Methodological issues limit interpretation of negative effects of satellite cell depletion on adult muscle hypertrophy

We are writing regarding the recent publication in Development entitled ‘Satellite cell depletion prevents fiber hypertrophy in skeletal muscle’ (Egner et al., 2016). Egner et al. claim to have ‘essentially repeated’ our work (McCarthy et al., 2011); however, we think that methodological differences between the two studies make it impossible to compare the results directly. More importantly, technical problems with their analyses and lack of controls make it impossible to attribute the lack of growth that they observed to the loss of satellite cells (SCs). We have not studied hypertrophy in the extensor digitorum longus (EDL) muscle, as the mechanics and function of the tibialis anterior (TA)/EDL do not support the relevance of a synergist ablation-mechanical overload model. The EDL is only activated during the ‘swing phase’ of walking, designed for velocity at the expense of force. It operates essentially unloaded, undergoing only concentric contractions to lift the weight of the foot during ambulation (Carlson-Kuhta et al., 1998), so removing the TA should have little effect on the size of EDL (Lieber and Ward, 2011). On the other hand, the gastrocnemius, soleus and plantaris muscles are designed for force production and, following removal of the gastrocnemius muscle, the plantaris experiences significant concentric and eccentric loading during walking, the latter being crucial for hypertrophy (Booth and Thomason, 1991; Gregor et al., 2006). Thus, we will limit our comments to the results reported from the plantaris following synergist ablation, a well-accepted hypertrophic model.

We have performed synergist ablation surgery on hundreds of Pax7-DTA mice treated with tamoxifen to deplete SCs or vehicle, starting when they are at least 4 months old, and consistently find no difference in the plantaris growth response, as defined by muscle mass and fiber cross-sectional area (CSA), in the first 2 weeks. We do not include tamoxifen-treated mice that are less than 90% depleted in our analyses, and we do not observe significant proliferation of the few remaining SCs at 2 weeks, or myonuclear accretion into fibers. By contrast, Egner et al. report an average of only 76% depletion, and a 420% increase in SCs in their SC-depleted, overloaded (SC−OL+) plantaris; this level of depletion should have limited effect on growth. However, the fact that they detect only about five SCs in the entire cross-section of the control SC+OL− plantaris (figure 1B in Egner et al., 2016), makes the entire quantification questionable. That SCs are not effectively depleted is most clearly demonstrated by the observation that markers of muscle regeneration [embryonic myosin heavy chain (eMyHC)-positive and centrally nucleated fibers] are significantly higher in SC− compared with SC+ plantaris with overload (figure 2B-D in Egner et al., 2016). We reported that SC-depleted muscle showed significantly fewer eMyHC+ fibers compared with vehicle-treated OL+, as expected, and severely impaired regeneration capacity after BaCl₂ injection. The higher regenerative response in plantaris from tamoxifen- compared with vehicle-treated mice in the Egner study suggests that there is actually more SC activity in response to overload in SC− mice, making the lack of growth even more inexplicable. Without the appropriate control – the parental Pax7-CreER strain tamoxifen treated (Fry et al., 2014; Jackson et al., 2015) and subjected to overload (Fry et al., 2014) – no conclusions regarding the specific effect of SC depletion, as opposed to tamoxifen toxicity, on growth can be drawn from the Egner study.

We think a much more reasonable explanation for the lack of growth reported by Egner et al. is that their tamoxifen-treated mice were unhealthy, as evidenced by lower body weight. They report that tamoxifen-treated mice had a 13% lower body weight than vehicle-treated mice. The tamoxifen-treated mice in our 2011 study showed 4% lower body weight compared with vehicle-treated mice (vehicle, 24.5±2.7 g; tamoxifen, 23.5±2.5 g), which was not statistically different. This is also the case for tamoxifen- compared with vehicle-treated mice from our various studies in both male (Fry et al., 2014, 2015; McCarthy et al., 2011) and female (Jackson et al., 2015, 2012) mice. As stated above, the parental Pax7-CreER strain must be tamoxifen treated to rule out tamoxifen toxicity accounting for the lack of growth in the Egner study.

In addition to being unhealthy, we posit that the difference in response to tamoxifen observed between the studies is due to the fact that the mice in the Egner study were less than 4 months old at the time of tamoxifen treatment, and thus were still growing. Femur and humerus growth increases rapidly from 4 to 12 weeks, after which time longitudinal bone growth appears to cease (Ferguson et al., 2003). We only use mice that are 4 months of age or older to be sure we are past the period of long bone growth, and therefore muscle growth, as we expect different SC requirements for hypertrophic growth in full grown adult compared with growing mice. Figure 7D in Egner et al., 2016 shows a larger myonuclear domain (MND) in SC−OL− compared with SC+OL− mice, indicating that control muscles were in fact growing, without addition of myonuclei, which is counter to the premise of the paper.

There are other technical differences contributing to the different results. As stated above, we have performed surgery on hundreds of mice and are very adept at the procedure, and do not experience the various problems reported by Egner et al. We are very careful not to disturb the blood supply or the nerve to the plantaris and do not observe the ‘rupture damage’ described by Egner et al., where they eliminated 5 of 30 ablated mice because the plantaris was ‘almost completely degenerated’. We agree with Egner's conclusion that their muscle weights are unreliable due to their inability to dissect the plantaris cleanly, resulting in overloaded plantaris muscle weights ranging from 20 to 45 mg. Given the small number of fibers quantified for CSA in their study, we also believe those values are unrepresentative. Using a grid, they only measured an average of 160 fibers (range 101-205) per cross-section, whereas we count essentially all fibers in the section (>1000). Measuring CSA with a grid could favor measuring larger fibers, because those fibers will have a higher chance of falling on a grid intersection. They report an unusually high number of very large fibers (>3000) which might be due to the fact that they counted relatively more of them, thereby skewing their results. Our results agree with those of Egner that hypertrophy is normally associated with myonuclear accretion, which is abolished in response to overload following tamoxifen treatment.

Although we do not see the extensive degeneration reported by Egner et al., we do agree that synergist ablation subjects the plantaris to significant mechanical overload that results in a regenerative response, which we documented in our 2011 Development paper (McCarthy et al., 2011). In our study, we saw a 30% increase in eMyHC+ fibers and centrally nucleated fibers in response to overload in the vehicle-treated plantaris. However, we saw no relationship between those two markers with each other or with fiber CSA, which also appears to be the case in the Egner et al. report (figure 2 in Egner et al., 2016), so the relationship of these markers to ‘degeneration’ is not clear. That is, as shown in the McCarthy et al. (2011) paper, some large fibers expressed eMyHC, and some large fibers were centrally nucleated, but not eMyHC+, which is why we chose to count all fibers in the entire cross-section in every muscle. Resident myonuclei might activate eMyHC expression as a repair response, independently of SCs, and eMyHC+ fibers could grow and contribute to the overall hypertrophic response, so we do not think it is appropriate to eliminate them from the analyses. Thus, as mentioned above, counting all fibers compared with counting 160 randomly selected fibers might contribute to the different results due to the heterogeneous and regional nature of the plantaris.

We reported in our 2011 paper that vehicle-treated overloaded mice had more very small, eMyHC+ fibers (<300 μm²), which are probably formed de novo as a result of SC-dependent regeneration, and those were quantified separately (McCarthy et al., 2011). However, they made up less than 1% of the total muscle area and did not contribute substantially to overall muscle mass. Re-analysis of all other fibers in bins in bar graph form with error bars (Fig. 1) shows that there is no significant difference in fiber size distribution between vehicle and tamoxifen groups.

We agree that on average, the increase in CSA that we observe at 2 weeks is relatively modest. We think that this overload model results in such a large increase in fiber CSA that by 2 weeks many of the largest fibers split. There is precedence for this in the literature in rats (Ho et al., 1980), mice (Vaughan and Goldspink, 1979) and humans (Larsson and Tesch, 1986; Tesch and Larsson, 1982) in response to a robust hypertrophic stimulus. As a result, although there is a rightward shift in fiber size distribution, average fiber CSA is only modestly larger, but there are more fibers and the muscle is significantly hypertrophied, evidenced by increased mass. (This might not have occurred in the Egner et al. study because they performed unilateral surgery and the mice may have favored the sham-operated limb.) Although we admit this is supraphysiological, the important point is that the response of SC-depleted muscle compared with muscle with its full complement of SCs is not different. Plantaris muscles with and without SCs grow.