Two E-Boxes Are the Focal Point of Muscle-specific Skeletal Muscle Type 1 Na+ Channel Gene Expression*

We have characterized a group of cis-regulatory elements that control muscle-specific expression of the rat skeletal muscle type 1 sodium channel (SkM1) gene. These elements are located within a 3.1-kilobase fragment that encompasses the 5′-flanking region, first exon, and part of the first intron of SkM1. We sequenced the region between −1062 and +311 and determined the start sites of transcription; multiple sites were identified between +1 and +30. The basal promoter (−65/+11) lacks cell-type specificity, while an upstream repressor (−174/−65) confers muscle-specific expression. A positive element (+49/+254) increases muscle-specific expression. Within these broad elements, two E boxes play a pivotal role. One E box at −31/−26 within the promoter, acting in part through its ability to bind the myogenic basic helix-loop-helix proteins, recruits additional factor(s) that bind elsewhere within the SkM1 sequence to control positive expression of the gene. A second E box at −90/−85 within the repressor controls negative regulation of the gene and acts through a different complex of proteins. Several of these cis-regulatory elements share both sequence and functional similarities with cis-regulatory elements of the acetylcholine receptor δ-subunit; the different arrangement of these elements may contribute to unique expression patterns for the two genes.

Sodium channels comprise a multigene family, the members of which are expressed in a tissue-specific and developmentally-regulated manner. The electrophysiological properties of the different channel isoforms differ subtly, and presumably reflect the underlying need for slight variations in electrical conduction in the different tissues. Several members of the sodium channel family have been implicated in various inherited disorders in muscle, heart, and neuronal tissues (1)(2)(3). Although significant progress has been made in understanding the structural defects of mutant channel proteins, the mechanisms that regulate transcriptional control of these genes have been examined in detail only for the rat brain II (RBII) 1 and skeletal muscle type 2 (SkM2) isoforms (4 -7). In this report, we characterize several of the cis-regulatory elements that control transcription of the skeletal muscle type 1 sodium channel (SkM1).
The two isoforms of skeletal muscle sodium channel, SkM1 and SkM2, are regulated differentially during development. SkM2 is expressed in embryonic muscle but is down-regulated postnatally; denervation of adult muscle leads to the reactivation of SkM2 transcription (8,9). SkM1 comprises most of the sodium channel expressed in adult skeletal muscle, with the mRNA levels up-regulating 10-fold postnatally (8 -10), and it is expressed in the same temporal manner as the ancillary ␤ 1 subunit (11).
Since the SkM1 isoform is expressed in skeletal muscle, it seemed likely that it is regulated in part by muscle-specific transcription factors from the MyoD and MEF2 families (see Refs. 12-14, for reviews). Unlike most skeletal muscle genes, which are expressed at highest levels before birth, SkM1 reaches peak expression in adult muscle. Only a few other muscle genes, such as myoglobin, exhibit this late rise in expression levels (15), and novel factors may be involved in the regulation of such genes (16,17). In addition, like the acetylcholine receptor (AChR), the sodium channel is expressed at higher levels beneath the neuromuscular junction than in the surrounding sarcolemma (18,19); comparison of the SkM1 regulatory sequences to those for the AChR subunits may reveal common regulatory pathways.
We anticipate that the expression of the SkM1 gene is regulated by a complex combination of tissue-and developmentalspecific factors working in conjunction with the general transcription apparatus. As a first step toward elucidating the regulation of this gene, we have carried out a detailed analysis of the cis-regulatory elements that control its expression, elucidated similarities between some of these elements and those of the AChR ␦-subunit, and studied in more detail the critical role played by two E boxes. One E box, located within the promoter of the gene, plays a critical role in positive, tissuespecific gene expression. We demonstrate that members of the MyoD family of transcription factors contribute to this positive regulation, although they are not wholly responsible for it. The second E box plays a critical role in the repression of SkM1 expression in non-muscle cells through binding a complex of proteins.

EXPERIMENTAL PROCEDURES
Isolation of Genomic Clones-A rat spleen genomic library in DASH was screened using probes derived from the 5Ј-untranslated region of the SkM1 gene (20), and five overlapping clones were obtained. Clones 14.2, 29.1, and 22.1A were positive in a Southern blot to a PstI fragment encoding sequences Ϫ310 to ϩ119 relative to the published translation start site of the SkM1 gene (20). Sequence analysis of 14.2, reported previously (21), encodes the Ϫ401 to Ϫ226 segment but then deviates from that of the cDNA clone (20), revealing the presence of an intron between Ϫ402 and Ϫ401. The library was screened using se-quences Ϫ451 to Ϫ402 as a probe, and two clones, designated 2633 and 2A.1.1, were obtained. A restriction site map was derived using the enzymes HindIII, EcoRI, BamHI, and KpnI, indicating that the five clones encompass almost 40 kb of DNA surrounding the transcription initiation site of the SkM1 gene. A 3.1-kb HindIII fragment obtained from clone 2A.1.1 that was positive in a Southern blot with the Ϫ451 to Ϫ402 probe was cloned into pBluescript (KS) (SkM1-3.1) and used for all subsequent analysis.
DNA Sequence Analysis-DNA sequencing (Sequenase; U. S. Biochemical Corp.) was carried out on the SkM1-3.1 clone from the 3Ј edge using T7 and a series of primers to the SkM1 sequence. Using the resulting sequence, additional primers were synthesized and sequencing performed on the complementary strand. Three clones containing sequence between Ϫ1062 (BglII) and Ϫ438 (SphI), Ϫ438 (SphI) and ϩ11 (SacI), and ϩ11 (SacI), and ϩ254 (XbaI) were cloned into pAlter (Promega) and single strand DNA generated. This DNA was both sequenced and used for generation of site-directed mutations.
RNase Protection and 5Ј-RACE-RNase protections were carried out as described (4), using a [ 32 P]UTP-labeled antisense RNA probe containing sequences Ϫ640 to ϩ127 hybridized to 50 g of total RNA obtained from rat skeletal muscle or liver.
For 5Ј-RACE, poly(A) ϩ mRNA was purified from rat skeletal muscle, and cDNA synthesized using a primer containing the Ϫ325 to Ϫ349 segment relative to the translational start site. A poly(G) tail was added to the cDNA by incubating in 25 units of terminal transferase, 40 M dGTP, 0.72 mM CoCl 2 , and the buffer supplied by the manufacturer (Boehringer Mannheim) for 15 min at 37°C. PCR was performed using this template, an oligo(C) anchor primer (CACTGCAGAAGCTTG-GATCCCCCCCCCCCCCCC), and a primer complementary to sequence contained in the second exon of the SkM1 gene (Ϫ357/Ϫ380). For cloning purposes, the PCR primers were synthesized with restriction sites (HindIII for the anchor primer and XbaI for the SkM1 primer). The PCR products were cloned into pBluescript (SK) (Stratagene) using the HindIII and XbaI sites and 21 individual clones sequenced. The first transcription initiation site observed by 5Ј-RACE was numbered ϩ1 and all other sites are numbered relative to it ( Fig. 2A). Unless otherwise indicated, numbering is shown relative to the start site of transcription.
Generation of Reporter Gene Constructs-The SkM1-3.1 clone was digested with HindIII (Ϫ2.8) and XbaI (ϩ49) and the resulting fragment was ligated into pCAT-Basic (Promega) digested with HindIII and XbaI using a linker (XbaI-XhoI-XbaI). This clone was designated SkM1CAT (Ϫ2800 to ϩ49). 5Ј deletion mutants Ϫ1062 and ϩ11 were created from SkM1CAT using naturally occurring restriction sites at Ϫ1062 (BglII) and ϩ11 (SacI) and linkers (HindIII-BglII; HindIII-SacI). The 5Ј deletion mutant Ϫ438 (SphI) was created by inserting the SphI to XbaI fragment from SkM1CAT into pCAT-Basic cut with the same enzymes. The 3Ј deletion extending from Ϫ2.8 (HindIII) to Ϫ438 (SphI) was created by inserting this fragment into pCAT-Basic cut with HindIII and SphI and the 3Ј deletion mutant ending at ϩ11 (SacI) was created by digesting SkM1CAT with SacI and XbaI and ligating using a linker (SacI-XbaI). The 5Ј deletion mutants Ϫ273, Ϫ174, Ϫ135, Ϫ95, Ϫ85, and Ϫ65 were created by PCR. The 5Ј primers contained 20 bp of SkM1 sequence starting at the designated point and a restriction site for enzyme indicated below for cloning purposes; the 3Ј primer was complementary to ϩ56 to ϩ78. The PCR products were digested with PstI and XbaI (Ϫ273 and Ϫ174), SalI and XbaI (Ϫ65), or HindIII and XbaI (Ϫ135, Ϫ95, and Ϫ85), and cloned into pCAT-Basic digested with the same enzymes. All mutants generated by PCR were sequenced. To generate the constructs XmaS and XmaG, the XmaI fragment encoding sequences ϩ50 to ϩ254 was cloned into the XmaI site of the SkM1CAT vector and clones corresponding to the native (XmaS) and reverse (XmaG) orientation were selected.
Generation of Linker-scanning Mutations-Linker-scanning mutations of the promoter were created by site-directed mutagenesis using the Alter Sites II kit (Promega) according to the manufacturer's directions. The entire region between Ϫ57 (HinfI) and ϩ11 (SacI) was sequenced and introduced back into the SkM1 sequence of either the construct XmaS or the core promoter (Ϫ65 to ϩ49) at those restriction sites.
Cell Culture and Transient Expression Assays-Culture of primary muscle cells and NIH 3T3 cells was carried out as described (5,11). Cells were transfected by CaPO 4 precipitation as described (22) using either 8.5 pmol of SkM1 reporter gene constructs (25-41 g of DNA), 5 g of pCAT-Control as a positive control, or 24.4 g (8.5 pmol) of pCAT-Basic as a negative control. Rous sarcoma virus-␤-galactosidase was co-transfected as a control for transfection efficiency. Primary muscle cultures were transfected on culture day 2 and harvested on culture days 6 or 7, following robust formation of myotubes. NIH 3T3 cells were split to 40% confluence prior to transfection and harvested 2 days following transfection. Cells were harvested and assayed for CAT activity as described (18); samples were also assayed for ␤-galactosidase activity (23). Conversion of [ 14 C]chloramphenicol was quantitated using a PhosphorImager (Molecular Dynamics). The ␤-galactosidase assays were used to normalize the samples within an individual set of transfected cells; for comparison between transfections or cell types, CAT activity was expressed relative to pCAT-Control or the RBII promoter (5).
Gel Shift and DNase Footprint Assays-Nuclear extracts were prepared from muscle or 3T3 cells by the method of Dignam et al. (24). The 32 P-labeled Ϫ174/Ϫ65 and Ϫ135/Ϫ82 repressor probes and wild-type and mutant Ϫ34/Ϫ23 promoter probes were purified from a nondenaturing acrylamide gel. Double-stranded competitor DNAs were generated by annealing the wild-type or mutant synthetic oligonucleotides. Gel shifts and DNase footprints were carried out as described previously (19) with the following modifications. The binding buffer for the repressor probes was 4 mM HEPES, pH 7.9, 1% glycerol, 1% Ficoll, 20 mM KCl, 50 M EDTA, and 0.1 mM dithiothreitol, and the gel was run in a low ionic strength buffer (6.4 mM Tris, pH 7.5, 3.3 mM sodium acetate, and 1 mM EDTA) at room temperature. The binding buffer and conditions described by Simon and Burden (25) were used for gel shifts involving myogenic factors, except that 10 g of protein was used. Supershift assays were carried out as described previously (25) using commercially available antibodies to MyoD (Vector) and myogenin (Dako).

Isolation and Sequencing of Genomic Clones and Identification of Transcriptional Start
Sites-Five SkM1 genomic clones were obtained from a rat spleen genomic library, and the overlaps determined by restriction mapping (Fig. 1). A 3.1-kb Hin-dIII fragment derived from 2A.1.1 was positive in a Southern blot using a probe to sequences Ϫ451 to Ϫ402 relative to the translational start site (20). This fragment was cloned into Bluescript KSII, and used as the starting material for all subsequent analysis. Sequence analysis between Ϫ1062 and ϩ311 revealed that this fragment contains 225 bp of the first intron, the first exon (87 bp), and approximately 2.8 kb of 5Ј-flanking sequence ( Fig. 2A).
Transcriptional start site analysis was carried out by both nuclease protection and 5Ј-RACE. RNase protection assays were performed with an 32 P-labeled antisense RNA probe containing sequences Ϫ640 to ϩ127 hybridized to total RNA prepared from both adult rat skeletal muscle and liver. Protected fragments ranged in size from 87 to 70 bp (Fig. 3), indicating multiple transcriptional start sites.
5Ј-RACE was used to confirm the transcriptional start sites.
FIG. 1. Isolation of genomic clones. Five overlapping clones spanning approximately 40 kb of sequence were isolated from a rat genomic library. The 3.1-kb HindIII fragment was positive in a Southern blot to a probe encoding sequences Ϫ451 to Ϫ402 relative to the published translational start site (20). This fragment was isolated and used as the starting material for all subsequent experiments.
Of 21 clones sequenced, 16 encoded the SkM1 5Ј-untranslated region and 11 of these initiated between ϩ1 and ϩ29, in good agreement with the distribution pattern observed by the RNase protection assay. The five remaining clones initiated between ϩ40 and ϩ78, and may represent transcripts that terminated prematurely. The sequence of 5 clones obtained with 5Ј-RACE is shown (Fig. 3). We numbered the first transcription initiation site observed by 5Ј-RACE as ϩ1 and all other sites are numbered relative to it, as is shown in the sequence ( Fig. 2A).
Inspection of the sequence just upstream of the transcriptional start sites indicates that the SkM1 promoter region lacks canonical TATA boxes, CAAT boxes, or binding sites for SP-1, and also lacks conserved initiator sequence surrounding the transcription start sites (26). There is a CG-rich sequence just upstream of the transcription initiation sites that includes a consensus binding site for the AP-2 transcription factor (GCCN 3 GGC) as well as an E box (CANNTG), the consensus binding site for the bHLH family of transcription factors that includes the MyoD family. The general structure of the SkM1 promoter is similar to that of the rat brain type II sodium channel promoter, which also lacks common promoter motifs and has an E box in its promoter region (4).
In addition, there are several similar motifs in common between the 5Ј-flanking sequence of the AChR ␦-subunit gene and the SkM1 gene, as is highlighted in the sequence comparison (Fig. 2B). The E box in the SkM1 promoter at Ϫ31/Ϫ26 and the surrounding sequence are similar to the ␦-subunit promoter and both are located at similar sites relative to the first transcriptional start site. The SkM1 sequence contains two E boxes (at Ϫ90/Ϫ85 and Ϫ31/Ϫ26), whereas the ␦-subunit sequence contains only one (at Ϫ22/Ϫ16). However, the sequence of the E box at Ϫ90/Ϫ85 in the SkM1 sequence is identical to the E box in the promoter of the ␦-subunit gene. As described below, the E box in the ␦-subunit promoter may control two functions that are split between the two separate E boxes of the SkM1 gene. The second set of similar motifs occurs at Ϫ85 to Ϫ60 in the SkM1 sequence.
Deletion Analysis of the SkM1 Genomic Sequence Reveals Multiple cis-Regulatory Elements That Control Expression of the Gene-We created a series of deletion mutants of the SkM1 genomic sequence linked to the reporter gene for chloramphenicol acetyltransferase (CAT) and transfected these mutants into either rat primary muscle cultures, which express the SkM1 gene, or NIH 3T3 cells, which do not express the gene. The series of 5Ј deletion mutants shown in Fig. 4A terminate at ϩ49. The longest construction of this group, beginning approximately Ϫ2800 bp upstream of the transcription start sites, and 5Ј deletion mutants to Ϫ174, produced approximately 10-fold higher CAT levels in rat primary muscle cells than in NIH 3T3 cells, demonstrating the ability of the 5Ј-flanking region to drive muscle-specific expression. Deletion of sequences from Ϫ174 to Ϫ65 resulted in a 10-fold up-regulation in the 3T3 cell line, with only a 2-fold increase in primary muscle cells. Further deletion to ϩ11, which falls within the cluster of transcription start sites, reduced expression in both lines to levels only 2-3-fold that of the pCAT-Basic vector itself. A 3Ј deletion mutant that includes sequence between Ϫ2800 to ϩ11 exhibited expression levels in primary muscle cells similar to that of the mutation that ends at ϩ49, whereas further 3Ј deletion to Ϫ438 abolished expression in primary muscle cells. Thus, the rough boundaries of the SkM1 "core" promoter fall between nucleotides Ϫ65 and ϩ11, although for convenience, our core promoter constructs extend from Ϫ65 to ϩ49. This core promoter is not muscle-specific; the repressor element located between Ϫ174 and Ϫ65 confers muscle-specific expression on the SkM1 promoter.
The addition of nucleotides between ϩ50 and ϩ254 to the Ϫ2800/ϩ49 sequence increased expression in primary muscle FIG. 2. A, SkM1 genomic sequence. The 3.1-kb clone was sequenced from Ϫ1062 to ϩ311 and found to encode approximately 2.8 kb of flanking sequence, the first exon (87 bp), and part of the first intron, designated by lowercase letters. The first exon is indicated by a light underline, with the sequence from Ϫ38 to Ϫ87 corresponding to Ϫ451 to Ϫ402 (relative to the translational start site) of the published 5Ј-untranslated region of the SkM1 gene (20). Up arrows below the sequence indicate the transcriptional start sites, and bent arrows indicate the starting points of deletion mutants created for functional assays. Two E boxes that have a critical function in the regulation of the gene are boxed, and regions of sequence homologous to regions of AChR ␦-subunit sequence are indicated with a heavy underline. B, the 5Ј-flanking sequence for the AChR ␦-subunit (27) and SkM1 genes are aligned at the first transcription initiation site, as indicated. The single underline indicates the similarities in the promoter regions and the boxed sequences indicate functionally important E boxes. The double underlines indicate motifs conserved between the SkM1 Ϫ85/Ϫ60 positive element and an enhancer region of the AChR ␦-subunit. cells, 10-fold, to achieve a final level 50-fold higher than in 3T3 cells (Fig. 4B). This increase in gene expression is partly dependent on orientation, since placement of the ϩ50/ϩ254 fragment in a reverse orientation reduced the effect by 50%. This fragment did not function in an enhancer trap assay with either the heterologous SV-40 promoter or its native SkM1 promoter (data not shown) and is, therefore, a position-dependent positive element.
We refer to the genomic sequences spanning nucleotides Ϫ2800 to ϩ254 as the full-length SkM1 regulatory sequence. Within this region, we have initially defined three broad elements important for the expression of the SkM1 gene: 1) the core promoter, which lies between Ϫ65 and ϩ11 and lacks muscle-specificity; 2) a repressor, which lies between Ϫ174 and Ϫ65 and confers muscle-specificity on the promoter; and 3) a muscle-specific positive element which lies between ϩ50 and ϩ254.
Analysis of the Ϫ174/Ϫ65 Repressor Element by Gel Shift and DNase Footprint Assays-Because the repressor located between Ϫ174 and Ϫ65 confers muscle-specific gene expression, we investigated this element in greater detail. A 32 Plabeled Ϫ174/Ϫ65 probe was analyzed in a gel shift assay using nuclear extracts prepared either from primary muscle cells at various stages of development or from NIH 3T3 cells (Fig. 5A). As an independent control for the quality of the nuclear extracts, a 32 P-labeled SP-1 recognition site synthetic oligonucleotide was tested with the same extracts. A prominent shift, designated as the upper band (UB), was detected at the same mobility in primary muscle cells at all stages of development and in 3T3 cells. Although the intensity of the upper band was greater in 3T3 cells than primary muscle cells, the presence of a shift in the primary muscle cultures once myotubes had fully formed (day 7) seemed at odds with the activity profile in the two cell types shown in Fig. 4A. However, as shown below, this factor is not only present in both cell types, but also functional in both.
To further delineate the boundaries of the protein binding region, we performed DNase footprint assays using extracts from both day 7 primary muscle and 3T3 cells (Fig. 5B). The footprints in the two cell types had in common a series of hypersensitive sites and a protected region extending from Ϫ135 to Ϫ82 that may indicate the binding of a factor common to both cell types. In the primary muscle cells, there was also additional protection both within the Ϫ135 to Ϫ82 region and upstream to Ϫ65, indicating either that an additional factor binds the DNA in the primary muscle cells or that a different factor binds. The combination of the gel shift and footprint analysis suggested that repressor-binding protein(s) are present in both cell types, but that the identity of the protein or protein complex may not be the same.
The footprinted Ϫ135/Ϫ82 region produced the same gel shift pattern as observed for the Ϫ174/Ϫ65 repressor, although two lower bands (labeled LB) are more easily visualized when using the smaller probe (Fig. 5C). These shifts were also observed with the Ϫ174/Ϫ65 probe, but were partly obscured by the unbound probe. A Ϫ93/Ϫ82 fragment encompassing only the Ϫ90/Ϫ85 E box failed to produce a gel shift (data not shown), but when present at micromolar concentrations, this E box displaced the probe from the upper band in a competition assay, demonstrating that the relevant factor binds to the E box, although with a lower affinity than for the larger fragment.
An E Box at Ϫ90/Ϫ85 Plays a Crucial Role in the Function of the Repressor-Extending the results of the footprint analysis, we constructed a series of mutations in which the Ϫ174/ Ϫ65 repressor element or various permutations of this sequence were transferred to a heterologous promoter. We chose a minimal rat brain type II sodium channel promoter contain- FIG. 4. A, 5Ј and 3Ј deletion analysis. A series of 5Ј deletion mutations extending from the indicated sequence to ϩ49 were fused to the reporter gene for CAT. These constructs were transfected into either primary muscle cells (black bars) or NIH 3T3 cells (hatched bars) and expression levels for test constructs were normalized relative to a pCAT-Control vector. Two 3Ј deletions extending from Ϫ2800 to the indicated point were also created and tested. The pCAT-Basic vector without regulatory sequences inserted was used as a negative control. Error bars in all graphs indicate S.E. B, addition of a positive element. Sequences ϩ50 to ϩ254 were added to the Ϫ2800 to ϩ49 construct in either the native position and orientation or the reverse orientation.
FIG. 3. Transcription start site analysis. RNase protections were performed using a probe that extended from Ϫ640 to ϩ127. A series of protected fragments were observed in skeletal muscle, indicating the presence of multiple transcription start sites; no signal was seen in liver. 5Ј-RACE was used to confirm these start sites, and the sequence of five selected clones is shown below. The asterisk indicates the lane in which the first nucleotide of the clone occurs; the sequence on the gel is antisense.
ing sequences Ϫ130 to ϩ177 (RBII) because it is expressed strongly in many cell types and has structural similarities to the SkM1 promoter. As shown in Fig. 6A, the Ϫ174/Ϫ65 fragment, a slightly longer Ϫ174/Ϫ50 fragment, and the footprinted region of Ϫ135/Ϫ82 all reduced expression of the rat brain promoter to approximately 20% of its normal levels in both cell types, while a fragment extending from Ϫ135 to Ϫ95 only reduced expression to 70% of control. When a Ϫ93/Ϫ82 fragment, containing the E box at Ϫ90/Ϫ85, was transferred to the heterologous promoter, the level of expression was also 20% of control levels. Thus, the sequence encoding the E box conferred the same level of repression as observed for the full Ϫ174/Ϫ65 repressor.
Smaller Deletions between Ϫ174 and Ϫ65 on the Native Promoter Reveal the Presence of a Muscle-specific cis-Regulatory Element-Deletions on the native promoter confirmed the importance of the E box between Ϫ95 and Ϫ85 in the function of the SkM1 repressor (Fig. 6B). While deletions to Ϫ135 and Ϫ95 had only small effects, deletions that included the E box produced a much higher expression level in primary muscle than 3T3 cells. Further deletion to Ϫ65 greatly reduced this musclespecific expression, indicating the presence of a positive element adjacent to the repressor. As noted above, there are two motifs within this region, AGATTG at Ϫ83/Ϫ78 and GTTTC at Ϫ64/Ϫ60, that are conserved exactly between the SkM1 gene and a region of the AChR ␦-subunit gene that acts as an enhancer (25,27,28).
An E Box at Ϫ31/Ϫ26 Coordinates a Positive Muscle-specific Interaction of the SkM1 Promoter with Factors That Bind Elsewhere in the Full-length SkM1 Regulatory Sequence-We eval-uated the SkM1 promoter by characterizing a series of linkerscanning mutations through the region homologous to the ␦-subunit promoter. All mutations were created within two different "backgrounds" to assess their impact both on the promoter itself and on the larger, full-length SkM1 regulatory sequence. Within the background of the core promoter, all mutations reduced expression to 20 -80% of wild-type levels in both the positive and negative cell types, and mutations within the E box itself did not have special significance (Fig. 7A). The basal promoter activity arose from a distributed sequence of nucleotides, such that all mutations diminished promoter activity while none abolished it.
In contrast, expression in the background of the full-length SkM1 regulatory sequence clearly indicates the unique importance of the region between Ϫ31 to Ϫ23 that encompasses the promoter E box. Although mutations upstream of Ϫ31 and downstream of Ϫ23 reduced expression in primary muscle cells to 40 -60% of control levels, the mutations between Ϫ31 and Ϫ23 reduced expression to 5-20% of control levels (Fig. 7A). All mutations within the Ϫ31 to Ϫ23 region were not equivalent; the most severe disruptions involved the conserved consensus sites within the E box, while less severe effects resulted from mutation of adjacent nucleotides. The difference between the effect of these mutations in the full-length SkM1 regulatory sequence background as compared with the core promoter suggests that a factor bound to the E box orchestrates strong positive muscle-specific SkM1 expression in part through interaction with another factor(s) bound elsewhere in the fulllength SkM1 regulatory sequence.
Although the expression level in primary muscle cells was FIG. 5. A, gel shift assay with Ϫ174/Ϫ65 repressor probe. The Ϫ174/Ϫ65 probe was tested in a gel shift assay with nuclear extracts isolated from either primary muscle cells on culture day 2 (D2), 3 (D3), 4 (D4), and 7 (D7) or NIH 3T3 cells (left set of lanes). The position of the free probe is indicated. Gel shifts using the same extracts were also performed on a SP-1 probe as a control (free probe not shown, right set of lanes). B, DNase footprint of the Ϫ174/Ϫ65 probe. DNase footprint analysis of the Ϫ174/Ϫ65 probe was carried out using nuclear extracts from either day 7 primary muscle cells or NIH 3T3 cells. The B indicates the protein-bound probe, and the F indicates the free probe. Asterisks indicate the hypersensitive sites, and lines indicate protected regions. A Maxam and Gilbert A ϩ G sequencing reaction was carried out on the same probe to mark the position of the footprint. The region of the probe extending from the lowest hypersensitive site to the highest protected part in the 3T3 lane corresponds to sequences Ϫ135 to Ϫ82. C, gel shift competition with repressor E box. The gel shift performed with the Ϫ135/Ϫ82 probe forms a series of complexes, as does the Ϫ174/Ϫ65 probe, although with the larger probe, the lower bands (LB) are partially obscured by the free probe. Increasing concentrations of a competitor DNA encoding the repressor E box at Ϫ90/Ϫ85 disrupt the formation of the upper band (UB) in the gel shift assay performed with a Ϫ135/Ϫ82 probe. severely reduced by mutations in the promoter E box, the expression level in 3T3 cells was not increased by any of the mutations within the context of the full-length SkM1 regulatory sequence, in contrast to previously reported results for the E box of the ␦-subunit promoter (25).
MyoD and Myogenin Form Part of the Complex of Proteins That Bind to the SkM1 Promoter E Box-A 32 P-labeled probe encompassing the E box (Ϫ34/Ϫ23) gave rise to a prominent group of shifted bands in primary muscle cells, while fewer bands were seen in 3T3 cells (Figs. 7B and 8). In primary muscle cells, this pattern consisted of an upper band (UB), two closely spaced middle bands (UMB and LMB), and a lower band (LB). Competition with both the wild-type sequence and the mutants revealed a close correlation between the functional assay and the degree to which each mutant disrupted the gel shift (Fig. 7B). In 3T3 cells, there was a band of similar size and intensity to the UMB and two lower bands of different size from those in primary muscle cells.
Using antibodies to MyoD and myogenin, supershift assays were carried out to determine whether these factors are part of the complex of proteins that give rise to the gel shift with the promoter E box probe. As expected, assays performed with 3T3 nuclear extracts did not yield supershifted bands, confirming that MyoD and myogenin are not bound to this E box in the non-muscle cell line. Both antibodies produced supershifted complexes with primary muscle nuclear extracts. The UB was shifted upward by the MyoD antibody, indicating that MyoD is present in the largest complex, while the intensity of the LB signal was greatly diminished by the myogenin antibody, suggesting that the LB is composed of two independent complexes, one of which includes myogenin. Because the myogenin supershift fell on top of the UB, it was not possible to discern whether or not the myogenin antibody shifted the upper band. Since not all bands shifted, there must be factors other than MyoD and myogenin that bind this region. DISCUSSION We have identified multiple cis-regulatory elements that control expression of the SkM1 gene in a primary muscle culture system. The combination of 5Ј and 3Ј deletion analysis suggests that the core promoter of the SkM1 gene is located between nucleotides Ϫ65 and ϩ11. Within this sequence lie consensus binding sites for two families of transcription factors. The CG-rich sequence (Ϫ20 to Ϫ9) is characteristic of housekeeping genes and includes a binding site for the AP-2 factor, while the E box at Ϫ31/Ϫ26 is a binding site for the basic helix-loop-helix (bHLH) proteins that include members of the muscle-specific MyoD family. Linker-scanning mutations of nucleotides Ϫ40 to Ϫ11 within the background of a basal promoter sequence of Ϫ65 to ϩ49 indicates that both families of factors are involved in basal expression of the gene, since no single mutation eliminates promoter function, although all reduce function. This core promoter drives transcription in both primary muscle cells and NIH 3T3 cells, indicating that it lacks muscle specificity.
The linker-scanning analysis of the nucleotides Ϫ40 to Ϫ11 within the larger background of the full-length SkM1 regulatory sequence reveals that the Ϫ31/Ϫ26 promoter E box plays a role in coordination of positive muscle-specific expression of the SkM1 gene. Because this property is seen only in the background of the full-length SkM1 regulatory sequence, factors that bind to this E box must work in concert with factors that bind elsewhere in the full-length SkM1 sequence. The gel shift competition and supershift assays implicate the myogenic bHLH factors in this function.
An element located between ϩ50 and ϩ254 exerts a powerful positive effect on the overall expression of the SkM1 gene. This region encodes the 3Ј end of the first exon and the first 167 nucleotides of the first intron. Because this element exerts at least 50% of its effect in the reverse orientation, it likely acts by binding a transcription factor rather than by stabilizing the RNA structure. Since this element is included in the full-length SkM1 regulatory sequence, it is a candidate for interacting with the promoter E box.
A repressor element located between Ϫ174 and Ϫ65 confers muscle-specific expression on the native SkM1 promoter, although this specificity is not seen with a heterologous promoter. Interactions between the native promoter and the repressor must produce the muscle specificity observed on the native promoter, although we have yet to discover the nature of this interaction. The footprint of the Ϫ174/Ϫ65 probe localizes the binding of the repressor complex of proteins to Ϫ135/Ϫ82, and this fragment confers the same degree of repressor activity as the larger fragment. Although the E box at Ϫ90/Ϫ85 exhibits only low-affinity binding restricted to the upper band, it is required for repressor activity, implicating the factor which gives rise to this gel shift in the activity of the repressor. This factor is present in both muscle and non-muscle cell types, which lack the MyoD family of proteins.
Immediately downstream of the repressor E box, there is an additional muscle-specific positive element at Ϫ85/Ϫ60, within which occur two sequence motifs, AGATTG at Ϫ83/Ϫ78 and GTTTC at Ϫ64/Ϫ60, also found in an AChR ␦-subunit enhancer (25). In our primary muscle culture model system, the Ϫ90/Ϫ85 FIG. 6. A, SkM1 repressor acts on the heterologous RBII promoter. The indicated sequences of the SkM1 flanking sequence were transferred to the previously characterized rat brain type II sodium channel (RBII) promoter (18,19). CAT activity is shown relative to the RBII promoter in both primary muscle cells and NIH 3T3 cells. B, fine mapping of the SkM1 repressor in the context of its native promoter. Small deletions were constructed between Ϫ174 and Ϫ65. CAT activity is shown relative to pCAT-Control in primary muscle cells and NIH 3T3 cells.
E box masks the activity of this element. The overall effect of the repressor is therefore both to counteract the activity of the Ϫ85/Ϫ60 element and to confer muscle specificity on the native SkM1 promoter. The possibility that the Ϫ85/Ϫ60 positive element mediates the muscle-specific aspect of the interaction between the repressor and promoter was tested by using a construction extending from Ϫ174 to Ϫ50 on the heterologous promoter, but inclusion of the Ϫ85/Ϫ60 element fails to supply muscle specificity. Clearly, the interaction between the repressor, promoter, and Ϫ85/Ϫ60 positive element is complex, and it is unclear that their activity in vitro reflects their activity in vivo. In the ␦-subunit, the region of the second motif (GTTTC) was shown to be important in the subsynaptic expression of that gene (28). Since sodium channels, like AChR, are clustered preferentially at the neuromuscular junction, it will be important to determine if these elements play a role in localized gene expression in vivo.
One candidate for the factor that binds the repressor E box is the transrepressor ZEB, which binds to a subset of E boxes that encode CACCTG (29 -31). Like our factor, ZEB is a direct transrepressor, and is found in both non-muscle cells and in adult skeletal muscle (29,31). Others have shown that there is competition between the bHLH proteins and ZEB for binding to E boxes and that this competition reduces ZEB's activity in muscle (29,31).
Based on this competitive mechanism, we suggest a model that could explain the difference in expression patterns of SkM1 and AChR ␦-subunit. Following denervation, SkM1 mRNA levels remain relatively constant (9), while the ␦-subunit mRNA levels increase 30-fold (32). Although there is considerable similarity between the 5Ј-flanking regions of the two genes, the layout of the cis-regulatory elements is different. The ␦-subunit contains one E box that controls both positive and negative regulation (25), while the SkM1 contains two E boxes that control these functions separately. For the SkM1 gene, we propose that the repressor E box is always bound by a repressor factor, and the promoter E box is always bound by a bHLH protein. The 40-fold increase in myogenin levels and 15-fold increase in MyoD levels following denervation (33) have minimal impact since both E boxes are already occupied. In contrast, the bHLH proteins could displace repressor factor(s) bound at the sole E box of the ␦-subunit, resulting in the dramatic increase in mRNA levels post-denervation.
In the course of this work, we have identified several cisregulatory elements that control expression of the SkM1 gene and have begun to address the question of how they work together. These elements include: 1) a core promoter, which lies between Ϫ65 and ϩ11 and lacks muscle specificity; 2) a repressor, which lies between Ϫ135 and Ϫ82 and confers muscle specificity on the promoter; 3) a muscle-specific positive element at Ϫ85 to Ϫ60, which partially overlaps the repressor and the action of which is masked by the repressor; and 4) a musclespecific positive element that lies between ϩ50 and ϩ254 and confers a 10-fold up-regulation of expression. The organization of these four elements, and others yet to be identified, contribute to aspects of SkM1 gene regulation shared with other muscle genes, as well as those unique to itself.
Within the these larger regions, two E boxes play a pivotal role in the regulation of the gene, with an E box at Ϫ90/Ϫ85 FIG. 7. A, linker-scanning analysis of the Ϫ31/Ϫ26 promoter E box. A series of linker-scanning mutations were created spanning nucleotides Ϫ40 to Ϫ11. The native sequence is indicated before the slash and sequence to which it was changed, after the slash. These mutations were tested both within the background of Ϫ65/ ϩ49 core promoter sequence and the Ϫ2800/ϩ254 full-length SkM1 regulatory sequence, which includes the repressor and positive elements. CAT activity is shown relative to pCAT-Control in primary muscle cells and NIH 3T3 cells. B, gel shift competition of Ϫ34/Ϫ23 promoter E box probe with linker-scanning mutations. Gel shift assays were performed using nuclear extracts from primary muscle cells using a 32 P-labeled probe encompassing sequences Ϫ34/Ϫ23. A series of protein-binding complexes were observed. Increasing concentrations of either this same probe (wt) as a competitor, or competitors bearing the indicated mutations were used to displace the gel shift formed using the promoter E box probe.
FIG. 8. MyoD and myogenin form part of the complex of proteins which bind to the ؊34/؊23 promoter E box probe. Gel shift assays were performed using nuclear extracts from either primary muscle cells or NIH 3T3 cells in the absence and presence of antibodies (Ab) to MyoD and myogenin (Mgn).
controlling negative regulation and an E box at Ϫ31/Ϫ26 controlling positive regulation. Although some of the transcription factors binding these elements are the same factors that control expression of other genes, the arrangement of the cis-regulatory elements in SkM1 may lead to interactions between those factors that give rise to the unique regulation of this gene.