Evolutionary loss of foot muscle during development with characteristics of atrophy and no evidence of cell death

Many species that run or leap across sparsely vegetated habitats, including horses and deer, evolved the severe reduction or complete loss of foot muscles as skeletal elements elongated and digits were lost, and yet the developmental mechanisms remain unknown. Here, we report the natural loss of foot muscles in the bipedal jerboa, Jaculus jaculus. Although adults have no muscles in their feet, newborn animals have muscles that rapidly disappear soon after birth. We were surprised to find no evidence of apoptotic or necrotic cell death during stages of peak myofiber loss, countering well-supported assumptions of developmental tissue remodeling. We instead see hallmarks of muscle atrophy, including an ordered disassembly of the sarcomere associated with upregulation of the E3 ubiquitin ligases, MuRF1 and Atrogin-1. We propose that the natural loss of muscle, which remodeled foot anatomy during evolution and development, involves cellular mechanisms that are typically associated with disease or injury.


Introduction
Muscles in the feet of birds, reptiles, and mammals were lost multiple times in the course of limb evolution, usually coinciding with the loss of associated digits and elongation of remaining skeletal elements (Hudson, 1937;Raikow, 1987;Pavaux and Lignereux, 1995;25 Botelho et al., 2014;Abdala et al., 2015;Berman, 1985;Cunningham, 1883;Souza et al., 2010). Despite its frequent occurrence, the developmental mechanisms that lead to the natural absence of adult limb muscle are not known. We focus here on a representative example of distal limb muscle loss in the bipedal three-toed jerboa (Jaculus jaculus), a small laboratory rodent model for evolutionary developmental biology, to determine if evolutionary muscle loss 30 conforms to expectations based on what was previously known about muscle cell biology.
The hindlimb architecture of the adult jerboa is strikingly similar by convergence to the more familiar hooved animals, like horses and deer, including the disproportionately elongated foot that lacks all intrinsic muscle (Berman, 1985;Cunningham, 1883). The tendons were retained and expanded in each of the anatomical positions where flexor muscles are absent 35 ( Figure 1A,B and Figure1 -figure supplement 1A,B) and serve to resist hyperextension when the terminal phalanx contacts the ground during locomotion (Lochner et al., 1980;Moore et al., 2017). The evolutionary origin of jerboa intrinsic foot muscle loss lies deep in the phylogenetic tree of Dipodoid rodents. Compared to the ancestral state, the number of intrinsic foot muscles are reduced from sixteen to six in pygmy jerboas (Stein, 1990) which diverged from the three-40 toed jerboa lineage more than 20 million years ago (Wu et al., 2012;Pisano et al., 2015).
The mechanisms of limb muscle development have been extensively studied in traditional model systems, and its degeneration has been studied after injury and during disease. Briefly, limb muscle progenitors are specified from mesodermal cells at the ventrolateral edge of the dermomyotome in somites aligned with the prospective limb. These 45 cells delaminate and migrate into the limb bud as dorsal and ventral muscle masses that proliferate and initiate a myoblast specification program (Chevallier et al., 1977;Christ et al., 1977;Hayashi and Ozawa, 1991;Murphy and Kardon, 2011). The muscle masses are then subdivided into individual muscle groups in response to cues from the developing muscle connective tissue, which is derived from limb field lateral plate mesoderm (Hayashi and Ozawa, 50 1991;Kardon, 1998;Kardon et al., 2003;Wortham, 1948). They then initiate a differentiation program, which includes cell fusion to form aligned multinucleated myofibers (Abmayr and Pavlath, 2012;Kelly and Zacks, 1969).
Each differentiated myofiber produces an assemblage of Z-body proteins, Actin filaments, and non-muscle Myosin that form premyofibrils (Ono, 2010;Rhee et al., 1994;Sanger 55 and Sanger, 2008;Sanger et al., 2002). Desmin, α -Actinin, and the Z-body portion of Titin also begin to organize (Furst et al., 1989;Sanger et al., 2002). Subsequent uncoiling of Titin increases Z-body spacing, and integration of embryonic skeletal muscle Myosin results in formation of nascent myofibrils (Ono, 2010;Sanger et al., 2010). Further maturation of the nascent myofibril into a mature myofibril involves incorporation of additional proteins that are 60 important for sarcomere structure and function, and Z-lines are aligned and properly spaced to bring sarcomeres into register (Ehler and Gautel, 2008;Sanger et al., 2010). Failure at any point of myoblast specification, migration, myofiber differentiation, or myofibril maturation compromises muscle function and manifests as muscle degenerative disease in humans (Bönnemann and Laing, 2004;Laing and Nowak, 2005;Morita et al., 2005). 65 Working backward in time from the adult jerboa phenotype, we found that two of the three flexor muscle groups differentiate as multinucleated myofibers that initiate sarcomere assembly, as in other species. However, almost all jerboa intrinsic foot muscle is lost within a few days shortly after birth. Despite the rapid and near complete loss of myofibers, we found no molecular or ultrastructural evidence of apoptotic or necrotic cell death, no accumulation of 70 autophagic vesicles, and no macrophage infiltration. Instead, we observed evidence of ordered sarcomere disassembly and upregulation of muscle-specific ubiquitin ligases, MuRF1 and Atrogin-1. Although the ultimate fate of intrinsic foot myofibers after loss of muscle identity remains unknown, these data suggest that the mechanism of myofiber loss is similar to atrophy, which is typically considered a pathological response to injury or disease. 75

Results
The absence of intrinsic foot muscle in the adult jerboa could be due to a failure of early myoblasts to migrate into and/or to differentiate in the distal limb. Alternatively, embryonic muscles may form but not persist through development to the adult. In transverse sections of 80 newborn mouse feet, immunofluorescent detection of skeletal muscle myosin heavy chain reveals each intrinsic muscle group (Figure1 -figure supplement 1G). In newborn jerboas, we observed two of the three groups of flexor muscles. While the m. lumbricales never form, the jerboa has a single m. flexor digitorum brevis and three pinnate m. interossei that are not present in adults (Figure1 -figure supplement 1H). 85 Postnatal growth of vertebrate skeletal muscle typically involves an increase in myofiber number (hyperplasia) within the first week, followed by an increase in myofiber size (hypertrophy) (Chiakulas and Pauly, 1965;Gokhin et al., 2008;White et al., 2010). In order to understand the dynamics of muscle growth and loss, we quantified the rate of myofiber hyperplasia at two-day intervals after birth of the mouse and jerboa, focusing on the 90 representative interosseous muscle that is associated with the third metatarsal ( Figure 1E,F). As expected in the mouse, we observed a steady increase in the average number of myofibers in cross section from birth to P8 ( Figure 1C). In contrast, the number of myofibers in the third interosseous of the jerboa foot rapidly declines beginning at approximately P4, and few myofibers remain by P8 ( Figure 1D). 95 It is possible that the rate of myofiber loss outpaces a typical rate of new cell addition such that muscles with the potential to grow are instead steadily diminished. Alternatively, myofiber loss may be accelerated by a compromised ability to form new myofibers and to add nuclei to growing myofibers. To distinguish these hypotheses, we analyzed cohorts of animals two days after intraperitoneal BrdU injection at P0, P2, or P4. Since multinucleated jerboa foot present within Dystroglycan+ myofiber membranes were added by myocyte fusion during the two-day window after they were labeled as myoblasts or myocytes in S-phase ( Figure 2A).
When normalized to the total number of myofiber nuclei, we found that myocytes fuse to form multinucleated myofibers in jerboa hand muscle at a consistent rate from P0 to P6. However, their incorporation into jerboa foot muscle decreased significantly after P2 ( Figure 2B). These 135 results suggest that myofiber loss, which begins at P4, is preceded by reduced myogenesis. The reduced rate of myocyte incorporation could be due to reduced numbers of muscle progenitor cells or to an inability of these cells to mature and fuse. To distinguish these possibilities by quantifying proliferative muscle progenitor cells, we analyzed animals two hours after BrdU injection at P0, P2, and P4 and counted the number of BrdU+ nuclei located between 140 the Dystroglycan+ myofiber membrane and the Laminin+ basal lamina (Fig. 2C). Normalized to the total number of myofibers, we found that the number of proliferative progenitor cells in to jerboa foot muscle significantly decreased from P0 to P4 compared to hand muscles that showed no change over time (Fig. 2D). These results suggest that a reduced number of muscle progenitor cells might contribute to the reduced prevalence of myocyte fusion events.
We next tested whether compromised proliferation and differentiation of jerboa foot 160 muscle progenitors is cell autonomous or non-cell autonomous. We isolated single cells, including myoblasts and myocytes but excluding myofibers, by mechanical trituration and enzymatic digestion of P1 jerboa and mouse lower leg and foot muscles (Danoviz and Yablonka-Reuveni, 2012a). After 6 days and 9 days of culture, we detected Myogenin+ differentiating myocytes and Myosin+ fully differentiated myofibers in primary cell cultures 165 isolated from each muscle (Fig. 2E). We did not detect a significant decline in the number of The rapid and almost complete loss of differentiated myofibers in vivo from P4 to P8 170 suggested these cells die, since individual cells or groups of cells are commonly eliminated by apoptosis during development (Brill et al., 1999;Fernández-Terán et al., 2006). We therefore tested the hypothesis that neonatal intrinsic foot muscles undergo apoptosis by implementing the TUNEL assay to detect DNA fragmentation and by immunofluorescent detection of cleaved Caspase-3, a key protein in the apoptotic program (Elmore, 2007). Each revealed keratinocyte 175 apoptosis in hair follicles, which are known to undergo programmed cell death, as a positive control in the same tissue sections (Magerl et al., 2001). However, TUNEL or cleaved Caspase-3 positive jerboa foot myofibers or cells in their vicinity were an extreme rarity (0.25% of myofibers) in animals ranging from P0 to P8 and comparable to mouse myofibers suggesting muscle is not eliminated by apoptosis ( Figure 3A Alternatively, myofiber loss may occur through a cell death mechanism that is first characterized by plasma membrane permeability, such as necrosis (Berghe et al., 2014). To test this hypothesis, we injected Evans blue dye (EBD), a fluorescent molecule that accumulates in cells with compromised plasma membranes (Hamer et al., 2002;Matsuda et al., 1995), into the peritoneum of P3 and P4 neonatal jerboas 24 hours before euthanasia. Although 185 we detected EBD in mechanically injured myofibers of the gastrocnemius as a control, we saw no EBD fluorescence in jerboa foot myofibers or in surrounding cells ( Figure 3C and Since we observed no direct evidence of cell death, we asked whether there was an immune response that might be an indirect proxy for undetected death. Dying muscle cells frequently recruit phagocytic macrophages that engulf cellular debris (Arnold et al., 2007;Londhe and Guttridge, 2015;Tidball and Wehling-Henricks, 2007). We predicted that myofibers that die by any mechanism that produces cellular debris might recruit macrophages that are 210 detectable by expression of the F4/80 glycoprotein. However, consistent with the lack of evidence of cell death in the jerboa foot, no F4/80 + macrophages were found among myofibers from birth to P7 ( The absence of any clear indication of muscle cell death motivated us to re-evaluate muscle maturation at greater resolution in order to capture the earliest detectable signs of muscle cell loss. We collected transmission electron micrographs of jerboa hand and foot 220 muscle at P0, P2, and P4. We identified criteria for three categories of maturation, as described previously (Borisov et al., 2008;Raeker et al., 2014;Sanger et al., 2006), and two categories of degeneration. Category A cells have pre-myofibrils with thick and thin filaments and poorly resolved Z-discs, but the M-lines and I-bands are not yet apparent ( Figure 4A). In Category B, Z-discs of nascent myofibrils are better resolved, and M-lines and I-bands are apparent, but 225 parallel sarcomeres are not in register ( Figure 4B). The mature myofibrils of Category C have Zlines that are aligned with one another ( Figure 4C). In Category D, early degeneration, some sarcomeres appear similar to Category A, but other areas of the cell contain disorganized filaments ( Figure 4D). Category E includes those in the worst condition where less than half of the cell has any recognizable sarcomeres, and much of the cytoplasm is filled with pools of 230 disorganized filaments and Z-protein aggregates ( Figure 4E). Additionally, Category D and E cells have membrane-enclosed vacuoles and large lipid droplets (Figure 4 -figure supplement 1). However, consistent with a lack of evidence for cell death, none of these cells or their organelles appear swollen, nuclear morphology appears normal, plasma membranes seem to be contiguous, and we do not observe an accumulation of autophagic vesicles that typically 235 characterize cell death associated with unregulated autophagy (Mizushima, 2007;Kroemer and Levine, 2008;Denton and Kumar, 2019).
We then coded and pooled all images of hand and foot myofibers from P0, P2, and P4 jerboas and blindly assigned each cell to one of the five categories. Quantification of the percent of myofibers in each category after unblinding revealed the progressive maturation of jerboa hand myofibers and the progressive degeneration of jerboa foot myofibers ( Figure 4G).
Compared to later stages, there is little difference in the maturation state of hand and foot sarcomeres at birth. Loss of ultrastructural integrity is therefore initiated perinatally, prior to 255 complete myofibril maturation in the jerboa foot.
Our analysis of transmission electron micrographs also revealed the presence of filamentous aggregates that we did not include in our quantifications because they are enucleate, lack all other recognizable organelles, and are not bounded by a plasma membrane.
Although these aggregates do not appear to be cellular, they are always closely associated with 260 cells of a fibroblast morphology, and most lie between remaining myofibers in a space we presume was also once occupied by a myofiber ( Figure 4F,H). To determine if these unusual structures contain muscle protein, we performed immunofluorescence on sections of P4 jerboa foot muscle and found similar aggregates of intensely fluorescent immunoreactivity to skeletal muscle myosin heavy chain. We also found that the surrounding cells, which correlate with the 265 positions of fibroblasts in electron micrographs, express the intracellular pro-peptide of Collagen I ( Figure 4I), the major component of tendon and other fibrous connective tissues and of fibrotic tissue after injury (Mann et al., 2011).
Given the apparent deterioration of nascent sarcomeres, we asked whether individual sarcomere proteins are lost from myofibrils in a temporal order or if proteins disassemble 270 simultaneously. We assessed the organization of sarcomere proteins by multicolor immunofluorescence at P0, P2, and P4. Alpha-Actinin, Desmin, Myomesin, Myosin, Titin, and Tropomyosin are each localized to an ordered series of striations in a subset of myofibers suggesting all are initially incorporated into immature sarcomeres ( Figure 5A and Tables S1-5).
By assessing all combinations of immunologically compatible primary antibodies, we identified 275 populations of cells where Desmin was no longer present in an ordered array, but each of the other proteins appeared properly localized to the sarcomere ( Figure 5B and Table S2). Although we could not distinguish such clear categories of mislocalization for each protein relative to all others, we inferred a relative timeline whereby Desmin disorganization is followed together by Tables S1-5).
Desmin forms a filamentous network that connects parallel sarcomeres to one another and coordinates myofibril contraction within cells and between neighboring cells (Bär et al., 2004;Capetanaki et al., 2015;Goldfarb et al., 2008). Mutations that cause desminopathies  illustrate that Desmin is essential to maintain sarcomere integrity (Clemen et al., 2013). In mouse models of muscle atrophy triggered by fasting or denervation, phosphorylation of Desmin removes the protein from the sarcomere and targets it for ubiquitination and proteolytic 340 degradation prior to degradation of other sarcomere proteins (Volodin et al., 2017). The observation that Desmin is the first of an ordered sarcomere disassembly in the jerboa foot may reflect targeted degradation of muscle proteins that is similar to muscle atrophy.
The ubiquitin-proteasome system is the main pathway through which cellular proteins are degraded during muscle atrophy, and MuRF1 and Atrogin-1 are E3 ubiquitin ligases among 345 the 'atrogenes' that are highly upregulated (Bodine and Baehr, 2014;Schiaffino et al., 2013). To test the hypothesis that muscle loss in the jerboa foot exhibits molecular hallmarks of atrophy, we performed quantitative reverse transcriptase PCR (qRT-PCR) of MuRF1 and Atrogin-1 mRNA from intrinsic foot muscles and the flexor digitorum superficialis (FDS) of the mouse and jerboa. The FDS, which originates in the autopod during embryogenesis and translocates to the 350 forearm (Huang et al., 2013), is the most easily dissected of the analogous forelimb muscles and serves as a control for typical muscle maturation in both species. When normalized to expression in the FDS at birth of each species, Atrogin-1 expression is 3.1-fold higher in the jerboa foot at P3 ( Figure 5G). MuRF1 mRNA expression is already significantly elevated at birth and remains elevated at P3 ( Figure 5H). 355 The NF-κB pathway is an upstream mediator of skeletal muscle atrophy (Li et al., 2008) and is both necessary and sufficient to induce MuRF1 expression (Cai et al., 2004;Wu et al., 2014). To lend further support to the hypothesis that jerboa foot muscle loss involves an 'atrophy-like' mechanism, we performed qRT-PCR of NF-κB2 and its binding partner, Relb. We observed that each gene is expressed greater than three-fold higher in jerboa foot at birth and 360 at P3 (Figure 5 -figure supplement 1A, B). The progressively disordered ultrastructure of the sarcomere that begins with loss of Desmin localization, the increased expression of multiple genes that are typically upregulated during atrophy, and the lack of evidence for cell death or macrophage infiltration are consistent with observations of atrophying muscle in mice and rats (Volodin et al., 2017;Bonaldo and Sandri, 2013;Sakuma et al., 2015;von Haehling et al., 365 2010).
Despite the similarities to muscle atrophy, myofiber loss in the jerboa foot does not seem to be simply explained by an atrophic response to denervation. First, and in contrast to the rapid rate of jerboa foot myofiber loss, chronic denervation in mice (100 days after nerve transection at P14) reduced the size but not the number of individual myofibers (Moschella and Ontell, 370 1987a). Additionally, we found that the post-synaptic Acetylcholine Receptor (AchR) exclusively coincides with the presynaptic neuronal protein, Synaptophysin, in neonatal jerboa foot muscles ( Figure 5 -figure supplement 1). In the mouse, AchR clusters are present in a broad domain of fetal muscle prior to innervation and are refined to nerve terminals in response to chemical synapse activity before birth (Yang et al., 2001). The refinement of AchR clusters in jerboa foot 375 muscles suggests that the muscles are not only innervated at birth but are also responsive to motor inputs.

Discussion
The natural process of muscle loss in the jerboa foot is surprising in the context of what 380 is known about muscle development and pathology. Although we found multinucleated myofibers in the feet of neonatal jerboas, all muscle protein expression rapidly disappears from the jerboa foot shortly after birth. We were perplexed to find no evidence of apoptotic or necrotic cell death by a variety of assays and throughout the time when muscle cells are lost, nor did we observe immune cells that are commonly recruited to clear the remains of dying cells. Instead, 385 we saw structural and molecular similarities to muscle atrophy, though atrophy in young mice leads to reduced myofiber size rather than number as in the jerboa (Bruusgaard and Gundersen, 2008;Moschella and Ontell, 1987a).
As for why the phenotype is limited to the distal limb, it is possible that disuse contributes to jerboa foot muscle loss, since jerboas and ungulates each fuse metatarsals into a single 390 cannon bone, which would be expected to physically impair lateral motion of the digit elements (Cunningham, 1883;Moore et al., 2015). However, the rapid and complete loss of myofibers in the neonatal jerboa foot does not appear to simply reflect a species-level difference in the animal's generalized response to disuse atrophy, since hindlimb denervation and immobilization in adults causes gradual loss of muscle mass, primarily through a significant reduction in the 395 diameter of individual myofibers (AlWohaib and Alnaqeeb, 1997; Aryan and Alnaqeeb, 2002).
These observations are very similar to what has been shown in disuse atrophy models in mice and in rats (Bonaldo and Sandri, 2013;Moschella and Ontell, 1987b) and differ from what we see in the neonatal foot.
Why would an embryo expend energy to form muscles that are almost immediately lost? 400 The formation and subsequent loss of intrinsic foot muscles in jerboas and hooved animals may simply reflect a series of chance events in each lineage with no fitness cost, or these similarities in multiple species may reveal true developmental constraints. Muscle is not required for autopod tendon formation or maintenance in mice, but the tendons that develop in a muscleless or a paralyzed mouse are thinner and less well organized (Huang et al., 2015). It is 405 therefore possible that muscle is initially required in the fetus and neonate for tendons to establish sufficient architecture from origin to insertion so that the tendon, after further growth, can withstand high locomotor forces in the adult (Lochner et al., 1980;Moore et al., 2017).
Regardless of whether these nascent muscles serve an essential purpose, we are left wondering what is the ultimate fate of jerboa foot myofibers. If these cells do indeed die, 410 perhaps death is too rapid for detection. However, programmed cell death is thought to occur over hours or even days from the initial trigger to the final corpse (Green, 2005). Alternatively, death may result from a mechanism that does not proceed through DNA fragmentation, plasma membrane permeability, macrophage recruitment, or stereotyped ultrastructural changes, and yet this would seem to eliminate most known forms of regulated cell death (Galluzzi et al., 2007) 415 Alternatively, multinucleated myofibers may transform to another cellular identity after degrading all sarcomere proteins. Although a fate transformation would be surprising, it would not be without precedent. The electric organ of fish that can produce an electric field (e.g. knifefish and elephantfish) is thought to be derived from skeletal muscle. Electrocytes of Sternopygus macrurus express skeletal muscle Actin, Desmin, and α -Actinin, and electrocytes 420 of Paramormyrops kingsleyae retain sarcomeres that are disarrayed and non-contractile (Gallant et al., 2014;Unguez and Zakon, 1998). If myofibers in the jerboa foot indeed transdifferentiate, it is possible they transform into the pro-Collagen I positive fibroblasts that are entangled with the filamentous aggregates, though these could also be phagocytic fibroblasts recruited to consume the enucleate detritus without stimulating inflammation (Heredia et al., 425 2013;Joe et al., 2010;Monks et al., 2005;Schwegler et al., 2015). Unfortunately, the lineage labeling approaches required to track the ultimate fate of jerboa myofibers are exceptionally challenging in this non-traditional animal model.
It is clear, however, that regardless of the ultimate fate of jerboa foot myofibers, their path passes through a phase marked by cell biology that is typical of atrophy, including the 430 ordered disassembly of sarcomeres and expression of the E3 ubiquitin ligases, MuRF1 and Atrogin-1. However, skeletal muscle atrophy is typically associated with pathology in the context of disuse, nerve injury, starvation, or disease. In this context, we were struck by a statement in the 1883 anatomical description of the fetal and adult suspensory ligament of four species of hooved mammals: "It is an instance of pathological change assisting a morphological process" 435 (emphasis his) (Cunningham, 1883). Indeed, there are remarkable similarities in the histology of jerboa and horse foot muscle compared to human clinical observations of tissue remodeling that follows rotator cuff tear characterized by muscle atrophy, myofiber loss, and fibrosis (Souza et al., 2010;Gibbons et al., 2017).
Foot muscle atrophy in the jerboa may be one of many cellular responses associated 440 with injury or disease in humans that is utilized in the normal development and physiology of other species. These data suggest that there is less of a clear divide between natural and pathological states than typically thought. Studies of non-traditional species may not only reveal the mechanisms of evolutionary malleability, but may also advance our understanding of fundamental biological processes that are typically associated with pathological conditions. 445

Animals
Jerboas were housed and reared as previously described (Jordan et al., 2011).

Immunofluorescence and TUNEL 470
Mouse and jerboa limbs were dissected and fixed in 4% PFA in 1x PBS overnight.
Tissues were washed in 1X PBS twice for 20 min and placed in 30% sucrose in 1x PBS overnight at 4 degrees Celsius. Tissues were then mounted in a cryomold in OCT freezing media, and blocks were frozen and stored at -80°C until cryosectioned.
Blocks were sectioned at 12 µm thickness, and sections were transferred to Super-475 Frost Plus slides (Thermo Fisher). For immunofluorescence, slides were washed for 5 min in 1x PBS and subject to antigen retrieval by incubation in Proteinase K (5 µg/mL) for 10 min followed by 5 min postfix in 4% PFA in PBS and three washes in 1x PBS. Slides were then blocked in a solution of 5% heat inactivated goat serum, 3% Bovine Serum Albumin, 0.1% TritonX-100, 0.02% SDS in PBS. Slides were incubated in the appropriate primary antibody dilution in block 480 overnight at 4°C. On the second day, slides were washed three times for 10 minutes in PBST (1x PBS + 0.1% TritonX-100) and incubated at room temperature in secondary antibodies and 1 µg/ml DAPI for one hour. Slides were then washed three times for 10 minutes in PBST and mounted in Fluoro Gel with DABCO (EMS).
For TUNEL, slides that had been previously processed for MF20 immunofluorescence 485 were placed immediately into the TUNEL reaction mixture following manufacturer's instructions (Roche In Situ Cell Death Detection Kit, TM-Red) for 60 min at 37°C, rinsed three times in 1x PBS, and mounted in Fluoro Gel with DABCO.

Myofiber count
Blocks containing embedded mouse or jerboa feet were cryosectioned at 12µm thickness in transverse orientation onto two serial sets of slides. Slides of the second series were used as back up in case certain sections of the first series contain folded tissue and 495 thus cannot be used. Slides of the first series were stained with MF20 & WGA and analyzed to locate the proximal and distal ends of the third interosseous muscle. Using this information we estimated the middle area of each muscle and selected 10 sections for subsequent analysis. We analyzed the third interosseous muscle of the hindlimb, spanning approximately 240 µm in muscle length. For each selected section, all cross-sectionally oriented myofibers 500 were manually counted and recorded using the plugin cell counter in ImageJ. The average number of myofibers from 10 sections represents an estimate of the myofiber number for the middle transverse section of the third interosseous muscle. For each developmental stage, data from three animals were collected, and one-way ANOVA with Tukey's multiple comparisons test was performed to determine the statistical significance of mean myofiber 505 number differences between developmental stages in each species.

Myocyte fusion assay
BrdU solution was intraperitoneally injected to achieve 100 μg/g (BrdU/ animal body weight) in P0, P2, and P4 jerboas. Injected animals were sacrificed two days later. The feet and 510 hands of each animal were fixed in 4% PFA/PBS overnight, processed through a sucrose series, and embedded in OCT freezing media. Blocks of embedded tissue were cryosectioned in transverse orientation at 12 µm thickness and placed in serial sets on Superfrost Plus slides.
Slides were stained with BrdU and Dystroglycan antibodies as indicated above. As in the methods to count myofibers, we chose ten sections near the midpoint of the interosseous 515 muscle associated with the third metatarsal and counted all BrdU+ nuclei within a Dystroglycan+ myofiber as well as all myofiber nuclei in each section. Data is represented as the total number of BrdU+ myofiber nuclei divided by the total number of myofiber nuclei, and this ratio was averaged for all 10 sections in each animal. The data was plotted using Prism8 (GraphPad), and the statistical significance between datapoints at each time interval was calculated with 520 one-way ANOVA with Tukey's multiple comparisons test in each of forelimb and hindlimb.

Short-term BrdU labeling
BrdU solution was intraperitoneally injected to achieve 100 μg/g (BrdU/ animal body weight) in P0, P2, and P4 jerboas. Injected animals were sacrificed two hours after injection. 525 The feet and hands of each animal were fixed in 4% PFA/PBS overnight, processed through a sucrose series, and embedded in OCT freezing media. Blocks of embedded tissue were cryosectioned in transverse orientation at 12 µm thickness and placed in serial sets on Superfrost Plus slides. Slides were stained with BrdU and Myosin or BrdU, Laminin, and Dystroglycan antibodies as indicated above for assessment of proliferation in myonuclei. As in 530 the methods of fusion assay, we chose ten sections near the midpoint of the interosseous muscle associated with the third metatarsal and counted all BrdU+ nuclei within a Laminin+ basal lamina and outside Dystroglycan+ myofiber membrane as well as number of Dystroglycan+ myofiber in each section. Data is represented as the total number of BrdU+ myofiber nuclei divided by the total number of myofiber, and this ratio was averaged for all 10 535 sections in each animal. The data was plotted using Prism8 (GraphPad), and the statistical significance between datapoints at each time interval was calculated with one-way ANOVA with Tukey's multiple comparisons test in each of forelimb and hindlimb.

Muscle stem/progenitor cell culture: 540
Intrinsic foot muscles (m. flexor digitorum brevis and m. interossei) and lower leg muscles (tibialis anterior and gastrocnemius) were manually dissected from three animals of P1 jerboas and mice and pooled. After connective tissues were manually removed with forceps, muscle stem/progenitor cells were isolated and cultured as described in (Danoviz and Yablonka-Reuveni, 2012b) Briefly, the tissues were enzymatically with 10 mg/ml Pronase (EMD 545 Millipore) and mechanically dissociated. The cells were plated onto matrigel-coated 8-well chamber slides (Nunc Lab-Tek, Thermo fisher) coated with Matrigel (Corning) at 1 × 10 4 cells/well. The cells were cultured for 9 days with DMEM (Thermo Fisher), 20% fetal bovine serum (Thermo fisher), 10% horse serum (Thermo Fisher) and 1% chicken embryonic extract (Accurate Chemical). During the culture period, the medium was changed at day 3, 6, and 8. 550 After 6 and 9 days, cells in replicate cultured wells were fixed with 4% PFA/PBS at 4℃ for 15 min and washed with PBS. After permeabilization with 1 % Triton-X 100 in PBS at room temperature for 10 min, the cells were blocked with 5% BSA/PBS for 30 min and stained with BrdU, anti-Myogenin and Myosin antibodies and secondary antibodies. At each time point of each experimental group, the total number of nuclei and nuclei within Myosin+ myofibers were 555 counted in 10 images taken from 8 wells using the Olympus compound microscope at 4x magnification. The numbers in 10 images were averaged and the difference between day 6 and day 9 were statistically analyzed with paired sample t-test in each experimental group.

Evans Blue Dye 560
We injected Evans Blue Dye as 1% solution by animal body weight (1mg EBD/100µl PBS/10g) 24 hours prior to sample collection (Hamer et al., 2002). As a positive control for EBD uptake, we create an injured muscle area by inserting a 21-gauge needle 2-3 times into the jerboa gastrocnemius muscle. Samples were fresh frozen in OCT and cryosection at 12 µm thickness. Slides were processed for MF20 fluorescence with primary antibody incubation for 1 565 hr at RT before secondary antibody incubation. Slides were mounted for analysis: EBD signal is detected using the Cy5 filter and imaged using the Olympus compound microscope or imaged using the Leica SP5 confocal laser 633nm.

Oil red O (ORO) staining 570
ORO stock solution: 2.5 g of Oil red O to 400 ml of 99% (vol/vol) isopropyl alcohol and mix the solution by magnetic stirring for 2 h at room temperature (RT; 20-25 °C). ORO working solution: 1.5 parts of ORO stock solution to one part of deionized (DI). Cryosections were fixed with 4%PFA in 1x PBS for 5 minutes. Slides were washed with 2x with PBS for 10 minutes each and stained with ORO working solution for 10 minutes followed three 30 second washes with DI 575 water. Slides were then washed in running tap water for 15 minutes followed by three 30 second washes with DI water and mounting in aqueous medium.

Transmission Electron Microscopy (TEM)
Animals were perfused with 2% glutaraldehyde and 2% PFA plus 2mM CaCl 2 in 0.15M 580 sodium cacodylate buffer, pH 7.4 @ 35°C for 2-3 minutes. The hands and feet were removed, skinned, and fixed on ice for 2 hours. Samples were then rinsed six times for 5 min in cold 0.15M cacodylate buffer and then post-fixed in 1% OsO4 in 0.15M cacodylate buffer on ice for 1 hour. Samples were then rinsed in cold double distilled water (DDW) six times for 5 min and placed into 1% uranyl acetate in DDW on ice overnight. Fixed tissue was then rinsed in ice cold 585 double distilled water three times for 3 min and dehydrated in an ethanol series (50%, 70%, 90% in DDW) on ice for 5 min each. Samples were further dehydrated into 100% ethanol twice for 5 min at room temperature and then transitioned to 1:1 ethanol:acetone for 5 min followed by two times 5 min in 100% acetone. Dehydrated samples were infiltrated with 1:1 acetone:Durcupan ACM resin for 1 hour at room temperature followed by 100% resin twice for 1 590 hour and then placed in fresh resin overnight. On the next day, samples were transferred to fresh resin, which was polymerized in a 60°C vacuum oven for 48-72 hours. Resin embedded samples were stored at room temperature until ready for sectioning. Seventy nanometer thick sections were stained in lead solution and image using Tecnai Spirit TEM scope (120 kV).

PCR)
Foot muscles were dissected and stored in RNAlater solution (Thermo Fisher) at -80°C until ready for use. RNA extraction was performed using the PicoPure RNA Isolation Kit (Thermo Fisher) according to the manufacture instructions. RNA was reverse transcribed to 600 generate cDNA using QuantiTect Reverse Transcription Kit. cDNA was used as template for quantitative PCR with PCR amplification detected with Sybr green (SYBR Green Real-time PCR master mixes, Invitrogen). See the table below for the sequences of primers used to quantify real time amplification.
Each quantitative reverse transcriptase PCR experiment was conducted twice with 605 technical triplicates in each experiment. Cq values that are significant outliers were determined using Grubb's test in GraphPad software and eliminated. Expression of MuRF-1, and Relb was normalized to SDHA, quantitation of gene expression was determined by the equation 2 −ΔΔCT , and the fold-change of mRNA expression was calculated relative to the mRNA level of P0 FDS samples in each species, which was set to 1. One-way 610 ANOVA with Tukey's multiple comparisons test was performed to determine the statistical significance of fold change differences between samples in each species.

mouseMuRF1_F
TGCCTGGAGATGTTTACCAAGC (Dogra et al., 2007) mouseMuRF1_R AAACGACCTCCAGACATGGACA (Dogra et al., 2007) Table S1. Information extracted from multicolor immunofluorescence of individual myofibers to infer the order of sarcomere protein disorganization in jerboa foot muscles. Related to Figure 4. "Good" represents striated localization of each protein to the sarcomere, and "bad" refers to no distinguishable banded pattern of protein expression. In group 1, we saw 960 myofibers with all three proteins properly localized suggesting disorganization follows an initial state of proper localization. When we compared the dRM and Drm categories, we saw loss of Desmin localization when Tropomyosin and Myosin were "good" and almost no myofibers where Desmin was good and the others were bad. This suggests that Desmin is disorganized prior to Tropomyosin and Myosin. In group 2a, there were cells in the rmT category and almost none in 965 the RMt category, suggesting Tropomyosin and Myosin are disorganized prior to Titin. Group 2b illustrates that both categories RmT and rMT appeared at similar frequency, suggesting it is unclear whether Tropomyosin or Myosin become disorganized prior to the other. In group 3, there were cells in the mtA category and none in the MTa category, suggesting Titin becomes disorganized before Alpha-actinin. Similarly, in group 4, there were cells in the mtY category but 970 not in the MTy category suggesting Titin becomes disorganized before Myomesin. Due to shared antibody isotype for Alpha-actinin and Myomesin, the order of disorganization between these two could not be discerned. See Table S2-5 for full details of the percentage of myofibers in each category for each combination of multicolor immunofluorescence. 975