME-180 cervix carcinoma cells, and HeLa cells. In addition, as a parameter for total nNOS mRNA expression, a sequence between exon 6 and exon 7, encoding parts of the oxygenase domain of nNOS (hem binding site that is essential for NOS activity (34)), was amplified. As internal controls GAPDH, HPRT, and TF2D mRNA expression were quantified. Real-time RT-PCR of the different first exons revealed high expression of nNOS exon 1c in the investigated human tissues (Fig. 2, A-C) and the TGWnu-I cell line (Fig. 2D), whereas ME-180 cells were nNOS exon 1c-negative (Fig. 2E). The other first exons showed varying expression patterns with a cell-and tissue-specific expression (see Fig. 2, A-F). Exon 1f and exon 1g are highly abundant in the brain and rectum and very lowly abundant in skeletal muscle (Fig. 2, A-C, F; where bars are not evident despite positive RT-PCR results in Fig. 2F, values are less than the resolution shown in the figure). In contrast exon 1a is highly expressed in skeletal muscle (Fig. 2B) but missing in TGW-nu-I cells (Fig. 2D) and lowly abundant in the rectum (Fig. 2C). HeLa cells were nNOS mRNA-negative but showed high expression of the investigated housekeeping genes (data not shown). Fig. 2F summarizes the distribution of alternative first exons of nNOS in the investigated tissues and cell lines. nNOS first exon expression was normalized against GAPDH (Fig. 2, A-E) and showed no significant difference when other housekeeping genes (HPRT, TF2D) were used as internal controls (data not shown).
Cloning of the 5Ј-Flanking Region of nNOS Exon 1c-To further determine the partially known promoter sequence of nNOS exon 1c (formerly called exon 1 5Ј3 ) (12), we used rapid amplification of genomic ends (RAGE) to obtain a 5939-bp DNA fragment of the 5Ј-flanking regulatory region of exon 1c spanning nucleotides (nt) Ϫ5891 to ϩ49 (Fig. 1B). This nucleotide sequence has been deposited in the EMBL/GenBank TM data base with accession number AJ308545. The transcriptional start site of exon 1c was identified previously to be located 84 bp upstream of the exon 1c/intron 1 splice junction at an adenine (ϩ1) (Fig. 1, B and C) (12). Computer-based analysis of the sequence immediately upstream of this region using the Mat-Inspector professional software, revealed no canonical TATA or CAAT boxes as initiators of transcription, but a putative transcription initiation site with overlapping consensus sequences for Staf (common binding site for ZNF76 and ZNF143 (23)), p53 half site, Olf-1, Ap2, Myc/Max, ZNF42 (alternative title MZF-1 (24)), Sp1 and MAZ between nt Ϫ90 and Ϫ47 bp relative to the transcription start site. The obtained 5Ј-flanking region of exon 1c contains in addition the sequence of nNOS exon 1a (nt 3656 -3755) derived from EMBL accession number AF049712 (11) and exon 1b (nt 5567-5635) derived from EMBL accession number AF049713 (11) (see Fig. 1B). Exon 1c is located at nt 5892-5975 of the submitted sequence with EMBL accession number AJ308545. Therefore exons 1a, 1b, and 1c are located in close proximity within 2400 bp, similar to the genomic structure of exons 1f and 1g (former called exons 1 5Ј2 and 1 5Ј1 ) of human (13) and exons 1b and 1c of rat nNOS (35). Furthermore, the BLASTn search revealed that the putative transcription initiation site of exon 1c is highly conserved between rat nNOS Exon 1c Basal Promoter Regulation  3. Functional analysis of the human nNOS exon 1c promoter in TGW-nu-I (A), HeLa (B), and ME-180 (C) cells. 5Ј and 3Ј deletions were introduced into the nNOS exon 1c 5Ј-flanking region, and the indicated DNA fragments were ligated into the promoter-/enhancerless firefly nNOS Exon 1c Basal Promoter Regulation and human with a sequence homology of 100% between nt Ϫ84 and Ϫ41 relative to the transcription start site of human exon 1c and nt 320 and 363 of rat nNOS exon 1b (EMBL accession number AF008911) (35). Note that the nomenclature of alternative first exons of nNOS is different in rat and human with human exon 1c corresponding to rat exon 1b (11,35).
Identification of the Basal Promoter Region of nNOS Exon 1c-Previously, we showed that exon 1c (designated exon 1 5Ј3 ) of nNOS is driven by a separate promoter within 332 bp upstream of the transcription start site in HeLa cells (12). However, neither the minimal promoter, nor the major transcription factors that regulate basal transcriptional activation were elucidated. To examine the sequence that is necessary for basal transcription of nNOS exon 1c, we analyzed various 5Ј and 3Ј deletions of the 5Ј-flanking region of exon 1c by reporter gene assays. The 5.9-kb promoter fragment of exon 1c (see Fig. 1B), obtained by 5Ј-RAGE, was cloned 5Ј to the firefly luciferase reporter gene of the pGL3-basic plasmid, resulting in pGL3Ϫ5891/ϩ49. Different 5Ј and 3Ј deletions of the exon 1c promoter were generated by PCR, or restriction endonuclease and exonuclease III digestion of the pGL3Ϫ5891/ϩ49 and pGL3Ϫ332/ϩ49 plasmids (Fig. 1, B and C). These constructs were transiently transfected into nNOS exon 1c-positive TGWnu-I neuroblastoma cells (see Fig. 2, D and F) and nNOSnegative HeLa cells (see Fig. 2F), as determined by quantitative real-time RT-PCR. After transfections of TGW-nu-I and HeLa cells with pGL3Ϫ5891/ϩ49, we obtained a 13.7-and 9.1-fold increase of the normalized promoter activity compared with that of promoter-/enhancerless pGL3-basic, respectively (Fig. 3, A and B). After deletion of 3117 bp from full-length pGL3Ϫ5891/ϩ49, a decrease of 57% (p ϭ 0.008) in promoter activity could be observed in TGW-nu-I cells (Fig. 3A). In contrast, an increase of 67% (p ϭ 0.007) was evident in HeLa cells (Fig. 3B). For both cell lines, progressive deletions from Ϫ2774/ ϩ49 to Ϫ1520/ϩ49 did not show a significant change of transcriptional activity. Further deletion to Ϫ332/ϩ49 resulted in a decrease of ϳ44% (p ϭ 0.012) in TWG-nu-I cells, whereas promoter activity showed no significant change in HeLa cells. These observations suggest the presence of cis regulatory sequences exhibiting positive effects on the promoter activity between Ϫ5891 and Ϫ2774 and between Ϫ1520 and Ϫ332 specifically in TGW-nu-I cells, whereas sequences between Ϫ5891 and Ϫ2774 negatively affect promoter activity in HeLa cells. This argues for a cell-specific regulation of nNOS exon 1c promoter activity by cis acting elements in the upstream 5Јflanking region of exon 1c.
Up to 242 bp (to position Ϫ90) could be further deleted in HeLa cells without a significant change of transcriptional activity. In contrast, in TGW-nu-I cells, we observed a drop in activity of ϳ33% (not significant) between Ϫ278 and Ϫ241, an increase between Ϫ241 and Ϫ131 of ϳ72% (p Ͻ 0.01), and an additional profound rise of ϳ108% (p Ͻ 0.001) after deletion of sequences between Ϫ131 and Ϫ90. However, deletion of an additional 27 bp (to position Ϫ63) abolished promoter activity completely in both cell lines (p Ͻ 0.001). Values seen for this Ϫ63/ϩ49 construct, as well as for further deletions up to ϩ18/ ϩ49, essentially reflected those of pGL3-basic. The antisense construct of pGL3Ϫ5891/ϩ49, named pGL3 ϩ 49/Ϫ5891, showed only background activity. As positive control the SV40 promoter/enhancer-directed pGL3-control vector was used in TGW-nu-I (Fig. 3A) and HeLa cells (data not shown).
These data demonstrate that the basal promoter for exon 1c is disrupted by deletion to Ϫ63 relative to the exon 1c transcription start site, and therefore, they strongly suggest that the promoter resides within the predicted region between Ϫ90 and Ϫ47. Interestingly, adjacent upstream 5Ј regions regulate this basal promoter activity differentially in nNOS-positive TGW-nu-I cells and nNOS-negative HeLa cells, demonstrating distinct mechanisms of activation in these two cell lines.
When 350 bp was removed from the 3Ј-end of pGL3Ϫ5891/ ϩ49, a Ϫ5891/Ϫ279 construct resulted that lacks the minimal promoter and the transcription start site of exon 1c but contains the putative promoters and transcription start sites of exon 1a and exon 1b. After transient transfection of this construct, normalized promoter activity showed just background activity in exon 1a and exon 1b mRNA-negative TGW-nu-I cells and in nNOS mRNA-negative HeLa cells (Fig. 3, A and B). These results demonstrate that the upstream 5Ј-flanking region of exon 1c, containing exons 1a and 1b and their putative promoters, can influence transcription of exon 1c but does not activate transcription of exon 1a and/or exon 1b in both cell lines under the investigated conditions.
Because exon 1c promoter constructs are active in endogenous nNOS-negative HeLa cells, we further investigated this discrepancy by reporter gene assays using nNOS-positive but exon 1c-negative ME-180 cells. After transient transfection luciferase expression vector pGL3-basic. As positive control an SV40 promoter/enhancer-directed firefly luciferase control vector (pGL3-control) was used (data not shown for HeLa and ME-180 cells in B and C, respectively). pGL3-basic and an antisense construct of pGL3Ϫ5891/ϩ49 named pGL3 ϩ 49/Ϫ5891 were used as a negative control. Constructs were transiently cotransfected into TGW-nu-I nNOS Exon 1c Basal Promoter Regulation with different nNOS exon 1c pGL3 promoter constructs, we observed a moderate, but significant increase in relative luciferase activity over that of promoter-/enhancerless pGL3-basic (Fig. 3C). Therefore, endogenous nNOS-negative HeLa cells and nNOS exon 1c-negative ME-180 cells are able to transactivate nNOS exon 1c reporter constructs.

Identification of Transcription Factors Binding to the Basal nNOS Exon 1c
Promoter-Computer-based sequence inspection of the GC-rich (66%) and TATA-less basal promoter region between Ϫ90 and Ϫ47 of exon 1c indicated a variety of putative cis-regulatory elements, like Staf (consensus binding site for ZNF143 and ZNF 76 (23)) (Ϫ85 to Ϫ64), p53 half-site (Ϫ79 to Ϫ70), Olf-1 (Ϫ78 to Ϫ57), Ap2 (Ϫ73 to Ϫ62), Myc/Max (Ϫ61 to Ϫ48), ZNF42 (alternative title MZF-1) (Ϫ70 to Ϫ63), and a low affinity GC box with binding sites for Sp1 (Ϫ69 to Ϫ57) and MAZ (Ϫ67 to Ϫ59). To assess transcription factor binding, a double-stranded 44-bp 32 P-labeled oligonucleotide (Ϫ90/Ϫ47) (see Table II for all oligonucleotides used in gel shift experiments, except consensus and mutant consensus oligonucleotides for Ap2, Myc-Max, NF-I, YY1, NFB, p53, Sp1, USF-1, which were purchased from Santa Cruz Biotechnologies), including these elements, was used in electrophoretic mobility shift assays (EMSAs). A series of five shifted protein-DNA complexes were observed after incubation of TGW-nu-I nuclear extracts with the labeled probe (Fig. 4A, lane 2), whereas only two protein-DNA complexes were evident using HeLa nuclear extracts (Fig. 4B, lane 2). These complexes were competed with a 100-fold molar excess of unlabeled Ϫ90/Ϫ47 probe, establishing binding specificity (Fig. 4, A and B, lane 3). Shifted bands without competition after addition of the unlabeled Ϫ90/Ϫ47 probe were considered as nonspecific binding and were marked by an asterisk in Fig. 4 (A-D). Using a 100-fold molar excess of an unlabelled Sp1 consensus oligonucleotide (containing the binding site for the transcription factors Sp1, Sp3, and Sp4) complexes I/III (TGW-nu-I cells, Fig. 4A, lane 7) and complex I (HeLa cells, Fig. 4B, lane 7) were completely competed, and complex II (both cell lines) was partially competed. There was no further competition of complex II, when a 200-and 500-fold molar excess of unlabeled Sp1 consensus oligonucleotides was used (both cell lines, data not shown). No competition of the complexes was observed with a 100-and 200-fold molar excess of a commercial mutant Sp1 oligonucleotide (Fig. 4, A and B,  lane 8). Using TGW-nu-I nuclear extracts a slight reduction in the protein-DNA complexes I, II, and III was seen with a Ϫ90/Ϫ63 oligonucleotide where the GC box is disrupted (Fig.  4A, lane 4) and with a Ϫ90/Ϫ47 oligonucleotide in which the Sp1 binding site was mutated (Fig. 4A, lane 5), whereas complexes IV and V were completely competed by these two oligonucleotides. With HeLa cell nuclear extracts complex I was slightly and complex II was completely competed using the Ϫ90/Ϫ63 and Ϫ90/Ϫ47 Sp1-Mut competitors (Fig. 4B, lanes 4  and 5). Furthermore, no competition of any complex was ob-served using an unrelated YY1 consensus oligonucleotide (TGW-nu-I cells, Fig. 4A, lane 23; HeLa cells, Fig. 4B, lane 21). When protein-DNA complexes were incubated with antibodies against Sp1, Sp2, Sp3, and Sp4 or combinations of these antibodies, supershifts of different complexes were observed with TGW-nu-I and HeLa cell nuclear extracts (Fig. 4, C and D). For the Sp1 antibody a supershift of complex I/III (TGW-nu-I cells) and complex I (HeLa cells), for the Sp2 antibody a supershift of complex I (TGW-nu-I and HeLa cells), for the Sp3 antibody a supershift of complex I/II/III (TGW-nu-I cells) and complex II (HeLa cells) was observed whereas for the Sp4 antibody no supershift was seen. A combination of Sp1, Sp2, and Sp3 antibodies led to different supershifts as shown in Fig. 4C for TGW-nu-I cells and Fig. 4D for HeLa cells. When all three antibodies were added to the binding reaction, complexes I/II/ III (TGW-nu-I cells) and complexes I/II (HeLa cells) were supershifted with a nearly complete abrogation. Using antibodies against goat and rabbit IgG, no supershift of any complex was evident (data not shown). Taken together these results demonstrate that Sp1, Sp2, and Sp3 can interact with the 44-bp minimal promoter of exon 1c. To further characterize the retarded complex II obtained with TGW-nu-I and HeLa cells, which was not completely competed by an Sp1 consensus oligonucleotide, and the unaffected complexes IV and V (TGWnu-I cells), consensus oligonucleotides for the transcription factors Staf (ZNF76 and ZNF143), p53, Olf-1, Ap2, Myc/Max, ZNF 42, MAZ, NF-〉, NF-1, YY1, and USF were used. Myc/Max, NF-1, USF, p53 (data not shown), and YY1 (Fig. 4A, lane 23;  Fig. 4A,  lanes 11 and 12) was evident using TGW-nu-I cell nuclear extracts and a 100-fold excess of unlabelled consensus and mutated consensus oligonucleotides for MAZ and Olf-1. In contrast, oligonucleotides containing Staf, ZNF42, Ap2, and NF-〉 binding sites were specific in competition studies. An oligonucleotide with a Staf binding site (consensus sequence for ZNF76 and ZNF143) competed partially complex II (both cell lines, Fig. 4, A and B, lane 9), whereas ZNF42 consensus oligonucleotides competed complexes I/III/IV/V (TGW-nu-I cells, Fig. 4A, lane 15) and complex I (HeLa cells, Fig. 4B, lane 15). Addition of oligonucleotides containing mutations in the Staf and ZNF42 binding sites failed to compete protein-DNA complexes (Fig. 4,  A and B, lane 16). A combination of Sp1 and Staf consensus oligonucleotides resulted in a complete competition of complex II (both cell lines, Fig. 4, A and B, lane 13), whereas Sp1 combined with other consensus oligonucleotides like Olf-1 (Fig.  4, A and B, In lanes 3, 4, and 5 a 100-fold molar excess of unlabeled Ϫ90/Ϫ47, Ϫ90/Ϫ63, and Ϫ90/Ϫ47 Sp1 Mut oligonucleotides were used as competitors. To characterize shifted protein-DNA complexes, unlabeled consensus and mutated consensus oligonucleotides for the indicated transcription factors were used, as well as combinations of these oligonucleotides (panel A: lanes 7-16, 20 -23; panel B: lanes 7-21). Lane 1 represents labeled probe alone. C and D, detection of protein-DNA complexes using antibodies against Sp1, Sp2, Sp3, Sp4, and combinations of these antibodies. EMSAs were done as described above using TGW-nu-I (C) and HeLa (D) cell nuclear extracts and the 32 P-labeled Ϫ90/Ϫ47 spanning probe of the basal nNOS exon 1c promoter. Supershift assays were performed by incubation of the binding reaction with the labeled probe for 30 min followed by the addition of 2 l of each antibody and an additional incubation for 30 min at room temperature. Nonspecific bands are indicated by an asterisk. E and F, EMSAs of TGW-nu-I (E) and HeLa (F) cell nuclear extracts binding to the Ϫ90/Ϫ63 region of the nNOS exon 1c promoter. Gel shift assays were performed as described above (A and B), except that a 32 P-labeled oligonucleotide probe spanning the Ϫ90 to Ϫ63 region of the minimal exon 1c promoter was used. Lane 1 represents probe alone, and 10 g of the respective nuclear extracts was added to the binding reaction in lane 2, resulting in four (panel E) and six (panel F) specific shifted protein-DNA complexes, respectively. A 100-fold molar excess of various unlabeled oligonucleotides were added as competitors in lanes 3-10 as indicated.

FIG. 5. Mutagenesis of transcription factor binding sites decreases transcription from the nNOS exon 1c minimal promoter.
TGW-nu-I (A, C, E) and HeLa (B, D, F) cells were transiently cotransfected with pRL-TK and different wild type or mutant reporter gene constructs containing targeted substitutions or deletions in transcription factor binding sites as described under "Experimental Procedures." To adjust for transfection efficiency, the firefly luciferase activity of the test plasmids was corrected for Renilla luciferase activity (pRL-TK). Data are expressed as percent luciferase activity relative to the respective "full-length" wild type pGL3 construct and represent the means Ϯ S.D. of three independent experiments in triplicate. A and B, the percent luciferase activities of wild type pGL3Ϫ90/ϩ49, 5Ј-deleted pGL3Ϫ83/ϩ49, and pGL3Ϫ48/ϩ49, the promoter-/enhancerless pGL3-basic vector, or mutant pGL3Ϫ90/ϩ49 reporter gene constructs containing the indicated targeted substitutions in the binding site of Sp1 (pGL3Ϫ90/ϩ49 Sp1-M2), the common binding sites of Sp1 and ZNF42 (pGL3Ϫ90/ϩ49 Sp1/ZNF42-M1), Ap2 and Olf1 (pGL3Ϫ90/ϩ49 Ap2/Olf1-M), ZNF76 and ZNF143 (called Staf binding site; pGL3Ϫ90/ϩ49 Staf-M) are plotted for TGW-nu-I (A) and HeLa (B) cells. C and D, the percent luciferase activities of wild type pGL3Ϫ5891/ϩ49, 5Ј-deleted pGL3Ϫ48/ϩ49, pGL3-basic vector, and the indicated mutant pGL3Ϫ5891/ϩ49 constructs containing targeted substitutions or deletions in the Sp1/ZNF42, Staf, and both the Sp1/ZNF42 and Staf binding element are plotted for TGW-nu-I (C) and HeLa (D) cells. E and F, the percent luciferase activities of wild type pGL3Ϫ332/ϩ49, 5Ј-deleted pGL3Ϫ48/ϩ49, pGL3-basic vector, and the indicated mutant pGL3Ϫ332/ϩ49 constructs containing targeted substitutions or deletions in the Sp1/ZNF42, Staf, and both the Sp1/ZNF42 and Staf binding element are plotted for TGW-nu-I (E) and HeLa (F) cells.
clear reduction in the protein-DNA complexes I/II/III (TGWnu-I cells, Fig. 4A, lanes 21 and 22) and complex I (HeLa cells, Fig. 4B, lanes 17 and 20), whereas mutated oligonucleotides failed to compete (Fig. 4B, lane 18 and data not shown). However, using Ap2 and NF-B (subunits p50 and p65) antibodies, we were unable to observe a supershift or abrogation of any complex (data not shown). The causal mechanism of this observation is unclear. Ap2 and NF-〉 may not be able to bind to the minimal promoter of exon 1c autonomously and thus do not play a direct role in exon 1c basal promoter activation. Therefore, competition of the retarded bands in the gel shift assays could be due to protein-protein interactions of Ap2/NF-B with other nuclear factors whose DNA binding affinity and specific-ity could be increased by the presence of Ap2 or NF-B. Such mechanisms have been demonstrated recently (36,37), and therefore Ap2 and NF-B could participate in the formation of multiple protein complexes that enhance or repress the transactivation potential of other transcription factors like Sp1.
To determine transcription factor binding in the absence of Sp and GC box activity, a 22-bp 32 P-labeled Ϫ90/Ϫ63 oligonucleotide with a disrupted GC box and Sp1 binding site was used in EMSAs. A series of four and six shifted protein-DNA complexes was observed after incubation with TGW-nu-I (Fig. 4E,  lane 2) and HeLa (Fig. 4F, lane 2) cell nuclear extracts, respectively. These complexes were completely competed with a 100fold molar excess of unlabeled Ϫ90/Ϫ63 probe (Fig. 4, E and F,   FIG. 5-continued nNOS Exon 1c Basal Promoter Regulation lane 3). Using a 100-fold molar excess of an unlabelled Sp1 consensus oligonucleotide, the signal intensity of the protein-DNA complexes II and III were amplified, and the signal intensity of complex IV was reduced with TGW-nu-I cell nuclear extracts (Fig. 4E, lane 4). In contrast, complete competition of complex I was observed using HeLa cell nuclear extracts (Fig.  4F, lane 4). When a ZNF42 consensus oligonucleotide was used, complex IV was completely competed and complexes II and III were shifted or competed resulting in a retarded band between complex II and III in TGW-nu-I cells (Fig. 4E, lane 5), whereas no competition was observed with a mutated ZNF42 consensus oligonucleotide (Fig. 4E, lane 6). The significance of the described observation remains unclear but demonstrates that ZNF42 is able to modulate transcription factor binding to the Ϫ90/Ϫ63 sequence. In HeLa cells ZNF42 consensus oligonucleotides resulted in only an unspecific competition of complex III (Fig. 4F, lanes 5 and 6). A Staf consensus oligonucleotide competed complex I/II/III (TGW-nu-I cells, Fig. 4E, lane 9) and complex II/III/IV/V (HeLa cells, Fig. 4F, lane 9). However, complex VI of Fig. 4F (HeLa cells) could not specifically be competed by any of the used consensus oligonucleotides, indicating that an as yet unknown transcription factor present in HeLa cells is able to bind to the Ϫ90 to Ϫ63 sequence of the exon 1c promoter. No competition was observed with mutant Staf (Fig. 4, E and F, lane 10), mutant Sp1 (data not shown), and unrelated YY1 (Fig. 4E, lane 7, Fig. 4F, lane 8) consensus oligonucleotides using nuclear extracts from both cell lines.
Collectively, these results identify a multiplicity of protein-DNA complexes within the nNOS exon 1c basal promoter involving Sp1, Sp2, Sp3, and members of the ZNF family of transcription factors. These factors have differential effects on nucleoprotein complex formation. Some of them increase, whereas others decrease, band intensities, suggesting modulatory effects on protein-DNA interactions.

Confirmation of Transcription Elements and Promoter Transactivation by Mutagenesis and Transient
Transfections-To further clarify the role of the different cis regulatory elements in the basal nNOS exon 1c promoter, reporter gene constructs for pGL3Ϫ90/ϩ49 containing point mutations in the Sp1/ZNF42 (pGL3Ϫ90/ϩ49 Sp1/ZNF42-M1), Sp1 (pGL3Ϫ90/ ϩ49 Sp1-M2), Ap2/Olf-1 (pGL3Ϫ90/ϩ49 Ap2/Olf-1-M), and FIG. 6. Sp1, the long isoform of Sp3, ZNF76, and ZNF143 transactivate the human nNOS exon 1c promoter in Drosophila Schneider cells. SL2 cells were cotransfected with the indicated nNOS exon 1c-promoter/pGL3-luciferase reporter gene constructs and different combinations and amounts of pPac-Sp1, pPac-Sp2, pPac-Sp3, pPac-USp3, pPac-Sp4, pPac-ZNF76, pPac-ZNF143, pPac-ZNF42, or empty pPac0 expression vector as described under "Experimental Procedures." Luciferase activity was assayed 48 h after transfection and normalized for total cellular protein values of lysed SL2 cells. Data are expressed as -fold induction of normalized luciferase activity relative to that obtained following cotransfection of the pGL3 reporter plasmids with empty pPac0, which does not express Sp/ZNF proteins. Values represent means Ϯ S.D. of three independent experiments in triplicate. Where bars and/or error bars are not evident, values are less than the resolution of the figure. Variations in the amount of the expression plasmids were compensated with the empty plasmid pPac0 to keep the total DNA content of pPac plasmids constant (0.5 g per transfection). A, Drosophila Schneider cells were cotransfected with 1.0 g of pGL3Ϫ90/ϩ49, 5Ј-deleted pGL3Ϫ48/ϩ49, and the promoter-/enhancerless pGL3-basic luciferase reporter plasmid along with 0.5 g of the expression plasmids pPac-Sp1, pPac-USp3, pPac-ZNF76, pPac-ZNF143, and pPac-ZNF42. B, cotransfection of 1.0 g of pGL3Ϫ90/ϩ49 along with increasing concentrations (0.025-0.5 g) of the expression plasmids pPac-Sp1, pPac-USp3, pPac-ZNF76, pPac-ZNF143, and pPac-ZNF42 into SL2 cells reveals a dose-dependent transactivation of the minimal nNOS exon 1c promoter for Sp1, the long isoform of Sp3, ZNF76, and ZNF143. C, specific Sp and ZNF family members of transcription factors exert a differential activation of the nNOS exon 1c promoter. Combinations of the expression plasmids pPac-Sp1, pPac-Sp2, pPac-Sp3, pPac-USp3, pPac-Sp4, pPac-ZNF76, pPac-ZNF143, pPac-ZNF42, and empty pPac0 were cotransfected along with 1.0 g of wild type pGL3Ϫ90/ϩ49 into SL2 cells. 0.25 g of DNA of each indicated pPac expression plasmid was used for transfections. Staf (pGL3Ϫ90/ϩ49 Staf-M) binding sites were constructed (see Fig. 5, A and B). These constructs were transiently expressed in TGW-nu-I (Fig. 5A) and HeLa cells (Fig. 5B), and promoter activities were compared with that of the wild type plasmid pGL3Ϫ90/ϩ49. Furthermore, 5Ј-deleted pGL3Ϫ83/ ϩ49, pGL3Ϫ48/ϩ49, and promoter-/enhancerless pGL3-basic were used as controls. Mutation of the common Sp1/ZNF42 binding site, as well as the Sp1 binding site, alone abolished promoter activity of nNOS exon 1c to that of pGL3-basic in both cell lines (Fig. 5, A and B). In TGW-nu-I cells, the pGL3Ϫ90/ϩ49 promoter construct containing a mutated Staf element resulted in a 53% decrease in luciferase activity compared with the pGL3Ϫ90/ϩ49 wild type construct (Fig. 5A), whereas a decrease of 37% was detected in HeLa cells (Fig. 5B). In TGW-nu-I and HeLa cells, constructs of pGL3Ϫ90/ϩ49 with mutations in the Ap2/Olf-1 binding site were 94% and 98.5% as active as wild type pGL3Ϫ90/ϩ49, respectively, indicating that the Ap2 and Olf-1 consensus binding sites in the exon 1c minimal promoter play no role for basal transcription (Fig. 5, A and B). Taken together, these data clearly demonstrate the critical cis-acting roles for Sp and ZNF binding motifs in the exon 1c minimal promoter.
To investigate the importance of this basal promoter region for the transactivation of the longer reporter constructs pGL3Ϫ5891/ϩ49 and pGL3Ϫ332/ϩ49, site-specific mutations and deletions in the Sp and Staf binding sites were introduced into the wild type plasmids. After transient expression in TGW-nu-I cells, promoter activities of all mutated/deleted pGL3Ϫ5891/ϩ49 plasmids reflected that of pGL3-basic (Fig.  5C). In HeLa cells, the construct pGL3Ϫ5891/ϩ49 Staf/Sp1/ ZNF42-M1, which contains a double mutation of the Staf and Sp1/ZNF42 binding sites, abolished promoter activity completely, and the construct pGL3Ϫ5891/ϩ49 Staf/Sp1/ZNF42del, with a deletion of both sites, was 11% as active as wild type pGL3Ϫ5891/ϩ49 (Fig. 5D). Plasmids containing a mutation (pGL3Ϫ5891/ϩ49 Sp1/ZNF42-M1) or deletion (pGL3Ϫ5891/ ϩ49 Sp1/ZNF42-del) of the Sp1/ZNF42 site displayed a drop of 87 and 79% in functional promoter activity relative to the wild type construct, respectively, and mutation (pGL3Ϫ5891/ϩ49 Staf-M) and deletion (pGL3Ϫ5891/ϩ49 Staf-del) of the Staf element resulted in a decrease of 68 and 71%, respectively (Fig.  5D). The same mutations and deletions were incorporated in to the pGL3Ϫ332/ϩ49 construct. After transient expression in TGW-nu-I and HeLa cells, relative functional reporter activities revealed no important differences compared with that of the mutations/deletions in the setting of the full-length Ϫ5891/ ϩ49 construct (Fig. 5, E and F).
Collectively, these data clearly indicate the major importance of the GC-rich region between Ϫ90 and Ϫ47 relative to the exon 1c transcription start site to promote nNOS exon 1c expression.
Expression of Sp1, Sp2, isoforms of Sp3, and Sp4 in SL2 cells after transient transfection with the different pPac plasmids was verified by Western blot analysis of SL2 cell nuclear extracts using selective antibodies against Sp1, Sp2, Sp3, and Sp4 (data not shown). Where no antibodies were available (ZNF42, ZNF76, and ZNF143), expression in SL2 cells after transient transfection was confirmed using EMSAs. Binding of expressed ZNF42, ZNF76, and ZNF143, as well as Sp1 and the short and long isoforms of Sp3 to the minimal exon 1c promoter (Ϫ90/Ϫ47) and the respective consensus oligonucleotides (ZNF42, Staf, Sp1) was seen in gel shift assays using nuclear extracts from SL2 cells transiently transfected with the different expression plasmids (pPac-Sp1, pPac-USp3, pPac-Sp3, pPac-ZNF42, pPac-ZNF76, pPac-ZNF143) (data not shown). In contrast to Sp1 and Sp3 isoforms, expressed Sp2 and Sp4 proteins showed no binding to the exon 1c minimal promoter. In addition, there was no binding of expressed Sp2 to the Sp1 consensus oligonucleotide as previously described (26,27,30,39) (data not shown). Control experiments showed that Sp1, Sp2, and Sp3 proteins and ZNF42, ZNF76, and ZNF143 mRNAs are expressed in TGW-nu-I and HeLa cells, enabling the regulation of nNOS exon 1c promoter activity in mammalian cells (data not shown). In contrast to Sp1, Sp2, Sp3, ZNF76, and ZNF143, mRNA for ZNF42 was not detectable in human brain, skeletal muscle, and rectum by RT-PCR, indicating that ZNF42 plays no role in the transcriptional regulation of nNOS exon 1c in these tissues under the investigated conditions. This is in agreement with previous findings, demonstrating that ZNF42 is a specific transcriptional regulator of myeloid differentiation (24).
In summary, these results identify Sp1, full-length Sp3, ZNF76, and ZNF143 as potent transcriptional activators of the nNOS exon 1c promoter, whereas the short isoforms of Sp3 and ZNF42 exhibit a specific repressive effect on Sp/ZNF-mediated transcriptional activation. DISCUSSION A variety of human nNOS mRNA variants have been described recently (3,4,(11)(12)(13)40). Among these, transcripts with different untranslated first exons are generated by alternative promoter usage (12,13). Real-time quantitative RT-PCR was used to determine the expression patterns and the quantitative distribution of nine alternative first exons of nNOS (exons 1a-1i) (11) and revealed a cell-and tissue-specific expression with exon 1c being one of the predominant variants in human brain, skeletal muscle, rectum, and TGW-nu-I neuroblastoma cells. Because exon 1c is highly expressed in human brain and skeletal muscle, it seems responsible for the largest proportion of nNOS mRNA in the body. Furthermore, nNOS exon 1c mRNA expression is significantly reduced in the pyloric sphincter of patients with infantile hypertrophic pyloric stenosis. 2 Therefore the transcriptional regulation of this variant is of special interest.
To characterize the structure and the expressional regulation of nNOS exon 1c, we cloned its genomic 5Ј-flanking region and analyzed the basal promoter. A 5939-bp genomic 5Ј-flanking DNA fragment of exon 1c, which contains in addition nNOS exon 1a and exon 1b, was obtained by 5Ј-RAGE PCR. By 5Ј and 3Ј deletion analysis of reporter plasmids, using nNOS exon 1c mRNA-positive TGW-nu-I, nNOS-negative HeLa and nNOS mRNA-positive, but exon 1c-negative ME-180 cells, the minimal promoter was localized to position Ϫ90 to Ϫ47, relative to the transcription start site of exon 1c. This region is highly conserved between different species with a sequence homology of 100% between rat (nt 320 to 363 of EMBL accession number AF008911) and man (nt Ϫ84 to Ϫ41 of exon 1c). Such a high degree of conservation argues for an important, biologically conserved function of this transcriptional control region (41).
Interestingly, nNOS exon 1c reporter plasmids were active in HeLa and ME-180 cells that lack endogenous nNOS exon 1c mRNA. This indicates that cell-specific expression of nNOS exon 1c is mediated by distinct mechanisms that cannot influence transcription of the investigated reporter gene constructs. Such cell-specific transcriptional control mechanisms can be due to repressive cis-acting elements that inhibit transcription in distinct cell types. They are often localized within the upstream 5Ј-flanking region or within the first intron of a gene, as shown for the growth-associated protein 43 (42). Therefore additional sequences in the far upstream region or in the first intron may mediate cell-specific expression of nNOS exon 1c. In addition the chromatin structure and acetylation state that predicts DNA sequence accessibility plays an important role in cell-specific gene regulation (for review see Ref. 43). Therefore condensation and thus silencing of the nNOS gene could also be responsible for the lack of endogenous nNOS expression in HeLa cells.
To detect transcription factors binding to the basal nNOS exon 1c promoter, EMSAs, including competition and supershift analysis were performed. They identified Sp and ZNF family members of transcription factors as the critical factors regulating the exon 1c Ϫ90/Ϫ47 basal promoter region. For example, Sp1 and isoforms of Sp3 are binding to the low affinity GC box (GGGAGGGG). Although this motif contains an A in place of the consensus C, it has been demonstrated to bind transcription factors of the Sp family, which activate or repress transcription substantially (30,41,44). Because the minimal promoter of nNOS exon 1c is GC-rich, TATA-less, and Spregulated, it resembles those for constitutively expressed genes, like dihydrofolate reductase, endothelial NOS, or the serotonin 1a receptor (41,(45)(46)(47). The regulation of such promoters is poorly understood. It has been shown that G/C-rich promoter regions, lacking a canonical TATA box, can bind Sp1 molecules that interact with multiple components of the transcriptional machinery (for review see Ref. 48), and therefore Sp1 plays a critical role in the assembly of the transcription initiation complex (45). A number of transcription factors has been documented as acting in combination with Sp1 or promoting its displacement from the same or an overlapping site (28,30,37,49). Among these, Sp2, Sp3, and Sp4 belong to the same transcription factor family (27,28,30). They share a highly homologous DNA binding domain, but only Sp1, Sp3, and Sp4 recognize the classic Sp1 GC-box in vitro with similar affinities (27,30,39). Sp3 mRNA encodes for different isoforms, arising from alternative translation initiation sites (50). These isoforms of Sp3 have been found to exert both activating and inhibiting effects on gene transcription (26, 28, 30, 49 -53). In contrast, Sp1 is typically an activator of transcription (30,48). Because Sp1, Sp2, and isoforms of Sp3 are ubiquitous expressed (28), this could explain why the investigated reporter plasmids under the control of the nNOS exon 1c minimal promoter are expressed in nNOS exon 1c-negative HeLa and ME-180 cells. When the effects of Sp transcription factors on the exon 1c promoter were examined in Drosophila Schneider cells that lack Sp, ZNF76, and ZNF143 binding activity (23,25,26,38), Sp1 and the long isoform of Sp3 potently stimulated transcription, whereas Sp2 and the short isoforms of Sp3 and Sp4 were transcriptionally inactive on their own. Expressed Sp1 and the short and long isoforms of Sp3 were able to bind to the exon 1c minimal promoter region in gel shift assays, whereas the Sp2 and Sp4 proteins did not form protein-DNA complexes. However, an Sp2-specific antibody resulted in a supershift of complex I in EMSAs using TGW-nu-I and HeLa cell nuclear extracts. This suggests that Sp2 is not able to bind to the minimal promoter of exon 1c autonomously. The supershift of the retarded band I in the gel shift assays could be due to protein-protein interactions of Sp2 with other nuclear factors, not present in SL2 cells, resulting in an increase in DNA binding affinity and specificity of Sp2 (52). Such mechanisms have been described for a variety of transcription factors like GATA members and the p65 subunit of NF-B (37,54). In cotransfection experiments the transactivating effects of expressed Sp1 and full-length Sp3 were completely additive, whereas Sp2 selectively repressed Sp1-mediated transcriptional activation of the nNOS exon 1c basal promoter. This was a moderate effect with a decrease from a 8.4-to a 6.7-fold induction of luciferase activity. However, this observation demonstrates a specific and differential modulation of promoter transactivation by different members of the Sp family of transcription factors. A similar effect was observed for the combination of short and long isoforms of Sp3, where the short isoform selectively repressed transactivation of the exon 1c basal promoter by full-length Sp3 (decrease from a 13.9-to a 7.5-fold induction). The mode of action could be due to formation or disruption of protein-protein (self-)interactions and multimerization that increase or decrease the DNA binding affinity of the respective transcription factors or to competition for common DNA recognition sites (30,37,49,50,52,54). In contrast to Sp2 and the short isoforms of Sp3, Sp4 did not alter promoter transactivation by Sp1 and the long isoform of Sp3. In addition to Sp transcription factors, we demonstrated that exon 1c promoter activity is regulated by different members of the ZNF family of transcription factors. ZNF42, ZNF76, and ZNF143 specifically bind to the minimal exon 1c promoter in gel shift experiments, and cotransfection of ZNF76 or ZNF143 resulted in a strong induction of the exon 1c promoter in Drosophila Schneider cells, whereas ZNF42 by itself was transcriptionally inert. Combining ZNF and Sp transcription factors in SL2 cells resulted in differential effects on nNOS exon 1c promoter activity. ZNF76 potentiated the stimulatory effects of Sp1 and the long isoform of Sp3, whereas ZNF143 combined with Sp1 showed an additive effect. In contrast, no effect was seen when ZNF143 was coexpressed with the long isoform of Sp3. ZNF42 significantly repressed exon 1c promoter transactivation of ZNF76, ZNF143, and Sp1. These findings demonstrate a differential regulation of the nNOS exon 1c promoter by Sp and ZNF family members of transcription factors and suggest highly coordinated protein-protein interactions and protein-DNA binding to overlapping cis-acting elements. This observation was confirmed by the results of gel shift competition studies. Complete competition of the retarded complex II obtained with TGW-nu-I cell nuclear extracts could be demonstrated only by using a combination of a 100-fold molar excess of unlabelled Sp1 and Staf (ZNF76 and ZNF143 binding site) oligonucleotides, whereas Sp1 or Staf alone were not able to compete this band completely.
The involvement of members of the Sp and ZNF families of transcription factors on transactivation of the basal human nNOS exon 1c promoter was verified by mutating the consensus sequences and transfecting different reporter constructs into HeLa and TGW-nu-I cells. The mutated GC box completely abolished promoter activity of all tested exon 1c reporter plasmids in TGW-nu-I cells and reduced promoter activity by ϳ78% or more in HeLa cells. Mutation of the Staf (ZNF76 and ZNF143) binding site decreased relative luciferase activity of the minimal promoter construct (pGL3Ϫ90/ϩ49) by ϳ53% in TGW-nu-I and ϳ37% in HeLa cells. This effect was even more pronounced in the longer promoter constructs pGL3Ϫ332/ϩ49 and pGL3Ϫ5891/ϩ49 with a drop of ϳ76 and ϳ90%, respectively, in TGW-nu-I cells and ϳ68 and ϳ60%, respectively, in HeLa cells. Collectively, these data stress the essential role for Sp and Staf binding motifs for the initiation of transcription from the TATA less human nNOS exon 1c promoter.
Little is known about the role of the ZNF transcription factor family members ZNF76 (55) and ZNF143 (56) in transactivation of mRNA promoters. Both transcription factors resemble the Drosophila Kruppel segmentation gene product due to the presence of repeated Cys2-His2 zinc finger domains that are connected by conserved sequences. They contain seven tandemly repeated zinc fingers, showing high homology to the Xenopus laevis transcriptional activator Staf (23), originally identified in Xenopus as the transactivator of the tRNA Sec gene (38). In addition Xenopus Staf possesses the capacity to stimulate expression from TATA-box containing RNA polymerase II mRNA promoters (23,57). ZNF76 and 143 are highly identical and functionally equivalent to the X. laevis Staf and, therefore, believed to be the human homologues (23). Both transcription factors are expressed in a broad range of tissues (23), including human brain, skeletal muscle, rectum, TGW-nu-I, and HeLa cells, and are able to bind tightly to Staf consensus-responsive elements present in the X. laevis tRNA Sec and the human U6 promoter with identical affinities (23). When expressed in Dro-sophila Schneider cells, lacking ZNF76 and ZNF143 activity (23,38), both factors activated transcription from the TATAbox containing thymidine kinase mRNA promoter through the Staf binding site (23). Here we demonstrate for the first time that the human homologues of Staf can stimulate transcription from a TATA-less promoter in Drosophila Schneider cells on their own. In addition, we demonstrate a synergistic activation of transcription by combinations of Sp1, full-length Sp3, ZNF76, and ZNF143, whereas ZNF42, Sp2, and the short isoforms of Sp3 repressed transcriptional activation in certain cases. It has been suggested that the cell content of distinct transcription factors is a critical factor influencing the transcription of TATA-less promoters (49). Therefore, differences in the cellular levels of Sp1, Sp2, Sp3 isoforms, ZNF42, ZNF76, and ZNF143 in different cell and tissue types or changes of the cellular levels in response to physiological and pathological conditions (30,51,58) could conceivably exert profound effects on the expression of nNOS exon 1c. This adds another level of complexity and cell specificity to the already intricate regulation of nNOS gene expression resulting from the use of multiple promoters (3,4,10,12,13,40).
In conclusion, our results demonstrate that nNOS exon 1c is the predominant nNOS mRNA variant in the body. The minimal exon 1c promoter could be localized to 44 bp, with a cooperative binding of several transcription factors of the Sp and ZNF families. This portends many possibilities for tissue-and cell-specific control in response to cellular requirements. Recent findings support a complex and tightly regulated gene expression of nNOS, showing profound changes in different physiological and pathophysiological states (1-4, 14, 16, 17, 21, 35). However, the causal mechanisms are unknown, and therefore the results of this study could provide insight into the roles of changes of nNOS gene transcription, including possible therapeutic strategies for increasing nNOS gene expression.