In 1961, Dr. Alexander Mauro first reported the existence of satellite cells in frog muscle [7]. Subsequently, satellite cells were discovered in mammalian skeletal muscle. The name is derived from its specific location. One side of a satellite cell is directly attached to a myofiber via cell-cell adhesion molecules such as M-cadherin, and the other side is attached to the basal lamina; therefore, satellite cells are located between the myofiber and basal lamina. Other types of cells are never found in this position. Before the satellite cell-expressing molecules were identified, electron microscopy was the only means to identify and observe satellite cells. However, identifying the proteins expressed in satellite cells and establishing methodologies (FACS analysis, single-myofiber culture, and so on) allow us to observe satellite cells more easily [11, 37–39].

A stem cell is defined as a cell having both differentiation and self-renewal potentials. In 2008, Sacco et al. transplanted a single satellite cell and demonstrated that the satellite cell has both potentials [6]. Although there are many reports indicating the existence of muscle stem cells besides satellite cells, as described above, many recent studies leave no doubt that the satellite cell is a physiological stem cell having both differentiation and self-renewal potentials [5, 25, 26, 40].

Like other stem cells, satellite cells are maintained in an undifferentiated and quiescent state [41]. This state is considered essential for maintaining a satellite cell pool. In general, each tissue stem cell can be found in a specific location, called its niche, composed of certain types of cellular and extracellular matrix proteins. The satellite cell locates beneath the basal lamina and attaches to a myofiber; therefore, both of them together are thought to create the niche for satellite cells. In addition, Christov et al. reported that satellite cells exist in close proximity to endothelial cells [42]. Therefore, endothelial cells might also be a niche cell for maintaining satellite cells, like hematopoietic, neural, and spermatogonic stem cells [42–44].

Recent studies are starting to reveal the molecular mechanism for keeping satellite cells in an undifferentiated and quiescent state [31]. A report by Seale and colleagues was the first that showed a decreased number of satellite cells in genetically modified animals. They found expression of Pax7 in myogenic cells and discovered the loss of the satellite cell pool in Pax7 knockout mice [12]. Because Pax7 is specifically expressed in satellite cells in skeletal muscle, Pax7 has been the best molecule for identifying satellite cells in skeletal muscle. Importantly, the expression of Pax7 has been also confirmed in across species (human, rodent, avian). Although Lepper et al. had reported that the role of Pax7 in adult satellite cells was dispensable [45], two independent groups demonstrated that it was essential even in adult satellite cells [46, 47]. The discrepancy is likely due to the construct of Pax7-floxed mice [46].

What is the role of Pax7 in satellite cells? Taking the abovementioned studies together, Pax7 is involved in both the generation and maintenance of satellite cells. Originally Seale et al. proposed the specification of muscle satellite cells by Pax7 [48]. Relaix et al. showed the antiapoptotic roles of Pax7 in satellite cells [49]. In addition to those roles, McKinnell et al. indicated that Pax7 directly regulates Myf5 expression by association with the Wdr5-Ash2L-MLL2 histone methyltransferase complex that causes methylation of histone H3 lysine 4 [50]. The phenotype of Myf5-null mice [51] is much milder than that of Pax7-null mice; therefore, other targets of Pax7 might be also important for the specification and/or survival of satellite cells. Significantly, Pax7 was used for generating satellite cell-like cells from human iPS cells [52], demonstrating that Pax7 is one of the important discoveries in satellite cell biology that promotes the achievement of cell therapy.

Another molecule essential for regulating satellite cells in adult skeletal muscle is the Notch signaling pathway, which is an evolutionarily conserved intracellular signaling system. When Notch signaling is activated by its ligand (Dll1, Dll3, Dll4, Jag1, or Jag2), the intracellular domain of Notch is cleaved by a protease like γ-secretase and transported to the nucleus [53]. Then, the target genes are transcribed through interaction with Rbpj. Rbpj-dependent Notch signaling is known as canonical Notch signaling, and the representative target genes of canonical Notch signaling are the Hes (Hairy and enhancer of split) and Hesr (Hes related, also known as Hey/Herp/Hrt/Ghf/Gridlock) families of bHLH transcriptional repressor genes [54]. Using conditional ablation of Rbpj in Pax7-CreER^T2 mice, two independent groups have indicated the essential roles of Rbpj for maintaining adult satellite cells in an undifferentiated and quiescent state [55, 56]. Intriguingly, when satellite cells in adult skeletal muscle lack Rbpj, they start to show precocious myogenic differentiation following protein expressions of MyoD and myogenin. They eventually fuse with myofibers and the satellite cell pool is diminished. When C2C12 cells or primary myoblasts are stimulated by Dll1 (Delta-like 1) or Dll4, Hesr1 and Hesr3 are upregulated [10, 57]. In addition, we observed premature differentiation and loss of satellite cell pools in Hesr1/3-double knockout mice [10]. Because each single knockout mouse did not show any significant defect in skeletal muscle including satellite cells, these results suggest that Hesr1 and Hesr3 have a redundant effect in maintaining satellite cells downstream of the canonical Notch signal [32]. Like Pax7, Notch signaling might contribute to the realization of cell therapy. Parker et al. compared the in vivo reconstitution potential of transplanted cells stimulated with Notch ligand and control cells and found that the former cells can engraft in vivo more actively than the latter [58]. The mechanism of Pax7 expression induction and the type of ligand/receptor for canonical Notch signaling in adult satellite cells have not been fully determined. Therefore, elucidation of these regulatory mechanisms will lead to the identification of the in vitro maintenance conditions for preparing cells for regenerative medicine.

Except for Pax7 and Notch-related genes knockout mice, decreased numbers of satellite cells are observed in some genetic knockout mice [59–63]. Most gene knockout mice show a loss of satellite cell pools at a relatively young age [31]. However, the phenotypes of Sprouty 1 (an inhibitor of receptor tyrosine kinase signaling) knockout mice are quite different. The loss of Sprouty 1 did not affect the number of satellite cells in uninjured muscle of young mice, but in aged mice (22 months), the satellite cell pool was diminished by half [64]. They also showed that this diminishment of satellite cells resulted from accelerated fibroblast growth factor (FGF) signaling. Therefore, maintenance genes can be divided to two categories: one is a gene for autonomous regulation, and the other is a gene affected by a change of environment. The elucidation of both mechanisms is important for the development of an ideal cell source for muscular disorders.

In addition to coding RNA, noncoding RNA is also essential for maintaining the satellite cell pool in uninjured muscle. Dicer is an enzyme for generating microRNA. When Dicer is conditionally and specifically deleted in adult satellite cells, they start to express proliferative markers and eventually undergo apoptosis [65]. Among quiescent stage-specific microRNA, Cheung et al. demonstrated that mir-489 is critical for a quiescent state of satellite cells through suppressing its target gene, Dek. Notably, mir-489 is located in the intron of the calcitonin receptor expressed specifically in quiescent satellite cells in skeletal muscle [9]. The expressions of both mir-489 and calcitonin receptor are downregulated after activation of satellite cells. During regeneration, re-expression of calcitonin receptors is restricted to the area where regeneration is mostly completed [9]. Although the physiological role of calcitonin receptor is unknown, the common expression mechanism of both mir-489 and calcitonin receptor might be extremely important to unravel the mechanism for establishment of quiescent satellite cells during developmental and regenerative processes.

As described above, adult satellite cells are maintained in a dormant and undifferentiated state, but when skeletal muscle is injured, these cells start to enter an activated state. Before proliferating, they begin to express MyoD protein and then enter the cell cycle. Proliferating satellite cells that express both Pax7 and MyoD are called myoblasts. When they become more committed to myogenic differentiation, they start to express myogenin and lose Pax7. Myogenin-expressing cells that can fuse with each other or myotubes and make multinucleated myotubes are called myocytes. After myotubes mature, myofibers are innervated and rebuilt. During the regeneration processes, satellite cells produce self-renewed satellite cells that are indispensable due to their repeated regenerative potentials. The critical mechanism of satellite cell self-renewal is unknown, but two basic models have been proposed. Using cultured single myofibers, Zammit et al. indicated that almost all satellite cells become Pax7⁺MyoD⁺ cells, and then a part of them return to a Pax7⁺MyoD⁻ state. On the other hand, Kuang et al. found that satellite cells can be divided into two populations based on YFP expression driven by Myf5-Cre and that the Pax7⁺Myf5⁻(YFP⁻) population has more stem cell-like characteristics than the Pax7⁺Myf5⁺(YFP⁺) population. Intriguingly, when Pax7⁺Myf5⁻ cells divide on a plane, most cells generated are Pax7⁺Myf5⁻ cells. On the other hand, when Pax7⁺Myf5⁻ cells do apical-basal division, both Pax7⁺Myf5⁻ and Pax7⁺Myf5⁺ cells (asymmetric division) are generated. However, these cells were observed in regenerating myotubes/myofibers having central nuclei. Furthermore, Wnt7s promote the number of Pax7⁺Myf5⁻ cells [66], and the effect of Wnt7a was accelerated by fibronectin, which is specifically expressed during myotube/myofiber formation [67]. Therefore, there is a possibility that symmetric and asymmetric division in Pax7⁺Myf5⁻ might occur after several cell divisions of Pax7⁺MyoD⁺ cells. The self-renewal mechanism of satellite cells is one of the important issues to be explained.

Fibrosis is a hallmark of pathological conditions with impaired regeneration in many tissues. In skeletal muscle, accumulation of adipocytes is also observed in progressed muscular dystrophy patients. In order to identify the origin of the cell populations involved in fat accumulation and fibrosis in skeletal muscle, we performed comprehensive FACS-based cell analyses. Although satellite cells were considered the original source of fibrosis and adipogenesis, we found that a non-satellite cell population exclusively contributed to both fat accumulation and fibrosis in skeletal muscle, and this cell population specifically expressed PDGFRα [3, 4]. Besides the potentials to produce adipocytes and fibrogenic cells, they can also differentiate into osteogenic cells [3, 68]. Therefore, we called these cells mesenchymal progenitors.

Our study clearly demonstrated that only PDGFRα⁺ mesenchymal progenitors differentiate into adipocytes and form ectopic fat tissue in skeletal muscle [3]. This conflicts with the adipogenic potentials of satellite cells proposed by experiments using a myogenic cell line and single isolated myofiber. Myogenic cell lines are extremely useful for investigation of many aspects of myogenic differentiation, but they do not reflect some aspects of characteristics of in vivo satellite cells. Isolated single myofibers are also widely used for studies of satellite cell biology, but we have to pay attention to the risk of contamination by non-satellite cells (including mesenchymal progenitors) attached to the myofiber. Satellite cells were also proposed as a potent cellular source of collagen-producing cells by a study using a myogenic cell line [69]. In fact, fibrosis-related genes are detected even in highly purified primary myogenic cells, but their expression levels in mesenchymal progenitors are much higher than those of myoblasts, and myogenic cells do not contribute to or differentiate into collagen-producing cells in the dystrophic condition [4]. Brack et al. showed that satellite cells in aged mice transdifferentiate into fibroblastic cells in response to canonical Wnt signaling [28]. Therefore, we cannot exclude the potential of satellite cells to become fibroblastic cells. However, given their highly fibrogenic nature, it seems highly plausible that mesenchymal progenitors are the best candidate for fibrosis-initiating cells in skeletal muscle. In addition, we could identify a human counterpart to these progenitors by the expression of PDGFRα [34, 70]. Therefore, their pathological relevance to muscular disorders is further convinced. Taken together, the mesenchymal progenitor is the most potent origin of adipogenesis and fibrosis in skeletal muscle.

Mesenchymal progenitors reside in the muscle interstitium, and thus, their location is also distinct from that of satellite cells. Mesenchymal progenitors are more frequently observed in the perimysium than in the endomysium, particularly in the perivascular space, but they are distinct from pericytes because they locate outside the vessel wall and basement membrane of capillaries [3]. The definition of a pericyte is based exclusively on its anatomic location, where it is bounded by a common basement membrane with endothelial cells [71, 72]; there is no specific pericyte marker. Mesenchymal progenitors account for about 5–10 % of whole mononucleated cells isolated from skeletal muscle [3], but skeletal muscle is known to have an extremely low percentage of pericytes. In skeletal muscle, there is only one pericyte per 100 endothelial cells [71, 73]. This is quite a contrast to pericyte-rich tissues such as the retina or central nervous system, where the pericyte to endothelial cell ratio is 1:1 [71, 73]. Because endothelial cells account for 60–70 % of total mononuclear cells in skeletal muscle, as described earlier, pericytes should constitute only 0.7 % of all mononuclear cells at most. Although pericytes have been reported to possess myogenic differentiative potential [74], mesenchymal progenitors cannot differentiate into myogenic cells. Where do mesenchymal progenitors arise developmentally? We do not know, but we are sure that they do not originate in bone marrow and that they reside in skeletal muscle before the beginning of inflammation.

The morphology of mesenchymal progenitors is similar to that of fibroblasts, and there is no marker that distinguishes fibroblasts and mesenchymal progenitors. Unlike fibroblasts, mesenchymal progenitors have the potential to differentiate into mesenchymal lineage cells including adipocytes, chondrocytes, and osteoblasts. Therefore, the ability to differentiate is the only way to characterize mesenchymal progenitors. Sudo et al. showed that primary fibroblasts derived from various human tissues can differentiate into at least one mesenchymal lineage cell [75], and mesenchymal progenitors and so-called fibroblasts have much more similarity than previously recognized [76]. Therefore, many tissues seem to have their own tissue-resident mesenchymal progenitors which have been considered as fibroblasts.