Novel Messenger RNA and Alternative Promoter for Murine Acetylcholinesterase*

A portion of the 5′-flanking region of murine acetylcholinesterase was cloned from genomic DNA by 5′-rapid amplification of genomic ends, identified in a mouse genomic library, and sequenced. Multiple potential binding sites for universal and tissue-specific transcription factors were suggestive of a promoter region within this DNA sequence. Potential promoter activity was confirmed by coupling the new sequence to the open reading frame of a luciferase reporter gene in transient expression experiments with nerve and muscle cells. 5′-Rapid amplification of cDNA ends with templates from multiple sources revealed a novel transcription start site (at position −626, relative to translation start), located 32 bases downstream from a TATAA sequence. This start site appeared to mark a novel exon (1a) comprising 291 base pairs between positions −335 and −626, relative to the translation start. Supporting this conclusion, polymerase chain reactions with cDNA from mouse brain, heart, and other tissues, consistently amplified a transcript containing the exon 1a sequence fused to the invariant sequence beginning at position −22 in exon 2, but lacking exon 1. Northern blot analyses confirmed the in vivo expression of exon 1a-containing transcripts, especially in heart, brain, liver, and kidney. These results indicate that the murine acetylcholinesterase gene has a functioning alternative promoter that may influence expression of acetylcholinesterase in certain tissues.

A portion of the 5-flanking region of murine acetylcholinesterase was cloned from genomic DNA by 5rapid amplification of genomic ends, identified in a mouse genomic library, and sequenced. Multiple potential binding sites for universal and tissue-specific transcription factors were suggestive of a promoter region within this DNA sequence. Potential promoter activity was confirmed by coupling the new sequence to the open reading frame of a luciferase reporter gene in transient expression experiments with nerve and muscle cells. 5-Rapid amplification of cDNA ends with templates from multiple sources revealed a novel transcription start site (at position ؊626, relative to translation start), located 32 bases downstream from a TATAA sequence. This start site appeared to mark a novel exon (1a) comprising 291 base pairs between positions ؊335 and ؊626, relative to the translation start. Supporting this conclusion, polymerase chain reactions with cDNA from mouse brain, heart, and other tissues, consistently amplified a transcript containing the exon 1a sequence fused to the invariant sequence beginning at position ؊22 in exon 2, but lacking exon 1. Northern blot analyses confirmed the in vivo expression of exon 1a-containing transcripts, especially in heart, brain, liver, and kidney. These results indicate that the murine acetylcholinesterase gene has a functioning alternative promoter that may influence expression of acetylcholinesterase in certain tissues.
The best characterized function of acetylcholinesterase (AChE) 1 is to terminate cholinergic neurotransmission by hydrolyzing synaptic acetylcholine. AChE is expressed in many tissues, in multiple molecular forms (1,2). These molecular forms share identical catalytic properties but differ in hydrodynamic behavior, subunit assembly, and mode of association with the cell surface (2,3). Since AChE in all vertebrate species is encoded by a single gene (4 -10), the structural diversity arises from alternative mRNA splicing and post-translational modification (5,8,9). Alternative polyadenylation signals with tissue-specific usage are also found (11)(12)(13).
In addition to alternative mRNA processing at the 3Ј end of the open reading frame in the gene, Ache, there are indications of variant processing in the 5Ј-untranslated region. Promoter elements controlling cellular expression of AChE have been studied in Torpedo (14), mouse (12,15), and human cells (16). The results have consistently demonstrated a proximal promoter that is active in both muscle and nervous tissue. This promoter, with no TATAA and CAAT elements, drives transcription from an initiation site in the 87-bp noncoding region designated as exon 1 (12,15). Other data suggest that an alternative upstream promoter may also contribute to Ache transcription. Multiple mRNA bands are typically seen in Northern blots from brain and muscle, while some RNA protection assays point toward upstream elements in the 5Ј-flanking region of Ache (7,12).
One of the mechanisms that can give rise to more than one mRNA from a single gene is alternative promoter usage. Functional alternative promoters have been observed for many genes (17) and can be involved in regulating gene expression in different tissues. For example, alternative promoters generate four tissue-specific alternative noncoding first exons of brainderived neurotrophic factor (18). By transcribing a single gene from multiple promoters, an organism gains flexibility in controlling gene expression (19).
In this study, we characterized a portion of the 5Ј-flanking region of mouse Ache beginning about 400 bp upstream from the proximal promoter (12). An unusual abundance of general and tissue-specific transcription factor motifs suggested the presence of an alternative promoter. For this reason, we decided to search for a distal Ache promoter in this region. Our analysis indicates that it does contain an active promoter, which uses an alternative initiation site to drive expression of a novel AChE transcript.

EXPERIMENTAL PROCEDURES
PCR Primers-Unless otherwise indicated, oligonucleotide primers were synthesized in the Mayo Molecular Biology Core Facility and were based on the schemes of Li et al. (12) and Rachinsky et al. (9) (Gen-Bank TM /EBI Data Bank, accession numbers L06620 and X56518). Each primer in the following list is designated by the location of its 5Ј end with respect to the first nucleotide of the translation start codon, designated as position 1 (positions upstream of Ϫ23 are numbered consecutively regardless of intron/exon structure). Antisense-oriented Ache gene-specific primers were GSR-1 (26 nt, position Ϫ404), GSR-2 (23 nt, position Ϫ438), GSR-3 (23 nt, position ϩ500), GSR-4 (19 nt, position ϩ46), GSR-5 (18 nt, position Ϫ74), GSR-6 (26 nt, position ϩ306), and GSR-7 (20 nt, position ϩ995). Sense-oriented Ache primers were GSF-1 (18 nt, position Ϫ626), GSF-2 (18 nt, position Ϫ1251), GSF-3 (18 nt, position Ϫ485), and GSF-4 (20 nt, position, ϩ590). Sense primer for the anchor sequence of oligo(dG)-tailed first strand cDNA ((dC) 15 -F) was (5Ј-AAA AGA TCT GTC GAC CCC CCC CCC CCC CC-3Ј). Primers specific for the adapter sequence on mouse brain Marathon-Ready cDNA were APr1 (27 nt, at 5Ј end) and APr2 (23 nt, nested 3Ј of APr1), both supplied by CLONTECH. Primers specific for pBluescript plasmid DNA were pBSK988 (26 nt sense orientation, located upstream of the multiple cloning site, 5Ј position at 988), pBSK832 (24 nt sense, nested 3Ј to pBSK988), pBSK551 (22 nt antisense, located downstream of the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF148849.
Rapid Amplification of 5Ј Genomic Ends (5Ј-RAGE)-BALB/c mouse genomic DNA (CLONTECH) was digested in separate 5-g batches with each of 10 different restriction enzymes. The same enzymes were used to digest 5-g batches of pBluescript II SK M13(ϩ) phagemid (Stratagene). Batches of mouse genomic DNA and plasmid DNA digested with a given restriction enzyme were ligated for 5Ј-RAGE PCR (21). First-round PCR reactions were performed with 0.5 l of ligation mixture and 0.50 M gene-specific primer, GSR-1, and plasmid-specific primer, pBSK988 or pBSK551 (sense and antisense orientations). Reactions (50 l) were carried out in 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 20 mM Tris-HCl, pH 8.8, 2 mM MgSO 4 , 0.1% Triton X-100, 250 M each dNTP, with 1.5 units of AmpliTaq DNA polymerase (Perkin-Elmer). After 3 min of heating at 94°C samples were amplified for 35 cycles: 94°C for 1 min, 55°C for 2 min, 72°C for 3 min, and final extension at 72°C for 10 min. The products (2% aliquots) were subjected to a second round of 5Ј-RAGE PCR amplification with nested gene-specific primer (GSR-2) and nested plasmid primer (pBSK832 or pBSK551). Amplification conditions were unchanged except for the annealing temperature (53°C). Aliquots of each reaction (20 l) were electrophoresed on a 1.0% agarose gel for size determination (with 1-kb DNA ladder from Life Technologies, Inc. and X174 RF/HaeIII DNA size marker from CLONTECH). Extracted DNA bands were sequenced bidirectionally.
Rapid Amplification of 5Ј cDNA Ends (5Ј-RACE)-Templates for 5Ј-RACE PCR (22,23) were 0.5-ng aliquots of adapter-ligated BALB/c mouse brain Marathon-Ready cDNA (CLONTECH); separate experiments were performed with cDNA from three different batches (45 ng used altogether). Reaction mixtures (50 l) contained 0.40 mM each of adapter-specific primer, APr1, and antisense-oriented gene-specific primer, GSR-3. Initial amplifications used the buffer conditions of the 5Ј-RAGE reactions. Only gene-specific primer was present for the first 10 cycles (94°C for 30 s, 63°C for 30 s, 68°C for 4 min). Adapter primer was then added for 20 more cycles of the same conditions except for a 7-min extension at 68°C.
A second round of 5Ј-RACE was performed with 2% portions of the first amplification products as templates. One set of reactions used the nested adapter primer APr2 and nested, gene-specific primer GSR-2. Another set used only gene-specific primers GSF-1 and GSR-4. All samples were heated for 2 min at 94°C, followed by amplification for 35 cycles (94°C for 30 s, 52°C for 30 s, 72°C for 1 min) with a 10-min final extension at 72°C. Product was electrophoresed on 2% agarose, eluted, and cloned into TA cloning vector (Invitrogen, San Diego, CA) for further amplification before sequencing.
Enzymatic Amplification of mRNA-To detect possibly rare AChE mRNA species, we used the method of Ausubel et al. (24). In six preparations, 1-g aliquots from three different batches of poly(A ϩ ) RNA from BALB/c mouse brain (CLONTECH) were annealed and reverse transcribed using SuperScript TM II reverse transcriptase (Life Technologies, Inc.) from GSR-4 or GSR-6 primers. Oligo(dG) tails were added with terminal transferase (nucleoside triphosphate:DNA deoxynucleotidylexotransferase, Roche Molecular Biochemicals). A total of 300 PCR reactions were then performed using 5.0 ng of oligo(dG)anchored first strand cDNA as a template with (dC) 15 -F and GSR-4 primers. Cycling conditions were as described for 5Ј-RACE (35 cycles per reaction), except for minor variations in annealing temperatures depending on the primer pair. PCR products were electrophoresed on 2% agarose, extracted, and sequenced with gene-specific primers (GSR-1 and GSR-4) to locate the 5Ј end of mRNA transcripts and exon junctions.
Sequence Analysis-Specific DNA bands excised from agarose gels were extracted by the Wizard purification system (Promega) and sequenced in the Mayo Molecular Biology Core Facility on an Applied Biosystems DNA sequencer. The Taq dideoxy dye terminator cycle method was used, both with plasmid/adapter and with sequence-specific oligonucleotide primers.
Northern Blots-Northern blots were purchased from CLONTECH, preloaded with RNA size markers and poly(A ϩ ) RNA from BALB/c mouse heart, brain, spleen, lung, liver, skeletal muscle, kidney, and testis (2 g/lane). The blots were prehybridized and hybridized in Ex-pressHyb TM solution (CLONTECH) at 65°C. Hybridization was carried out with antisense cRNA probes (see below) as described by Ausubel et al. (25). Final washes were in 0.1ϫ SSC, 0.1% SDS at 65°C. All blots were exposed at Ϫ80°C.
Antisense cRNA Probes-Probes were prepared from the following DNA constructs, whose identity and orientation were confirmed by sequencing with vector-specific primers: 1) "exon 1a" (222 bp) generated by PCR with GSF-1 and GSR-1 primers using our mouse genomic AChE clone (see "Results") as a template (the product was cloned into a pNoTA/T7 shuttle vector (5Prime-3Prime), and the construct was linearized with EcoRI); 2) "exon 2/3" (598 bp) from position 542 to 1140 of the Ache coding region (9), kindly supplied by Dr. Palmer Taylor (Department of Pharmacology, San Diego, University of California) as a clone in pBluescript II KS(Ϫ), and linearized with HindIII; 3) "cyclophilin" (510 bp) generated by PCR with CYF and CYR primers, using mouse genomic DNA as a template. The product was cloned into pCR TM 2.1 vector (Invitrogen), and the construct was linearized with BamHI. Antisense cRNA was prepared from each of these linearized constructs and labeled with [␣-32 P]UTP (800 mCi/mmol, NEN Life Science Products) using T7 RNA polymerase and other components from the MaxiScript transcription kit (Ambion). Unincorporated radiolabel was removed on G-25 Sephadex columns (Roche Molecular Biochemicals).
Assay of Promoter Activity with Luciferase Reporter Genes-Candidate promoters were obtained by PCR amplification of three fragments from the mouse Ache genomic clone isolated in the present study: A) 411 bp (nt Ϫ485 to Ϫ75) containing the proximal Ache promoter region (12) with introduced 5Ј MluI and 3Ј BglII restriction sites; B) 813 bp (nt Ϫ1251 to Ϫ439) containing the contiguous 767-bp 5Ј-RAGE DNA sequence (see "Results") with introduced 5Ј KpnI and 3Ј BglII sites; C) 1177 bp (nt Ϫ1251 to Ϫ75) spanning fragments A and B, with introduced 5Ј MluI and 3Ј BglII sites. Amplified Ache fragments were double-digested with the appropriate restriction enzymes to permit ligation in appropriate orientation. The fragments were then cloned into a pGL2-Enhancer vector (Promega), which incorporates an SV40 enhancer and a luciferase gene as a reporter element. Sequence identities were confirmed before and after cloning into reporter plasmids.
N1E.115 murine neuroblastoma cells for the promoter assay were provided by Dr. Elliott Richelson (Mayo Clinic, Jacksonville, FL); L6 rat skeletal muscle myoblasts were from the American Type Culture Collection. Both cell lines were maintained in 10% CO 2 at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells at Ϸ50% confluence on 100 mm dishes were transiently transfected (26) with equimolar amounts (0.7 pmol) of expressing plasmid. Promoterless plasmid was added to correct for differences in construct size, fixing total DNA at 2.8 g/plate.
Before transfection, cells were washed twice with Opti-MEM I reduced serum medium (Life Technologies, Inc.). Plasmid DNA was introduced for 22 h as a 10% emulsion in LipofectAMINE in 5 ml of Opti-MEM reduced serum medium (both from Life Technologies, Inc.). After incubation for another 24 h in serum-containing media, cells were harvested and lysed for analysis of luciferase activity (27) and protein concentration (28). A promoterless pGL2 enhancer-vector defined the base line for promoter activity, while a pGL2 luciferase gene construct with SV40 promoter and enhancer served as a positive control for luciferase expression and a monitor for transfection efficiency in every experiment. Data were rejected when the luciferase activity of this control, per unit protein, differed by more than a factor of 2 from the overall mean. Initial experiments used co-transfection with an SV40-␤-galactosidase reporter vector for normalization of the data. This approach was abandoned because of strong interaction between the two types of promoter constructs (substantial ␤-galactosidase activity in single transfections, barely detectable activity in co-transfections), as reported in other systems (29). Instead we normalized luciferase activities to the amount of protein. All normalized values were then expressed as a fraction of the activity per unit protein obtained with the luciferase construct driven by the proximal AChE promoter.

RESULTS
Characterizing the 5Ј-Flanking Region of Mouse Ache-In 5Ј-RAGE PCR of mouse genomic DNA, many restriction enzyme digests were used but only XhoI digests yielded a distinct product (Figs. 1 and 3). The 5Ј end of this Ϸ1-kb product consisted of 175 bases of plasmid sequence, and the 3Ј end overlapped the known sequence of Ache. Including the 23 bases of primer, there was a perfect match with the first 55 bases in the previously reported, 5Ј-flanking region of Ache (12). Hence, the genomic location was presumed to lie upstream of the proximal promoter. The 767 nucleotides of novel sequence, designated "RAGE-DNA," began with the expected XhoI site (at Ϫ1260) and also contained, among many others, two HindIII restriction sites (positions Ϫ513 and Ϫ888).
To confirm that the novel sequence actually derived from  2. Luciferase promoter-reporter assays. Left, for directional insertion into a pGL2-Enhancer vector with integral luciferase reporter element, Ache fragments were PCR amplified with hybrid primers to incorporate specific restriction sites (see "Experimental Procedures"). The boundaries of tested constructs A, B, and C are indicated with respect to the Ache genomic sequence (top line); P indicates the proximal promoter. Construct D was a promoterless pGL2luciferase plasmid with SV40 enhancer, used as a base-line control. Right, seven separate times, constructs A-D were transfected into murine neuroblastoma or rat myoblast cells (see "Experimental Procedures"), and cellular luciferase activity was measured 48 h later. Transfection efficiency was monitored with a reporter consisting of an SV40 promoter and enhancer upstream of the luciferase open reading frame. Raw data in lumens/g of protein (mean Ϯ S.E.) were normalized with reference to activity simultaneously measured in cells transfected with construct A (set at 1.0). Statistical significance was evaluated with the non-parametric Mann-Whitney U test: *, p Ͻ 0.001 versus construct A; **, p Ͻ 0.001 versus construct A or B. tion, the 5Ј portion of this product matched perfectly the RAGE-DNA, while the 3Ј portion matched the previously sequenced 5Ј-flanking region of mouse Ache (12).
Promoter Activity of RAGE-DNA in a Reporter Gene System-The 5Ј-RAGE-DNA sequence contained a surprisingly rich array of consensus sequences for transcription factor binding (Fig. 1). These included potential Ets-1 and GAGA sites and multiple motifs for MyoD, NFB, and PEA3, in addition to Sp1, AP2, CCAAT box, CCAAT-binding protein (CP1) motif, and an atypical TATAA box. The sequence also contained possible cAMP-responsive elements. In view of these findings, we decided to test whether the 5Ј-flanking region of murine Ache might contain another functional promoter.
To assess promoter activity in a reporter-gene system, we amplified three parts of our genomic clone (Fig. 2): fragment A, containing the proximal promoter; fragment B, containing the entire RAGE-DNA sequence; and fragment C, containing both RAGE-DNA and proximal promoter. These DNA fragments were cloned into pGL2 luciferase reporter-vectors, and the resulting plasmid constructs, A, B, and C, were transfected into two AChE-expressing cell lines, NIE.115 murine neuroblastoma and L6 rat skeletal myoblasts. Promoter function was assessed in terms of luciferase activities in cell lysates pre-pared 48 h after transfection. Each experiment involved separate transfections of all three constructs; altogether, seven experiments met the criteria for analysis (see "Experimental Procedures").
In neuroblastoma cells, the promoter activity of construct B (RAGE-DNA) was nearly identical to that of construct A (proximal promoter), which was 0.4 times the positive control (SV40 promoter) and 12.5 times background (promoterless plasmid). The adjacent DNA sequences were markedly synergistic, as construct C was 6 times more active than either A or B. Results in L6 rat myoblasts differed. First, overall luciferase production was about 40-fold greater with every construct, including promoterless plasmid. This difference might mean that the L6 cells were more efficiently transfected or, alternatively, that they contained a different complement of co-activators. Second, construct A was as active as the SV40 promoter. Construct B, however, was only double the promoterless control and 12% as effective as construct A. Despite this weak direct action of the RAGE-DNA, there was still synergism, as construct C was twice as active as the proximal promoter.
Evidence for Exon 1a Sequence in AChE cDNA-The foregoing results showed that the RAGE-DNA can function as a promoter in neural cells, at least in vitro. To test whether this DNA actually functions as an Ache promoter in vivo, 5Ј-RACE PCR was performed with adapter-ligated mouse brain Marathon-Ready cDNA as template (see "Experimental Procedures"). The experiment involved two consecutive PCR reactions. Reaction 1 used adapter primer, APr1, and gene-specific primer, GSR-3. Reaction 2 (with 1 l of the product mixture as template) used only gene-specific primers, GSF-1 (selected for its position 32 nucleotides downstream of the putative TATAA box) and GSR-4 (positioned in exon 2; see "Experimental Procedures").
The 5Ј-RACE PCR generated a 360-bp DNA product that appeared to be a hybrid of exon 2 and the RAGE DNA (Figs. 3  and 4). This product was 100% identical with exon 2 from the 5Ј end of the reverse primer (at position ϩ46) to the 5Ј end of the invariant region of exon 2 defined by Li et al. (13), i.e. position Ϫ22 relative to the translation start site. Beyond that point, the sequence showed 100% identity with the RAGE-DNA starting at position Ϫ335 and running to position Ϫ626, the 5Ј end of the forward 5Ј-RACE primer. Most striking, the 5Ј-RACE PCR product lacked exon 1. Thus, a run of 312 nucleotides from the genomic sequence was entirely missing, including the proximal promoter and its site for transcription initiation. These comparisons with the extended genomic sequence of Ache suggest a new "exon 1a" and a new splicing pattern.
To confirm the splice junction between exon 2 and the RAGE-DNA, and to identify the transcription start site, cDNA was reverse-transcribed from poly(A ϩ ) mouse brain RNA with gene-specific AChE primers (see "Experimental Procedures"). A 3Ј-oligo(dG) adapter was then attached. Separate amplifications, using gene-specific (GSR-4) and adapter primers ((dC) 15 -F), yielded cDNAs of sizes 180, 390, or 440 bp. Three examples of the 180-bp product were analyzed and found to comprise AChE cDNA with the exon 1 sequence immediately downstream of the adapter primer. This sequence started from position Ϫ109, continued to Ϫ23, and was spliced to position Ϫ22 in exon 2 (data not shown). The five clones of the 390-bp product were found to represent an alternative AChE cDNA, beginning just downstream of the adapter primer. This alternative cDNA contained the whole exon 1a sequence (position Ϫ626 to Ϫ335) spliced to position Ϫ22 in exon 2. A minor variation of this theme was found in the seven clones of the 440-bp fragment. All of these were spliced like the 390-bp fragment, except for an additional 48 nt interpolated between the adapter sequence and position Ϫ626. The same 48-nt sequence was also present in one of the clones generated by 5Ј-RACE using mouse brain Marathon-Ready cDNA as a template. These findings suggest that the 48-nt sequence could well represent an upstream exon driven by an additional, unidentified promoter. This possibility is the subject of continuing investigation.
Size and Tissue Distribution of Exon 1a Transcripts-To detect exon 1a expression in vivo, we also performed reverse transcription-PCR with GSF-1 (specific for exon 1a) and GSR-4 primers, using freshly isolated total RNA from BALB/c mouse tissues as template. Amplification yielded robust DNA fragments of expected size from brain and heart (data not shown).
In view of this finding, we proceeded to characterize the distribution of exon 1a expression by Northern blotting. Blots of poly(A ϩ ) RNA from different murine tissues were probed with an antisense exon 1a-specific riboprobe. A transcript of approximately 2 kb was observed in all surveyed tissues and was a predominant species in kidney, liver, brain, and heart, but was barely apparent in muscle (Fig. 5A). Liver and testis showed additional signals (2.5 and 1.5 kb, respectively). A different expression pattern was obtained with a probe for the invariant coding sequence (designated "Exon2/3"), which labeled a 2-kb transcript in all tissues, most prominently in skeletal muscle, and an additional 2.5-kb transcript in brain and liver (Fig. 5B).
Mapping the Transcription Start Site-To map the potential transcription start of the new Ache promoter, we repeated 5Ј-RACE PCR with mouse brain Marathon-Ready cDNA. Again the initial PCR reaction used APr1 and GSR-4 as primers. This time, however, in order to obtain the extreme 5Ј terminus of the cDNA, the second reaction used GSR-2 (located well within exon 1a) and APr2, the nested adapter primer.
The repeated 5Ј-RACE PCR generated a 220-bp DNA with 186 bp of Ache sequence, in which the adapter sequence was ligated to Ache cDNA at position Ϫ626 (Fig. 3). This location is The blot was exposed for 8 days after probing with antisense cRNA specific for exon 1a, position Ϫ404 to Ϫ626 (cf. Figs. 3 and 6B). The major species labeled in most tissues was a 2-kb mRNA (black arrow), with an additional signal at 1.5 kb in testis and 2.5 kb in liver. B, transcripts for AChE coding region and cyclophilin. The blot from A was stripped and re-probed with antisense cRNAs specific for cyclophilin and for the invariant sequence of exons 2 and 3, positions 542-1140 (cf. Fig. 6A). Exposure was for 24 h. The main AChE species labeled in all tissues was a 2.0-kb mRNA (black arrow), with an additional 2.5-kb signal in brain; cyclophilin mRNA consistently was labeled as a 0.74-kb band (white arrow). C, exon 1a expression in relation to coding-region AChE mRNA; relative intensity of 2-kb bands in A and B (corrected for exposure time). midway through the RAGE fragment, 32 nt downstream from the TATAA sequence, and happens to coincide with the 5Ј end of the sense primer used in the earlier 5Ј-RACE PCR experiment. Position Ϫ626 represents the most distal position of the transcription start site for this upstream promoter (filled circles in Fig. 3). The transcription start site of exon 1a was also tested by enzymatic amplifications of RNA in which GSR-4 and GSR-6 primers were used for reverse transcription of poly(A ϩ ) RNA from BALB/c mouse brain. The longest cDNA obtained in this way began near position Ϫ626, and all started within 17 nucleotides of this position (open circles in Fig. 3). These results, together with all other information regarding the 5Јflanking region of mouse Ache, are depicted schematically in Fig. 6.

DISCUSSION
Organization of the 5Ј-Flanking Region in Murine Ache-A previously unreported 1.0-kb stretch of the murine Ache gene, directly upstream of the 5Ј-flanking region analyzed by Li et al. (12), has been obtained using the 5Ј-RAGE method. The new Ache sequence shows no significant homology with any other nucleotide sequence in the GenBank, which is typical for noncoding regions of DNA. This novel region presents a GC-rich pattern with many consensus motifs for transcription factors. Among the observed motifs were potential CCAAT, CP1, Sp1, and TATAA sites (30); two MyoD motifs (31); an immediate early growth response factor-1 (Egr-1) sequence (32); a motif for cAMP-responsive elements; and an AP2-binding motif (33). There were also multiple motifs for NFB (34). Motifs of the early transcription response elements family, Ets-1 and PEA3, were identified (30). In addition, there were multiple E box motifs, a GATA-1 motif (35), a GAGA motif (36), and a motif for the liver-specific factor HNF-5. We do not know which, if any, of these sites are functional, but their abundance is suggestive of a promoter that might operate under specific circumstances in a variety of tissues.
Existence of Alternative Promoter(s) in Mouse Ache-An alternative promoter in murine Ache was predicted earlier from Northern blots and RNA protection data (12), although corresponding transcripts could not be identified in a cDNA library. The suggested location of this alternative promoter was about 4 kb upstream of exon 1. The present data indicate a new promoter lying much closer to the open reading frame, in the distal 1 kb of the 5Ј-flanking region.
Evidence that our RAGE-DNA contains a functional Ache promoter is that: 1) this sequence drove reporter gene expression in neuroblastoma cells as efficiently as the proximal promoter; 2) a specific portion of the sequence was detected by PCR methods, both in a commercial preparation of adapterligated cDNA and in freshly isolated RNA; 3) the sequence could be detected by Northern blotting with specific riboprobes. There are several possible reasons why this distal promoter might have been overlooked. First, the new sequence lies in a GC-rich region and would be predicted to have a complicated folding pattern that may be problematic for primer extension assays (we did attempt classical primer extension assays using primers from the exon 1a region, but a clear-cut product was not obtained). Second, the promoter may generate only a minor fraction of the AChE mRNA. On the other hand, in a nuclease protection assay (12), DNA fragments including the region of exon 1a were reported as "weakly protected." Thus, the present findings are not inconsistent with available information.
One or more additional Ache promoters, still farther upstream, may also be functional. That is a reasonable inference from our finding of eight independent cDNA clones sharing a common 5Ј sequence of 48 nt directly preceding the exon 1a sequence. The existence of multiple alternative promoters is a legitimate subject of further investigation, but it remains an open question at present.
Synergism of Ache Promoters-One of the more interesting aspects of the new promoter was its synergism with the proximal promoter in a luciferase reporter system. Synergism occurred not only in cells where the new promoter functioned well alone (neuroblastoma), but also where it did not (myoblasts). Thus, construct C, incorporating both promoter regions, always led to super-additive effects, i.e. luciferase expression was greater than the sum of the expression driven by either promoter alone. A possible explanation is that, in this setting, elements of the alternative promoter work as enhancers. Another possibility is that DNA folding permits interaction between Sp1 proteins or other transcription factors bound to both FIG. 6. Structure of mouse Ache gene. A, scheme of Taylor and Radic (50). showing a possible alternative exon (1␣) and its hypothesized splicing to exon 1, along with alternative splicing at the 3Ј end of the gene. B, organization of the novel 5Ј-flanking region, with the splicing patterns implied by 5Ј-RACE PCR data. In A and B, respectively, the dimensions and locations of exon 1a-specific and exon 2/3 probes for Northern blotting are also indicated.
promoters (37). These possibilities could be resolved by further studies involving systematic deletions and mutations in the distal and proximal promoter regions, to determine what combination of promoters operates under different circumstances.
Mapping Exon 1a -The sequences from multiple clones, obtained by 5Ј-RACE PCR and by enzymatic amplifications of RNA, confirm the existence of an AChE mRNA transcript from which 312 nt, comprising exon 1 and the proximal promoter region (12), were excised. These sequences also suggest that transcription driven by the distal promoter starts at or near position Ϫ626, relative to the initiating ATG codon. That location is 32 nt downstream of the TATAA sequence, as one would expect if the TATAA box were functional. The location of the CCAAT box relative to this transcription start site is also in line with the consensus structure of other vertebrate promoters containing functional CCAAT and TATAA elements. A modified FoldRNA analysis (38) indicates that the minimum destabilizing energy of the novel promoter region is in the same (low) range as for the previously reported mouse Ache promoter (data not shown). Finally, the cap signal (CAG) matches base preferences in the majority of known vertebrate promoter sequences (39). Thus, the new promoter appears to be organized along conventional lines.
Genes of various species are known to carry multiple promoters. Selective usage of these promoters usually generates transcripts differing in 5Ј noncoding sequence but identical in protein coding sequence (40,41). It is worth noting that the nucleotide sequence of exon 1a contains two ATG codons upstream of the ATG that initiates the Ache open reading frame. A similar feature is shared by the 5Ј-flanking regions of the AChE genes in Drosophila (42) and Torpedo (43). In murine Ache, a favorable Kozak sequence precedes the ATG codon at position Ϫ601 (44). Since stop codons occur within 30 nucleotides downstream of each potential translation start site, exon 1a is unlikely to participate in protein coding or yield variant amino acid sequences. On the other hand, short open reading frames upstream of the coding sequence in many genes serve to control the expression of the main product at the translational level (45). We should therefore consider the possibility that exon 1a might have a role in regulating AChE translation from the open reading frame.
Tissue Distribution of Exon 1a-In the luciferase reporter data, the promoter's strength in neuroblastoma cells and its lower activity in myoblasts were striking. It is true that the transfected myoblasts were from rat, not mouse. However, the proximal promoter of mouse was highly effective in these same cells. Furthermore, at least in the portion that has been sequenced, the 5Ј-flanking regions of mouse and rat Ache are nearly identical (46).
Northern blot analysis confirmed the luciferase reporter data indicating that the new, distal, promoter is much less active in mouse muscle than the proximal promoter. The pattern of expression of exon 1a differs in other ways from the pattern detected with a probe for the main coding region, which labeled a double band in brain mRNA. In most studies of murine AChE, the size of the major AChE mRNA species has been estimated as 2.4 kb (9 -13, 47-49). Our observation of Ϸ2-kb transcripts might reflect differences in use of polyadenylation sites, or simply technical factors such as different sizing standards, electrophoresis conditions, etc. In any case, the numerous tissue-specific and general transcription factor motifs in the 5Јflanking region of murine Ache, together with the Northern blot data, suggest that the distal AChE promoter is active in many mouse tissues. Further studies are needed to define the relative roles of the distal and proximal promoters. Meanwhile, it seems that alternative promoter usage and synergistic promoter action, along with alternative splicing, could influence expression of AChE mRNA in tissues of both neural and non neural origin.