To our knowledge, this is the first report testing the myogenic capacity of MDSCs isolated from transgenic mice with inactivation of the myostatin gene, in comparison to the WT MDSC, both in vitro and in the injured muscle of the aged mdx mice in vivo. Our main findings were (a) in contrast to WT MDSCs, Mst KO MDSCs were unable to form myotubes in vitro, although no major differences were found between both MDSC cultures in terms of morphology, replication rates, expression of most members of a subset of key embryonic-like stem cell and other markers, and nonmyogenic multilineage differentiation; (b) however, a fundamental difference is that the expression of key genes in myogenesis seen in WT MDSCs such as actc1, acta1, and myoD, was virtually obliterated in Mst KO; (c) surprisingly, both types of MDSCs were refractory in vitro to the modulation or induction of myotube formation by well-known regulators of this process, or of myofiber number in vivo, such as demethylating agents, myostatin inhibition or overexpression, or follistatin, although myostatin receptors are expressed in MDSC cultures; (d) the myofiber regeneration and anti-lipofibrotic capacities of WT MDSCs were evident even in the environment of a severely injured mdx gastrocnemius at an age at which lipofibrotic degeneration is considerable; (e) in turn, these capacities, blocked in cell culture, were recovered in Mst KO MDSCs when they were implanted in the injured mdx aged-muscle setting, even if not at the level expected from the supposed paracrine effects triggered in the MDSCs by the absence of myostatin.
It should be noted that although notexin-induced injury is not clinically relevant for DMD, it is experimentally convenient by stimulating cell engraftment on implantation and also inducing more lipofibrotic degeneration both in mdx and Mst KO mice [56, 57], thus providing an adequate environment for testing the MDSC-repair effects. The high variability in the repair response that is often associated with notexin injection was not observed in the current work.
The WT MDSC used here as control, fulfill all the criteria that have been extensively defined as potential tools for skeletal muscle, cardiac, and osteogenic repair on implantation into the target organs [29, 34]. In the current work, MDSCs were isolated as the pP6 fraction by using a modification of the extensively validated preplating procedure on collagen-coated flasks and Sca1 selection, and shown to have the expected morphology, rapid replication for at least 50 passages, express MDSC markers such as Sca1, CD44, and CD34, and the stem cell gene Oct 4, and the ability to differentiate in vitro into multiple cell lineages. The latter capability includes a robust formation of multinucleated and branched myotubes that is assumed to translate in vivo into their ability to donate their nuclei to injured skeletal myofibers or most likely to stimulate paracrinely their regeneration through paracrine trophic effects [32–34]. This is evidenced by a much higher number of centrally located nuclei, and even some central location of the DAPI-labeled implanted nuclei. In previous studies, we showed that WT MDSC generate at least smooth muscle and epithelial cells when implanted into urogenital tissues [27, 28], adding to the extensive demonstration of their stem cell nature [7, 12, 26, 58] related to their putative origin as myoendothelial stem cells in the muscle and other tissues .
Another novel finding here is that WT MDSCs have some embryonic-like stem cell features, mainly the expression of nuclear Oct 4 A, myc, LIF, and other embryonic stem cell genes. Oct 4 is a key not only for embryonic stem cell programming, but also for iPS generation, where it can act virtually by itself . Our MDSC cultures contain some tiny rounded cells similar to the very small embryonic-like stem cells (VSELs) described in many adult organs , and other larger ones.
An important finding is the unexpected observation that myotube formation by the WT MDSCs in vitro is refractory to modulation by agents that are well known to affect this process, or skeletal muscle mass in vivo. The fact that myotube formation by WT MDSCs was not influenced by (a) demethylating agents like azacytidine that stimulate 'stemness" in cell lines ; (b) downregulation or overexpression of myostatin, despite the detectable expression of its receptor (ActIIb); (c) counteracting myostatin activity by the respective antibodies or follistatin, that in vivo stimulate myofiber growth [17, 19, 20]; poses questions related to the role of MDSCs during normal myogenesis. A study showing that myostatin stimulated fibroblast proliferation in vitro and induced its differentiation into myofibroblasts, while increasing TGF-β1 expression in C2C12 myoblasts, did not examine MDSC differentiation . The claim of a small inhibitory effect of myostatin on the fusion index in MDSCs  may indicate less fusion efficiency but might not entirely reflect the actual effects on the number and size of myotubes, as determined here. This question requires further clarification in terms of the actual modulation of MDSC differentiation.
It may be speculated that satellite cells rather than MDSCs are the only myogenic progenitors during normal myofiber growth, as opposed to repair of damaged fibers . Therefore the selected in vitro conditions may not mimic the repair process, or alternatively, unknown in vivo paracrine or juxtacrine modulators may modify the response of MDSCs to the better-characterized agents tested in this work. Another possibility is that myostatin and other modulators investigated here would stimulate in vivo satellite cell replication and fusion to the adjacent myofibers to induce hypertrophy, without truly affecting MDSC differentiation or fusion.
We are unaware of any report on the isolation or characterization of MDSCs from the Mst KO. Therefore, it is also both novel and unexpected to find that these cells obtained from the same skeletal muscles as the WT MDSCs, by using identical procedures, and displaying rather similar nonmyogenic pluripotency and stem cell-marker features, are however completely unable to form myotubes in vitro. In fact, our prediction was that the Mst KO MDSCs should be more myogenic than the WT MDSCs because of the absence of the myogenic inhibitor myostatin, The fact that Mst replenishment, either as recombinant protein or as cDNA, does not counteract the unexpected myogenic blockade found in the Mst KO MDSCs, suggests speculatively that these cells have been imprinted in the embryo by the myostatin genetic inactivation through downstream pathways that have become unresponsive to the in vitro myostatin modulation that we explored here. This may involve genes in other myogenic pathways whose expression may be altered, as we observed in Mst KO MDSCs. However, validation of this assumption requires further investigation.
An interesting corollary is the activation of the in vitro-suppressed myogenesis in Mst KO MDSCs, and/or their ability to fuse with preexisting myofibers, after their implantation into the notexin-injured mdx gastrocnemius. At the age selected (10 months), this muscle experiences the considerable damage that occurs in the diaphragm much earlier [3, 4], and this is compounded by injury. It may be speculated that the restoration of myotube (myofiber) formation by Mst KO MDSCs in this setting occurs by paracrine or juxtacrine modulation, possibly of some of the key genes silenced in these cells. Estimation of their products and proof-of-function approaches may elucidate this issue. The fact that although Mst KO MDSCs are able to fuse with or differentiate into new myofibers, they do not increase the muscle-repair process in a clearly more efficient way than do WT MDSCs, may possibly result from the persistent myostatin expression in the fibers that may counteract its absence in Mst KO MDSCs. This suggests the need to block myostatin systemically in the host muscle [63, 64], not just in the implanted MDSCs, and our findings do not contradict the potential use of this approach
One of the genes that may be involved in the silencing of Mst KO MDSC myogenesis in vitro and its reactivation in vivo is the cardiac α-actin (Actc), the major striated actin in fetal skeletal muscle and in adult cardiomyocytes, but strongly downregulated in adult skeletal muscle to 5% of the total striated actin , and whose mRNA is highly expressed in the proliferating (nondifferentiating) WT MDSCs but at very low level in the Mst KO MDSCs. The same applies to the α1-actin (Acta1) mRNA, the adult protein encoding thin filaments . Because actins are so crucial for cell division, motility, cytoskeleton, and contraction, and mutations are associated with severe myopathies, it would not be surprising that their downregulation could cause the lack of myogenic commitment in vitro in Mst KO.
Similarly, the striking transcriptional downregulation of myoD, a critical early gene in skeletal myogenesis , confirmed at the protein level, and of secreted phosphoprotein 1, or osteopontin, a gene mostly involved in ossification, inflammation, and fibrosis, but postulated recently to participate in early myogenesis and skeletal muscle regeneration , may also trigger the absence of myogenic capacity in Mst KO. Interestingly, the fact that Pax 3 mRNA, upstream of MyoD in the myogenic signaling  is expressed in Mst KO MDSCs at higher levels than in WT MDSCs, suggests that the myogenic commitment of Mst KO and mdx MDSC is arrested at some point between these genes. Because a critical regulator of skeletal muscle development, Mef2a (myocyte enhancer factor 2a) , is expressed similarly in both MDSCs (as is Pax 3), albeit at very low levels, the silencing may occur at the level of the satellite cell marker, Pax 7. Therefore, it is not surprising that expression of a member of the cadherin family (cadherin-15) that is involved in later stages, such as myoblast differentiation and fusion , is so downregulated in Mst KO MDSCs.