Muscle satellite (stem) cell activation during local tissue injury and repair

Abstract

In post-mitotic tissues, damaged cells are not replaced by new cells and hence effective local tissue repair mechanisms are required. In skeletal muscle, which is a syncytium, additional nuclei are obtained from muscle satellite (stem) cells that multiply and then fuse with the damaged fibres. Although insulin-like growth factor-I (IGF-I) had been previously implicated, it is now clear that muscle expresses at least two splice variants of the IGF-I gene: a mechanosensitive, autocrine, growth factor (MGF) and one that is similar to the liver type (IGF-IEa). To investigate this activation mechanism, local damage was induced by stretch combined with electrical stimulation or injection of bupivacaine in the rat anterior tibialis muscle and the time course of regeneration followed morphologically. Satellite cell activation was studied by the distribution and levels of expression of M-cadherin (M-cad) and related to the expression of the two forms of IGF-I. It was found that the following local damage MGF expression preceded that of M-cad whereas IGF-IEa peaked later than M-cad. The evidence suggests therefore that an initial pulse of MGF expression following damage is what activates the satellite cells and that this is followed by the later expression of IGF-IEa to maintain protein synthesis to complete the repair.

Introduction

Satellite cells in skeletal muscle were first described by Mauro (1961) and it is now realized that these cells provide the extra nuclei for post-natal growth (Moss & Leblond, 1970; Schultz, 1996) and that they are also involved in repair and regeneration following local injury of muscle fibres (Grounds, 1998). In normal adult undamaged tissue the satellite cells are quiescent and usually detected just beneath the basal lamina. They express M-cadherin (M-cad) (Bornemann & Schmalbruch, 1994; Irintchev et al. 1994) and co-express myogenic factors including c-met, MyoD and myf5 and later myogenin (Cornelison & Wold, 1997; Beauchamp et al. 2000; Qu-Petersen et al. 2002). The origins of satellite cells are still somewhat uncertain as they were thought to be residual myoblasts (reviewed by Seale & Rudnicki, 2000) but there is accumulating evidence that they may also originate from pluripotent stem cells derived from progenitor cells of the vasculature (De Angelis et al. 1999). Pluripotent stem cells from bone marrow cells (Ferrari et al. 1998), as well as epidermal cells (Pye & Watt, 2001), have also been shown to fuse and adopt the muscle phenotype when introduced into dystrophic muscle.

It has been established that even in normal muscle, local injury does occur from time to time (Wernig et al. 1990) but in certain diseases such as the muscular dystrophies, the muscle fibres are markedly more susceptible to damage, in particular to the membrane (Cohn & Campbell, 2000). The contractile system of muscle fibres also sustains damage during eccentric contractions, i.e. when the muscle is activated while being stretched. It is interesting to note that the forces generated by activation combined with stretch exceed even those of maximal isometric contraction. In the muscle fibres involved, the sarcomeres may be pulled out to such a degree that there is no longer any overlap of the actin and myosin filaments, thus causing damage (Lieber & Friden, 1999).

During embryonic differentiation, mononucleated myoblasts first proliferate and then fuse to form myotubes that become innervated and develop into muscle fibres. Following fusion, no further mitotic divisions occur within the myotubes or muscle fibres. The extra nuclei required for growth are provided by satellite cells fusing with muscle fibres principally at their termini (Aziz-Ullah & Goldspink, 1974), which is the region responsible for longitudinal growth (Griffin et al. 1971; Williams & Goldspink, 1971, 1976; Tabary et al. 1972). When muscle fibres sustain damage they have to obtain extra nuclei for the repair process reasonably quickly, to avoid cell death, which would result in a decrease in muscle mass and a permanent functional deficit. The extra set of genes required for protein synthesis during repair is derived from satellite cells typically located between the basal lamina and the sarcolemma. However, it has not been apparent which factors are involved in activating these cells to multiply and fuse with damaged or growing muscle fibres. As insulin-like growth factor-I (IGF-I) has been implicated (see review by Adams, 1998), it seemed relevant to measure expression levels of two insulin-like splice variants following imposed local damage. These were the systemic IGF-IEa and an autocrine splice variant produced by muscle (MGF). The latter was recently cloned from stretched/stimulated muscle (Yang et al. 1996) and, for this reason, and the fact that it has a different sequence to systemic IGF-I, it has been called mechano growth factor (MGF). IGF-I is reportedly involved in satellite cell activation (reviewed by Chakravarthy et al. 2001), although these in vitro studies may not accurately reflect what is happening in vivo, particularly in mature muscle when subjected to damage. Recent in vivo studies have indicated that MGF has different expression kinetics to IGF-IEa (Haddad & Adams, 2002) and this and other studies (Owino et al. 2001; Yang & Goldspink, 2002) suggest they have different modes of action.

Repair following skeletal muscle damage has been observed in experimental models and certain features are common. Fibre degeneration with subsequent influx of leucocytes into the damaged area predominates in the first few days. Regeneration begins once the phagocytic inflammatory cells clear necrotic tissue. This phase of muscle remodelling is characterized by activation of undifferentiated skeletal muscle precursor cells or satellite cells. Cell adhesion molecules, for example M-cad and N-CAM, have previously (Irintchev et al. 1994; Qu-Petersen et al. 2002) been shown to be expressed in activated satellite cells (myoblasts) and on myotubes during the regeneration process. As IGF-I and other growth factors have been implicated in satellite cell activity, it was important to ascertain what type of IGF-I may be involved.

For this purpose, local mechanical damage was induced by electrical stimulation of stretched muscles, mimicking a type of damage that occurs during eccentric muscle contraction. In another series of experiments, damage was induced by a myotoxic agent to determine if damage per se, as well as mechanical stress, would up-regulate those factors implicated in satellite cell activation. As satellite cells initially proliferate and then fuse with damaged fibres, a change in expression of certain adherins occurs. Their nuclei become reprogrammed to express muscle genes, and therefore M-cad and MyoD were chosen as marker genes to monitor this genetic reprogramming of the satellite cells in vivo.

As tissue repair is a local and continuing process throughout life, it is essential to use morphological methods to assess in which cell type the relevant genes are expressed and how this relates to the damaged area. In a complementary study (Hill & Goldspink, 2003) we have evaluated gene expression using real time RT-PCR, which permitted quantitative measurements in single rat muscles subjected to muscle damage. It was felt, however, that this present study should involve a more morphological examination of the muscle repair in vivo. The initiation of the IGF-I isoform expression could then be correlated with the appearance and distribution of satellite cell markers to establish which forms of IGF-I are likely to be involved in satellite cell activation.

Materials and methods

Induction of local muscle damage

Bupivacaine protocol for local muscle damage and repair

Sprague-Dawley rats, 250–300 g in body weight and 10–12 weeks of age, were divided equally into five experimental and two control groups (n = 6). The latter included untreated animals plus a sham control group injected with saline only. Young animals were studied because they have a greater potential for muscle regeneration than older subjects (Schultz & Lipton, 1982). Anaesthesia in the experimental and sham control animals was induced with approximately 3% halothane in oxygen at a flow rate of 2 L min⁻¹ and subsequently maintained at ∼1–2%. The left hind quarter was shaved to disclose the tibialis anterior (TA) and a 0.3-mL injection of either 0.5% bupivacaine hydrochloride (1-Butyl-N-[2,6-dimethylphenyl]-2-pipiridine carboxamide Hydrochloride; Sigma) in 0.9% sodium chloride, or 0.9% sodium chloride only, was made into the middle of the TA muscle. A 26-gauge × 13-mm-length needle was introduced at the midpoint of the muscle, inserted at an angle and advanced proximally along the muscle's longitudinal axis. The needle was then slowly withdrawn as the muscle expanded. In addition to sampling the contralateral muscle, controls included animals injected with physiological saline only, as well as normal rats of the same age and weight, which received no injection. Animals regained consciousness 10–15 min later. Regular checks were made to ensure rats were not in pain during the recovery period. The experimental animals were killed at intervals up to 25 days and the muscle used for morphological analysis as well as the determination of the corresponding mRNA levels.

Local muscle damage induced by the stretch combined with electrical stimulation

Sprague-Dawley rats, 250–300 g in body weight and 10–12 weeks of age, were divided into three experimental and two control groups. The TA of three groups of six rats was subjected to continuous stretch in the extended position with 1 h of stimulation of the peroneal nerve at a frequency of 30 Hz on days 1, 5 and 7 while the animals were anaesthetized. Six sham-operated controls, in which electrode wires were inserted near the peroneal nerve but the electrical stimulation circuit was not switched on, were also used. A further group of six rats subjected to no stretch or stimulation were used as normal controls. In all groups, muscle from the contralateral limb was also examined and served as a subgroup control. In the experimental and sham-operated animals, anaesthesia was induced as before and the left hindquarter of the hind limb was immobilized in plantar flexion using a fibreglass cast. Care was taken to ensure that blood flow to the foot was not compromised by omission of plaster cast around the ankle and by placing a cotton bud on the hind limb to ensure casting was not too tight. Electrical stimulation involved introducing stainless steel electrodes on either side of the peroneal nerve that were attached to a microstimulation circuit. When activated, the stimulators delivered bi-directional pulses of 1-ms duration, supramaximal intensity (3 V) at a frequency of 30 Hz (protocol taken from McKoy et al. 1999; altered and adjusted to the present model). At the end of stimulation, absorbable sutures (VICRYL 4/0) were used to close the incision and 0.3 mL of an analgesic (Tamgesic) was given subcutaneously. Animals regained consciousness 10–15 min later and the plaster cast still remained in position until the animals were killed. Regular checks were made to ensure that there was no swelling of the limb due to tight casting and that sutures were still in place. To analyse for morphological aspects of this type of local damage, muscles from these animals were killed at three time points up to 7 days and again used for the expression of certain morphological markers as well as their mRNA levels.

Tissue preparation

At the time points stated above for both studies, animals were killed using CO₂ and death was ensured by cervical dislocation. In the bupivacaine study, the TA muscle from both hind limbs was quickly removed, weighed under cold conditions and cut into two parts. One section of the TA was taken from the mid-belly region, covered in cryo-preservative (Tissue-Tek II, OCT Compound), snap-frozen by immersion in isopentane that had been cooled by liquid nitrogen and stored at −70 °C until further processing. Remaining TA muscle was processed for total RNA isolation. In the stretch and stimulation study, one part was taken from the distal tendon end and the other from the mid-belly region and processed as described above.

Assessment of muscle fibre damage and repair

Histological assessment

Cryostat sections were prepared and some of these were stained using the conventional H&E method and the area of damage determined using image analysis.

Expression of embryonic myosin

The primary monoclonal antibody to the embryonic myosin heavy chain (emb. MyHC) was used as a marker of muscle regeneration. The MyHc330 and secondary antibodies were a gift from A. F. M Moorman (Anatomy and Embryology Department, AMC, Amsterdam). The established protocol was performed on gelatin-embedded single fibres and detection was based on an indirect unconjugated immunoperoxidase technique (PAP) according to Moorman et al. (1984). However, the protocol was modified to use biotin–streptavidin detection: sections were fixed for 5 min in a 4% (w/v) paraformaldehyde in 100 mM potassium phosphate buffer (pH 7.4), and washed twice in 150 mM NaCl, 50 mM Tris/HCl, pH 7.6 (TBS). Endogenous peroxidase activity was quenched by immersing the slides in 0.3% hydrogen peroxide (H₂O₂) in methanol for 20 min, using a shaker. Sections were pre-incubated in a mixture of 5% horse serum, 0.5% Triton X-100 in TBS for 1 h at room temperature in a humidified chamber, before incubation with the emb. MyHC330 monoclonal antibody at room temperature overnight. This was followed by washing three times in TBS for 5 min and incubation with a horse antimouse secondary antibody (rat adsorbed) diluted 1 : 200 in 5% horse serum in TBS for 90 min at room temperature. Slides were washed three times for 5 min each in PBS, then incubated with ABC peroxidase reagent (Vector Laboratories kit) for 30 min at room temperature, washed three times for 5 min in PBS and the immunocomplex visualized by incubation with DAB substrate (Vector Laboratories kit) for 5 min. The colour reaction was stopped by washing sections in water and, following dehydration in ethanol washes of 50, 75, 90 and 100%, sections were mounted using DPX mounting medium (BDH).

M-cad used for identification of satellite cell marker proteins

M-cad rabbit polyclonal primary antibody was used according to a protocol based on that of Irintchev et al. (1994). Sections of 6 µm thickness were fixed in methanol for 4 min at 4 °C. Blocking solution of 20% normal goat serum (NGS) in PBS was applied for 30 min at room temperature in a humidified chamber. The solution was aspirated from sections before they were incubated at 4 °C overnight with M-cad primary antibody (produced by Professor Wernig's laboratory). This was diluted 1 : 50 in PBS containing 0.7% lambda carrageenan (Sigma) and 0.02% sodium azide. Following washing in PBS, sections were pre-incubated with 20% NGS diluted in PBS for 30 min to enhance specificity prior to incubation with a biotin-conjugated goat antirabbit secondary antibody (Jackson Immunoresearch Laboratories) diluted 1 : 200 in PBS-carrageenan solution for 1 h at room temperature. After washes in PBS, a fluorescein (DTAF)-conjugated streptavidin antibody (Jackson Immunoresearch) diluted 1 : 200 in PBS was applied for 30 min at room temperature. Sections were washed in PBS. To confirm the localization of the putative satellite cells, additional staining of nuclei with bis-benzimide and laminin with blue fluorescence was used to reveal the basal lamina. Following the last wash to remove M-cad secondary fluorescein antibody, sections were incubated with mouse antihuman laminin monoclonal antibody (Chemicon), diluted 1 : 1000 in PBS-carageenan solution, for 1 h at room temperature. After three 5-min washes in PBS, a secondary antimouse Texas Red-labelled antibody (Molecular Probes), diluted 1 : 200 in PBS, was incubated for 45 min at room temperature. Following washing in PBS, 1 µg mL⁻¹ of bis-benzimide (Hoechst 33258, Sigma) diluted in PBS was incubated for 5 min to stain nuclei. Washed sections were then mounted in Fluoromount (Agar Scientific).

Image analysis

Images were acquired on a Nikon TE300 inverted microscope with fluorescent attachment (Nikon) and Photonic Science low-light-level, peltier-cooled, CCD camera (Photonic Science), controlled by Kontron KS400 image analysis software (Zeiss Microscience). In order to analyse the total percentage damage/regeneration with H&E staining, whole muscle sections were scanned under bright-field conditions at 10× magnification using a motorized XY-stage (Prior) mounted on the Nikon inverted microscope and controlled by the KS400 image analysis software. Thereby, multiple microscope fields were collected (up to 10 × 10 fields) using the montage macro of the KS400 image analysis software to produce large montages of the whole muscle section. From these montages the areas lacking organized cellular structure could be delineated. For M-cad-specific staining, a minimum of five random fields per section containing a maximum of 100 muscle fibres were acquired from three sections per slide, over four slides at 20× magnification using fluorescence illumination with standardized imaging conditions for all specimens. Image analysis and H&E staining was performed using a custom-written KS400 macro program that allowed the user interactively to draw around damaged muscle fibres and express the identified area as a percentage of the total muscle area. It was apparent that there was some staining in the control sections as well as background fluorescence and the levels of detection were adjusted to remove this baseline staining. At higher magnifications than those used for image analysis, the fluorescence M-cad protein antibody complex was seen to be located mainly around the periphery of the muscle fibres and was detected even in the sections of the control muscles, but at a much lower level. Image analysis of the percentage increase in M-cad-positive staining was assessed by a semi-automatic segmentation macro, which allowed some limited interaction by the operator and expressed the results as AREA% of positive M-cad staining over the identified field.

RNA isolation and real-time RT-PCR

The single-step method of RNA isolation using acid guanidinium thiocyanate–phenol–chloroform after Chomczynski & Sacchi (1987) was used to isolate total RNA, from muscles that were used for morphological analysis. RNA concentrations were measured in a Genespec instrument (Shimatzu). First-strand cDNA synthesis for RT-PCR was performed using a Roche Diagnostics kit and cDNA samples were stored at −20 °C until required. In specific cDNA synthesis reactions, i.e. for MGF RT-PCR, 25 pmol of the sequence-specific primer MGF-rt was used in addition to the 25 pmol of random primers. Specific primers for IGF-IEa, MGF, MyoD and M-cad were used for the determination of these transcript levels as described in Owino et al. (2001) and Hill & Goldspink (2003). The RNA transcript levels for the different experimental and control muscles were analysed simultaneously and runs were carried out in triplicate. A negative control was present in each run where the template DNA was replaced with PCR-grade water. Briefly, reactions for the individual genes were optimized using SYBR Green I as the detection method and concentrations of the specific mRNAs given as picograms per microgram of total RNA as described in Owino et al. (2001) and Hill & Goldspink (2003).

Results

Muscle wet weight post stretch and stimulation

Muscle weight

After 1 day of induced damage by stretch and stimulation, the TA muscle weight of both experimental and contralateral muscles remained approximately the same, as in the sham and normal control groups. After 5 days the TA muscle wet weight of the experimental limb was 11.6% less (P < 0.001) compared with the right contralateral TA and the 1-day group. Thereafter, the muscle weight increased again. Greater weight loss was evident in the bupivacaine-treated muscles (−33% at 4 days) but by 24 days of recovery the muscle weight was significantly greater (10%) than for their contralateral controls (P < 0.01).

Time course and extent of morphological changes

Figure 1 shows examples of the sections that were stained for routine histological (H&E staining) and immunohistochemical (emb. MyHC) examination to assess local damage. None of the sham control muscles or contralateral muscles to the stretched and stimulated muscles showed any damage and were similar to the normal muscle group. Conversely, the bupivacaine-injected and the stretched/stimulated muscles showed extensive damage. Using the KS400 Image Analyser, it was found that in response to the bupivacaine insult (Fig. 2a) the percentage of damaged–regenerated area at day 4 was 67% and, thereafter, decreased gradually until day 24 when most of the muscle fibre architecture had returned to normal. Two-way anova revealed that there were significant differences (P < 0.05) among the five time points concerning the duration of recovery of muscle fibres towards normal muscle morphology except between days 14 and 24.

Fig. 1.

Transverse sections of rat TA muscle stained with haematoxylin and eosin demonstrating maximal damage at (a) 4 days after bupivacaine injection, (b) 5 days following stretch and stimulation and (c) recovery at 14 days following bupivacaine injection where central nuclei were present in the regenerated fibres. Bupivacaine injection caused massive muscle fibre degeneration as seen in (a) with a small number of fibres in the periphery of the section that survived the bupivacaine insult. (d–f) Transverse sections of the same TA muscles as above, stained with embryonic myosin heavy chain to determine the regenerating muscle fibres at each time point following both damage models. At both 4 and 5 days following bupivacaine injection and stretch and stimulation, embryonic myosin heavy chain stained heavily whereas at 14 days only a few regenerating muscle fibres were present (f). Scale bar = 50 µm.

Fig. 2.

(a) Mean percentage of damaged–regenerating muscle fibre area in relation to the whole muscle section in both damage models. There is a continuing decrease in the damaged–regenerating area after 4 days following bupivacaine injection and 5 days after stretch and stimulation where maximal damage was present in both. (b) The same pattern was seen with the embryonic myosin heavy chain staining the regenerating area in both models. Two-way analysis of variance (anova) was used to determine significant differences among the means. N = 4 for bupivacaine model and N = 6 for stretch/stimulation model for each time point. Experimental muscles were compared with those of the untreated animals at zero time and all differences were significant at P < 0.01 up to 15 days post-injury.

Muscle repair following local damage was also confirmed by emb. MyHC labelling. This was absent from all muscle fibres in the control groups and the contralateral muscle of the bupivacaine-injected animals and normal muscle fibres that survived the bupivacaine insult. In agreement with the data from the mean percentage of damaged–regenerating area shown with the H&E analysis, the mean percentage of embryonic myosin-positive muscle fibres declined after 4 days to reach zero at 24 days (Fig. 2b).

Interestingly, the degree of damage in the stretched/stimulated muscles was found to be higher in the distal parts of the muscle than those in the middle of the muscle. Both show maximum areas of damage at 5 days with the myotendon end exhibiting an area of damage of 50% and the middle region 30%. Two days later the damage area had reduced to 30% in sections from the myotendon region and almost to zero for the middle region. The difference between the regions was also reflected in embyronic MyHC expression, with 90% of the muscle fibres in the distal region and 70% of the muscle fibre showing expression, and at 7 days there was a reduction to 50% expressing emb. MyHC.

Bupivacaine-treated TA muscles exhibited a sequence of degenerative and regenerative changes. At 4 days after bupivacaine injection, most of the muscle fibres had degenerated, except from the periphery where muscle fibres were still intact. Macrophages filled areas of necrosis in which ghost-like remnants of the original fibres could occasionally be seen. The remaining fibres displayed features indicative of partial damage: a circular shape and ‘moth-eaten’ appearance, hyaline cytoplasm and pyknotic (but peripheral) myonuclei. However, dispersed colonies of regenerative fibres with central myonuclei could also be seen amongst the necrotic and normal fibres. At 7 days, a substantial fraction of the fibre population consisted of small regenerating fibres with peripheral nuclei migrating to the centre (defined as fibres that had an area less than 50% of the average for control muscle fibres) and many with central myonuclei as seen with the H&E stain. These fibres were larger than at day 4 and most of them reacted strongly with the emb. MyHC antibody. On day 11, regenerating fibres were larger and the number of these with central myonuclei and embryonic myosin-positive fibres was still large. After 14 days, the regenerating fibres were larger compared with those at 7 and 11 days. In addition, the reaction with emb. MyHC antibody had decreased compared with that at 11 days. Finally, on day 24, the fibre differentiation and morphology appeared normal and most of the myonuclei were now located at the periphery.

Time course of MGF and IGF-IEa expression

The mRNA levels of the two types of IGF-I at different time intervals are shown in Fig. 3(a,b). From Fig. 3(a) it can be seen that MGF expression had peaked by the first measurement at 1 day in the case of mechanical damage and 4 days following bupivacaine injection. By contrast, the expression of IGF-IEa (Fig. 3b) was much slower and took 12 days to peak following bupivacaine injection. In the case of mechanical damage, IGF-IEa was still rising at 7 days whereas MGF mRNA levels had already declined to their original (non-damaged) control levels by this time.

Fig. 3.

mRNA levels of MGF and IGF-IEa isoforms in the two models of muscle damage. MGF was maximally expressed as early as 1 day following stretch and stimulation and 4 days following bupivacaine injection (a), whereas IGF-IEa was maximally expressed later at 7 and 11 days following injury (b). Two-way analysis of variance (anova) was used to determine significant differences among the means. N = 6 for stretch/stimulation model and N = 4 for bupivacaine model for each time point. MGF measurements were significant at P < 0.01 for up to 5 days and IGF-IEa for up to 11 days.

Expression of M-cad in stretched and stimulated TA muscle

M-cad mRNA also peaked surprisingly early at 5 days following damage (Fig. 4a). This was also the time for maximal expression of M-cad protein (Fig. 4b). Therefore, both M-cad mRNA and protein declined rapidly, indicating that satellite cell activation occurs within a relatively short period. M-cad antibody labelled small muscle fibres in a ring-like shape (Fig. 5) in damaged–regenerated muscle sections in response to stretch and stimulation. The reaction product was confined to the plasma membrane and never observed in the cytoplasm. Double labelling with laminin, a major component of the basal lamina of muscle fibres, was performed to determine the location of cells expressing M-cad. M-cad-positive cells are deemed to be muscle satellite cells contained within the basal lamina of muscle fibres and between this and the plasma membrane. The M-cad immunoreactivity in the stretched and stimulated muscle was associated with the damage–regeneration phase. One day following stretch and stimulation, no significant changes in the muscle morphology were observed. However, after 5 days the muscle sections in which there was marked inflammatory response and regeneration showed strong M-cad staining beneath the basal lamina of small regenerating fibres. This was very noticeable at the tendon region 7 days after injury, where regenerating fibres close to the tendon demonstrated strong M-cad staining localized inside the laminin-positive fibres but the overall staining was less compared with that on day 5. In sham-operated control and normal muscle, no increase of M-cad expression was evident.

Fig. 4.

M-cad mRNA and protein expression in damaged TA muscle following stretch and stimulation and bupivacaine injection. M-cad mRNA levels (a) seemed to peak at 4–5 days after bupivacaine injection after stretch and stimulation and (b) M-cad protein levels peak shortly after but rapidly decrease once regeneration has commenced. All measurements of M-cad mRNA and protein were significantly different between the experimental muscles up to 11 days compared with normal resting values.

Fig. 5.

M-cad staining in a cross-section of TA muscle (a) 4 days after bupivacaine injection, (b) 5 days following stretch and stimulation and (c) 14 days following bupivacaine injection. In order to verify the identity of the M-cad-positive cells as satellite cells in (d), triple staining with laminin antibody (e) to identify the basal lamina and bis-benzimide (f) for nuclei staining was performed in order to check that M-cad was staining the satellite cells underneath the basal lamina. M-cad staining (arrows in d–f) was seen in the form of positive rings at the periphery of the muscle fibres but underneath the basal lamina which was much more obvious in the fibres of the damaged muscles (d). Scale bar = 50 µm.

Discussion

The aim of the work described was to investigate the role of two IGF-I splice variants under conditions of damage and further regeneration of skeletal muscle. The application of highly sensitive PCR technology enabled amplification of low-abundance transcripts for the quantative analysis of the locally produced insulin-like growth factors in muscle. During regeneration of skeletal muscle in young rats following ischaemia- or myotoxin-induced damage, elevated expression of IGF-I has been reported (Jennische & Hansson, 1987; Jennische et al. 1987; Edwall et al. 1989), which was diminished by day 15 of recovery (Marsh et al. 1997). However, the present study is the first to look at the distinct IGF-I isoforms, IGF-IEa and MGF, under such conditions and relate this to the activation of muscle satellite (stem) cells in vivo.

Results of the experiment in which damage was induced by bupivacaine demonstrated a surge of IGF-IEa mRNA expression that was maximal at 11 days and diminished thereafter to similar levels as those in the non-injected animals. By contrast, MGF mRNA showed a much earlier transient response that peaked at 4 days post bupivacaine injection and decreased thereafter; following mechanical damage, MGF peaked even earlier. It seems that in both myotoxin- and mechanical activity-induced damage models the temporal expression pattern for each IGF-I splice variant showed similar differential gene splicing sequences, with MGF peaking before IGF-IEa. This temporal difference in expression of the two muscle IGF-I RNA transcripts has also been described in the rat following commencement of resistance exercise (Adams, 2002). As M-cad expression peaked well before IGF-IEa, whether it was measured as mRNA or protein, it is unlikely that the systemic type of IGF-IEa is responsible for initial activation of satellite cells. However, it is not possible to determine from these data whether this was due to an increase in number of satellite cells because it is known that quiescent satellite cells do stain to some extent for M-cad protein (Rosenblatt et al. 1999). Nevertheless, this does represent a marked increase of M-cad, whether in existing satellite cells or an increase in the number of these cells or both.

MGF and IGF-IEa splice variants apparently yield the same mature peptide, which is derived from the highly conserved exons 3 and 4 of the IGF-I gene. These exons present in all the known IGF splice variants are known to code for the IGF-I receptor ligand domain. A mechanism of extracellular endoproteolysis of the IGF-I pro-hormone results in the same mature peptide (Gilmour, 1994), even though the splice variants of IGF-I may have different 3′ sequences including the E domain. It has been suggested that IGF-I precursors could be pluripotent, in a form analogous to that of pro-hormone propiomelanocortin and proglucagon (Siegfried et al. 1992). The observation that a synthetic peptide derived from the rat Eb domain induces proliferation in epithelial cells is noteworthy (Siegfried et al. 1992). The role of the growth-promoting properties of the E peptide in MGF, acting as an independent growth factor, is supported by the recent cell culture experiments of Yang & Goldspink (2002), in which stable transfection with MGF was shown to stimulate myoblast proliferation but differentiation was suppressed. The addition of a synthetic MGF peptide or the medium from MGF-transfected cells onto normal C2C12 cells also inhibited their differentiation. Yet this inhibition was reversed when the peptide or the medium were withdrawn. By contrast, cells of the liver type of systemic IGF-I (IGF-IEa)-positive clone did form myotubes and the normal cell lines showed less cellular proliferation as well as forming myotubes. Of particular interest was the observation that when an IGF-I receptor antibody was added to the muscle cell cultures, cell proliferation induced by MGF was not inhibited whereas their stimulation to increase in mass and to form myotubes by IGF-I was repressed. This result strongly suggests that MGF is involved in another signalling pathway in addition to that associated with the IGF-I receptor.

As satellite cells appear to play an important role in muscle repair and adaptation it was important to investigate the expression of a satellite cell marker under conditions of damage and regeneration and to relate this to the temporal expression of the mechanosensitive MGF and or the systemic type IGF-IEa. One of the most useful and suitable markers for the identification of satellite cells for this work proved to be the cell surface protein M-cad, because it has been shown to play a significant role in alignment and fusion of myoblasts to form and expand developing myotubes (Cifuentes-Diaz et al. 1995) and has been detected in satellite cells of normal muscle and during regenerative responses after muscle damage (Moore & Walsh, 1993; Irintchev et al. 1994). The early and acute surge of MGF mRNA following mechanical and myotoxic damage in this study strongly suggests that it is this splice variant of IGF-I that is involved in initiating the proliferation and differentiation of satellite cells. M-cad expression had already peaked when damage was evident, i.e. at 4 days post bupivacaine injection (Fig. 4a,b) and 5 days post stretch and stimulation (Figs 2a and 4b), and started to decrease once regeneration, fusion of myoblasts, had begun. The mRNA results were confirmed by the presence of M-cad protein in the activated satellite cells of the damaged muscles. This was very evident at 5 days by antibody staining (Fig. 5), and its RNA levels were seen to peak by 4 days following bupivacaine injection in the stretch and stimulation model (Fig. 4a). As MGF expression precedes M-cad mRNA and protein expression, this strongly suggests that this splice variant rather than IGF-IEa is involved in satellite cell activation. The latter (IGF-IEa) is expressed and peaks at 10 days following the insult. Although IGF-IEa is probably not involved in the initial activation of satellite cells, it is important that the repair process continues after the initial events and IGF-IEa is expressed at higher levels than MGF and is therefore a greater source of the mature peptide (IGF-I ligand domain). IGF-IEa expression may therefore be regarded as the second phase of local tissue repair as it is necessary to maintain protein synthesis rates in order to restore muscle mass. A specific monoclonal antibody is being generated to the MGF peptide. This will, we hope, permit its expression to be studied and compared with that of the mature IGF-I peptide, which is encoded by the RNA of both splice variants expressed in muscle following damage.

The results of these studies provide additional insight into the complexity of the IGF-I system and its implications under conditions of damage and subsequent regeneration. IGF-IEa and MGF are produced by active muscle in rodents and have been shown to be positive regulators of muscle hypertrophy (Goldspink, 1999; McKoy et al. 1999; Owino et al. 2001). However, as reported here, the MGF isoform is acutely induced, whereas IGF-IEa has a delayed effect that is sustained during the later phase of regeneration. When comparing mechanical damage with myotoxin damage it is apparent that both involve a relatively rapid expression of the MGF splice variant, although it may seem that this growth/repair factor has been misnamed ‘mechanogrowth factor’. However, even in the case of myotoxin damage it is likely that the damaged tissue mass is subjected to increased mechanical strain that results in the same cellular response. As the expression of the autocrine splice variant (MGF) precedes satellite cell activation, it is likely that this form of IGF-I is associated with satellite cell activation, not the systemic IGF-IEa type. This is in accord with the finding that MGF is not expressed in dystrophic muscles (Goldspink et al. 1996) and the decrease in MGF mRNA levels in response to mechanical overload in older muscles (Owino et al. 2001). There is a deficiency of active satellite cells in both these situations, in which local tissue repair becomes increasingly impaired. Experiments to investigate the expression of the two transcripts and activation of satellite cells in young and old muscles after the therapeutic application of MGF and IGF-IEa to ameliorate muscle loss are in progress.