Embryonic activation of the myoD gene is regulated by ... - Development

Development 121, 637-649 (1995) Printed in Great Britain © The Company of Biologists Limited 1995

637

Embryonic activation of the myoD gene is regulated by a highly conserved distal control element David J. Goldhamer1,*, Brian P. Brunk2,†, Alexander Faerman3, Ayala King3, Moshe Shani3 and Charles P. Emerson, Jr2,† 1Department of Cell and Developmental Biology, University of Pennsylvania 2Institute for Cancer Research, Fox Chase Cancer Center, Philadelphia, PA 3Institute of Animal Science, The Volcani Center, Bet Dagan 50250, Israel

School of Medicine, Philadelphia, PA 19104, USA 19111, USA

*Author for correspondence: Department of Cell and Developmental Biology, 219 Anatomy-Chemistry Building, University of Pennsylvania School of Medicine, Philadelphia, PA 19104, USA. E mail: [email protected] †Present address: Department of Cell and Developmental Biology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104, USA

SUMMARY MyoD belongs to a small family of basic helix-loop-helix transcription factors implicated in skeletal muscle lineage determination and differentiation. Previously, we identified a transcriptional enhancer that regulates the embryonic expression of the human myoD gene. This enhancer had been localized to a 4 kb fragment located 18 to 22 kb upstream of the myoD transcriptional start site. We now present a molecular characterization of this enhancer. Transgenic and transfection analyses localize the myoD enhancer to a core sequence of 258 bp. In transgenic mice, this enhancer directs expression of a lacZ reporter gene to skeletal muscle compartments in a spatiotemporal pattern indistinguishable from the normal myoD expression domain, and distinct from expression patterns reported for the other myogenic factors. In contrast to the myoD promoter, the myoD enhancer shows striking conservation between humans and mice both in its sequence and its distal position. Furthermore, a myoD enhancer/heterologous promoter construct exhibits muscle-specific expression in transgenic mice, demonstrating that the myoD promoter is dispensable for myoD activation. With the exception of E-boxes, the myoD enhancer has no

apparent sequence similarity with regulatory regions of other characterized muscle-specific structural or regulatory genes. Mutation of these E-boxes, however, does not affect the pattern of lacZ transgene expression, suggesting that myoD activation in the embryo is E-box-independent. DNase I protection assays reveal multiple nuclear protein binding sites in the core enhancer, although none are strictly muscle-specific. Interestingly, extracts from myoblasts and 10TG fibroblasts yield identical protection profiles, indicating a similar complement of enhancerbinding factors in muscle and this non-muscle cell type. However, a clear difference exists between myoblasts and 10TG cells (and other non-muscle cell types) in the chromatin structure of the chromosomal myoD core enhancer, suggesting that the myoD enhancer is repressed by epigenetic mechanisms in 10TG cells. These data indicate that myoD activation is regulated at multiple levels by mechanisms that are distinct from those controlling other characterized muscle-specific genes.

INTRODUCTION

motifs known as E-boxes, sequences present in the regulatory regions of most muscle-specific genes. Transfection assays indicate that myoD family members also directly or indirectly positively regulate each other’s expression (Thayer et al., 1989; Braun et al., 1989; Edmondson et al., 1991, 1992). These crossand auto-regulatory functions may amplify expression of the myogenic factors and stabilize the muscle phenotype (Thayer et al., 1989). Importantly, these genes can induce myogenesis in a variety of non-muscle cell types when expressed from a constitutive promoter (reviewed by Emerson, 1990; Olson, 1990; Weintraub et al., 1991), a finding consistent with a function in determination of the skeletal muscle lineage. In vertebrates, skeletal muscle progenitor cells are derived predominantly from the somites, which are formed by seg-

An understanding of the molecular mechanisms that govern the determination and differentiation of skeletal muscle lineages has progressed rapidly since the discovery of the myoD family of myogenic regulatory genes. These genes, myoD (Davis et al., 1987), myogenin (Edmondson and Olson, 1989; Wright et al., 1989), myf5 (Braun et al., 1989) and MRF4 (Rhodes and Konieczny, 1989; also known as myf6 [Braun et al., 1990] and herculin [Miner and Wold, 1990]), which are expressed exclusively in skeletal muscle, encode structurally related transcription factors of the basic helix-loop-helix (bHLH) class (Murre et al., 1989). These myogenic factors activate muscle-specific gene transcription in differentiating cells by binding to DNA

Key words: myoD, transcription, myogenesis, basic helix-loop-helix factors, transgenic mice

638

D. J. Goldhamer and others

mentation of the paraxial mesoderm in a rostral-to-caudal sequence. The myotome gives rise to axial muscles, whereas trunk and limb muscles are derived from myogenic progenitor cells that migrate away from the ventrolateral edge of the maturing somite (Wachtler and Christ, 1992; Ordahl and Le Douarin, 1992). Myogenic cells of the branchial arches, which form facial, jaw and throat musculature, originate both from anterior somites and from cranial paraxial mesoderm (Noden, 1991; Couly et al., 1992). Gene-targeting experiments in mice showed that the elaboration of the myogenic phenotype is dependent on myogenic regulatory gene function. Mice homozygous null for both myoD and myf5 completely lack differentiated skeletal muscle by both histological and biochemical criteria (Rudnicki et al., 1993). In these mice, muscle-forming regions of the embryo are devoid of myoblasts, as evidenced by the absence of desmin-positive cells (Rudnicki et al., 1993). Interestingly, mice homozygous null for either myf5 or myoD alone exhibit no muscle defects at birth, indicating that the functions of myoD and myf5 in myogenesis are at least partially redundant (Braun et al., 1992; Rudnicki et al., 1992, 1993). In contrast, although myogenin knockout mice show severe muscle differentiation defects, myoblasts are present in approximately normal numbers and positions in these embryos (Hasty et al., 1993; Nabeshima et al., 1993). These data indicate that myoD and myf5 serve upstream functions in the myogenic developmental program to determine, expand, or maintain myogenic lineages, whereas myogenin, although not required for commitment of cells to the myogenic lineage, is required for normal biochemical and morphological differentiation of skeletal muscle (reviewed by Weintraub, 1993). Each of the myogenic genes is expressed in a unique spatiotemporal pattern in developing skeletal muscle (Buckingham, 1992; Faerman and Shani, 1993), indicating that members of the myoD family are regulated by distinct developmental signals. Given the formative role of myoD and myf5 in establishing the skeletal muscle lineage (Rudnicki et al., 1993), a mechanistic understanding of lineage determination will require detailed information of how these genes are transcriptionally controlled. Cis and trans analysis of the myoD enhancer offers a powerful means to define upstream signaling pathways and transcriptional events that govern myoD activation in the embryo. Previously, we showed by transgenic analyses that the embryonic expression of the human myoD gene is regulated by a distant enhancer localized to a 4 kb fragment approximately 18 to 22 kb upstream from the start of myoD transcription (Goldhamer et al., 1992). In the present study, we define and characterize the distal enhancer that regulates the embryonic activation of the human myoD gene. These data indicate that myoD activation is regulated by a highly conserved and complex regulatory system that is distinct from mechanisms that regulate the other myogenic regulatory genes.

DNA constructs A 1.7 kb ApaI/PstI fragment derived from the 4 kb enhancer-containing fragment (fragment 3 in Goldhamer et al., 1992; see Fig. 2) was subcloned into pBluescript KS+ (KS+; Stratagene) by blunt-end ligation into the unique EcoRV site of the multiple cloning site. Bidirectional nested deletions were created by exonuclease III digestion using the Erase-a-base kit (USB), or by partial digestion with PvuII at nucleotide 95, to create ∆15 (3′-5′). ∆14KS+ (3′-5′) contains 258 base pairs (bp) of enhancer sequence, and is referred to as the core enhancer throughout. The reporter plasmid ptkCAT∆EH (derived from pBLCAT2 [Luckow and Schutz, 1987] by deletion of the NdeI/HindII fragment of pUC 18) was the parental plasmid used for transfection experiments (Goldhamer et al., 1992). A 2.7 kb genomic fragment containing human myoD 5′ flanking sequences extending from an EcoRI site at −2.5 kb to +198 relative to the TATA box (−37 relative to the translational initiation codon) was generated from a sequencing deletion in KS+. −2.5CAT was produced by excising this promoter-containing fragment from KS+ by digestion with SacI and KpnI, followed by blunt end-ligation into the XbaI and BglII sites of ptkCAT∆EH. (XbaI and BglII digestion removes all thymidine kinase promoter sequences.) myoD enhancer/promoter CAT constructs were generated by digesting enhancer deletion constructs in KS+ with SalI and XbaI and cloning into unique SalI and XbaI sites of −2.5CAT. (The XbaI site was derived from KS+ during the construction of −2.5CAT.) The lacZ vector pPD46.21 (kindly provided by Andrew Fire) was used for transgene constructions. pPD46.21 is identical to pPD1.27 (Fire et al., 1990) except that it lacks the sup-7 gene. It contains an initiation codon and SV40 T antigen nuclear localization signal just upstream from the bacterial lacZ gene, and polyadenylation sequences from the SV40 early region. Enhancer/myoD promoter lacZ constructs were prepared by liberating enhancer/promoter inserts from the CAT vector by digestion with SalI and XhoI, and cloning into the SalI site of pPD46.21. Orientation was determined by restriction digests. tklacZ was prepared by digesting ptkCAT∆EH with BamHI and BglII and inserting the tk promoter fragment (from −105 to +51 of the HSVtk gene) into the BamHI site of pPD46.21. Orientation was determined by sequencing. To produce 258tklacZ, the BamHI/BglII tk promoter fragment was cloned into the BamHI site of ∆14KS+. After orientation was determined by sequencing, the 258tk fusion was excised by digestion with HindIII and XbaI and cloned into the corresponding sites of pPD46.21. The E-box mutant construct 258(E1-E3)/−2.5lacZ was cloned by excising the mutagenized core enhancer (see below) from ∆14KS+ with SalI and XbaI, and cloning the fragment into the SalI and SpeI (derived from the KS+ multiple cloning site during excision of the myoD promoter fragment) sites of −2.5lacZ.

MATERIALS AND METHODS

Sequence analysis The sequence of the human and mouse myoD enhancer was determined on both strands by dideoxy sequencing using the Sequenase kit (version 2.0; USB). The UWGCG sequencing package was used for sequence analysis, and transcription factor site searches utilized signal scan software (Prestridge, 1991). The nucleotide sequence of the

Unless otherwise noted, all molecular biological techniques were conducted using standard methods (Sambrook et al., 1989). All DNAs were purified by double banding in cesium chloride equilibrium gradients.

Cloning of the mouse myoD enhancer Mouse sequences homologous to the human enhancer were detected by Southern analysis of PstI-digested mouse genomic DNA. To create a size-limited library, the hybridizing band was excised from an agarose gel, ligated into PstI-digested KS+, and electroporated into NM554 bacteria using the BioRad Gene Pulser. This library was screened by standard methods using a probe from the human core enhancer to isolate the corresponding mouse sequences. Linkage to the mouse myoD gene was verified by hybridization of enhancer-containing cosmid clones (cosmid library was kindly provided by Yoshimichi Nakatsu) with a mouse myoD cDNA.

Regulation of the myoD gene in mouse embryos human and mouse myoD core enhancer has been submitted to the GenBank database. Mutagenesis E-box mutations were created by the PCR-based overlap extension method as previously described (Ho et al., 1989), using Vent polymerase, and ∆14KS+ as the template. By using mutant derivatives as the template for subsequent rounds of mutagenesis, a construct was obtained in which all three conserved E-boxes were mutated. All Eboxes were changed from the wild-type sequence CANNTG to CTNNTA, which destroys the minimal sequence required for binding of bHLH myogenic factors. Mutations were confirmed by sequencing both strands of the core enhancer in ∆14KS+. Transfections and CAT assays Culturing of 23A2 myoblasts (Konieczny and Emerson, 1984), transfections, cell extract preparation and CAT assays were done as previously described (Goldhamer et al., 1992). A minimum of three independent experiments were conducted for each DNA construct. Identical molar amounts (0.8 pmoles) of each plasmid were used (between 2.2 µg and 6 µg, depending on the size of the plasmid), which was adjusted to 25 µg with pUC 8 carrier plasmid DNA. Protein concentrations in cell extracts were measured by the modified Bradford assay (Bio-Rad) using bovine serum albumin as the standard. 15 µg to 25 µg of protein was used in each CAT assay, which yielded CAT activities within the linear range of the assay. Nuclear extract preparation C2C12 myoblasts, 10TG fibroblasts and JEG-3 choriocarcinoma cells were purchased from the American Type Culture Collection. HMP8 and FC1010 cells (kindly provided by Mark Lovell and Jerome Freed, respectively) are primary human myoblasts and foreskin fibroblasts, respectively. C2C12, 10TG, and JEG-3 cells were grown in DMEM (Gibco: with high glucose and sodium pyruvate) supplemented with 10% (10TG and JEG-3) or 15% (C2C12) fetal bovine serum (FBS; Hyclone). HMP8 cells were grown in Ham’s F-10 supplemented with 20% FBS and 0.5% chick embryo extract. FC1010 cells were grown in RPM-I supplemented with 10% FBS. All media was supplemented with penicillin (100 units/ml) and streptomycin (100 µg/ml). Cells were fed fresh medium every 3 days and harvested at about 50 to 80% confluence. Typically, 20 to 40, 15 cm plates were used for each nuclear extract preparation. Nuclear extracts were prepared according to the method of Zaret (personal communication) as follows. After rinsing cells twice with calcium- and magnesium-free phosphatebuffered saline at room temperature, 5 ml of ice-cold PSDP (0.15 M NaCl, 20 mM sodium phosphate pH 7.4, 0.35 M sucrose, 0.5 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride [PMSF], 5 µg/ml leupeptin) was added per plate, and cells were scraped into 50 ml tubes on ice. Cells were pelleted at 2,000 g for 5 minutes at 4°C. All subsequent procedures were conducted in a cold room on ice. Cell pellets were washed two times with PSDP and gently resuspended (2 ml for each original 50 ml of cell suspension) in an ice-cold hypotonic buffer (buffer A; 10 mM KCl, 10 mM Hepes pH 7.9, 1.5 mM MgCl2, 0.5 mM DTT, 1 mM PMSF, and 5 µg/ml leupeptin). Cell suspensions were combined, swelled on ice for 5 minutes, and centrifuged at 2,000 g for 5 minutes at 4°C. Cell pellets were resuspended in 3 ml of buffer A containing 0.5% NP-40, incubated on ice for 5 minutes, and cells lysed with a dounce homogenizer (10-15 strokes, pestle A). The cell lysate was transferred to a 15 ml disposable tube, 6 ml of buffer B (60 mM KCl, 15 mM NaCl, 15 mM Tris-HCl pH 7.4, 0.2 mM EDTA, 0.2 mM EGTA, 0.5 mM spermine, 0.15 mM spermidine, 1 mM DTT, 2 mM PMSF, and 5 µg/ml leupeptin) was added, and the cell lysate was centrifuged at 1,600 g for 5 minutes at 4°C. After gently washing the nuclear pellet with 2 ml of buffer B and centrifuging as above, the nuclear pellet was gently resuspended in 2 ml of hypertonic buffer C (0.42 M NaCl, 20 mM Hepes pH 7.9, 1.5 mM MgCl2, 0.2 mM EDTA, 25% glycerol, 1 mM DTT, 2 mM PMSF, 5 µg/ml leupeptin). The

639

nuclear suspension was transferred to two microfuge tubes and incubated on ice for 30 minutes, gently inverting the tubes every 5 minutes. Samples were microfuged at 3,000 revs/minute at 4°C for 5 minutes, and the supernatants dialyzed (6,000 to 8,000 Mr cutoff) with 1 liter of buffer D (60 mM KCl, 20 mM Hepes pH 7.9, 20% glycerol, 0.2 mM EDTA, 0.5 mM DTT, 1 mM PMSF) with two changes. After dialysis, extracts were microfuged for 5 minutes at full speed at 4°C, and small aliquots were quick-frozen in liquid nitrogen. Protein concentrations were determined by the modified Bradford assay (BioRad) using bovine serum albumin as the standard. DNase I protection assays The 258 bp core enhancer in ∆14 KS+ was 5′ end-labeled on the forward or reverse strand with [γ-32P]ATP (6,000 Ci/mmole; NEN) using unique SalI and XbaI restriction sites flanking the insert. Labeled fragments were purifed on non-denaturing 5% acrylamide gels, and after elution, were concentrated by ethanol precipitation. For each 50 µl reaction, 20 µg of nuclear protein and 5,000 to 10,000 cts/minute (1.4 fmoles) of labeled fragment was incubated in binding buffer (12 mM Tris pH 8.0, 50 mM KCl, 1 mM DTT, 1 mM MgCl2, 1 mM CaCl2, 5 mM NaCl, 100 µg/ml bovine serum albumin, 5% glycerol, 2% polyvinyl alcohol, and 0.5 µg poly(dI/dC)) on ice for 1 hour. Reactions were equilibrated to room temperature, 50 µl of binding buffer was added and, 1 minute later, DNase I (Worthington) was added to a final concentration of 1 µg/ml (experimental lanes) or 6 or 12 ng/ml (control lanes containing bovine serum albumin in place of extract). After exactly 2 minutes, stop buffer (50 mM EDTA, 0.2% SDS, 100 µg/ml yeast tRNA) was added, proteinase K was added to 75 µg/ml and tubes were incubated at 37°C for 30 minutes. Samples were extracted with an equal volume of phenol and the DNA was precipitated with ethanol for 5 minutes at room temperature. After centrifugation, pellets were washed two times in 70% ethanol, dried in a Speed-Vac, and resuspended in 8 µl of loading buffer (40% formamide, 1 mM EDTA). Samples were heated to 80°C and 3 µl of each sample was resolved on a 6% denaturing polyacrylamide gel. Dried gels were exposed to X-OMAT AR X-ray film (Kodak) with intensifier screens for 2 to 4 days. Positions of footprints were determined by comparison with G+A sequencing ladders prepared by standard Maxam and Gilbert chemical cleavages. At least two independently prepared extracts were tested for each cell type. DNase I protection assays were conducted a minimum of 3 times with each extract. DNase I hypersensitivity blots All cell types were from the American Type Culture Collection. The NB41A3 neuroblastoma cell line was grown in F-10 medium supplemented with 15% horse serum and 2.5% FBS. All other cell types were grown in DMEM with 10% FBS. Cells were grown in 10 cm dishes and fed every 3 days to approximately 80% confluence. Cells were harvested by scraping in calcium- and magnesium-free phosphate-buffered saline, permeabilized in NP-40, and treated with a range of DNase I concentrations (from 40-120 µg/ml) for 3 minutes on ice by the method of Rigaud et al. (1991). DNA was isolated by standard methods. Restriction enzymes and probes used are shown in Fig. 6. Hybridization and washing was done as described (Church and Gilbert, 1984). Whole-mount in situs Whole-mount in situs utilized a digoxigenin-labeled probe as described (Conlon and Rossant, 1992). The probe was from nucleotides 751 to 1785 of the mouse myoD cDNA (Sassoon et al., 1989; Faerman and Shani, 1993). Transgenic mice lacZ fusions were liberated from vector sequences and fragments purified as previously described (Goldhamer et al., 1992). Transgenic mice were produced by pronuclear injection of FVB/N 1-cell-stage

640


embryos (Hogan et al., 1986; Shani, 1986). To produce stable transgenic lines, DNA-positive male mice, or male offspring of female transgenic mice were mated to FVB/N mice. For analysis of injected, Fo mice, at least four independently derived lacZ-positive embryos were analyzed for each construct. DNA-positive mice were detected by Southern blot analysis of tail (stable lines) or placental (Fo mice) DNA as described (Shani, 1986), using a probe specific to lacZ sequences.

RESULTS A distal 258 bp core enhancer controls the embryonic activation of myoD We previously identified a muscle-specific enhancer contained in a 4 kb fragment (fragment 3 in Goldhamer et al., 1992) located 18 to 22 kb upstream of the human myoD gene. In combination with the myoD promoter, this enhancer drives expression of a bacterial lacZ gene in all skeletal muscle compartments of mouse embryos, conferring the endogenous myoD pattern of expression in 11.5 days post-coitum (d.p.c.). transgenic mice (Goldhamer et al., 1992). To analyze further the temporal and spatial pattern of activation of the transgene, we created stable transgenic lines that harbor the 4 kb enhancercontaining fragment juxtaposed to the myoD promoter (contained in 2.5 kb of myoD 5′ flanking sequence) cloned upstream of the lacZ gene. The spatiotemporal pattern of lacZ expression throughout embryogenesis coincides with that of the endogenous mouse myoD gene (Fig. 1; Faerman et al., unpublished data). Several interesting features of transgene expression patterns in somites are described below. In most somites anterior to the level of the forelimb bud of all three independently derived transgenic lines, lacZ-positive cells are scattered throughout the dorsal-ventral myotomal axis (Fig. 1A-C) and transgene expression appears to be coordinately activated along the dorsal-ventral axis beginning at 1010.5 d.p.c. (Fig. 1B). The most intense staining in these anterior somites is in the dorsal region of the myotome (Fig. 1A,C; see also Fig. 3A). In somites posterior to the forelimb bud, however, lacZ-expressing cells initially accumulate in the ventral portion of the myotome (Fig. 1B), followed by expression in dorsal myotomal cells, including cells of the dorsal medial lip of the dermamyotome (Fig. 1A,D). In contrast to more rostral somites, the most intense staining in these somites is in the ventral region of the myotome (Fig. 1A,B,D), which is likely due to a greater density of β-galactosidase (β-gal)-positive cells (Fig. 1D). Interestingly, a population of myotomal cells between these ventral and dorsal populations does not express the transgene (Fig. 1A,D) throughout myotomal stages, a result confirmed by analysis of serial sections. Whole-mount in situ analysis of endogenous mouse myoD mRNA yielded a similar pattern of myotomal expression that is dependent on axial position; message accumulation is highest dorsally in the myotome of the most anterior somites and ventrally in the myotome of more posterior somites (Fig. 1E). These data demonstrate the sufficiency of the identified regulatory elements in recapitulating the normal spatial pattern of myoD expression in somites. To further localize enhancer activity, overlapping restriction fragments and nested deletions were tested in transient transfection assays for their ability to enhance transcription from

the myoD promoter in 23A2 myoblasts. The myoD promoter alone has a very low basal level of CAT reporter gene activity, only about 5-fold above a promoterless CAT construct. Fragment 3 typically increases this basal level of transcription 10- to 20-fold in 23A2 myoblasts (Goldhamer et al., 1992). An internal 1.7 kb ApaI/PstI fragment was identified that yields approximately 50% of the CAT activity of fragment 3 (Fig. 2A). No other subfragment residing entirely outside of this 1.7 kb fragment exhibits enhancer activity, although all of the activity of fragment 3 could be reconstituted by the additive activities of two fragments that overlap the 1.7 kb fragment; the 2.1 kb KpnI fragment (see Fig. 2B), which includes the 5′ half of the 1.7 kb fragment, exhibits approximately 70% of the activity of fragment 3, and a 1.4 kb KpnI/EcoRI, which includes the 3′ half of the 1.7 kb fragment constitutes the remaining activity (data not shown). Importantly, the 1.7 kb fragment, and subfragments therein, yield the appropriate pattern of muscle-specific expression in transgenic mice (see below). Therefore, sequences outside of the 1.7 kb fragment were excluded from further analysis. Transfection analysis of bi-directional nested deletions of the 1.7 kb fragment identified two non-contiguous DNA elements with modest enhancer activity (Fig. 2A). Enhancer 1, defined by the 3′ to 5′ deletion 14, was localized to a 258 bp sequence at the 5′ end of the 1.7 kb fragment (Fig. 2A, bottom). Further deletion to nucleotide +95 abolishes enhancer activity (Fig. 2A; deletion 15). Approximately 1 kb downstream, a second enhancer was identified whose 5′ end falls between the endpoints of the 5′ to 3′ deletions, 7 and 8. The 3′ limit of this enhancer lies between the endpoint of the 3′ to 5′ deletion 1, and the 3′ end of the 1.7 kb fragment. Precise definition of the enhancers’ boundaries is difficult because of their inherently low activity in transfection assays. No activity was detected in sequences between enhancers 1 and 2 (data not shown). As with fragment 3 (Goldhamer et al., 1992), the 1.7 kb fragment, as well as subfragments containing enhancer 1 or 2, are active when transfected into both muscle and non-muscle cells in culture. Deletion constructs containing either enhancer 1 or enhancer 2 were tested transgenically for their ability to direct expression of a lacZ reporter gene in the typical myoD expression pattern (Fig. 2B, 3). Transgenes driven by enhancer 2 and the myoD promoter are only ectopically expressed with no consistent pattern, which is typical of results observed with promoter sequences alone (Goldhamer et al., 1992). In contrast, both subfragments containing enhancer 1 (Fig. 2B) are expressed specifically in skeletal muscle in transgenic mice; all five Fo transgenic mice and a stable line containing the minimal 258 bp enhancer 1 and the myoD promoter exhibit a pattern of expression indistinguishable from the larger 1.7 kb fragment or the 4 kb fragment 3 (Fig. 3A). Because enhancer 1 directs the myoD pattern of expression in transgenic mice, this 258 bp sequence will hereafter be referred to as the myoD core enhancer. The myoD core enhancer was tested in combination with the heterologous herpes simplex virus thymidine kinase (tk) promoter to assess the relative contributions of the enhancer and myoD promoter to the regulation of myoD expression. As shown previously, the tk promoter alone exhibits only ectopic expression due to integration site position effects (Allen et al., 1988; data not shown). In contrast, all seven DNA-positive

Regulation of the myoD gene in mouse embryos

641

Fig. 1. Spatial domains of myoD expression in somites. (A-D) β-gal expression in stable F3/−2.5 lacZ transgenic embryos. (A) Whole-mount 11.5 d.p.c. embryo showing the overall pattern of β-gal expression. Transgene expression is predominantly in the dorsal region of the myotome of the most anterior somites (white arrows) and in the ventral region of myotomes (black arrows) posterior to the forelimb bud (fb). Staining is also seen in forelimb and hindlimb buds (hb) and in branchial arches (ba). Ectopic staining in the neural tube is likely due to position effects. The nasal epithelium (asterisk) usually shows transgene expression in independent lines, for unknown reasons. (B) 10.5 d.p.c. whole-mount embryo from same transgenic line shown in A. Transgene expression is most prominent in the ventral-most portion of thoracic somites (black arrows). Note staining along the entire dorsal-ventral axis of the most anterior somites. Weak staining in the dorsal-most region of the myotomes of thoracic somites is obscured by ectopic neural tube staining. (C) Transverse section anterior to the forelimb bud of an 11.5 d.p.c. embryo from an independent transgenic line. Nuclear localized β-gal expression is observed throughout the dorsal-ventral axis of the myotome with the most intense staining dorsally (white arrow). asterisk, hypoglossal premuscle mass, corresponding to the more dorsal strip of stained cells ventral to anterior somites in A (see Faerman and Shani, 1993). Ectopic expression is restricted to dorsal root ganglia (drg) in this line. nt, neural tube. (D) Transverse section posterior to the forelimb bud from the same embryo shown in C. β-gal expression in the myotome of this thoracic somite is most prominent ventrally (black arrow). Note the population of β-gal negative cells (bracket) under the dorsal-most region of the myotome (white arrow). (E)Whole-mount in situ localization of endogenous mouse myoD mRNA in a 10.5 d.p.c. embryo. The spatial pattern of myoD message accumulation closely matches the pattern of transgene expression. black arrows, localization of myoD mRNA in the ventral region of myotomes of thoracic somites. white arrows, myoD mRNA is most abundant in the dorsal region of anterior somites. Low level expression in the forelimb bud and branchial arches is not apparent in this preparation. Staining of the otic vesicle (ov) is non-specific background.

embryos injected with a construct containing the core enhancer in combination with the tk promoter show clear musclespecific expression at 11.5 d.p.c., exhibiting a rostrocaudal gradient of expression in myotomes, as well as expression in branchial arches, limb buds, and other myogenic centers (Fig. 3B). As with previous constructs, the most anterior somites and thoracic somites could be distinguished by their prominent dorsal and ventral staining, respectively. Interestingly, however, most of these transgenic mice exhibit lacZ-positive cells scattered throughout the dorsal-ventral myotomal axis of all somites, regardless of their axial position (Fig. 3B). This

indicates a contribution either of sequences in fragment 3 outside of the core enhancer, or of myoD promoter sequences, in restricting the spatial expression of myoD within the somite myotome. The myoD core enhancer is highly conserved between humans and mice As a means of identifying critical regulatory motifs within the enhancer, we compared myoD enhancer sequences between humans and mice, reasoning that important regulatory sequences would be most highly conserved. For this analysis,

642


Fig. 2. (A) Localization of myoD enhancer activities by transient transfection assays in 23A2 myoblasts. The 1.7 kb ApaI/PstI fragment within fragment 3 (see B) was the parental molecule used to create nested deletions. All deletion constructs were assayed for enhancer activity in combination with the myoD promoter. Numbers are relative to the CAT activity of the promoter alone, which was arbitrarily set to a value of 1. Values (± s.e.m.) are the average of at least three independent experiments. Arrows denote the maximum limits of the two enhancers. (B) Summary of transgenic data, localizing the muscle-specific regulatory region to a 258 bp core enhancer. All fragments shown were assayed in combination with the myoD promoter. The approximate positions of enhancer activities defined by transient transfection assays are shown (boxes).

the mouse myoD enhancer was cloned from a size-selected plasmid library using the human core enhancer as a probe. Restriction mapping and Southern blot analyses of enhancercontaining cosmid clones demonstrated linkage with the mouse myoD gene, and revealed that the mouse myoD enhancer, like the human enhancer, is located approximately 20 kb 5′ of the start of transcription (data not shown). The human and mouse core enhancer show extensive sequence similarity (89% identity), particularly within the first 160 bp, in which the enhancer in the two species is 94% identical (Fig. 4). Sequence similarity drops dramatically outside of the core enhancer, although small regions of homology exist throughout the 1.7 kb ApaI/PstI fragment as well as 5′ of the core enhancer (unpublished observations). The human core enhancer sequence contains four E-boxes,

three of which (E-1, E-2 and E-3) are conserved in sequence and position in mice (Fig. 4). Both central and flanking nucleotides, which strongly influence the affinity of bHLH protein binding (Blackwell and Weintraub, 1990; Sun and Baltimore, 1991; Wright et al., 1991), are conserved between species in E-boxes 2 and 3. E-boxes 2 and 3 represent potential high affinity binding sites for MyoD/E12-E47 and MyoD homodimers, respectively (Blackwell and Weintraub, 1990; Sun and Baltimore, 1991). With the exception of E-boxes, the myoD core enhancer is distinguished by the lack of binding sites for factors known to regulate the expression of other muscle-specific genes (Fig. 4). Motifs that are absent include MEF-2 (Gossett et al., 1989; Cserjesi and Olson, 1991), CArG (Minty and Kedes, 1986), MHOX (Cserjesi et al., 1992), and M-CAT (Mar and Ordahl, 1990) sites. The A-T rich sequences

Regulation of the myoD gene in mouse embryos from nucleotides 16 to 25, and from nucleotides 31 to 41 (Fig. 4) resemble consensus CArG and MEF-2 motifs, but differ at nucleotides required for serum response factor and MEF-2 binding (Treisman, 1986; Pollock and Treisman, 1990, 1991; Cserjesi and Olson, 1991). A search of the transcription factor database revealed consensus sequence binding sites for several widely expressed transcription factors, including the Ets protein, PEA3 (Xin et al., 1992); nts 28-33, reverse strand), AP-1 (nts 143-149), NF-1 (nts 176 to 187), and H4TF-1 (Dailey et al., 1988; nts 165-173), a regulator of the histone H4 gene (Fig. 4). Mutation of the enhancer E-boxes does not affect enhancer activation in the embryo To test the functional significance of the conserved E-boxes, an enhancer construct in which all three conserved E-boxes (E1 through E-3; Fig. 4) were mutated, was tested in transgenic mice. In this construct, the canonical E-box motif CANNTG was changed to CTNNTA, thereby destroying the minimal sequence required for bHLH factor binding. This construct (258(E1-E3)/−2.5lacZ) contains the mutant core enhancer cloned upstream of the myoD promoter in the lacZ vector used above. Fo mouse embryos were analyzed at 11.5 d.p.c., which is about 1 to 1.5 days after the myotomal activation of both the endogenous myoD gene (Sassoon et al., 1989; Faerman and Shani, 1993), and myoD enhancer lacZ fusions. Defects in the activation function of the enhancer would result in a delay or loss of lacZ expression, as observed with E-box mutations in the myogenin promoter (Cheng et al., 1993; Yee and Rigby, 1993). Analysis of four lacZ-positive embryos, however, revealed no apparent difference between the pattern of expression of the mutant construct and the wild-type constructs (Fig. 3C). Also, there appears to be no delay in activation, since the caudal extent of expression along the rostrocaudal expression gradient is similar to that of wild-type constructs at 11.5 d.p.c. (compare to Figs 1A, 3A). In addition, expression of the mutant transgene was detected in the hindlimb bud at 11.5 d.p.c., a time coincident with the initial detection of wild-type enhancer constructs. An enhancer construct entirely lacking E-boxes (the three conserved E-boxes and the fourth, non-conserved E-box; see Fig. 4), also is expressed normally at 11.5 d.p.c (unpublished observations). We conclude that E-boxes in the myoD core enhancer are not required for myoD gene activation. Nuclear trans factors interact with multiple sequence elements in the core enhancer DNase I protection assays were used to identify enhancer sequences that interact in vitro with nuclear factors from human and mouse cells. Mouse cell types analyzed were C3H10TG (10TG) fibroblasts and C2C12 myoblasts. Human cells used were primary fibroblasts (FC1010), primary myoblasts (HMP8) and the choriocarcinoma cell line, JEG-3. A representative experiment is shown in Fig. 5. Extracts from HMP8 myoblasts, C2C12 myoblasts and 10TG fibroblasts show identical DNase I footprinting profiles (protected regions and hypersensitive sites). Five protected regions and many hypersensitive sites were detected, nearly spanning the core enhancer (Figs 4, 5). The hypersensitive sites not associated with protected regions likely reflect lower affinity DNA/protein interactions, or binding of lower abundance proteins. Protected regions 4 and 5, which encompass

643

consensus binding sites for AP-1 (Lee et al., 1987) and H4TF1 (Dailey et al., 1988), respectively, are produced with nuclear extracts from all cell types tested, including 23A2 azamyoblasts, primary chicken liver cells and BNL liver cells (Fig. 5; data not shown). These sites represent the highest affinity sites and/or are bound by the most abundant factors because they typically exhibit complete protection, even under more stringent conditions (10 µg protein, 4 µg poly(dI/dC); non-specific competitor; data not shown). Protection of Eboxes was not observed, however E-box binding is difficult to detect using standard DNase I protection assays (Buskin and Hauschka, 1989). Both primary human fibroblasts (FC1010) and JEG-3 cell extracts yield cell-specific qualitative and quantitative differences in their footprinting profiles. Extracts from JEG-3 cells, a cell line in which the transfected enhancer is inactive (Goldhamer et al., 1992), does not show detectable DNase I protection over region 2, and only partial protection over region 1. Note also the absence of a hypersensitive site between regions 1 and 2, and just 5′ of region 4 in JEG-3 cell extracts (Fig. 5, forward strand). Also JEG-3 extracts yield partial protection of sequences between regions 2 and 3, and produced a hypersensitive site between regions 3 and 4 (Fig. 5, forward strand). Finally, protected region 3 extends slightly further 5′ with JEG-3 extracts than with the other extracts. FC1010 extracts exhibit partial or complete protection over all five regions, although DNase I hypersensitivity was reduced or absent at most sites denoted in Figs 4 and 5. In addition, protection over region 4 extends slightly further 3′ with FC1010 extracts (Fig. 5), suggesting a qualitative difference in DNA/protein interactions over this site. The endogenous myoD enhancer exhibits musclespecific DNase I hypersensitivity The identical DNase I protection profiles produced using nuclear extracts from non-myogenic 10TG cells (in which the endogenous gene is inactive), C2C12 myoblasts and HMP8 myoblasts suggest that the constellation of enhancer binding proteins are highly similar in these cell types. In addition, the myoD enhancer is active in 10TG cells when introduced by transfection (Goldhamer et al., 1992). These data raise the possibility that repression of the endogenous myoD gene in 10TG cells is mediated by epigenetic mechanisms, such as packaging into inactive chromatin, that could restrict accessibility of critical cis-acting enhancer sequences to positive trans factors. We used DNase I hypersensitivity as an indicator of chromatin structure to assess whether chromatin structural differences exist between myogenic and non-myogenic cells. For this analysis, the endogenous mouse myoD enhancer was assayed in C2C12 and 23A2 myogenic cells in addition to several nonmyogenic cell lines, while the endogenous human myoD enhancer was assayed in primary human fibroblasts (FC1010) and myoblasts (HMP8). Permeabilized cells were treated with increasing concentrations of DNase I, and the presence and position of hypersensitive sites determined by Southern blotting using indirect end-labeled probes. Three distinct hypersensitive sites are present in chromatin derived from C2C12 and 23A2 myoblasts, two of which map to the core enhancer, with the third mapping just 5′ of the enhancer (Fig. 6). In contrast, no hypersensitive sites were detected in chromatin from non-myogenic mouse cells, including 10TG

644


cells, BNL liver cells, or NB41A3 neuroblastoma cells (Fig. 6). Similarly, chromatin from primary human myoblasts exhibits DNase I hypersensitivity within the core enhancer, whereas chromatin from primary human fibroblasts is 5- to 10fold more resistant to DNase I digestion, although a very weak signal was detected (data not shown). These data provide evidence that the chromatin structure of the core enhancer is altered in myogenic cells, perhaps reflecting greater accessibility to trans factors.

DISCUSSION A distal core enhancer governs myoD gene activation in the embryo We have identified and characterized regulatory sequences responsible for myoD expression to begin to define upstream signaling pathways and transcriptional events that govern myoD activation in the embryo. Cis-acting sequences that regulate the embryonic expression of the human myoD gene were localized to a 258 bp core enhancer, which is located approximately 20 kb upstream from the start of myoD transcription (Goldhamer et al., 1992). A lacZ reporter gene, under the control of the myoD enhancer and promoter, is expressed in a muscle-specific and spatiotemporal pattern that coincides with the endogenous mouse myoD expression domain, as revealed by in situ hybridization (present study; Buckingham, 1992; Faerman and Shani, 1993; Faerman et al., in preparation). Transfection assays revealed a second, weak enhancer approximately 1 kb 3′ of the core enhancer, which may contribute to quantitative levels of myoD expression in vivo. This second region, however, in combination with the myoD promoter, was neither necessary nor sufficient for musclespecific expression of the lacZ transgene. We conclude that the 258 bp core enhancer mediates myoD activation in all embryonic skeletal muscle compartments, including somitic myotomes, limb buds and branchial arches. The relative contributions of the myoD core enhancer and promoter in directing myoD expression was tested by replacing the myoD promoter with the heterologous tk promoter. We showed that the myoD promoter is dispensable for activation of the myoD gene in the embryo, as this heterologous construct also yields muscle-specific expression of a lacZ reporter gene that closely resembles the endogenous pattern of myoD transcript accumulation. This experiment also establishes that myoD mRNA accumulation is dictated primarily by transcriptional control mechanisms rather than by mRNA turnover, because no transcribed human sequences were present in the heterologous promoter construct. Consistent with these functional data, the core enhancer is highly conserved in sequence Fig. 3. The 258 bp core enhancer directs the appropriate, myoD pattern of expression in transgenic mouse embryos. (A) β-gal expression of 11.5 d.p.c. whole-mount mouse embryo derived from a stable line harboring the construct, 258/−2.5lacZ. β-gal expression is detected in all myogenic compartments, including somites, limb buds, and branchial arches. The spatial pattern of lacZ-expressing cells is similar to that of larger enhancer constructs (see Fig. 1). (B) β-gal expression of 11.5 d.p.c. Fo whole-mount mouse embryo injected with the construct, 258/tklacZ. Expression with this heterologous promoter construct is observed in all myogenic compartments. Expression is most prominent in the dorsal portion of the most anterior somites and in the ventral portion of thoracic somites, similar to constructs with the myoD promoter. Somites posterior to the forelimb bud, however, show continuous lacZ expression along the dorsal-ventral myotomal axis, normally observed only in the most anterior somites. Ectopic, position effect staining in the brain and dorsal root ganglia is observed in this embryo. (C) β-gal expression of 11.5 d.p.c. Fo whole-mount mouse embryo injected with the E-box mutant construct, 258(E1-E3)/− 2.5lacZ. The pattern of β-gal expression is the same as that of wildtype constructs. Ectopic staining in the neural tube is observed in this embryo. (A-C) Arrows; β-gal expression in branchial arches. hb; hindlimb bud.

Regulation of the myoD gene in mouse embryos

645

Fig. 4. Sequence alignment of the human myoD core enhancer and the corresponding sequence of the mouse. Sequence similarity is 89% overall, and 94% in the first 160 bp. Data from DNase I protection assays (Fig. 5) is also summarized. The positions of DNase I protection (bars below sequence) and hypersensitive sites (asterisks above sequence; the JEG-3-specific hypersensitive site is denoted by a ‘+’) are shown. Not all of the hypersensitive sites shown are produced with FC1010 and JEG-3 nuclear extracts (see Fig. 5). Protected region 2 is not produced with JEG-3 nuclear extracts. Consensus sequence binding sites are shown for known transcription factors that reside within regions of protection, or that have been shown or postulated to function in muscle gene regulation.

(89% overall; 94% in the first 160 nucleotides) and position between humans and mice, whereas 5′ flanking sequences are only 66% similar within 250 bp of the start of transcription, with no extended regions of sequence similarity present upstream of the TATA box (unpublished observations). When

the tk promoter was used, however, the spatial patterning of transgene expression was partially disrupted; the transgene was expressed along the entire dorsal-ventral myotomal axis in all somites, regardless of axial position (see below). Thus, the myoD promoter may function to refine spatial domains of myoD expression within the somite myotome. Also, it is possible that the promoter regulates aspects of postnatal myoD expression; a mouse genomic clone consisting of the myoD promoter and a regulatory element at −5 kb (Tapscott et al., 1992) that shares no obvious sequence similarity with the core enhancer, appropriately confers transgene expression preferentially in fast glycolytic fibers of adult muscle (Hughes et al., 1993). Our results, however, clearly establish that neither the promoter nor the more proximal control element at −5 kb is required for muscle-specific activation of myoD in embryos.

myoD is subject to complex transcriptional regulation The extensive sequence similarity between the human and mouse myoD enhancer and the complex patterns of DNA/protein interactions observed in vitro indicate that myoD expression in the embryo is subject to complex regulation. DNase I protection assays identified five protected regions and eight regions with one or more hypersensitive sites (Figs 4, 5). Several hypersensitive sites are not associated with detectable protection, probably reflecting interactions with proteins of lower abundance or affinity. Only protected regions 4 and 5 encompass consensus binding sites for known factors; region Fig. 5. DNase I protection assay of the human myoD core enhancer using nuclear extracts from myogenic (C2C12, HMP8) and nonmyogenic (10TG, FC1010, JEG-3) cell types. Five protected regions (boxes) and multiple hypersensitive sites (asterisks; the JEG-3specific hypersensitive site is denoted by a ‘+’) were detected. JEG-3 extract does not protect region 2 and yields partial protection of region 1 (most apparent on forward strand). FC1010 and JEG-3 extracts exhibit substantially different patterns of DNase I hypersensitive sites compared to the other cell types. 10TG nuclear extract yields the same footprinting profile as C2C12 and HMP8 myoblast nuclear extracts. The forward strand corresponds to the sequence shown in Fig. 4. Protected region 1 on the reverse strand is not very apparent due to band compression near the top of the gel. BSA, control lanes in which bovine serum albumin replaced nuclear extract. G+A, purine-specific sequence ladder.

646


Fig. 6. Southern blot of DNase I treated chromatin, revealing muscle-specific DNase I hypersensitive sites in the endogenous myoD core enhancer. Chromatin from mouse myogenic cells (C2C12 and 23A2) exhibits three DNase I hypersensitive sites; two map to the core enhancer (arrows) and one maps just 5′ of the core enhancer (arrowhead). Non-myogenic mouse 10TG cells, NB41A3 neuroblastoma cells, and BNL liver cells do not exhibit DNase I hypersensitivity. The degree of DNAse I digestion of total DNA is comparable in all cell types, as determined by ethidium bromide staining of agarose gels. The relative position of the enhancer, and the restriction enzymes and probe used are shown diagrammatically.

4 includes a consensus AP-1 site, representing a potential target for fos/jun complexes (Curran and Franza, 1988), whereas region 5 contains a consensus sequence important in histone H4 gene expression (Dailey et al., 1988). While we do not know the molecular species bound to these sites, linkersubstitution mutants encompassing these regions adversely affect enhancer activity in transfection assays (Lukitsch and Goldhamer, unpublished observations). With the exception of 10TG cells (see below), extracts from non-myogenic cells yield footprinting profiles substantially different from those of myogenic cells. These differences observed with human primary fibroblasts (FC1010) and human choriocarcinoma cells (JEG-3) do not simply reflect human and mouse species differences because extracts from human primary myoblasts (HMP8) exhibit a footprinting profile similar to mouse C2C12 (and 23A2) myoblasts. Numerous qualitative and quantitative differences in hypersensitive sites exist with FC1010 and JEG-3 extracts, as detailed in the results. In addition, extracts from primary human fibroblasts yield a unique pattern of protection and hypersensitivity encompassing and adjacent to the consensus AP-1 site (Figs 4, 5). Thus, the protein species bound to this site may be unique to FC1010 cells. Recently, a hematopoietic-specific factor was shown to regulate globin gene expression by binding to a ‘ubiquitous’ AP-1 site within the β-globin locus control region (Andrews et al., 1993). The most dramatic difference between myogenic and non-myogenic binding activities, however, was the lack of protection over region 2 and only partial protection over region 1 with nuclear extracts from ectodermally derived JEG-3 cells (Figs 4, 5). As the transfected enhancer is inactive in JEG-3 cells (Goldhamer et al., 1992), region 2 may interact with tissue-restricted positive trans factors required for enhancer activity; whether the protein(s) is restricted to mesodermal cells will require further analysis.

Footprinting experiments and previous transfection data suggest that cis-acting epigenetic mechanisms repress myoD expression in 10TG cells. Inspection of footprinting gels reveals no differences in protection or hypersensitivity profiles produced from 10TG extracts and C2C12, HMP8 and 23A2 myoblast extracts (Fig. 5; data not shown). Assuming that less stable muscle-specific interactions were not missed in this assay, these data suggest that the complement of regulatory factors that bind the myoD enhancer are similar, if not identical, in 10TG cells and myogenic cells. In addition, the transfected enhancer is active in 10TG cells (Goldhamer et al., 1992), indicating that repression of the endogenous myoD gene is not a consequence of inactivation of these trans factors by post-translational modification or heterodimerization with negative regulators. We found, however, that the chromatin within and immediately surrounding the myoD core enhancer exhibits DNase I hypersensitivity in myogenic cells, but not in 10TG or other non-myogenic cells. We propose, therefore, that the myoD core enhancer is packaged in inactive chromatin in 10TG cells, rendering it inaccessible to positive trans factors required for gene activation. In theory, regulated, musclespecific changes in accessibility of trans factors to myoD control elements, mediated by chromatin structural changes or other epigenetic modifications such as DNA methylation, could be a critical prerequisite for myoD activation, with such changes being stably passed on to progeny cells (see Groudine and Weintraub, 1982; Razin and Cedar, 1991). In this regard, demethylation of specific CpGs in the mouse myoD enhancer correlates with myoD expression both in cell culture and in mouse embryos (Brunk et al., unpublished data). Analysis of the temporal relationship between DNA demethylation, changes in chromatin structure, and the expression of myoD will address whether these epigenetic modifications are a cause or an effect of myoD gene activation. E-boxes in the core enhancer are not required for myoD activation The presence of E-boxes in the myoD core enhancer raised the possibility that myoD activation is regulated by direct transactivation by other bHLH myogenic factors. Auto- and cross-regulatory interactions between the myogenic factors have been well-documented in cell culture systems (reviewed by Emerson, 1990; Olson, 1990). In addition, myogenin promoter function in embryos is E-box dependent, suggesting that myogenic factors regulate myogenin expression in vivo (Cheng et al., 1993; Yee and Rigby, 1993). We found, however, that enhancer constructs lacking the three conserved E-boxes (present study) or lacking all four E-boxes (see Fig. 4; unpublished observations), exhibit the wild-type pattern of expression at 11.5 d.p.c. In addition, both the wild-type and mutant myoD enhancers are activated in somites to the same caudal extent, indicating no delay in transgene activation. Although, we cannot formally rule out the possibility that Eboxes in promoter sequences functionally substitute for the enhancer E-boxes in this construct, this is unlikely because the myoD promoter shows no muscle specificity in transgenic mice when assayed alone (Goldhamer et al., 1992) or in combination with enhancer 2 (Fig. 2B), and is dispensable for musclespecific transgene expression. Thus, unlike many musclespecific genes (see below), activation of myoD is not likely to be mediated by E-boxes. Importantly, the present experiments

Regulation of the myoD gene in mouse embryos do not address the possible function of E-boxes in the maintenance of myoD expression, once activated; because β-gal is a relatively stable protein (see Paterson et al., 1991), E-box mutant constructs will need to be investigated at later stages of development further removed from the initial activation of myoD.

myoD expression is controlled by regulatory mechanisms distinct from other characterized muscle-specific genes myoD exhibits a distinct temporal and spatial pattern of expression (reviewed by Buckingham, 1992; Faerman and Shani, 1993; present results), indicating that regulatory mechanisms controlling myoD expression are unique. In the most anterior somites of the mouse, for example, the myogenic factors are activated sequentially over a 2.5 day period from 8 d.p.c. to 10.5 d.p.c. in the order; myf5, myogenin, MRF4 and myoD (Buckingham, 1992). Also, while myf5 (Tajbakhsh and Buckingham, 1994) and myogenin (Cheng et al., 1992; Yee and Rigby, 1993) follow a strict rostral to caudal sequence of activation in somites, myoD transcripts and myoD transgene expression are first detected in thoracic somites at the level of the forelimb bud, followed approximately a half day later by expression in more anterior somites (Faerman et al., in preparation). In addition, spatial domains of myoD expression within somites exhibit a distinctive axial position-dependent pattern of expression. In somites anterior to the forelimb bud, myoD transgene expression is most prominent in the dorsal region of the myotome, whereas in somites posterior to the forelimb bud, transgene expression is most prominent ventrally. This axial position-dependent patterning of expression is also observed for endogenous myoD transcripts. Finally, lacZ-positive cells are found scattered throughout the dorsoventral myotomal axis of the most anterior somites, whereas in somites posterior to the forelimb bud, dorsal and ventral populations of lacZpositive cells are separated by a spatially restricted population of myoD-negative myotomal cells. In contrast, myogenin and myf5 regulatory elements drive expression of lacZ along the entire dorsal-ventral myotomal axis regardless of axial position (Cheng et al., 1992; Yee and Rigby, 1993; Tajbakhsh and Buckingham, 1994). The embryological signals that dictate these unique expression patterns remain to be defined. Clear differences also are emerging in the organization and regulation of cis sequences that control expression of the myogenic genes, consistent with their distinct patterns of expression. While myoD is controlled by a distal enhancer 20 kb upstream of the gene, which is highly conserved between humans and mice, myogenin is regulated by highly conserved proximal promoter elements (compare sequences in Salminen et al., 1991 and Edmondson et al., 1992) within 200 bp of the start of transcription (Salminen et al., 1991; Cheng et al., 1992; Yee and Rigby, 1993). myf5 and myf6 are also regulated, at least in part, by 5′ flanking sequences (within 6 kb) as transgenes containing these sequences recapitulate some aspects of the expression pattern of the corresponding endogenous genes (Patapoutian et al., 1993). In addition, the myoD core enhancer exhibits no extended regions of sequence identity with the myogenin promoter or other muscle-specific enhancers or promoters (specific sequences controlling the embryonic expression of myf5 and MRF4 genes have not yet been defined). Furthermore, among the specific DNA motifs known

647

to regulate other muscle genes, including E-boxes, MEF-2 sites, CArG boxes, and MCAT sites, only E-boxes are present in the core enhancer. The E-boxes, however, are not required for myoD activation. In striking contrast, the myogenin promoter contains an E-box and a MEF-2 site, both of which are critical for normal myogenin expression in embryos (Cheng et al., 1993; Yee and Rigby, 1993). Together with possible epigenetic mechanisms regulating the myoD enhancer by control of chromatin structure and methylation status noted above, these data indicate that myoD is regulated by a control system distinct from the other characterized muscle-specific genes. Analysis of the regulation and expression of the myogenic factors has revealed heterogeneity among myotomal cell populations. Mutational studies of myogenin promoter function revealed MEF-2-dependent and -independent populations of cells (Cheng et al., 1993; Yee and Rigby, 1993). In the present study, we document the existence of spatially restricted populations of myoD-positive and myoD-negative myotomal cells. In addition, previous immunocytochemical studies identified myosin-positive cells that express either MyoD or Myogenin, neither protein, or both proteins (Cusella-De Angelis et al., 1992). Finally, Myf5 protein appears to be expressed by more cells than MyoD in cultures derived from somites (Smith et al., 1993), although individual cells have not been assayed for both markers. Whether these differential patterns of myogenic gene expression reflect different fates or developmental potentials among these cell populations is presently unclear. Given the formative but functionally redundant roles of myf5 and myoD in skeletal muscle formation, it will be important to determine the degree of concordance of myoD and myf5 expression in individual myogenic cells. This will help distinguish whether functional redundancy arises because myf5 and myoD have similar biochemical properties within the same cell or whether distinct myf5-positive and myoD-positive myogenic populations can regulate their size and compensate for the loss of the other (see Emerson, 1993; Weintraub, 1993). Using myoD enhancer-lacZ transgenes to track myoD expression domains in myf5 knockout mice will also address cellular compensatory mechanisms. Activation of myogenic gene expression is initiated very early in presumptive myotomal cells (Ott et al., 1991; Pownall and Emerson, 1992), whereas presumably committed myogenic cells migrate from the lateral somite (Ordahl and Le Douarin, 1992) into the developing limb before expression of the myogenic genes is detected (Sassoon et al., 1989; Cheng et al., 1992; Yee and Rigby, 1993; Tajbakhsh and Buckingham, 1994). This raises the questions of whether these distinct skeletal muscle lineages are established and maintained by similar mechanisms, and whether myogenic cell fate in presumptive limb muscle is governed by additional unknown factors. As the myoD core enhancer functions as a molecular target to integrate ‘upstream’ embryonic signals and activate myoD in all skeletal muscle lineages, functional dissection of the myoD enhancer will serve as a directed molecular approach to investigate upstream regulatory processes, including the mechanisms of skeletal muscle lineage determination in the embryo. We thank Yoshimichi Nakatsu for providing the mouse cosmid library, and Marisa Bartolomei and Kristen Lukitsch for critical comments on the manuscript. This work was supported by grants from

648


the American Cancer Society (IRG-135N) and the National Institutes of Health (NIH; AR-42644) to D. J. G., by a CORE grant (CA-06927) from the NIH and an appropriation from the Commonwealth of Pennsylvania, by research grants from the Muscular Dystrophy Association and the NIH (HD-07796) to C. P. E., and by a grant from the United States-Israel Binational Agricultural Research and Development Fund (BARD) to C. P. E. and M. S.

REFERENCES Allen, N. D., Cran, D. G., Reik, S. C. B. S. H. R. and Surani, M. A. H. (1988). Transgenes as probes for active chromosomal domains in mouse development. Nature 333, 852-855. Andrews, N. C., Erdjument-Bromage, H., Davidson, M. B., Tempst, P. and Orkin, S. H. (1993). Erythroid transcription factor NF-E2 is a haematopoietic-specific basic-leucine zipper protein. Nature 362, 722-728. Blackwell, T. K. and Weintraub, H. (1990). Differences and similarities in DNA-binding preferences of MyoD and E2A protein complexes revealed by binding site selection. Science 250, 1104-1110. Braun, T., Bober, E., Buschhausen-Denker, G., Kotz, S., Grzeschik, K. and Arnold, H.-H. (1989). Differential expression of myogenic determination genes in muscle cells: possible autoactivation by the Myf gene products. EMBO J. 18, 3617-3625. Braun, T., Bober, E., Winter, B., Rosenthal, N. and Arnold, H.-H. (1990). Myf-6, a new member of the human gene family of myogenic determination factors: evidence for a gene cluster on chromosome 12. EMBO J. 9, 821-831. Braun, T., Buschhausen-Denker, G., Bober, E., Tannich, E. and Arnold, H. H. (1989). A novel human muscle factor related to but distinct from MyoD1 induces myogenic conversion in 10TG fibroblasts. EMBO J. 8, 701-709. Braun, T., Rudnicki, M. A., Arnold, H.-H. and Jaenisch, R. (1992). Targeted inactivation of the muscle regulatory gene Myf-5 results in abnormal rib development and perinatal death. Cell 71, 369-382. Buckingham, M. (1992). Making muscle in mammals. Trends Genet. 8, 144148. Buskin, J. N. and Hauschka, S. D. (1989). Identification of a myocyte nuclear factor that binds to the muscle-specific enhancer of the mouse muscle creatine kinase gene. Mol. Cell. Biol. 9, 2627-2640. Cheng, T.-C., Hanley, T. A., Mudd, J., Merlie, J. P. and Olson, E. N. (1992). Mapping of myogenin transcription during embryogenesis using transgenes linked to the myogenin control region. J. Cell Biol. 119, 1649-1656. Cheng, T.-C., Wallace, M. C., Merlie, J. P. and Olson, E. N. (1993). Separable regulatory elements governing myogenin transcription in mouse embryogenesis. Science 261, 215-218. Church, G. M. and Gilbert, W. (1984). Genomic sequencing. Proc. Natl. Acad. Sci. USA 81, 1991-1995. Conlon, R. A. and Rossant, J. (1992). Exogenous retinoic acid rapidly induces anterior ectopic expression of murine Hox-2 genes in vivo. Development 116, 357-368. Couly, G. F., Coltey, P. M. and Le Douarin, N. M. (1992). The developmental fate of cephalic mesoderm in quail-chick chimeras. Development 114, 1-15. Cserjesi, P., Lilly, B., Bryson, L., Wang, Y., Sassoon, D. A. and Olson, E. N. (1992). MHox: a mesodermally restricted homeodomain protein that binds an essential site in the muscle creatine kinase enhancer. Development 115, 1087-1101. Cserjesi, P. and Olson, E. N. (1991). Myogenin induces the myocyte-specific enhancer binding factor MEF-2 independently of other muscle-specific gene products. Mol. Cell. Biol. 11, 4854-4862. Curran, T. and Franza Jr., B. R. (1988). Fos and jun: The AP-1 connection. Cell 55, 395-397. Cusella-De Angelis, M., et al. (1992). MyoD, myogenin independent differentiation of primordial myoblasts in mouse somites. J. Cell Biol. 116, 1243-1255. Dailey, L., Roberts, S. B. and Heintz, N. (1988). Purification of the human histone H4 gene-specific transcription factors H4TF-1 and H4TF-2. Genes Dev. 2, 1700-1712. Davis, R. L., Weintraub, H. and Lassar, A. B. (1987). Expression of a single transfected cDNA converts fibroblasts to myoblasts. Cell 51, 987-1000. Edmondson, D. G., Brennan, T. J. and Olson, E. N. (1991). Mitogenic repression of myogenin autoregulation. J. Biol. Chem. 266, 21343-21346.

Edmondson, D. G., Cheng, T. C., Cserjesi, P., Chakraborty, T. and Olson, E. N. (1992). Analysis of the myogenin promoter reveals an indirect pathway for positive autoregulation mediated by the muscle-specific enhancer factor MEF-2. Mol. Cell. Biol. 12, 3665-3677. Edmondson, D. G. and Olson, E. N. (1989). A gene with homology to the myc similarity region of MyoD1 is expressed during myogenesis and is sufficient to activate the muscle differentiation program. Genes Dev. 3, 628-640. Emerson Jr., C. P. (1990). Myogenesis and developmental control genes. Curr. Op. Cell Biol. 2, 1065-1075. Emerson Jr., C. P. (1993). Skeletal myogenesis: genetics and embryology to the fore. Curr. Opin. Genet. Dev. 3, 265-274. Faerman, A. and Shani, M. (1993). The expression of the regulatory myosin light chain 2 gene during mouse embryogenesis. Development 118, 919-929. Fire, A., White-Harrison, S. and Dixon, D. (1990). A modular set of lacZ fusion vectors for studying gene expression in Caenorhabditis elegans. Gene 93, 189-198. Goldhamer, D. J., Faerman, A., Shani, M. and Emerson Jr., C. P. (1992). Regulatory elements that control the lineage-specific expression of myoD. Science 256, 538-542. Gossett, L. A., Kelvin, D. J., Sternberg, E. A. and Olson, E. N. (1989). A new myocyte-specific enhancer-binding factor that recognizes a conserved element associated with multiple muscle-specific genes. Mol. Cell. Biol. 9, 5022-5033. Groudine, M. and Weintraub, H. (1982). Propagation of globin DNAase Ihypersensitive sites in absence of factors required for induction: a possible mechanism for determination. Cell 30, 131-139. Hasty, P., Bradley, A., Morris, J. H., Edmondson, D. G., Venuti, J. M., Olson, E. N. and Klein, W. H. (1993). Muscle deficiency and neonatal death in mice with a targeted mutation in the myogenin gene. Nature 364, 501-506. Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K. and Pease, L. R. (1989). Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene 77, 51-59. Hogan, B., Costantini, F. and Lacy, E. (1986). Manipulating the Mouse Embryo: A Laboratory Manual. Cold Spring Harbor Press, Cold Spring Harbor, NY. Hughes, S. M., Taylor, J. M., Tapscott, S. J., Gurley, C. M., Carter, W. J. and Peterson, C. A. (1993). Selective accumulation of MyoD and Myogenin mRNAs in fast and slow adult skeletal muscle is controlled by innervation and hormones. Development 118, 1137-1147. Konieczny, S. F. and Emerson Jr., C. P. (1984). 5-azacytidine induction of stable mesodermal stem cell lineages from 10TG cells: evidence for regulatory genes controlling determination. Cell 38, 791-800. Lee, W., Mitchell, P. and Tjian, R. (1987). Purified trancription factor AP-1 interacts with TPA-inducible enhancer elements. Cell 49, 741-752. Luckow, B. and Schutz, G. (1987). CAT constructions with multiple unique restriction sites for the functional analysis of eukaryotic promoters and regulatory elements. Nucleic Acids Res. 15, 5490. Mar, J. H. and Ordahl, C. P. (1990). M-CAT binding factor, a novel transacting factor governing muscle-specific transcription. Mol. Cell. Biol. 10, 4271-4283. Miner, J. H. and Wold, B. (1990). Herculin, a fourth member of the MyoD family of myogenic regulatory genes. Proc. Natl. Acad. Sci. USA 87, 10891093. Minty, A. and Kedes, L. (1986). Upstream regions of the human cardiac actin gene that modulate its transcription in muscle cells: presence of an evolutionarily conserved repeated motif. Mol. Cell. Biol. 6, 2125-2136. Murre, C., McCaw, P. and Baltimore, D. (1989). A new DNA binding and dimerisation motif in immunoglobulin enhancer binding, daughterless, MyoD, and myc proteins. Cell 56, 777-783. Nabeshima, Y., Hanaoka, K., Hayasaka, M., Esumi, E., Li, S., Nonaka, I. and Nabeshima, Y. (1993). Myogenin gene disruption results in perinatal lethality because of severe muscle defect. Nature 364, 532-535. Noden, L. (1991). Vertebrate cranial facial development. Brain Behav. Evol. 38, 190-210. Olson, E. N. (1990). MyoD family: a paradigm for development? Genes Dev. 4, 1454-1461. Ordahl, C. P. and Le Douarin, N. M. (1992). Two myogenic lineages within the developing somite. Development 114, 339-353. Ott, M.-O., Bober, E., Lyons, G., Arnold, H. and Buckingham, M. (1991). Early expression of the myogenic regulatory gene, myf-5, in precursor cells of skeletal muscle in the mouse embryo. Development 111, 1097-1107. Patapoutian, A., Miner, J. H., Lyons, G. E. and Wold, B. (1993). Isolated sequences from the linked Myf-5 and MRF4 genes drive distinct patterns of muscle-specific expression in transgenic mice. Development 118, 61-69.

Regulation of the myoD gene in mouse embryos Paterson, B. M., Walldorf, U., Eldridge, J., Dubendorfer, A., Frasch, M. and Gehring, W. J. (1991). The Drosophila homologue of vertebrate myogenic-determination genes encodes a transiently expressed nuclear protein marking primary myogenic cells. Proc. Natl. Acad. Sci. USA 88, 3782-3786. Pollock, R. and Treisman, R. (1990). A sensitive method for the determination of protein-DNA binding specificities. Nucleic Acids Res. 18, 6197-6204. Pollock, R. and Treisman, R. (1991). Human SRF-related proteins: DNAbinding properties and potential regulatory targets. Genes Dev. 5, 2327-2341. Pownall, M. E. and Emerson Jr., C. P. (1992). Sequential activation of three myogenic regulatory genes during somite morphogenesis in quail embryos. Devel. Biol. 151, 67-79. Prestridge, D. S. (1991). SIGNAL SCAN: A computer program that scans DNA sequences for eukaryotic transcriptional elements. CABIOS 7, 203206. Razin, A. and Cedar, H. (1991). DNA methylation and gene expression. Microbiol. Rev. 55, 451-458. Rhodes, S. J. and Konieczny, S. F. (1989). Identification of MRF4: a new member of the muscle regulatory factor gene family. Genes Dev. 3, 20502061. Rigaud, G., Roux, J., Pictet, R. and Grange, T. (1991). In vivo footprinting of rat TAT gene: dynamic interplay between the glucocorticoid receptor and liver-specific factor. Cell 67, 977-986. Rudnicki, M. A., Braun, T., Hinuma, S. and Jaenisch, R. (1992). Inactivation of MyoD in mice leads to up-regulation of the myogenic HLH gene Myf-5 and results in apparently normal muscle development. Cell 71, 383-390. Rudnicki, M. A., Schnegelsberg, P. N. J., Stead, R. H., Braun, T., Arnold, H.-H. and Jaenisch, R. (1993). MyoD or Myf-5 is required for the formation of skeletal muscle. Cell 75, 1351-1359. Salminen, A., Braun, T., Buchberger, A., Jurs, S., Winter, B. and Arnold, H.-H. (1991). Transcription of the muscle regulatory gene MYF4 is regulated by serum components, peptide growth factors and signaling pathways involving G proteins. J. Cell Biol. 115, 905-917. Sambrook, J., Fritsch, E. F. and Maniatis, T. 1989. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Press, Cold Spring Harbor, NY. Sassoon, D., Lyons, G., Wright, W. E., Lin, V., Lassar, A., Weintraub, H. and Buckingham, M. (1989). Expression of two myogenic regulatory factors myogenin and MyoD1 during mouse embryogenesis. Nature 341, 303-307. Shani, M. (1986). Tissue-specific and developmentally regulated expression of

649

a chimeric actin-globin gene in transgenic mice. Mol. Cell. Biol. 6, 26242631. Smith, T. H., Block, N. E., Rhodes, S. J., Konieczny, S. F. and Miller, J. B. (1993). A unique pattern of expression of the four muscle regulatory factor proteins distinguishes somitic from embryonic, fetal and newborn mouse myogenic cells. Development 117, 1125-1133. Sun, X.-H. and Baltimore, D. (1991). An inhibitory domain of E12 transcription factor prevents DNA binding in E12 homodimers but not in E12 heterodimers. Cell 64, 459-470. Tajbakhsh, S. and Buckingham, M. E. (1994). Mouse limb muscle is determined in the absence of the earliest myogenic factor myf-5. Proc. Natl. Acad. Sci. USA 91, 747-751. Tapscott, S. J., Lassar, A. B. and Weintraub, H. (1992). A novel myoblast enhancer element mediates MyoD transcription. Mol. Cell. Biol. 12, 49945003. Thayer, M. J., Tapscott, S. J., Davis, R. L., Wright, W. E., Lassar, A. B. and Weintraub, H. (1989). Positive autoregulation of the myogenic determination gene MyoD1. Cell 58, 241-248. Treisman, R. (1986). Identification of a protein-binding site that mediates transcriptional response of the c-fos gene to serum factors. Cell 46, 567-574. Wachtler, F. and Christ, B. (1992). The basic embryology of skeletal muscle formation in verterbrates: the avian model. Semin. Dev. Biol. 3, 217-227. Weintraub, H. (1993). The MyoD family and myogenesis: redundancy, networks, and thresholds. Cell 75, 1241-1244. Weintraub, H., et al. (1991). The myoD gene family: nodal point during specification of the muscle cell lineage. Science 251, 761-766. Wright, W. E., Binder, M. and Funk, W. (1991). Cyclic amplification and selection of targets (CASTing) for the myogenin consensus binding site. Mol. Cell. Biol. 11, 4104-4110. Wright, W. E., Sassoon, D. A. and Lin, V. K. (1989). Myogenin, a factor regulating myogenesis, has a domain homologous to MyoD. Cell 56, 607617. Xin, J. H., Cowie, A., Lachance, P. and Hassell, J. A. (1992). Molecular cloning and characterization of PEA3, a new member of the Ets oncogene family that is differentially expressed in mouse embryonic cells. Genes Dev. 6, 481-496. Yee, S.-P. and Rigby, P. W. J. (1993). The regulation of myogenin gene expression during the embryonic development of the mouse. Genes Dev. 7, 1277-1289. (Accepted 30 November 1994)