Introduction

Summary Introduction Materials & Methods Results Discussion References

Nicotinic acetylcholine receptors are ligand-gated ion channels expressed at the neuromuscular junction, in autonomic ganglia, and in several areas of the vertebrate CNS. They are also present on adrenal chromaffin cells, where they control the release of catecholamines. Like their muscular counterpart, neuronal-type nicotinic receptors have a pentameric stoichiometry and are composed of ligand-binding and structural subunits, which are encoded by a large family of genes homologous to those encoding the muscle isoforms. Recombinant DNA technology and heterologous expression in Xenopus laevis oocytes have resulted in the identification of eight ligand-binding subunits (2-9) and three structural subunits (2-4) that can combine to form different receptor subtypes with distinctive pharmacological and electrophysiological properties. More precisely, some agonist-binding subunits (7-9) have been shown to work as homo-oligomeric channels, whereas 2-4 isoforms must be coexpressed with at least one structural subunit to generate functional molecules (for reviews, see Refs. 1 and 2). In vivo, the structure of native receptors is likely to be more complex, with the possibility that more than one type of agonist binding and/or structural subunit assembles into the same receptor molecule with multiple stoichiometries (2, 3). It implies that a stringent qualitative and quantitative control on the expression of each subunit must be exerted to generate specific receptor subtypes with different functions in synaptic pathways. In situ hybridization and immunocytochemical studies corroborate this idea, demonstrating that certain subunits are preferentially or exclusively expressed in the CNS or in autonomic ganglia; in the former case, the expression can be restricted to very few brain structures (4, 5).

The regional distribution pattern of each subunit seems to be under the control of fine regulatory mechanisms that start to operate early during development; the expression of some subunits is often precocious and sometimes coincides with the early events of neuronal differentiation. Interestingly, some nicotinic isoforms are transiently present in certain neural structures, disappearing during the perinatal period (5). Although the the majority of the anatomic organization of the neuronal cholinergic-nicotinic system seems to be defined during ontogenesis, the expression levels of certain subunits undergo important modifications in postnatal life, as in those areas of the CNS that show structural rearrangement and progressive neurochemical maturation during the first 21 days after birth (6).

These highly regulated spatial and temporal expression patterns are likely to rely on sophisticated genetic mechanisms that can be partially approached through the characterization of the regulatory sequences that control the transcription of the different neuronal nicotinic subunit genes. Although information for some chicken and rat genes is available (7-15), studies on the human genes have not been performed. In the current study, we describe for the first time a human promoter controlling the expression of the 3 nicotinic subunit gene.¹

Several lines of evidence have demonstrated that autonomic ganglia are not simply relay stations but rather display a high degree of integrative activity. In light of the relevant role of the 3 nicotinic subunit in ganglionic transmission (1, 2), the comprehension of the mechanisms that govern its transcription may also help to obtain new insights into the functions of the autonomic nervous system, especially concerning adaptive responses to environmental cues.

	Materials and Methods

Summary Introduction Materials & Methods Results Discussion References

Isolation of the 5-flanking region of the human 3 nicotinic subunit gene. The genomic clone c16, which has been previously shown to contain coding sequences of both 4 and 3 genes (16), was digested with ApaI and analyzed by Southern blotting. A 5.1-kb ApaI fragment was recognized by both an ApaI/EcoRI cDNA probe, corresponding to part of the 3 UT of the 4 subunit (17), and an EcoRI/NcoI cDNA probe, corresponding to the 5 UT of the 3 subunit (18). This fragment was purified, ligated into the ApaI site of pBluescript II KS⁺ (Stratagene, La Jolla, CA), and characterized by restriction analysis and sequencing (TAQuence, version 2.0; United States Biochemical, Cleveland, OH).

RNA preparation. Total RNA were isolated from different cell lines by the use of RNAfast-II (Molecular Systems, San Diego, CA) according to the manufacturer's instructions. Briefly, ~10⁸ cells were collected by centrifugation and lysed with a solution containing guanidine salts and phenol. RNA was extracted with chloroform and purified on the RNA-binding resin. Poly(A)⁺ mRNA was prepared with Oligotex-dT (Qiagen, Studio City, CA) according to the manufacturer's instructions, with minor modifications. Total and polyadenylate RNAs were quantified with spectrophotometry.

RNase protection analysis. A 201-bp fragment encoding part of the cytoplasmic domain of the human 3 nicotinic subunit (nucleotides 976-1176 according to Ref. 18) and a 338-bp BglII/SfiI fragment, corresponding to the region immediately upstream of the 3 start codon (Fig. 1) were subcloned in pBluescript II KS⁺. On linearization with appropriate restriction enzymes, constructs were transcribed in vitro by the use of T3 or T7 polymerase (MAXIscript; Ambion, Austin, TX) to obtain [-³²P]UTP-labeled antisense RNA. Because of the presence of part of the pBluescript II KS⁺ polylinker, the full-length riboprobes were 262 and 391 bp. A 316-bp fragment of the human GAPDH gene (pTRI-GAPDH-human antisense control template; Ambion), in vitro transcribed by T7 polymerase, was used to check the quality of the different RNAs. The full length riboprobe was 404 bp, including part of the vector polylinker. The full-length RNA probes were purified from 6% acrylamide gels containing 8 M urea. RNase protection assays were carried out with the RPA II kit (Ambion) with some modifications. The 262- and 391-bp purified RNA probes were hybridized overnight with 3 µg of poly(A)⁺ at 42° and 60°, respectively. The GAPDH RNA probe was hybridized overnight with 0.5 µg of poly(A)⁺ at 42°. Single-stranded RNA was digested with 5 units/ml RNase A and 200 units/ml RNase T₁ for 1 hr. Yeast RNA (10 µg) was used instead of cellular RNA as a negative control. The products of the RNase protection assay were analyzed on 6% acrylamide gels containing 8 M urea. DNA sequence ladders were used as molecular weight markers.

View larger version (41K): [in this window] [in a new window]   Fig. 1.   Human genomic cluster containing the three nicotinic subunits (4, 3, and 5) and nucleotide sequence of the 1-kb KpnI/NcoI fragment corresponding to the 5-flanking region of the human 3 gene. Line at the top, gene segment contained in the c16 phage clone. Arrows in the genomic cluster, direction of transcription; enlargement below, 5.1-kb cloned region between 4 and 3 plus some restriction sites (A, ApaI; K, KpnI; N, NcoI); striped region, 5 UT of the cloned 3 cDNA in the 5.1-kb fragment; filled region, part of the coding region of the human 3 gene in the 5.1-kb fragment. The nucleotide sequence of the KpnI/NcoI fragment is shown: positions are numbered from the initiator codon. Arrows, transcriptional start sites identified by RNase Protection (see legend to Fig. 3); *, First nucleotide at the 5 end, previously identified by cDNA cloning. Relevant restriction sites used for subcloning and putative DNA binding sites for the indicated transcription factors are shown. Shaded region, Alu sequence; overlined, Alu core repeats; superscript dashes, reducer sequences; AP1, activator protein-1; AP-2, activator protein-2; EGR1, ; GRE, glucocorticoid response element.

Construction of human 3 promoter-luciferase fusion plasmids. All restriction enzymes were purchased from Promega (Madison, WI). Different segments of the 5-flanking region of human 3 were subcloned into the pGL3-basic vector (Promega), upstream of the luciferase reporter gene (Figs. 1 and 4-6). When not differently identified, the constructs are named with a number, identifying the length of the subcloned fragment, followed by the letters indicating the restriction sites into which they were ligated (K = KpnI, B = BglII, W = BsiWI, A = AccIII, N = NcoI, S = SacII, P = PstI, and X = BstXI). The 4.3 KN construct was generated as follows: the 5.1-kb ApaI fragment in pBluescript II KS⁺ was separately digested with NcoI and KpnI or KpnI alone. The digestions resulted in the excision of a 1-kb KpnI/NcoI fragment and a 3.3-kb KpnI fragment that was generated because of the presence of a KpnI site in the polylinker of pBluescript II KS⁺, immediately upstream of the ApaI site. The two inserts were gel purified by the use of Qiaex (Qiagen) and ligated by three-way ligation into the NcoI and KpnI sites of pGL3-basic. The correct orientation and the identity of the two inserts were verified by restriction analysis and DNA sequencing. The resulting construct, 4.3 KN, contains the 3 UT of 4, the 5 UT of 3, and the intervening region (Fig. 1). 5 and 3 deletions (Fig. 4) of this construct were obtained as follows: 1 KN was generated by digestion of the 4.3 KN plasmid with KpnI, followed by religation; and the 0.35 BN construct was obtained by subcloning the purified 0.35-kb BglII/NcoI fragment into the corresponding restriction sites of pGL3-basic. The 0.2 AN plasmid was obtained by subcloning the purified AccIII/NcoI fragment into the XmaI and NcoI sites of pGL3. The constructs used to investigate the functional role of the Alu sequence were obtained as follows: the 0.8 PN derives from the 1 KN plasmid on digestion with KpnI and PstI; after gel purification, the extremities were blunted by T4 DNA polymerase and religated by T4 DNA ligase. Alu derives from the 1 KN plasmid, which was digested with PstI and BstXI, gel purified, blunt ended, and religated. The 0.6 XN construct again derives from the 1 KN plasmid, digested with KpnI and BstXI, gel purified, blunt ended, and religated. The 0.6r (r = reducer) represents a 30-bp 5 deletion of the previous construct that was obtained by digesting the 0.6 XN plasmid with SspI. A band of 4.5 kb was gel purified and digested with NcoI; after gel purification, the resulting SspI/NcoI fragment was cloned into the SmaI and NcoI sites of pGL3-basic. Notably, in this set of constructs, the 5 UT of the luciferase was removed, and the coding region of the reporter gene was directly fused to the 5 UT of the 3 nicotinic subunit at the NcoI site. The natural context for initiation of translation was not affected in these constructs. The 0.16 BA construct was obtained by digestion of the 0.35 BN plasmid with AccIII and XmaI; the purified XmaI/AccIII fragment was then subcloned into the XmaI/SmaI site of pGL3 basic, in appropriate orientation. The 3.3 K and 0.6 KB constructs were generated by subcloning the purified fragments into the polylinker of pGL3-basic. The plasmids derived from the 0.35 KN construct, bearing mutations in the 3 end of the promoter, were obtained as follows: NF-1 was generated by replacing the SfiI/NcoI fragment with a double-strand oligonucleotide bearing two point mutations in the NF-1 site. AP-2 contains a 3-bp deletion in the downstream AP-2 sequence, obtained through SfiI digestion, followed by removal of the overhangs and religation. SacII contains a deletion of the 150-bp SacII fragment. NFB contains a 4-bp deletion in the NFB sequence; the mutation was created in the SacII fragment and subcloned in a derivative of pGEM-5Z (Promega), which lacks BanII restriction sites. The plasmid was digested with BanII, blunt ended, and religated. The mutated SacII fragment was subcloned in SacII, in the appropriate orientation. MED-1 contains a 4-bp mutation in the 6-bp consensus sequence of MED-1. The 250-bp SacI/BsiEI fragment contained in the 0.35-kb promoter was gel purified. A double-strand oligonucleotide, corresponding to the 32-bp BsiEI/SacII region of the promoter, was generated bearing the MED-1 mutation. The oligonucleotide and the 250-bp fragment were cloned in pBluescript II KS⁺ by three-way ligation. The SacII fragment was gel purified and subcloned in SacII plasmid. AP-2-NFB was obtained from NFB by applying the same procedures used to obtain AP-2. AP-2-SacII was obtained from AP-2 by removing the SacII fragment. pGL3 control + 0.2 kb was generated as follows: pGL3 control was digested with HindIII, blunt ended by T4 DNA Pol, and digested again with NcoI to remove the DNA region specifying the 5 UT of the luciferase gene. The 0.2 AN was digested with NheI, blunt ended, and digested again with NcoI to excide the 0.2 AccIII/NcoI fragment. The fragment was gel purified and ligated blunt/NcoI in the pGL3 control plasmid previously prepared as described. All mutations were confirmed by DNA sequencing. The RSV--galactosidase plasmid, used to normalize for transfection efficiency, was obtained from pGL3-basic by replacing the luciferase gene with the -galactosidase gene, derived from pNASS (Clontech, Palo Alto, CA). The RSV promoter was obtained from pRc/RSV (InVitrogen, San Diego. CA) and subcloned into the polylinker of pGL3-basic, upstream of the reporter gene.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Functional delineation of the human 3 nicotinic subunit gene promoter in SY5Y and SK-N-BE human neuroblastoma cell lines. Left, chimeric constructs, in which different parts of the 3 5 regulatory region are fused to the luciferase (LUC) reporter gene. Note that the 3.3-kb KpnI fragment is not to scale. Open box, luciferase gene coding region; shaded portion of box, DNA region specifying the 5 UT of the luciferase. Note that this region, which is 90 bp, is not to scale. In the other constructs, the entire DNA region specifying the 5 UT of the 3 mRNA was directly fused to the coding sequence of the reporter gene. In the pGL3 control plasmid, the SV40 enhancer is located downstream of the luciferase gene and is not indicated. Bars, transcriptional activity of the construct calculated as described in Materials and Methods and expressed as fold increase over the transcriptional activity of the promoterless plasmid pGL3-basic. The standard deviation is indicated. Each construct was tested at least three times in triplicate with two independent plasmid preparations.

Cell lines and cultures. The human neuroblastoma cell lines SY5Y and SK-N-BE and the human rhabdomyosarcoma cell line TE671 (CRL-8805, American Type Culture Collection, Rockville, MD) were grown in RPMI 1640, 10% fetal calf serum, 50 units/ml penicillin, 50 mg/ml streptomycin, and 2 mM L-glutamine. HeLa cells were grown in Dulbecco's modified Eagle's medium, 10% fetal calf serum, 50 units/ml penicillin, 50 mg/ml streptomycin, 2 mM L-glutamine, and 110 µg/ml sodium pyruvate.

Transient transfections. Plasmid DNA was purified with Qiagen columns and transfected into logarithmically growing cells using a Gene Pulser electroporator (BioRad, Hercules, CA).

Luciferase and -galactosidase assays. Cells were harvested, washed twice in phosphate-buffered saline, and lysed in the Reporter Lysis Buffer (Promega) for 15 min at room temperature. After a brief centrifugation to remove cellular debris, 10 µg of cellular extract in a final volume of 20 µl was added to 100 µl of Luciferase Assay Reaction (Promega) and tested for luciferase activity with a Berthold Lumat LB 9501 luminometer for 60 sec. In parallel, 10 µg of the same extracts were processed for -galactosidase expression by using the Luminescent -Galactosidase Detection Kit (Clontech), following the supplied protocol. Samples were previously heat treated at 50° for 1 hr to destroy any endogenous -galactosidase activity. This treatment does not affect the bacterial -galactosidase expressed by the RSV--Gal plasmid (19); this procedure was also verified for each cell line that we used in our laboratory. -Galactosidase activity was determined with the luminometer for 5 sec. In both luciferase and -galactosidase assays, the background activities, which were measured in the absence of cellular extract, were subtracted. The protein content of the cellular extracts was measured with a BCA Protein Assay (Pierce Chemical, Rockford, IL).

Analysis of the data obtained by transient transfections. For each construct, the values of luciferase (expressed as relative luminescent units) that were obtained in the different experiments were plotted versus the corresponding values of -galactosidase (also expressed as relative luminescent units). Linear correlations were obtained with correlation coefficients ranging from 0.75 and 0.99. The transcriptional activity of each construct was defined as the slope of the straight line and was expressed as fold increase over the transcriptional activity of the promoterless plasmid pGL3-basic. The statistical significance of the differences in the transcriptional activities among constructs was assessed by analysis of variance (F < 0.05).

Computer-assisted analysis. The presence of putative transcription factor binding sites (Fig. 1) was evaluated with the MacPattern program (Macintosh) in conjunction with the EMBL Transcription Factor Database. Statistical analysis was carried with StatWorks software (Macintosh).

	Results

Summary Introduction Materials & Methods Results Discussion References

Isolation of the 5-flanking region of the human 3 nicotinic subunit gene. The human 4, 3, and 5 nicotinic subunit genes are located on chromosome 15, forming a genomic cluster that is conserved across species (Fig. 1). The genomic clone c16, which has been shown to contain coding sequences of both 4 and 3 genes (16), was analyzed by restriction digestions, followed by Southern blotting and hybridization to specific probes. c16 contains a 5.1-kb ApaI fragment that was recognized by two cDNA probes corresponding to the 3 UT of the 4 subunit and the 5 UT of the 3 subunit (data not shown); this result indicated that the entire genomic region between the 4 and 3 genes was contained in the ApaI/ApaI fragment. The cDNA sequencing of the human 3 subunit has been demonstrated to contain the ATG start codon in an NcoI site (18). After a preliminary restriction analysis, a 1-kb KpnI/NcoI fragment was purified from the 5.1-kb ApaI fragment (Fig. 1), subcloned into the pGL3-basic vector. and sequenced through the use of GLprimer 2 (Promega). This genomic fragment overlapped the 5 UT of the 3 cDNA by 201 bp (Fig. 1). A 3.3-kb ApaI/KpnI fragment and a 0.8-kb NcoI/ApaI fragment located upstream and downstream of the 1-kb KpnI/NcoI segment, respectively, were also characterized by DNA sequencing and proved to contain the 3 UT of the 4 subunit and part of the coding region of 3 subunit gene, including the signal peptide encoding sequence, respectively. These results confirmed that the ApaI fragment contained the whole genomic region located between the 4 stop codon and the 3 start codon. The entire sequence of the 1-kb KpnI/NcoI fragment, which was likely to contain the 3 gene promoter, was determined (Fig. 1). The results of the sequence analysis revealed that the 1-kb KpnI/NcoI fragment can be divided by the BglII restriction site into two subregions with different nucleotide composition: a 0.35-kb BglII/NcoI segment, proximal to the start codon, which is very rich in GC (75%), and a 0.6-kb KpnI/BglII upstream segment with a 52% GC content. The latter subregion contains an Alu repeat that displays 85% identity with a canonical Alu sequence (Fig. 1).

Detection of 3 transcripts in human neuroblastoma cell lines by RNase protection assay. To verify that SY5Y and SK-N-BE human neuroblastoma cell lines expressed 3 transcripts in our culture conditions, we carried out an RNase protection assay (Fig. 2A). We used a cRNA antisense probe, complementary to part of the region coding for the cytoplasmic domain of 3, which is the least conserved part among the different subunits of the nicotinic receptor family. A protected band of the expected size was detected in neuroblastoma cell lines and was prominent in SY5Y cells. A much less intense band, a few nucleotides shorter, was also detected and was probably the result of RNase "nibbling." The negative control in which yeast RNA was substituted for cellular RNA revealed no bands. No signals were detected in the non-neuronal cell lines, indicating that at least in HeLa and TE671 cells, the tissue specificity of the 3 expression is conserved even in tissue culture conditions. The same RNA preparations from the four cell lines were also tested by using a GAPDH probe (Fig. 2B). A strong band, with the correct size and the same intensity, was obtained in all RNA samples but not in the negative controls, confirming that the differences in the 3 expression between neuronal and non-neuronal cell lines were not due to RNA degradation.

View larger version (26K): [in this window] [in a new window]   Fig. 2.   Expression of the human 3 mRNA in neuronal (SY-5Y and SK-N-BE) and non-neuronal cell lines (HeLa, TE671). A, A 262-bp [-32P]UTP-labeled antisense riboprobe corresponding to 201 bp of the region encoding part of the cytoplasmic domain of human 3 nicotinic subunit and to 61 bp of the pBluescript II KS+ polylinker was hybridized to mRNA purified from the indicated cell lines. 1 and 2, undigested probe and negative control, respectively, in which the probe was incubated with yeast RNA and then digested with RNases. B, A 404-bp [-32P]UTP-labeled antisense riboprobe corresponding to 316 bp of the ubiquitous gene GAPDH and to 88 bp of the vector polylinker was hybridized to the same mRNA preparations used in A. 1 and 2, undigested probe and negative control, respectively, in which the probe was incubated with yeast RNA and then digested with RNases. Numbers on the left, size of the full-length antisense riboprobes; numbers on the right, size of the protected bands.

Mapping of the transcription start sites of the human 3 nicotinic subunit. To map the 5 termini of the human 3 mRNA, we carried out an RNase protection assay by using as probe a 338-bp BglII/SfiI fragment that at its 3 end overlaps with the 5 UT of the 3 cDNA (Fig. 1). Several transcription start points were identified in the two neuronal cell lines, with different preferential use (Fig. 3). No bands were detected in the negative control or in the non-neuronal cell line RNA. All of the transcription start sites except one were clustered in a region of ~80 bp, which is almost completely overlapping the 5 UT of the 3 cDNA (Fig. 1). The origin of the 341 transcription start site (Figs. 3 and 1), which is prominent in SK-N-BE cells, is unclear. Its presence has been also confirmed by using a 280-bp SspI/BsiWI riboprobe (Fig. 1 and data not shown), and an investigation of whether it reflects the existence of an additional upstream promoter is under way.

View larger version (29K): [in this window] [in a new window]   Fig. 3.   Mapping of the transcription start sites of the human 3 nicotinic subunit. A 391-bp [-32P]UTP-labeled antisense riboprobe corresponding to the 326-bp BglII/SfiI fragment of the 5-flanking region (see Fig. 2) and to 65 bp of the pBluescript II KS+ polylinker was hybridized to mRNA purified from the indicated cell lines. Numbers on the right, size of the different protected bands; number on the left, full-length antisense riboprobe. 1 and 2, undigested probe and negative control, respectively, in which the probe was incubated with yeast RNA and then digested with RNases.

Functional delineation of the human 3 nicotinic subunit gene promoter. To localize the region necessary and sufficient for promoter activity, we generated chimeric constructs containing different genomic fragments of the 5-flanking region of the human 3 nicotinic subunit gene fused to the luciferase reporter gene. Because both the ATG start codons of the human 3 subunit and the luciferase reporter gene are embedded in an NcoI restriction site, we were able to generate chimeric constructs in which different segments of the 5 regulatory region were directly fused to the luciferase coding sequence. The different constructs were transfected into the human neuroblastoma cell lines SY5Y and SK-N-BE along with an RSV--galactosidase plasmid as internal control for transfection efficiency. After 48 hr, luciferase and -galactosidase activities were measured. Similar but not identical results were obtained in the two cell lines (Fig. 4). In both neuroblastomas, the 4.3 KN construct, containing the whole intergenic region between 4 and 3, displayed a consistent promoter activity, 20 times over the background, evaluated as the luciferase expression driven by the promoterless pGL3 basic plasmid and 40% of the activity of pGL3 control, which contains the well-characterized SV40 promoter/enhancer. When a 5 deletion of 3.3 kb was carried out, the resulting 1 KN plasmid showed a 50% decrease in the expression of the reporter gene only in SY5Y cells, whereas its transcriptional activity did not change in SK-N-BE cells. The 3.3-kb KpnI deleted fragment displayed negligible activity when tested alone. The 1 KN construct was further dissected by removal of the 0.6-kb KpnI/BglII 5 region to generate the 0.35 BN plasmid. The promoter strength of this construct was as much as 3.5 times more than that of the 1 KN construct in both cell lines, indicating the presence of a negative element in the 0.6-kb deleted region. The 0.6 KB construct showed only basal activity.

Functional characterization of the AccIII/NcoI region specifying the 5 UT of the 3 mRNA. Our data indicated that the AccIII/NcoI fragment, corresponding to the 3 end of the BglII/NcoI 3 promoter region and specifying most of the 5 UT of the 3 mRNA, had a relevant positive effect on gene expression. To test whether this DNA region could also exert its function in conjunction with an heterologous promoter, we engineered pGL3 control plasmid, which contains the SV40 promoter/enhancer, by replacing the DNA region specifying the 5 UT of the luciferase gene with the AccIII/NcoI fragment of the 3 gene. The construct (pGL3 control + 0.2 kb) was transfected in SY5Y neuroblastoma cells, and the resulting luciferase activity was compared with that determined by the wild-type pGL3 control. Only a very modest increase in the ability to drive the expression of the reporter gene was observed with pGL3 control + 0.2 kb [+30%, F = 0.163 (not statistically significant)] (Fig. 5), suggesting that the AccIII/NcoI region could properly work only in the context of the human 3 promoter, where its presence determined an increase in luciferase activity of 400% (Figs. 4 and 5). This result also made unlikely the hypothesis that the AccIII/NcoI fragment, being incorporated into the luciferase mRNA, could increase its stability or translation. Indeed, if the difference in luciferase activity between the 0.35 BN and the 0.16 BA human 3 promoter constructs (400%) had been due to a post-transcriptional effect of the AccIII/NcoI region, we could have reasonably expected a similar difference between pGL3 control and pGL3 control + 0.2 kb. On the other hand, it was possible that the functional properties of the AccIII/NcoI fragment relied on the ability to bind specific transcription factors; actually, this DNA region contains putative cis-acting elements for the transcription factors NFB, AP-2, NF-1, and MED-1. To determine whether these putative cis-acting elements were functional, we carried out a mutation analysis (Fig. 5). Although a mutation in the NF-1 DNA site did not affect promoter activity (not shown), mutants in which the AP-2 or the NFB putative binding sites were destroyed displayed a 50% decrease in the expression of the reporter gene. Therefore, we generated the construct AP-2-NFB, with mutations in both sites. The plasmid showed a further 50% decrease in the promoter activity compared with the singly mutated parental plasmids, having a promoter strength identical to that of the 0.16-kb construct. These data indicated that the putative cis-acting elements for AP-2 and NFB were actually functional and could account for the properties of the AccIII/NcoI fragment. However, we also identified in this fragment a recently discovered cis-acting element, MED-1, which has been shown to be common to many TATA-less, multistart site promoters (21). In the 3 nicotinic subunit promoter, mutation of MED-1 determined a 50% decrease in the expression of the reporter gene (Fig. 5), confirming the functional role of this element in a novel TATA-less promoter. However, this finding also raised questions regarding the role of MED-1 in the context of the activity of the AccIII/NcoI downstream promoter region and of the functional correlations with AP-2 and NFB. To approach these problems, we generated the SacII plasmid, in which both MED-1 and the putative NFB cis-acting elements were removed. SacII displayed the same promoter strength as MED-1 and NFB but without any further decrease in the expression of the reporter gene, as expected by the sum of the effects of the single mutations. When SacII was mutated in the downstream AP-2 putative binding site to obtain SacII-AP-2, its transcriptional activity decreased by 50%, becoming similar to that of the 0.16 BA construct.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Functional characterization of the AccIII/NcoI region in SY5Y cells. Left, scheme of the different constructs. The putative cis-acting elements for the indicated transcription factors are represented by specific symbols. The mutations of these elements are indicated by omitting the corresponding symbols in the specific construct. In the pGL3 control plasmid, the SV40 enhancer is located downstream of the luciferase (LUC) gene and is not indicated. Bars, transcriptional activity of the construct calculated as described in Materials and Methods and expressed as fold increase over the transcriptional activity of the promoterless plasmid pGL3-basic. The standard deviation is indicated. Each construct was tested at least three times in triplicate with two independent plasmid preparations. The difference in transcriptional activity between pGL3 control and pGL3 control + 0.2 kb was not statistically significant (F = 0.163) Differences in transcriptional activity between 0.35 BN and the mutated constructs, compared pairwise, were statistically significant (F < 0.01). Differences in transcriptional activity between 0.16 BA and the mutated constructs, compared pairwise, were statistically significant (F < 0.05) except the following pairs: 0.16 BA versus AP-2-NFB and 0.16 BA versus AP-2-SacII.

Functional characterization of the 0.6 KpnI/BglII region. The most relevant structural feature of the 0.6-kb KpnI/BglII region is the presence of an Alu repeat (Fig. 1) in opposite orientation with regard to the direction of gene transcription that displays an 85% identity to the modern, primate-specific Alu consensus sequence described by Britten (22). Because Alu sequences have been shown to have regulatory functions on the activity of proximal promoters (23, 24), we decided to determine whether the 3 Alu repeat could account for the inhibitory influence of the 0.6-kb KpnI/BglII region on gene expression (Fig. 6). For this reason, we generated the Alu construct, in which most of the Alu repeat was removed. On transfection, the transcriptional activity of the construct seem to be increased by ~2.5-fold compared with that of the parental 1 KN plasmid, indicating that the deleted fragment significantly participates in the repression of the downstream promoter. Nevertheless, the Alu construct does not possess the same promoter strength as the 0.35 BN plasmid. Because Alu still contains the extremities of the Alu repeat and some outside flanking regions, we generated new constructs to test the influence of these sequences on 3 gene expression. The functional analysis of the 0.8 PN and 0.6 XN plasmids demonstrated that the region upstream of the Alu repeat, contained in the KpnI/PstI fragment, and the 5 end of the Alu itself do not seem to play any relevant role in the inhibition of 3 gene expression. At the same time, it became clear that the element or elements responsible for the residual inhibitory activity on the downstream promoter should be contained either in the 3 end of the Alu sequence or in the region between the SspI and BglII restriction sites. Interestingly, the 3 end of the Alu repeat contains an element that was previously shown to work as a reducer in a monkey kidney fibroblast cell line (25). For this reason, we generated the 0.6-kb r construct, in which in addition to the Alu sequence, the putative reducer element was removed. To our surprise, the transcriptional activity decreased by 50%, indicating that at least in our system, the reducer actually behaves as a positive element.

View larger version (27K): [in this window] [in a new window]   Fig. 6.   Functional characterization of the 0.6-kb KpnI/BglII region in SY5Y cells. Left, constructs; top, extent of the Alu sequence; open box, luciferase (LUC) gene coding region; bars, the transcriptional activity of the construct calculated as described in Materials and Methods and expressed as fold increase over the transcriptional activity of the promoterless plasmid pGL3-basic. The standard deviation is indicated. Each construct was tested at least three times in triplicate with two independent plasmid preparations. Differences in transcriptional activity between 0.35 BN and all other constructs, compared pairwise, were statistically significant (F < 0.05). Differences in transcriptional activity between 1 KN and all other constructs, compared pairwise, were statistically significant (F < 0.05) except the following pairs: 1 KN versus 0.8 PN and 1 KN versus 0.6 r.

Expression analysis of the human 3 nicotinic subunit 5 regulatory region in non-neuronal cell lines. To understand the transcriptional mechanisms underlying the tissue-specific expression of the human 3 nicotinic subunit gene, we carried out transfection experiments in the non-neuronal cell lines TE671 and HeLa, which are not able to express the endogenous 3 subunit gene (Fig. 2A).

View larger version (17K): [in this window] [in a new window]   Fig. 7.   Expression analysis of the human 3 nicotinic subunit 5 regulatory region in non-neuronal cells. Left, chimeric constructs in which different parts of the 3 5 regulatory region are fused to the luciferase (LUC) reporter gene. Note that the 3.3-kb KpnI fragment is not to scale; open box, luciferase gene coding region; shaded portion of the box, DNA region specifying the 5 UT of the luciferase. In the other constructs, the entire DNA region specifying the 5 UT of the 3 mRNA was directly fused to the coding sequence of the reporter gene. Bars, transcriptional activity of the construct calculated as described in Materials and Methods and expressed as fold increase over the transcriptional activity of the promoterless plasmid pGL3-basic. The standard deviation is indicated. Each construct was tested at least three times in triplicate with two independent plasmid preparations.

	Discussion

Summary Introduction Materials & Methods Results Discussion References

The 3 nicotinic subunit is abundantly expressed in the postganglionic neurons of vertebrates, including humans (26). All ganglion cells respond to preganglionic stimulation by a fast excitatory postsynaptic potential, which triggers the initiation of the postsynaptic spike (27). The fast excitatory postsynaptic potential is due to the activation of nicotinic-acetylcholine receptors that contain 3 as essential agonist binding subunit (2).

The transcript encoding this subunit is also easily detectable in human neuroblastoma cell lines. Neuroblastomas are pediatric tumors that are thought to arise from migratory cells of the embryonal neural crest; several clonal cell lines have been stabilized from these tumors that display most of the features of the autonomic neurons, including the ability to express ganglionic-type nicotinic receptors (17, 18, 28). We exploited two human neuroblastoma cell lines, SY5Y and SK-N-BE, to characterize the gene promoter responsible for the expression of the human 3 nicotinic subunit gene.

To this purpose, we isolated a 4.3-kb KpnI/NcoI fragment that corresponds to the genomic region contained between the coding sequences of the 4 and 3 genes.

RNase protection assays along with the functional dissection of this intergenic region allowed us to identify a 0.35-kb fragment immediately upstream of the start codon that was able to drive the expression of the luciferase reporter gene in both neuroblastoma cell lines with a strength comparable to that of the SV40 promoter/enhancer. The region displayed the features of a GC-rich, TATA-less, and CAAT-less promoter, with multiple transcription start points, located downstream of a series of putative Sp1 and AP-2 binding sites. In the past, TATA-less, CAAT-less multistart promoters were believed to be typical of housekeeping genes, but more recently it has become clear that the expression of neuron-specific genes can be driven by such a regulatory sequences.

In the current study, we focused on the downstream region of the human 3 promoter and the Alu sequence, upstream of the promoter; we also started to investigate the mechanisms responsible for the neurospecific expression of the 3 gene. Our results are discussed below.

Functional characterization of the downstream region of the human 3 promoter. The functional role of cis-acting elements located downstream of the transcription start point and incorporated into the mature transcript has been reported (13, 28). In a recent report, the region downstream of the human immunodeficiency virus type 1 promoter, which specifies the gag leader sequence, was proved to contain some putative cis-acting elements, whose mutations severely affected gene transcription (29). The AccIII/NcoI region contains putative cis-acting elements for the transcription factors AP-2 and NFB. AP-2 has been shown to be expressed in cells derived from the neural crest (30), including the postganglionic neurons. NFB actually identifies a family of inducible transcription factors with a wide range of activities in different tissues (31). Although in many tissues the first step of the NFB activation from different external stimuli is the translocation of the factor from the cytoplasm to nucleus, in neuronal cells some members of the family can be constitutively localized in the nucleus (32), where they likely participate in the regulation of the expression of the target genes. Notably, both NFB and AP-2 DNA binding activities have been documented in unstimulated SY5Y nuclear extracts (33, 34). Although the molecular identity of the transcription factors that bind to the DNA consensus motifs in the AccIII/NcoI region remains to be confirmed, we showed that a double mutation in the AP-2 and NFB putative cis-acting elements reduces by 4-fold the transcriptional activity of the 0.35 BN construct, bringing it down to the level of that of the 0.16 BA construct. Thus, the two putative cis-acting elements not only are functional but also could account for the entire activity of the AccIII/NcoI region. However, it is likely that the molecular mechanisms that take place in the context of the AccIII/NcoI region are more complex than the simple sum of the positive activities of two transcription factors. We reached this conclusion by exploring the functional role of a novel cis-acting element, MED-1, that has been shown to be common to many TATA-less, multistart site promoters (21). In the pgp1 promoter, the element is able to bind a nuclear protein, as demonstrated by gel shift analysis, and its mutation was proved to reduce the expression to 25% of that of the wild-type promoter. The element and its cognate protein might act as selectors or activators of multiple start sites and regulate them as a cassette rather than individually. However, the precise molecular mechanism underlying the functional role of MED-1 is not yet known.

An Alu sequence, upstream of the 3 promoter, participates in gene expression regulation. Alu sequences are transposable elements specific to primates that by moving into positions of significance to gene expression are believed to be a source of evolutionary change (36). In particular, Alu sequences have been shown to have regulatory functions when located near promoters, with either positive or negative effects on gene expression (23, 24). Our data are in line with this evidence, demonstrating that the 3 Alu repeat behaves as a composite region in which positive and negative elements coexist. Surprisingly, the positive element corresponds to a DNA sequence of the Alu repeat previously characterized as a "reducer" in monkey kidney fibroblast cell lines CV-1 and COS-1 (25).

The 5 regulatory sequences of the 3 gene drive the expression of the reporter gene preferentially, but not exclusively, in neuronal cells. In humans, as in other species, the expression of the 3 nicotinic subunit has been demonstrated by in situ hybridization in the CNS (39), autonomic ganglia (26), and adrenal medulla.² In the current study, we analyzed the behavior of some constructs in non-neuronal cells, with the aim of understanding whether the 4-3 intergenic region contained the elements responsible for the restricted expression of the 3 subunit. We used Hela cells (data not shown) and TE671, a rhabdomyosarcoma cell line that is able to express muscle-type but not neuronal-type nicotinic receptors. Although the 4.3 KN construct displayed a preferential neurospecific expression pattern and some DNA regions were proved to work preferentially or exclusively in neuroblastoma cells, we could not demonstrate a completely neurospecific activity of the human 3 5 flanking region; there are several possible explanations. In the past few years, an NRSE and its cognate trans-acting factor, neuron restrictive silencer factor/RE1-silencing transcription factor, have been identified that are able to switch off the expression of neurospecific genes in non-neuronal cells (for a review, see Ref. 40). It is possible that we missed an NRSE because it is located outside of the 4-3 intergenic region. Interestingly, an NRSE was recently identified and proved to be functional in the mouse 2 nicotinic subunit promoter (13). It is also possible that tissue-specific DNA modifications, such as methylation, or proper interactions with chromatin are needed to reach a full neurospecific activity of the 3 regulatory region. These issues, which cannot be addressed with transient transfection experiments, are under study through the generation of stable transfectants.