Nuclear Factor 1 Interferes with Sp1 Binding through a Composite Element on the Rat Poly(ADP-ribose) Polymerase Promoter to Modulate Its Activity in Vitro *

Poly(ADP-ribose) polymerase-1 (PARP-1) catalyzes the rapid and extensive poly(ADP-ribosyl)ation of nuclear proteins in response to DNA strand breaks, and its expression, although ubiquitous, is modulated from tissue to tissue and during cellular differentiation. PARP-1 gene promoters from human, rat, and mouse have been cloned, and they share a structure common to housekeeping genes, as they lack a functional TATA box and contain multiple GC boxes, which bind the transcriptional activator Sp1. We have previously shown that, although Sp1 is important for rat PARP1 (rPARP) promoter activity, its finely tuned modulation is likely dependent on other transcription factors that bind the rPARP proximal promoter in vitro. In this study, we identified one such factor as NF1-L, a rat liver isoform of the nuclear factor 1 family of transcription factors. The NF1-L site on the rPARP promoter overlaps one of the Sp1 binding sites previously identified, and we demonstrated that binding of both factors to this composite element is mutually exclusive. Furthermore, we provide evidence that NF1-L has no effect by itself on rPARP promoter activity, but rather down-regulates the Sp1 activity by interfering with its ability to bind the rPARP promoter in order to modulate transcription of the rPARP gene.

Poly(ADP-ribose) polymerase-1 (PARP-1) 1 is a nuclear enzyme which catalyzes the addition of ADP-ribose units from nicotinamide adenine dinucleotide (NAD ϩ ) onto itself and other nuclear proteins such as histones and topoisomerases (reviewed in Refs. 1 and 2). It is made up of three functional domains, namely the amino-terminal DNA-binding domain, the central automodification domain, and the carboxyl-termi-nal catalytic domain (3). Although PARP-1 is often associated with DNA repair, because of its rapid and extensive activation following DNA damage (4,5), it has also been implicated in other major nuclear functions such as transcription (6,7), DNA replication (8,9) and recombination (10). PARP-1 is also important for cell differentiation (11)(12)(13) and is involved in cell death, likely acting as a molecular switch between apoptosis and necrosis (14).
PARP-1 is expressed in all organs, albeit at varying degrees, with highest mRNA expression found in brain, thymus, heart, and testis (15,16). PARP-1 mRNA level is regulated at the cell cycle level, reaching its peak at either the G 1 (17)(18)(19)(20) or the S phases (21,22). It has also been shown that a decrease in PARP-1 mRNA levels is associated with cellular differentiation (23)(24)(25)(26) and senescence (27), whereas an increase is observed upon activation of lymphocytes (17,28) or peripheral blood mononuclear cells (18). All these studies show that, although PARP-1 is ubiquitously expressed, its modulation, likely through complex transcriptional regulation, is critical to major cellular functions.
In order to better understand the transcriptional mechanisms regulating PARP-1 expression, the PARP-1 gene promoter has been identified and cloned from three mammalian species, human (29,30), rat (31), and mouse. 2 All three mammalian proximal promoters share a similar structure proper to housekeeping genes; they lack a functional consensus TATA box, are GC-rich, and contain a consensus initiator sequence surrounding the transcription start site. 3 The human promoter has been shown to contain binding sites for transcription factors Sp1, AP-2 (30), YY1 (32), and Ets (33), whereas the mouse promoter was recently shown to be down-regulated by a complex of adenovirus E1A protein and pRb (34). In the case of the rat PARP-1 (rPARP) proximal promoter, we have shown that it is highly but not completely dependent on five Sp1 binding sites (35,36), and that other nuclear proteins, including one likely belonging to the nuclear factor 1 (NF1) family of transcription factors (37), bind to and potentially regulate rPARP promoter transcription.
In this study, we further characterize the rPARP promoter and demonstrate that it binds the rat liver form of NF1 known as NF1-L (38). Furthermore, we provide evidence that NF1-L does not actively modulate rPARP transcriptional activity but rather prevents Sp1 from binding the rPARP promoter in vitro and thus causes a concentration-dependent down-regulation of rPARP promoter activity. * This work was supported by the Natural Sciences and Engineering Research Council of Canada Grant OGP0138624 (to S. L. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: PARP-1, poly(ADP-ribose) polymerase-1; bp, base pair(s); CAT, chloramphenicol acetyltransferase; CM-Sepharose, carboxymethyl-Sepharose; DMS, dimethylsulfate; EMSA, electrophoretic mobility shift assay; EtBr, ethidium bromide; hGH, human growth hormone; NAD ϩ , nicotinamide adenine dinucleotide; NF1, nuclear factor 1; PCR, polymerase chain reaction; rPARP, rat PARP-1; SL2, Drosophila Schneider line 2.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-The rPARP promoter fragment pCR3 (Fig. 1), spanning region Ϫ103 to ϩ13 relative to the mRNA start site, and its Sp1 triple-mutant derivative pCR3/F2-F3-F4m, have been described previously (36). The site-directed mutant pCR3-L3m was produced by the polymerase chain reaction (PCR), using pCR3 as template and the synthetic oligomers L3m (5Ј-GCATGCCTGCAGTCGCGCTAAAAAAA-AACCCCG-3Ј, mutated nucleotides are shown in boldface) and pCAT-3Ј (5Ј-CTCAGATCCTCTAGAGTCG-3Ј). Unique PstI and XbaI sites (underlined in the primer sequences) were included for cloning purposes. Amplifications were performed in a final volume of 50 l, and reaction mixtures contained 100 ng of DNA, 250 M each dNTP (Amersham Pharmacia Biotech, Baie-d'Urfé, Quebec, Canada), 50 mM KCl, 1.5 mM MgCl 2 , 10 mM Tris-HCl, pH 9.0, 2 M each synthetic primer, and 3.5 units of Taq DNA polymerase (Amersham Pharmacia Biotech). Using a GeneAmp 2400 thermal cycler (PE Biosystems, Streetsville, Ontario, Canada), the samples were subjected to an initial denaturation at 94°C for 180 s and were then processed through 35 cycles of denaturation at 94°C for 20 s, annealing at 61°C for 20 s, and elongation at 72°C for 40 s, followed by a final elongation of 120 s at 72°C. The resulting PCR products were run through a 1.5% agarose gel in 1ϫ TAE (40 mM Tris acetate, 1 mM EDTA) containing 0.1 g/ml ethidium bromide (EtBr) and were isolated using the QIAEX II gel extraction kit (Qiagen, Mississauga, Ontario, Canada). The purified DNA fragments were then digested with PstI (Amersham Pharmacia Biotech) and XbaI (Amersham Pharmacia Biotech) and ligated upstream of the chloramphenicol acetyltransferase (CAT) reporter gene in the PstI-XbaI-linearized vector pCATbasic (Promega, Madison, WI). The DNA insert of each recombinant plasmid was sequenced by chain-termination dideoxy sequencing (39) to confirm the mutations.
The expression plasmid pPacNF1-L was constructed by amplifying the NF1-L insert from vector NF1-21/pBS (kindly provided by Dr Paolo Monaci, Istituto di Ricerche di Biologia Molecolare, Rome, Italy), using the synthetic oligonucleotides 5Ј-NF1 (5Ј-GAATTCCTCGAGGCAGT-TATGTATTC-3Ј) and 3Ј-NF1 (5Ј-GAATTCCTCGAGGGTGGTCT-GTCTGG-3Ј). A unique XhoI site (underlined in the primer sequences) was included for cloning purposes. The PCR reaction was performed in a final volume of 50 l, and the mixture contained 200 ng DNA, 250 M each dNTP, 50 mM KCl, 1.5 mM MgCl 2 , 10 mM Tris-HCl, pH 9.0, 2.5 M each synthetic primer, and 2.5 units of Pfu DNA polymerase (Stratagene, La Jolla, CA). Using a GeneAmp 2400 thermal cycler, the samples were subjected to an initial denaturation at 94°C for 30 s and were then processed through 32 cycles of denaturation at 94°C for 30 s, annealing at 47°C for 30 s, and elongation at 72°C for 120 s, followed by a final elongation of 420 s at 72°C. The resulting PCR product was run through a 1% agarose gel in 1ϫ TAE, 0.1 g/ml EtBr, and was isolated using the QIAEX II gel extraction kit. The purified DNA fragment was then digested with XhoI (Amersham Pharmacia Biotech) and ligated into the XhoI-linearized vector pPac, which was obtained by removing the XhoI-inserted Sp1 cDNA from vector pPacSp1 (kindly provided by Dr. Guntram Suske, Institut fü r Tumorforschung, Philipps Universitä t, Marburg, Germany). The recombinant plasmid was then sequenced by chain-termination dideoxy sequencing (39) to confirm the reading frame.
Protein Extracts and Electrophoretic Mobility Shift Assay (EMSA)-Crude nuclear extracts from rat pituitary GH4C1 cells (kindly provided by Dr David D. Moore, Massachusetts General Hospital, Boston, MA) were prepared essentially as described previously (36,40) and kept frozen in small aliquots at Ϫ80°C until use. The NF1-L-containing rat liver carboxymethyl (CM)-Sepharose fraction has been described previously (41).
Competition experiments were performed using EMSA conditions similar to those described above, except that the protein extracts were incubated with the pCR3-labeled probe in the presence of molar excesses (as specified in the caption to Figs. 2 and 5A) of unlabeled double-stranded oligonucleotides bearing the target sequence of CTF/NF-I (see above), the high affinity binding site for Sp1 (5Ј-GATCATATCTG-CGGGGCGGGGCAGACACAG-3Ј) (44), the PARP-1 promoter initiator sequence (5Ј-GATCGCGCCGCCAGGCATCAGCAATCTATCCTG-3Ј), 3 L3, or L3m (see above). In the case of the pCR3/F2-F3-F4m fragment, the GH4C1 protein extract was incubated with the labeled probe in the presence of molar excesses (as specified in the caption to Fig. 6) of the HindIII-XbaI pCR3 fragment (which was isolated from a 1.5% agarose gel in 1ϫ TAE, 0.1 g/ml EtBr and purified using the QIAEX II gel extraction kit), or the unlabeled oligonucleotides CTF/NF-I or Sp1 (see above).
For supershift experiments, similar EMSA conditions as described above were used, except that following formation of the DNA-protein complexes, 1.5 l of pre-immune rabbit serum, anti-NF1-L rabbit serum (kindly provided by Prof. P. C. van der Vliet, Laboratory of Physiological Chemistry, Utrecht University, Utrecht, The Netherlands) or anti-Sp1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was added and incubation allowed to proceed for another 8 min.
DNA Footprinting Procedures-Deoxyribonuclease I (DNase I) footprinting analysis of the rPARP promoter fragment pCR5 (36), which encompasses the pCR3 region, was performed using 5 l of the NF1-Lcontaining CM-Sepharose fraction. Briefly, after pCR5 was 5Ј-end-labeled on the top strand, 3 ϫ 10 4 cpm labeled probe was incubated for 10 min at room temperature with the protein extract and treated with DNase I (Worthington, Freehold, NJ) as described previously (45). Digestion products were resolved by electrophoresis on a 8% polyacrylamide sequencing gel.
For dimethylsulfate (DMS) methylation interference, the 140-bp HindIII-XbaI pCR3 fragment was 5Ј-end-labeled on the top strand and partially methylated with DMS essentially as described (46). Once methylated, 6 ϫ 10 4 cpm labeled probe was incubated with 10 l of the NF1-L-containing CM-Sepharose fraction in buffer D and DNA-protein complexes were separated by native polyacrylamide gel electrophoresis as described above. The NF1-L-DNA complex was then visualized by autoradiography and isolated by electroelution (47). The isolated labeled DNA was finally treated with piperidine (46) and further analyzed on a 8% sequencing gel.
The GH4C1 cells were transiently transfected with either pCR3 or its mutant derivative pCR3-L3m by the calcium phosphate precipitation method as described (48), using 15 g of test plasmid and 5 g of the human growth hormone (hGH) gene-encoding plasmid pXGH5 (49). In the case where the expression vectors NF1-L/pSI (kindly provided by Dr Winnie Eskild, Institute of Medical Biochemistry, University of Oslo, Oslo, Norway) or NF1-X/pcDNA3 (kindly provided by Dr. Bin Gao, Medical College of Virginia, Richmond, VA) were added, GH4C1 cells were transfected with GeneSHUTTLE-20 cationic liposome (Quantum Biotechnologies, Montréal, Quebec, Canada), using 1 g of test plasmid, 0.5 g of pXGH5, and the expression vectors as described in the caption to Fig. 5B. The SL2 cells were transfected by the calcium phosphate precipitation method, using 13 g of test plasmid, 5 g of the ␤-galactosidase (lacZ) gene-encoding plasmid pAc5/V5-His/LacZ (Invitrogen, Carlsbad, CA), and the pPacSp1 or pPacNF1-L expression vectors as described in the legend to Fig. 6A. CAT activities were measured as described previously (50) and normalized to either hGH secreted into the culture medium (for GH4C1), which was assayed using a radioimmunoassay kit (Medicorp, Montréal, Quebec, Canada), or lacZ produced in the cells (for SL2), which was measured as described (51). Each single value is expressed as either 100ϫ (% CAT in 4 h)/ng of hGH for GH4C1, or (% CAT in 4 h)/(lacZ/min/ml) for SL2, and corresponds to the mean of at least two individual transfection experiments performed in triplicate. Standard deviation for the CAT assay is provided for each transfected plasmid and for each cell type. Only values 3 times greater than the background level caused by the reaction mixture alone (usually corresponding to ϳ0.15% chloramphenicol conversion) were considered significant.

Rat NF1-L Binds the rPARP Promoter in Vitro at a Site
Overlapping an Sp1 Binding Site-We have previously demonstrated that, although Sp1 specifically binds to (Fig. 2, A and B) and strongly transactivates the GC-rich rPARP promoter, it is not sufficient to account for all of its transcriptional activity (36). Furthermore, a nuclear protein likely belonging to the NF1 family of transcription factors was shown to bind to the rPARP minimal promoter region, which extends from Ϫ103 to ϩ13 (pCR3, see Fig. 1 and Ref. 37). In fact, EMSA analysis, using a heparin-Sepharose-and CM-Sepharose-enriched rat liver nuclear extract, revealed the specific binding of a protein that is clearly distinct from Sp1 ( Fig. 2A). The interaction of this rat liver DNA-binding protein with the rPARP promoter likely occurs through a putative NF1 binding site (L3) previously identified on the pCR3 fragment (36,37), since formation of the DNA-protein complex was completely abolished by competition with an excess of unlabeled oligonucleotide bearing the target sequence for human CTF/NF-I (42) (Fig. 2A). In contrast, competition using unlabeled oligonucleotides bearing unrelated sequences such as that for Sp1 or the initiator sequence from the PARP promoter 3 had no effect on pCR3-rat liver protein complex formation ( Fig. 2A). However, when using a crude nuclear protein extract from rat pituitary GH4C1 cells, only Sp1 forms a complex with the pCR3 fragment, since its formation is prevented by an Sp1-specific antiserum but not by a pre-immune or an NF1-L-specific antiserum (Fig. 2B).
Using DNase I and DMS methylation interference footprinting, we precisely mapped the binding site of the NF1-like protein to the L3 region of the rPARP proximal promoter (Fig.  3, A and B), which we had previously identified (37) as a potential NF1 target site due to its similarity both to the consensus sequence recognized by members of the NF1 family (42) and to the proximal silencer-1 element from the rat growth hormone gene (52). Interestingly, the footprinted L3 region completely overlaps the F2 region (see Figs. 1 and 3), which was shown to bind Sp1 (36).
In order to unequivocally identify the rat liver L3-binding protein as a genuine member of the NF1 family, we performed supershift experiments using an antiserum specific to the rat liver form of NF1 known as NF1-L (38). As shown in Fig. 4A, the rat liver DNA-binding protein formed a complex (NF1-L) of similar electrophoretic mobility using either the NF1 consensus binding site or the L3 sequence as the labeled probe, and this complex was efficiently supershifted by the anti-NF1-L antiserum (C1). As shown in Fig. 4B, the same rat liver protein bound the pCR3 DNA fragment to form a specific complex (NF1-L), which was supershifted by the anti-NF1-L antiserum (C1), but not by the pre-immune rabbit serum (PS) or by an anti-Sp1 antibody (Sp1Ab). We therefore conclude that the rat liver protein that binds to the rPARP proximal promoter region is NF1-L.
Members of the NF1 Family Indirectly Down-regulate the rPARP Promoter Activity-To characterize in more detail the effect of NF1 on the rPARP promoter, the L3 element was mutated at positions critical for its recognition by NF1 (Fig. 3 and Ref. 37) in order to yield the mutant derivative pCR3-L3m. This mutated L3 sequence was previously shown to abolish NF1 binding (37) and thus would give valuable information in assessing the functional role of NF1 on rPARP promoter transcriptional activity. When the mutated pCR3-L3m construct was transiently transfected into rat pituitary GH4C1 cells, its activity was similar to that directed by wild-type pCR3 (Fig.  5A), suggesting that NF1 had no obvious effect on rPARP transcription. This is consistent with previous observations using the mutant derivative pCR3/F2-F3-F4m, in which all three Sp1 binding sites (F2, F3, and F4; see The rPARP promoter sequence that is contained on the plasmid pCR3 is depicted, along with the position of the three previously identified Sp1 binding sites (F2, F3, and F4) (36). In addition, the position of a target sequence for an NF1-like nuclear protein that binds the rPARP promoter is provided (L3; boxed nucleotides) (37). Numbers indicate positions relative to the rPARP mRNA start site (indicated by the arrow).

FIG. 2. EMSA analysis of NF1-L and Sp1 binding to the rPARP basal promoter.
A, the 140-bp HindIII-XbaI pCR3 fragment from the rPARP promoter was 5Ј end-labeled and incubated with crude nuclear extracts obtained from rat pituitary GH4C1 cells or with a CM-Sepharose enriched preparation of NF1-L (CM-Sep), either alone or in the presence of a 500fold molar excess of unlabeled, doublestranded oligonucleotides (NF1, Sp1, or Inr). DNA-protein complexes were then resolved by EMSA as detailed under "Experimental Procedures." The position of both the NF1-L/pCR3 (NF1-L) and Sp1/ pCR3 (Sp1) DNA-protein complexes is shown, along with that of the free probe (U). P, labeled probe with no proteins added. B, the 140-bp HindIII-XbaI pCR3 labeled probe used in A was incubated with crude nuclear proteins from GH4C1 cells either alone or in the presence of antibodies directed against either Sp1 (Sp1Ab) or NF1-L (NF1Ab). As a negative control, incubation was also performed in the presence of pre-immune serum (PS). The position of the Sp1/pCR3 complex (Sp1) is provided, along with that of the free probe (U). scriptional activity when transiently transfected into GH4C1 cells (36) despite the fact that GH4C1 cells displayed efficient NF1 binding activity on the pCR3/F2-F3-F4m DNA fragment (Fig. 5B). Indeed, binding of an NF1 protein from GH4C1 cells to labeled pCR3/F2-F3-F4m was efficiently competed by ex- FIG. 3. Footprinting analysis of NF1-L binding to the rPARP promoter. A, a 282-bp HindIII-XbaI obtained from the recombinant plasmid pCR5 (36) was 5Ј end-labeled at its HindIII site and incubated with the CM-Sepharose-enriched preparation of NF1-L and then subjected to DNase I digestion. A control sample treated with DNase I but with no protein added (U) and a DNA "G" sequencing reaction performed on the same probe (G) were also included. The DNA sequence of the L3 protected region is indicated along with its positioning relative to the rPARP mRNA start site. B, the 140-bp HindIII-XbaI pCR3 fragment used in Fig. 2 was 5Ј-end-labeled on the top strand, partially methylated with DMS and incubated with CM-Sepharose-enriched NF1-L. The NF1-L/pCR3 complex (C) was resolved by EMSA and isolated by electroelution, along with the free probe (U). The labeled DNA from the complex was then treated with piperidine and further analyzed on a sequencing gel. The position of the two G residues whose methylation by DMS interferes with binding of NF1-L is shown, along with the surrounding rPARP promoter sequence.

FIG. 4. Binding of NF1-L to the rPARP promoter as revealed by supershift analyses in EMSA.
A, CM-Sepharose-enriched NF1-L (CM-Sep) was incubated in the presence of a 5Ј end-labeled doublestranded oligonucleotide bearing either the sequence of the high affinity CTF/NF-I consensus binding site (NF1) or that of the rPARP L3 element (L3). The binding reaction was performed in either the absence or the presence of a polyclonal antibody directed against rat liver NF1-L (NF1Ab). Formation of DNA-protein complexes was then monitored by EMSA as in Fig. 2. The position of both the NF1-L and a supershifted complex (C1) resulting from the binding of the NF1Ab to the NF1-L/ labeled probe complex is indicated, along with that of a nonspecific signal (NS) that often appears in EMSA. U, unbound fraction of the labeled probes. B, the 140-bp HindIII-XbaI pCR3 labeled probe used in Fig. 2 was incubated with the CM-Sepharose-enriched preparation of NF1-L either alone or in the presence of antibodies directed against either NF1-L (NF1Ab) or Sp1 (Sp1Ab). As a negative control, incubation was also performed in the presence of pre-immune serum (PS). The position of both the NF1-L/pCR3 complex (NF1-L) and its corresponding supershifted counterpart (C1) is provided, along with that of the free probe (U). cesses of unlabeled NF1 oligonucleotide and by unlabeled pCR3 wild-type fragment (Fig. 5B). These results suggest that an NF1 protein from GH4C1 cells can bind to but has no direct effect on the rPARP promoter. This GH4C1 protein displays different electrophoretic mobility than liver-enriched NF1-L (compare Figs. 4B and 5B); this may suggest that those are two distinct proteins from the NF1 multigene family (53) or that, as suggested by Paonessa et al. (38), liver NF1-L is a cleavage product of a larger polypeptide which may be the GH4C1 NF1 protein. However, we believe the latter to be unlikely, as the anti-NF1-L antiserum does not supershift the NF1 protein from GH4C1 nuclear extract (data not shown).
Cotransfection of the wild-type pCR3 along with plasmids encoding high expression levels of members of the NF1 family, such as NF1-L (38) and NF1-X (54), led to a 2-fold decrease of transcriptional activity directed by pCR3 in GH4C1 cells (Fig.  6A, black columns, and data not shown). In contrast, overexpression of NF1-L practically had no influence on the NF1-site mutated derivative pCR3-L3m fragment (Fig. 6A, white columns). To gain more insight into the mode of action of NF1 on the rPARP promoter, we took advantage of the Drosophila SL2 cells, which lack classical NF1 and Sp1 activities (55,56). NF1-L alone had no effect on the transcriptional activity of pCR3, in contrast to Sp1, which increased pCR3 activity over 3-fold (Fig. 6B). However, coexpression of both Sp1 and NF1-L led to a concentration-dependent decrease of the transcriptional activation mediated by Sp1 on the pCR3 promoter fragment (Fig. 6B). In fact, when equal amounts of Sp1 and NF1-L expression vector DNA were cotransfected, NF1-L completely abolished Sp1-mediated activation of pCR3 transcription, which returned to basal level (Fig. 6B).
We then used the pCR3 fragment and its Sp1 mutant derivative pCR3/F2-F3-F4m labeled probes in EMSA with the NF1-L-enriched rat liver protein extract and the GH4C1 nuclear protein extract, which supports strong Sp1 binding to the pCR3 probe (see Fig. 2A and Ref. 36). When the pCR3 probe was incubated with a constant amount of NF1-L and increasing amounts of GH4C1 extract, NF1-L binding to pCR3 was abolished by the increasing amounts of Sp1 (Fig. 7A, left panel). On the other hand, NF1-L binding to the Sp1 mutant probe was expectedly unaffected by increasing amounts of GH4C1 nuclear extract (Fig. 7A, right panel). The reverse is also true, i.e. increasing amounts of NF1-L abolished Sp1 binding to pCR3 (Fig. 7B, left panel), suggesting that binding of NF1-L and Sp1 to the rPARP promoter is mutually exclusive. This and the SL2 transfection experiments show that NF1-L acts indirectly on rPARP promoter transcriptional activity by preventing Sp1 from binding to its target and therefore leads to decreased transcriptional activation of the rPARP promoter.

DISCUSSION
Poly(ADP-ribosyl)ation is a major post-translational modification potentially involved in chromatin architecture (1) and likely to affect important nuclear functions such as DNA repair, recombination, replication, and transcription. Although PARP-1, the main poly(ADP-ribosyl)ating enzyme (2), is expressed in a variety of cells and organs, its transcriptional regulation during cell cycle progression, cell proliferation, and differentiation is likely to impact on the timing and extent of PARP-1 activity and function. Indeed, it has been shown that PARP-1 mRNA levels increase just before maximal enzymatic activity during thymocyte proliferation (22), astrocyte proliferation and differentiation (25), and lymphocyte activation (28). Furthermore, elevated levels of PARP-1 mRNA apparently account for the enhanced poly(ADP-ribosyl)ation observed in Ewing's sarcoma cells (24).
In this study of the rPARP proximal promoter, we investigated the regulatory mechanisms involved in its expression and have highlighted the roles of both Sp1 and NF1 in finetuning rPARP gene transcription. Whereas Sp1 clearly upregulated rPARP transcription (Ref. 36 and Fig. 6B), the role of NF1 appeared more ambiguous; although mutation of the NF1 binding site in the rPARP fragment pCR3 had no effect on its transcriptional activity in rat pituitary GH4C1 cells (Fig. 5A), overexpression of NF1-L or NF1-X in the same cells downregulated pCR3 activity (Fig. 6A and data not shown). In Dro- FIG. 5. Transient transfection and EMSA analyses of pCR3 derivatives that bear mutations in either the NF1 or Sp1 target sites. A, the L3 element from the rPARP promoter was mutated in the context of the pCR3 plasmid in order to prevent its recognition by NF1. Both the wild-type pCR3 and its L3 mutated derivative (pCR3-L3m) were transiently transfected into rat pituitary GH4C1 cells in order to assess the effect of NF1 binding to the L3 element on the transcriptional activity of the rPARP pCR3 promoter fragment. CAT activities were measured and normalized as detailed under "Experimental Procedures." Standard deviation is provided for both plasmids. B, a 140-bp HindIII-XbaI fragment derived from the recombinant plasmid pCR3/ F2-F3-F4m that bears mutations in all three rPARP promoter proximal Sp1 binding sites was 5Ј end-labeled and incubated with crude nuclear proteins from GH4C1 cells in the presence of increasing concentrations (150-, 400-, and 1250-fold molar excesses) of various unlabeled competitor DNAs (the wild-type pCR3 fragment or double-stranded oligonucleotides bearing either the NF1 or Sp1 consensus sequences). Binding of NF1 to the Sp1-mutated labeled probe was monitored by EMSA as in Fig. 2. The position of the NF1 complex is indicated, along with that of a weak but specific complex of unknown identity (*). U, unbound fraction of the labeled probe. Cϩ, labeled probe incubated with proteins but without competitor DNA. sophila SL2 cells, which lack endogenous Sp1 and NF1 (55,56), NF1-L had no direct effect on the transcriptional activity mediated by the pCR3 fragment, but inhibited the Sp1-mediated up-regulation of this rPARP promoter fragment (Fig. 6B). NF1-L likely acted in a concentration-dependent fashion by preventing and/or displacing Sp1 from binding to the rPARP proximal promoter (Fig. 7). NF1-L has been previously shown to negatively regulate rat growth hormone (41) and peripherin (57) gene transcription.
Disruption of the Sp1 gene has revealed that, although important for embryonic development, it is dispensable for cell growth and differentiation (58), suggesting that it may not be relevant to differentiation-and cell cycle-related PARP-1 transcription. On the other hand, Sp1 is one of at least 16 members of the Sp/XKLF family of transcription factors able to bind GC or GT boxes (59), and it is therefore plausible that other members of this family take over Sp1 function in Sp1-deficient cells. Nevertheless, a number of genes have been shown to rely on Sp1 for either cell cycle-regulated expression, including thymidine kinase (60) and dihydrofolate reductase (61), or differen-tiation-associated expression, such as acetylcholine receptor ⑀ (62) and keratin K3 (63). Interestingly, these Sp1-related events are dependent upon modification of Sp1 activity, either by its association with other regulators such as cell cyclerelated transcription factor E2F (60,61), by its phosphorylation (62) or by its competition with other transcription factors such as AP-2 (63). The analysis of the keratin K3 promoter (63) is particularly relevant to the present study, in that the ratio between Sp1 and AP-2, rather than actual regulation of Sp1, seems to be critical to the modulation of keratin K3 expression.
The NF1 family is composed of four members encoded by distinct genes, NF1-A, -B, -C, and -X (53), producing distinct protein products, which can form homo-or heterodimers (64). In addition, all four NF1 mRNAs can be differentially spliced to yield a multitude of proteins with subtle differences in their transactivation capacities (58,65). Disruption of the NF1-A gene leads to severe neurological defects, with the majority of mice dying perinatally, yet other organs seem to develop normally (66). As in the case of the Sp1 knockout study, other members of the NF1 family may compensate for the absence of FIG. 6. Co-transfection experiments in rat GH4C1 and Drosophila SL2 cells. A, both the wild-type rPARP promoter-bearing plasmid pCR3 (black columns) and its mutated derivative pCR3-L3m (white columns) were transiently transfected into GH4C1 cells along with the NF1-L expression plasmid (0.5 or 1 g). As a negative control, pCR3 was also cotransfected along with the empty vector pSI (which was used for constructing the NF1-L expression plasmid). CAT activities were measured and normalized as detailed under "Experimental Procedures." Values are expressed as percentage of CAT activity relative to the level directed by either pCR3 and pCR3-L3m when cotransfected with the control plasmid pSI. Standard deviation is provided for each value shown. B, the rPARP promoterbearing plasmid pCR3 was transfected into Drosophila SL2 cells along with the empty vector pPac, or with expression plasmids encoding either NF1-L (0.2, 0.5, or 1 g) or Sp1 (1 g), either individually or in combination. CAT activities were measured and normalized to ␤-galactosidase as described under "Experimental Procedures." Standard deviation is provided for each selected condition. NF1-A. Nevertheless, proteins of the NF1 family have been implicated in adipocyte differentiation (67) and in liver regeneration (68). In adipocytes, the adipocyte p2 gene expression, which is dependent on NF1, is induced upon differentiation (67), as is expression of PARP-1 (13), even though Sp1 expression is down-regulated in the early stages of adipocyte differentiation (69). On the other hand, in liver regeneration, NF1 concentration decreases (68) whereas PARP-1 expression increases (20), which is consistent with Sp1 activity being more prominent (70). That NF1 proteins exhibit stimulatory activity in one case (i.e. adipocyte p2 gene expression) and inhibitory activity in the other (i.e. the rPARP promoter) has been previously observed and shown to be dependent on the promoter or the cell type (71).
Some correlation is observed between the level of PARP-1 mRNA in mouse organs (15) and that of Sp1 (72), although it is not complete. Indeed, mouse thymus and testis express nearly as much PARP-1 (15) despite that Sp1 levels are 8 times higher in thymus (72), whereas high expression of Sp1 mRNA in the lung (72) does not result in proportional PARP-1 expression (15), thereby confirming that Sp1 is an important but not the only transcription factor involved in PARP-1 gene expression. Interestingly, NF1 proteins are also highly expressed in the lung (65) and, in light of the present study, may contribute to down-regulation of PARP-1 expression in the lung.
Composite elements are transcriptionally active DNA sequences, which bind two or more transcriptional regulators, and they have been found in many vertebrate gene promoters (73). They provide an efficient way of fine-tuning gene expression in different cellular situations, by exploiting either the synergistic or antagonistic effects of the transcription factors they bind to respectively overactivate or down-regulate transcription. In the present study, we provide evidence that the rPARP gene promoter is regulated by the antagonistic effect of NF1 over Sp1 transactivation on a composite element. Both of these transcription factors have been found in composite elements on other gene promoters. Antagonistic binding of Sp1 has been reported with erythroid cell-specific factor NF-E1 on the human -globin promoter (74), where it may account for tissue-specific expression, with AP-2 on the rabbit K3 keratin promoter (63), where it is associated with differentiation-specific expression, and with NF-B on the human fas promoter (75), where it is involved in lymphocyte activation-related upregulation. In the case of NF1, its mutually exclusive binding has been reported with AP-2 on the hGH promoter (76), but also, and consistent with our results, with Sp1, on the rat liver-type arginase promoter (77), on the murine collagen ␣1 (1) promoter (78), and on the rat ␣ 1B adrenergic receptor gene promoter (79). In the latter two studies, mutually exclusive binding of Sp1 and NF1 has been suggested to confer an important mechanism of tissue-specific expression. Interestingly, and in contrast to the results presented here on the rPARP promoter, Sp1 actually down-regulates NF1-mediated activation of the collagen ␣1 (1) promoter (78).
We have shown that Sp1-activated transcription of the rPARP gene promoter is susceptible to subtle regulation via the mutually exclusive binding of NF1-L. As Sp1 and NF1-L are both individual members of extended families of transcription factors, it is expected that transcriptional regulation through a composite Sp1/NF1 element, such as those reported here on the rPARP promoter, on the rat liver-type arginase promoter (77), the murine collagen ␣1 (1) promoter (78), or the rat ␣ 1B adrenergic receptor gene promoter (79), will lead to complex patterns of activation/down-regulation depending on the cell type or the cellular situation analyzed, as members of both families may be differentially expressed in such instances.