Two Myogenin-related Genes Are Differentially Expressed in Xenopus laevis Myogenesis and Differ in Their Ability to Transactivate Muscle Structural Genes*

Among the myogenic regulatory factors, myogenin is a transcriptional activator situated at a crucial position for terminal differentiation in muscle development. It is unclear at present whether myogenin exhibits unique specificities to transactivate late muscular markers. During Xenopus development, the accumulation of myogenin mRNA is restricted to secondary myogenesis, at the onset of the appearance of adult isoforms of (cid:1) -tro-pomyosin and myosin heavy chain. To determine the role of myogenin in the isoform switch of these contractile proteins, we characterized and directly compared the functional properties of myogenin with other myogenic regulatory factors in Xenopus embryos. Two distinct cDNAs related to myogenin, XmyogU1 and XmyogU2, were differentially expressed during myogenesis and in adult tissues, in which they preferentially accumulated in oxidative myofibers. Animal cap assays in Xenopus embryos revealed that myogenin, but not the other myogenic regulatory factors, induced expression of embryonic/larval isoforms of the (cid:1) B , the PCR products were submitted to Southern blot detection of each Xenopus myogenin transcript with primers hU1, hU2, and hU12, specific to XmyogU1 ( U1 ), XmyogU2 ( U2 ), and both sequences ( U1 (cid:5) U2 ), respectively. The h3 primer was specific to the 3 (cid:2) -end sequence of long Xenopus myogenin transcripts ( 3 (cid:2) ). Increasing concentrations (2, 4, and 8 ng) of XmyogU1 and XmyogU2 plasmid PCR products were used to check the hybridization reaction specificity ( CU1 and CU2 ). In the adult study ( C ), the Southern blot was hybridized with radiolabeled S1 primer. Because the PCRs were not performed under the same conditions, no relevant comparison could be drawn on the basis of the hybridization signal intensity between adult muscle and/or regenerating or developing muscle.

The members of the MyoD gene family, including myoD (1), myogenin (2), myf5 (3), and MRF4 (4,5), encode myogenic transcription factors (MRFs) 1 able to convert non-muscle cells to a muscle phenotype in culture (6,7). First identified in mammals, all four proteins share a highly conserved central region related to the basic helix-loop-helix domain of the c-Myc superfamily. The MRFs are specifically expressed in skeletal muscle, with non-identical patterns of expression (8), and form a regulatory network controlling muscle determination and/or differentiation. Experiments using various MRF knockout mice have progressively elucidated the hierarchical relationships among the MRFs and established that functional redundancy is a feature of the MRF regulatory network. Thus, MyoD and Myf5 play overlapping roles in myoblast specification, whereas myogenin and either MyoD or MRF4 are required for differentiation (9). However, the redundant functions of MyoD and MRF4 appear not to overlap with those of myogenin (10). Knockout mice lacking the myogenin gene die at birth due to severe muscle deficiency, despite normal levels of MyoD and Myf5 (11,12). Recent studies have shown that the role of myogenin in muscle formation is distinct from that of MyoD and that this difference is due to functional specialization, and not just regulation of expression (13,14).
Surprisingly little is known about the identity of muscle genes that are selectively activated by myogenin and that cannot be activated by the other myogenic regulators. Indeed, all the MRFs can transactivate muscle-specific gene expression by interacting with the consensus nucleotide motif, CANNTG, also called the E box, and they are all believed to bind to their target sequences as heterodimers with members of the E-protein family (E12/E47). Several genes, including those encoding muscle creatine kinase, myosin light chain 1/3, and acetylcholine receptor subunits (15,16), were shown to be activated by MRFs. Although the MRFs showed different binding affinities for the promoter sequences and different transactivation potential (17), the concept of individual sets of genes each placed under the control of a specific MRF has not been substantiated.
In this context, we wondered whether myogenin is able to transactivate muscle structural genes that are not activated by the other MRFs. For this question, early Xenopus embryos may provide a convenient model. Xenopus homologs of MyoD (18,19), Myf5 (20), and MRF4 (21) have been identified, and their temporal and spatial expression patterns during early embryogenesis have been analyzed (21,22). In vitro and in vivo analysis indicated that Xenopus Myf5 and MyoD transcripts can induce a myogenic program following ectopic microinjection (18,23,24). Only a partial genomic clone of Xenopus myogenin has been isolated (21); and more recently, we reported that, in Xenopus, the accumulation of myogenin mRNA is restricted to secondary myogenesis, including 1) muscle regeneration (25), 2) the formation of new muscles in developing limbs, and 3) formation of new dorsal muscles during body remodeling at metamorphosis (26). At the onset of metamorphosis, the primary myotomal myofibers die and are progres-sively replaced by secondary multinucleated myofibers arising from fusion of recently migrated adult-type myoblasts. Interestingly, these new myofibers express myogenin (26) as well as adult isoforms of muscle contractile proteins, adult-type myosin heavy chain (MHC) and ␤-tropomyosin (␤-TM) (26,27), making Xenopus highly suitable for establishing the basis for myogenin's specific function in the differentiation of skeletal muscle.
In this report, we describe the cloning of two cDNA homologs of myogenin in Xenopus. These two cDNAs, designated XmyogU1 and XmyogU2, code for proteins closely related to each other and are likely to be expressed from duplicated Xenopus myogenin genes. We show that these two isoforms are differentially expressed during development and muscle regeneration, suggesting a functional difference. We found that, following ectopic expression and animal cap assay in Xenopus embryos, XmyogU1 and XmyogU2 display differential abilities to transactivate muscle structural genes.

EXPERIMENTAL PROCEDURES
Animals-Xenopus laevis adults were maintained at 22°C in tap water and fed once a week.
Muscle Injury and Sampling-Animals were anesthetized with tricaine methanesulfate (MS-222), and pure cardiotoxin from Naja mossambica nigricollis venom (10 Ϫ5 M in 0.9% NaCl; Latoxan, Rosans, France) was injected into the right anterior brachial muscle of the fore limb and into the right posterior crural muscle of the hind limb as described by Nicolas et al. (25).
RACE/PCR-Total RNA from a brachial muscle and a crural muscle after 15 days of regeneration was prepared using the RNeasy kit (QIA-GEN Inc., Valencia, CA). 5Ј-and 3Ј-RACE adaptor-ligated, oligo(dT)primed cDNAs were prepared using the Marathon cDNA amplification kit (CLONTECH, Palo Alto, CA). PCRs were performed using the KlenTaq enzyme mixture (CLONTECH). The oligonucleotide primers are detailed in Table I. For cDNA cloning, PCR products were cloned into the pGEM-T vector (Promega, Madison, WI). For Southern blot analysis, the PCR products were separated on 1% alkaline agarose gels, transferred onto Hybond-N nylon membrane (Amersham Biosciences, Inc.), and hybridized overnight at 30°C with 32 P-labeled 15-mer primers and at 50°C with 32 P-labeled 20-mer primers. The primers hU1, hU2, hU12, and h3 (see Table I), specific for XmyogU1 cDNA, XmyogU2 cDNA, both sequences, and the 3Ј-end of the long form of Xenopus myogenin transcripts, respectively, were 32 P-labeled at their 3Ј-ends by incorporation of [ 32 P]dCTP using terminal transferase (Invitrogen) according to the manufacturer's recommendations. The blots were washed twice at room temperature with buffer containing 2ϫ SSC and 0.1% SDS and once with 0.1ϫ SSC and 0.1% SDS at room temperature for the blots probed with the 15-mer primers and at 50°C for those probed with the 20-mer primers. Signals were detected by autoradiography.
Sequence Determination and Comparison-Plasmids were sequenced (Sequenase Version 2.0, Amersham Biosciences, Inc.) in both directions by primer walking. The final sequence encoding the two myogenin isoforms represents the consensus sequence from eight overlapping cDNAs. The sequences were aligned at the DNA and protein levels using the LALIGN program (28).
In Vitro Translation-In vitro transcription/translation reactions were completed using a TNT kit (Promega) as specified by the manufacturer. For each reaction, 1 g of plasmid DNA was used in a 500-l total reaction volume. A reaction without addition of plasmid DNA was performed as a control. Reactions were supplemented with translationgrade [ 35 S]methionine (PerkinElmer Life Sciences, Brussels, Belgium). Then, 10 l of each reaction was diluted at 50% in a loading dye and separated on a 10% SDS-polyacrylamide gel. Radioactive components were detected by fluorography, and protein bands were visualized by Coomassie Blue staining.
DNA Isolation and Southern Analysis-Genomic Southern blots, made essentially as described by Scales et al. (19), were probed with a 480-bp (nucleotides 1-480) cDNA fragment from the coding region of XmyogU2 labeled with 32 P by a random primer labeling technique (Amersham Biosciences, Inc.).
RT-PCR Analysis-First-strand cDNAs were synthesized from 1 g of total RNAs by oligo(dT) priming using Superscript reverse transcriptase (Invitrogen) at 37°C for 1 h. PCR assays were performed as described elsewhere (26). Controls for PCR and hybridization were performed using plasmids carrying XmyogU1 and XmyogU2 cDNAs as templates. Increasing concentration of control PCR products (2, 4, and 8 ng) were analyzed as described above. When enzyme digestion was performed to discriminate between XmyogU1 and XmyogU2, StyI (New England Biolabs Inc.) was used to digest the PCR fragments amplified with F1 and Ro (see Table I).
In Situ Hybridization-A sample of the iliofibularis muscle was removed from a Xenopus female adult. The tissue was washed in water at room temperature and frozen in liquid nitrogen. Sections 10 m thick were cut in a cryostat at Ϫ22°C, collected on slides treated with Vectabond (Vector Laboratories, Inc., Burlingame, CA), air-dried for 10 min at room temperatures, and stored at Ϫ80°C until used. Xenopus myogenin template is the 522-nucleotide fragment corresponding to the partial genomic DNA sequence isolated by Jennings (21), subcloned into pGEM4Z (Promega), linearized with HindIII (Roche Molecular Biochemicals), and transcribed using SP6 RNA polymerase (Promega). cRNA probes were used as described by Nicolas et al. (25).
Succinate Dehydrogenase Histochemistry-The succinate dehydrogenase activity of 10-m-thick cross-sections of the unfixed part of the iliofibularis muscle was determined as described by Pool et al. (31) to establish the oxidative capacity in individual muscle fibers. Sections were incubated for 10 min at 37°C in medium consisting of 0.4 mM nitro blue tetrazolium (Sigma) and 75 mM sodium succinate (Sigma) in 37.5 mM sodium phosphate buffer (pH 7.6). The staining intensity obtained by this method correlates with the maximal rate of oxygen consumption of single muscle fibers of X. laevis (32).
Embryo Injection and Animal Cap Assay-Xenopus embryos were placed in 1ϫ Niu-Twitty medium with 4% Ficoll and injected bilaterally with mRNA at the two-cell embryo stage. Injected embryos were transferred to 0.1ϫ Niu-Twitty medium with antibiotics.

RACE/PCR, cDNA Cloning, and Sequence Determination of
Xenopus Myogenin-To obtain a full-length cDNA of myogenin in Xenopus, 5Ј-and 3Ј-RACE reactions were performed using regenerating muscle mRNAs and primers designed from the partial genomic sequence of Xenopus myogenin (21) (Fig. 1A).
During regeneration of Xenopus skeletal muscle, the forming myotubes produce a large amount of myogenic factors, including myogenin (25). Thus, total RNA was extracted from brachial muscles at 15 days of regeneration and a "Marathon cDNA" population was synthesized.
In the first round of 3Ј-RACE/PCR, Xenopus myogenin-spe- . Identical nucleotides are indicated by dots. Dashes represent gaps introduced for optimal sequence alignment. The sequences were aligned using the LALIGN software (28). C, in vitro translation of XmyogU1 and XmyogU2 cDNAs. Proteins were synthesized in a reticulocyte lysate system using [ 35 S]methionine as a labeled precursor with plasmid cDNA corresponding to XmyogU1 or XmyogU2 and no added cDNA (U1, U2, and C, respectively). Translation products were analyzed by SDS-PAGE. Protein molecular mass markers are indicated in kilodaltons. cific primer (S1) and an adaptor primer (AP1) yielded two PCR fragments of 680 and 920 bp in length that were cloned into pGEM-T and named XmyogU1 and XmyogU2, respectively. Sequencing confirmed that the two fragments were 3Ј-extensions of Xenopus myogenin, including a poly(A) tail. Comparison with the partial sequence of Xenopus myogenin previously characterized (21) showed that the sequence of XmyogU2 (680 bp) matched exactly the 3Ј-end of the genomic clone of Xenopus myogenin, whereas the sequence of XmyogU1 (920 bp) was only 97% similar to this sequence. RACE fragments displayed 90% sequence homology to each other. These data indicated also that the difference in size between the two RACE fragments was due to an additional sequence of 240 bp at the 3Ј-end of XmyogU1. Additional specific RT-PCR led us to isolate a long form of XmyogU2 that contains a sequence of 240 bp at its 3Ј-end. The XmyogU2 sequence contains two consensus motifs for a poly(A) tail signal (5Ј-AATAAA-3Ј) at positions 1054 and 1328, whereas the XmyogU1 sequence contains only the second poly(A) tail signal at position 1349. These results suggest the use of at least two different poly(A) tail signals in XmyogU2, generating two transcripts with different 3Ј-ends.
Because these data strongly suggest that two different transcripts related to myogenin are actually expressed in Xenopus muscle, we decided to characterize the entire coding regions of both XmyogU1 and XmyogU2. For this purpose, a second set of primers was designed based on the 3Ј-end of each fragment (Fig. 1A) to allow the specific amplification of XmyogU1 or XmyogU2. 5Ј-RACE/PCRs were performed on the Marathon cDNA using the XmyogU1-and XmyogU2-specific primers (rU1 and rU2, respectively) and the adaptor primer (AP1). Sequence data showed that the resultant RACE fragments were two isoforms of the 5Ј-coding region of Xenopus myogenin cDNA. As expected, XmyogU2 displayed a 5Ј-region identical to the genomic sequence of Xenopus myogenin previously described (21).
To obtain reliable sequences, amplification of the entire cDNA was performed with a Pfu polymerase. A set of fU1 and fU2 primers was designed based on the 5Ј-end of each 5Ј-RACE/ PCR fragment and was used in combination with the 3Ј-reverse primers (rU1 and rU2). This third round of RT-PCR generated fragments of 880 bp.
The two cDNAs (represented in Fig. 1B) are 1370 bp (XmyogU1) and 1349 bp (XmyogU2) long (excluding the poly(A) tail). Nucleic sequence alignment showed 90.7% identity between the two Xenopus myogenins. The two putative coding sequences are 95.3% identical at the DNA level. In the 3Јuntranslated region, the two cDNAs are 84% identical.
Both the XmyogU1 and XmyogU2 cDNAs contain a unique open reading frame, encoding predicted polypeptides of 235 and 236 amino acid residues, respectively. The two myogenins potentially initiate at two different start codons. The predicted sequences of these two polypeptides are 95.3% identical. The amino acid differences are distributed throughout the coding region, but are more numerous in the N-and C-terminal regions. The helix-loop-helix region is conserved completely.  In Vitro Translation of XmyogU1 and XmyogU2-XmyogU1 and XmyogU2 display sequence differences in the 5Ј-region that predominate around the initiation site of translation. To demonstrate the ability of the two cDNAs to encode proteins, the pGEM-T plasmids containing the complete coding sequence and the 5Ј-flanking region of each myogenin cDNA (nucleotides 1-880) were used in a coupled in vitro transcription/translation system to make [ 35 S]methionine-labeled translation products. Fig. 1C shows the separation of in vitro translation products by SDS-PAGE. Each cDNA generated a single product with an apparent molecular mass of 40 kDa. The difference between the molecular mass of 25 kDa predicted from the coding sequence and the estimated mass of 40 kDa seen after translation has already been observed for in vitro synthesized myogenin in the mouse (36).
Evidence for Distinct Genes-Several pieces of evidence suggest that XmyogU1 and XmyogU2 are encoded by distinct genes. First, X. laevis contains a pseudotetraploid genome (37), making genomic duplication a likely mechanism for producing the two myogenin messengers. Second, the differences between the two sequences are widely scattered throughout the sequence, including the coding sequence. Third, probing a genomic Southern blot with a cDNA expected to hybridize with both sequences consistently produced at least two bands (Fig.  2). Three hybridizing bands of unequal size were detected after digestion with PstI, which cut once within the XmyogU1 coding region. Taken together, our results suggest that these two transcripts are very likely to be the products of different genes.
Muscle-restricted Expression of the Two Myogenin Genes-The expression of myogenin mRNAs in Xenopus tissues was investigated by Northern blotting. Total RNA was extracted from multiple adult tissues, including adult muscles and different stages of regeneration of the brachial and crural muscles. The Northern blots were probed with the 3Ј-region (930-bp fragment, nucleotides 440 -1370) of the XmyogU1 cDNA. Rep-robing the blots with the entire coding region gave the same results (data not shown).
Three patterns of hybridization signals were observed for myogenin expression: two RNA populations, only one RNA population, or no myogenin mRNA. In the brachial muscle, although no hybridization signal was observed in normal adult muscle (data not shown; see Ref. 26), two mRNAs of ϳ1.3 and 1.5 kb were detected during muscle regeneration (Fig. 3A). The levels of expression of these two messengers were similar. In the crural muscle, only the 1.3-kb mRNA was clearly detected in the regenerating muscle, although two transcripts of 1.3 and 1.5 kb were found in the adult muscle (Fig. 3B). The levels of expression of the two myogenin mRNAs were lower in normal adult muscles than during regeneration. Finally, no hybridization signal was detected in tissues other than skeletal muscles, including the heart (data not shown).
Identification of the Two mRNA Species of Myogenin-To determine which of the mRNAs detected by Northern blot analysis corresponds to XmyogU1 and which to XmyogU2, we performed 3Ј-RACE experiments. Total RNA were extracted from a brachial muscle and a crural muscle, both after 15 days of regeneration, and Marathon cDNA populations were synthesized. 3Ј-RACE/PCR was performed on each Marathon cDNA using S1 and the adaptor primer (AP1). The resulting Southern blots were probed with the radiolabeled primers hU1, hU2, hU12, and h3 (Table I), which hybridized with XmyogU1, XmyogU2, both sequences, and the 240-bp 3Ј-extension of the longest myogenin transcripts, respectively (Fig. 4A). The hybridization data clearly showed different patterns of expression for the two myogenin isoforms. Although strongly expressed during regeneration of the brachial muscle, XmyogU1 was faintly detected during regeneration of the crural muscle. In contrast, XmyogU2 displayed similarly strong levels of expression during regeneration of both muscles. A faint hybridization signal corresponding to the long isoform of XmyogU2 was also Additional tests were performed with regenerating crural (cru) and brachial (bra) muscles at 15 days of regeneration and with dorsal muscle at developmental stage 56 (st56). Myogenin sequences were amplified using the F1/Ro primers, aiming for the indiscriminate amplification of XmyogU1 and XmyogU2 cDNAs. The PCR products were digested with StyI, allowing for selective discrimination between XmyogU1 (lower bands) and XmyogU2 (upper bands). In all cases, the PCR experimental conditions were designed to avoid the saturation of PCR. In A and B, the PCR products were submitted to Southern blot detection of each Xenopus myogenin transcript with primers hU1, hU2, and hU12, specific to XmyogU1 (U1), XmyogU2 (U2), and both sequences (U1ϩU2), respectively. The h3 primer was specific to the 3Ј-end sequence of long Xenopus myogenin transcripts (3Ј). Increasing concentrations (2, 4, and 8 ng) of XmyogU1 and XmyogU2 plasmid PCR products were used to check the hybridization reaction specificity (CU1 and CU2). In the adult study (C), the Southern blot was hybridized with radiolabeled S1 primer. Because the PCRs were not performed under the same conditions, no relevant comparison could be drawn on the basis of the hybridization signal intensity between adult muscle and/or regenerating or developing muscle. detected in the brachial muscle, but not in the crural muscle. No short form of XmyogU1 was detected.
Comparison of these data with those obtained by Northern blot analysis permitted us to identify the transcripts encoded by the XmyogU1 and XmyogU2 genes. The larger RNA was composed of XmyogU1 plus a minor population of XmyogU2, and the smaller RNA was composed only of XmyogU2. It was noted that the difference in size corresponded to the difference between the sequences of the 3Ј-noncoding ends. This result suggests that at least two different poly(A) tail signals are used in vivo for XmyogU2, generating two different 3Ј-ends depending on the muscle, and that only one poly(A) tail signal is used for XmyogU1. These results indicate that a specific control governs the expression of each myogenin gene according to the muscle type.
Expression of the Two Myogenin Genes during Development-Because these data substantiate the hypothesis of a dynamic pattern of expression specific to each myogenin gene, we investigated by RT-PCR whether XmyogU1 and XmyogU2 are differentially induced during development. The results shown in Fig. 4B reveal different patterns of expression for the two genes. During development, XmyogU1 was not expressed, whereas XmyogU2 displayed a biphasic pattern of expression. It was possible to detect a progressive increase in the XmyogU2 transcription level from stages 51 to 56 at the onset of myoblast fusion. After stage 56, XmyogU2 expression decreased, and after stage 60, a faint XmyogU1 signal was observed. These data suggest that the two myogenin genes are differentially expressed, with one isoform (XmyogU2) expressed during development and in the adult and the other isoform (XmyogU1) expressed only in adult muscles.
Restricted Expression of Myogenin in Oxidative Myofibers of Adult Muscles-To determine the pattern of expression of each Xenopus myogenin isoform in adult muscles, RT-PCR and in situ hybridizations were performed using adult muscles, including cruralis, gluteus maximus, adductor brevis, semimembranosus, iliofibularis, tibialis anterior, tibialis posterior, gastrocnemius, and sartorius. The two myogenin mRNAs were detected in all adult muscles studied, as evidenced by the results of RT-PCR/restriction enzyme digestion experiments that discriminated between XmyogU1 and the XmyogU2 expression (Fig. 4C). Our data provide evidence for the persistent expression of Xenopus myogenin mRNA in many Xenopus adult muscles.
To determine in which type of myofibers myogenin mRNA is expressed, in situ hybridizations were performed in a Xenopus adult muscle. We chose to analyze iliofibularis muscle, the most widely studied muscle in the frog. Diameter determination and succinate dehydrogenase activity allowed us to identify the five different fiber types, referred to as types 1-5 (38). Histochemical and functional studies clearly indicated that type 1-3 fibers are fast twitch fibers, whereas type 5 fibers are slow tonic fibers; yet classification of type 4 fibers in the fast or slow group remains a matter of speculation (38). We showed here that the small fibers with the lowest succinate dehydrogenase activity (type 5 fibers) and the large fibers with low succinate dehydrogenase activity (type 1 fibers) did not react with myogenin cRNA probes (Fig. 5). In contrast, the three types of myofibers characterized by high succinate dehydrogenase activity (type 2-4 fibers) strongly expressed myogenin mRNA, with the small type 3 fibers with the higher succinate dehydrogenase activity being the myofibers expressing the higher levels of myogenin transcripts. These results show that only oxidative myofibers express myogenin mRNA in Xenopus adult muscles.
XmyogU1 and XmyogU2 Differ in Their Abilities to Activate the MHC Adult Isoform-We used the Xenopus animal cap assay to compare the specificity of each MRF for the transactivation of muscle structural genes. In this assay, synthetic Xenopus MyoD, Myf5, MRF4, XmyogU1, XmyogU2, or elongation factor-1␣ mRNA was bilaterally injected into the animal pole of two-cell stage embryos. At the blastula stage, animal FIG. 5. Myogenin mRNA localization in adult iliofibularis muscle. In situ hybridization using antisense riboprobes to Xenopus myogenin (A and C) and succinate dehydrogenase histochemistry (B and D) were carried out on serial transverse sections. Dark-field photomicrographs are shown (A and C). Using sense riboprobes, we did not detect hybridization signals (data not shown). Numbers identify the five different fiber types, referred to as types 1-5 (38). poles were dissected and cultured until the late gastrula stage for molecular analysis. Normally, cultured animal pole explants form epidermis, but they can be converted to other cell fates following injection of appropriate differentiation factors. We chose to analyze animal caps by RT-PCR for the differential expression of the Xenopus cardiac ␣-actin gene, the first muscle marker (8), and for the differential expression of two isoforms of muscle structural genes, the embryonic/larval and adult isoforms of the MHC (34) and ␤-TM (35) genes.
The cardiac ␣-actin gene was found to be activated in all the animal caps, except in the Xenopus elongation factor-1␣ control (Fig. 6). This result is consistent with data previously obtained with Xenopus MyoD and Myf5 (20). We examined the expression of embryonic/larval isoforms of MHC and ␤-TM and found that both were activated by Xenopus myogenin, but not by Xenopus MyoD, Myf5, or MRF4.
We then examined transactivation of the adult isoforms of MHC and ␤-TM. As is evident from Fig. 6, mRNA for the A7 isoform of MHC was detected only in the XmyogU1-injected samples, although expression of the adult isoform of the ␤-TM gene was not induced. DISCUSSION A clue to the mechanism of the transcriptional activity of myogenic factors was provided by the evidence of conserved structural domains, the basic helix-loop-helix domain, responsible for dimerization and DNA binding (39,40), and specific domains that play important roles in dictating the unique specificities of each factor in the activation of specific sets of genes (41) such as the Pbx-Meis/Prep1 domain (42) and the helix III domain (14) in myogenin (Fig. 7). Alignment of the Xenopus sequences with chick, Cyprinus carpio, rat, and human myogenins revealed additional conserved domains that are candidate regions for specific transactivation. A tryptophan residue (Trp 186 ) is actually conserved in all six proteins, within a region that also exhibits a remarkable homology between the Xenopus and other species sequences. Of note, within the conserved Pbx-Meis/Prep1 motif in myogenin, cooperative interac-tions were reported to be mediated by an essential Trp residue (42). Xenopus myogenins also present a serine-rich domain (between residues 170 and 180) within the C-terminal region that is highly conserved among species. Interestingly, Pelpel et al. (43) have recently pointed out a serine residue in this transactivation domain as a potential target for phosphorylation by a protein kinase of the Mos family. In the myogenic factor MyoD, phosphorylation of Ser 237 by Mos regulates the activity of MyoD during the course of muscle differentiation (43). In fact, several studies emphasize the role of phosphorylation of serine/threonine-rich sites in regulating the transcriptional activity of myogenin heterodimeric complexes (44,45). One striking example of such a regulation is the protein kinase C-dependent inhibition of myogenin activity that involves the direct phosphorylation of a conserved threonine residue located within the basic domain of the protein (46). It will be of interest to determine whether the C-terminal serine-rich region of Xenopus myogenin is a target for phosphorylation of the protein that might affect its transcriptional activity.
The two forms of myogenin in X. laevis show remarkable similarity in size and sequence, but the position, nature, and clustering of the amino acid differences between XmyogU1 and XmyogU2 may account for functional specificity. Notably, substitutions involving proline at positions 12 (Ser for Pro) and 229 (Pro for Thr) are concentrated at both extremities of the proteins in the transactivation domains (41). Prolines have the capacity to twist the shape of the protein away from a helical configuration or out of a ␤-sheet. Therefore, the two myogenins may adopt different conformations that still allow interaction with regulatory proteins, but that could modify their activity or potential gene targets.
Searching for expression of XmyogU1 and XmyogU2 in different contexts, including development, muscle regeneration, and adult muscles, has provided evidence for differential expression of the two Xenopus myogenin genes. The expression patterns of Xenopus myogenin were complicated by the structural heterogeneity of XmyogU2, which presents two different  (14) are underlined. In the C-terminal region, the residues shown in boldface are conserved in at least six proteins. The sequences were aligned using the LALIGN software (28). X-U1, XmyogU1; X-U2, XmyogU2. 3Ј-ends. Interestingly, this structural heterogeneity is dependent on the muscle. Because the structure of the mRNA 3Ј-end is thought to play a role in determining the half-life of the molecule, it will be of interest to investigate the functional significance of this phenomenon.
At the cellular level, the restricted accumulation of Xenopus myogenin mRNA in muscle cells that present a predominantly oxidative metabolism shows an expression profile expected of a protein that is part of the phenotype maintenance of these cells. This hypothesis is substantiated by the fact that myogenin can induce a shift from glycolytic to oxidative metabolism in muscle fibers in which it is overexpressed (47). These observations led to the concept that myogenin may play two different roles depending on the differential status of the cells in which it was activated, the terminal differentiation of myoblasts during myogenesis, and the control of oxidative metabolism in fully differentiated myofibers.
For all that is known about MRF gene expression patterns, structural features of the proteins, and post-translational modifications, several important issues remain unanswered regarding our understanding of how the MRFs function. Different molecular mechanisms could account for the functional specification of the MRFs and explain their implication in myoblast determination and/or in myotube terminal differentiation. First, it may be that each MRF targets a specific array of genes with a high degree of specificity. This model accounts for the spatial, temporal, and quantitative pattern of expression specific for each MRF and the fact that, as evidenced by the myogenin Ϫ/Ϫ mutant (11), myogenin plays a unique role in muscle differentiation. But this function of myogenin is in complete contrast to the functions of Myf5, MyoD, and MRF4, none of which is needed for formation of skeletal muscle in mice (9). Another model was recently proposed on the basis of structural differences between the MyoD and myogenin proteins (14), according to previous studies that emphasized the abilities of determination factors to remodel the chromatin at the promoter of silent muscle genes (48). In this model, the determination factors initiate the expression of an array of genes that are subsequently enhanced and maintained in transcription by the differentiation factors, permitting the full level of gene expression needed for terminal differentiation. Finally, Valdez et al. (10) have propose an alternative model in which each MRF has evolved a specialized as well as a redundant role in muscle differentiation, depending on the promoter structure of the target muscle gene.
Thus, in this study, we have shown that the expression of embryonic/larval MHC and ␤-TM genes is activated by myogenin and not by the other MRFs. Interestingly, the expression of the adult fast isoform of MHC is induced by XmyogU1 and not by XmyogU2. These results suggest that, despite the high degree of structural homology between the two isoforms of myogenin, i.e. only 11 different amino acid residues, XmyogU1 and XmyogU2 target different sets of genes. These findings substantiate the hypothesis of the existence of a specificity of MRFs in muscle gene targeting and favor the model of Valdez et al. (10), who predicted that some muscle genes could discriminate among MRFs for their transactivation.
These results also provide evidence for an alternative pathway in the activation of MHC and ␤-TM genes during early Xenopus development. Indeed, the expression of these structural genes is not correlated with the presence of their specific activator because the embryonic/larval isoforms of MHC and ␤-TM appear at stages 21 and 25, respectively, when there is still no myogenin expression, and the adult isoforms of MHC appear at stage 52, when only XmyogU2 is expressed. This findings are consistent with previous studies showing that the relationship between MRF and MHC expression is not causal (49). It can be speculated that the activation of expression of these structural genes by myogenin in Xenopus takes place in adult tissues in the maintenance of a particular phenotype and during the muscle-regenerating process.
A better knowledge of the MRF target genes is an important issue for understanding the orchestration of gene expression in the myogenic cascade. Future work will address the mode of myogenin gene regulation and its functional significance in the process of myogenesis. Functional analysis of the two myogenin genes in Xenopus is likely to provide important insights into the regulation and physiological roles of this MRF in muscle.