Transcriptional regulation of neuronal nicotinic acetylcholine receptor genes. Functional interactions between Sp1 and the rat beta4 subunit gene promoter.

To date, 11 members (α2-α9 and β2-β4) of the neuronal nicotinic acetylcholine receptor gene family have been identified. These genes encode subunits that form distinct receptors with different pharmacological and physiological profiles in temporally and spatially restricted patterns within the nervous system. Distinct molecular mechanisms probably orchestrate the expression of various receptor subtypes, yet little is known of specific transcriptional regulatory elements and their associated factors that are responsible for this segregated pattern of expression. Here we report the identification of an element, in the 5′-flanking region of the rat β4 subunit gene, containing a CA box that is necessary for β4 promoter activity in a transiently transfected cholinergic cell line, SN17. This element was shown to interact with a protein(s) in SN17 nuclear extracts that is antigenically related to the transcriptional activator Sp1. Furthermore, co-transfection experiments confirmed that Sp1 can transactivate a β4 promoter-reporter gene construct, indicating that Sp1 is necessary, at least in part, for transcriptional activation of the β4 subunit gene.

To date, 11 members (␣2-␣9 and ␤2-␤4) of the neuronal nicotinic acetylcholine receptor gene family have been identified. These genes encode subunits that form distinct receptors with different pharmacological and physiological profiles in temporally and spatially restricted patterns within the nervous system. Distinct molecular mechanisms probably orchestrate the expression of various receptor subtypes, yet little is known of specific transcriptional regulatory elements and their associated factors that are responsible for this segregated pattern of expression. Here we report the identification of an element, in the 5-flanking region of the rat ␤4 subunit gene, containing a CA box that is necessary for ␤4 promoter activity in a transiently transfected cholinergic cell line, SN17. This element was shown to interact with a protein(s) in SN17 nuclear extracts that is antigenically related to the transcriptional activator Sp1. Furthermore, co-transfection experiments confirmed that Sp1 can transactivate a ␤4 promoter-reporter gene construct, indicating that Sp1 is necessary, at least in part, for transcriptional activation of the ␤4 subunit gene.
Neuronal nicotinic acetylcholine (nACh) 1 receptors belong to a large family of related neurotransmitter receptors that are expressed within the central and peripheral nervous systems (CNS and PNS). Little is known of their function within the CNS, although a recent report suggests that presynaptic nACh receptors may modify fast excitatory transmission in the CNS (1). Additionally, the loss of cholinergic neurons in pathological states involving alterations in cognition and memory (e.g. Alzheimer's disease) implicates nACh receptors in these processes (2). In support of this premise, transgenic mice lacking the nACh receptor ␤2 subunit performed poorly in passive avoidance testing, suggesting a defect in associative memory (3). Further underscoring the importance of nACh receptors within the CNS is a recent report describing the development of au-tosomal dominant, nocturnal, frontal lobe epilepsy in humans, resulting from a missense mutation in the ␣4 subunit gene (4). It is clear that a significant amount of work remains to be done in order to fully understand the function of nACh receptors, including the identification and characterization of individual subunit genes.
Eleven members of the nACh receptor gene family have been identified to date, and they include ␣2-␣9 and ␤2-␤4 (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). These subunits have been demonstrated in either Xenopus oocytes or in vivo to form a variety of distinct heteromeric (␣2, ␣3, and ␣4 with either ␤2 or ␤4) and homomeric (␣7, ␣8, and ␣9) receptors with different pharmacological and physiological profiles (21)(22)(23). This functional heterogeneity likely results from incorporation of different ␣ and ␤ subunits into mature receptors (19,24). Further complicating the characterization of nACh receptors is the observation that individual nACh receptor subunit gene expression occurs in distinct yet overlapping temporally and spatially restricted patterns within the CNS and PNS (23,(25)(26)(27)(28)(29). This segregated pattern of expression may influence a variety of neuronal processes such as synaptic changes during development and modulation of neurotransmitter release by influencing presynaptic neuronal excitability (1,23). Characterizing the molecular mechanisms leading to the diversity in expression and function of nACh receptors will further the understanding of the development and maintenance of the nervous system and the role nACh receptors play in these processes.
Despite the information gained concerning the expression patterns and varied physiological profiles of distinct nACh receptors, very little is known of the molecular mechanisms underlying this diversity although evidence indicates that positive and negative regulatory mechanisms are involved (30 -37). Furthermore, the segregation in nACh receptor gene expression suggests that transcriptional mechanisms lead to the heterogeneous expression within the nervous system, although the evidence identifying transcriptional control elements and their cognate DNA-binding proteins is limited. As such, our laboratory has focused upon characterizing the molecular mechanisms leading to the expression of the rat ␤4 nACh receptor subunit gene. Interestingly, the ␤4 subunit gene is clustered within a 60-kilobase region of the rat genome with the genes encoding the ␣3 and ␣5 subunits, indicating that there may be common regulatory mechanisms leading to their expression (11). Identification of mature nACh receptor subtypes containing ␣3, ␣5, and ␤4 expressed within the chick ciliary ganglion supports this hypothesis (38).
Our earlier investigations led to the identification of a 19base pair (bp) segment of DNA within the 5Ј-flanking region of the ␤4 subunit gene (Ϫ82 to Ϫ63 relative to the transcriptional start site) that when deleted or mutated resulted in a significant loss of promoter activity in transient transfection assays using ␤4 promoter-luciferase reporter gene constructs (35). Interestingly, deletion of the 19-bp element resulted in greater than 80% loss in activity, whereas site-directed mutagenesis of this sequence resulted in an approximately 60% loss in luciferase activity. These disparate results prompted investigations to isolate other potential cis elements within this region of the promoter that are also involved in ␤4 gene regulation.
In this report, we demonstrate that a CACCC motif (CA box; immediately downstream of the 19-bp element) is required for ␤4 gene expression. Mutagenesis of the CA box resulted in drastically reduced levels of luciferase activity in transient transfection analyses. Results from electrophoretic mobility shift assays (EMSAs) revealed that two major protein complexes bound specifically to this element. Furthermore, these complexes were effectively competed with unlabeled oligonucleotides containing the CA box or a consensus binding site for the transcription factor Sp1. Subsequent results from mobility supershift experiments using an Sp1-specific antibody demonstrated that the complexes contained Sp1 or an antigenically related protein(s). Last, results from transactivation experiments in Drosophila Schneider SL2 cells revealed that Sp1 can transactivate the ␤4 promoter. Taken together, these data indicate strongly that Sp1, in conjunction with the 19-bp element-binding protein and possibly other cellular factors, participates in ␤4 gene expression.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-SN17 cells (39) were maintained as described previously (35). Transient transfections in SN17 cells were performed essentially as described (35). Briefly, SN17 cells were plated at 4.0 ϫ 10 5 cells/60-mm culture dish approximately 24 h prior to the addition of DNA precipitates. Wild type and mutant ␤4-promoter/luciferase constructs (see Figs. 1 and 2) were transfected at 5.0 g/60-mm culture dish along with 5.0 g of a ␤-galactosidase expression plasmid, pCH110 (Pharmacia Biotech Inc.) by the calcium phosphate method (40). The cells were harvested approximately 48 h later, and luciferase assays were performed using a commercially available kit (Promega). Luciferase values were normalized to ␤-galactosidase activity to correct for discrepancies in transfection efficiencies.
Drosophila Schneider SL2 cells were maintained at room temperature in modified Schneider's Drosophila medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum and antibiotics. Cells were seeded at 2 ϫ 10 6 /60-mm culture dish immediately prior to transfection. DNAs were introduced into the cells by lipofectamine-mediated transfection using 6 l of lipofectamine/sample (Life Technologies, Inc.) and 2.0 g of target DNA (pX1B4-FHwt) in the absence or presence of 0.2 g of effector DNA (Sp1 expression construct (pActSp1) containing Sp1 cDNA inserted downstream of the Drosophila actin 5C promoter (41)) per dish. The target DNA constructs were also transfected with the vector pAct (minus Sp1 coding sequences) as a negative control for luciferase expression. Lipofectamine complexes were added to the cells and incubated for 5 h. Following the incubation period, the cells were overlaid with 4 ml of complete medium. The cells were harvested, luciferase assays were performed 48 h after transfection, and values were normalized to total protein concentrations in the extracts.
Generation of Mutant Constructs and DNA Probes-Site-directed mutagenesis of wild type pX1B4FH (Ref. 35; see Fig. 1) was accomplished by polymerase chain reaction using a commercially available kit (Perkin-Elmer) with the oligonucleotides used to construct pX1B4FHmut1-4 (Fig. 2). These oligonucleotides contain a BamHI recognition site and were used in polymerase chain reaction containing wild type pX1B4FH purified DNA and a 3Ј oligonucleotide (5Ј-GTT GGT GTC GAC CCA GCT TCT GTG-3Ј, containing a SalI restriction site) corresponding to the noncoding strand. The products were directionally cloned into the BamHI/SalI restriction sites of pBluescript KSϩ (Stratagene). Clones containing inserts were sequenced, and those positive for the indicated sequence changes were cloned into the promoterless luciferase vector pXP1 (42) to create mutant clones pX1B4FHmut1-4 ( Fig. 2).
Oligonucleotides used in the EMSA were generated with 30-nmol DNA synthesis columns using an Oligo 1000 DNA synthesizer (Beckman). Oligonucleotides were purified using a commercially available kit (Beckman), annealed, and subsequently 5Ј-end-labeled with T4 polynucleotide kinase (Promega) and [␥-32 P]ATP (DuPont NEN). Probes used in DNase I footprinting experiments were obtained by single restriction enzyme digestion with HindIII of clone pBLSKBH (34) containing the BamHI to HindIII sequence shown in Fig. 1. The DNA was end-labeled at the HindIII restriction site as before with T4 polynucleotide kinase, purified, and subsequently digested with BamHI to release the single end-labeled DNA fragment. The insert was purified through a 4% polyacrylamide gel, excised, and eluted overnight in elution buffer (0.2 M NaCl, 20 mM EDTA, pH 8.0, 1% SDS, and 1 mg/ml tRNA) at 37°C.
Electrophoretic Mobility Shift Reactions and Mobility Supershift Experiments-EMSAs were performed as described (35). Radiolabeled double-stranded DNA oligonucleotides were incubated in the presence of SN17 nuclear extract in the presence of 2 g of nonspecific competitor poly(dI⅐dC). Specificity in the binding reactions was demonstrated in competition experiments using double-stranded oligonucleotides corresponding to E1, E2, or E3 (Figs. 1 and 2) and commercially purchased double-stranded oligonucleotides containing consensus binding sites for transcriptional regulators Sp1 (5Ј-ATT CGA TCG GGG CGG GGC GAG C-3Ј), AP1 (5Ј-CGC TTG ATG AGT CAG CCG GAA-3Ј), and TFIID (5Ј-GCA GAG CAT ATA AGG TGA GGT AGG A-3Ј) (Promega). Competitor oligonucleotides were preincubated with nuclear extracts 5 min prior to the addition of probe. Following the addition of specific probe, the reactions were incubated on ice for 45 min to 1 h. The reactions were then separated through 6.0% polyacrylamide gels (prerun at 100 V for 30 min) in 0.5 ϫ TBE buffer (150 V for approximately 4 h). For mobility supershift experiments, specific antiserum to Sp1 (␣-PEP2, Santa Cruz Biotech) or a nonrelated antiserum was preincubated with SN17 cell nuclear extract for 5 min followed by the addition of specific probe. The reactions were incubated on ice for 1-3 h followed by separation through 6% polyacrylamide gels and visualization by autoradiography.
DNase I Footprinting-DNase I footprinting was performed essentially as described (43). Briefly, 20 -50 fmol of DNA probe (see above), end-labeled at the HindIII site of the BamHI/HindIII fragment shown in Fig. 1, were incubated in standard binding reactions as described above (total volume of 55 l), with SN17 nuclear extracts ranging in protein content from 12. DNA sequences that interact with protein(s) within SN17 nuclear extracts, as judged by EMSA and DNase I footprinting experiments, are indicated. *, element 1, the 19-bp sequence shown to interact with specific protein complexes distinct from SP1 (35); ٙ, element 2, the CA box shown to interact with Sp1-like protein(s); #, element 3, containing a core AP1-like DNA binding sequence; ϭ, potential CA boxes. The transcription start is indicated as ϩ1; the BamHI and HindIII recognition sites demarcate the sequence of the probe used in DNase I footprinting (Fig. 3). The wild type clones used in transfection experiments (Fig. 4) are indicated (pX1B4FH and pX1B4D4).
g/ml tRNA), placed on ice, extracted once with phenol/chloroform, and then ethanol-precipitated. The digested products were separated through 6% polyacrylamide, 7 M urea sequencing gels and visualized by autoradiography. Chemical sequencing reactions were performed with the end-labeled BamHI/HindIII to determine the boundaries of protected regions. Experiments using competitor oligonucleotides were performed as described above for EMSA.

RESULTS
DNase I Footprinting Analysis of the ␤4 5Ј-Flanking Region-Previous experiments to identify 5Ј-flanking sequences involved in expression of the ␤4 subunit gene revealed a 226-bp fragment (pX1B4FH; see Fig. 1) that could direct expression of promoterless luciferase expression constructs in a transiently transfected neuronal cell line SN17 (35). Further analysis of this region revealed the presence of a C/T-rich 19-bp sequence (Ref. 35; element 1, see Fig. 1) that when deleted (pX1B4D4, sequence shown in Fig. 1) resulted in an approximately 80% decrease of luciferase activity in transfected cells relative to that obtained with the larger subclone (pX1B4FH; sequence shown in Fig. 1). Site-directed mutagenesis of this 19-bp sequence resulted in an approximately 60% decrease in luciferase activity (35). These results indicated that the 19-bp sequence acted positively in directing ␤4 gene expression and also indicated that additional positively and/or negatively acting sequences were involved in directing ␤4 gene expression. A further examination of the 5Ј-flanking DNA sequence within the ␤4 gene revealed the presence of additional C/T-rich sequences similar to element 1 (E1) and three CA boxes (Fig. 1). Therefore, to begin to understand the protein interactions with the 5Ј-flanking region of the ␤4 gene, DNase I footprinting was performed using the BamHI/HindIII fragment (Ϫ255 to ϩ151 of the ␤4 5Ј-flanking DNA) shown in Fig. 1. The BamHI/Hin-dIII fragment was radiolabeled specifically at the HindIII site, and approximately 20 -50 fmol were incubated in DNA-binding reactions (see "Experimental Procedures") with increasing concentrations of SN17 cell nuclear extract (12.5-50.0 g). One large protected region (Fig. 3) was visualized and encompasses nucleotides Ϫ90 to Ϫ46, which contains E1 plus additional bases downstream and adjacent to E1. This downstream region contains a consensus CA box that has been shown to be important in regulating several genes (44 -49). For clarity, we have labeled this downstream region containing the CA box as element 2 (E2) to delineate it from the 19-bp (C/T-rich) element (E1). A second, smaller DNase I-protected region was also observed and corresponds to nucleotides Ϫ119 to Ϫ98, which contain an AP1-like sequence shown to interact specifically with protein(s) in SN17 cell nuclear extracts. 2 This region was labeled element 3 (E3). Competition experiments using unlabeled oligonucleotides corresponding to E1, E2, and E3 (Figs. 1 and 2) and the consensus binding site for the transcriptional activator Sp1, were done to identify specific interactions. Interestingly, E2 and an oligonucleotide containing a consensus binding site for Sp1 competed for protein binding to the footprinted region between nucleotides Ϫ90 and Ϫ46, as did E1 but with apparently a lower efficiency (see "Discussion"). The footprinted region between nucleotides Ϫ119 and Ϫ98 (E3) was inhibited specifically by nonradioactive excess E3 in competition reactions. Taken together, these data indicated that several protein complexes are able to interact with 5Ј-flanking sequences of the ␤4 gene. The experiments presented below focus upon the characterization of E2 and its role in ␤4 gene expression.
Functional Analysis of E2-Since the CA box has been implicated in interacting with a variety of DNA-binding proteins involved in positively regulating several genes (44 -49), we investigated whether the CA box in E2 (see Figs. 1 and 2) plays a functional role in regulating the ␤4 gene as well. Site-directed mutagenesis of E2 was performed using the oligonucleotides listed in Fig. 2 (see pX1B4FHmut1-4). Subsequently, wild type 2 C. B. Bigger, unpublished observations. pX1B4FH (pX1B4FHwt) along with four mutants containing base changes in E1 (pX1B4FHmut1 and pX1B4mut2), E1 and E2 (pX1B4FHmut3), or E2 (pX1B4FHmut4) were transiently transfected into SN17 cells. Approximately 48 h following transfection, the cells were harvested and assayed for luciferase activity. The data (Fig. 4) revealed results similar to those reported earlier for the mutants containing base changes within E1 ( Fig. 1; Ref. 35). Constructs containing mutations within E1 exhibited an approximately 30 -60% decrease in luciferase activity (pX1B4FHmut1 and -2), whereas constructs containing mutations within E1 and E2 (pX1B4FHmut3) or E2 alone (pX1B4FHmut4) resulted in an approximately 80% reduction in luciferase activity, similar to luciferase values obtained with the E1 deletion construct (pX1B4D4; Fig. 4). These results indicate that E2 is required for ␤4 gene expression.

Electrophoretic Mobility Shift Analysis of SN17 Cell Nuclear
Extracts and Element 2-To identify nuclear protein(s) capable of interacting with E2, EMSA experiments were performed using radiolabeled E2 (Fig. 2) and SN17 nuclear extract. Approximately 15-20 fmol of probe were incubated in binding reactions containing increasing concentrations of protein. The products were resolved on 6% polyacrylamide gels in low ionic strength buffer and viewed by autoradiography. Three prominent complexes (Fig. 5, A, B, and C) were detected. To confirm the specificity of the complexes interacting with E2, competition experiments using oligonucleotides corresponding to E1, E2, and E3 (Fig. 2) and the consensus binding sites for AP1, Sp1, and TFIID were preincubated in the binding reactions prior to the addition of labeled E2. Two complexes (Fig. 5, A and  B) bound E2 specifically and were competed only by unlabeled E2 and, interestingly, by an oligonucleotide corresponding to the consensus binding site for Sp1. These data indicated, then, that SN17 nuclear extracts contained Sp1-like protein(s) that interacted with E2 in a specific manner. Proteins Antigenically Related to Sp1 Interact with Element 2-To confirm whether Sp1 or antigenically related proteins interacted with E2, mobility supershift experiments were performed using an antiserum specific for Sp1 or a nonrelated antiserum obtained from a nonimmune rabbit prior to immunization. Standard binding reactions were performed in the presence of specific antiserum (␣-Sp1) or nonrelated serum. An additional shifted band (Fig. 6, complex A) was visualized in lanes containing binding reactions with ␣-Sp1 and either radiolabeled E2 or an oligonucleotide containing the consensus binding sequence for Sp1, whereas those lanes containing SN17 nuclear extract alone or with a nonrelated serum (NRS) exhibited the two complexes previously observed to interact specifically with E2 (compare A and B in Fig. 5 with B and C in Fig. 6). These data indicate that Sp1 or an antigenically related protein(s) can associate specifically with the CA box in E2.
Sp1 Transactivates Wild Type ␤4 Promoter Expression-Results from the preceding experiment (Fig. 6) demonstrated that a protein(s) antigenically related to the transcriptional activator Sp1 interacted with the CA box in E2. To confirm that Sp1 could functionally activate the ␤4 promoter containing a wild type CA box, Drosophila Schneider SL2 cells (which lack Sp1) were co-transfected with pX1B4wt and an Sp1 expression construct (pActSp1; Ref. 41) under the control of the Drosophila actin promoter. To confirm that Sp1 expression was involved in transactivation, the target DNAs were transfected with the pAct vector alone (minus Sp1 coding sequences). A positive control for transactivation included the pSLUC2 vector containing the luciferase gene under the control of the SV40 early promoter (42). The results of these experiments (Fig. 7) indicated that Sp1 can strongly transactivate (10-fold higher) the ␤4 promoter/luciferase construct, pX1B4FHwt.

DISCUSSION
Despite the abundance and variation of nACh receptors within the CNS and PNS, little is known of their roles in the establishment of memory and cognitive responses. Clearly these receptors play an important role in these processes, since the addictive effects of nicotine in increasing the number of nACh receptors within the brain, improving learning tasks, and affecting short term memory and anxiety responses have been well documented (1, 50 -51). Furthermore, the loss of cholinergic neurons in certain diseases (e.g. Alzheimer's disease) affecting memory and the targeted deletion of specific nACh receptor subunit genes (i.e. ␤2) in mice, resulting in impaired memory, implicate nACh receptors in cognition and memory (2)(3). Understanding the mechanism(s) leading to nACh receptor gene expression will enhance our understanding of the processes involved in learning and memory and the "cross-talk" between different neurotransmitter systems.
To date, little is known of the molecular mechanisms leading to the expression of nACh receptors. Both positive and negative regulatory mechanisms have been described (30 -37); however, few reports have described specific DNA-protein interactions involved in activation or repression of specific nACh receptor subunit genes. Members of the Brn-3 POU family have been shown to activate or repress the nACh receptor chicken ␣2 subunit gene (36); transcriptional activators Sp1 and AP2 have been demonstrated or implicated, respectively, in the regulation of the rat ␣3 subunit gene (33); and our laboratory has identified a C/T-rich sequence of 19 bp (E1, Figs. 1 and 2) within the 5Ј-flanking region of the rat ␤4 subunit gene that has been demonstrated to interact with a novel protein(s) (which we have termed neuronal ACh receptor promoter-binding protein(s), or NARP) 3 in extracts from a neuronal cell line and in rat brain (35). Additional work is necessary to identify specific cis-acting sequences and their cognate DNA-binding proteins prior to understanding the molecular mechanisms leading to the expression of distinct nACh receptors within the 3 Q. Du, unpublished observations. In this report we demonstrated conclusively that the transcriptional activator Sp1 associates with a CA box in the 5Јflanking region of the rat ␤4 subunit gene and can activate ␤4 promoter/luciferase constructs in transient transfection experiments. The CA box has been demonstrated to bind distinct proteins in a variety of cell types and participates in the regulation of such genes as the T-cell receptor (44), the erythropoietin receptor (45), members of the globin gene family (47)(48)(49), the glucocorticoid receptor (46), and others (52)(53)(54). Indeed, a mutation within the CA box of the ␥-globin gene promoter has been shown to result in the development of ␤-thalassemia (56 -57).
We have shown that the protein(s) interacting with E1 is distinct from Sp1 (Figs. 3 and 5; Ref. 35), yet we report here that Sp1 or antigenically related proteins interact with the CA box sequence in E2, as demonstrated by DNase I footprinting experiments and EMSA. That E1 competed less efficiently in DNase I footprinting assays is puzzling, although it is possible that protein interactions (e.g. cooperative binding) stabilize complex formation at E1, such that the E1 oligonucleotide alone is less efficient at competing for binding than the 409-bp fragment used in the DNase I footprint analysis in Fig. 3. Competition experiments revealed that the complexes associating with E2 were competed only with excess, unlabeled E2 or with an oligonucleotide corresponding to a consensus Sp1-binding site. Interestingly, E2 competed more effectively with complex formation than the oligonucleotide containing the canonical Sp1 binding sequence. Mobility supershift experiments using a commercial antiserum to Sp1 indicated that Sp1 did indeed associate with the CA box in E2. Importantly, mutagenesis experiments revealed the CA box to be required for ␤4 gene expression. Sp1 could transactivate the ␤4 promoter/luciferase construct (pX1B4FHwt), confirming that Sp1 interacts with the CA box and strongly indicating that Sp1 is required for ␤4 subunit gene expression.
Mutagenesis of the CA box alone in E2 results in an approximately 80% reduction in ␤4 promoter activity. This is not surprising, since the ␤4 gene belongs to the TATA-less class of genes (34). Sp1 has been shown to be important in the regulation of many TATA-less promoters and may function to recruit or stabilize the TFIID initiation complex via a "tethering factor" (58 -63). Additionally, multiple Sp1 binding sites may function to increase basal promoter activity (64,65), and the presence of additional CA boxes in the 5Ј-flanking region of the ␤4 gene distinct from E2 may also function similarly. Preliminary mutagenesis experiments of the remaining CA-like sequences shows this to be likely. 2 Interestingly, deletion of the 19-bp sequence in E1 resulted in an approximately 80% loss in promoter activity, whereas mutagenesis of E1 (see pX1B4FHmut1, Fig. 4) resulted in an approximately 60% loss of promoter activity. Mutagenesis of E2 alone, however, (pX1B4FHmut4; see Fig. 4) resulted in almost a complete loss of promoter activity. Because E2 is present in the deletion clone (pX1B4D4; see Fig. 1), a plausible explanation for the drastic reduction in promoter activity may be that although Sp1 binding to the CA sequence in E2 plays a pivotal role in activating ␤4 gene expression, the complexes formed at E1 and E2 interact. This is consistent with the results discussed above from the DNase I footprinting studies (Fig. 3). Sp1 has been shown to interact synergistically with a variety of nuclear factors in mediating gene expression (e.g. FIG. 8. A simple model for interactions between NARP (E1binding protein) and Sp1. Elements 1, 2, and 3, as defined by DNase I footprinting, are depicted schematically. Potential protein interactions with E3 are indicated with a question mark. Initially, NARP binds to E1 and facilitates or strengthens interactions between E2 and the transcriptional activator Sp1, which then interacts with factor(s) associated with the transcription initiation complex, leading to ␤4 gene expression.
FIG. 7. Sp1 can transactivate a ␤4 promoter/luciferase construct containing element 2. Drosophila SL2 cells were transfected with pX1B4FHwt alone or with an Sp1 expression construct under the control of the Drosophila actin promoter or vector plasmid devoid of Sp1 coding sequences. Luciferase values were normalized to total protein concentrations.
with GATA-1 in mediating expression of the ␥-globin (48) and erythropoietin receptor genes (66) in erythroid cells; with PU.1 in mediating gene expression of myeloid-specific genes (67)(68); with YY1 (41); and with transcription factor E2F (69)). Currently we are investigating whether the protein(s) (NARP) interacting with E1 can interact with the protein(s) binding to E2. It is likely that NARP interacting with E1 or even the protein(s) binding to E3 function to recruit Sp1 to the CA box in E2, which in turn recruits other members of the transcription initiation complex (i.e. TFIID), resulting in effective transcription of the TATA-less ␤4 subunit gene as depicted in Fig. 8. Finally, it would be of interest to determine the role of the remaining CA boxes (see Fig. 1) in regulating ␤4 subunit gene expression.
In summary, we have identified three cis-acting sequences within the 5Ј-flanking region of the nACh receptor ␤4 subunit gene (Fig. 1). Here we report that Sp1 interacts with E2 and is probably required for ␤4 gene expression. Furthermore, the data also lead to the conclusion that NARP is probably not related to Sp1, because even in the context of a wild type E1 in pX1B4FHmut4, a NARP-binding site (E1) alone is not sufficient to direct ␤4 gene expression. Additionally, E1 did not compete with protein binding to radiolabeled E2 in EMSA (Fig.  5), further supporting the hypothesis that a protein(s) distinct from Sp1 binds to E1. These data contrast with a previous report indicating that Sp1 binds to a C/T sequence, similar to E1, within the 5Ј-flanking region of the nACh receptor ␣3 gene (33). That purified Sp1 protein was used in the ␣3 EMSA and DNase I footprinting experiments whereas SN17 cell nuclear extract was used in this report may explain these discrepancies. We are currently investigating whether the protein(s) binding to E1 interacts with Sp1 bound to the CA box in E2 in addition to investigating the functional significance of E3.