The two myostatin genes of Atlantic salmon (Salmo salar ) are expressed in a variety of tissues Tone-Kari Østbye 1 , Trina Falck Galloway 2 , Christer Nielsen 1 , Irene Gabestad 1 , Tora Bardal 2 and Øivind Andersen 1 1 Institute of Aquaculture Research, Aas, Norway; 2 Norwegian University of Science and Technology, Department of Zoology, Trondheim, Norway Two myostatin isoforms were identified in Atlantic salmon (Salmo salar ) by RT-PCR, and genomic sequences encoding this negative muscle growth factor were for the first time isolated from a nonmammalian species. Salmon myostatin isoform I is transcribed in white skeletal muscle as a 2346-nucleotide mRNA species that encodes a pre- cursor protein of 373 amino acids. Salmon myostatin I shows 93% sequence identity with isoform II which was isolated from white muscle as a partial cDNA sequence of 1409 nucleotides. In contrast to the restricted gene expression of myostatin in mammals, salmon myostatin I and II mRNAs were identified by RT-PCR in multiple tissues, including white muscle, intestine, brain, gills, tongue and eye. In addition, isoform I mRNA was found in red skeletal muscle, heart, spleen, and ovarian tissue. Using polyclonal antibodies against both isoforms, a 55-kDa precursor protein was detected by Western blot analysis in the red and white skeletal muscle, heart, intestine, and brain. Immunoreactive peptides of 35–40 kDa were identified in the gills, tongue, spleen, and head kidney, while the 25-kDa mature myostatin was found in the eye and serum, and in vitro expressed in rabbit reticulocyte lysate. Salmon myostatin was immunohistochemically localized in the sarcoplasma of red and white muscle fibres, in intestinal epithelial cells, at the basis of the branchial primary lamellae, and in odontoblasts and ameloblasts of the tongue teeth. The results indicate that the role of fish myostatin may not be restricted to muscle growth regulation, but may have additional functions similar to the growth/differentiation factor-11 in mammals. Keywords: myostatin; Atlantic salmon; Salmo salar; GDF-11; TGF-b. Myostatin is a recently identified member of the transform- ing growth factor-b (TGF-b) superfamily [1], which com- prises multiple growth/differentiation factors (GDF) that play important roles in the regulation of embryonic devel- opment and maintenance of tissue homeostasis in adults. At early embryonic stages, myostatin is restricted to the myo- tome compartment of the developing somites, and myostatin has been proposed to play an essential role in skeletal, but also cardiac, muscle growth and development [1,2]. The first muscle fibers form in the embryo when the myoblasts cease dividing and fuse into multinucleated myotubes [3]. The involvement of myostatin in these events was suggested from embryonic studies of cattle [4,5], pig [6], and chicken [7], demonstrating that the developmental pattern of myo- statin mRNA abundance coincides roughly with the progression of muscle fiber formation. The inhibitory effect of myostatin on myoblast proliferation [8] was further elucidated by showing that myostatin prevents the pro- gression of myoblasts from G1 to S phase of the cell cycle [9]. Myostatin as a negative regulator of myoblast number, and hence fiber number, was confirmed by studying myo- statin knock-out mice, which showed a dramatic increase in skeletal muscle mass due to muscle fibre hyperplasia, but also hypertrophy [1]. Furthermore, in the myostatin coding sequence of double-muscled cattle breeds, several mutations have been identified that were predicted to disrupt the function of the protein [1,4,10,11]. Myostatin cDNA sequences have been isolated from representatives of the five vertebrate groups [1], and addi- tional genomic sequences have been reported from human, porcine, and ovine myostatin [12,13] (GenBank accession no. AF266758). A highly related factor termed GDF-11, which plays a unique role in patterning both mesodermal and neural tissues in the developing embryo, was recently identified in multiple tissues in mammals [14 – 16]. Here we report on the isolation of two genetic distinct myostatin isoforms in Atlantic salmon, and show that their expression patterns are not limited to skeletal muscle, but similar to those of mammalian myostatin and GDF-11 together. During the preparation of this manuscript, two studies of myostatin in tilapia, white bass, and brook trout were presented [17,18]. Altogether the results indicate that the functional role of fish myostatin may not be restricted to muscle growth regulation. We propose that a common ancestral gene duplicated after the origin of fishes and diverged into the two factors myostatin (formerly GDF-8) and GDF-11, which have adopted distinct functions in mammals. Correspondence to Ø. Andersen, Institute of Aquaculture Research, PO Box 5010, N-1432 Aas, Norway. Tel.: 1 47 64949500, Fax: 1 47 64949502, E-mail: [email protected]Note: web page available at: http://www.akvaforsk.no (Received 5 July 2001, accepted 15 August 2001) Abbreviations: TGF, transforming growth factor; GDF, growth/ differentiation factor; LAP, latency associated peptide Eur. J. Biochem. 268, 5249–5257 (2001) q FEBS 2001
Transcript
The two myostatin genes of Atlantic salmon (Salmo salar ) areexpressed in a variety of tissues
Myostatin is a recently identified member of the transform-ing growth factor-b (TGF-b) superfamily [1], which com-prises multiple growth/differentiation factors (GDF) thatplay important roles in the regulation of embryonic devel-opment and maintenance of tissue homeostasis in adults. Atearly embryonic stages, myostatin is restricted to the myo-tome compartment of the developing somites, and myostatinhas been proposed to play an essential role in skeletal, butalso cardiac, muscle growth and development [1,2]. The firstmuscle fibers form in the embryo when the myoblasts ceasedividing and fuse into multinucleated myotubes [3]. Theinvolvement of myostatin in these events was suggestedfrom embryonic studies of cattle [4,5], pig [6], and chicken[7], demonstrating that the developmental pattern of myo-statin mRNA abundance coincides roughly with theprogression of muscle fiber formation. The inhibitory effectof myostatin on myoblast proliferation [8] was furtherelucidated by showing that myostatin prevents the pro-gression of myoblasts from G1 to S phase of the cell cycle
[9]. Myostatin as a negative regulator of myoblast number,and hence fiber number, was confirmed by studying myo-statin knock-out mice, which showed a dramatic increase inskeletal muscle mass due to muscle fibre hyperplasia, butalso hypertrophy [1]. Furthermore, in the myostatin codingsequence of double-muscled cattle breeds, several mutationshave been identified that were predicted to disrupt thefunction of the protein [1,4,10,11].
Myostatin cDNA sequences have been isolated fromrepresentatives of the five vertebrate groups [1], and addi-tional genomic sequences have been reported from human,porcine, and ovine myostatin [12,13] (GenBank accessionno. AF266758). A highly related factor termed GDF-11,which plays a unique role in patterning both mesodermaland neural tissues in the developing embryo, was recentlyidentified in multiple tissues in mammals [14–16]. Here wereport on the isolation of two genetic distinct myostatinisoforms in Atlantic salmon, and show that their expressionpatterns are not limited to skeletal muscle, but similar tothose of mammalian myostatin and GDF-11 together.During the preparation of this manuscript, two studies ofmyostatin in tilapia, white bass, and brook trout werepresented [17,18]. Altogether the results indicate that thefunctional role of fish myostatin may not be restricted tomuscle growth regulation. We propose that a commonancestral gene duplicated after the origin of fishes anddiverged into the two factors myostatin (formerly GDF-8) andGDF-11, which have adopted distinct functions in mammals.
Correspondence to Ø. Andersen, Institute of Aquaculture Research, PO
Eur. J. Biochem. 268, 5249–5257 (2001) q FEBS 2001
M A T E R I A L S A N D M E T H O D S
Isolation of RNA and genomic DNA
Juvenile Atlantic salmon at the stages prior to (parr) andafter (smolt) seawater adaptation, and adults were obtainedfrom a local fish farm at the Agricultural University ofNorway, Aas. The fish were killed, and red and whiteskeletal muscle, heart, mid-intestine, brain, liver, gills, headkidney, spleen, tongue, eye, and gonads were immediatelydissected and frozen in liquid nitrogen together with citratedplasma. Total RNA was isolated from the sampled organsusing Trizol (Boehringer Mannheim), and the cDNA wasobtained by reverse transcribing the mRNA using aT-primed First-Strand kit (Pharmacia Biotech). GenomicDNA was isolated from skeletal muscle tissue (DNeasyTissue Kit, Qiagen).
Cloning of salmon myostatin I and II encoding sequences
The strategy used for PCR amplifying the salmon myostatincDNAs of white skeletal muscle is outlined in Fig. 1. Theinitial PCR was performed on the oligo(dT)-tailed primedmuscle cDNA using the degenerate primers Myo1 and Myo2located in conserved regions of myostatin [1]. Upstreamsequences of salmon myostatin I were amplified using a50 RACE kit (Gibco BRL) according to the manufacturer’sinstructions. The supplied 50 RACE anchor primer and theuniversal amplification primer were used in combinationwith Myo3 and Myo4. The 30 end of myostatin I wasamplified by 30 RACE using the tailed oligo(dT) primer andthe adaptor primer [19] in combinations with the specificprimers Myo5, Myo6, and Myo7. The resulting full-lengthcDNA sequence was then confirmed by amplifying andsequencing the region from nucleotide 225 to nucleotide1362 using Myo11 and Myo12. In order to identify intronsequences of the salmon myostatin I gene, overlappingregions of the cDNA sequence were PCR amplified withdifferent primer combinations, and compared with the
corresponding genomic PCR fragments generated fromsalmon genomic DNA. The primers Myo21 and Myo22were included in order to sequence the 1.2-kb intron 2.
Based on the recent report on two isoforms of brook troutmyostatin [18], the initial PCR clones of salmon myostatincDNA were further examined. A second isoform of salmonmyostatin was identified, and upstream and downstreamsequences were amplified using the primer sets Myo23–Myo24 and Myo26–Myo12, respectively.
All PCRs were run on a Gene Amp PCR system2400 (PerkinElmer) by denaturating at 94 8C for 4 min,followed by 35–40 cycles of amplification at 94 8C for 30 s,58–65 8C for 30 s, and 72 8C for 30–90 s. Amplified frag-ments were visualized by agarose gel electrophoresis andligated into pGEM-T Easy vector (Promega). Plasmidswere prepared from transformed JM109 competent cells(Promega), and the inserts were automatically sequenced(ABI 377) in both directions using Big Dye DNAsequencing kit (PE Biosystems).
Myostatin I and II mRNA expression
The expression patterns of myostatin I and II mRNA wereexamined by RT-PCR. One microgram of total RNA wasisolated from the different organs and reverse transcribedinto cDNA as outlined above. The sensitivity of the RT-PCRtechnique was enhanced by performing two successivePCRs using a 100 times dilution of the first PCR products astemplates in the second PCR. Due to the sequence similaritybetween myostatin I and II all the amplified products wereexamined by sequencing. For detection of myostatin I cDNA,a 560-nucleotide cDNA fragment was amplified usingprimers Myo8 and Myo9 in the first PCR followed by Myo8and the nested primer Myo10 in the second PCR. Myo8is located at the junction of exon 1 and intron 1, thuspreventing the amplification of genomic DNA. Myostatin IIcDNA was identified by amplifying an 809-nucleotideregion using the two primer sets Myo23–Myo24 andMyo23–Myo25 in the two successive PCRs, respectively.
Fig. 1. Schematic diagram of the PCR amplification of Atlantic salmon myostatin I and II encoding sequences. The scale denotes numbers of
nucleotides of the cDNA sequence. Location and orientation of the primers are indicated by arrows. The positions of the introns are indicated.
5250 T.-K. Østbye et al. (Eur. J. Biochem. 268) q FEBS 2001
Both Myo24 and Myo25 are specific for salmon myostatinisoform II. The amplified region includes intron 1, and anygenomic DNA contaminations would result in a PCRproduct of about 1300 nucleotides.
Protein analysis
Antiserum preparation. Antibodies were produced againstthe highly conserved C-terminal part of salmon myostatin Iand II in order to identify the precursor, the propeptide,and the mature myostatin. The hydrophilic peptide frag-ment N367-I382 (Fig. 2) was synthesized and conjugated tokeyhole limpet hemocyanin by EuroGentec, Belgium. Theconjugate was emulsified in Freunds complete adjuvants andinjected subcutaneously into two rabbits after collection ofpreimmune serum. The rabbits were finally bled after fourmonths of treatment.
Western blot analysis. The sampled organs were homo-genized in NaCl/Pi with 1 mM phenylmethyl sulfonylfluoride. The extracted proteins (50–250 mg) were separatedby SDS/PAGE under reducing conditions, and transferred tonitrocellulose membrane by electroblotting. After blockingwith 3% BSA, the blot was incubated with the specific rabbitantisera diluted 1 : 200 000 in NaCl/Tris. The membranewas washed in NaCl/Tris with 0.1% Tween 20, and thenincubated in a 1 : 5000 dilution of goat anti-(rabbit IgGH 1 L) Ig (Southern Biotechnology) conjugated to horse-radish peroxidase. Salmon myostatin was visualized with3,30diamino-benzidine tetrahydrochloride (Sigma) as sub-strate according to the manufacturer’s protocol.
Immunohistochemistry. Skeletal muscle, tongue, eye, mid-intestine, gills, heart, and kidney tissue from three salmonparr were frozen in 2-methylbutane precooled in liquidnitrogen, and stored at 280 8C until further processing.Prior to sectioning, the tissue blocks were acclimatized to220 8C for 1 h in the cryostat. Frozen sections, 4–6 mmthick, were cut, air dried and fixed in cold acetone. Thesections were then rinsed in NaCl/Pi, immersed in 50 mM
ammonium chloride and incubated for 3 h in a 1 : 500dilution of the primary antiserum in 10% fetal bovine serumin NaCl/Pi. After washing in NaCl/Pi, the sections wereincubated for 30 min in a fluorescence-labelled goat anti-(rabbit IgG) secondary Ig (Alexa 488; Molecular Probes,Eugene) at a final concentration of 10 mg:mL21. Finally, thesections were washed in NaCl/Pi, mounted under cover slipswith poly(vinyl alcohol), and studied in a Zeiss fluorescencemicroscope with a no. 16 filter set. Negative controls(preimmune or no rabbit serum) were included in theprocedure.
In vitro expression. The protein-coding region of the salmonmyostatin I cDNA was PCR amplified using primers Myo11and Myo12 and ligated into pGEM-T Easy vector(Promega) downstream of the T7 promoter. Myostatin Iwas in vitro expressed by adding 1 mg of the circularplasmid to a rabbit reticulocyte lysate (TNT kit, Promega),which was incubated at 30 8C for 90 min according to themanufacturer’s protocol. The expressed myostatin wasanalysed by Western blotting after SDS/PAGE of 5 mglysate under reducing conditions.
Fig. 2. Atlantic salmon white skeletal muscle myostatin I (S.s. I) and II (S.s. II) cDNAs. The predicted amino-acid sequences are given in single
letters, and amino-acid differences between isoform I and II are indicated by I/II. The putative signal sequence is given in italics, and antibodies were
raised against the hydrophilic fragment shown in bold type. The tetrabasic cleavage site is boxed. The positions of intron 1 and 2 are indicated. The
cDNA sequences have been deposited into the GenBank database (accession nos. AJ297267 and AJ344158).
q FEBS 2001 Two Atlantic salmon myostatin genes (Eur. J. Biochem. 268) 5251
A full-length cDNA of 2346 nucleotides encoding Atlanticsalmon myostatin isoform I was isolated by performingRT-PCR on pooled RNA isolated from the white skeletalmuscle of four juveniles. Initially, an < 700-nucleotidecDNA fragment was PCR amplified using the primers Myo1and Myo2 that replicate the conserved sequences D132–F139 and I365–S373, respectively (Fig. 2). The subsequent50 and 30 RACE resulted in the isolation of the upstream anddownstream sequences of the salmon myostatin I cDNAincluding the 117-nucleotide 50 UTR and the 1110-nucleo-tide 30 UTR, respectively. The very recent identification oftwo distinct isoforms of brook trout myostatin [18]prompted us to search for a second type of salmon myo-statin among the initial PCR clones. Sequence alignmentrevealed a second isoform, and upstream sequences of thismyostatin II were amplified from white skeletal musclecDNA using the specific primer Myo24 together withMyo23 (Fig. 2). Downstream sequences were amplifiedusing the primer set Myo26–Myo12 resulting in a 1409-nucleotide cDNA sequence encoding salmon myostatin
isoform II. The predicted myostatin I and II proteins consistof 373 amino acids, with an estimated molecular mass of42 kDa. As other members of the TGF-b superfamily, thesalmon myostatins seem to be synthesized as precursorproteins, each possessing two putative processing signalsequences (Fig. 2). Removing the putative 22-amino-acidhydrophobic leader sequence gives a propeptide estimatedto nearly 40 kDa. The proteolytic cleavage of this predictedlatent form of myostatin at the highly conserved tetrabasiccleavage site KRSRR would then separate the 12-kDaC-terminal mature peptide from the 242-amino-acid sequence termed the latency associated peptide (LAP)[20].
Intron sequences of the salmon myostatin genes wereidentified by comparing the length of amplified cDNAfragments with the corresponding genomic DNA fragments.Two primer sets, Myo18–Myo4 and Myo19–Myo20,amplified genomic fragments that were longer than expected,suggesting the presence of intron sequences. Two intronswere identified in both genes, intron 1 is inserted betweennucleotide 373–374 and intron 2 between nucleotide 741–742 (Fig. 2). Only the introns of the myostatin I gene werecompletely sequenced. Intron 1 consists of 529 bp, andintron 2 is approximately 1200 bp long due to a variableregion (T.-K. Østbye, O. Frang & Ø. Andersen, unpublishedresults; Genbank accession no. AJ316006).
Protein analyses
Western blot analysis of the sampled salmon organs usingantibodies against the highly conserved C-terminus of salmonmyostatin resulted in the identification of four majorimmunoreactive peptides with calculated molecular mass ofapproximately 55, 40, 35, and 25 kDa (Fig. 3). The amountof the myostatin peptide forms varied between the differentorgans and the examined individuals. In general, the putativeprecursor of 55 kDa was identified in white and red skeletalmuscle and in the intestine, but also in the heart and brain(not shown). Several bands close to this 55 kDa specieswere shown at varying intensities, in addition to a 45-kDapeptide observed in the skeletal muscle of some individuals(not shown). Shorter immunoreactive peptides of 40 and35 kDa were found in the gills and tongue, respectively. The35-kDa peptide was also observed in the spleen and headkidney (not shown). Salmon myostatin I was expressed invitro as a 25-kDa peptide in rabbit reticulocyte lysate, that issimilar to the myostatin peptide form found in the eye and
Fig. 3. Western blot analysis of Atlantic
salmon myostatin. Extracted proteins from the
sampled organs were separated by SDS/PAGE
under reducing conditions. 1, intestine; 2, white
muscle; 3, red muscle; 4, gills; 5, tongue; 6,
myostatin expressed in vitro; 7, eye; 8, serum.
In order to visualize the different immunoreactive
peptides each lane of the blotted membrane had
to be treated separately in the 3,30diamino-
benzidine tetrahydrochloride colour reaction.
Molecular markers (kDa) are shown on the left
side.
5252 T.-K. Østbye et al. (Eur. J. Biochem. 268) q FEBS 2001
serum. No immunoreactive myostatin was detected in theliver and ovary.
Salmon myostatin was immunohistochemically localizedto the sarcoplasm of white and red muscle fibres of thesalmon parr (Fig. 4A), and there seemed to be quantitativelymore in white than in red fibres. In the tongue, myostatinwas found in early developmental stages of the teeth
localized here (Fig. 4B). Myostatin was identified in thedental pulp (odontoblasts) and at the surface of the enamel(ameloblasts) of the tongue teeth (Fig. 4C), but also intongue epithelial cells at the basis of goblet cells. In gills,myostatin was found at the basis of the primary lamellae incells identified either as mucus or chloride cells (not shown).Both heart muscle cells and some intestinal epithelial cellscontained myostatin, whereas the signals observed in theretina were not conclusive. No myostatin was immunohis-tochemically identified in the head kidney.
Expression of myostatin I and II mRNAs
The gene expression of salmon myostatin in juvenile fishwas further examined by RT-PCR by performing twosuccessive PCRs using nested primers. In the first PCR, the686-nucleotide myostatin I cDNA fragment flanked by theprimers Myo8 and Myo9 was easily amplified from whiteskeletal muscle and brain, but a faint signal was alsoidentified in the intestine, gills, tongue, and eye, in additionto ovarial tissue of adult fish. To increase the sensitivity ofthe method, a second PCR was performed using these PCRproducts as template, and Myo10 as nested primer togetherwith Myo8. The result was a single band of 560 nucleotidesamplified from all the examined organs, except the liver andhead kidney (Fig. 5A). The amplified products weresequenced and shown to be identical to the myostatinisoform I. PCR amplifications of genomic DNA failed due tothe location of primer Myo8 at the junction between exon 1and intron 1.
Similar to isoform I, myostatin II mRNA was mainlyfound in the white skeletal muscle and brain of juvenile fishafter the first PCR round, but also in the intestine, tongue,and eye. In the second PCR, myostatin II cDNA was highlyamplified from all these organs, in addition to the gills(Fig. 5B), which contained only low amounts of myostatin IcDNA. In contrast, no myostatin II was detected in redmuscle, heart, spleen, and ovary. All the amplified productswere sequenced and shown to be identical to the myo-statin II encoding sequences. The specificity of the involvedmyostatin II primers was further confirmed by the lackof PCR products using myostatin I cDNA as template(MSTN I ctrl lane, Fig. 5B). The PCR product of about 1300nucleotides amplified from liver, spleen, and ovary wasshown to include intron sequences from genomic DNA,which presumably is a result of contamination from theRNA isolation procedure. The corresponding productamplified from salmon genomic DNA is hardly visibleon the gel, as this control was only included in the secondPCR.
Sequence comparison of vertebrate myostatins
The salmon myostatin I and II precursors share 93% overallsequence identity (Fig. 6A). Salmon myostatin I shows 95%identity with the brain/muscle (b/m) myostatin isoform ofbrook trout, compared to 91% identity with the ovarianisoform of brook trout. In contrast, salmon myostatin II ismore similar (99% identity) to the ovarian isoform thanto the b/m isoform (92%). In comparison, the sequenceidentities between the two salmon myostatins and theorthologs of the two nonsalmonid teleost species tilapia andwhite bass are about 80%. Furthermore, salmon myostatin I
Fig. 4. Immunohistochemical detection of Atlantic salmon myo-
statin. (A) White skeletal muscle fibres (W) with signals in sarcoplasm.
(B) Early stage of developing tongue tooth showing signals seen in
odontoblasts of dental pulp (DP), but not in the dentin (D). (C) Detail
from (B) with signals in ameloblasts (AB). All scales are 50 mm. No
signals were found in the negative controls.
q FEBS 2001 Two Atlantic salmon myostatin genes (Eur. J. Biochem. 268) 5253
and II show 61 and 62% identity, respectively, with humanmyostatin, compared to 57 and 58% identity with humanGDF-11. Finally, the salmon myostatin isoforms and thesecond known salmonid member of the TGF-b superfamilyshare only 14% overall sequence identity. However, the
sequences of the mature TGF-b members are highlyconserved, including the location of nine cysteines involvedin the dimerization and the formation of the characteristiccysteine knot structure of this superfamily [21,22] (Fig. 6B).Interestingly, the mature myostatins of salmon, tilapia, and
Fig. 6. Sequence comparison of teleost myostatins. (A) The overall amino-acid sequence identity (%) between the myostatins of Atlantic
salmon (S.s. I and S.s. II), brook trout ovarian isoform (O.t. b/m) and brain/muscle isoform (O.t. ov), zebrafish (D.r.) and Tilapia (O.m). Human
myostatin (H.s.) and GDF-11, and a rainbow trout TGFb member (accession no. AJ007836) are included for comparison. (B) Sequence alignment
of the predicted mature myostatins of salmon, Tilapia, zebrafish, and man. Human GDF-11 and a rainbow trout TGFb member are included
for comparison. Residues identical to the salmon myostatin sequence are indicated by dots (:), and spaces (–) are introduced to optimize the
alignment.
Fig. 5. Expression of Atlantic salmon
myostatin I. cDNA sequences of (A) myostatin I
and (B) myostatin II were amplified by RT-PCR
from total RNA of different organs performing
two successive PCRs for each isoform. Myostatin I
cDNA was included as a positive control in (A),
but as a negative control in (B). Hae III digested
fX174 DNA marker is indicated on the right
side.
5254 T.-K. Østbye et al. (Eur. J. Biochem. 268) q FEBS 2001
zebrafish are almost as similar to GDF-11 (86% identity) asto mammalian myostatins (88%).
D I S C U S S I O N
This study presents two distinct genes of Atlantic salmonencoding two isoforms of the negative muscle growthregulator myostatin. In contrast to mammalian myostatin,which is restricted to skeletal and cardiac muscle, mammarygland, and adipose tissue [2,6], the two isoforms of salmonmyostatin are expressed in a variety of tissues, includingwhite skeletal muscle, brain, intestine, gills, tongue, and eye.Interestingly, only salmon myostatin I was identified in redmuscle, heart, spleen, and ovary. The sequence comparisonindicates that the salmon isoform I corresponds to theisoform termed b/m of brook trout myostatin, which wasshown by Northern blot to be expressed in brain and redmuscle, but not in ovarian tissue [18]. On the other hand,salmon isoform II, which is very similar to the ovarianisoform of brook trout, was not identified in ovarian tissueby the sensitive technique of RT-PCR. Furthermore, incontrast to salmon isoform II the brook trout ovarian isoformwas also identified in red muscle using the unspecificovarian isoform cDNA as probe in the Northern blot analysis[18]. The discrepancies might also be due to variations inexpression levels during the different stages of reproductionand muscle development. In addition, species-specific expres-sion of myostatin in red and white muscle was reported infour teleost species [18]. The expression patterns of salmonmyostatin I and II together are highly similar to that oftilapia myostatin, including the lack of transcripts in liverand head kidney [17]. The origin of the myostatin identifiedin the salmon head kidney by Western blot analysis istherefore unknown.
Salmon myostatin was immunohistochemically localizedin odontoblast and ameloblasts of the developing tongueteeth. Fish in general exhibit well-developed dentition, andin salmon the teeth are found on the tongue, jaws, andpharynx. In mammals, members of the TGF-b superfamilyregulate the early steps of tooth morphogenesis, and differ-entiation of odontoblasts and ameloblasts [23]. By screeningdental pulp RNAs from mouse embryos, Nakashima et al.[15] identified the novel factor GDF-11, which is highlyexpressed in the terminally differentiated odontoblasts.Mouse GDF-11 was also found in the mandibular and thehyoid arches [15], which represents the first and secondvisceral arches, whereas the gill arches of fish constitute the3rd–7th visceral arches. Thus, the expression of myostatinin the salmon gills corresponds to the expression of GDF-11in the mice branchial arches. Finally, immunohistochemicalstudies of the salmon eye indicated the expression of myo-statin in the retina, which is similar to the report on GDF-11expressed in the inner layer of the optic cup, the outer layerof the retina, and weakly in the lens of the developing mouseeye [15]. Altogether, the expression patterns of the salmonmyostatins are highly consistent with the patterns of mam-malian myostatin and GDF-11 together. Furthermore,sequence aligment reveals that the mature myostatins ofthe unrelated species salmon, tilapia, and zebrafish arenearly as similar to mammalian GDF-11 (86% identity) asto mammalian myostatin (88%), and several residues areunique to fish myostatin and mammalian GDF-11 (Fig. 6B).However, in contrast to all the known myostatins to date, the
three identified mammalian GDF-11 contain conservedrepeats of 11 alanine and six glycine residues. Thus, theduplicated genes of Atlantic salmon and brook trout bothencode myostatin, and not GDF-11 orthologs. The presenceof two isoforms of myostatin in salmonids are probably aresult of the tetraploid evolution of the salmonid genome[24]. The resulting duplicated genes may then have acquirednew functions by modifications of the protein structureand/or the expression pattern [25]. Whereas the primarysequences of the biologically active part of salmonmyostatin I and II are almost identical, their expressionpatterns are different. Due to the mutual exclusive expres-sion domains of myostatin and GDF-11 in mice, Gamer et al.[14] suggested that these two factors adopted differentfunctions as they evolved from a common ancestral gene.A common receptor for the two factors may still existsas suggested by Lee & McPherron [26], which is supportedby the observation that GDF-11 and myostatin appearto have similar activity in the Xenopus animal cap assay[14]. The ubiquitous expression of myostatin in both tilapiaand salmonids might suggest that the fish myostatin generepresents the proposed ancestral gene. The gene dupli-cation and divergence of the myostatins in salmonids, whichdid not give rise to the terrestrial vertebrates, may however,demonstrate how this ancestral fish gene duplicated andradiated during the evolution of vertebrates into the twomammalian TGFs myostatin and GDF-11.
All members of the TGF-b superfamily, includingmyostatin, are synthesized as a precursor protein, whichis normally secreted as a homodimeric propeptide afterremoval of the signal peptide [20]. TGF-b is then activatedby proteolytic cleavage of the latent form that releases themature C-terminal region from the complexed N-terminalLAP domain [27]. In salmon, both the precursor and severalprocessed forms of myostatin were identified, as previouslyreported from mammalian in vivo and in vitro studies. The55-kDa peptide identified in extracts of salmon skeletalmuscle, intestine, heart, and brain corresponds to the pre-cursor identified in bovine myoblasts [9] and in ovine heartextract [2]. Shorter forms of 35–45 kDa were identified inthe tongue, gills, head kidney, spleen, and skeletal muscle.This is similar to the myostatin monomers of 35 and 37 kDafound in rat plantaris and soleus muscle, respectively [28],and the 37-kDa processed form identified in the skeletalmuscle of normal mice [29]. However, these authors did notdetect this processed form in transgenic mice expressingmyostatin lacking its normal cleavage site. This result isconsistent with the idea that these immunoractive salmonpeptides are the latent propeptide of myostatin containingthe N-terminal LAP. The LAP region is apparently removedin the 25-kDa salmon myostatin expressed in vitro in rabbitreticulocyte lysate. This putative mature salmon myostatinwas also identified in the eye and in serum under reducingconditions, which corresponds with the 26-kDa monomer ofbovine myostatin expressed in E. coli, and identified in thebovine and human skeletal muscle and plasma [2,9,12]. Incontrast, tilapia and mouse myostatin were expressed invitro in human embryonal kidney cells and Chinese hamsterovary cells, respectively, as mature forms of 13 and 15 kDa,in addition to the mouse precursor of 52 kDa [1,17]. As themyostatin dimer is expected to be cleaved under the givenexperimental conditions, the observed differences betweenthese putative monomers of the mature myostatin have been
q FEBS 2001 Two Atlantic salmon myostatin genes (Eur. J. Biochem. 268) 5255
explained by species- or cell-specific differences in theregulatory mechanisms of post-translational modifications[12]. Glycosylation is common in the TGF-b superfamily[30,31], and the potential N-glycosylation sites N305 andN71 (Fig. 2) are highly conserved among the vertebratemyostatins. Such modifications of the peptide may alsoexplain the discrepancy between the observed and calcu-lated molecular masses of myostatin. It should be noted theprepared antibodies would not discriminate any differencesin post-translational modifications between the two salmonmyostatin isoforms.
In fish, with some exceptions, muscle fibre recruitmentcontinues into the adult stages of the life cycle [32]. Thenumbers of white muscle fibres per myotome in Atlanticsalmon were estimated to be 180 000 in sea-water adaptedsmolts and more than one million after two winter at sea[33,34]. The hypertrophic effect of myostatin was suggestedto be independent of its hyperplasic effect in transgenic miceexpressing a dominant negative myostatin mutant [29]. Inthe developing tilapia larvae the expression of myostatinseemed to be regulated and peaked with the appearance ofmuscle activity [17]. In contrast, Northern blot analysis ofadult tilapia skeletal muscle RNA failed to identify myo-statin. In juvenile salmon, myostatin was widely, but nothighly, expressed in the skeletal muscle fibers. The lowerlevel of salmon myostatin in the red vs. white fibers isconsistent with that of little tunny (Euthynnus alletteratus ),but opposite to the myostatin expression in king mackerel(Scomberomus cavalla ) and yellow perch (Perca flavescens )[18]. Due to the hypertrophic and hyperplasic postem-bryonic muscle growth and the distinct location of the fibertypes, fish should provide an excellent experimental modelfor further elucidating the roles of myostatin.
