Analysis of myostatin gene structure, expression and function in zebrafish

Myostatin (or GDF-8), a member of the Transforming Growth Factor-β(TGF-β) superfamily, was first identified in mice by McPherron et al.(1997) and has been demonstrated to negatively regulate skeletal muscle growth in several mammalian species. Myostatin knockout mice show a dramatic increase of skeletal muscle mass, and the increase results from a combination of hyperplasia and hypertrophy (McPherron et al., 1997). The `Double muscle' breeds of cattle that have significantly more muscle mass than standard breeds were found to carry natural mutations in the myostatin gene(McPherron and Lee, 1997; Kambadur et al., 1997; Grobet et al., 1997, 1998). In vitrostudies have demonstrated that Myostatin functions by inhibiting myoblast proliferation and differentiation (Thomas et al., 2000; Rios et al.,2001; Taylor et al.,2001; Langley et al.,2002). This is, in part, accomplished by downregulating myogenic gene expression (Langley et al.,2002; Amthor et al.,2002). The myostatin gene has been cloned from over 20 different vertebrate species including several fish species(McPherron and Lee, 1997; Rodgers and Weber, 2001; Rodgers et al., 2001;Maccatrozzo et al., 2001a,b, 2002; Kocabas et al., 2002; Roberts and Goetz, 2001; Ostbye et al., 2001; Rescan et al., 2001; Radaelli et al., 2003). In some fish species, two distinct myostatin genes were found(Roberts and Goetz, 2001; Ostbye et al., 2001; Rescan et al., 2001). Comparison of Myostatin sequences revealed that Myostatin was extremely well conserved throughout evolution. Remarkably, the murine, rat, human, porcine,chicken and turkey Myostatin sequences are all identical in the active C-terminal region of the protein, suggesting that the function of this gene might be conserved in all vertebrates. However, this has not been tested.

Like other members of the TGF-β family, Myostatin is synthesized as a prepro-peptide that undergoes two steps of proteolytic cleavage to generate the biologically active C-terminal domain(Thomas et al., 2000; Rios et al., 2001). The bioactive C-terminal domain dimerizes and binds to membrane receptors on target cells (Thomas et al.,2000). The mature TGF-β C-terminal dimer often forms an inactive complex with the N-terminal prodomain of TGF-β(McPherron and Lee, 1996). This observation suggested that the Myostatin C-terminal active domain might also exist as a secreted latent complex with the Myostatin prodomain(Lee and McPherron, 1999; Thies et al., 2001). Recently,Thies and colleagues (2001)showed that Myostatin prodomain was able to bind to the bioactive Myostatin C-terminal active domain and inhibit its biological activity, presumably by preventing Myostatin active domain from binding to its receptor on the cell surface. Transgenic mice expressing the Myostatin prodomain in muscle cells showed enhanced muscle growth similar to myostatin knockout mice(Lee and McPherron, 2001; Yang et al., 2001).

Studies in mice and cattle have demonstrated that myostatin mRNA was specifically expressed in developing somite and skeletal muscles(McPherron et al., 1997; Kambadur et al., 1997). Recent studies, however, revealed that in addition to muscle cells, myostatin mRNA was also expressed in several other tissues, such as cardiomyocytes, mammary glands and, at a lower level, in adipose tissue(Ji et al., 1998; Sharma et al., 1999). In fish, myostatin mRNA was found to be expressed in muscles, eyes, gill filaments, spleen, ovaries, gut, brain and, to a lesser extent, in testes(Rodgers et al., 2001; Maccatrozzo et al., 2001a). Myostatin expression in skeletal muscles appeared to be restricted to certain types of muscle fibers. Carlson et al.(1999) demonstrated that myostatin mRNA was mainly expressed in fast muscle fibers in mice. Roberts and Goetz (2001)reported that myostatin mRNA was primarily expressed in red muscles in brook trout, king mackerel, and yellow perch, but expression in the little tunny is in the white muscles. Consistent with the study by Roberts and Goetz(2001), Rescan et al.(2001) showed that trout myostatin-2 mRNA was predominantly expressed in red muscles. It remains to be determined if Myostatin is involved in different muscle growth of fast and slow muscles in fish.

Although myostatin cDNA has been cloned from zebrafish(McPherron and Lee, 1997), the temporal and spatial pattern of its expression is unknown. It is not clear when and where it is expressed in zebrafish embryos. Is it expressed in developing somites and skeletal muscles as in mammals? Is the muscle expression restricted to specific types of muscle fibers? Moreover, does Myostatin inhibit muscle growth in fish? We have reported here the isolation and characterization of myostatin genomic gene from zebrafish and analysis of its expression in zebrafish embryos, larvae and adult skeletal muscles. Our data showed a weak myostatin expression in early stage embryos and a strong expression in swimming larvae and juveniles, and the skeletal muscles of adult zebrafish. Overexpression of the Myostatin prodomain in skeletal muscles of transgenic zebrafish had no effect on the expression of myogenic regulatory genes. However, the adult transgenic fish showed hyperplasia in their skeletal muscles compared with non-transgenic controls.

The genomic gene of zebrafish myostatin was isolated as several DNA fragments by polymerase chain reaction (PCR) from the zebrafish GenomeWalker libraries (Du et al.,2003). A 1.2 kb DNA fragment containing the 5′ flanking sequence and part of the first exon were generated by PCR using the myostatin specific primers (myostatin E1.1/E1.2) close to the ATG start codon together with the adapter primers AP1 and AP2 at the 5′ end of the flanking region. A 2.5 kb fragment including the 3′flanking region and part of the third exon were amplified from the libraries using a set of myostatin 3′ specific primers(myostatin E3.1/E3.2) together with the AP1 and AP2 primers. The middle part of the gene, including part of the first exon, intron 1, exon 2,intron 2 and part of exon 3, was generated by regular PCR from zebrafish genomic DNA using two sets of myostatin-specific primers, myostatin E1.3/E2.1R and myostatin E2.1F/E3.3. PCR for GenomeWalker was carried out in 50 μl reaction containing 1 μmol l^-1 of each primer, four dNTPs at 0.2 mmol l^-1 for each,and 5 units of Advantage Tth polymerase (Clontech, Palo Alto, CA, USA). PCR was carried out in two-step cycle program with 7 cycles at 94°C for 25 s and 72°C for 3 min, and followed by 32 cycles at 94°C for 25 s and 67°C for 3 min. The regular PCR was carried out in a 50 μl reaction solution containing 1 μmol l^-1 of each primer, four deoxyribonucleotide triphosphates at 0.2 mmol l^-1 for each, and 2.5 units of Taq DNA polymerase (Promega, Madison, WI, USA). PCR was carried out for 30 cycles. Each cycle included 30 s at 94°C, 30 s at 60°C and 3 min at 72°C. All DNA fragments were cloned into pGEM-T Easy Vector(Promega) and sequenced with an ABI automated DNA sequencer. Sequence alignment and searches were performed using the BLAST, TRANSFAC and TFSEARCH databases: Ap1, 5′-GTAATACGACTCACTATAGGGC-3′; Ap2, 5′-ACTATAGGGCACGCGTGGT-3′; myostatin E1.1,5′-GCTGATGTTTGGAGCCTGCTTGAGTCG-3′; myostatin E1.2,5′-CTTGAGTCGGAGTTTGCTAAGAATTTG-3′; myostatin E1.3,5′-CAAATTCTTAGCAAACTCCGACTCAAG-3′; myostatin E2.1R,5′-CTGCCAAGACGTGACTCCTGCGTTCA-3′; myostatin E2.1F,5′-TGAACGCAGGAGTCACGTCTTGGCAG-3′; myostatin E3.1,5′-ACTCCCACCAAGATGTCTCCCATCAAC-3′; myostatin E3.2,5′-GGCAAAGAGCAGCTCATCTACGGCAAG-3′; myostatin E3.3,5′-CTTGCCGTAGATGATCTGCTCTTTGCC-3′;

The chromosomal position of zebrafish myostatin gene was mapped using the LN 54 radiation hybrid panel generated by Hukriede et al.(1999). The panel was produced by irradiation of zebrafish fin fibroblasts prior to fusion with mouse B78 melanoma cells. DNA from 93 hybrid cell lines was used for PCR analysis to detect the presence of the zebrafish myostatin gene using E1.3/E2.1R primers. The PCR was done as described above. The results of the PCR from the radiation hybrid panel were scored according to Hudson et al.(1995) using a web-based interface RHVECTOR at http://mgchd1.nichd.nih.gov:8000/zfrh/beta.cgi.

The myostatin-GFP gene construct was obtained by amplifying the 5′ flanking sequence of zebrafish myostatin by PCR using the Ap2 primer at the 5′ and a gene-specific primer myostatin5R1(5′-GCGTCGACGTTCCAAGGCGTGCTAAAGGATG-3′) based on the 5′ UTR sequence of the zebrafish myostatin gene. To minimize mutation introduced by PCR, pfu DNA polymerase was used in the PCR reaction. PCR was carried out in a 50 μl reaction solution containing 1 μmol l^-1 of each primer, four deoxyribonucleotide triphosphates at 0.2 mmol l^-1 for each, and 2.5 units of pfu DNA polymerase (Stratagene,La Jolla, CA, USA). PCR was carried out for 35 cycles. Each cycle included 30 s at 94°C, 30 s at 60°C and 5 min at 72°C. A SalI site was introduced at the end of the 5′ UTR primer. The 1.2 kb 5′flanking sequence was first cloned into the SmaI site of Bluescript SK. The 5′ flanking sequence was then released from the Bluescript SK vector by SalI digestion and cloned into the SalI site of the GFP construct (Du and Dienhart,2001). The resulting plasmid was confirmed by DNA sequencing and designated as myostatin-GFP.

To construct the transgene encoding the Myostatin prodomain, the complete coding region of zebrafish myostatin was first amplified by PCR using pfu DNA polymerase. A BamHI and a XhoI site were introduced by the PCR primers at the 5′ and 3′, respectively. 5′-primer: 5′-GGATCCAACATGCATTTTACACAGGTTTT-3′,3′-primer: 5′-CTCGAGGGTTCATGAGCAGCCACAGCGG-3′. PCR was carried out in a 50 μl reaction solution containing 1 μmol l^-1 of each primer, four deoxyribonucleotide triphosphates at 0.2 mmol l^-1 each, and 2.5 units of pfu DNA polymerase (Stratagene). PCR was carried out for 30 cycles. Each cycle included 30 s at 94°C, 30 s at 60°C, and 3 min at 72°C. The PCR fragment was purified on a gel and then cloned into a plasmid Bluescript SK SmaI site. To generate the Myostatin prodomain construct, a stop codon (TAA) was introduced immediately after the RIRR proteolytic cleavage site at position 783 in the myostatin coding region using the QuickChange Site Directed Mutagenesis Kit (Stratagene). The gene construct encoding the Myostatin prodomain and its signal peptide were released from the plasmid by EcoRI digestion and cloned downstream of the rat mylcpromoter/enhancer at the EcoRI site(Donoghue et al., 1991). In addition, SV 40 polyadenylation and transcription termination signals were included at the 3′ end of the Myostatin preprodomain to ensure proper transcription termination and polyadenylation of the mRNA encoding the Myostatin prodomain.

The plasmids were diluted in distilled water to a final concentration of 50μg ml^-1. Phenol Red was added to the injection solution at a final concentration of 0.1% to facilitate visualization during microinjection. Approximately 2 nl of DNA solution were microinjected into the cytoplasm of zebrafish embryos at the one- or two-cell stage. Microinjection was carried out under a dissection microscope (MZ8, Leica, Deerfield, IL, USA) using a PLI-100 pico-injector (Medical System Corp., New York, NY, USA). 500 embryos were injected for each construct.

The whole-mount in situ hybridization and antibody staining was carried out as previously described (Du and Dienhart, 2001). The Dig-labeled probe was synthesized by T7 RNA polymerase in vitro using BamHI linearized plasmid pBSMSTN that contains the full-length zebrafish myostatin cDNA(McPherron and Lee, 1997). For embryos of 24 h and older, proteinase K treatment was performed to enhance the permeability of the embryos, as described by Du and Dienhart(2001).

Total RNA was extracted from zebrafish Danio rerio embryos,larvae, juvenile and adult muscle with TRIzol reagent (Invitrogen Corp.,Carlsbad, CA, USA) according to the manufacturer's instructions. cDNA was synthesized using the first strand cDNA synthesis kit (Life Sciences Inc., St Petersburg, FL, USA). The expression of myostatin mRNA was analyzed by reverse transcriptase (RT)-PCR using two pairs of myostatinprimers. One pair is a set of transgene-specific primers(myostatin5pc/E2.1r). The myostatin5pc primer(5′-CTGCAGCCCGGATCCAACATGCAT-3′) was based on part of the 5′UTR sequence of the rat mylc promoter and part of the myostatin coding region near the ATG site. The E2.1R primer(5′-CTGCCAAGACGTGACTCCTGCGTTCA-3′) was based on part of the sequence in myostatin exon 2. PCR using this pair of primers generated a 627 bp fragment from the mylc-myostatin^protransgene. Another pair of primers was myostatin E1.3/E2.1R. E1.3(5′-CAAATTCTTAGCAAACTCCGACTCAAG-3′) was based on myostatin exon 1, while primer E2.1R was from the exon 2 sequence shown above. PCR using this pair of primers generated a 432 bp fragment from the endogenous myostatin gene and the prodomain transgene. Elongation factor 1-α (Ef-1α) was used as RT-PCR control. The primers for the Ef-1α 5′ and 3′ primers were EF-1α-5′(GCATACATCAAGAAGATCGGC) and Ef-1α-3′ (GCAGCCTTCTGTGCAGACTTTG),respectively. PCR was carried out for 35 cycles (30 s at 94°C, 30 s at 60°C and 1 min at 72°C) in a 25 μl reaction solution containing 1μmol l^-1 of each primer, four deoxyribonucleotide triphosphates at 0.2 mmol l^-1 for each, and 0.5 units of Taq DNA polymerase(Promega). 20 μl of the amplified product was analyzed by electrophoresis on a 1% agarose gel.

To screen germline transgenic fish, DNA was extracted from a batch of 100 F₁ embryos from each cross. Briefly, 500 μl of lysis buffer (50 mmol l^-1 KCl, 10 mmol l^-1 Tris, pH 8.8, 1.5 mmol l^-1 MgCl₂, 0.1% Triton X-100) was added to a group of 100 embryos, 48 h.p.f. (hours post fertilization). The embryos were gently homogenized in the lysis buffer. The homogenate was boiled for 5 min, and was then digested with proteinase K (100 μg ml^-1) for 1 h at 55°C. Proteinase K was inactivated by boiling for 10 min after the digestion. The samples were centrifuged at 12 000 g for 5 min and 1.5 μl of the supernatant was used for the PCR reaction. PCR was carried out in a 25 μl reaction solution containing 1 μmol l^-1 of each primer, four deoxyribonucleotide triphosphates at 0.2 mmol l^-1 for each, and 0.5 units of Taq DNA polymerase (Promega). PCR was carried out for 35 cycles. Each cycle included 30 s at 94°C, 30 s at 60°C and 1 min at 72°C. 10 μl of the amplified product was analyzed by electrophoresis on a 1% agarose gel.

To screen adult transgenic offspring, DNA was extracted from a small portion of the caudal fin as described (Du and Dienhart, 2001) and used for PCR analysis. PCR was carried out using transgene-specific primers (mylc-p2/E2.1R) based on the mylc5′ flanking sequence (mylc-p2,5′-CACCACTGCTCTTCCAAGTGTCA-3′) and part of exon 2 (E2.1R)of zebrafish myostatin. This PCR yields a 704 bp product. Control PCR primers HH-F (5′-GGACGGTGACACTTGGTGATG-3′) and HH-R(5′-CGAGTGGATGGAAAGAGTCTC-3′) were derived from the sense and antisense strand of the tiggy winkle hedgehog exon 3 sequence,respectively. PCR using this set of control primers produced a 615 bp DNA fragment. PCR reaction was carried out as described above for the embryos.

Expression of MyoD, Myf5 and myogenin mRNAs was analyzed by real-time PCR in transgenic and non-transgenic fish at 2, 7, 45 and 240 days of age. For 2- and 7-day-old fish, total RNA was extacted from 100 embryos or larvae using TRIzol. For 45-day-old fish, individual fish were used for RNA extraction. For 240-day-old fish, ∼60 mg of muscle tissue was used for RNA extraction. The total RNA was digested with DNAses to remove endogenous DNA contamination. First-strand cDNA was synthesized by reverse transcriptase using oligo-dT primer (RETROscript Kit, Ambion, Austin, TX,USA). Real-time RT-PCR was carried out using the following primers. MyoD was amplified using MyoD 5′- and 3′-primers. The PCR product was a 288 bp fragment. Similarly, Myf5 and myogenin were amplified with their specific primers that generated a 482 bp and 319 bp fragment, respectively: MyoD 5′:5′-AGACGGAACAGCTATGACAGCTCT-3′; MyoD 3′,5′-ATTTTAAAGCACTTGATAAATG-3′; Myf-5 5′,5′-CACTCAGAAACCTTCAACACCAA-3′; Myf5 3′,5′-ATGCTCTCTGAGCAGCTGGAGGA-3′; Myogenin 5′,5′-TCTAGTGATCAGGGCTCTGGCAGCA-3′; Myogenin 3′,5′-TAAGCCCTCCAAGGCTTGTCTAACTTGC-3′.

Real-time PCR was carried on Sequence Detector (PRISM 7700; ABI, Foster City, CA, USA). Each reaction (50 μl) contained 3 μl cDNA and primers at a final concentration of 2 ng μl^-1. SYBR Green was included in the PCR reaction as described in the protocol of SYBR Green PCR (ABI). The samples were first heated to 50°C for 2 min followed by 95°C for 10 min. The PCR reactions were carried out for 45 cycles at 95°C for 15 s and 60°C for 1 min. SDS v1.7a software was used to define the cycle in which each sample attained the threshold value.

To determine the muscle structure in transgenic zebrafish at juvenile and adult stages, F₂ transgenic fish were generated by crossing transgenic males with wild-type females. Approximately 100 F₂ fish,including both transgenic and non-transgenic fish, were raised in the same tank. These fish were weighed at 2.5, 3 and 5.5 months of age, after anesthetization with 0.016% tricaine at two time points. The transgenic fish were identified by PCR using DNA extracted from the tail fins. Mean body mass was calculated for transgenic and non-transgenic males and females of different ages. For fiber size and number analysis, five individual fish from each group (female or male and transgene or wild type) were fixed with Bouin's solution for 24 h followed by routine paraffin sectioning and Haematoxylin/Eosin staining. A horizontal section at the base of the first pin of the anal fin was selected for quantification of the number of muscle fibers and measurement of the diameter of muscle fibers. A Student t-test was used to determine whether significant differences existed between wild-type and transgenic fish.

To better understand the myostatin gene structure, function and regulation of expression in fish, the complete genomic sequence of zebrafish myostatin was isolated by polymerase chain reaction (PCR) using the GenomeWalker method as described in Materials and methods. Sequence analysis revealed that the zebrafish myostatin gene spanned 6.4 kb, including a 1.2 kb 5′ flanking sequence and a 5.2 kb transcriptional unit that includes the 3′ flanking sequence (GenBank accession number AY323521). Comparison with the zebrafish myostatin cDNA sequence revealed that the zebrafish myostatin gene contained three exons and two introns(Fig. 1A). The intron and exon junctions are highly conserved with myostatin genes of other vertebrate species. Sequence analysis of the zebrafish myostatin5′ flanking sequence identified a putative `TATA' box at position 110 bp upstream from the ATG start site (Fig. 1B). In addition, several putative MyoD binding sites (CAxxTG)that confer muscle-specificity, known as E boxes, were identified in the 5′-flanking region (Fig. 1B). Two of the potential E boxes (E4, E5) are closely located in the 5′ flanking sequence of the zebrafish myostatin gene. Of particular interest, two closely linked E boxes (E5, E6) were present in a similar position in the bovine myostatin promoter(Fig. 1A), and critical for its muscle-specific expression (Spiller et al., 2002). Considering that MyoD often functions as a dimer, the presence of closely linked E boxes in the conserved region of myostatin gene promoters suggests that MyoD may be involved in regulation of myostatin gene expression.

Zebrafish is widely viewed as an excellent system for genetic study of gene function. To map the position of the myostatin gene in zebrafish for future identification of potential myostatin mutants, linkage group analysis was carried out using the LN 54 radiation hybrid panel generated by Hukriede et al. (1999). The results placed the zebrafish myostatin gene in chromosome 9, at approximately 0.5 centiRay (1 cR isapproximately 148 kb) from the genetic marker Z8363.

There is much evidence that myostatin mRNA is primarily expressed in developing somite and skeletal muscles in mammals. Recent studies in fish have also demonstrated that myostatin mRNA is predominantly expressed in skeletal muscles. However, its expression pattern in early stage fish embryos is not clear. A clear understanding of myostatin expression in fish embryos would provide important insights into the role(s) of Myostatin in regulating muscle development and growth in fish. To determine the temporal and spatial pattern of myostatin mRNA expression in zebrafish embryos, whole-mount in situ hybridization was carried out with zebrafish embryos at several developmental stages (14 h post fertilization to 4 days post fertilization). The results showed that myostatin mRNA could not be detected by in situ hybridization in zebrafish embryos(Fig. 2A,B), suggesting that myostatin mRNA was not expressed or only at a very low level in early stage zebrafish embryos. To confirm the in situ data and to further investigate its temporal pattern of expression in zebrafish, RT-PCR was used to analyze myostatin expression using total RNA from whole zebrafish embryos, larvae, juvenile and skeletal muscles of adult zebrafish. The results confirmed that myostatin mRNA was weakly expressed in the early stage zebrafish embryos (Fig. 2C). However, compared with day-4 embryo, myostatin mRNA expression appeared to be increased in 2-week-old swimming larvae, 1-month-old juveniles and skeletal muscle of 3-month-old adult zebrafish based on results from regular RT-PCR (Fig. 2C).

Analysis of the zebrafish myostatin 5′ flanking sequence revealed several putative E box sites that may be involved in its muscle-specific expression. To determine if the 5′ flanking sequence could drive gene expression in muscle cells, the 1.2 kb DNA sequence of zebrafish myostatin 5′ flanking region was ligated with the GFP reporter gene, and the DNA construct myostatin-GFP(Fig. 3) was microinjected into zebrafish embryos for transient expression analysis. As indicated by GFP expression and anti-GFP antibody staining in the injected embryos(Fig. 3A,B,D,E), the zebrafish myostatin 5′ flanking sequence strongly targeted GFP expression in muscle cells of zebrafish embryos. Approximately 69% (N=148) of the injected embryos showed muscle expression, although some of the embryos also showed weak non-muscle expression in floor plate and head regions(Fig. 3C,F). These results indicate that the zebrafish myostatin 5′ flanking sequence contained regulatory elements required for muscle expression and possibly other regulatory sequences for myostatin expression in other tissues.

To investigate Myostatin function in fish muscle development and growth, a transgenic approach was used to express Myostatin prodomain in zebrafish muscle cells that could suppress Myostatin function. To identify a strong muscle-specific promoter with which to target the expression of the Myostatin prodomain in zebrafish skeletal muscles, we analyzed whether or not the rat myosin light chain (mylc) gene promoter and its 1/3 enhancer were muscle-specific in zebrafish. The mylc promoter/enhancer has been shown to be muscle-specific in other vertebrates(Donoghue et al., 1991; Lee and McPherron, 2001). The rat mylc gene promoter/enhancer was linked with the GFP reporter gene(Fig. 4A). The mylc-GFP construct was microinjected into zebrafish embryos. GFP expression was analyzed by direct observation using a fluorescence microscope or immunostaining using anti-GFP antibody. Over 80% (N=157) of the injected embryos exhibited muscle-specific GFP expression(Fig. 4C,D). The remaining embryos that failed to express GFP in muscle cells did not exhibit GFP expression in other tissues. These data demonstrated that the rat mylc promoter/enhancer was muscle-specific in zebrafish and could be used to drive the expression of Myostatin prodomain in zebrafish muscle cells.

A gene encoding the zebrafish Myostatin prodomain was linked with the mylc promoter/enhancer (Fig. 4B). The resultant construct mylc-myostatin^pro was microinjected into zebrafish embryos for transgenic fish production. Two independent germline transgenic founders (1 and 33) were identified (out of 184) by PCR screening of their F₁ embryos. To determine the temporal and spatial expression of the transgene encoding the Myostatin prodomain, whole-mount in situ hybridization was used to analyze the expression of myostatin prodomain transgene in F₁ embryos at 14, 16 and 24 h.p.f. Because the expression of the endogenous myostatin gene was undetectable by whole-mount in situ hybridization at these stages(Fig. 2A,B), the in situ expression pattern, thus represented the expression of the transgene. The results showed that the myostatin transgene was strongly expressed in developing somite and embryonic muscles of the transgenic fish embryos (Fig. 5A,C,E). The two transgenic lines exhibited different patterns of transgene expression. In line 33, the transgene was exclusively expressed in skeletal muscles, and it appeared that the expression was fast-fiber specific(Fig. 5H,I). In contrast, line 1 showed expression in skeletal muscle and weak expression in the brain and spinal cord (data not shown). Because line 33 exhibited a strong muscle-specific expression, this transgenic line was primarily used in this study.

The inheritance of the transgene in line 33 was analyzed in the F₁, F₂ and F₃ generations. The transgene was inherited to the F₂ and F₃ generations in a Mendelian fashion, suggesting that it was integrated in a single site (data not shown). To determine if the transgene was expressed in skeletal muscles of adult transgenic fish, two sets of RT-PCR were employed to analyze the expression of the transgene and the endogenous myostatin gene(Fig. 6). The results showed that the transgene was strongly expressed in the skeletal muscles of adult transgenic fish. RT-PCR using the transgene-specific primers produced a predicted PCR product from the skeletal muscles of the transgenic fish(Fig. 6). In contrast, no PCR product of the transgene was found in non-transgenic fish. RT-PCR using another set of primers that amplify RNA transcripts from both the transgene and the endogenous myostatin gene generated a PCR product from both the transgenic and the non-transgenic fish(Fig. 6). Moreover, there were clearly more PCR products generated from the transgenic fish than the non-transgenic control (Fig. 6). This was probably because the transgenic fish contained RNA transcripts from both the endogenous myostatin gene and the transgene, whereas in the non-transgenic fish, the PCR product was solely derived from the endogenous gene transcripts. The PCR fragments were cloned,sequenced and confirmed to be derived from the myostatin gene or the transgene. Together with the in situ data(Fig. 5), these studies demonstrated that the myostatin prodomain transgene was inherited and expressed in a muscle-specific manner in the developing somite of transgenic zebrafish embryos. Moreover, this transgene is strongly expressed in adult skeletal muscles of the transgenic fish.

The development and growth of skeletal muscles were analyzed in the transgenic fish at the morphological, histological and molecular levels. Overall, there were no obvious morphological changes with the transgenic fish. Transgenic fish showed normal development and growth. To determine if inhibiting Myostatin function affected the expression of myogenic regulatory genes, we analyzed the expression of MyoD, myogenin and Myf-5 in transgenic fish embryos by in situ hybridization. There was no significant change in MyoD, myogenin or Myf5expression in transgenic zebrafish embryos(Fig. 7). In addition, we analyzed the formation of embryonic myofibers by F59 and MF20 antibody staining, and found no difference between transgenic and non-transgenic fish(data not shown). These results demonstrate that Myostatin may not be involved in myogenesis in early stage embryos. To determine if inhibiting Myostatin function affected the expression of myogenic regulatory genes at later stage when myostatin mRNA expression was increased, we analyzed the expression of MyoD, myogenin and Myf-5 in transgenic fish at days 7, 45 and 240 by quantitative RT-PCR. The results showed the expression levels of MyoD, myogenin and Myf-5 were very similar between transgenic and non-transgenic fish (Fig. 8, Table 1). Statistical analysis did not reveal any significant difference in myogenic gene expression between transgenic and non-transgenic fish at 2, 7, 45 and 240 days of age.

To determine whether or not Myostatin plays a role in regulating muscle growth in adult fish, F₂ transgenic fish were generated by crossing F₁ (hemizygous) transgenic males with transgenic females, or transgenic males with non-transgenic females. The F₂ generations from these crosses contained both homozygous and hemizygous transgenic fish as well as non-transgenic offspring. The body mass and muscle structures of the transgenic and non-transgenic fish were compared at several developmental stages. The hemizygous transgenic fish did not grow significantly larger than the non-transgenic siblings when analyzed at 3 and 5.5 months of age(Table 2). However, there appeared to be a significant increase in mass (10-15%) when homozygous transgenic fish were included in the analysis at 2.5 months of age(Table 2).

To determine whether or not the muscle growth was affected in the transgenic fish, samples of representative fish from transgenic and non-transgenic sibling were sectioned to determine fiber size and for number analysis at 3 and 5.5 months. The results showed that the transgenic fish had approximately 10% more myofibers than the non-transgenic fish that were in the similar body mass group (Table 3). To determine whether inhibiting Myostatin function caused hypertrophic muscle growth, fiber size was compared in transgenic and non-transgenic fish. Because the fiber size varies significantly in different cross section regions, we chose to compare myofibers from a defined region of the epaxial muscles where the fibers appear to be relatively uniform in size. The data revealed that there was no significant difference between fiber size in transgenic and non-transgenic fish(Table 3). Together, these data indicate that Myostatin may play an inhibitory role in hyperplastic myofiber formation, but has little effect on controlling the size of myofibers in zebrafish. Therefore, inhibiting Myostatin function in zebrafish results in a small but significant increase in muscle hyperplasia, but not in hypertrophy.

In this study, we isolated the zebrafish myostatin genomic gene and analyzed its expression and function in zebrafish embryos, larvae and adult skeletal muscles. Our data demonstrated that the zebrafish myostatin mRNA was weakly expressed in early stage zebrafish embryos. Its expression increased significantly in swimming larvae, juvenile and adult skeletal muscles. Transient expression analysis demonstrated that the zebrafish myostatin 5′ flanking sequence contained regulatory elements for muscle expression. To develop a zebrafish model for the functional study of Myostatin in fish, transgenic zebrafish expressing the Myostatin prodomain were generated. The Myostatin prodomain was specifically expressed in skeletal muscle cells using a muscle-specific promoter. The transgenic fish developed normally and showed no defect in muscle development of early stage embryos. The transgenic fish exhibited an increased hyperplastic muscle growth, but no obvious increase in hypertrophic muscle growth compared to non-transgenic siblings. Similar to the results observed in the present report with zebrafish, a significant increase in muscle fiber number (i.e. hyperplasia) has been observed in rainbow trout muscle transfected with a morpholino oligonucleotide directed against myostatin (T. Bradley, personal communication). These data demonstrated that inhibiting Myostatin function in fish had a positive effect in stimulating muscle growth.

Although fish myostatin genes share high sequence identity with their mammalian counterparts, myostatin mRNA expression in fish is different compared with that in mammals. In mice, myostatin mRNA is strongly expressed in developing somite and skeletal muscles, and weakly expressed in cardiomyocytes, mammary glands and adipose tissue(McPherron et al., 1997; Ji et al., 1998; Sharma et al., 1999). In fish,several studies have demonstrated that, in addition to muscle expression, myostatin mRNA was expressed in eyes, spleen, gill filaments,ovaries, gut and brain and, to a lesser extent, in testes(Rodgers et al., 2001; Maccatrozzo et al., 2001a; Roberts and Goetz, 2001; Ostbye et al., 2001; Radaelli et al., 2003). Moreover, in contrast to strong expression in developing somites in mouse embryos, little or no myostatin mRNA expression could be detected in developing somites of fish embryos by whole-mount in situhybridization. Myostatin mRNA expression in early stage zebrafish embryos could only be detected by RT-PCR. This is consistent with a recent report in seabream by Maccatrozzo et al.(2001a), and in zebrafish by Vianello et al. (2003).

The different pattern of myostatin expression in fish and mice raised the question of whether the myostatin that we focused on in this study was the mammalian homologue, and whether or not additional myostatin genes are present in zebrafish, given that zebrafish have more duplicated genes compared to mammals(Wittbrodt et al., 1998; Robinson-Rechavi et al.,2001). Our BLAST search found two additional myostatin-related genes in the zebrafish genome sequence. They share approximately 60-70% identity with the zebrafish myostatin gene used in this study. These two myostatin related genes, however, did not show any expression in developing somites when examined by in situhybridization (data not shown) and their DNA sequences share less identity with the mouse myostatin gene compared with the myostatingene characterized in this study.

Recently, a closely related gene GDF-11/BMP-11 was identified in human,mouse and frog. GDF-11 is expressed in many tissues other than the skeletal muscles, such as brain and eye (Nakashima et al., 1999; McPherron et al., 1999; Gamer et al.,1999). GDF-11 and myostatin are thought to be derived from the same ancestral gene through gene duplication. The myostatin-like genes in zebrafish could represent the zebrafish GDF-11. These data suggest that the duplication event that generated myostatin and GDF-11 might occur before the divergence of the fish species. It is unknown at present whether the functions of Myostatin and GDF-11 are highly specific, as in mammals. Further studies, especially the characterization of fish GDF-11 expression and function, will provide more insight into the possible function of these two highly related genes in fish.

Histological analysis revealed that the transgenic fish exhibited stratified hyperplasia (data not shown). Stratified hyperplasia generates new fibers along a distinct germinal layer. This type of hyperplasia is found in all fish species. In addition to stratified hyperplasia, another type of hyperplasia, mosaic hyperplasia, results in a large increase in total fiber number during juvenile growth, and is therefore very common in commercially important aquatic species that grow to a large size. Mosaic hyperplasia is greatly reduced or entirely lacking in species such as zebrafish, guppies and other fish that remain small (Van Raamsdonk et al., 1983; Weatherley and Gill, 1984, 1985; Weatherley et al., 1988). The lack of a dramatic effect on enhancing muscle growth in zebrafish could be due to the lack of mosaic hyperplasia in zebrafish. Nevertheless, inhibiting Myostatin function resulted in a small but significant increase in fiber numbers in the transgenic fish. This is consistent with the recent finding that growth hormone transgenic zebrafish only grow 20% faster than non-transgenic control (Morales et al.,2001). It will be interesting to determine if blocking Myostatin function in large aquatic species has a more dramatic effect in stimulating muscle growth.

The lack of a significant effect on hypertrophic growth in transgenic fish differs from previous findings in mice. In the myostatin knockout mice, the marked increase in muscle mass was attributed to both hypertrophy and hyperplasia (McPherron et al.,1997), while the transgenic mice expressing the Myostatin prodomain or the dominant negative form of Myostatin exhibited primarily hypertrophy (Zhu, 2000; Yang et al., 2001). The different response to Myostatin in fish and mice could be due to the different types of muscle growth in postnatal or post-larval stages. Postnatal muscle growth in mammals is largely contributed by hypertrophy. In contrast, in most fish species, muscle growth in post-embryonic life is attributed to continuous hyperplastic and hypertrophic growth (reviewed by Rowlerson and Veggetti, 2001). The less dramatic effect in fish could also be due to other reasons. Several studies have demonstrated that there is no correlation between changes in myostatin mRNA and Myostatin protein levels. Moreover, in zebrafish,it has been shown that most of the Myostatin proteins exist as precursor protein (Vianello et al.,2003). Therefore, overexpression of the Myostatin prodomain may not have a dramatic effect in inhibiting Myostatin activity in zebrafish.

The possibility that myostatin related genes may be also involved in the process should not be overlooked. Recently, Lee and McPherron(2001) demonstrated that overexpression of Follistatin, a TGF-β/BMPs inhibitor, in skeletal muscles of transgenic mice induced hyperplasia and hypertrophy. Interestingly,the muscle phenotype is more dramatic than that obtained from the myostatin knockout, suggesting that Follistatin may have an additional function than simply blocking Myostatin activity in skeletal muscles (Lee and McPherron,2001). Follistatin has been cloned in zebrafish and was found to be expressed in developing somite and skeletal muscles(Bauer et al., 1998). It remains to be determined if overexpressing Follistatin in fish skeletal muscles has a more dramatic effect in stimulating fish muscle growth.

Myostatin inhibits myoblast specification and differentiation by downregulating Pax3, Myf-5 and MyoD expression in myoblasts(Langley et al., 2002). In this study, we demonstrated that expression of MyoD, Myf-5 and myogenin was not significantly affected in transgenic zebrafish expressing the Myostatin prodomain. This result is not surprising considering that only a 10% increase in hyperplasia was found in transgenic fish.

Although myostatin mRNA is predominantly expressed in skeletal muscles, several studies have demonstrated that it is also expressed in other tissues. To understand its regulation of expression in muscle cells, we analyzed the activity of the zebrafish myostatin 5′ flanking sequence in zebrafish embryos. We found that the zebrafish myostatin5′ flanking sequence could drive GFP expression in muscle fibers of zebrafish embryos. We also noticed that GFP expression in myofibers driven by the myostatin flanking sequence was much stronger than myostatin mRNA expression observed by in situ hybridization. The discrepancy could be due to the fact that the fluorescent signal from GFP is more sensitive than staining from the whole-mount in situhybridization. In addition, for microinjection, a large number of copies(10⁶) of the transgene were injected into zebrafish embryos. Compared with the myostatin mRNA expression from the two copies of the endogenous gene, GFP expression from the injected transgene is expected to be significantly stronger. Moreover, it cannot be ruled out that the myostatin 5′ flanking sequence used in this study lacks some of the inhibitory elements that restrict high levels of myostatinexpression in early stage embryos.

Myostatin expression is restricted to specific types of muscle fibers (Kambadur et al., 1997; Ji et al., 1998; Kocamis et al., 1999; Roberts and Goetz, 2001; Rescan et al., 2001). Because of the mosaic nature of the transient expression assay, it was difficult to determine if the expression of the myostatin-GFP transgene was restricted to certain types of myofibers in zebrafish embryos. Future studies are required to generate the myostatin-GFP transgenic model that could be used to study the expression and regulation of myostatinexpression in zebrafish and to clarify whether its expression in zebrafish is restricted to certain types of myofibers during development and growth.

We would like to thank Bosheng Zhang and Tracy Robinson for assistance in screening the transgenic fish, Se-Jin Lee (Johns Hopkins University), Phil Hamilton and Bob Curtis (Cape Aquaculture Technologies) for helpful discussions during the execution of this project, and John Stubblefield for editorial assistance. We would also like to thank the two anonymous reviewers for their very constructive comments on the manuscript. This work was supported in part by a NIH grant to SJD (RO1GM58537) and a research contract from Cape Aquaculture Technologies. This publication is contribution number 04-606 from the Center of Marine Biotechnology, University of Maryland Biotechnology Institute.