Developmental regulation of creatine kinase gene expression by myogenic factors in embryonic mouse and chick skeletal muscle

Creatine kinases (CK) catalyze the regeneration of ATP by transphosphorylation of energy-rich phosphate from phosphoryl creatine to ADP produced in energyconsuming processes like muscle contraction, pumping of ions, etc. A phosphoryl creatine metabolic circuit was proposed to function in cells such as sperm, muscle and nerve, and in cells of the visual system where sudden changes of energy demand may occur (Walliman and Eppenberger, 1990). There are several genes encoding the different CK subunit types, which have different tasks in cellular physiology and are regulated in developmental- and tissue-specific manner. BCK is expressed in brain, smooth muscle, heart and several other tissues (Wirz et al. 1990), while MCK is specific for striated muscle (Perriard et al. 1989). In addition, there are at least two different genes for mitochondrial CK (MiCK), producing the MiCKs localized in the mitochondria (Hossle et al. 1986).

In developing muscle and in myogenic cell cultures, there is a transition from BCK to MCK expression (Perriard et al. 1978, 1989; Olson et al. 1983). In myogenic cell cultures, an early increase of BCK synthesis is observed (Caravatti et al. 1979). MCK synthesis is induced during differentiation and myotube formation while BCK expression is leveling off (Caravatti et al. 1979; Perriard, 1979; Chamberlain et al. 1985). BCK is expressed in a variety of embryonic tissues: at high levels in neural tissues, smooth muscle and chicken heart, and at lower levels in a wide variety of cell types. The results presented here will focus on the accumulation of its mRNA in embryonic skeletal muscle, and, to a lesser extent, in embryonic cardiac muscle.

Analysis of MCK gene expression in mouse muscle cell Unes (Jaynes et al. 1988; Buskin and Hauschka, 1989) has revealed a muscle-specific enhancer in the 5’ promoter which is involved with the upregulation of transcription in differentiating myocytes. Recently it was shown that the myogenic differentiation factor, MyoDl, binds specifically to two regions of the MCK enhancer (Lassar et al. 1989). This binding is important for in vivo expression of MCK. Constitutive expression of the MyoDl cDNA in C3H10T1/2 cells leads to the transactivation of a cotransfected CAT reporter gene linked to the MCK 5’ promoter containing the enhancer element (Davis et al. 1990; Yutzey et al. 1990). Using a CAT construct with an MCK promoter lacking the enhancer, no transactivation is observed. Another myogenic differentiation factor, myogenin (Wright et al. 1989), also has the ability to bind to the MCK enhancer (Brennan and Olson, 1990) and to transactivate the MCK gene (Yutzey et al. 1990).

In light of these findings, we investigated the pattern of expression of BCK and MCK genes in mouse embryos using in situ hybridization and compared these results with the expression of MyoDl and myogenin. We show that, in myotomes, the first skeletal muscle to form in mouse embryos, BCK, is coexpressed with myogenin and, two days later, with MyoDl, but, surprisingly, MCK is not detected until two days after MyoDl is first expressed. In order to determine if this sequence of expression of muscle genes is the same in other vertebrates, we also investigated BCK and MCK gene expression in chick embryos, and correlated our findings with the pattern of expression of CMD1 (Lin et al. 1989), the chick homologue of MyoDl, and of chick myogenin. Although the chick develops more rapidly than the mouse, first traces of MCK transcripts are detected approximately 24 h after the onset of CMD1 and myogenin expression. We show that the developmental sequence of expression of two of the myogenic factors in chick embryos differs from that in mouse embryos. CMD1, the chick homologue of MyoDl, is detected prior to chick myogenin, whereas in mouse embryos, myogenin precedes MyoDl (Sassoon et al. 1989).

The protocol that was used to fix and embed chick and mouse embryos and to prehybridize the 5-7 micron paraffin sections is described in detail in Lyons et al. (1990a). Our standard procedure for comparing sections is as follows. Two immediately adjacent sections are hybridized per slide and the next slide in the series contains the two adjacent sections (serial sections).

The cRNA transcripts were synthesized according to manufacturer’s (Stratagene) conditions and labelled with ³⁵S-UTP (>1000Ci mmol^-1; Amersham). cDNA probes were subcloned into the indicated transcription vectors to generate both sense and antisense probes. Transcripts greater than 100 nucleotides were alkali hydrolyzed to generate probes with a mean length of 70 bases for efficient hybridization. The following probes were used:

1. Rat BCK (Benfield et al. 1988). For the 1100 nucleotide probe, the Bluescribe + plasmid was linearized with EcoRI and T3 RNA polymerase was used.

2. Mouse MCK (Buskin et al. 1985). For the 330 nucleotide probe, the SP65 plasmid was linearized with A/mdlH and SP6 RNA polymerase was used.

3. Mouse myf-5 (Ott et al. 1991). For the 310 nucleotide probe, the Bluescribeplasmid was linearized with Wmdlll and T7 polymerase was used.

4. Rat myogenin (Wright et al. 1989). For the 700 nucleotide probe, the Bluescribe M13— plasmid was linearized with PstI and T7 polymerase was used.

5. Mouse MyoDl (Sassoon et al. 1989). For the 1000 nucleotide probe, the Henikoff modified Bluescribe plasmid was linearized with Mlul and T3 RNA polymerase was used.

6. Chick BCK (Hossle et al. 1986). For the 225 nucleotide probe, the pGEM-3zf(—) plasmid was linearized with BamHI and SP6 RNA polymerase was used.

7. Chick MCK (Ordahl et al. 1984). For the 220 nucleotide probe, the pGEM-3zf(—) plasmid was linearized with Kpnl and T7 RNA polymerase was used.

8. CMD1 (Lin et al. 11)89). For the 619 nucleotide probe, the Bluescript KS plasmid was linearized with Pvull and T7 RNA polymerase was used.

9. Chick myogenin (Masood and Paterson, unpublished data. The sequence is identical to that published by Fujisawa-Sehara et al. 1990). For the 305 nucleotide probe, the pGEM-3zf(—) plasmid was linearized with SacII and T7 polymerase was used.

The hybridization and posthybridization procedures for the mouse probes are as described by Lyons et al. (1990a). The following wash stringencies were used for the chick probes:

Chick BCK: 2×SSC/50% formamide 60min/65°C

CMD1: 2×SSC/50% formamide 60min/65°C

Chick myogenin: 2×SSC/50% formamide 90min/65°C

Chick MCK: l×SSC/50% formamide 120min/65°C

The slides were dipped in undiluted Kodak NTB-2 nuclear track emulsion and exposed for one week (unless stated otherwise) in light-tight boxes with desiccant at 4°C. Photographic development was carried out in Kodak D-19. Slides were analyzed using both light- and dark-field optics of a Zeiss Axiophot microscope or a Zeiss Tessovar optical system with dark-field equipment.

BCK gene transcripts are detected in developing somites by in situ hybridization when myotomes are formed. Fig. 1A–D shows sections of an 8.5 day post coitum (p.c.) mouse embryo (8–12 somites, Theiler, 1989). Serial sections were hybridized with the BCK probe (A,B), the myf-5 probe (C) and the myogenin probe (D). Due to the rotation of the embryo that occurs at this stage, the somites at the left (arrowhead Fig. 1A), which are rostral, are cut in.a frontal plane, whereas the somite at the right (arrow), which is more caudal, is cut in a transverse plane. Each of the three muscle mRNAs are expressed in the rostral somites, which have formed myotomes, but only myf-5 transcripts are detected in a caudal somite which is just beginning to form a myotome (arrow Fig. 1C). This pattern of muscle gene expression follows the rostro-caudal gradient of somite formation and development (Rugh, 1990; Theiler, 1989).

In the chick, CMD1 (the chick homologue to MyoDl, 80% amino acid sequence homology with the mouse MyoDl sequence, Lin et al. 1989) gene transcripts are detected in a series of myotomes in a Hamburger-Hamilton (1951) stage 13 embryo (Fig. 2C). Similarly, qmfl, the quail homologue of MyoDl, is first detected at stage 13 in quail embryos (qmfl has 96% amino acid sequence homology with MyoDl in the basic and helix-loop-helix regions of the molecule, but only 57 % homology with myf-5 (Charles de la Brousse and Emerson, 1990)). BCK mRNAs (Fig. 2B) are coexpressed with CMD1 gene transcripts. Myogenin transcripts are not detectable at this stage (data not shown). BCK transcripts are also expressed at high levels in the neural tube of both mouse (Fig. 1B) and chick (Fig. 2B) embryos.

MCK transcripts are not detected either in mouse or in chick myotomes at these early stages. In Fig. 3A–C, serial transverse sections of an 11.5 day p.c. mouse embryo were hybridized with cRNA probes in the following order: (A) MyoDl, (B) MCK and (C) BCK. These results show that one day after MyoDl transcripts are first detectable (Sassoon et al. 1989), MyoDl and BCK mRNAs are coexpressed in developing inyotomes, but MCK mRNAs are not detectable.

In contrast to the pattern of expression in thé mouse (Sassoon et al. 1989), myogenin transcripts are first detected in chick myotomes after CMD1 mRNAs are expressed at a high level. In Fig. 4A–D, serial frontal sections of a stage 15 chick embryo were hybridized with cRNA probes to: (B) BCK, (C) CMD1 and (D) chick myogenin. Beginning at this stage in the chick, CMD1 and the chick homologue of myogenin are coexpressed with BCK, but MCK transcripts are not detectable (data not shown). Following a rostrocaudal pattern of somite formation (Rugh, 1990; Theiler, 1989), myotome formation (Kaehn et al. 1988) and muscle gene expression (Sassoon et al. 1988, 1989; Charles de la Brousse and Emerson, 1990; Lyons et al. 1990 a,c), myogenin transcripts first appear at a low level in myotomes (Fig. 4D arrowheads) which are more rostral than the majority of those in which CMD1 (Fig. 4C) and BCK (Fig. 4B) are already expressed.

MCK transcripts are first detected in mouse embryos at 13 days p.c. (Fig. 5A–D). MCK mRNAs are expressed at low levels in both cardiac (Fig. 5B) and skeletal (Fig. 5D) muscle at approximately the same time. In skeletal muscle, MCK mRNAs are initially expressed most abundantly in muscles in the head and neck region (Fig. 5D) but rapidly accumulate in other skeletal muscles (Fig. 7D). In cardiac muscle, these transcripts first appear in the myocytes of the ventricular wall (Fig. 5B) and at later stages in all cardiac myocytes (Fig. 7D).

MCK mRNAs begin to be detected at a low level in rostral myotomes in developing chick embryos by stage 19. Fig. 6 shows parallel transverse sections of a stage 19 chick embryo hybridized with probes to: (B) BCK, (C) CMD1, (D) myogenin and (E–F) MCK. Unlike the mouse, MCK transcripts are not detected in chick cardiac muscle in ovo (Fig. 6E; Perriard et al. 1978; Schafer and Perriard, 1988) or in adult chick heart at significant levels.

Serial parasagittal sections of a 15 day p.c. mouse embryo (Fig. 7A–D) show that, after 2 days, the levels of MCK message have risen sharply in cardiac and skeletal muscle (Fig. 7D) whereas the amount of BCK message has decreased (Fig. 7C). BCK transcript levels have increased in smooth muscle at this stage (Fig. 7C arrowheads). MyoDl transcripts (Fig. 7B) continue to be expressed in skeletal muscle but are not detected in cardiac myocytes.

CK switching in chick embryos occurs within a different time frame from that in mouse embryos. At stage 22, the latest stage shown, BCK (Fig. 8C) message levels are still higher than those of MCK (Fig. 8F) although both CMD1 (Fig. 8D) and myogenin (Fig. 8E) transcripts are abundant. The time period shown for the chick from the first appearance of MCK transcripts (stage 19) to stage 22 spans only 12 h while the period in the mouse covers 2.5 days (Fig. 7). In chick embryos, similar regulatory events have been observed at later stages by SI nuclease protection analysis (Perriard et al. 1989; Perriard et al. unpublished data).

Table 1 summarizes the results for in situ hybridization with each of the probes used on mouse and chick embryonic sections.

Our results show that BCK gene transcripts, in addition to being expressed at high levels in the neural tube and elsewhere in the embryo, are early markers for myogenesis in mouse embryos. BCK mRNAs are coexpressed in newly formed myotomes, the first skeletal muscle to form in developing embryos, with mRNAs for a-actin (Sassoon er al. 1988), myogenin (Sassoon et al. 1989; results presented here) and myf-5 (Ott et al. 1991). We have shown that BCK mRNAs are also early myogenic markers in chick embryos and are coexpressed in myotomal cells with CMD1. Consistent with these results, Charles de la Brousse and Emerson (1990) have shown that, in quail embryos, transcripts of the qmfl gene, the quail homologue of MyoDl, are early myogenic markers with cardiac a-actin transcripts. We will discuss our observations in terms of: functional differences between CKs, cell-specific expression of myogenic factors, and protein-protein or protein-DNA interactions between transcription factors.

Myotomal myocytes are elongated, mononucleated cells, which express contractile proteins prior to fusion (Holtzer et al. 1957; Jockusch et al. 1984; Vivarelli et al. 1988; Fürst et al. 1989; Sweeney et al. 1989; Lyons et al. 1991). The myosins and actins are assembled into functional myofibrils because myotomal myocytes have been observed to contain cross-striations and to contract (Holtzer et al. 1957). Thus, by most criteria, myotomes consist of differentiated myocytes. Although myogenic factors are expressed, MCK mRNA was not detected, but high levels of BCK transcripts were observed in embryonic myotomes. It is possible that the task of CK in energy metabolism of myotomal cells can be fulfilled by BCK. There are differences in kinetic parameters between various CK isozymes (Quest et al. 1990; Walliman and Eppenberger, 1990). Other properties of the isozymes are important for the interaction with the M-band (Schafer and Perriard, 1988) or other sites that appear to become more important in fully differentiated muscle cells, where an additional CK isoprotein, the mitochondrial MiCK is expressed (Perriard et al. 1989) and a metabolic circuit is thought to function (Walliman and Eppenberger, 1990).

At present, there has been no demonstration, although it has not been excluded either, that the BCK gene promoter contains binding sites for myogenic factors (Wirz et al. 1990; Horlick et al. 1990; Hobson et al. 1990). This observation, and high levels of expression in other cell types, suggest that the BCK gene is not regulated by myogenic factors in skeletal muscle.

One possible explanation for the lack of MCK gene expression in newly formed mouse myotomes is that MyoDl, which is known to bind to the MCK enhancer (Lassar et al. 1989) and to upregulate transcription of MCK in differentiating myocytes (Davis et al. 1990; Yutzey et al. 1990), is not expressed at detectable levels for the first two days during which myotomes are forming (Sassoon et al. 1989). However, our results show that MyoDl mRNAs are expressed for two full days prior to the first detectable MCK transcripts in mouse embryos. Similarly, CMD1 mRNAs are expressed for a full day before MCK transcripts are first detected in chick embryos.

MyoDl is known to be expressed at varying levels in proliferating myoblasts grown in vitro without activation of the MCK gene (Tapscott et al. 1988; Mueller and Wold, 1989). Therefore a second possible explanation for the delayed appearance of MCK transcripts is that the high level of MyoDl expression detected in mouse embryos on days 11 and 12 p.c. is due to expression only in proliferating myoblasts. This possibility cannot be ruled out without the use of cold probes or antibodies for colocalization of MyoDl in single embryonic cells with other markers for myocyte differentiation. However, it seems unlikely given the similarity of its distribution with myogenin, which is known to be expressed only in differentiating myocytes (Wright et al. 1989), and which also binds to the MCK enhancer and upregulates MCK gene expression (Yutzey et al. 1990). MCK is expressed in differentiated BC3H1 cells which do not express detectable levels of MyoDl (Davis et al. 1987). In addition to myogenin, these cells also express myf-5 (Braun et al. 1989). myf-5 is expressed from very early stages of myogenesis in the mouse embryo (Ott et al. 1991), and is no longer an abundant transcript at the time when MCK begins to be expressed. Myogenin transcripts are detected in myotomes for four days before MCK mRNAs are first seen. In the chick embryo, myogenin is first detected at stage 15 in myotomes and precedes the first detectable expression of MCK (stage 19) by approximately 18 h.

A third possible explanation is that MyoDl and myogenin may not be free to interact with the enhancer binding site due to heterodimerization with other proteins (see Weintraub et al. 1991) or due to the presence of inhibitory factors. In the case of the MCK enhancer, both proteins have been shown to interact with it as heterodimers with the widely expressed E12 DNA-binding protein (Murre et al. 1989a,b;Davis et al. 1990). E12 availability is modulated by the protein Id (Benezra et al. 1990), another member of the helixloop-helix superfamily, which is known to be down regulated during differentiation of myoblasts. In situ hybridization experiments with an Id probe or a probe from the E12 class of sequences on embryo sections would yield further information about the possible involvement of this protein in the observed pattern of MCK gene expression. Myogenic factor proteins are known to be phosphorylated, which suggests that post-translational modifications of MyoDl and myogenin may prevent their binding to and activation of the MCK enhancer (see Olson, 1990).

Since other muscle structural genes, which are probably also activated by a similar mechanism, e.g. MLC1F (Donoghue et al. 1988), are already expressed in myotomes at an earlier stage (Lyons et al. 1990a), it seems unlikely that MyoDl or myogenin are not acting as transcription factors prior to MCK gene expression. Different genes may require different quantitative levels of these muscle-specific factors for activation, and indeed threshold levels of other proteins implicated in MCK transcription may not be reached until a relatively late stage for this gene. Alternatively, the MCK enhancer may not be available for binding of myogenic differentiation factors, possibly due to DNA configuration, since in vivo protein-DNA interactions detected at the MCK enhancer were confined to differentiated myocytes in which MCK is actively transcribed (Mueller and Wold, 1989).

At a later stage in myotome development, day 4 in the chick (Holtzer et al. 1957) and day 12–13 in the mouse (Vivarelli and Cossu, 1986), the mononucleated myocytes fuse to form multinucleated myotubes. Interestingly, the earliest stages at which MCK transcripts become detectable in mouse and chick embryos correlate roughly with this onset of fusion, but it is not clear if the two events are related. However, neither MyoDl nor fusion to form multinucleated myotubes is necessary for MCK expression, since the BC3H1 cell line, which does not express MyoDl (Davis et al. 1987) and does not fuse, does express MCK under differentiation conditions (Mueller and Wold, 1989).

MCK mRNAs first appear in mouse cardiac myocytes at approximately the same time as they are first detected in skeletal muscle. This is in contrast to other muscle genes such as cardiac actin and the cardiac myosins which are expressed much earlier in the heart (Lyons et al. 1990b). Transcription of the MCK gene in the heart is not regulated by any of the known myogenic differentiation factors, since none of these are expressed in the cardiac myocytes (Wright et al. 1989; Lyons et al. 1990b,c). BCK is an early marker for cardiac myogenesis in mouse (data not shown) and is the only cytosolic isoform of CK expressed in chick cardiac muscle (Turner and Eppenberger, 1973; Schafer and Bernard, 1988).

The developmental program of creatine kinase gene expression in embryonic skeletal muscle is complex and involves more than the myogenic factors that have been characterized to date (Perriard et al. 1989; Horlick et al. 1990). In the initial phase of skeletal muscle formation, i.e. in myotomes, BCK is the only cytosolic isoform of CK that is expressed and is sufficient to meet the energy requirements of differentiating myocytes, which make myofibrils and contract. The expression of this gene is not known to be regulated by myogenic factors identified to date. MCK is expressed later, well after myogenic differentiation factor transcripts appear, and approximately at the time that myocytes begin to fuse to form multinucleated myotubes. MCK expression is not likely to be regulated by myf-6, since these transcripts are not detected in fetal muscle until 15 days p.c. (Bober et al. 1991), two days after the initial detection of MCK transcripts. Subsequently, in the mouse, MCK mRNA levels increase rapidly and BCK transcript levels fall off in striated muscle. In the chick, a similar behavior is observed, although within a different time frame. The decrease in BCK mRNA levels is delayed approximately 10 days in skeletal muscle until 13 days in ovo when MCK mRNA accumulation is induced (Perriard et al. 1989; Perriard et al. unpublished data). In the mouse, the rapid increase in MCK levels is likely to be associated with the formation of secondary myotubes which begins around 15 days p.c. (Ontell and Kozeka, 1984). Further elucidation of the developmental regulation of BCK and MCK in muscle cells will require quantitative and qualitative characterization of the transcription factors that regulate the activation of these genes, and analysis of the question of the accessibility of these genes to transcription due to chromosomal structure.

It is striking that the onset of expression of the myogenic factor, MyoDl, differs between chick and mouse embryos. A myf-5 homologue in the chick has not yet been described, and its possible role in the early activation of chick MyoDl expression cannot be discounted. MyoDl is expressed relatively early in the chick, and is a major transcript before myogenin in the early myotome, whereas in the mouse, MyoDl is only detectable two days after the onset of myogenin expression (Sassoon et al. 1989). This result based on in situ hybridization data in the mouse has recently been confirmed by PCR analysis (Montarras et al. 1991). Either these two genes have evolved independently in the two species from a common ancestor, or, as the sequences tend to suggest, we are dealing with equivalent coding sequences which are regulated differently. The implication would be that this is also the case for the muscle structural genes, activated during myotome formation in the chick, which are probably regulated by these myogenic factors.

The authors would like to thank Dr P. Benfield for the rat BCK probe, Drs J. Buskin and S. Hauschka for the mouse MCK probe, Drs E. Bober and H. Arnold for the mouse myf-5 probe, Dr W. Wright for the myogenin probe and Drs A. Lassar and H. Weintraub for the MyoDl probe. G.L. would like to thank Mme Odette Jaffrezou for her invaluable technical assistance. J-C.P. would like to acknowledge the expert technical assistance of S. Keller. p.L. held a NIH/CNRS fellowship from the Fogarty International Center. This work was supported by grants from the Pasteur Institute, C.N.R.S., I.N.S.E.R.M., A.F.M., and A.R.C. to M.B. The work at the Swiss Federal Institute of Technology was supported by Swiss National Science Foundation grant 31-2775689, and training grants to S.M. and A.M. from E.T.H.