Regulation of transcription in the neuronal nicotinic receptor subunit gene cluster by a neuron-selective enhancer and ETS domain factors.

Expression of neurotransmitter receptors encoded by the nicotinic acetylcholine receptor (nAchR) subunit gene cluster depends on coexpression of the beta4, alpha3, and alpha5 subunits in certain kinds of neurons. One way in which coexpression might be achieved is through the regulation of promoters in the cluster by neuron-selective enhancers. The beta43' enhancer is located between the beta4 and alpha3 promoters and it directs cell type-specific expression in cell lines. It is not known, however, whether beta43' is active in neurons. Therefore, we assayed beta43' in dissociated rat sympathetic ganglia cultures, which contain nAchR-positive neurons as well as nAchR-negative non-neuronal cells. Reporters controlled by the alpha3 promoter and beta43' were expressed in a neuron-selective manner; greater than 90% and up to 100% of luciferase expression was detected in neurons. Neuron selectivity was maintained when beta43' was placed next to ubiquitously active viral promoters. In contrast, replacing beta43' with the SV40 enhancer eliminated neuron selectivity. The enhancer is composed of at least two separate but functionally interdependent elements, each of which interacts with a different type of ETS domain factor. These findings support a model in which beta43' controls neuronal expression of one or more genes in the cluster through interactions with a combination of ETS factors.

Neuronal differentiation depends on selective expression of certain genes such as those encoding ion channels, axon guidance molecules, neurotransmitter synthetic enzymes, transporters, and receptors in subsets of neurons. Selective expression is likely to occur through a combinatorial interaction of transcriptional activators and repressors with specific regulatory sequences in or near a particular gene (1). A rich assortment of transcriptional activator proteins are expressed in the nervous system. However, little is known about the sequences they interact with to direct expression of their target genes to certain kinds of neurons. Elucidating these interactions is likely to provide additional insight into the mechanisms underlying neuronal differentiation and development of neuronal diversity.
The family of neuronal nicotinic acetylcholine receptors (nAchR) 1 subunit genes is an example of genes whose members are expressed in different populations of neurons. The subunits encoded by these genes are assembled into different heteromeric excitatory ligand-gated ion channels (2). These neurotransmitter receptors mediate and modulate synaptic communication between a wide variety of central and peripheral neurons (3,4). Three of these genes are clustered in the order ␤4, ␣3, and ␣5 in the vertebrate genome (5). They are coexpressed in sympathetic neurons, adrenal chromaffin cells, and probably in some central neuron cell types (6 -8). The three encoded subunits are likely assembled together into at least one heteromeric neurotransmitter receptor subtype (9). Gene targeting suggests that the ␣3 and ␤4 genes are essential for the expression of the majority of nAChRs in autonomic neurons (10,11).
In contrast to the well known function of these genes in the peripheral nervous system, the mechanisms that direct synthesis of their RNAs to neurons are largely unknown. Using adrenal chromaffin cell-derived PC12 cells, which express the clustered genes, we identified an enhancer positioned within the ␤4 3Ј-untranslated exon. This enhancer, ␤43Ј, is responsible for cell-type specific activity in cell lines of a 2.8-kilobase pair fragment that constitutes the immediate ␣3 upstream region (12,13). ␤43Ј contains consensus ETS domain transcription factor-binding sites and can be activated in co-transfection assays by the ETS factor Pet-1 (14). Several ETS factors are expressed in the vertebrate nervous system during development and in the adult (14 -17), but their function in neurons has not yet been determined.
Although ␤43Ј is active in the PC12 neuroendocrine line it is not clear whether it is active in normal neurons that express the clustered genes. To investigate this, we assayed ␤43Ј activity in dissociated sympathetic ganglia cultures. Electroporation was used to introduce ␤43Ј-controlled luciferase reporters into these cultures and reporter expression was scored immunocytochemically with an antibody specific for luciferase protein.
We also investigated whether ␤43Ј directly interacts with ETS factors. The findings presented indicate that ␤43Ј is a neuronselective enhancer and is a target of ETS factors. Thus, ␤43Ј and its interaction with ETS factors are likely to be important for neuronal expression of the cluster.

Plasmids
Luciferase Reporters-Reporters for cell line transfections were prepared using pGL2 series vectors and those used for sympathetic ganglia culture transfections were prepared using pGL3 series vectors (Promega, Madison, WI). The ␤43Ј enhancer and ␣3 promoter reporters used here were derived from the Ϫ2732/ϩ47 region of the ␣3 gene (12,18,19). For most reporters, enhancer sequences were placed upstream of either the SV40 or ␣3 promoters using synthetic oligonucleotides and convenient restriction sites or by polymerase chain reaction. The ␣3 promoter used in this article is the minimal promoter, which includes Ϫ238/ϩ47 sequences (19). The ␤43Ј/␣3P reporter in which ␤43Ј is upstream of the ␣3 promoter was made by cloning ␤43Ј sequences 1-107 obtained from (107)␣3-luc (12) into pGL3b. The same enhancer fragment was cloned upstream of SV40P or AdMLP. The AdMLP construct contains sequences Ϫ55/ϩ10 of the adenovirus 2 major late promoter cloned upstream of the luciferase reporter. To prepare the SV40E/␣3P construct, an ␣3 promoter restriction fragment was cloned into pGL3c in place of the SV40p. Luciferase reporters containing four GAL4binding sites upstream of the ␣3 promoter (4XG-␣3P) or AdMLP (4XG-AdMLP) were prepared from pGL2-4XG (20) using convenient restriction sites.
␤43Ј reporters prepared using synthetic oligonucleotides were confirmed by dideoxy sequencing using Sequenase reagents according to the manufacturer's protocol (Amersham Pharmacia Biotech). All other constructs were confirmed by restriction digests. Oligonucleotides were synthesized by Life Technologies (Gaithersburg, MD) and plasmids for transfections were prepared using Qiagen reagents (Qiagen, Santa Clarita, CA). At least two different plasmid preparations were used for each construct and all constructs were tested in at least three independent experiments.

Cell Culture and Transfections
Cell Lines-The PC12 and Rat 2 cell lines were maintained in culture, transfected, and assayed for reporter activity as described previously (12).

Luciferase Reporter Assay in Dissociated SG Cultures
Transfected cells were fixed for 25-30 min at room temperature with 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4. Immunocytochemistry was carried out as described (22) using the following antibody dilutions: rabbit anti-luciferase (Accurate Chemicals, Westbury, NY) 1:1000, mouse anti-␤III tubulin (Sigma) 1:3000, fluorescein isothiocyanate-conjugated goat anti-rabbit IgG and rhodamine beta isothiocyanate-conjugated goat anti-mouse IgG (Cappel, Durham, NC) 1:200. Coverslips were mounted in phosphate-buffered saline/glycerol. Immunocytochemistry data were analyzed on a Nikon microscope. Cell counts were performed directly from immunostained coverslips. Photomicrographs were digitally processed from color slides using Adobe Photoshop.

Preparation of Recombinant Pet-1 and ETS-2 Protein
Pet-1-To express Pet-1 in bacteria, the Pet-1 cDNA from p73-7Z (14) was cloned into pGEM-3Z and modified to optimize the translation start site (23) creating an NcoI site. Convenient restriction sites were used to clone the Pet-1 coding sequence into pQE-210 (Qiagen) that had been modified to include an HA epitope tag downstream of the hexahistidine tag, creating plasmid pHH-Pet. Pet-1 was purified on Ni ϩ -NTA agarose (Qiagen) from XL1 Blue Escherichia coli under native conditions according to the manufacturer's protocol. After elution with 250 mM imidazole, Pet-1 was dialyzed into 20 mM HEPES, pH 7.9, 50% glycerol, 100 mM KCl, 0.2 mM EDTA, 0.1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride. Pet-1 was analyzed by SDS-polyacrylamide gel electrophoresis on 12% gels and Western blotted with antisera to Pet-1. Pet-1 purified from bacteria consists of a major band at about 33 kDa, and a minor band at 21.5 kDa. The 33-kDa band was recognized on Western blots with anti-Pet-1 antiserum and both the 33and 21.5-kDa bands were recognized by an antiserum that recognizes the conserved ETS domain (data not shown). Other minor bands from ϳ40 -35 kDa were also recognized by both antisera, suggesting the 33-kDa band may be a partially degraded form of Pet-1.

Electrophoretic Mobility Shift Assays (EMSA)
PC12 and Rat 2 nuclear extracts were prepared as described previously (12) and contained protease inhibitors, 1 mM phenylmethylsulfonyl fluoride, 4 g/ml leupeptin, 1 g/ml aprotinin, and 20 g/ml pepstatin. EMSAs were performed using double stranded oligonucleotides (Life Technologies) containing different segments of the enhancer and various competitor sequences (Table II). PEA3 and TCR␣ sequences were obtained from Refs. 24 and 25, respectively. Oligonucleotide probes were prepared by end labeling with [ 32 P]ATP and T4 DNA polynucleotide kinase (Roche Molecular Biochemicals). Binding reactions (20 l, 30 min, 4°C or room temperature) contained 100,000 cpm probe, 4 -5 g of nuclear extract, 20% glycerol, 2 g of dI-dC in 0.5 ϫ Tris glycine buffer (25 mM Tris, 190 mM glycine, 1 mM EDTA, pH 8.5). Binding reactions containing transcription factor antisera were performed in 20 mM HEPES, pH 7.9, 100 mM KCl, 0.5 mM dithiothreitol, and 1-2 l ETS-1/ETS-2 or Oct-1 antibody (Santa Cruz Biotechnologies Inc., Santa Cruz, CA) for 30 min at 4°C before adding probe. Oligonucleotide competitions were performed for 10 min with the indicated fold molar excess of unlabeled double stranded sequences (Table II) at 4°C or room temperature before addition of probe. Complexes were resolved on 4% polyacrylamide gels in 0.5 ϫ Tris glycine buffer at 4°C. For EMSA with recombinant Pet-1, approximately 200 ng of Pet-1 was incubated with competitors for 20 min at room temperature in 1 ϫ TGE, 30 mM KCl, 4 mM MgCl 2 , 5% glycerol, with 10.5 g of bovine serum albumin, and 1 g of dI-dC; total reaction volume was adjusted to 14 l with water. Radiolabeled probe (0.1 pmol) was added and the reactions further incubated 10 min at 37°C. Reactions were separated on 6% acrylamide, 2% glycerol, 0.5 ϫ TGE gels as described above and exposed 12-24 h to a PhosphorImager screen (Molecular Dynamics, Sunnyvale, CA).

DNase I Footprinting and Methylation Interference
DNase I footprinting was carried out with a footprinting probe that was derived from plasmid ␤43ЈFP, containing the sequence Ϫ2816/ Ϫ2545 of the ␣3 gene cloned into the XbaI site of pGEM-7Z (18). To label the antisense strand (GGA-containing), 10 -15 g of plasmid was cut with XbaI and NsiI. The ␤43Ј enhancer fragment was gel purified and the XbaI site filled in with Klenow (Roche Molecular Biochemicals) in the presence of [␣-32 P]dCTP (Amersham Pharmacia Biotech) followed by G-50 spun column chromatography. Binding reactions were carried out in 50 l of EMSA binding buffer with approximately 800 ng of recombinant Pet-1 prepared in bacteria. Radiolabeled probe (20 -40,000 cpm) was added and the reactions incubated 30 min at room temperature followed by 15 min at 37°C. Reactions were brought to room temperature, and DNase I added in 50 l of cofactor solution (12.5 mM CaCl 2 , 10 mM MgCl 2 ) for 2 min. Reactions were stopped with 100 l of stop solution (1% SDS, 30 mM EDTA, 200 mM NaCl, 100 g/ml yeast tRNA), extracted with phenol-chloroform and ethanol precipitated. Reactions were separated on denaturing (8 M urea) 6% polyacrylamide gels in 1 ϫ TBE for 2.5 h at 55 watts. Gels were fixed in 10% methanol, 10% acetic acid, dried, and exposed to a PhosphorImager screen for 1-4 days. Methylation interference assays were carried out exactly as described (26) with the same probe used for DNase I footprints.

␤43Ј Is a Neuron-selective Enhancer in Sympathetic Ganglia
Cultures-The activity of luciferase reporters in which ␤43Ј is placed immediately upstream of the ␣3 promoter (␤43Ј/␣3P) was determined immunocytochemically in neurons and nonneuronal cells of dissociated SG cultures. About 30% of the cells in these cultures were neurons based on their immunoreactivity with antibodies to neuron-specific tubulin; the remainder consist of GFAP-immunoreactive glial cells and a small number of contaminating fibroblasts. In experiments where the ␤43Ј/␣3P luciferase reporter was transfected between 93 and 100% of the luciferase-expressing cells were neurons ( Fig. 1, A, B, and E; Table I). To demonstrate that both neurons and non-neuronal cells were efficiently transfected, sister aliquots of cells in each experiment were transfected with a reporter driven by SV40 promoter and its own enhancer (SV40E/P). As expected for these viral elements, both neurons and non-neuronal cells expressed SV40PE/P driven luciferase ( Fig. 1, C and D) so that only about 30% of the luciferase-expressing cells were neurons ( Fig. 1E; Table I).
Analysis of the ␣3 promoter in several cell lines has suggested that it does not possess cell type-specific activity (12,18). However, promoters such as those for the avian ␤3 nAChR subunit gene promoter and the rat GluR2 subunit gene (27,28) show neuron-selective characteristics when assayed in primary neuronal cultures. Therefore, we wanted to assay the cell selective activity of the ␣3 promoter in SG cultures. The ␣3 promoter is only weakly active in primary cultures, as are the ubiquitously active AdMLP and the SV40 promoter, so that too few cells were positive by the immunocytochemistry assay to reliably characterize the expression pattern directed by any of these promoters on their own. Therefore, to determine the possible contribution of the minimal ␣3 promoter to neuronselective reporter expression, several additional constructs were tested. First, the ␣3 promoter was placed in cis to the SV40 enhancer (SV40E/␣3P). When SG cells were transfected with SV40E/␣3P, more non-neuronal cells than neurons expressed detectable luciferase, similar to what was seen with SV40E/P, although fewer luciferase-expressing cells were detected with SV40E/␣3P than with SV40E/P ( Fig. 2A; Table I). This indicates that the presence of the ␣3 promoter is not sufficient to direct a neuronal pattern of gene expression. A second approach to evaluate the ␣3 promoter was to amplify its activity using the herpes simplex virus VP16-activation domain. Four binding sites for the yeast GAL4 DNA-binding domain were cloned upstream of the ␣3 promoter or the AdMLP and these luciferase reporters were co-transfected into SG cultures with an effector construct encoding the chimeric activator GAL4-VP16. Using immunocytochemistry, we found relatively large numbers of luciferase positive cells in each experiment with either reporter when co-transfected along with the effector construct. However, the majority of these luciferase-expressing cells were non-neuronal ( Fig. 2A, Table I).
We then tested the ability of ␤43Ј to direct neuron-selective expression from heterologous promoters by placing ␤43Ј in each orientation upstream of either SV40 promoter or AdMLP and assessing reporter activity with immunocytochemistry. In SG cultures transfected with either reporter, the majority of the luciferase-expressing cells were neurons (Fig. 2B, Table I) indicating that the neuron-selective activity of ␤43Ј does not require the ␣3 promoter. The comparison between ␤43Ј/SV40P and SV40E/P is of particular interest because it demonstrates the differential cell type selectivity of the two enhancers in the context of the same promoter (Figs. 1E and 2; Table I). Thus ␤43Ј is a neuron-selective enhancer and this property is not  Table I) and then the percentages were averaged to produce mean Ϯ S.E. Schematics represent the cis elements present in each construct and arrows indicate orientation of ␤43Ј. dependent on the presence in cis of its natural promoter. ␤43Ј Is Composed of at Least Two Interacting cis Elements-␤43Ј was originally defined as being composed of two 37-bp direct repeats separated by a 6-bp spacer sequence (12). To determine which of these subregions are important for its activity, a series of reporters were prepared which contain different portions of the enhancer placed upstream of either the SV40 promoter or the ␣3 promoter (Fig. 3). Because of the relatively small number of neurons present in sympathetic ganglia we found it difficult to quantitate the activity of large numbers of reporters and to perform biochemical assays of protein/DNA interactions in these cells. Therefore, PC12 cells were chosen for these assays because: 1) each of the clustered genes are expressed in these cells whether or not they are differentiated with NGF (5). 2) These genes are expressed in undifferentiated PC12 cells at levels that are comparable to those in sympathetic neurons (6). 3) Our previous findings showed optimal ␤43Ј activity in undifferentiated PC12 cells relative to several other neural and non-neural lines (12). Thus, ␤43Ј activity correlates well with expression of the cluster in cell lines. 4) We have shown that the sites of ␣3 transcription initiation and their relative levels of usage are virtually identical in PC12 cells and sympathetic neurons (18,20), which supports similar mechanisms of ␣3 transcription in these cells types. 5) Data presented in Table I and Figs. 1, and 2 indicate that the cell type specific activity of ␤43Ј in PC12 cells is not simply an artifact of the transformed state of these cells.
We found that a single copy of either isolated repeat sequence was unable to stimulate reporter activity when cloned upstream of either ␣3 promoter or SV40 promoter (Fig. 4,  reporters 3 and 4). However, a spacer-containing fragment, which includes the spacer sequences plus a small part of each repeat (Fig. 3) retained about 40% of full enhancer activity (Fig.  4, reporter 2). This fragment was active in a cell type-specific manner as it showed no activity in Rat 2 cells, which do not express any of the clustered genes (12). These results indicate that ␤43Ј is dependent on interactions between factors bound to the spacer region and those bound to the repeats. This conclusion is consistent with the lack of any enhancer activity in an additional reporter in which a 4-bp insertion was placed in the spacer sequence (Fig. 4, reporter 5). The insertion most likely disrupted elements in the spacer region evident in reporter 2 and obligatory interactions between essential spacer and repeat elements. These findings suggest further that the activity of the repeat element(s) depends on interactions with elements in the spacer region.
␤43Ј Activity Depends on Two Different ETS Factor-binding Sites-To begin to determine the kinds of DNA binding factors that might control ␤43Ј activity we inspected its sequence for the presence of consensus transcription factor-binding sites. Interestingly, three different regions of the enhancer contain the sequence 5Ј-GGA(A/T), which is the core binding motif for ETS domain transcription factors (29) (Fig. 3). Moreover, each of these motifs is flanked by certain nucleotides that together with the core motif resemble particular kinds of ETS-binding sites. Each repeat has one of these ETS-like sites and both of them are most similar to PEA3-type ETS-binding sites (30). The third site is present in the spacer region and is similar to one of the ETS-binding sites present in the T cell receptor-␣ gene enhancer (TCR␣) (25). To determine whether any of these potential ETS sites are important for enhancer activity they were eliminated and then tested in PC12 reporter assays. The sites were eliminated in reporters by making clustered point substitutions in the ETS core sequence (Fig. 3), which are known to prevent ETS factor interactions (29). As shown in Fig.  5, elimination of either repeat site alone had no effect on enhancer activity. In contrast, a reporter in which both sites were eliminated resulted in a 75% loss of enhancer activity. Similarly, elimination of the spacer site resulted in a 70% loss of enhancer activity. We next asked whether or not these sequences are genuine ETS factor-binding sites by using recombinant ETS factors for mobility shift assays and footprinting. We used the PEA3binding protein, Pet-1 (14), to determine whether it could bind, in vitro, to the putative PEA3-type sites in the enhancer repeats. First, the ability of recombinant Pet-1 to bind was tested by EMSA. This recombinant Pet-1 bound to the repeat 1 (R1) probe (Table II, Fig. 6A) and the complex formed was completed by an excess of either R1 or R2, but not by modified R1 or R2 in which the ETS sites were eliminated (Fig. 6A, lanes 3 and 4, and data not shown). Identical EMSA results were obtained with in vitro translated Pet-1 (data not shown).
DNase I footprinting was then performed to determine the extent of the Pet-1-binding sites in the repeats, and to determine whether it can bind to the potential ETS site in the spacer region (Fig. 6B). This analysis revealed a footprint in each of the repeats, both of which include sequences within and flanking the GGA core motif. The Pet-1 footprint on the second repeat site is somewhat longer than that of the first repeat (sequence covered: R1, CAAGGAAGTG; R2, CAAGGAAAT-GACA). No footprint was detected in the spacer region. However, binding of Pet-1 to R1 and R2 induced two hypersensitive sites in the potential ETS GGAT motif located in the spacer. Pet-1 binding to the R1 and R2 ETS sites may therefore alter the conformation of the DNA within the spacer, which could affect binding of transcription factors or interactions among enhancer bound factors. Methylation interference was then used to determine which bases are near bound Pet-1. This analysis indicated that methylation of the guanines in the GGA core of R1 or R2 interfere with Pet-1 binding (Fig. 6C), similar to what has been found for binding of other ETS factors (29). Methylation of other guanines in the enhancer did not interfere with Pet-1 binding.

FIG. 2. ␤43 does not require the ␣3 promoter to direct neuronselective reporter expression in sympathetic ganglion cultures.
A and B, percent neuron selectivity is calculated as the percent luciferase positive cells that are neurons. This was calculated for each of several experiments (n, Table I) with the constructs shown, and then the percentages were averaged to produce mean Ϯ S.E. Similar results were obtained for reporters in which the orientation of ␤43Ј was flipped (Table I). In A, luciferase reporters were co-transfected with 3 g of the GAL4-VP16 effector construct. Schematics represent the cis elements present in each construct and arrows indicate orientation of ␤43Ј.
Although the spacer region was not able to bind recombinant Pet-1 protein its similarity to the TCR␣ enhancer ETS site suggested that it may still be an ETS site that interacts with a different kind of ETS factor. To investigate this, we tested whether recombinant ETS-2 could bind the spacer. This factor was chosen because it is expressed in PC12 cells and has a virtually identical DNA-binding domain to the TCR␣ enhancerbinding protein, ETS-1 (25,31). We found that ETS-2 prepared in vitro was able to bind specifically to a probe prepared from the spacer region (Table II). Binding was dependent on an intact ETS sequence as a spacer probe in which the ETS core motif sequence was eliminated did not bind ETS-2 (Fig. 6D).
To determine whether factors present in PC12 nuclear ex-tracts could discriminate between these different ETS sites, we performed competitive EMSAs with PC12 nuclear extracts and previously characterized ETS-binding sites present in other genes. R1 probe (Table II) reproducibly formed three specific complexes and a nonspecific complex when incubated with PC12 nuclear extract (Fig. 7A). Formation of each of the specific complexes depended on an intact repeat ETS site as competition with unlabeled R1 eliminated them but competition with GGA-substituted R1 did not (Fig. 7A). We also found that PEA3 ETS sequences could eliminate formation of both complexes but TCR␣ ETS sites could not (Fig. 7A). These data suggest that the repeats are able to bind ETS factors present in PC12 cell nuclear extracts and provide further evidence that these ETS sites interact specifically with PEA3-like binding proteins but not with TCR␣-like ETS factors. Incubation of spacer probe (Table II) with PC12 nuclear extracts led to the formation of five specific complexes and a nonspecific complex (Fig. 7B). In contrast to the specific complexes formed on the repeat probe, the specific complexes on the spacer could be eliminated by molar excess of TCR␣ competitor sequences but not by GGA-substituted TCR␣ competitor nor by molar excesses of PEA3 sequences (Fig. 7B). The formation of each of the specific complexes detected on the spacer segment depended on the ETS site as complex formation was nearly eliminated on a spacer probe carrying point substitutions in the ETS site (Fig. 7C). Two additional experiments further support the interaction of ETS factors in PC12 cells with the spacer site. First, incubation of an antibody raised against the DNA-binding domain of ETS-1 and ETS-2, which prevents DNA binding of these factors, resulted in elimination of most of the specific complexes formed on the spacer probe (Fig. 8A). In contrast, an antibody raised against the Oct-1 POU factor and used here as a negative control had no effect on complex formation (Fig. 8A). Second, expression of an ETS-2 dominant negative effector reduced the activity of intact ␤43Ј to the level of a reporter carrying a mutation in the spacer ETS site but had no effect on the activity of this mutated reporter or the SV40 promoter (Fig. 8B). DISCUSSION The ␤43Ј enhancer is being investigated as a regulatory element involved in the expression of the clustered neuronal nAchR genes in vertebrate neurons. It was originally identified in the PC12 neuroendocrine cell line. Assay of its activity in neural and non-neural lines indicated that it is most active in PC12 cells (12). To determine whether or not the enhancer might be relevant for expression of the clustered genes in neurons we used an electroporation protocol to measure ␤43Ј   FIG. 3. ␤43 enhancer. Top, location of the enhancer in the neuronal nAchR gene cluster is indicated as a crosshatched box in the ␤4 3Ј-untranslated region (open rectangle). Bottom, the sequence of the enhancer repeats (unidirectional horizontal arrows) and 6-bp spacer segment is shown. Sequences underlined indicate three ETS core motifs. The GGA core sequences are on the complementary strand. The last 9 bp on the right-hand side of the sequence are not required for enhancer activity but are present in some of the reporters assayed. Lowercase letters above the enhancer sequence show nucleotide substitutions used to make repeat ETS site mutations (gta) or the spacer ETS site mutation (ta). Vertical arrow shows position of a four base insertion (gtac) used to make reporter 5 assayed in Fig. 4. Sequence of enhancer deletions used in Fig. 4 are indicated by a numbered line below enhancer sequence. Enhancer sequences in reporter 2 and 3 are identical to those in spacer and R1 oligonucleotides, respectively (Table II).
FIG. 4. Activity of ␤43 fragments. The activities of reporter constructs containing the indicated portions of ␤43Ј were analyzed in PC12 (filled bars) and Rat 2 (cross-hatched bars) cell lines. The repeat was tested in both orientations with similar results and the spacer element was tested only in the correct orientation. ϩENHANCER/ϪENHANC-ER indicates the ratio of activity of the indicated enhancer containing reporter to that of the indicated promoter alone. At least three separate transfections were done in duplicate and were corrected for transfection efficiency with a co-transfected RSV-␤gal reporter to obtain mean Ϯ S.D. activity in dissociated sympathetic ganglia cultures. The main findings presented here are: 1) the enhancer is active in sympathetic neurons and has little or no activity in non-neuronal cells. 2) It is sufficient for neuron-selective reporter expression in sympathetic neurons.
3) The activity of ␤43Ј is dependent on ETS factor interactions.
Neuron-Selective Activity of ␤43Ј-The neuron-selective activity of ␤43Ј is supported by the observation that the SV40 promoter together with the SV40 enhancer did not show neuron-selective activity and actually seemed to be more active in non-neuronal cells. Therefore electroporation is capable of transfecting neurons and non-neuronal cells. Moreover, the preferential expression of luciferase in non-neuronal cells, upon amplification of the ␣3 promoter by GAL4-VP16, provides independent confirmation that neurons are not being favored by electroporation. Our conclusions are reinforced by the fact that we estimate that only 30% of the cells in our cultures are neurons, which should favor transfection of non-neuronal cells. The neuron-selective activity of ␤43Ј is not dependent on the presence in cis of its natural promoter as neuron selectivity was maintained when the enhancer was positioned next to two different ubiquitously active viral promoters. Thus ␤43Ј is a genuine neuron-selective enhancer as it shows cell type-selective activity not only in cell line models of neurons (12), but also in primary neuron cultures.
As the ␣3 minimal promoter does not have cell-type specific activity in cell lines (12) or neuron-selective activity in sympathetic ganglia cultures (Fig. 2) we hypothesize that expression of ␣3 and perhaps ␤4 and ␣5 is achieved through the action of neuron-selective enhancers. Several regulatory elements have been identified in the neuronal nAchR gene cluster (18,19,(32)(33)(34). However, ␤43Ј is currently the only known enhancer in or near the cluster that can control gene expression in a neuron-selective manner. Selectivity for neurons strongly suggests that ␤43Ј is biologically important for cell type-specific control of the cluster. Its location may allow ␤43Ј to function as a shared element in order to coordinate neuronal expression of ␤4 and ␣3. ␤43Ј may act as a pan-neuronal enhancer that must be silenced in certain neuronal populations. However, we have been unable to detect silencer elements, such as the neuronrestrictive silencer elements (35,36) or the GAP-43 repressive element (37) between the ␣3 promoter and ␤43Ј enhancer. Moreover, except for PC12 cells, ␤43Ј activity is low or absent in several neural and non-neural cell types (12). Further assays in other peripheral as well as central primary neuronal populations may help to determine whether or not silencing is required to limit ␤43Ј activity to appropriate neuronal cell types.
ETS Factor Interactions-The neuron-selective activity of ␤43Ј raises the question of which kinds of transcription factors are required for this property. As there are few neuron-selective enhancers that have been identified in vertebrates ␤43Ј offers an opportunity to investigate mechanisms through which transcription restricted to neurons can be achieved. Several lines of evidence indicate that ETS factor interactions are required for ␤43Ј activity. First, there are three good matches to ETS consensus binding sites present in the enhancer. One of these is located in the spacer region while the other two are located in each repeat. Mutation of the spacer site or combined mutation of the repeat sites eliminates 70 and 75% of ␤43Ј activity, respectively. Second, each of these sites can bind recombinant ETS factors in a specific manner. Third, complexes formed on these sites when incubated with PC12 nuclear extracts can be eliminated specifically by competition with genuine ETS-binding sites. TCR␣ sites but not PEA3 sites can eliminate complexes on the spacer site whereas the converse is found for the repeat sites. These results suggest that the spacer site binds an ETS factor that is different from that binding the  . Functional analysis of consensus ETS sites. Transient transfections were performed in PC12 cells with either the ␤43Ј enhancer or modified enhancers in which putative ETS-binding sites were eliminated in the first repeat alone, the second repeat alone, both repeats, or the spacer. Activity of the indicated reporters is presented relative to the activity of a reporter containing an intact enhancer, which is set at 100%. Also shown is the activity of the SV40 promoter alone determined in parallel.
Error bars indicate the mean Ϯ S.D. Results are from at least three separate transfections done in duplicate and are corrected for transfection efficiency with co-transfected RSV-␤gal reporter. Sequences used to eliminate the sites are described in the legend to Fig. 3. repeats. Fourth, the binding of most of the PC12 factors bound to the spacer site can be prevented with an antisera raised against a conserved portion of the ETS-1 and ETS-2 DNAbinding domain. Fifth, a dominant negative form of ETS-2 decreased enhancer activity to the level of ␤43Ј carrying a mutation in the spacer ETS site. This effect is specific to the ETS site as no effect was seen on the SV40 promoter or on ␤43Ј carrying a mutation in the spacer ETS site.
These data suggest that ETS-1, ETS-2, or both are the endogenous ETS factors that bind the spacer ETS site in PC12 cells. However, they are not likely to be sufficient for the cell-type specific activity of the enhancer as their patterns of expression (38) are dramatically different from that of the clustered nAchR genes. A more likely mechanism is that these ETS factors function in a combinatorial manner with other kinds of factors. Expression of some of these other factors may be largely restricted to PC12 cells and to neurons expressing the clustered genes. They may bind to other sites in the en- FIG. 7. Two kinds of ETS-like factors present in nuclear extracts interact with the enhancer. A, PEA3-like ETS factors present in PC12 nuclear extracts bind to ETS sites in the enhancer repeats. EMSAs were done with a radiolabeled R1 probe (all lanes), PC12 nuclear extracts (lanes 2-9), and the indicated competitors (Table II). Note that the maximum molar excess used for the TCR␣ competitor was twice more than that used for the PEA3 competitor. B, TCR␣ ETS sites but not PEA3 sites prevent ETS complex formation on spacer site. EMSAs were done with spacer probe (all lanes), PC12 nuclear extracts (lanes 2-7 and 9 -11), and the indicated competitors (Table II). Note that the maximum molar excess used for the PEA3 competitor was 2.5 times more than that used for the TCR␣ competitor. C, ETS site is required for formation of several complexes on spacer. Spacer probe (lanes 1 and 2) or ETS site-mutated spacer probe (lanes 3 and 4) were incubated with PC12 nuclear extracts (lanes 2 and 4) or run free (lanes 1 and 3). Arrowheads to the left of capital letters in panels A-C indicate specific protein/DNA interactions. N, nonspecific interaction.

FIG. 6. Three ETS-binding sites are present in the enhancer.
A-C, Pet-1 binds the putative repeat ETS sites. A, EMSA was carried out with an R1 radiolabeled probe and the indicated competitors used at ϫ 400. Arrows indicate the major protein-DNA complexes formed. The complexes are more prominent in lane 4 due to the presence of competitor oligonucleotides that routinely produced this effect. B, DNase I footprint of Pet-1 bound to R2 and R1. Lines indicate the protected sequences shown at the right. Pet-1 does not bind to the GGAT motif present in spacer, but introduces hypersensitive sites in this region (arrows). This experiment is representative of at least five replicates. The open circles indicate guanines whose methylation (C) interferes with binding. C, methylation of guanines in the R1 and R2 ETS core motifs prevent Pet-1 binding. The lines indicate methylated guanines that are under-represented in the bound fraction. Similar results were obtained in four experiments. D, ETS-2 binds the putative spacer ETS site. In vitro translated ETS-2 protein was incubated with enhancer spacer probe (Table II) hancer or are recruited to the enhancer through protein-protein interactions. Assembly of a specific combination of these factors on the enhancer would result in selective transcription in neurons. The elimination of nearly all of the PC12 specific complexes formed on the spacer when its ETS site is destroyed (Fig. 7C) is consistent with a combination of ␤43Ј bound factors. Additional evidence supporting a combinatorial mechanism is that neither enhancer repeat in isolation has activity and insertion of sequences into the spacer region completely eliminates enhancer activity. Only when placed in cis to the spacer segment is the contribution of the repeats evident.
Pet-1 is a candidate for the endogenous PEA3-type ETS factor that interacts with the enhancer repeats in PC12 cells and adrenal chromaffin cells. Evidence in support of Pet-1 is that Pet-1 RNA is expressed in PC12 cells and adrenal medulla but few other cell types, Pet-1 can bind the repeats, and it can activate transcription, albeit weakly, in an enhancer-dependent manner (14). However, despite the presence of quantities of Pet-1 RNA comparable to that of ␤-actin in PC12 cells, several antisera raised against Pet-1 have failed to supershift PC12 FIG. 8. ETS-1/2 or a related factor interacts with the spacer ETS site. A, PC12 nuclear extracts were incubated with spacer probe and either an antibody to the ETS-1/ETS-2 DNA-binding domain, or an Oct-1 antibody. Complexes A-E are eliminated when reactions are preincubated with the ETS antibody. B, expression of the ETS-2 DNA-binding domain decreases ␤43Ј enhancer activity through an interaction with the spacer ETS site. Transient co-transfections were done in PC12 cells with 10 g of the indicated luciferase reporters and either 5 g of a dominant negative ETS-2 expression vector, CGS ETS-2 281-445 , (ϩ), or 5 g of an empty cytomegalovirus promoter containing plasmid (Ϫ). Data are presented relative to the activity of a reporter containing an intact enhancer obtained in the absence of the ETS-2 expression vector. The modified (ϫ) enhancer construct contains a substitution in which the GGAT ETS core sequence within the spacer was changed to GGta. Activities are corrected for transfection efficiency by co-transfected RSV-␤gal. Data are from three separate experiments where error bars represent the mean Ϯ S.D. complexes formed on the repeat. 2 On the other hand, neither the ETS1/2 antibody nor any of several other commercially available ETS factor-specific antisera could supershift or compete with probe for the repeat ETS factor complex. 2 Nonetheless, our data suggest that transcription directed by ␤43Ј is achieved, at least in part, through the interaction of different kinds of ETS domain factors.
ETS domain factors are primarily known for their roles in the development of different hematopoietic lineages as well as for roles in oncogenesis (29,39). Interestingly, expression of several of these genes have been detected in the nervous system by Northern blotting. However, few have been localized to specific populations of neurons and none of these have been shown to be expressed in sympathetic neurons (15)(16)(17). This general lack of information makes it difficult to identify specific ETS factor candidates that may interact with ␤43Ј in neurons. As some of these factors have highly restricted expression patterns (16,17,40) perhaps different ETS factors in the different neuronal populations control expression of the clustered genes.
␤43Ј is one of few cis elements present within genes expressed in neurons that have been shown to interact with ETS factors.Experimental support for a role of ETS factors have been obtained for the synapsin II, peripherin, and presenilin I genes (41)(42)(43). We reported recently that expression of Pet-1 in the brain is restricted to hindbrain 5-HT neurons and can bind specifically to ETS sites present in the human and mouse serotonin transporter, tryptophan hydroxylase, and 5-HT1a receptor genes (17). An interesting hypothesis currently being tested is that Pet-1 or related ETS factors regulate genes required for expression of two different neurotransmitter systems in different kinds of neurons.
Our results identify ETS factor interactions as a regulatory feature common to neuronal and muscle nAchR gene expression. Synapse-specific regulation of muscle nAchR ␦ and ⑀ subunit genes by the extracellular signaling factor neuregulin depends on ETS-binding sites in their promoters (44 -46). The nuclear factors interacting with these sites and the targets of the neuregulin-activated intracellular signaling pathway are believed to be the ETS factor GABP␣ and its genetically unrelated dimerization partner GABP␤ (44 -46).
In summary, we have shown that ␤43Ј is a neuron-selective enhancer controlled by ETS factors. Our findings support ␤43Ј as biologically relevant for expression of the cluster in neurons. Its interactions with ETS factors may be part of a mechanism that controls the differentiation of certain neuronal cholinergic synapses. Molecular genetics approaches in the mouse should help to reveal the biological role of these interactions.