Introduction

Neuregulins (NRGs) (also known as heregulin or neu differentiation factor) belong to a complex family of proteins, and were identified as glial growth factors. They are mainly present at neuromuscular junctions (NMJs) and have been shown to accumulate the AChR molecule at NMJs [8, 14, 17]. Four related NRG genes (NRG-1, NRG-2, NRG-3 and NRG-4) have been identified and, in particular, the function of NRG-1 has been studied in detail.

NRG has been shown to participate in the myelination of Schwann cells and oligodendrocytes, and in the specialization of astrocytes [1]. NRG promotes embryonic myogenesis [4, 10] and the glucose transport of skeletal muscles [2, 22]. Recently, Lebrasseur et al. [12] showed that the amount of NRG and ErbB proteins increased in several hindlimb muscles of rats after electrical stimulation and treadmill running.

In myogenesis in vitro, Ford et al. [5] demonstrated that NRG-1 potently induces mitogenesis in L6 myoblasts, but inhibits myogenic differentiation by inhibiting myogenin expression. In contrast, Florini et al. [4] and Kim et al. [10] suggested a differentiation-promoting role of neuregulin using L6 rat myotubes. However, all of these studies investigating the functional role of NRG-1 utilized an in vitro culture model. This model did not necessarily reflect the adaptive changes in satellite cells in in vivo muscle regeneration. In vitro, myogenesis of L6 or C2C12 cells ultimately ends up in the formation of myotubes but not myofibers, while in an in vivo muscle regeneration model, quiescent satellite cells migrate to the damaged site where they proliferate, differentiate and fuse with the damaged fibers or form new fibers [7].

Neurotrophic factors produced by regenerating muscle fibers are carried into the neuronal cell body by retrograde transport [13, 23]. Kami et al. [9] found, using in situ hybridization, that leukemia inhibitory factor (LIF) and glial cell-line derived neurotrophic factor (GDNF) were upregulated in intramuscular nerves and damaged muscles, and their receptors in motoneurons were also up-regulated following muscle contusion. Loeb et al. [16] showed that the amount of NRG mRNA and proteins increased after adding several neurotrophic factors, such as GDNF, and brain-derived neurotrophic factor (BDNF), to cultured motoneurons. Therefore, NRG-1 production may be influenced by the upregulation and/or downregulation of these factors in motoneurons or damaged muscle.

The purpose of this study was to investigate the changes in NRG-1 expression in regenerating processes after muscle damage using RT-PCR and immunoblotting. Furthermore, using immunofluorescence, we clarified the expression of NRG-1 in skeletal muscles at the stages of proliferation and differentiation following muscle damage, as well as in motoneurons.

Materials and methods

Experimental animals

Adult male Wistar rats (10 weeks of age) were used. The rats were housed in a temperature- (22 ± 2°C) and humidity-controlled (60 ± 5%) room regulated to provide an alternating 12-h period of light and darkness. Animals were allowed to feed (commercial rat chow) and drink ad libitum. This experimental procedure was approved by the Committee for Animal Research, Kyoto Prefectural University of Medicine.

Bupivacaine treatment

Muscle regeneration was induced in the left TA muscles of each rat (n = 40) by an intramuscular injection of 0.5 ml of 0.5% bupivacaine hydrochloride prepared in 0.9% saline solution. Rats were euthanized with an overdose of pentobarbital, and the TA muscles of the injected legs and spinal cords from the 4th lumbar vertebra (L4) to the 1st sacral vertebra (S1) were dissected at 0 (n = 10), 1 (n = 3), 2 (n = 4), 4 (n = 4), 6 (n = 11), 10 (n = 4) and 14 (n = 4) days after surgery. We used non-operated TA muscles and the spinal cords of animals sacrificed at 0 days after surgery as controls.

Primary antibodies

The antibodies employed in the present study were as follows: affinity-purified rabbit polyclonal antibody to neuregulin-1 (1:100, SC-348, Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) and myogenin (1:100, Santa Cruz), affinity-purified mouse monoclonal antibody to heregulin (1:100, Sigma), PCNA (1:100, Pharmingen, Becton Dickinson Company, USA), myogenin (1:100, NeoMarkers), bromodeoxyuridine (1:400, Pharmingen), dystrophin (1:100, Sigma) and c-Met (1:100, Santa Cruz) or affinity-purified goat polyclonal antibody to microtubule-associated protein-2 (MAP-2; 1:100, Santa Cruz).

Tissue preparation, gel electrophoresis and immunoblotting

Each tissue was homogenized in 10–20 vols of 50 mM Tris–HCl pH 7.4, 5 mM EDTA, 10 μg/ml phenylmethylsulfonylfluoride, 0.5 μg/ml leupeptin, 0.2 μg/ml aprotinin, 0.2% NP-40, 0.1% Triton X-100, 0.05% mercaptoethanol, and 1 mM Na3VO4 using a polytron (PCU-2, Kinematica, Steinhofhalde, Switzerland) for 30 s. The homogenized tissues were centrifuged for 25 min at 15,000g at 4°C, and the protein concentration of the supernatant was determined colorimetrically (Bio-Rad protein determination kit, Bio-Rad Laboratories, Richmond, CA, USA). Sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis were performed according to the methods of Laemmli [11]. Proteins separated by SDS-PAGE were transferred electrophoretically onto nitrocellulose membranes (Hybond-ECL Western, Amersham, Arlington Heights, IL, USA). The blots were incubated with a blocking buffer of 0.1% Tween-20 and 1% gelatin in 10 mM Tris-buffered saline (TBS, 10 mM Tris, 135 mM NaCl, 1 mM KCl, 0.02 % NaN3, pH 7.4) for 30 min. The blots were incubated with primary antibodies in a cold room overnight and with alkaline phosphatase conjugated anti-rabbit IgG (1:300, Promega Co. Madison, WI, USA) for 30 min. They were visualized with Western Blue™ Stabilized Substrate for Alkaline Phosphatase (Promega Co.). Densitometric analysis of each blot was performed with a computerized image processing system (NIH image 1.59).

RNA isolation and RT-PCR

Total RNA was isolated from the TA muscles and spinal cords using Sepasol-RNA II Super (Nacalai Tesque, Inc., Kyoto, Japan), according to the manufacturer’s protocols. First-strand cDNA synthesis was performed with 1 μg RNA according to a 1st Strand cDNA Synthesis Kit for RT-PCR (AMV; Takara Bio Inc., Otsu, Japan). Complementary DNA was then used as a template for PCR in a 20 μl-reaction volume including 1 μM of each primer, 20 μM MgSO4, 20 μM dNTP, 4 μM KOD plus-Buffer and 0.2 μl of KOD plus (KOD- Plus-PCR kit, Toyobo Co., Ltd.). The primers (SIGMA GENOSYS, Inc.) used for amplification were 5′-ACTCCTCACTCAGATTCAAA-3′ and 5′-AGGCTGTCTCAGTCGAGGCT-3′ corresponding to 532–551 and 783–764 base pairs, respectively, of rat HRG (sequence with total identity to all neuregulin isoforms), 5′-TGCCATCCAGTACATTGAGCG-3′ and 5′-CCACCGTCCATTCACATAAGACT-3′ corresponding to 406–426 and 846–869 base pairs, respectively, of rat myogenin and 5′-TCCTGCACACCAACTGCTTAGCC-3′ and 5′-CCCACTAAAGGGCATCCTGGGCTA-3′, corresponding to 1294–1317 and 1647–1670 base pairs, respectively, of rat glyceraldehyde phosphate dehydrogenase (GAPDH). PCR conditions for the analysis of the expression of each gene were designed to avoid PCR saturation and to enable semi-quantitative determination. The amplicons were resolved by electrophoresis on 1.5% agarose gels and visualized with ethidium bromide staining. Each sample was run in duplicate, normalized to each data point against GAPDH, averaged and statistically compared. The densitometric analysis was performed with NIH image 1.59.

Double-labeling immunofluorescence

As described previously [20], 5-bromo-2′-deoxyuridine (BrdU; Roche Diagnostics), a non-radioactive marker for DNA synthesis [4], was injected (100 mg/kg, intraperitoneally) 1 h before sampling for the labeling of proliferating satellite cells. Several 7 and 10 μm transverse sections of TA and spinal cord, respectively, made with a cryostat (CM-502 Coldtome, Sakura seiki Co., Ltd., Tokyo, Japan) were mounted on silanized slides (Dako Japan, Tokyo). TA muscles and spinal cords were isolated and frozen in isopentane. Cryosections were incubated in blocking solution (10% normal goat serum in PBS) for 10 min at room temperature. Sections were incubated with primary antibodies or α-bungarotoxin-tetramethylrhodamine (1:100, Sigma), overnight in the cold room. After washing in PBS, sections were incubated with anti-mouse or anti-goat FITC-conjugated (1:100 final dilution; Rockland Immunochemicals, Inc., USA) and anti-rabbit Rhodamine-conjugated (1:100 final dilution; Chemicon International Inc., USA) secondary antibodies. The sections were mounted using the slowfade antifade kit (Molecular Probes, Eugene, Oregon, USA). To increase the reliability of our data on NRG-1 expression in normal and regenerating muscle, we carried out several experiments using similar procedures. We investigated NRG-1 immunoreactivity in normal and regenerating muscle without primary antibody, with another NRG-1 antibody (anti-heregulin, Sigma), and after treatment with recombinant human NRG1-β1/HRG1-β1 EGF domain (R&D systems) using three different amounts (0.5, 1.0, 2.0 μg) of serum along with the primary antibody. Images were acquired on an Olympus IX70 inverted microscope with a fluorescent attachment (Olympus) and a Photonic Science CCD camera (Olympus DP50), controlled by the Viewfinder Lite (version 1.0) software (Pixera Corporation, Kawasaki, Japan).

Statistical analysis

All values were expressed as means ± SEM. Unpaired t tests were used to evaluate the significance of differences in each NRG-1 mRNA and protein levels between control and bupivacaine-treated rats. P < 0.05 was considered significant.

Results

The immunoreactivity of c-Met and dystrophin or NRG-1 were visualized using FITC- or Rhodamine-conjugated secondary antibodies, respectively. The expression of NRG-1 was detected in some of the quiescent satellite cells expressing c-Met (Fig. 1a, b). Some c-Met negative mononuclear cells located near the membrane of muscle fibers also possessed NRG-1 immunoreactivity in control TA muscles of rats (Fig. 1a, b). NRG-1 positive cells were detected in within (myonuclei) and outside (satellite cells) of the plasma membrane labeled with dystrophin (Fig. 1c, d).

Fig. 1
figure 1

In the control TA muscle of rats, the expression of NRG-1 was detected in some quiescent satellite cells expressing c-Met (a, b). Several c-Met negative mononuclear cells (myonucleus) existing near the membrane of muscle fibers also possessed NRG-1 immunoreactivity (b). These NRG-1 positive cells existed in the interior (myonuclei) and exterior (satellite cells) of the plasma membrane labeled with anti-dystrophin (c, d). C-Met (a) and dystrophin (c) immunoreactivity was visualized with FITC-, and NRG-1 (b, d) was visualized with Rhodamine-conjugated secondary antibodies. White circles and squares indicate the same fibers on different immunoimages. White arrows denote the satellite cells that expressed both c-Met and NRG-1 or NRG-1 at the exterior of the plasma membrane labeled with dystrophin, and yellow arrows denote the myonuclei expressing NRG-1 but not c-Met or NRG-1 at the interior of the plasma membrane. Bar = 50 μm

Full size image

Previous studies showed the existence of multiple bands of NRG-1 at molecular weights of 183, 115, 81, 64, 50, 37 and 26 kDa using extensor digitorum longus (EDL) and soleus muscles of rats [12]. We observed major bands of NRG-1 at 81, 50 and 26 kDa (Fig. 2 a) in the TA muscle of adult Wistar rats. When the muscle was damaged, Western blot analysis showed that all the NRG-1 isoforms in the regenerating muscles decreased on the 1st day after damage, which was most pronounced on the 4th day. It increased gradually after the 4th day up to 2 weeks following damage (Fig. 2a). On the other hand, NRG-1 and myogenin mRNA in the muscles similarly increased from 2 to 6 post-surgical days (Fig. 2b). Densitometric analysis at 6 days after bupivacaine treatment revealed a significant increase in NRG-1 mRNA levels compared with the control TA muscle (Fig. 2c).

Fig. 2
figure 2

Western blot (n = 1/each timepoint) shows that NRG-1 proteins in the regenerating muscles decrease on the 1st day and increase after the 4th day up to 2 weeks (a). Lower lane is an α-actin band. Equal amounts of protein were loaded in each lane. NRG-1 and myogenin mRNA in the muscle (n = 1/each timepoint) increases from 2 to 6 post-surgical days (b). Densitometric analysis showed a significant increase in NRG-1 mRNA in the regenerating muscles at 6 days after bupivacaine treatment compared to those of the controls (c). The integrated optical density (IOD) of NRG-1 normalized against the IOD of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA (arbitrary units). Values are means ± SEM (n = 5/group). Asterisks denote significant differences with the bupivacaine-treated rats

Full size image

In normal TA muscle of the rat, we observed α-bungarotoxin immunoreactivity at neuromuscular junctions (Fig. 3a). In the regenerating muscle at 1 day post-surgery, loss of selective accumulation of α-bungarotoxin occurred at neuromuscular junctions of all muscle fibers (Fig. 3b). In order to investigate whether the proliferating satellite cells have expressed NRG-1 protein in regenerating muscles, we performed immunofluorescence using anti-BrdU, which identifies these cells. BrdU-positive proliferating satellite cells did not possess NRG-1 protein in the regenerating muscle at 2 days postsurgery (Fig. 3c, d). At 4 days post-surgery, we did not observe clear co-localization between NRG-1 and myogenin in the regenerating muscle (Fig. 3e, f). NRG-1 expression was detected on myogenin-positive satellite cells at 6 (Fig. 3g, h) and 10 (Fig. 3i, j) days post-bupivacaine injection.

Fig. 3
figure 3

In normal TA muscle of the rat, we observed α-bungarotoxin immunoreactivity at neuromuscular junctions (a). In the regenerating muscle at 1 day post-surgery, loss of selective accumulation of α-bungarotoxin occurred at neuromuscular junctions of the muscle fiber (b). BrdU-positive satellite cells (c) did not possess NRG-1 protein in the regenerating muscle at 2 days post-surgery (d). At 4 days post-surgery, we did not observe clear co-localization between NRG-1 and myogenin in the regenerating muscle (e, f). NRG-1 expression was detected on the myogenin-positive satellite cells at 6 (g, h) and 10 days (i, j) post-bupivacaine injection. BrdU (c) and myogenin (e , g and i) immunoreactivity were visualized by FITC-, and α-bungarotoxin (a, b) and NRG-1 (d, f, h and j) was visualized by Rhodamine-conjugated secondary antibodies. White circles and squares indicate the same fibers on different immunoimages. White arrows denote the same satellite cells. Bar = 50 μm

Full size image

To confirm our data on NRG-1 expression in normal and regenerating muscle, we carried out several control experiments [19]. Another NRG-1 antibody (anti-heregulin, Sigma) made using a non-overlapping epitope of SC-348 (Santa Cruz) detected co-localization of NRG-1 and myogenin in the differentiating satellite cells (Fig. 4c, d). Moreover, NRG-1 immunoreactivity was lost after the treatment of recombinant human NRG-1 using different amounts (0.5, 1.0, 2.0 μg) of serum along with the primary antibody (Fig. 4e, f).

Fig. 4
figure 4

In the regenerating TA muscle of rats, there was no primary antibody that showed clearly NRG-1 immunoreactivity in myogenin-positive satellite cells 10 days after damage (a, b). Another NRG-1 antibody (mouse anti-heregulin, Sigma) detected the co-localization between NRG-1 (c) and myogenin (d) in the differentiating satellite cells at 10 days. In many satellite cells possessing myogenin at 10 days (e), NRG-1 immunoreactivity was lost after treatment with recombinant human NRG-1 of 1.0 μg (f), 0.5 or 2.0 μg (data not shown) of serum along with the primary antibody. Myogenin (a, e) and NRG-1 (c) immunoreactivity was visualized by FITC-, and NRG-1 (b, f) and myogenin (d) were visualized by Rhodamine-conjugated secondary antibodies. White circles and squares indicate the same fibers on different immunoimages. White arrows denote the same satellite cells expressing myogenin. Bar = 50 μm

Full size image

To clarify changes in NRG-1 expression in the spinal cord following muscle damage, we conducted immunoblotting and RT-PCR at 0, 1, 2, 4, 6, 10 and 14 days postsurgery. In the spinal cord, we observed several bands of NRG-1 at 115, 81, 50 and 26 kDa, with bands at 115 and 26 kDa being decreased and bands at 81 and 50 kDa being increased after the first and second day (Fig. 5a). Densitometric analysis at 6 days after injection showed a significant increase in the amount of NRG-1 protein at 50 kDa compared with the controls (Fig. 5b).

Fig. 5
figure 5

Western blot analysis showing the gradual increase in NRG-1 proteins in the spinal cord after the first day following damage (a). Ponceau staining (lower lane) showed an equal amount of protein loaded to each lane. Densitometric analysis 6 days after bupivacaine treatment shows a significant increase in the NRG-1 protein levels compared with the controls (b). Values are expressed as percentages of the immunostaining intensity of NRG-1 in the spinal cord in age-matched normal rats, which is taken as 100 %, and are given as means ± SEM (n = 8/group). Asterisks denote significant differences with the bupivacaine-treated rats. NRG-1 mRNA in the spinal cord was not influenced after muscle damage (c). In the normal rat, NRG-1 expression was not detected in many motoneurons identified by MAP-2 immunoreactivity (d, e). In contrast, NRG protein was abundantly expressed in cell bodies of many of the motoneurons innervating the TA muscle subjected to bupivacaine injection (f, g) NRG-1 (d, f) immunoreactivity was visualized by Rhodamine-, and MAP-2 (e, g) was visualized by FITC-conjugated secondary antibodies. White arrows denote the MAP-2 positive spinal motoneurons. Bar = 100 μm

Full size image

On the other hand, the amount of NRG-1 mRNA in the spinal cord did not significantly change after muscle damage (Fig. 5c). To determine whether NRG-1 protein is expressed in the spinal motoneurons after muscle damage, we used an antibody for microtubule associated protein-2 (MAP-2). MAP-2 is associated with dendrite stability in mature neurons and normally exists in nerve cell bodies and dendrites. In the normal rat, NRG-1 expression was not detected in many of the MAP-2 positive motoneurons (Fig. 5d, e). In contrast, NRG-1 protein was abundantly expressed in the cell bodies of many motoneurons innervating the TA muscle subjected to bupivacaine injection (Fig. 5f, g).

Discussion

Two main conclusions can be drawn from the present study. First, NRG-1 immunoreactivity was observed in differentiating satellite cells in regenerating muscle. Second, NRG-1 protein was expressed in several motoneurons in the spinal cord innervating the regenerating muscle.

We showed that NRG-1 protein was expressed in some quiescent satellite cells expressing c-Met. It is well known that satellite cells play important roles in the regeneration of muscle fibers; however, in the quiescent stage, satellite cells have no proliferating activity. It is possible that NRG-1 plays an important role in the differentiation of satellite cells in the regenerating TA muscle of rats. Indeed, we did not detect NRG-1 expression in the same nuclei of proliferating satellite cells possessing BrdU at the 2nd day post-injection. Ford et al. [5] demonstrated that NRG-1 promotes the proliferation of myoblasts and inhibits myoblast differentiation in early myogenesis in vitro. Additionally, our observation of the NRG-1 immunoreactivity in the many myogenin-positive differentiating satellite cells may coincide with previous studies that NRG stimulates the differentiation of L6 rat myotubes in culture [4, 10]. Since myogenin induces myoblasts to withdraw from the cell cycle and fuse to form multinucleated myotubes, our finding further supports the differentiation-enhancing role in NRG-1 against satellite cells in the regenerating muscle in vivo.

Our results did not differ from those in the report on the L6 myoblasts, which demonstrated that NRG-1 inhibits the formation of myotubes by downregulating the myogenin expression [5]. Although NRG-1 may inhibit early muscle differentiation (2–3 days after muscle damage in vivo) as suggested by Ford et al. [5], 6 days following damage seems to be a late period of muscle differentiation. Therefore, it is unlikely that NRG-1 co-localized with myogenin at this stage interferes with the potent muscle differentiation-promoting activity of myogenin. Although the molecular mechanism has not been understood in the regulation of myogenin expression by NRG-1 at fusing myotubes, this regulation was suggested using several neutralizing antibodies for NRG and its receptor, ErbB3 [10].

Using immunofluorescence, we clearly observed NRG-1 immunoreactivity in c-Met negative mononuclear cells near the fiber membrane or in the interior of the plasma membrane in normal TA muscle. Several previous studies have shown a localization of NRG-1 at the position of nucleus in various cells of humans or rats [6, 15, 18]. For example, immunofluorescence and Western blot of subcellular fractionated samples demonstrated an immunoreactivity of NRG-1α, NRG–1β, ErbB2 and ErbB3 in the nucleus of Schwan cells, as well as in the cytoplasm [18]. The exact mechanisms by which NRGs or ErbB receptors are localized to the nucleus are not known. However, the translocation of NRG-1 isoforms and/or ErbB receptors to the nucleus may be to initiate transcription of genes for a specific cellular response, such as glucose transport. This nuclear localization of NRG-1, a transmembrane protein, resembled the distributing pattern of another membrane protein, presenilin-1, in the denervated tibialis anterior muscles of rats [21].

Western blot analysis in our study showed increased levels of NRG-1 protein at 81 and 50 kDa in the spinal cord with time following muscle damage. In addition, we showed abundant expression of NRG-1 protein in the cell bodies of many motoneurons on the damaged side. It is possible that NRG protein is associated with the innervating damaged muscle fibers but not with the motoneurons themselves, because there were no significant changes in NRG-1 mRNA in the spinal cord after muscle damage. Indeed, loss of α-bungarotoxin accumulation at NMJ in the damaged fibers (Fig. 3b) indicates that some of the neuromuscular junctions were damaged after bupivacaine injection for the TA muscle of the rat. Various neurotrophic factors produced in the muscles are transported retrogradely to neural cell bodies of motoneurons after peripheral damage to skeletal muscle [13, 23], although there are differences in the muscle regeneration model. In contrast, muscle contusion increases the amount of LIF and GDNF in intramuscular nerves and damaged muscle, and the amount of these receptors in motoneurons, respectively [9]. Since the amount of NRG mRNA and protein increased in cultured motoneurons following the addition of various neurotrophic factors (GDNF, BDNF, NT-3 and NT-4) in vitro [16], the NRG-1 level in motoneurons or muscle may be upregulated by the GDNF expression in these positions after muscle damage. These lines of evidence raise the possibility that the increased NRG-1 protein in motoneurons is made within the regenerating muscles and would be transported retrogradely from muscles to motoneurons. NRG-1 in motoneurons of the spinal cord may promote the reinnervation of muscles by intramuscular sprouting of motor axonal terminals and/or proliferation and migration of Schwann cells.

In summary, the present study showed that the expression of NRG-1 increased during the regeneration of skeletal muscles and spinal cord. We suggest that NRG-1 promotes the differentiation of satellite cells, and that NRG-1 is transported retrogradely from muscles to motoneurons in vivo.