The occurrence of hyperplasia in healthy adult skeletal muscle is widely debated, fuelled by observations of “splitting” myofibres. However, in principle, such features could be explained by fusion of a myotube to its “parent” myofibre. In our detailed examination of healthy human muscle undergoing adult regenerative myogenesis after myofibre necrosis, we find support for fusion rather than splitting, which argues against hyperplasia, in line with earlier hypotheses that observations of branching/splitting represent fusion and are a physiological process in healthy muscle.

It is important to note that our study examines regenerating myofibres, as shown by positive staining for the developmental myosins (neonatal and embryonic MHC), in healthy adult skeletal muscle. While there are some parallels between embryonic development and adult regenerative myogenesis [35], our findings may not necessarily represent other situations, such as development or muscle growth following heavy loading. This distinction is important, as data in larva show that, specifically during metamorphosis, fibre splitting is a common feature in this species [15]. However, more commonly, changes in muscle mass are investigated following heavy loading, a situation characterised by repair of segmental muscle damage or growth. In any case, changes in muscle mass are most often observed by looking at cross-sections of either whole animal muscles or muscle biopsies in humans. In a comprehensive review from 2019, Murach and colleagues describe fibre splitting following extreme loading conditions in animals [7]. Among others, they use data from Roy and Edgerton [36] to illustrate how changes in fibre pennation angle can make it difficult to assess fibre number from a tissue cross-section. While this is correct, change in fibre length is an additional critical factor, as is pointed out in a letter to the editor by Jorgenson and Hornberger [37]. Fibre lengthening is primarily seen; following eccentric resistance training [38] represents an additional challenge to counting fibres on muscle cross-sections. In this study, we show that branching seen during regeneration can further complicate this analysis. Figure 6 is a schematic representation, inspired by the illustrations by Murach et al. [7] and Jorgenson and Hornberg [37] of how changes to muscle morphology, and the angle and depth of the cross section, can influence total fibre number counts. This emphasises that under non-control situations, e.g. following training, during regeneration or in pathological states, muscle cross sections should be used with caution if the goal is to count the total number of fibres in a muscle. Furthermore, these methodological issues have likely clouded the data pertaining to hyperplasia.

Illustration of the challenges of counting myofibre number on muscle cross-sections, due to changes in pennation angle, myofibre length and myofibre branching

We set out to assess whether muscle fibre branching represents complete fibre splitting, and, with this, myofibre hyperplasia, or alternatively if the presence of branching fibres could be explained by fusion of myotubes rather than splitting. In this systematic analysis of regenerating human muscle, 30 days after injury induced by electrical stimulation, we find multiple signs of fibre splitting or branching, in line with earlier reports in powerlifters and anabolic drug users [19, 20]. By following these fibres through a series of 22 consecutive sections, we frequently observed the branch fusing with the parent myofibre again (Fig. 1), with a median branch length of 144 μm (range 24–264 μm). The same fibre thus appears split in portions of its length and “normal” in other parts. We often observed a single fibre branching into several smaller fibres, similar to observations in rodents in pathological states [4] and following strenuous exercise [10]. This has previously been described as evidence of muscle fibre hyperplasia. Thus, when examined alone, a single cross section can be difficult to interpret.

While the serial sectioning analysis confirmed branching as a regular feature at this time point of 4 weeks post injury, a continuation of this process towards complete splitting could still be a viable explanation for our observations. To investigate further, we turned to high-resolution confocal imaging for viewing single fibres longitudinally in 3 dimensions. Small myofibre branches were observed, tightly associated with parent myofibres. These branches were often positive for neonatal and in some cases embryonic myosin (Fig. 2), as well as staining positive for nestin, which has been shown to stain fibres in the regenerating or un-innervated states [32, 33, 39]. Importantly, both the neonatal myogenic marker and nestin-positive staining indicate that the branch is at an earlier stage of development in the regeneration process than the parent fibre. This is further supported by the lack of desmin + striations in these branch segments (Figs. 3 and 4). Disruption to the sarcomeric striations reported by Crameri and colleagues in the days after maximal eccentric voluntary exercise [40] is a sign of fibre damage and degeneration, while at the 4-week time point in the present study, it represents a different process — the formation of a new myofibre. Theoretically, fibre splitting would result in two, or more, similar fibres, rather than one fibre with mature sarcomeres and one without. Therefore, we find it most likely that the fibres are not splitting and dividing, but rather that the branch is in an earlier state of regeneration than the parent myofibre, where fusion between the branch and parent myofibre is an ongoing process. The chains of myonuclei extending from the point of branch fusion support this, as these nuclei likely mark the point of recent fusion between branch and parent sarcolemmas. Further substantiation for fusion can be found in the transmission electron microscopy images. Firstly, we observed myofibres with different states of sarcomere arrangement separated by non-membranous borders (Fig. 4), again supporting fusion of an immature branch with a more mature parent myofibre. Secondly, between a small myofibre (potentially a branch) and a larger closely associated myofibre, we observed membrane-associated electron-dense plaques. Electron-dense plaques are shown in Drosophila to be needed for the initial step of membrane fusion between myoblast and myotube [34, 41, 42]; however, electron-dense plaques are also observed, in kette mutants, where fibres do not fuse [42]. As such, electron-dense plaques are not indicant of fusion but can be interpreted as membranes in the phase of fusing. Taken together, we propose that the branching signs we, and others, observe can be explained by incomplete, ongoing, regeneration following fibre necrosis, and not splitting of myofibres leading to hyperplasia.

How these branches occur in the first place deserves some consideration. The appearance of split or branched myofibres has earlier been attributed to as “incomplete lateral fusion of myotubes during regeneration” [21, 22], which fits well with our understanding of how necrotic muscle fibres are replaced by new myofibres. Successful muscle regeneration requires the preservation of the myofibre basement membrane. The original basement membrane is eventually shed after providing essential scaffolding for myogenesis [16, 43]. Basement membrane formation distinguishes foetal muscle development from adult myogenesis because the basement membrane does not form until later stages of muscle development [43]. Within the original basement membrane, satellite cells are activated, proliferate, differentiate and eventually fuse with each other to form myotubes. Thus, one basement membrane scaffold will contain many myotubes, which in turn fuse with each other to form a single myofibre. With this in mind, it does not seem unreasonable that a myotube occasionally does not fuse in a synchronised manner with the other myotubes and becomes partially orphaned from the parent myofibre, forming its own sarcolemma and basement membrane. Alternatively, it is possible that the branch represents a lone myotube formed at a later stage of regeneration than the other myotubes.

While our observations support incomplete lateral myotube fusion as a possible explanation for the presence of branching myofibres in healthy regenerating muscle, alternative interpretations should be considered. The branches that are followed from the first point of branching (or fusion), through the 22 consecutive sections, do in some cases fuse again with the parent myofibre, while other branches were not observed to refuse and remained as branches with only a single point of attachment to the parent myofibre. This could be explained by the fibres not being followed far enough to detect fusion with the parent myofibre; however, it is also possible that these branches do not fuse, but rather split, for instance via myocyte grafting as suggested by Murach and colleagues [7]. A dual response could be expected if some fibres suffer only partial destruction. However, from a previous analysis in muscle biopsies taken from these same subjects 7 days following injury, we know that all fibre fragments studied were either completely normal or regenerating (alternating necrotic and regenerating zones, with macrophage infiltration) along their entire length [24], so segmental damage does not seem to be a feature of the electrical stimulation model used, in healthy humans, in the present study. Furthermore, based on the earlier regeneration state of the branch, the centralised nuclei potentially marking site of recent fusion between the branch and parent myofibre (Fig. 3), and the presence of electron dense plaques (Fig. 4) pointing towards the cells being on a path towards fusion, we find it more likely that fusion is ongoing rather than splitting. Whether some branches undergo fusion, while others split completely, could certainly be a feature of other models of muscle overload, injury or pathology, and are potentially species specific.

One of the other interesting observations from the present study was the difficulty in categorising regenerating myofibres as clear type I or type II. The fibres in the non-stimulated leg were straightforward to classify as type I or type II, whereas for the regenerating muscles, we had to create additional profiles, seven in total, most of which expressed developmental myosins which has been well documented previously supporting the notion that there is a re-expression of these during regeneration [44]. The different myosin profiles most likely represent different stages of regeneration, as in the case of the branches, developmental myosin representing an immature state as is seen in the prenatal environment [45]. In general, the MyHC I and II antibody staining was in agreement with the ATPase fibre categories. Although some fibres stained positive with both myosin type I and myosin type II antibodies, in other cases, the ATPase staining profile was inconclusive. Thus, a combination of antibody staining and ATPase staining is helpful for defining a fibre as primarily expressing either type I or type II myosin, under regeneration conditions. The other outcome from this analysis was that the regenerating fibres were almost exclusively type II. Crameri and colleagues have previously suggested, but were not able to show conclusively, that electrical stimulation primarily targets type II fibres [46]. However, Gregory and Bickel suggest in a perspectives paper [47] that it is more likely that electrical stimulation has a stochastic/nonselective fibre-type recruitment pattern when used as a model of exercise. We speculate that the finding that 99.5% of the affected fibres being type II in nature is not a feature of the electrical stimulation model per se but rather a combination of [1] the recruitment pattern seen in the case of very high strain; in this case, electrical stimulation in combination with eccentric contractions selectively recruits large motor units, at least initially, and with this the fast type II fibres and [2] this regeneration stage, 4 weeks after the stimulated contractions. In rat soleus, a predominantly slow muscle, regenerating fibres initially express fast myosin transcripts, and then switch to slow myosin, upon innervation [48]. This supports type II myosin as the default myosin heavy chain type, while the fibre is still developing (and not yet innervated). However, it is not known if this occurs in healthy adult human muscle, and it should be noted that the overall percentage of type I fibres (47%) did not differ between the regenerating muscle and the control muscle, which further supports the regenerating fibres being type II in nature. Our observations relating to branching and fusion therefore only reflect events in type II fibres, in a mixed muscle such as the vastus lateralis.