Tumor Necrosis Factor-α Inhibition of Skeletal Muscle Regeneration Is Mediated by a Caspase-Dependent Stem Cell Response


  • Viviana Moresi,

    1. Department of Histology and Medical Embryology, Sapienza University, Rome, Italy
    2. Interuniversity Institute of Myology, Sapienza University, Rome, Italy
    Search for more papers by this author
  • Alessandro Pristerà,

    1. Department of Histology and Medical Embryology, Sapienza University, Rome, Italy
    2. Interuniversity Institute of Myology, Sapienza University, Rome, Italy
    Search for more papers by this author
  • Bianca M. Scicchitano,

    1. Department of Histology and Medical Embryology, Sapienza University, Rome, Italy
    2. Interuniversity Institute of Myology, Sapienza University, Rome, Italy
    Search for more papers by this author
  • Mario Molinaro,

    1. Department of Histology and Medical Embryology, Sapienza University, Rome, Italy
    2. Interuniversity Institute of Myology, Sapienza University, Rome, Italy
    Search for more papers by this author
  • Laura Teodori,

    1. Biotechmed, ENEA-Casaccia, Rome, Italy
    Search for more papers by this author
  • David Sassoon,

    1. Myology Group INSERM U787, Université Pierre et Marie Curie, Paris, France
    Search for more papers by this author
  • Sergio Adamo M.D.,

    Corresponding author
    1. Department of Histology and Medical Embryology, Sapienza University, Rome, Italy
    2. Interuniversity Institute of Myology, Sapienza University, Rome, Italy
    • Department of Histology and Medical Embryology, Via A. Scarpa 14, 00161 Rome, Italy. Telephone: 39-0649766756; Fax: 39-0649766643
    Search for more papers by this author
  • Dario Coletti

    1. Department of Histology and Medical Embryology, Sapienza University, Rome, Italy
    2. Interuniversity Institute of Myology, Sapienza University, Rome, Italy
    Search for more papers by this author


Skeletal muscle is susceptible to injury following trauma, neurological dysfunction, and genetic diseases. Skeletal muscle homeostasis is maintained by a pronounced regenerative capacity, which includes the recruitment of stem cells. Chronic exposure to tumor necrosis factor-α (TNF) triggers a muscle wasting reminiscent of cachexia. To better understand the effects of TNF upon muscle homeostasis and stem cells, we exposed injured muscle to TNF at specific time points during regeneration. TNF exposure delayed the appearance of regenerating fibers, without exacerbating fiber death following the initial trauma. We observed modest cellular caspase activation during regeneration, which was markedly increased in response to TNF exposure concomitant with an inhibition in regeneration. Caspase activation did not lead to apoptosis and did not involve caspase-3. Inhibition of caspase activity improved muscle regeneration in either the absence or the presence of TNF, revealing a nonapoptotic role for this pathway in the myogenic program. Caspase activity was localized to the interstitial cells, which also express Sca-1, CD34, and PW1. Perturbation of PW1 activity blocked caspase activation and improved regeneration. The restricted localization of Sca-1+, CD34+, PW1+ cells to a subset of interstitial cells with caspase activity reveals a critical regulatory role for this population during myogenesis, which may directly contribute to resident muscle stem cells or indirectly regulate stem cells through cell-cell interactions.

Disclosure of potential conflicts of interest is found at the end of this article.


Skeletal muscle is capable of extensive regeneration following injury. Injury induces an early inflammatory response followed by the activation of resident muscle precursors (satellite cells), which subsequently differentiate to replace and repair muscle fibers [1]. A subpopulation of satellite cells re-enters a quiescent state, providing a reserve myogenic population [2]. These phenomena rely on asymmetric self-renewal and commitment of satellite cells [3]. Multipotent stem cells are also present in adult skeletal muscle, although their anatomical localization has not been clearly resolved and the extent of their contribution to regenerating muscle is unclear [4]. Although satellite cells are located underneath the basal lamina, non-lineage-restricted precursor cells with myogenic potential, such as mesoangioblasts, bone marrow-derived cells, and side population cells, display various localizations in the interstitial space [5, [6]–7] Mesoangioblasts have successfully been exploited in the treatment of muscular dystrophy in mice and dogs [8]. Transplant experiments showed that bone marrow-derived cell incorporation into myofibers occurs at a low frequency and increases following muscle injury [9]. Striking changes in the number of nuclei per fiber over time were reported, negatively affected by atrogenic stimuli such as denervation [10, 11]. The occurrence of fibers with centrally located nuclei associated with muscle hypertrophy in the absence of damage has been reported [12, 13]. These reports highlight the importance of precursor cell recruitment and incorporation into fibers during muscle growth and homeostasis. How cytokines influence muscle precursor cell behavior and muscle regeneration is of relevance in several pathological conditions characterized by elevated levels of cytokines associated with muscle wasting [14]. In particular, cachexia is a devastating fat and skeletal muscle wasting syndrome displayed by patients with chronic diseases, including cancer, AIDS, chronic heart, and kidney failure. Cachexia interferes with therapies and increases morbidity and mortality [15, 16]. Inflammatory cytokines promote cachexia and are targets of therapeutic approaches [15]. Several cytokines, including IL-1, IL-6, and tumor necrosis factor-α (TNF; abbreviation is in accordance with [17]), as well as other factors of tumor (proteolysis inducing factor) or host (interferon-γ, leukemia inhibitory factor, transforming growth factor-β) origin, have been identified as promoters of cachexia [14, 18]. We have demonstrated that TNF is sufficient to induce cachexia, as well as inhibition of muscle regeneration [19]. TNF has also been shown to downregulate the myogenic factor MyoD in vivo [20]. Muscle atrophy is coupled with an impairment in myogenic potential of muscle precursor cells [21, 22], confirming that these cells are required for muscle recovery. Multiple pathways are involved in the TNF-mediated inhibition of myogenesis, including the downregulation of MyoD and myogenin [23], decrease in MyoD protein stability [24], and induction of proliferation through cyclin D1 [25]. We showed that TNF-mediated inhibition of muscle differentiation requires the p53 cell-death effectors PW1 and Bax [26, 27]. Furthermore, we reported that muscle stem cells show constitutively activated p53 and that loss of p53 function alters muscle stem cell number [28]. In addition, tumor-bearing p53-null mice are resistant to cachexia. We have also shown that the p53 effector PW1 cooperates with p53 in regulating stem cell number and muscle atrophy in vivo [28]. In cultured myogenic cells, PW1 recruits p53-dependent apoptotic pathways, including downstream caspase activation, which participates in the regulation of myogenesis [27]. Whereas apoptosis is a most likely cellular outcome of caspase activation, several studies demonstrate that caspases also play a role in mediating cell differentiation in specific lineages [29, [30]–31].

TNF levels are barely detectable in uninjured skeletal muscle and increase following muscle injury [32, 33]. TNF localizes to the infiltrating inflammatory cells during the first 3 days following injury and is subsequently detected in regenerating myofibers [32, 34]. Although in vivo studies reveal a role for TNF during muscle regeneration [20, 32, 34, [35]–36], the molecular and cellular pathways triggered by TNF in this process remain poorly understood.

A direct role for myogenic stem cells during cachexia remains to be clearly demonstrated, although deregulation of stem cell number or behavior leads to decreased muscle mass [37, 38]. We recently reported that cachexia is associated with more hematopoietic stem cells in skeletal muscle, coupled with a compromised regenerative capacity of the musculature [19, 39]. In this context, it is important to address the mechanisms underlying such impaired regenerative potential of the muscle. We propose that TNF abrogates stem cell function, which in turn delays or inhibits muscle regeneration.

In this study, we demonstrated that TNF inhibits muscle regeneration by upregulating caspase activity in a subpopulation of interstitial cells that can be identified by specific stem cell markers, such as Sca-1+, CD34+, and PW1+. Caspase activation occurs in the absence of apoptosis and in a PW1-dependent manner. These results provide new insights in the molecular mechanisms underlying TNF-mediated effects on muscle regeneration.

Materials and Methods

Experimentally Induced Muscle Injury and Other In Vivo Manipulations

To induce freeze injury, the tip of a steel probe precooled in dry ice was applied to tibialis anterior muscle belly of anesthetized adult (7–8 weeks old) female CD1 mice for 10 seconds. As depicted in supplemental online Figure 1A and 1B, this procedure induced a focal injury extending distally from the spike of the tibia and spreading over approximately one-third of the muscle. As labeled by Evans Blue Dye (EBD) 1 day (1d) following injury, the lesion site has a hemispheric shape. As a measure of its dimensions, we evaluated its average length (2.5 × 103 μm) and maximal cross-sectional area (1.4 × 106 μm2). The injury dimensions in independent replicates had SEMs of 8% and 14% of the above values, respectively: we considered these results satisfactory for the reproducibility of the amount of injury.

Injections were performed twice per muscle, by inserting the needle of a 0.3 ml/29-gauge syringe (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com) for 1 mm in the distal part of the muscle (i.e., in a region that is not taken into account for histological analysis). We noted that such procedure did not induce muscle fiber damage, inasmuch as EBD uptake was not detectable following injection in an adult muscle (data not shown). However, needle puncture has been reported by others to produce muscle regeneration [40]. To evaluate the damage induced by the combination of our experimental manipulations (i.e., two injections followed by one DNA injection and electroporation) used in a subset of experiments, we performed a time course of the fiber permeability to EBD following freeze injury in the absence or presence of such manipulations (supplemental online Fig. 2). As expected, altogether the experimental manipulations appeared to boost fiber damage at 4d following injury and to prolong and enhance the inflammatory phase. In this context, we also noticed that EBD+ fibers were detectable up to 4d following injury, whereas regenerating fibers (i.e., with centrally located nuclei) were detectable from 8d following injury.

To treat muscles, 25 μl of 5 μg/ml TNF (7.3 μmol per muscle) or 25 μl of phosphate-buffered saline (PBS) was injected in injured tibialis anterior 2 and 4 days after injury. When needed, 25 μl of 1 mM (2.5 nmol per muscle) pan-caspase inhibitor Z-VAD-FMK (benzyloxycarbonyl-Val-Ala-Asp fluoromethylketone; Trevigen, Gaithersburg, MD, http://www.trevigen.com) was injected alone or in combination with 25 μl of 5 μg/ml TNF. In the latter case, no leakage was observed, with the exception of sporadic experiments that were aborted. The animals were sacrificed at the following time points, on the basis of the kinetics of muscle damage and regeneration reported in Figure 1 and supplemental online Figures 1 and 2: 2 and 4 days following injury to detect fiber damage, apoptosis, and caspase activation; and 7, 14, and 28 days following injury to detect regenerating fibers in either nonelectroporated or electroporated muscles. Treatment of mice was in accordance with the guidelines of the institutional Animal Care and Use Committee.

Figure Figure 1..

TNF negatively affects skeletal muscle regeneration. Tibialis anterior muscle was subjected to freeze INJ, injected with PBS or TNF 2 and 4 days later, and analyzed at 1, 2, and 4 W following INJ. (A): Representative H&E-stained muscle cross-cryosections showing the regeneration area (upper panels) 1 W after INJ. In CTR muscle, the regeneration area could not be depicted by a single microphotograph, whereas in TNF-treated muscle, the regeneration area was significantly smaller. At higher magnification (lower panels), the presence of muscle fibers characterized by centrally located nuclei (i.e., regenerating fibers) was detectable in the presence of both treatments, but regenerating fibers exposed to TNF appeared fewer and smaller. In TNF-treated muscle, a higher number of infiltrated cells was also manifest. Scale bars = 50 μm. (B): The graph shows the average number of the regenerating fibers per muscle cross-section at 1, 2, and 4 W after INJ, of CTR and TNF-treated muscles. TNF decreased the total number of regenerating fibers detectable per cross-section. *, p < .005 versus TNF, by the Tukey honest significant difference (HSD) test. (C): Histograms showing the distribution of the regenerating fiber cross-sectional areas (fiber number/cross-section) in different size classes (cross-sectional area), at 1, 2, and 4 W of regeneration. Red lines indicate fiber size medians. TNF induces a shift in the distribution of the regenerating fiber size toward the smaller classes of area in the first 2 W of regeneration. *, p < .05 versus TNF, by the Tukey HSD test. (D): Reverse transcription-polymerase chain reaction on RNAs from adult and regenerating muscles at 4, 7, and 10 days after INJ. Myog and Neo-MHC transcripts were analyzed as marker of differentiation, whereas GAPDH expression is reported as a loading CTR. TNF delayed expression of differentiation markers in regenerating muscle. Abbreviations: C, control; CTR, control; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; INJ, injury; Myog, myogenin; Neo-MHC, neonatal myosin heavy chain; T, tumor necrosis factor-α-treated muscles; TNF, tumor necrosis factor-α; W, week(s).

DNA Delivery by Electroporation

To induce expression of ΔPW1-green fluorescent protein (GFP), we used a construct having the cDNA for an N terminus-deleted form of PW1 under control of the cytomegalovirus obtained by subcloning ΔPW1 [41] in a GFP-C1 (Clontech, Mountain View, CA, http://www.clontech.com). Four and a half days after freeze injury, the tibialis anterior was injected with 25 μg of ΔPW1-GFP or soluble NSF attachment protein (Snap)-GFP (generously provided by Pr. Tullio Pozzan, University of Padua, Padua, Italy) and immediately subjected to electroporation. Electroporation was performed by delivering six electric pulses of 20 V each (three with the anode placed on the frontal side of injured tibialis anterior, followed by three pulses at inverted polarity). A pair of 3 × 5-mm Genepaddle electrodes (BTX, San Diego, http://www.btxonline.com), placed on opposite sides of the muscle, was used to deliver the electric pulses. The electrodes cover an area corresponding to the vast majority of the muscle surface and therefore totally overhang the area of injury. Supplemental online Figure 1E and 1F depicts examples of muscle expression of exogenous DNA (Snap-GFP), demonstrating that the expression of exogenous DNA delivered by injection followed by electroporation is highly consistent throughout a muscle. In the same context (supplemental online Fig. 1D), it is worth mentioning that electroporation on a regenerating muscle at 4d following injury targets single interstitial cells and small regenerating fibers, as well as uninjured fibers.

Immunohistochemical and Histological Analysis

Muscles were sectioned throughout the whole length, and the injury/regeneration site was identified by extemporary toluidine blue staining. At the level of the injury/regeneration site, all the sections, 8 μm thick, were collected. The section containing the maximal cross-sectional area of injury/regeneration (described above; supplemental online Figure 1) was identified and defined as center of freeze injury. Only the sections within 400 μm of the center of freeze injury were analyzed. For each sample, the whole cross-section was analyzed.

Cryosections were fixed in 4% paraformaldehyde, permeabilized in methanol at −20°C, and rehydrated in PBS. After incubation in 2% bovine serum albumin + 5% goat serum in PBS (BB) for 1 hour, the samples were incubated with a 1:250 dilution of polyclonal anti-activated capsase-3 (Becton Dickinson), a 1:100 dilution of polyclonal anti-laminin antibody (Ab) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), a 1:4,000 dilution of anti-PW1 antibody [42] in BB, a 1:50 dilution of fluorescein isothiocyanate-conjugated anti-mouse Sca-1 monoclonal antibody (Becton Dickinson), or a 1:50 dilution of anti-CD34 rabbit polyclonal antibody (H-140; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com). The unconjugated primary Ab were detected by anti-rabbit-Alexa 488-conjugated Ab or -Alexa 568-conjugated Ab (Molecular Probes, Eugene, OR, http://probes.invitrogen.com), diluted 1:500 in BB. Alternatively, the anti-laminin Ab was detected by anti-rabbit-Cy5-conjugated Ab diluted 1:600 in BB (Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com). The Cy5 signal was processed (hue, −150) with Photoshop (Adobe Systems Inc., San Jose, CA, http://www.adobe.com) to convert it to an orange hue that would be better distinguished from red signals. Nuclei were stained by a 1-minute incubation with 5 ng/ml Hoechst 33342 in PBS (Sigma-Aldrich). For the morphological evaluation of skeletal muscle, cryosections were stained with hematoxylin and eosin (H&E; Sigma-Aldrich) using standard methods.

To stain damaged muscle fibers, the plasma membrane-impermeable Evans Blue Dye was i.p. injected in mice (10 μl/g body weight of a 1% solution in PBS; Sigma-Aldrich) 7 hours before sacrifice, as described previously [43]. Animals were sacrificed at the indicated times and muscles were frozen in liquid nitrogen-cooled isopentane. EBD signal was recorded on paraformaldehyde-fixed sections by using a 516/560 excitation/emission pair.

Caspase Activity

Enzymatic Assay.

A fluorometric analysis exploiting the CaspGLOW red active caspase staining kit (Biovision Inc., Mountain View, CA, http://www.biovision.com) was developed in our laboratory; it consisted of a macro (2-ml cuvettes) assay of skeletal muscle lysates incubated with the Red-conjugated pan-caspase inhibitor VAD-FMK [44]. The injured area of tibialis anterior was isolated, finely minced, and transferred in 1.5-ml tubes. After centrifugation at 400g, samples were washed twice in the manufacturer's wash buffer for 10 minutes each and lysed in 1 ml of lysis buffer (5 mM Tris-HCl, pH 8; 10 mM EDTA; 0.5% Triton) for 30 minutes at 4°C. The cytosolic fraction was used to perform the enzymatic assay, whereas the nuclear pellet was used to measure DNA content as described previously [45]. Caspase activity was measured by incubating the cytosolic fraction with a working solution of Red-VAD-FMK (1:300 in PBS) at 37°C for 45 minutes. The incubation was followed by extensive washes to eliminate unbound Red-VAD. Fluorometric readings were performed at a wavelength pair of 540/570 nm excitation/emission for caspase activity and 365/460 nm excitation/emission for DNA content. Samples were normalized by DNA content and expressed as fold increase over controls (injured, PBS-treated muscles). As positive and negative controls, we used differentiated C2C12 cell cultures treated or not treated with 2.5 μg/ml puromycin for 24 hours, respectively.

In Situ Staining.

We developed a novel assay to detect active caspases in situ on skeletal muscle cryosections, exploiting the CaspGLOW red active caspase staining kit (Biovision). To validate this approach, we performed experiments of caspase activity colocalization both in vitro and in vivo by using the CaspGLOW substratum and/or its nonfluorescent counterpart Z-VAD in combination with an antibody-antiactivated caspase-3 (supplemental online Fig. 3). Puromycin treatment used to induce apoptosis in C2C12 cells resulted in a dramatic increase of the number of cells positive for both the CaspGLOW signal and the activated capsase-3 (supplemental online Fig. 3A), concomitant with massive caspase-3 expression upregulation (supplemental online Fig. 3B). In vivo, we found that CaspGLOW activated capsase-3 colocalization in muscle fibers characterized by an altered morphology (i.e., swollen and invaded by several cells), consistent with fiber death, 2d following injury (supplemental online Fig. 3). To further validate the specificity of the in situ reaction, a nonfluorescent form of caspase substratum (i.e., Z-VAD) was used alone or in excess compared with the red-VAD in a competition experiment: in both cases, the CaspGLOW signal was absent from structures showing caspase-3 activity.

For the above assay, the cryosections were incubated with Red-VAD-FMK diluted 1:300 in PBS in a humidified chamber at 37°C for 45 minutes, washed in two changes of wash buffer, and fixed in 2% paraformaldehyde in PBS for 20 minutes at room temperature. After three washes in PBS, the specimens were processed for further immunohistochemical analysis.

Analyses of Apoptosis

Caspase-3 Staining.

Caspase-3 staining is described in Immunohistochemical and Histological Analysis.

Terminal Deoxynucleotidyl Transferase dUTP Nick-End Labeling Assay.

The denatured DNA of apoptotic cells was detected in muscle cryosections using an anti-single strand DNA Ab (Chemicon, Temecula, CA, http://www.chemicon.com) according to the manufacturer's instructions. Immunofluorescence for laminin was performed as described above, following the terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay.

DNA Ladder.

The chromatin was isolated from muscles 4.5 and 7 days following injury, and DNA was quantified by Hoechst fluorometric analysis as described above. Per well, 8 ± 1.5 μg was loaded on 1.2% agarose gel. A positive control for DNA laddering was obtained by treating C2C12 myotubes with puromycin for 12 hours.

Transmission Electron Microscopy.

The damaged region of the muscle was isolated, cut in 1 × 1 × 2-mm pieces, fixed in 4% glutaraldehyde, and processed by standard methods. Ultrathin cross-sections were analyzed with a Hitachi H7000 electron microscope (Hitachi Ltd., Tokyo, http://www.hitachi.com). Approximately 20 semithin sections sampled every 5 μm were extemporarily examined by toluidine blue staining to detect the presence of muscle damage. Two series of ultrathin sections were prepared at two different levels in this range per sample and analyzed.

Reverse Transcription-Polymerase Chain Reaction

Total RNA was prepared from tibialis anterior using Trizol reagent (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) following the manufacturer's protocol. Reverse transcriptase-polymerase chain reaction (RT-PCR) was performed using 2 μg of total RNA reverse-transcribed using Moloney murine leukemia virus reverse transcriptase (Invitrogen). PCRs were carried out in a final volume of 50 μl in a buffer containing 1 μl of RT reaction, 200 μM dNTP, 1.5 mM MgCl2, 0.2 μM each primer, and 1 U of Taq-DNA polymerase (Invitrogen). The PCR products were analyzed in 2% agarose gel. The following specific primers were used: myogenin, forward, 5′-CTGGGGACCCCTGAGCATTG-3′; reverse, 5′-ATCGCGCTCCTCCTGGTTGA-3′; neonatal myosin heavy chain, forward, 5′-AACTGAGGAAGACCGCAAGAATG-3′; reverse, 5′-AAGTAAACCCAGAGAGGCAAGTGACC-3′. The following oligonucleotides were used to detect glyceraldehyde 3-phosphate dehydrogenase transcript (used as internal control): GAPDH, forward, 5′-AACATCAAATGGGGTGAGGCC-3′; reverse, 5′-GTTGTCATGGATGACCTTGGC-3′.

Morphometric Analysis and Statistics

Photomicrographs of all the regenerating fibers (identified by morphological criteria, i.e., centrally located nuclei in H&E-stained cryosections) were taken at standard (1,300 × 1,030-pixel) resolution and analyzed using Scion Image software (version beta 4.0.2; Scion Corporation, Frederick, MD, http://www.scioncorp.com). For morphometric evaluation of fiber size, 200–1,000 cross-sectioned fibers per sample were analyzed. All data are expressed as mean ± SEM. Differences between the regenerating fiber numbers were determined by two-way analysis of variance (ANOVA) and post hoc tests using SPSS 12.0 software (SPSS, Chicago, http://www.spss.com). Alternatively, statistical analysis was performed by Student's t test.


TNF Inhibits Muscle Regeneration Without Increasing Muscle Damage Following Injury

The tibialis anterior was subjected to freeze injury, which yields reproducible muscle damage (supplemental online Fig. 1A, 1B). The extent of muscle damage was followed using EBD, a marker of plasma membrane leakage (supplemental online Fig. 1A). The kinetics for the two experimental settings used in our study are summarized in supplemental online Figure 2. Freeze injury was followed by an inflammatory phase (1–3 days after damage) characterized by the presence of numerous mononucleated cells in the injury site (supplemental online Figs. 1C, 2). Regenerating fibers (identified by centrally located nuclei) were first detected on day 4, indicating the onset of overt regeneration. Following the formation of new fibers, regeneration proceeded with fiber size increase, until the newly formed/repaired fibers could be distinguished only by the presence of centrally located nuclei (supplemental online Figs. 1C, 2). On the basis of the kinetics of regeneration in this model, muscles were treated with TNF by intramuscular injection on days 2 and 4 after injury. We found that 1 week after injury, TNF-treated muscles showed a reduction in the extent of the regenerating zone and a decrease in the regenerating fiber cross-sectional area, compared with controls (PBS-treated) (Fig. 1A). By morphometric analysis, we found that the regenerating fiber number peaked 1 week after injury in controls, whereas TNF-treated muscle showed a 60% reduction in the regenerating fiber number (Fig. 1B). The number of regenerating fibers decreased in the controls at all later time points (Fig. 1B), likely as the result of the attainment of a mature fiber phenotype [1]. In contrast, the number of fibers with centrally located nuclei did not change significantly in TNF-treated muscles after the 1st week (Fig. 1B). As shown by two-way ANOVA, TNF treatment and time following injury interacted in negatively affecting the number of fibers with centrally located nuclei, with TNF further reducing the number of fibers with centrally located nuclei decreasing with time. Post hoc analyses revealed that the effects of TNF were highly significant 1 week after damage (p < .01; Tukey honest significant difference test). Analysis of the regenerating fiber cross-sectional area revealed the dynamics of fiber size change over time and allowed us to calculate the medians of the regenerating fiber size upon the different treatments. Two-way ANOVA revealed that both TNF and time affected the median fiber size and that these variables interacted (F = 13.58, p < .01, df = 1; F = 4.28, p < .03, df = 2; and F = 3.72, p < .05, df = 2 for treatment, time, and interaction, respectively). Furthermore, the comparison of median fiber size between TNF-treated and control muscles showed that TNF induced a significant decrease in size compared with controls at the 1- and 2-week time points following injury (Fig. 1C). In addition, we noted a marked reduction in the number of both small and large regenerating fibers 2 weeks after injury in TNF-treated muscles, suggestive of delayed fiber formation, as well as maturation (Fig. 1C). We further investigated these phenomena using RT-PCR for markers of muscle development. In control regenerating muscle, myogenin and neonatal myosin heavy chain were expressed 4.5 days after injury and were downregulated by day 7 following injury (Fig. 1D). In contrast, TNF-treated muscle displayed delayed expression of both genes (Fig. 1D).

Although our data demonstrate that TNF delays regeneration, we further investigated whether this delay is accompanied by an increase in muscle fiber death. We analyzed the extent of fiber damage on days 4.5, 8, and 14 following injury (Fig. 2; supplemental online Fig. 2). EBD staining [43] was almost absent in uninjured muscle. In contrast, freshly injured muscle had numerous EBD+ fibers (Fig. 2A). We found no detectable spatial or temporal differences in the damaged area between control and TNF-treated muscles (Fig. 2; supplemental online Fig. 2). We immunostained damaged muscle for the activated form of caspase-3 at various time points to test whether cell death pathways were specifically activated (Fig. 2B). We found that caspase-3 activation followed the occurrence of EBD signal and was localized in most EBD+ fibers on 2d following injury. Caspase-3 activation decreased by 4d following injury, a time point when some EBD+ fibers were still detectable. TNF treatment did not affect these kinetics compared with the control (Fig. 2B).

Figure Figure 2..

TNF does not enhance fiber death following freeze injury. (A): Uninjured or freshly injured (postinjury: 1d) muscles were injected with phosphate-buffered saline (PBS) (CTR) or TNF 2 and 4 d following injury and analyzed for the presence of fiber damage by Evans Blue Dye (EBD) staining (red) at 4 and 8 d following injury. Nuclei were counterstained by Hoechst (blue). The epimysium is indicated by arrows. Spatial and temporal extension of fiber damage was unaffected by TNF treatment. (B): More detailed kinetics of fiber damage (Evans Blue Dye+, red) was combined with activated caspase-3 immunostaining (green) and nuclear Hoechst staining (blue). Also shown are the corresponding three-color merged images. Portions of cryosections from uninjured, PBS-treated (CTR), and TNF-treated muscle are shown. A transient caspase-3 activation followed plasma membrane damage and peaks at 2 d following injury. Abbreviations: CTR, control; d, day(s); TNF, tumor necrosis factor-α.

TNF Increases the Number of Cells with Caspase Activity Infiltrating the Regenerating Muscle in the Absence of Apoptosis

To investigate the molecular mechanisms underlying TNF-mediated inhibition of skeletal muscle regeneration, we analyzed the levels of activated caspases. To verify our fluorometric assay for caspase activation, we induced apoptosis in skeletal myogenic cells in culture by puromycin treatment, which induced an ∼3.5-fold upregulation of caspase activity compared with controls (Fig. 3A, left panel). Whereas uninjured muscle showed barely detectable levels of activated caspases, we observed that regenerating muscle showed a measurable activity (Fig. 3A, right panel). Interestingly, we detected a statistically significant increase (∼30%) in the amount of activated caspase induced by TNF compared with regenerating control muscles (Fig. 3A, right panel). Intramuscular injection of the pan-caspase inhibitor Z-VAD 3 days prior to the fluorometric assay decreased caspase activation (Fig. 3A, right panel).

Figure Figure 3..

Caspase activity is detectable in regenerating muscle and is increased by TNF treatment in the absence of apoptosis. The tibialis anterior was injured and injected with phosphate-buffered saline (PBS) (CTR), pan-caspase inhibitor (Z-VAD), and/or TNF 2 and 4 d later. (A): Positive control (PURO) and negative CTR were obtained by treating C2C12 cell cultures with PURO or vehicle, respectively, and the corresponding caspase activation is shown in the left panel. In the right panel, caspase activity as detected in regenerating muscle lysates at d 4.5 after injury by fluorometric analysis is compared with uninjured muscle (adult). Regenerating muscle showed a significant increase in caspase activity compared with adult muscle. This activity was inhibited in vivo by Z-VAD administration. TNF treatment significantly increased caspase activity in regenerating muscle compared with the control. The results are the means ± SEM of five independent experiments performed in triplicate. Student's t test: *, p < .03 versus CTR; **, p < .01 versus CTR. (B): Cryosections of tibialis anterior immunohistochemically stained for caspase activity at 4.5 and 14 d following injury (red, upper panels and lower panels, respectively). Laminin (green) and nuclei (blue) were also labeled to highlight muscle architecture, by immunofluorescence and Hoechst stain, respectively. The inset shows a higher magnification of cells with activated caspases, more abundant in TNF-treated muscle. (C): On other sections of the same samples, the immunostaining above was combined with TUNEL assay to detect apoptotic cells at 4.5 d following injury (TUNEL+ cells shown in red). The inset shows a higher magnification of an apoptotic cell. The histograms show the occurrence of cells positive for either caspase activity or TUNEL as measured by counting the positive cell number normalized by surface of cryosection analyzed. (+cells per mm2). TNF treatment increased the number of cells with active caspases but not that of apoptotic cells. The means ± SEM of five independent experiments are shown. Student's t test: *, p < .01 versus CTR. Scale bars = 100 μm. (C): DNA ladder in muscles collected 4.5 d (left) and 7 d (right) following injury. Positive CTR (PBS-injected; CTR+) consisted of DNA extracted from apoptotic C2C12 myotubes. (D): Transmission electron microscopy images of regenerating tibialis anterior injected with TNF or PBS (CTR) 2 and 4 d after injury and analyzed on d 4.5 following injury. In both cases, extensive fiber damage was visible and apoptosis was sporadically detectable in mononucleated cells (inset). No modulation of apoptosis by TNF was detectable with any of the approaches above. Abbreviations: CTR, control; d, day(s); MW, DNA molecular weight; PURO, puromycin; TNF, tumor necrosis factor-α; TUNEL, terminal deoxynucleotidyl transferase dUTP nick-end labeling.

To identify cells with active caspases in regenerating muscle, we applied the in situ caspase activity assay to muscle cryosections. Validation of this novel histochemical method is described in supplemental online Figure 3. Using this approach, we localized caspase activity to a subset of mononuclear interstitial cells (Fig. 3B). The caspase-positive cells did not appear to be associated with any specific anatomical structure (e.g., vasculature, tendons, nerves, etc.). In agreement with the quantitative analysis (Fig. 3A), we observed an increase in the density of cells with activated caspases in TNF-treated muscle compared with the control (Fig. 3B, graph). TNF-mediated effect was detectable by 4.5 days of regeneration and persisted for up to 2 weeks following injury (Fig. 3, upper and lower panels, respectively). To verify whether caspase activity was coupled with apoptosis in regenerating muscle, we performed TUNEL and DNA ladder assays and transmission electron microscopy analysis on muscle samples at 4.5, 7, and 14 days following injury. Although we detected a sporadic presence of apoptotic cells in regenerating muscle, we did not observe TNF-induced changes in the amount of apoptosis at any of these time points (Fig. 3C, panels and graph; Fig. 3D, 3E; data not shown).

The interstitial localization of cells with activated caspase excludes the satellite cell compartment (Fig. 4A). By confocal analysis, we also observed that the vast majority of interstitial cells with activated caspases (as monitored by red-VAD-FMK) did not express the proapoptotic activated caspase-3. The latter colocalized with the red-VAD-FMK signal exclusively in large fibers invaded by several nuclei and in rare interstitial cells, characterized by a nuclear morphology different from that of the caspase-3-negative cells (Fig. 4A). To further characterize the cell population showing activation of caspases different from caspase-3, we combined the red-VAD-FMK caspase staining with immunofluorescence analyses for stem cell markers in cryosections of control and TNF-treated muscle at 4.5 days of regeneration and found that these cells expressed stem cell antigen-1 (Sca-1), CD34, and PW1 (Fig. 3B; Table 1). Colocalization of caspase activity and stem cell markers was independent of treatment (Table 1), although we noted a higher number of cells with detectable caspase activity in TNF-treated muscle.

Figure Figure 4..

Interstitial cells with active caspases other than caspase-3. (A): The tibialis anterior was injured and injected with phosphate-buffered saline (CTR) or TNF 2 and 4 days later. Muscle cryosections at 4.5 d following injury are shown. Four-color merged images depict the differential localization of activate caspase signal (red-VAD-FMK), active caspase-3 (green) and laminin (orange). Nuclei were Hoechst-stained (blue). Signal colocalization of the pan-caspase marker red-VAD-FMK and of the immunolabeled active form of caspase-3 was restricted to dying fibers and to rare cells characterized by a large nucleus. Most of interstitial cells that showed pan-caspase activity (arrowheads), although more abundant in TNF-treated muscle, did not show activated caspase-3, in agreement with a nonapoptotic role for caspases in response to TNF. (B): Cells with activated caspases in regenerating muscle expressed stem cell markers. Muscle cryosections were histochemically stained for caspase activity (red) and Sca-1, or CD34 or PW1 as indicated (green). Micrographs of either labeling merged with nuclear Hoechst staining (blue) are shown. Arrows indicate examples of colocalization, also shown at higher magnification in the red, blue, and green insets. Scale bars = 100 μm. Abbreviations: CTR, control; TNF, tumor necrosis factor-α.

Table Table 1.. %Percentage of cells with caspase activity also expressing stem cell markers
original image

Caspase Inhibition Promotes Skeletal Muscle Regeneration and Counteracts TNF Effects

To verify whether caspase activity contributes to TNF-mediated inhibition of regeneration, we treated regenerating muscles with Z-VAD, at 2 and 4 days following injury, in the absence or presence of TNF. One week after injury, Z-VAD treatment resulted in an increase in the number of small regenerating fibers (<600 μm2) compared with controls (Fig. 5). Two weeks after injury, Z-VAD-treated muscles showed an increase in both number and size of the regenerating fibers (Fig. 5). In combination with TNF, Z-VAD treatment resulted in a marked increase in the number and size of regenerating fibers compared with TNF-treated muscles at both 1 and 2 weeks following injury (Fig. 5). These data demonstrate that caspase activity is required for the TNF-dependent inhibition of myogenesis in vivo.

Figure Figure 5..

Caspase activation inhibits muscle regeneration. (A): Injured tibialis anterior was injected with phosphate-buffered saline (CTR), pan-caspase inhibitor (Z-VAD), TNF, or a combination of TNF and Z-VAD 2 and 4 days after injury. Representative photomicrographs of H&E-stained muscle cross-cryosections show regenerating fibers (characterized by the presence of centrally located nuclei) 1 week after injury (left panels). Scale bars = 50 μm. (B): Histograms showing the distribution of the regenerating fiber cross-sectional areas (fiber number/cross-section) in different size classes (cross-sectional area), 1 week (middle panels) and 2 weeks (right panels) after injury. The application of caspase inhibitor promoted regeneration and totally inhibited the negative effects of TNF, as shown by the increased number and cross-sectional area of the regenerating fibers in the presence of Z-VAD. Abbreviations: CTR, control; TNF, tumor necrosis factor-α.

PW1 Regulates Caspase Activation and Muscle Regeneration

We observed that PW1 is restricted to interstitial and sublaminal cells or nuclei in adult muscles (Fig. 6A) and confirmed that PW1 expression is upregulated during muscle regeneration [28]. Immunofluorescence analyses revealed a striking colocalization of PW1 expression and caspase activation in vivo. To test whether PW1 plays a role during muscle regeneration, we delivered a GFP-dominant-negative PW1 fusion protein (ΔPW1-GFP) [38], or a Snap-GFP vector as a control, to regenerating muscle by electroporation-mediated gene transfer on day 4 following injury. Gene delivery targeted interstitial cells, as well as small regenerating fibers and nonregenerating fibers (described in supplemental online Fig. 1). We noted that ΔPW1-GFP localized exclusively in the nucleus (like endogenous PW1 in vivo), whereas Snap-GFP showed a broader localization (Fig. 6B). Regenerating fibers were identified by the presence of centrally located nuclei (Fig. 6B). We measured the cross-sectional area of regenerating, transfected fibers at 2 weeks of regeneration. TNF significantly reduced the regenerating fiber cross-sectional area with respect to PBS-treated controls in Snap-GFP-expressing muscle fibers (Fig. 6C). In addition, we found that ΔPW1-GFP expression was able to overcome the TNF-mediated reduction of regenerating fiber cross-sectional area (Fig. 6C).

Figure Figure 6..

PW1 regulates caspase activation and muscle regeneration. (A): Adult and regenerating muscle cryosections were stained for PW1 (green) and laminin (red). Scale bars = 100 μm. PW1 expression was upregulated in regenerating muscle. (B): Regenerating tibialis anterior was injected with phosphate-buffered saline (PBS) (CTR) or TNF 2 and 4 days after injury. Gene delivery by electroporation of an expression vector for ΔPW1-GFP (ΔPW1) or Snap-GFP was performed 4.5 days after injury; samples were analyzed at 2 weeks of regeneration. Muscle cryosections were stained for laminin (red) and nuclei (blue). Green indicates exogenous DNA expression. Gene expression was delivered prevalently to myofibers at 2 weeks of regeneration, with ΔPW1-GFP showing a nuclear localization, whereas Snap-GFP appeared diffused within the fiber. Scale bars = 100 μm. Arrowheads point to regenerating fibers expressing exogenous DNA, which were exclusively considered for the following quantitative analysis. (C): CTR transfection with the Snap-GFP construct (MOCK, black column) and ΔPW1-GFP (ΔPW1, hatched columns) transfection were combined with PBS or TNF (TNF, gray column) injections. Histogram shows the cross-sectional area of regenerating fiber expressing exogenous DNA (size of regenerating fibers). Values are the mean ± SEM obtained from at least four independent experiments. Expression of the dominant-negative form of PW1 blocked the negative effects of TNF on regenerating fiber size. Student's t test: *, p < .02 versus MOCK; **, p < .001 versus MOCK TNF. (D): Samples treated as described above were analyzed 1 day after the electroporation by staining for caspase activity. Three-color, merged images are shown: yellow arrows indicate colocalization of caspase activity (red), exogenous DNA expression (green), and nuclei (blue). Caspase activity was never localized in ΔPW1-expressing cells, in either the absence or the presence of TNF. Scale bars = 200 μm. (E): Mechanistic model for TNF effects on muscle regeneration. In response to TNF, PW1 mediates caspase activation in muscle interstitial cells expressing PW1, Sca-1, and CD34. Caspase activation hampers the contribution of these cells to the onset of regenerating fibers. The negative effects of TNF on muscle regeneration can be counteracted by either gene delivery of a dominant-negative form of PW1 (ΔPW1) or a pharmacological form based on a caspase inhibitor (Z-VAD). Abbreviations: CTR, control; GFP, green fluorescent protein; TNF, tumor necrosis factor-α.

To investigate whether PW1 affects caspase activity in vivo, we performed in situ staining for caspase activity 5.5 days following injury. The electroporation procedure did not alter caspase activity either in the absence or presence of TNF (Fig. 6D, upper panels). In contrast, we did not detect ΔPW1-GFP-expressing cells with active caspases in the presence or absence of TNF (Fig. 6D, lower panels). These results demonstrate that PW1 is necessary for caspase activation and mediates fiber size reduction during regeneration in response to TNF.


TNF plays a key role in the early inflammatory response following muscle damage, which can only be examined in vivo [46], whereas the effects of TNF have been examined primarily using in vitro models [27, 47]. TNF is a mitogen for muscle satellite cells [36], although it is still controversial whether TNF receptor expression is necessary for proper muscle regeneration [32, 35]. We have recently reported that exposure to chronic low levels of circulating TNF inhibits muscle regeneration [19]. Here, we demonstrate that exogenous TNF delivered at specific time points following muscle injury inhibits regeneration. TNF-treated muscle shows an overall reduction of regenerating fibers and a delayed myofiber maturation. The degenerative process appears to be unaffected by TNF exposure. This is particularly relevant in the light of a recent report demonstrating that TNF exacerbates muscle damage through neutrophil recruitment in dystrophic mice [48]. Damaged myofibers release signals that are important not only for resident stem cell activation but also for homing of circulating stem cells that may contribute to muscle repair [9].

We have previously reported that caspases are involved in TNF-mediated inhibition of muscle differentiation in vitro [27]. The evidence that caspase activity is undetectable in uninjured adult muscle and is upregulated during muscle regeneration supports the hypothesis that caspases play a role in muscle regeneration. Upon TNF treatment, caspase activity is further increased in regenerating muscle. Caspase activity is involved in the TNF-dependent impairment of regeneration, as shown by the Z-VAD-mediated rescue of TNF effects. The increased number of small regenerating fibers treated with TNF and Z-VAD at 2 weeks can be explained by a selective block of the TNF-mediated pathways that hamper regeneration, not affecting pathways that, instead, promote the onset of regenerating fibers. This would ultimately result in enhanced regeneration. It has recently been reported that inhibition of caspase-9 activation by Bax inactivation protects against myocardial ischemia-reperfusion injury [49]. We have shown that primary cultures of myogenic cells differentiate regardless of caspase-3 expression [27]; however, a significant increase in the activity of caspase-1, 3, 8, and 9 in skeletal muscle has been reported in muscle wasting [50], where caspases play a critical role in initiating muscle protein degradation [51]. The caspase activity detected in muscle during cachexia could be involved in altering protein metabolism, as well as in controlling cell fate. The contribution of apoptosis to muscle wasting is controversial [50, 52]. On the basis of our data, the reported caspase activity in models of cachexia may underlie impaired regeneration. A regulatory role for caspase during nonapoptotic processes has previously been reported. Specifically, caspase-1 controls inflammation in various animal models [50], and we have demonstrated that proapoptotic caspases govern differentiation in the absence of apoptosis [27]. The TNF-dependent increase in caspase activity over PBS-treated regenerating muscle (approximately 30%) is 1 order of magnitude smaller than what was observed in muscle cells induced to undergo apoptosis, suggesting that caspases play a nonapoptotic role at lower activity levels. Although we detected caspase-3 activity in necrotic fibers (i.e., EBD+ and invaded by cells), we found that the vast majority of interstitial cells with activated caspases did not activate caspase-3 (Figs. 3, 4). We did not detect overt apoptosis coupled with caspase activity in muscle using a variety of detection assays. We therefore propose that caspase activation regulates skeletal muscle regeneration by nonapoptotic pathways.

We observe that caspase-positive cells are interstitial (i.e., outside the basal lamina) and express the stem cell markers Sca-1 and CD34. Expression of Sca-1 characterizes muscle-derived stem cells (including mesoangioblasts), as well as endothelial cells, and is downregulated in satellite cells [53, 54]. In agreement with these observations, ectopic Sca-1 expression inhibits myoblast proliferation and differentiation [55]. CD34, another hematopoietic stem cell and endothelial marker, is also expressed in muscle-derived stem cells [56] and in satellite cells [57] and correlates with the maintenance of the quiescent state [53, 56]. CD45+/CD34+/Sca-1 cells are common ancestors to myogenic and endothelial cells [58]. In our experimental model, we observed that caspase-positive cells are not associated with the microvasculature or nerves (Figs. 3, 4). On the basis of these observations, we propose that caspase activity regulates the fate of stem cells in regenerating muscle and impairs their participation in muscle regeneration. The significance of the contribution of muscle progenitor cells other than satellite cells to muscle regeneration is debated [59]. Our data support a model whereby TNF-mediated caspase activation in a subset of interstitial cells regulates muscle regeneration.

In skeletal muscle cell cultures, PW1 is a key mediator of caspase activation in response to TNF [27]. PW1 is expressed in the developing skeletal muscle during embryogenesis and in most of the myogenic cell lines and primary cultures [42]. PW1 expression is absent in MyoD-mutant primary myoblasts [60], suggesting a link between myogenic potential and PW1 expression. In the adult, PW1 is expressed in stem cells with myogenic potential, such as the mesoangioblasts (G. Marazzi and G. Cossu, personal communication), and in still-uncharacterized cells located both within and outside the basal lamina in skeletal muscle. PW1 contributes to the progression of cachexia, highlighting the importance of PW1 in maintaining muscle homeostasis [28]. The evidence that cells with caspase activity also express PW1 suggests that PW1 is a modulator of the caspase pathway as well as of muscle regeneration in vivo (Fig. 4). Interference with PW1 activity by ΔPW1 abolished caspase activation and rescued regenerating myofiber maturation upon TNF treatment (Fig. 6). These data demonstrate that PW1 mediates caspase activation in response to TNF, leading to hampered muscle regeneration. We demonstrated that primary myoblasts obtained from ΔPW1 transgenic mice are resistant to the inhibitory effects of TNF upon myogenesis [38]. Furthermore, ΔPW1-expressing muscle fibers are resistant to tumor-induced cachexia in vivo [28]. These observations highlight the pivotal role PW1 plays in controlling muscle homeostasis by affecting stem cell function.


Our data highlight a novel mechanism whereby TNF inhibits muscle regeneration, that is, by upregulating caspase activity, in the absence of apoptosis, in a subpopulation of interstitial stem cells. PW1 is necessary for TNF-dependent caspase activation in these cells. These results provide new insights into the molecular regulation of muscle regeneration and reveal an important role for interstitial stem cells in muscle homeostasis.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


The expert technical assistance of Carla Ramina is gratefully acknowledged. Confocal images were obtained in the Department Confocal Facility, funded by a Grandi attrezzature 2003 grant of Sapienza University (to M.M.) This work was funded by Association Française contre les Myopathies (Project 11788-SR-2006 and 12668-2007); Agenzia Spaziale Italiana and Progetti di Ateneo of Sapienza University, Rome; Ministero dell′ Università e della Ricerca-Rientro dei cervelli 2003 (to D.C.); the National Cancer Institute (P01-CA80058-04, project 4); and the Muscular Dystrophy Association (to D.S.).