Journal of Muscle Research and Cell Motility I3, 551-572 (1992)

Myosin heavy chain composition of single fibres and their origins and distribution in developing fascicles of sheep tibialis cranialis muscles A L F R E D M A I E R 1., J O H N C. M c E W A N 2, K E N N E T H G. D O D D S 3, D O N A L D A. F I S C H M A N 4, R O B I N B. F I T Z S I M O N S 5 and A. J O H N

HARRIS 2

IDepartment of Cell Biology, University of Alabama at Birmingham, Birmingham AL 35294, USA 2Centre for Neuroscience and Department of Physiology, University of Otago, Dunedin, New Zealand 3MAF Technology, Invermay, Private Bag, Mosgiel, New Zealand 4Cornell University Medical College, Department of Cell Biology and Anatomy, 1300 York Avenue, New York, N Y 10021, USA SMRC Laboratory of Molecular Biology, Hills Road, Cambridge CB2 2QH, UK Received 24 October 1991; revised 25 February 1992; accepted 7 March 1992

Summary The myosin heavy chain (MHC) composition of single muscle fibres in developing sheep tibialis cranialis muscles was examined immunohistochemically with monoclonal antibodies to MHC isozymes. Data were collected with conventional microscopy and computerized image analysis from embryonic day (E) 76 to postnatal day (PN) 20, and from adult animals. At E76, 23% of the young myofibres stained for slow-twitch MHC. The number of these fibres considerably exceeded the number of primary and secondary myotubes. By EI00, smaller fibres, negative for slow-twitch MHC, encircled each fibre from the initial population to form rosettes. A second population of small fibres appeared in the unoccupied spaces between rosettes. Small fibres, whether belonging to rosettes or not, did not initially express slow-twitch MHC, expressing mainly neonatal myosin instead. These small fibres then diverged into three separate groups. In the first group most fibres transiently expressed adult fast myosin (maximal at EI10-E120), but in the adult expressed slow myosin. This transformation to the slow MHC phenotype Commenced at EII0, was nearing completion by 20 postnatal days, and was responsible for approximately 60% of the adult slow twitch fibre population. In the other two groups expression of adult fast MHC was maintained, and in the adult they accounted for I4% (IIa MHC) and 17% (IIb MHC) of the total fibre numbers. We conclude that muscle fibre formation in this large muscle involves at least three generations of myotube. Secondary myotubes are generated on a framework of primary myotubes and both populations differentiate into the young myofibres which we observed at E76 to form rosettes. Tertiary myotubes, in turn, appear in the spaces between rosettes and along the borders of fascicles, using the outer fibres of rosettes as scaffolds.

Introduction Recent electron microscopic evidence from developing sheep tibialis cranialis muscle (Wilson et al., 1992) suggests that in addition to the primary and secondary generations of myotubes already described in small laboratory rodents (Kelly & Zacks, 1969; Ontell & Dunn, 1978; Harris, 1981), a tertiary generation of myotubes (Draeger et al., 1987) may also make a significant contribution to adult muscle fibre numbers. Studies in developing rat muscles (Duxson et al., 1989) show that *To whom correspondence should be addressed. 0142-4319 9

1992 Chapman & Hall

secondary myotubes begin to form by the fusion of secondary myoblasts near the neuromuscular junction of primary myotubes. Subsequently, longitudinal growth of the smaller fibre is guided by complex infolding of its sarcolemma into that of the primary myotube (Duxson & Usson, 1989), so that primary myotubes serve as a 'cellular skeleton' (Kelly, 1983) to define the structure of the adult muscle. In laboratory rodents, this mechanism of alignment of secondary and primary myotubes, and their eventual detachment from each other, produces the final number of muscle fibres in an adult muscle (Ross et al., 1987). One difficulty in applying this model to large muscles, such as the lateral head of the tibialis cranialis

552

MAIER, McEWAN, DODDS, FISCHMAN, FITZSIMONS and HARRIS

muscle in the sheep, is to understand how the 440 primary myotubes in this muscle (Wilson et al., 1992) can provide sufficient sarcolemmal surface for the alignment of the appropriate number of secondary myotubes to produce the final number of 30 000 fibres seen in single crosssections in the adult muscle. For this to be accomplished, each primary myotube would have to accommodate 70 secondary myotubes, an unlikely event if one considers the multiple sets of alignment and detachment this would take. An alternate and intuitively simpler mechanism by which to arrive at the total fibre population would be through third and even higher orders of generation of myotubes aligned on appropriate lower order generation myofibres. Evidence supporting this possibility has recently been provided from a study of human fetal muscles (Draeger et al., 1987), and in an electron microscope study of early development in the sheep tibialis cranialis (Wilson et al., 1992). With currently available technology it would be extremely time-consuming to examine the development of an entire muscle the size of tibialis cranialis by single fibre analysis. Fortunately, its fascicles resemble in many ways the whole muscle, and they are more manageable in size. Except for a narrow region at the periphery of the muscle, its fascicles are homogeneous in their fibre type composition. This study was designed to explore, using single fibre analysis, the contributions of different myofibre generations to the various groups of fibre type and hence to the total fibre pool of typical fascicles located in the more central regions of a mixed muscle. Our system employed the relative degree of reactivity of individual fibres with four monoclonal antibodies,: against myosin heavy chains (MHC), fibre cross-sectional areas, and the geometric distribution of fibres within fasicles. No previous studies have described the ontogeny of MHC isozyme expression, or characterized monoclonal antibodies to MHC in sheep, so this was an integral part of the study. Development of the tibialis cranialis muscle has already been described up to E76 (Wilson et al., I992) so we felt that new information would most likely accrue from

examination of fascicles beginning at that date and continuing through the early postnatal period to adulthood. Data were collected using conventional light microscopy, and a computerized image analysis system.

Materials and m e t h o d s

Immunohistochemistry Immunohistochemical preparations from a series of sheep fetuses of embryonic age E76 (n = 3), E88 (1), E105 (3), El10 (1), El13 (1), E121 (2), E130 (1) and E141 (3), two postnatal specimens (PN1 and PN20), and from two adult ewes, 4 years of age, were examined. The day of mating (E0; gestation length in sheep 147 days) was determined by individual observation of ewes with synchronized oestrous cycles, mated to harnessed rams. Ewes were electrically stunned and killed by exsanguination. The uterus was removed in toto, the location of fetuses noted, and their weight recorded. Fetuses up to El10 were perfused through the heart while all others and the postnatal animals were perfused through the abdominal aorta or through the common iliac artery. Perfusion involved an initial flushing with normal heparinized saline followed by infusion of fixative, with both solutions at 37~ Fixative contained 10% formalin in 100 mM sodium phosphate buffer, pH 7.2, with 100 mM sucrose. Entire legs were skinned and removed from the animal and immersed in fresh fixative for 1 day at 4~ The tibialis cranialis muscle (May 1970) was then dissected and washed several times in PBS (phosphate buffered saline, NaC1 100 mM, sodium phosphate buffer 20 mM pH 7.4), taken through a series of alcohols and xylene and embedded in paraplast. Muscles were stored embedded in paraffin until required for analysis. Serial cross sections 10 #m thick were cut, mounted on subbed slides and dried on a slide warmer for 24 h. Sections were then dewaxed, washed in PBS and postfixed in 4% paraformaldehyde in PBS (pH 7.4) for 20 rain. Fixative was removed by washing three times in PBS (5 min, 10 min, 5 min), the second wash containing 0.1 M glycine. Sections were then treated with protease K (Merck), 2 x 10-6 mg m1-1 in Tris HC1 50 mM, EDTA 5 raM, pH 7.0. This was followed by dipping slides for 2 rain in 4 mM CaC12 in TBS (Tris buffered saline, NaC1 100 mM, Tris 20 raM, pH 7.4) to quench the enzyme, and washing them in two changes of PBS. Sections were incubated overnight with the primary anti-MHC antibodies at room

Table 1. Monoclonal antibodies used for immunohistochemistry

Antibody designation*:'~

M H C reactivity

Working dilutionw

Reference

MY32

Neonatal, IIa**, IIb

1:500

NOQ.7.1A:I=I: MF20

Slow-twitch All MHCsw167

1:100 1:100

B1

IIa, IIb

1:10

Sigma Chemical Co. Harris et al. (1989a) Harris et al. (1989a) Bader et al. (1982) Vivarelli et al. (1988) D. Fischman (unpublished data)

*Second antibody: biotinylated sheep anti-mouse IgG diluted 1:100 (Amersham). ~Third antibody: biotin-streptavidinHRP complex (Amersham), diluted I:100. DAB was used as chromogen. w antibody diluent 0.5% Triton X-100 and 0.5% BSA in PBS. ~Second and third antibody diluent 0.5% BSA in PBS. **Response to type IIa MHC reduced by fixation. C-tAbbreviated in text as IA. w167 to type lib MHC inactivated and response to type IIa MHC reduced by fixation.

553

Development of muscle fascicles temperature, and antibody binding revealed with the biotinstreptavidin horse-radish peroxidase (HRP) sandwich technique using diaminobenzidine (DAB) as chromogen (see Table 1 for details). Control sections were treated in the same way, except that incubation with primary antibodies was omitted. Small portions of the iateral head of tibialis cranialis were removed from one flesh unfixed adult muscle and frozen in isopentane cooled in liquid nitrogen. Tissue was frozen lightly stretched to approximate its in situ length. Serial 10/2m cross-sections were cut in a cryostat and mounted on subbed slides. Sections used to demonstrate myosin ATPase activity were air-dried for I h and processed as described by Guth and Samaha (1970). Alkaline preincubation was carried out at pH 9.6 for I5 min and acid preincubation at pH 4.55 for 8 rain. Incubation time with the substrate was 45 min (pH 9.4). All incubations were done at room temperature. Sections used for immunohistochemistry were kept at -80~ for 24 h and then air-dried for 2 h. These sections were either processed identically to the paraffin sections, commencing with the paraformaldehyde postfixation step, or with the posffixation and protease K steps omitted. In either case, series of four consecutive sections (monoclonal antibodies MY32, NOQ.7.1A, MF20, B1 (see Table I), were prepared for comparison with the appropriate ATPase slides. The NOQ.7.1A antibody will henceforth be abbreviated as 1A.

Characterization of antibodies Myosin extracts were obtained from tissue of sheep fetuses of known ages that had been snap frozen in liquid nitrogen and stored at -80~ until needed. Similar samples were obtained from other species as required. These samples were used to elucidate the expression of myosin isozymes in sheep and characterize the monoclonal antibodies used in this study. The techniques of ELISA, gel electrophoresis and Western blotting were used in addition to immunohistochemistry described previously. Unless otherwise stated all the procedures used are identical to those described in Harris and colleagues (1989a). Myosin extracts were obtained from E61 and E76 whole sheep hindlimbs, El00, EI40, PN28 and adult semitendinosus muscle, and from adult vastus intermedius and masseter muscles. The samples were separated using SDS and native gel electrophoresis, and silver stained. In some cases myosin samples from other species were also run in adjacent tracks on the same gel. Western blots were developed according to the procedures of Harris and colleagues (1989a) except that 0.25% BSA was substituted for 1% non-fat skim milk powder for the antibody incubation steps. MY-32 was used at 1:8000 dilution, 1A at 1:100, 2B6 (a gift from Dr Rubenstein; Gambke & Rubenstein (1984)) at 1:2000, B1 at 1:12.5 and MF20 at 1: 2000. ELISA assays were used to define the relative affinity of each monoclonal antibody to myosin extracts from different muscles, ages and species. All gels, Western blots and ELISAs presented used the same samples and all samples were adjusted and used at the equivalent protein loading. Immunohistochemistry was conducted using methods described previously. Specific investigations were conducted on the ability of the monoclonal antibodies to recognize various myosin isozymes in frozen sections after different periods of fixation in paraformaldehyde, and to compare antibody reactions with ATPase staining patterns. As the subsequent data will show, fixation is necessary to identify fast subtypes IIA and IIB.

Antibody dilutions used for Western blotting and immunohistochemistry were chosen to provide good signal (maximal binding to specific MHC) with low background and no cross-reactions with related epitopes.

Data acquisition About the same number of fascicles from both the larger medial head and the smaller lateral head of tibialis cranialis were sampled, enough in each case to assemble complete series for the entire range of ages. As no significant differences in reactivity or uniformity in fibre type distribution were noted between fascicles of the two heads, data from both muscle parts were pooled. Tissue for the immunohistochemical preparations was taken from the proximal half of the muscle at approximately the point where medial and lateral heads merge to form a single muscle belly. A block was also prepared from the distal half of one adult muscle. Examination of tissues from either source showed no variation in the immunohistochemical properties of the muscle throughout its length. There were fewer slow-twitch fibres in a narrow band near the periphery of the muscles; no fascicles were sampled from this region. The initial analysis of sections was done visually to gain an overview of fibre reactivity throughout the developmental sequence. Photographic records were made at each developmental stage, and the number of slow-twitch fibres in the lateral head of the muscle counted from E76 to EI21 by viewing and marking them through a camera lucida. Counting the number of fibres in the large medial head with this technique would have been a formidable task and was not expected to yield different results. In some instances, individual fibres were traced through blocks of up to 2 mm in length to obtain information on how far they extended through the developing muscle. The main part of the analysis used data collected with an image analysis system (SAMBA, Alcatel-TITN, Grenoble, France) run on a Compaq 386/20 computer fitted with a Matrox video card with a neighbourhood coprocessor. A program specifically relevant to the questions addressed in this study to run within the SAMBA environment was written by one of us (J.C.McE). Fascicles were selected that could be clearly followed through the four consecutive sections that were incubated with the MY32, 1A, MF20 and B1 antibodies, respectively. Using a camera lucida, every fibre in each section was drawn and numbered. Slides were then mounted on a Zeiss microscope fitted with a computer-driven motorized stage, and images captured with a Panasonic CCD video camera. Using the camera lucida drawings as a template, each fibre was reidentified on the computer monitor and labelled. Computer-drawn images of each fascicle were printed out and checked for accuracy against the camera lucida drawings. The image analysis system was used to obtain absolute values of antibody staining intensity from individual fibre cross-sections, and to provide data on fibre numbers, cross-sectional area and Cartesian coordinates for individual fibre location. Staining intensity was defined as 255 (1 - I/I0), giving a range of 256 grey scale values with higher values being darker. After the four sections of a fascicle had been processed all its component fibres had been analysed with respect to immunohistochemical profile, fibre size, and spatial distribution within the fascicle. The files of raw data were merged into a single file and checked for internal consistency, then transferred to a DEC VAX mainframe computer for statistical analysis using the SAS statistical package (SAS Institute Inc., Cary, N.C.).

554

MAIER, McEWAN, DODDS, FISCHMAN, FITZSIMONS and HARRIS

Results

Fig. 1. SDS gels and Western blots of muscle samples from developing and adult sheep muscles separated by 5% SDS-gel , electrophoresis. (a), E61; (b), E76 bulk sheep hindlimb muscle; (c), El00; (d), E140; (e), PN28; (f), adult semitendinosus muscle; (g), adult vastus intermedius muscle; (h), adult masseter muscle. Molecular weight calibrations (arrows) are shown in kDa. SDS, silver-stained gels, 200 ng protein per lane. Embryonic and neonatal MHCs could not be resolved with this separation system, so that gel electrophoresis resolved four bands of MHC, defined as IIa, IIb, embryonic/neonatal and slow MHC, from slowest to fastest migrating, respectively; these bands are shown in schematic form to the right of the silver stained gel records. Western blots, I #g per lane, were incubated with the monoclonal antibodies MF20, 2B6, 1A, MY32, or B1. Slow (type I) MHC was identified by its presence in both early and adult muscle, and by staining with antibody 1A. Embryonic MHC, recognized by 2B6, was not present after El00. When comparing degrees of staining it must be taken into account that the weak 2B6 staining at E61, relative to staining at E76, is due to the E61 sample having less myosin per unit protein (see Fig. 2). Neonatal MHC was already present at E61, but was no longer evident by PN28; it was identified by staining with antibody MY32. Isoforms IIa and IIb, also identified with MY32 give faint bands at El00 and are clearly obvious at E140. MF20 stained all MHC isoforms on Western blots and, additionally, stained some minor bands which we assume to be degradation products. Antibody B~ stained only the fast MHC isoforms.

CHARACTERIZATION OF MYOSIN HEAVY CHAIN MARKERS Semipurified sheep myosin extracts from different muscles and developmental ages were separated on 5% SDS gels and silver stained. Several protein loadings were examined; the results of loading 200 ng per lane are illustrated in Fig. 1 (top row). The MHC isozymes are present as strong bands with molecular weights of approximately 200-205 kDa. At E61 (lane a) two major bands are present. The faster migrating band consists of slow MHC, and the slower migrating broad band includes both embryonic and neonatal MHC. These latter two isozymes could not be unambiguously resolved using the electrophoretic procedure employed (Harris et al., 1989a). A similar pattern was present at E76 (lane b). By El00 (lane c), two faint bands were present migrating slightly slower than the previous slower band which now consisted predominantly of neonatal MHC (compare staining with 2B6 (embryonic) and MY32 (neonatal)). This was more apparent at higher loadings of protein (results not presented). By E140 (lane d), four bands were resolved. The relative quantities of the two faint slower migrating bands present at El00 had markedly increased. On the basis of comparison with other species the bands have been classed, in order of slowest to fastest migrating, as IIa, IIb, neonatal and slow MHC (B/ir & Pette, 1988). At this age, approximately equal amounts of each isoform are present in the semitendinosus. At PN28 (lane e) the neonatal isoform is absent and a pattern similar to the adult semitendinosus muscle (lane f) is present, with IIa, IIb and slow MHC being resolved. Adult vastus intermedius and masseter muscles (lanes g, h) contain only the slow MHC isoform. When adult sheep semitendinosus muscle was run on SDS gels with samples from rat, mouse and rabbit (results not presented), no differences in gel mobility for IIa and slow MHC were observed. The pattern of migration of IIb MHC, however, was slower than the equivalent neonatal MHC while in rats IIb MHC migrates faster than neonatal MHC (B/Jr & Pette, 1988; Harris et al., 1989b). It seems likely that the isoforms have diverged, although we cannot formally exclude the possibility that some other isoform (such as IId) is being mistaken for IIb. For the sake of convenience and for consistency with the previous literature on sheep muscle we assume it to be IIb.

Fig. 2. ELISA analysis of the specificity of the monoclonal antibodies MF20, MY32, 2B6, B1 and 1A, using antibody dilution curves. Sheep myosin samples were those run on the gels illustrated in Fig. I (E61 and E76, bulk hindlimb muscle; El00, E140, PN28 and adult semitendinosus muscle; adult vastus intermedius; adult masseter muscle). The curves for MF20, which is a pan anti-sarcomeric MHC in unfixed tissue, show equality of loading of total MHC in each sample except for the E61 sample, which was isolated from a tissue block which contained some skin and bone in addition to muscle. The curves for MY32 show high reactivity in all samples except vastus intermedius and masseter muscle, which contain only slow MHC. The curves for 2B6, specific for embryonic MHC, show reactivity only up to El00, demonstrating that embryonic myosin is not expressed in large quantities after this age. The curves for B1 show that it reacts only with samples which contain fast MHC, consistent with its being specific to adult fast MHCs. The curves for 1A show variable reactivity with samples from all ages but are highest in the muscles containing only slow MHC, consistent with other evidence that this antibody is specific for slow (type I) MHC.

Development of muscle fascicles

555 2

2

MF20

11

I

i

o

!

10-6

10-5

10-4

0

u

10-3

2

2

11

10 -4

10-8

10-2

10 "!

10 0

1 0 - 3 10-2

10-1

1

10-7

10-6

10-5

10"4

!

10 -6

!

1

10-5

2 ] 2B6

104

A n t i b o d y dilution

'~ -* -:" A A

0 10-7

10 -3

1

o

0

10 -5

10 -2

10-6

10-5

104

10-3

10"2

A n t i b o d y dilution Fig. 2.

E61 E76 El00 E140 PN28 Adult semitendinosus Adult vastus intermedius Adult masseter

556

MAIER, McEWAN, DODDS, FISCHMAN, FITZSIMONS and HARRIS

Western blots of equivalent 5% SDS gels (Fig. 1) were used to confirm the patterns of specificity of the monoclonal antibodies used in this study (Table 1), which in all cases were consistent with results from other mammalian species. Antibody MF20 stained all MHC isoforms as well as several minor components of lighter molecular weight, probably degradation products. Antibody 2B6 stained the slower migrating band at EOl, E76 and EI00, consistent with its specificity for embryonic MHC in rats (Harris et aI., 1989a). The staining was weak at E61 (due to the sample having less myosin per unit protein; Fig. 2), strongest at E76 and very weak by El00. Antibody MY32 stained the two slow migrating bands (IIa, lib) and the band identified by comparison with other species as neonatal/embryonic MHC. Note the different patterns of staining of this band with MY32 and 2B6. In sheep embryonic and neonatal MHC isoforms migrate together, but in rats MY32 recognizes only neonatal and not embryonic MHC (Harris et al., 1989b). Antibody 1A stained only slow MHC at all ages and muscles used, and antibody B1 stained only the IIa and IIb MHC bands present in the E140, PN28 and adult semitendinosus. The ELISA results (Fig. 2) give information additional to that obtained from the Western blots. Antibody MF20 (pan-sarcomeric MHC) dilution curves are similar for all ages, except that the E61 sample has a lower maximum. This is probably due to a lower quantity of myosin in the semi-purified sample obtained at this age as the sample initially contained skin and bone as well as muscle. All other ELISA curves for this age need this observation incorporated in their interpretation. Antibody 2B6 recognizes embryonic myosin, which is present at E61, highest at E76 and is diminishing by El00, with essentially none present at older ages. Antibody 1A recognizes slow MHC which is present at least to some extent in all samples, and is the sole MHC in the vastus intermedius and masseter muscles. Antibody MY32, which recognizes neonatal and adult fast MHC, bound most strongly to muscles containing large proportions of these isoforms. When the antibodies employed in this study were used on wax embedded sections the staining observed could not be reconciled with the Western blot or ELISA results. Further investigations (Figs 3 and 4) demonstrated that this was a result of paraformaldehyde fixation

preferentially masking epitopes of certain MHC isozymes recognized by MF20 and MY32. The presence of this masking was illustrated for MF20 by making serial frozen sections from adult rat EDL muscles and adult sheep tibialis cranialis muscles, and comparing their reactivity with the antibodies after various periods of paraformaldehyde fixation (Fig. 3). Here, MY32 staining of unfixed tissues is used to distinguish between fast-twitch (MY32-positive) and slow-twitch (MY32negative) fibres. The MF20 antibody stained all fibres in unfixed tissues from both species, and strongly stained slow fibres even after I h of fixation. In the sheep, some fast-twitch fibres (type IIa) retained their reactivity following fixation, while others (type IIb) lost their reactivity; the classification into IIa and lib fibres was based on the frequency, size and position of these fibre classes. In the rat, type IIb fibres were faintly stained, and type IIa fibres moderately stained with antibody MF20, following I h of fixation, making it possible to distinguish three fibre types in a single section. Fixation of Western blots of 5% SDS gels, subsequently stained with MF20, resulted in reduced binding to all MHC isozymes (results not presented) indicating that intact tertiary or quaternary structure may be needed for the selective epitope masking by fixation to occur. As a further comparison, serial frozen sections of adult sheep muscle were incubated with the antibodies, or reacted for myosin ATPase (Fig. 4a-h). Antibody IA detected fibres containing slow MHC at all ages in fixed and unfixed tissue and this staining was consistent with the ATPase staining and the other techniques used. Antibody B1 reacted with the adult fast twitch MHCs (IIa, IIb) and binding was unaffected by fixation. It was noted that, in contrast to results from rats and mice (Brooke & Kaiser, 1970), ATPase staining could not differentiate between IIa and lib MHC isozymes in sheep. This is consistent with the slightly different mobility of the sheep IIb myosin, suggesting that a divergence for this isozyme has occurred between species. Antibody MY32 detected both neonatal and adult fast MHCs in sheep. In fixed tissue, however, the epitope of the IIa MHC was partially masked (probably by crosslinking), resulting in lighter staining of these fibres (Figs 3 and 4). The neonatal MY32 MHC epitope did not appear to be affected by fixation, but this aspect was not examined in detail. The reactivity

Fig. 3. Serial frozen sections from adult rat EDL (a--d) and adult sheep tibialis cranialis (e-h) muscles to demonstrate the effect of 4% paraformaldehyde fixation on the epitope recognized by MF20. The panels depict unfixed sections stained with MY32 (1:1000) (a and e), unfixed sections stained with MF20 (b and f), sections stained with MF20 after 20 rain fixation (c and g) and sections stained with MF20 after I h of fixation (d and h). MF20 was diluted 1:10 for sheep and 1:100 for rat tissue. Fibres unstained by MY32 stained strongly with MF20 even after fixation (arrows; slow-twitch fibres), whereas some fibres which stained with MY32 progressively lost their ability to stain with MF20 when subjected to longer periods of fixation (arrowheads; lib fibres). The effect was similar at higher MF20 dilutions except that the effect of fixation appeared more rapidly. The fixation effect is more heterogeneous in rat EDL muscle, where MF20 staining allows discrimination of types I, IIa (asterisk-markedarrowhead) and IIb in single sections after fixation, than in sheep tibialis cranialis where types I and IIa were stained at approximately equal intensities, but type IIb lost its reactivity. Bar= 50 gm.

Development of muscle fascicles

557

Fig. 3.

Development of muscle fascicles

559

of MF20 in fixed tissue was consistent with that described above, i.e. it reacted strongly with slow and neonatal MHC, less strongly with IIa fibres, and negligibly with IIb fibres (Fig. 4i-l). In summary, the antibodies examined reacted with sheep MHC with specificities identical to those observed in rat muscles, except for a minor difference in IIa-MF20 reactivities. The novel observation was the selective masking of MHC epitopes observed after paraformaldehyde fixation. FASCICULAR ARCHITECTURE Fascicles from E88 to E130 animals could be partitioned into smaller units which, from their characteristic form, we term rosettes. In analysing our results, we use a working nomenclature ('C', 'P' and 'F') for the fibres making up a rosette, so as not to pre-empt our conclusions about their embryonic origins. A rosette consists of a large, round central (C) fibre surrounded by a ring of smaller peripheral (P) fibres, usually also round (Fig. 5). Adjacent rosettes, and especially those at the periphery and near angular portions of the fascicle perimeter, could not always completely fill the fascicular space. These spaces became occupied by solitary fibres or small groups of fibres which were not arranged in an obvious geometric pattern. Because they filled space not taken up by rosettes we term them fill-in (F) fibres. The central fibre always reacted positively with antibody 1A, but not with antibody B1, which, together with its large size, allowed it to be unambiguously identified (Figs 5-7). Owing to changing size relationships and fibre type transformation, rosettes could not be identified unequivocally after E130. IMMUNOHISTOCHEMICAL STAINING OF DEVELOPING MUSCLES

Antibody IA (anti-slow) On E76, about 25% of all fibres were positive, and all fibres positive for this antibody during the period E76El00 stained intensely. These fibres were usually apposed to each other in rows, separated by regions of unstained tissue (Fig. 6b). By E88 this pattern began to change, and fibres within rows became increasingly separate from one another so that each individual fibre was now surrounded by unstained tissue (Fig. 6f). At about El10, small fibres about one-half the diameter of the original 1A-positive population, expressing low levels of reactivity, were seen

Fig. 5. Antibody B1 staining of a paraffin-embedded crosssection from an EI21 tibialis cranialis muscle illustrating the grouping of fibres into rosettes. Central fibres (C) are unreactive. Some of the surrounding P (P) and F (F) fibres react strongly while others, which are transforming into slow-twit& fibres, give light to moderate staining reactions. Peripheral fibres are sometimes shared between neighbouring rosettes (arrowhead). The distinguishing feature of F fibres is that they are separated from C fibres by at least one intervening P fibre. Bar = I0 #m. in previously unstained regions of tissue (Fig. 7b). The gradual increase in reactivity observed over time indicated that these new 1A-positive fibres arose by a process of transformation, rather than expressing slow-twitch MHC from their time of formation. Transformation continued through PN20, the date when the last immature muscle was examined. By E130 the staining intensity of all smaller 1A-positive fibres had increased to the point that it was no longer an adequate criterion for separating the two populations (Fig. 7f), although the two groups could still be distinguished by size and by their location within rosettes. In the early postnatal period, fibre outlines became

Fig. 4. Characteristics of fibre types in adult tibialis cranialis muscles. The upper eight panels provide comparisons of myosin ATPase reactions (a-d) and antibody staining (e-h) in paired sections. Fast-twitch fibres give a dark ATPase reaction after alkaline preincubation (a and b) and a light reaction after acid preincubation (c and d). The reverse is true for slow-twitch fibres. The MY32 and 1A sections are from frozen unfixed tissue (fast-twitch subtypes after incubation with MY32 are recognized only in fixed tissues), while the MF20 and B1 sections are from frozen fixed tissue. Corresponding fibres between each pair (first and second rows) are marked with arrowheads. The lower 4 panels (i-l) are consecutive paraffin-embedded cross-sections incubated with antibodies MY32, MF20, 1A and B1. Corresponding fibres expressing slow(s), IIa or lib MHC are marked in the successive sections. Note the discrimination between types IIa and IIb with antibodies MY32 and MF20 (which also stains slow fibres), whereas antibody B1 stains both fast isoforms. Bars = 50 #m.

560

MAIER, McEWAN, DODDS, FISCHMAN, FITZSIMONS and HARRIS

Fig. 6. Comparisons between MY32, 1A, MF20 and B1 reactivities in fibres of consecutive paraffin-embedded tibialis cranialis cross-sections from animals aged E76 and E105. The 1A-positive C fibres are larger than P fibres at E76 (compare a and b), and more widely spaced at EI05 than at E76 (compare b and f). Presumptive IIb fibres, recognized by their strong reactivity with MY32, already line the periphery of fascicles at E105 (e). Antibody B1 also stained a subclass of fibres on the periphery of fascicles at E76 (d; arrowheads); these reacted strongly with both MY32 and MF20, and may represent presumptive IIb fibres. Bar = 10 ~m.

D e v e l o p m e n t of muscle fascicles

561

Fig. 7. Comparisons between M'Y32, 1A, MF20 and B1 reactivities in fibres of consecutive paraffin-embedded tibialis cranialis cross-sections from animals aged E121 and E14I. At E121, small P and F fibres are transforming towards a 1A-positive profile (arrowheads); note their smaller diameters compared with the original 1A-positive C fibres. The intensely staining small 1A-positive fibres present at E121 (b) are most probably fibres that began transforming at an earlier age than those marked with arrowheads. At EI41, P fibres have differentiated into type IIa or IIb (e), or have transformed into slow fibres which are stained by antibodies 1A or MF20, but not MY32 or B1. Bar = 10 I~m.

562

MAIER, McEWAN, DODDS, FISCHMAN, FITZSIMONS and HARRIS

increasingly polygonal, and it was difficult to identify the first and second generations of 1A-positive fibres by size; in effect, the new population had blended with the older one. In the adult, IA positive fibres made up 69% of the muscle (Fig. 4j).

E76

M~32

MF20

El10

C (P,2~

~"

= S L 0 W

]B1

Antibody MY32 (anti-neonatal/fast) Except for the 1A-positive (slow MHC) fibres of rosettes, which were essentially negative, at E76 fibres stained strongly with this antibody, reflecting the expression of neonatal MHC (Fig. 6a). The diameters of the MY32positive fibres varied in size, but on average they were much smaller than the 1A-positive fibres (Fig. 6a and b). By E105, two populations of MY32-positive fibres were apparent (Fig. 6e). Many fibres, generally of small diameter, were lightly stained. Intensely staining fibres, with diameters approaching those of the large 1A-positive fibres, were present around the periphery of rosettes. From their characteristic distribution we identify these as the precursors of adult lib fibres, which appeared relatively early in development and retained a fast-twitch profile through to adulthood. Disappearance of the neonatal isoform in the postnatal muscle caused the percentage of reacting fibres to decline, and in the adult only fast fibres (3I% of total fibres) were stained (Fig. 4i).

Antibody MF 20 (anti-embryonic/neonatal ~slow/IIa) All fibres reacted with this antibody to some degree during the part of the prenatal period examined here. At E76 (Fig. 6c), 1A-positive fibres were less strongly stained by MF20 than 1A-negative fibres but, with time, stronger staining shifted to presumptive slow-twitch fibres so that 1A-positive fibres and transforming fibres became the most strongly reacting cells during El10-E141 (Figs 6g and 7g). Postnatally, MF20 staining dramatically decreased in the subgroup of fast-twitch fibres (IIb) that reacted most strongly with MY32, and adult fibres of this type (I9% of total fibres) were essentially MF20-negative; type IIa (I2% of total fibres) stained lightly, and type I fibres more strongly (Fig. 4k).

Antibody BI (anti-fast) This antibody strongly stained intrafusal fibres, which served as internal positive controls (not shown). At E76, among extrafusal fibres, small 1A-negative fibres were lightly stained, while large fibres on the periphery of fascicles reacted more strongly (Fig. 6d). On the basis of their characteristic size and position, it is likely that

I~2

~

B!

Ila 9 r~2

,~,,,,-.0

-,0,

-

B1

MY32 1~'20 B1

~ llb

Fig. 8. Diagram illustrating the major stages of MHC transition in developing slow- or fast-twitch fibres in tibialis cranialis (the terms 'slow' and 'fast' refer only to adult muscle fibres). The degree of shading within the symbols represents the intensity of staining from no reaction (C)) to intense (Q). Letters and numbers indicate whether a given sequence occurs in central (C), peripheral (P) or fill-in (F) fibres, and which generation of myotubes (I ~ 2~ 3~ most probably contributes to each population. these fibres were the presumptive IIb fibres that later stained intensely with MY32. Beginning on E105 and extending to PN20 all fibres except the 1A-positive group stained with B1 (Figs 5, 6h, 7d and h), and subpopulations of presumptive fast-twitch fibres were no longer distinguished. During this time, previously fasttwitch fibres in the course of transformation into adult slow-twitch fibres progressively lost their reactivity to B1 (Fig. 5). The percentage of reacting fibres continued to decrease after that time due to fast-to-slow transformation, and in adult muscles only fast-twitch fibres gave positive reactions (Fig. 41). A summary of the major steps in the sequence of antibody staining during development is presented in Fig. 8. FIBRE NUMBER AND LENGTH The total number of muscle fibres in single cross-sections of the lateral head of tibialis cranialis at El10 (when

Fig. 9. Sample of three-dimensionalplots derived from cluster analysis of individual fibre-staining patterns at different embryonic ages, and in the adult. The symbols identify different clusters. Pyramids, 1A-positive fibres; cylinders, strong reaction with MY32 and MF20 (expression of neonatal MHC); diamonds, strong reaction with MY32, little or no reaction with MF20 (expression of IIb MHC); spheres, moderate reaction with both MY32 and MF20 (expression of IIa MHC). The staining intensity values have been normalized within batches and individuals, and are expressed in units of SD from the mean staining intensity (=0). For each age, the plots presented have only 100 randomly selected points from one animal, and even with this small number many plotted points represent several fibres, owing to overlap. Note that the IA axis is inverted.

Development of muscle fascicles E

563

76

E

76

MF20 3

o

-3

3

-6

E

3

88

E

88

E

121

MF20 3

I

0

-3 -2

3

-6

-4

--1

E

121

MF20 3

J

J

o

-2

3 -4

-1

ADULT

ADULT MF20

01

3

4

2

0

-3

-6

i 1

I

0

J

-2

~

MY32

-2

Fig. 9.

"

2

564

MAIER, McEWAN, DODDS, FISCHMAN, FITZSIMONS and HARRIS

fibre formation is largely completed) was about 30 000. Terminating or beginning fibres were found in serial sections cut for I mm in an El13 muscle (30% of total fibre population) and for 2 mm in an E143 muscle (5% of total fibre population). The lower incidence of short fibres in the older animal (near full term) indicates that fibres continue to grow in length throughout the prenatal period (Swatland & Cassens, 1972, I973). Large adult muscles with a tendon-to-tendon rather than a pennate form of muscle architecture are likely to be made up of serially arranged fibres with individual lengths ~ 1-2 cm, the fibres having finely tapered ends with no evident tendinous insertions (Richmond et al., 1985; English & Weeks, 1987; Loeb et al., 1987; Gans et al., 1989; Gaunt & Gans, 1990; Wilson et al., I992). This possibility does not affect the assessment of fibre numbers, fibre type composition and fascicular architecture at the single level chosen for this study. COMPUTERIZED IMAGE ANALYSIS At least five fascicles were examined at each developmental age. A total of 4500 fibres from 82 fascicles, divided into two series (batches), were analysed. Analysis of data collected with the SAMBA image analysis system gave results congruent with those gained by visual inspection.

Cluster analysis Because of variability in reaction product density between batches, data were treated on a within-animal and withinbatch basis. A fast-cluster procedure based on the K-means algorithm (MacQueen, 1967) was used for all final analyses. A minimum size limit was set on the size of clusters identified. Preliminary computations using other algo~thms gave similar results. Parameters for all analyses were the staining intensities for each fibre with each of the four monoclonal antibodies. The additional parameter of fibre cross-sectional area was significant only at early ages where it contributed to identification of the already well defined 1A-positive fibres. Consequently, it was deleted from subsequent analyses. At E76, two widely separated groups of compact shape could be distinguished (Fig. 9). One group, consisting of 23% of the total fibre number in the individual presented, stained strongly for 1A, while the other group stained strongly for MY32. Most fibres stained little or not at all for B1. It is concluded that at this time most fibres were expressing either neonatal or slow MHC, and were not expressing adult fast MHC. By E88, there was a greater range of variation in MY32 staining of the fast-twitch group. There was also a reduction in staining intensity with MF20 in presumptive fast-twitch fibres relative to presumptive slow-twitch fibres. The graphic effect (Fig. 9) was a diagonal oblong cluster with fibres having lower values of MF20 staining also showing reduced staining for MY32. Slow-twitch fibres accounted for 9.4% of the fibre population in the individual presented (Fig. 9).

By E121, the fast-twit& fibres could be separated into two broad groups based on their staining with MY32. The groups staining weakly or strongly for MY32 consisted of 54% and 25% of the total fibre numbers respectively, while the slow-twitch fibres had increased to 21% in the individual presented. In addition, there was a transitional region between 1A-positive fibres and the fast-twitch group, consisting of fibres staining lightly for MY32. This region presumably identifies fibres beginning to transform from fast-twitch to slow-twitch. Relative to slow-twitch fibres, there was a further decrease in MF20 staining in the fast-twitch group. This was largely achieved by reduced MF20 staining in fibres which reacted strongly with MY32 (Fig. 9). Three clusters of fibres could be identified in adult muscles. In the individual illustrated, the majority of fibres (69%) were slow-twitch. Two smaller clusters could be identified within the fast-twitch group: 12% stained lightly and 19% stained strongly for MY32. The lighter staining fibres reacted less with MF20 than did the 1A-positive fibres, and the fibres reacting strongly with MY32 were essentially unstained by MF20 (Fig. 9).

Proportions and growth rates of different fibre types To examine the relative numbers and fibre growth rates of different fibre types at various ages, fibres of all ages were partitioned into two broad categories: fast-twitch and slow-twitch fibres. This was done using cluster analysis on the variables of staining intensity to B1 and 1A monoclonal antibodies. The results from the analysis

100, go.

~

/

80. 70,

j j

"~ 50" ~ 4O.

20

"9

9

9 - -.

9149149

- - - Slow

. . .

o E75

10.

100

125

~r

II 20

. . . . . Adult

Age

Fig. Graph of the percentage of 'fast-twitch' and 'slowtwitch' fibres within a fascicleat different embryonic ages (birth occurs at EI47; asterisk). Fibres were classified on the basis of staining with B1 and 1A antibodies. The bars represent the least difference that would be significant when comparing fast- and slow-twitch percentages at each age. The initial divergence in percentage of the differentfibre types is the result of a difference in the rate of formation of new fibres, while the later increase in slow4witch and corresponding decrease in fast-twitch fibres is from transformation of existing fibres from fast to slow.

Development of muscle fascicles

565

were almost identical to those obtained using additional parameters to subdivide the two categories. The results for the proportions of the fibre types at different ages (Fig. 10) were derived from a residual maximum likelihood (REML) model (Patterson & Thompson, I971) where batch number was fitted as a fixed effect and animal as a random effect. The proportion of slow-twitch fibres initially decreased, from 24% at E76 to 10% at EI10, and then increased to 70% in the adult muscle, most of the increase occurring postnatally. The initial changes over the E76 to El10 period occurred at a time when total muscle fibre numbers in the tibialis cranialis were still increasing. Camera lucida counts of 1A-positive (slowtwitch) fibres in the lateral head of the tibialis cranialis increased from 1700 to 2600 over this period, suggesting that during this time the absolute numbers of fast-twitch fibres increased even more. In contrast, the changes in fibre-type profiles occurring later in fetal development, and postnatally, are a result of fibre transformation rather than generation of new fibres. Fibre cross-sectional areas increased throughout the period studied. The values plotted in Fig. 11 are again derived from an REML model including batch as fixed and animal and fascicle as random effects. The data were logarithmically transformed before analysis. The rate of increase in fibre cross-sectional area was nearly exponential until PN20; however, until E120 the fast-twitch fibres were significantly smaller than the slow-twitch fibres. At later ages, when fast-twitch fibre subgroups could be reliably distinguished, it was found that fibres in the cluster staining moderately for MY32 had slightly smaller cross-sectional areas than fibres in the other clusters. The fibres staining intensely for MY32 were smaller than

the 1A-positive fibres at E121, but the two groups were not significantly different by PN20. In the adult, values for the three groups were: 1A-positive, 998 ~tm2; low MY32 group, 806 ~tm2, high MY32 group 927 ~tm2, with an average SED of ___75 ~tm2.

Fibre position in relation to C fibres The proportions and characteristics of the C, P and F fibres were investigated over the E76-EI13 period. It was possible to use positive staining with 1A to define C fibres, as over this period transforming fibres were not present or accounted for only a small percentage of fibres (Fig. I0). P fibres were those that met the criterion (I-rl+r2)/dist))