FGF2 rescues denervation-induced muscle atrophy in Ambystoma mexicanum

Abstract

Background

Denervation occurs as a consequence of disease or injury and is typically accompanied by skeletal muscle atrophy. Although the relationship between nerves and muscle atrophy has been studied, the direct molecular contributions of nerves remain unclear.

Methods

We used the axolotl (Ambystoma mexicanum), a vertebrate with robust regenerative capacity and optically accessible musculature, to investigate the role of nerve-derived signals in muscle maintenance.

Results

Forelimb denervation produced significant muscle atrophy in axolotls. Quantitative imaging showed that reduced muscle fiber size—rather than fiber loss—underlies this atrophy. To investigate the molecular basis of denervation-induced atrophy, we focused on fibroblast growth factor 2 (FGF2), a nerve-derived factor previously implicated in axolotl limb regeneration. Electroporation of FGF2 into denervated muscles significantly preserved muscle fiber size compared with controls. Conversely, pharmacological inhibition of FGF signaling with SU5402 reduced fiber size, supporting a requirement for FGF signaling in maintaining muscle mass.

Conclusions

These findings demonstrate that FGF2 is sufficient to mitigate denervation-induced muscle atrophy and support a requirement for FGF signaling in the maintenance of muscle mass. Together with the neural expression of Fgf2, our data support a model in which nerve-derived FGF2 contributes to muscle maintenance. This work positions the axolotl as a tractable model for dissecting neuromuscular signaling and identifies FGF2 as a promising therapeutic candidate for neuromuscular atrophy. Understanding this mechanism may inform strategies to preserve muscle mass after nerve injury or in neurodegenerative diseases.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13395-026-00413-w.

Background

In humans, skeletal muscle comprises approximately 40% of the total body weight and plays an essential role in both movement and energy metabolism [1]. It is innervated via the neuromuscular junction (NMJ), which is crucial for maintaining normal muscle structure and function [2]. However, trauma or neurological injury can lead to motor neuron loss, resulting in denervation. Following denervation, the absence of neural input prevents the transmission of excitation signals to the muscle, thereby impairing contraction and ultimately causing a loss of motor function. This sequence of events leads to a reduction in muscle mass and strength, resulting in muscle atrophy. Given that muscle function is fundamental to daily activities, muscle atrophy significantly impairs quality of life (QOL).

Despite its importance, the precise mechanisms by which nerves contribute to the maintenance of muscle remain incompletely understood. In particular, it remains debated whether nerves maintain muscle mass through direct molecular mechanisms or indirectly via motor activity. Previous studies in mammals have proposed several interventions to prevent denervation-induced muscle atrophy. Among them, electrical stimulation (ES) has been widely accepted as a method to preserve muscle mass and strength before reinnervation [3]. ES is known to inhibit the activation of the ubiquitin–proteasome pathway, thereby preventing proteolysis and delaying muscle atrophy. Similarly, mechanical stimulation through rehabilitation has also been shown to prevent muscle loss [4, 5]. It has been reported that mechanical stimulation can activate protein synthesis in muscle fibers through signal transduction via the extracellular matrix (ECM). These observations demonstrate that muscle mass can be maintained by direct stimulation of the muscle in place of neural input, thereby supporting the notion that nerves preserve muscle mass through their motor function. On the other hand, clinical evidence indicates that stimulating the sympathetic nervous system—without affecting motor function—can still enhance skeletal muscle strength, as shown in recent murine studies [6]. Within 1–2 weeks after sympathectomy, NMJ size decreases, accompanied by muscle atrophy and functional decline, suggesting a critical role for the sympathetic nervous system in maintaining skeletal muscle integrity [7, 8]. These findings imply that nerves may directly contribute to the preservation of muscle size and mass, beyond their role in motor function. Elucidating the mechanisms by which nerves regulate muscle mass holds promise for understanding and addressing a variety of neuromuscular diseases and disorders.

The axolotl represents an ideal model for such investigations. Despite its structural similarity to mammals, the axolotl offers exceptional tissue transparency, enabling direct visualization of neural trajectories in vivo [9, 10]. Moreover, axolotl cells are considerably larger than those in mammals, facilitating both microscopic observation and genetic manipulation. These unique advantages, along with its comparable capacity for muscle regeneration and maintenance, position the axolotl as an optimal in vivo model for skeletal muscle research relevant to human biology.

To date, however, no studies have investigated denervation-induced muscle atrophy in the axolotl. Most muscle-related research in this species has focused on its extraordinary regenerative capacity. The axolotl is well known for its ability to regenerate various organs, including limbs. Over 200 years ago, it was discovered that nerves play a crucial role in appendage regeneration [11, 12]. Since then, numerous nerve-derived factors have been identified as key contributors to this process. Notably, the activation of BMP and FGF signaling pathways by nerves has been shown to drive appendage regeneration, with specific molecules such as FGF2, FGF8, and BMP2 or BMP7 playing prominent roles [13, 14]. While the contribution of nerves to appendage regeneration has been extensively studied, the functional interplay between nerves and the individual tissues comprising the limb—particularly skeletal muscle—remains poorly understood.

In the present study, we investigated the effects of denervation on limb muscle in axolotls, with a particular focus on the role of the nerve-derived factor FGF2. We aimed to determine whether FGF2 can prevent muscle mass loss following denervation. This work provides evidence for the direct involvement of nerves in muscle mass regulation, highlights FGF2 as a candidate nerve-derived regulatory molecule, and further establishes the axolotl as a powerful model system for dissecting neuromuscular interactions.

Methods

Animal harvesting

Axolotls (A. mexicanum) were obtained from the Amphibian research center of Hiroshima University. Animals were kept in aerated water at a temperature of 22 ℃. For the surgical experiments conducted in this study, axolotls with nose-to-tail lengths ranging from 9 to 11 cm were used, with no specific bias towards any sex. The care and treatment of the animals in this study were executed under protocols approved by the Animal Care and Use Committee of Okayama University. All animal experiments adhered strictly to the guidelines provided by the Animal Care and Use Committee of Okayama University. Every possible measure was taken to minimize animal suffering, in line with the NIH Guide for the Care and Use of Laboratory Animals.

Surgical procedure of denervation and isolation of muscle

Axolotls were anesthetized using 0.1% MS-222 (Sigma-Aldrich, St. Louis, Missouri) at pH 8.0, for roughly 20 min, depending on the size of the animal. For denervation, we cut the skin of the proximal forelimb (stylopod), exposed the blood vessels and nerves, and severed two nerves—the Nervus medianus and the Nervus ulnaris—near the brachial artery (Arteria brachialis). As a control, the skin of the contralateral forelimb was also incised, and the nerves were gently pinched without cutting.

For muscle isolation, axolotls were anesthetized as described above, and the forelimb was amputated and all skin was peeled off. Using fine forceps, the muscles were carefully dissected from the limb from tendon to tendon, taking care not to damage the muscle tissue.

Whole mount immunofluorescence staining

The axolotls were anaesthetized one week after denervation surgery, and their forelimbs were amputated at the base of the humerus and fixed in 4% paraformaldehyde (PFA)/phosphate-buffered saline (PBS) at room temperature for one day. Zeugopod muscles were then isolated and washed twice with Tris Buffered Saline with Tween 20 (TBST). Dehydration was performed using 50% ethanol (EtOH)/PBS, 75% EtOH/PBS, 100% EtOH, and 100% methanol (MetOH), followed by storage at − 26 °C for one day. Rehydration was carried out using 100% EtOH, 70% EtOH, and PBS, followed by two washes with TBST. The tissue was then transferred to blocking buffer and incubated at room temperature for 30 min to prevent non-specific antibody binding. Primary antibody reaction was performed at 4 °C for 16 h. In this study, Anti-Neurofilament (DSHB, 3A10-s, 1:20) was used as the primary antibody. The following secondary antibodies were used: anti-mouse IgG Alexa 488 conjugated antibody (Invitrogen, A32723, 1:1000). Muscle tissues were treated with Tissue-Cleaning Reagent CUBIC-R+(N) (TCI, T3983) after immunofluorescence staining. The stained tissues were mounted using Mounting Solution (RI 1.520) (TCI, M3294).

Sectioning and immunofluorescence staining

The samples were fixed in 4% paraformaldehyde (PFA)/PBS at room temperature for one day. For immunofluorescence staining, the samples were decalcified in a 10% Ethylenediaminetetraacetic Acid (EDTA) solution for one day after fixation. The samples were immersed in a 30% sucrose/PBS for 12 h and underwent cryoprotection. The samples were then embedded in O.C.T. compound (Sakura, Finetek) and frozen. Sectioning was performed using a Leica (Nussloch, Germany) CM1850 cryostat. Section thickness was 4–14 μm. The sections were dried at room temperature for 3 h.

Dried sections were washed three times with TBST. Subsequently, blocking buffer was applied at room temperature for 30 min to prevent non-specific binding of antibodies. Primary antibody incubation was then performed at room temperature for 2 h. In this study, the following antibodies were used as primary antibodies: Anti-Type III collagen antibody (Proteintech, 22734-1-AP, 1:500), Anti-myosin heavy chain antibody (DSHB, MF20, 1:250), Anti-Neurofilament antibody (DSHB, 3A10-s, 1:20), Anti-Phospho-FGF Receptor (Cell Signaling Technology, 3471, 1:100), and Anti-Phospho-p44/42 MAPK(T202/Y204) (Cell Signaling Technology, 9101 S, 1:400). For staining Type III collagen, after TBST washing, the samples were heat-treated at 130 °C for 30 min, then transferred to Proteinase K/TBST and incubated at room temperature for 20 min before transferring to the blocking buffer. The secondary antibody reactions were performed for 2 h at room temperature (Goat anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ Plus 488 antibody (Invitrogen, A-32723, 1:1000) and Donkey anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 594 (Invitrogen, A-21207, 1:1000) or Donkey anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 594 (Invitrogen, A-21203, 1:1000) and Donkey anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488 (Invitrogen, A-21206, 1:1000)). Nuclei were stained with Hoechst 33,342 (Nacalai Tesque, 19172-51) and mounted with Fluoromount (DBS, K 024).

Electroporation

We used an FGF2 plasmid vector (cmv: FGF2-p2A-GFP) containing a ubiquitous promoter-Fgf2-p2A-gfp-SV40 polyA signal based on the pCS2 plasmid, as well as a GFP plasmid vector (cmv: GFP). The plasmids (5 µg/µL) were mixed with Fast Green dye for visualization and injected into the muscle fibers of the ventral side of the forelimb using a glass capillary. Immediately after injection, electric pulses (20 V, 50 ms pulse length, 950 ms interval, 10 times) were administered.

SU5402 and FGF2 treatment in cultured muscle

The dissected muscles were cultured in medium (50% DDW, 40% DMEM (glutamic acid), 10% FCS, gentamicin, and 0.01 M HEPES). A working stock solution of SU5402 (30 mM, #193-16733, Wako) was prepared by dissolving the reagent in DMSO (#08904-14, Nakalai Tesque) and the stock was diluted to 30 µM in the culture medium. The control group was treated with a solution containing the same amount of DMSO. The muscles were cultured for 3 days.

For FGF2 treatment, the isolated muscles were precultured in a serum-free medium (50% DDW, 50% Opti-MEM, and 0.4 mg/ml gentamicin) for 24 h. FGF2 was then added to the medium at a final concentration of 1 µg/mL (R&D Systems, Inc. 3139-FB). The muscles were further incubated for 12 h and processed for immunofluorescence analysis.

Creation of color-coded images based on muscle fiber cross-sectional area and normalization via polar coordinate transformation

The perimysium visualized by COL3A1 immunofluorescence was traced using ImageJ, and the resulting image was then binarized. Fiber coloring was performed using Python Code 1 (Supplemental Data 1). In Code 1, the filename of the image to be processed was specified in line 7. For each individual, the mean of the 20 largest cross-sectional areas of control muscle fibers was entered in line 28, and a color-coded image was generated. This color-coded image was overlaid onto the fascia image, and the epimysium was traced in white (RGB: 255, 255, 255). A white dot (RGB: 240, 240, 240) was placed at the outer edge of the fascia between the FDC and FACR to serve as the 0° reference point for the polar coordinate analysis.

Polar coordinate transformation was then performed using Python Code 2 (Supplemental Data 2), in which the filename of the image to be processed was specified in line 8 to generate the transformed images for subsequent quantitative analysis.

Calculation of individuals’ average muscle fiber size and visualization of the effects of denervation or electroporation of FGF2

An average image was generated from multiple color-coded muscle images using Python Code 3 (Supplemental Data 3). In Code 3, the filenames of the images to be processed were specified in line 36. Next, regions affected by denervation or by the overexpression of Fgf2, Fgf8, or Bmp2 were visualized using Python Code 4 (Supplemental Data 4). In Code 4, the filenames of the images to be loaded were specified in lines 47 and 48. The difference between the average scores of control and denervated muscles was then visualized using a new color map. Re-circularization after polar coordinate transformation was performed using Python Code 5 (Supplemental Data 5), in which the filename of the image to be processed was specified in line 5. The resulting image was rearranged such that the vertical axis represents radial distance and the horizontal axis represents circumferential angle.

Quantitative RT-PCR

RNA preparation for qRT-PCR was performed using TriPure isolation reagent (Roche). After the addition of 200 µl chloroform, samples were centrifuged at 3000 rpm for 5 min. The upper aqueous phase (300 µl) was carefully collected, and RNA was precipitated by adding 400 µl of isopropanol, followed by centrifugation at 15,000 rpm for 30 min at 4 °C. The resulting RNA pellet was washed with 70% ethanol and resuspended in 20 µl of RNase-free water. The reverse transcription reaction was performed using Prime Script II (Takara) at 42 °C for 60 min, and the enzymes were then inactivated at 95 °C for 5 min. Quantitative PCR was performed using KAPA SYBR FAST qPCR Master Mix (Kapa Biosystems) on StepOne real-time PCR system (Thermo Fisher). The primer sequences are as follows.

• Fgf2 forward: TCTTCCTTCGCATCAACCCC

• Fgf2 reverse: TTTCATTGCCATCAACCGCC

• Spry2 forward: GTCCGGACACAAAGAGCGTC

• Spry2 reverse: CTGAAATCCATGTCGCACGC

• Ef-1α forward: AACATCGTGGTCATCGGCCAT

• Ef-1α reverse: GGAGGTGCCAGTGATCATGTT

RNA was prepared from ventral forelimb muscle from at least four limbs at 0 days (intact) and 7 days after denervation. All samples were normalized to Ef-1α.

In situ hybridization

In situ hybridization was performed as follows. For probe synthesis, Fgf2 cloned into the pTAC-2 vector (BioDynamics Lab., Inc.) was amplified by PCR with KOD-Plus-Neo (Toyobo) using M13 primers. PCR fragments were purified and used as an RNA probe template. RNA probe synthesis was performed with T7 RNA polymerase (Takara) for 3 h, and RNA was hydrolyzed for 6 min. The sections were washed in PBT to remove the O.C.T. compound, treated with proteinase K (10 µg/mL) (Invitrogen)/PBT for 10 min at room temperature, washed in PBT, treated with 4% PFA/PBS for 20 min at room temperature, washed in PBT, and then probes were hybridized at 62.5 °C for approximately 16 h. The sections were washed in wash buffer 1 (formamide: H₂O: 20 × SSC [3 M NaCl: 0.3 M sodium citrate, pH 5.0] = 2:1:1), and then in wash buffer 2 (formamide: H₂O: 20 x SSC = 5:1:4). The samples were then incubated with Anti-Digoxigenin-AP Fab fragments (Sigma-Aldrich, 1/1000) for 2 h at room temperature. Samples were stained with BCIP (Nacalai Tesque) and NBT (Nacalai Tesque) in alkaline phosphatase buffer (0.1 M NaCl, 0.1 M Tris-HCl [pH 9.5], 0.1% Tween20) for 48 h at room temperature after washing in TBST.

Results

Denervation causes loss of muscle volume in axolotl limb

We first established a denervation procedure to assess its effect on axolotl forelimb muscles (Fig. 1). Two nerves—the Nervus medianus and the Nervus ulnaris—run alongside the brachial artery (Arteria brachialis) in the proximal forelimb (stylopod) (Fig. 1A) [15]. Denervation was achieved by excising a segment of these nerves. Successful denervation was confirmed in the distal forelimb (zeugopod) by immunofluorescence using an anti-neurofilament antibody (Fig. 1B–E). The size of each muscle tissue in the zeugopod was evaluated 7 days after denervation. Six distinct muscle tissues were identified in the zeugopod (Fig. 1F). Denervation resulted in a visibly smaller limb, particularly due to a decrease in ventral muscle mass (Fig. 1G, H). To quantify this observation, we measured the area of each muscle (Fig. 1I–N). Although all muscles appeared to be affected by denervation, statistically significant reductions were observed specifically in the ventral muscles: Extensor antebrachii ulnaris (EACU), Flexor digitorum communis (FDC), and Flexor antebrachii et carpi radialis (FACR). These results suggest that the Nervus medianus and the Nervus ulnaris primarily regulate ventral muscle groups to maintain muscle volume.

Fig. 1.

Effects of denervation on forelimb muscles of axolotls. A Schematic diagram of the denervation surgery, shown from the ventral view of the forelimb. HAB: Humeroantebrachialis muscle, CBL: coracobrachialis longus muscle. B–E Axons were visualized by immunofluorescence (NF, green). B, D Innervated limb. C, E Denervated limb. The asterisks and the dotted line mark regions where axons were lost due to the denervation surgery. Arrowheads indicate nerves, which were not affected by the denervation surgery. The denervated samples were harvested one week after the denervation. The forelimbs from the same animal were used, with one designated as CTRL and the other as denervated. F Segmentation of muscles in the zeugopod: Muscle of Extensor antebrachii ulnaris (EACU), Extensor digitorum communis (EDC), Extensor antibrachii radialis (EACR), Flexor antibrachii et carpi ulnaris (FACU), Flexor digitorum communis (FDC), and Flexor e antebrachii t carpi radialis (FACR). G, H Muscle tissues were visualized by immunofluorescence (MHC, green). G Innervated limb. H Denervated limb. Anatomical axes are indicated in the top right. D, V, A, and P denote dorsal, ventral, anterior, and posterior, respectively. I–N Quantification of the area of each muscle (n = 6). CTRL: control muscle (innervated). DeN: denervated muscle. Data are presented as means ± SEM. *p < 0.05, **p < 0.005, n.s., not significant; statistical analysis was performed using a two-tailed paired Student’s t-test. Scale bars: 200 μm B–E and 1 mm F–H. Representative tissue images are shown from four independent samples

Quantification of muscle loss by denervation

We next focused on each muscle fiber in the FDC and FACR. These two muscle tissues are relatively large and amenable to experimental analysis. The muscle fibers were visualized by immunofluorescence using an anti-myosin heavy chain (MHC) antibody (Fig. 2A, E, I, M). The outlines of the muscle fibers and nuclei were visualized using an anti-type III collagen antibody and Hoechst 33,342, respectively (Fig. 2B, C, F, G, J, K, N, O). No apparent histological changes were observed between the control (innervated) limb and the denervated limb, although the muscle fibers in the denervated limb appeared smaller (Fig. 2D, H, L, P). To quantify the decrease in fiber size in the FDC and FACR, we developed a method to objectively classify muscle fiber area (Suppl. Figure 1). Fibers were categorized into six ranges based on fiber size, and this classification was applied to tissue sections (Fig. 3A, B, F, G). Then, the number of fibers classified in each color category was counted (Fig. 3E, J). Denervation caused a decrease in larger fibers and an increase in smaller fibers. We also counted the number of nuclei and muscle fibers and found that neither was affected by denervation (Fig. 2C, D, H, I). These observations suggest that the decrease in muscle size is caused by fiber shrinkage, rather than by the decrease in the numbers of fibers or nuclei.

Fig. 2.

Comparison of muscle fibers in CTRL (control, innervated) and denervated limbs. The denervated samples were harvested one week after the denervation. MHC (green) and COL3A1 (red) expression patterns in FDC (A–H) and in FACR (I–P). The dotted lines outline the perimysium of FDC or FACR. Hoechst (blue) stains nuclei. A2–G2 and I2–O2 show higher-magnification views of the boxed regions in A1–G1 and I1–O1, respectively. D, H, L, P Representative high-magnification images. The upper right insets show merged images of MHC, COL3A1, and Nuclei within the same section. Scale bars: 800 μm (A1–G1, I1–O1), 400 μm (A2–G2, I2–O2), and 50 μm (D, H, L, P). Representative images are shown from six independent samples

Fig. 3.

Classification and quantification of muscle fiber size. A, B, F, G Representative images of muscle fibers classified into six color-coded categories. The black regions represent MHC-negative (MHC⁻) areas. Details of the classification method are provided in Supplemental Figure 1. C, H Average number of nuclei in the sections of FDC (C) and FACR (H) (n =6). CTRL: control muscles (innervated) and DeN: denervated muscles. D, I Relative muscle size (normalized to the value in control muscles; n = 6). E, J Proportion of fibers in each color category in control and denervated muscles (n = 6). Data in panels C and H are presented as box-and-whisker plots, in which the center line indicates the median, the box represents the interquartile range (IQR), and the whiskers indicate the minimum and maximum values. Data in panels D and I are presented as means ± SEM. *p < 0.05, **p < 0.005, ***p < 0.0005, n.s., not significant (p ≥ 0.05); statistical analysis was performed using a two-tailed Student’s t-test (D, H, I) and a two-tailed Welch’s t-test (C, E, J. Scale bars: 400 μm A, B, F, G

The classified map indicates that denervation exerted different effects depending on position. Quantifying such spatial variation was challenging, because comparisons across different samples required careful normalization. We developed a novel normalization method to account for anatomical variability (see Suppl. Figures 2–3 and the Materials and Methods section for details). Briefly, a polar coordinate transformation was applied to normalize anatomical variability and the normalized images were merged and averaged to identify the overall trend. For improved visualization, both the transformed images and the averaged images were subsequently re-circularized. This method enabled the analysis of spatial regulation using multiple samples with varied muscle shapes.

Using the established method, we generated average color maps from control and denervated samples. In both FDC and FACR, muscle fibers in the lateral (marginal) region tended to be thinner, whereas those in the medial (centroid) region were thicker (Fig. 4A, B, H, I). This trend was maintained even after denervation (Fig. 4C, D, I, K). To assess denervation-induced atrophy, changes in muscle size were visualized by generating data in which the score was obtained by subtracting the denervated data from the control data (Suppl. Figure 3 C; Fig. 4E, L). In the difference maps, blue areas indicate atrophy and red areas indicate hypertrophy (Fig. 4E, L). Viewed as a whole, atrophy (blue) appeared predominant in both the FDC and FACR. To quantify denervation-induced atrophy in segmented regions, the color-map information was transformed into bar graphs (Fig. 4F, G, M, N). In the FDC, size reduction was dominant across all areas (Fig. 4F). The degree of atrophy varied across “Areas a–d”, suggesting differences in sensitivity to denervation. Within zones 1–4, the lateral region (Zone 4) appeared less sensitive to denervation (Fig. 4G). In the FACR, a similar pattern of sensitivity to denervation was observed as in the FDC (Fig. 4M, N). Likewise, in the FACR, overall fiber size reduction was observed after denervation, although area c and zone 4 were less affected (Fig. 4L–N).

Fig. 4.

Visualization of atrophy using a color map. A, C, H, J Representative color-coded images of muscle tissue transformed using polar coordinates. B, D, I, K Averaged images were generated from multiple color-coded muscle images (n = 6). The averaging method is provided in Supplemental Figure 3A. A’–D’ and H’–K’ Re-circularized images. Details are provided in Supplemental Figure 3B. E, L Visualization of regions affected by denervation. The blue and red colors indicate pixels showing positive scores (atrophy) and negative scores (hypertrophy), respectively. Details are provided in Supplemental Figure 3C. The circle was divided into four areas (“Areas a–d”) at 90° intervals, and the radial distance was subdivided into four concentric zones (“Zones 1–4”) by quartering the radius. F, M Sum of pixel-level differences in per-degree average values between control and denervated muscles in “Areas a–d” of panels E and L. G, N Sum of pixel-level differences in radial (r) 1/100–interval average values between control and denervated muscles in “Zones 1–4” of panels E and L. Data are presented as means ± SEM. **p < 0.005, ***p < 0.0005, n.s., not significant (p ≥ 0.05); statistical analysis was performed using Tukey’s range test

Together, these data demonstrate that denervation induces muscle fiber atrophy and that specific muscle regions exhibit differential sensitivity to denervation.

FGF2 application rescues loss of muscle volume by denervation

Having clarified the role of nerves in maintaining muscle fibers, we next focused on nerve-derived molecules that may support this function in axolotl muscle. We previously demonstrated that nerves secrete growth factors to initiate limb regeneration in urodele amphibians. FGF2, FGF8, and BMP2 have been identified as nerve-derived factors in axolotl limb regeneration [13]. First, we focused on FGF2 among the identified nerve-secreted factors. Plasmids carrying FGF2-p2A-GFP were delivered to the FACR muscle via electroporation (Fig. 5). Electroporation was performed 5 days before denervation, and GFP signals were confirmed both 5 and 12 days post-electroporation (dpe). Muscle fibers that received the plasmids were identifiable by GFP fluorescence (Fig. 5A–D). Tissue sections were prepared from samples at 12 dpe. Immunofluorescence confirmed that electroporation successfully targeted a portion of the FACR (Fig. 5E–L). Electroporation did not cause any severe histological changes in the muscle (Fig. 5E–L). To assess the spatial effects of introduced FGF2, we made color-classified map images using the same analytical procedures as in Figs. 3 and 4 (Fig. 6A, B). Then, we counted the distribution of muscle size (Fig. 6F). Focusing on the larger fiber sizes, statistical analysis revealed that the FGF2-electroporated muscle fibers had a greater number of large fibers than the GFP-electroporated muscle fibers (Fig. 6F). We also quantified the number of nuclei (Fig. 6C), the number of muscle fibers (Fig. 6D), and the relative size of muscles (Fig. 6E). We observed no significant differences in the numbers of nuclei and myofibers. This suggests that changes in muscle size after electroporation were not due to changes in the numbers of fibers or nuclei. We further investigated the spatial effects of FGF2 introduction by electroporation (Fig. 6G–K). The color maps were transformed using polar coordinates, normalized, and re-circled (Fig. 6G–J). Fig. 6H and J represent merged data from four independent samples. FGF2 delivery appeared to alleviate denervation-induced muscle atrophy. To quantify this effect, we subtracted the color map data of the control (GFP-electroporated and denervated) from the FGF2-electroporated and denervated samples (Fig. 6K). Viewed as a whole, the FGF2-introduced muscles had larger muscle fiber sizes than the control (Fig. 6K). When examined within each defined area, all vertical-horizontal quadrants were dominated by positive scores, indicating that FGF2-electroporated muscle had larger muscle size than the control (Fig. 6L). Focusing on zones 1–4, there appears to be a trend that the medial zone has a higher score than the lateral zone (Fig. 6M). However, since the muscle size before electroporation was different, it is not appropriate to interpret this trend in our analysis. Importantly, all areas a–d show positive values, indicating that FGF2-electroporation contributed to the maintenance of muscle fiber size after denervation. Next, we investigated the relationship between the sites of the electroporated fibers and the rescue effects (Suppl. Figure 4). Regardless of the site of electroporation, the muscle rescue effects could be observed throughout the muscle tissue. FGF2 was secreted from the electroporated fibers and spread throughout the muscle tissue. These findings demonstrate that FGF2 can mitigate muscle atrophy induced by denervation.

Fig. 5.

Overexpression of Fgf2 in the denervated limb. A, B, E–H Electroporation of pCS2:FGF2-p2A-GFP (C, D, I–L) Electroporation of pCS2:GFP. A–D Ventral views. From top to bottom: bright-field, GFP, and merged images. Images were taken at 5 and 12 days post-electroporation (dpe). Denervation surgery was performed 5 days after electroporation in both groups. Merged images (A3–D3) were generated in Affinity Photo 2 using the “difference” merge function, with GFP-positive (GFP+) regions shown in magenta. E–L GFP, COL3A1, and nuclei expression patterns in FACR. E2–G2 and I2–K2 show higher-magnification views of the boxed regions in E1–G1 and I1–K1, respectively. H, L Representative high-magnification images. The upper right insets show merged images of GFP, COL3A1, and Hoechst. Scale bars: 5 mm (A–D), 800 μm (E1–G1, I1–K1), 400 μm (E2–G2, I2–K2), and 50 μm H, L. Representative images are shown from six independent samples

Fig. 6.

Visualization of areas affected by Fgf2 overexpression. A, B Muscle fibers are classified into six color-coded categories. C Average number of nuclei in the section of FACR (n = 6). D Average number of muscle fibers in the section of FACR (n = 6). E Relative muscle size (normalized to the value in Fgf2-overexpressing muscles; n = 6). F Proportion of fibers in each color category in Fgf2- and Gfp-overexpressing muscles (n = 6). Data in panel C are presented as box-and-whisker plots, in which the center line indicates the median, the box represents IQR, and the whiskers indicate the minimum and maximum values. Data in panels D and E are presented as means ± SEM. *p < 0.05, **p < 0.005, n.s., not significant (p ≥ 0.05); statistical analysis was performed using a two-tailed Student’s t-test (C–E) and a two-tailed Welch’s t-test F. G, I Images transformed into polar coordinates. H, J Averaged images from multiple color-coded muscle samples (n = 6). G’–J’ Re-circularized images. K Visualization of regions affected by Fgf2-overexpression. The orange and purple colors indicate pixels showing positive scores (muscle size is maintained) and negative scores (atrophy), respectively. L Sum of differences in average scores in per-degree average values between Fgf2- and Gfp-overexpressing muscles in the “Areas a–d”. M Sum of pixel-level differences in radial (r) 1/100–interval average values between Fgf2- and Gfp-overexpressing muscles in “Zones 1–4” of panels in K. Because Fgf2-overexpressing regions were merged differently across experiments, statistical comparisons among areas/zones are not meaningful and were therefore not performed. Scale bars: 400 μm (A, B)

Regarding the other nerve-secreted factors, FGF8 and BMP2, we did not observe any clear effect on denervation-induced atrophy (Suppl. Figure 5). We overexpressed Fgf8 and Bmp2 under the same conditions as Fgf2; however, neither Fgf8 nor Bmp2 electroporation mitigated denervation-induced muscle atrophy (Suppl. Figure 5). To assess the effects of introduced FGF8 and BMP2, we made color-classified map images using the same analytical procedures as in Fig. 3 (Suppl. Figure 5 A, B, G, H). We then quantified the distribution of muscle fiber sizes; however, no differences were observed following the overexpression of Bmp2 or Fgf8 (Suppl. Figure 5 F, L). We also collected data on the number of nuclei, the number of muscle fibers, and the relative muscle size; however, no significant differences were observed (Suppl. Figure 5 C–E, I–K). These results indicate that, among the nerve-derived factors examined, only Fgf2 effectively contributes to muscle maintenance.

FGF signaling inhibition causes muscle atrophy in axolotls

Next, we investigated whether decreased FGF signaling in the muscle leads to muscle atrophy in axolotls. We employed an organ culture system with or without SU5402, an FGF signaling inhibitor. The FACR muscle tissue was dissected between the proximal and distal tendons and incubated in culture medium for 3 days. In the control and SU5402-treated samples, cultured muscle fibers retained sufficient integrity to maintain myosin heavy chain expression (Fig. 7A, D). The outlines of muscle fibers and nuclei were visualized using an anti-type III collagen antibody and Hoechst 33,342, respectively (Fig. 7B, C, E, F). Muscle fibers were size-classified as described previously (Fig. 7G, H). Comparison between control and SU5402-treated samples revealed a reduction in the proportion of larger fibers in the SU5402 group (Fig. 7I). We also confirmed that the numbers of nuclei and muscle fibers remained unchanged between the two conditions (Fig. 7J, K). Regarding the size of the muscle tissue, the mean value was lower in the SU5402-treated group than in the control; however, this difference did not reach statistical significance (p = 0.058; Fig. 7L). Pharmacological inhibition of FGF signaling reduced fiber cross-sectional areas, supporting a requirement for FGF signaling in the maintenance of muscle size.

Fig. 7.

Inhibition of the FGF signaling pathway in FACR. The dissected FACR was incubated for three days with DMSO (CTRL) or the FGF signaling inhibitor (SU5402: 30 µM). A–F MHC, COL3A1, and nuclei expression patterns in the cultured FACR. A2–F2 show higher-magnification views of the boxed regions in A1–F1. G, H Color-coded images. I Proportion of fibers in each color category in the DMSO- and SU5402-treated FACR (n = 6). J Average number of nuclei in the section of FACR (n = 6). K Average number of muscle fibers in the section of FACR (n = 6). L Relative muscle size (normalized to the value in DMSO-treated muscles; n = 6). Data in panels J and K are presented as box-and-whisker plots, in which the center line indicates the median, the box represents IQR, and the whiskers indicate the minimum and maximum values. Data in panel L are presented as means ± SEM. *p < 0.05, **p < 0.005, n.s., not significant (p ≥ 0.05); statistical analysis was performed using a two-tailed Student’s t-test (J–L) and a two-tailed Welch’s t-test (I). Scale bars: 400 μm (A1–F1, G, H), 100 μm (A2–F2). Representative images are shown from six independent samples

FGF2 is sufficient to maintain FGF signaling in muscles

Finally, we investigated whether FGF2 derived from nerves contributes to limb muscle maintenance. We confirmed that axolotl Fgf2 is expressed in the dorsal root ganglia (DRG) neurons, which project axons into the limb (Fig. 8A, B). This is consistent with our previous study [13].

Fig. 8.

Neural Fgf2 expression and FGF signaling changes in muscle after denervation. A Fgf2 expression in the dorsal root ganglia (DRG) visualized by in situ hybridization. B Negative control (no RNA probe). A2 and B2 show higher-magnification views of the boxed regions in A1 and B1, respectively. C, D Expression patterns of phosphorylated FGFR1 (phosFGFR1) and MHC in innervated and denervated muscles. Panels C1 and D1 show phosFGFR1, and panels C2 and D2 show MHC. E Quantification of phosFGFR1 intensity in muscles. CTRL: control muscle (innervated) and DeN: denervated muscles. F Quantification of Fgf2 expression levels in innervated and denervated muscles. G Quantification of Spry2 expression levels in innervated and denervated muscles. Data in panels F and G are presented as box-and-whisker plots, in which the center line indicates the median, the box represents IQR, and the whiskers indicate the minimum and maximum values. Data in panel E are presented as means ± SEM. *p < 0.05, **p < 0.005, ***p < 0.0005, n.s., not significant (p ≥ 0.05); statistical analysis was performed using a two-tailed Student’s t-test. Scale bars: 400 μm (A1, B1), 100 μm (A2, B2, C, D). Representative images are shown from three independent samples

We also examined whether denervation led to a reduction in phosphorylated FGFR1 (phosFGFR1) levels in the muscle fibers (Fig. 8C1, D1). The muscle fibers were visualized by immunofluorescence using an anti-myosin heavy chain antibody (MHC; red) (Fig. 8C2, D2). FGF2 binds to FGFR1 during signal transduction, triggering tyrosine autophosphorylation of the receptor [16]. Thus, phosphorylation of FGFR1 serves as an indicator of active FGF signaling. In innervated muscle, phosFGFR1 was detectable along the muscle fibers, localized to the membrane and cytoplasm (Fig. 8C). By contrast, the signal was reduced in denervated muscle (Fig. 8D).

To quantify this reduction, we measured phosFGFR1 signal intensity via immunofluorescence (Fig. 8E). The normalized pixel-based intensity was significantly lower in denervated muscle, confirming a reduction in phosFGFR1 levels. To further characterize downstream signaling in muscle fibers, we examined ERK activation in isolated muscles cultured with exogenous FGF2 (Suppl. Figure 6). The muscle fibers were labeled with anti-myosin heavy chain antibody (MHC; red), and nuclei were labeled with Hoechst 33,342 (Nuclei; blue). Immunofluorescence for phosphorylated p44/42 MAPK (pERK; green) suggested that FGF2 treatment enhanced the nuclear accumulation of pERK in muscle fibers compared with control (Suppl. Figure 6 A– F). To quantify the extent of nuclear translocation, we calculated the ratio of nuclear to cytoplasmic pERK signal intensity based on immunofluorescence images (Suppl. Figure 6G). The nuclear-to-cytoplasmic pERK ratio was significantly higher in FGF2-treated muscles, confirming enhanced nuclear translocation of pERK. To assess whether the reduction of FGFR1 levels could be due to indirect effects—such as fibroblast inactivation—we performed qRT-PCR using muscle tissue samples with or without denervation (Fig. 8F, G). Because axonal projections lack transcriptional activity, dissected tissues should contain minimal transcripts from DRG neurons. Transcription occurs primarily in neuronal cell bodies located in the trunk. Interestingly, denervation led to an increase in Fgf2 expression within the muscle tissue (Fig. 8F). These results suggest that although denervation activates Fgf2 transcription within muscle tissue, this muscle-derived FGF2 does not contribute significantly to phosFGFR1 signaling. We also examined Spry2 expression as an indicator of FGF signaling [17]. Denervated muscle tissue showed reduced Spry2 transcript levels (Fig. 8G). This finding further supports the interpretation that the elevated Fgf2 expression observed in denervated muscle does not translate into functional FGF signaling. Taken together with the neural expression of Fgf2, our findings are consistent with a model in which nerve-derived FGF2 is an important contributor to the maintenance of muscle mass in axolotls.

Discussion

Establishment of denervation-induced muscle atrophy model in axolotls

In this study, we established a denervation-induced muscle atrophy model in axolotls. A key feature of this model is that muscle atrophy can be induced selectively in specific muscle groups through partial denervation. Specifically, we surgically damaged two nerves, the Nervus medianus and Nervus ulnaris, and observed that only the EACU, FDC, and FACR muscles exhibited atrophy (Fig. 1). Although denervation caused complete paralysis for a certain period, not all muscle tissues showed significant atrophy (Fig. 1). This indicates that the atrophy observed in the FDC and FACR results from a selective sensitivity to denervation rather than paralysis alone. It should be noted that paralysis in axolotls is temporary, as their nerves regenerate rapidly. Therefore, prolonged paralysis, as seen in other animals, may lead to more generalized muscle atrophy. Importantly, axolotl limb denervation enables separation of the effects of paralysis from those of denervation, supporting its value as a model for studying denervation-induced muscle atrophy.

We developed and applied a method to analyze multiple samples with differing muscle shapes. The axolotl FDC and FACR exhibit substantial anatomical variability, which complicates quantitative comparison. To address this, we applied color-based classification of fiber size and a geometric image transformation. Fiber size was discretized into stepwise bins, resulting in color maps (Suppl. Figure 2). We defined a centroid for each transverse outline, converted the color map to polar coordinates with θ = 0 aligned to the lateral edge of the fascia between the FDC and FACR, and remapped it to a square grid (Suppl. Figure 2, 3). To normalize shape, all images were rescaled to a common frame. This workflow enables direct comparison across samples despite differences in muscle shape. Because fiber size scales with overall animal size, we restricted the analysis to animals of similar size. Under these conditions, the method supports reproducible comparisons of fiber-size distributions across the entire muscle.

Regional heterogeneity of denervation-induced atrophy

In mice and humans, denervation typically causes atrophy primarily in fast-twitch muscle fibers [18]. However, in axolotls, due to the absence of specific antibodies to distinguish between fast and slow fibers, it is difficult to categorize muscle fibers definitively. It is clear that muscle atrophy occurs throughout the muscle tissue (Fig. 4E, L). On the other hand, it is also apparent that the extent of atrophy varies across different muscle fibers, with some fibers showing more noticeable atrophy than others (Figs. 2 and 3). This may result from the difference in muscle type. Further investigation is needed on this point.

In this study, we observed denervation-induced atrophy with apparent regional differences (Fig. 4). However, baseline heterogeneity in muscle-fiber size across the muscle must be considered. For example, area c of the FACR contains smaller fibers than the other areas (Fig. 4I), and fibers in the lateral (marginal) region of the FDC and FACR are smaller than those near the centroid (Fig. 4B, I). Because our quantitative approach is less sensitive to small absolute changes in regions composed of inherently small fibers, it may under-detect subtle decreases there. If denervation reduces fiber size by a uniform proportion across the muscle, a 10% decrease, for instance, would produce a larger absolute change in large fibers than in small fibers. We acknowledge this as a technical limitation; therefore, conclusions regarding regional differences in denervation effects on axolotl muscle atrophy should be interpreted with caution.

FGF2 as a candidate nerve-derived factor for muscle maintenance

In this study, we focused on FGF2 as a potential nerve-derived factor that could mitigate denervation-induced muscle atrophy. Previous research has shown that FGF2 is expressed in the dorsal root ganglia (DRG) of axolotls and can be transported along nerve axons to the limb [9, 13]. Our findings showed that electroporation of FGF2 into denervated muscle fibers significantly mitigated muscle atrophy (Fig. 5). The broad effect of FGF2, a secreted growth factor, suggests that nerve-derived factors could regulate muscle mass in a more global manner, rather than targeting specific muscle fibers. This raises the possibility that partial nerve damage could lead to a decrease in the overall concentration of FGF2, contributing to muscle atrophy.

Therapeutic implications of FGF2 in muscle atrophy

Our findings support FGF2 as a therapeutic candidate for muscle atrophy. Prior studies in rats have reported anti-atrophic actions of FGF2 [19], yet the mechanistic basis remains incompletely defined. Here, we show that peripheral nerves contribute to the maintenance of muscle size and that denervation likely reduces FGF signaling within muscle fibers, precipitating atrophy. In this context, exogenous FGF2 represents a mechanistically grounded approach to preserve muscle mass after denervation. The downstream signaling pathways mediating FGF2-dependent maintenance require further definition; the axolotl, with its relatively large cell size, provides a tractable system for such analysis. Although the full cascade that safeguards muscle fiber size is not yet resolved, our data demonstrate that FGF2 rescues denervation-induced loss of muscle volume. Collectively, these results highlight a direct role for nerves in maintaining muscle mass and identify FGF2 as a promising candidate for therapeutic development in neuromuscular atrophy.

Conclusion

We established an axolotl model of denervation-induced muscle atrophy and developed a normalization pipeline that enables cross-sample, spatially resolved comparison of muscle fiber size. Using this framework, we show that atrophy reflects reduced fiber size rather than loss of fibers and exhibits positional heterogeneity across the muscle. We further demonstrate that exogenous FGF2 is sufficient to mitigate atrophy after denervation, and that pharmacological blockade of FGF signaling reduces fiber size, supporting a requirement for this pathway in muscle maintenance. These findings position the axolotl as a tractable system for dissecting neuromuscular trophic signaling and identify FGF2 as a promising therapeutic candidate. Limitations include reduced sensitivity to subtle changes in regions with intrinsically small fibers and the pathway-level (rather than ligand-specific) nature of SU5402 inhibition. We were also unable to directly assess changes in protein synthesis or degradation induced by denervation or FGF2 supplementation, primarily because suitable antibodies for axolotl tissues are lacking. Future work using genetic or neutralizing approaches and optimized FGF2 delivery should refine the mechanism and translational potential.

Abbreviations

BMP

Bone morphogenetic protein

CBL

Coracobrachialis longus muscle

CMV

Cytomegalovirus (immediate-early) promoter

COL3A1

Collagen type III alpha 1 chain

DeN

Denervated

DMEM

Dulbecco’s Modified Eagle Medium

DMSO

Dimethyl sulfoxide

DRG

Dorsal root ganglia

dpe

Days post-electroporation

EACR

Extensor antebrachii radialis

EACU

Extensor antebrachii ulnaris

EDC

Extensor digitorum communis

ECM

Extracellular matrix

FACR

Flexor antebrachii et carpi radialis

FACU

Flexor antebrachii et carpi ulnaris

FDC

Flexor digitorum communis

FGF

Fibroblast growth factor

FGFR1

Fibroblast growth factor receptor 1

HAB

Humeroantebrachialis muscle

MHC

Myosin heavy chain

NF

Neurofilament

PBS

Phosphate-buffered saline

Authors’ contributions

H.N. designed the study, conducted experiments, analyzed data, prepared figures, and wrote the manuscript. A.O., S.F., and S.Y. provided animals, performed experiments, and analyzed data. A.S. designed the study, analyzed data, developed Python code, wrote the manuscript, and secured funding.

Funding

This work was supported by JSPS KAKENHI Grant-in-Aid for Scientific Research (B) (24K02034 to A.S.) and JSPS KAKENHI Grant-in-Aid for Transformative Research Areas (A) (24H01953).

Data availability

No datasets were generated or analyzed during the current study. All results necessary to support the conclusions are included in the manuscript. Primary/raw data are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

As described in the Methods section. All experimental procedures were approved by Okayama University and conducted in accordance with institutional guidelines. This study did not involve human participants or clinical trials; therefore, informed consent was not required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.