3D Timelapse Analysis of Muscle Satellite Cell Motility

Skeletal muscle repair and regeneration requires the activity of satellite cells, a population of myogenic stem cells scattered throughout the tissue and activated to proliferate and differentiate in response to myotrauma or disease. While it seems likely that satellite cells would need to navigate local muscle tissue to reach damaged areas, relatively little data on such motility exist, and most studies have been with immortalized cell lines. We find that primary satellite cells are significantly more motile than myoblast cell lines, and that adhesion to laminin promotes primary cell motility more than fourfold over other substrates. Using timelapse videomicroscopy to assess satellite cell motility on single living myofibers, we have identified a requirement for the laminin-binding integrin α7β1 in satellite cell motility, as well as a role for hepatocyte growth factor in promoting directional persistence. The extensive migratory behavior of satellite cells resident on muscle fibers suggests caution when determining, based on fixed specimens, whether adjacent cells are daughters from the same mother cell. We also observed more persistent long-term contact between individual satellite cells than has been previously supposed, potential cell-cell attractive and repulsive interactions, and migration between host myofibers. Based on such activity, we assayed for expression of “pathfinding” cues, and found that satellite cells express multiple guidance ligands and receptors. Together, these data suggest that satellite cell migration in vivo may be more extensive than currently thought, and could be regulated by combinations of signals, including adhesive haptotaxis, soluble factors, and guidance cues. Stem Cells 2009;27:2527–2538


INTRODUCTION
Skeletal muscle, which comprises up to 40% of human body mass, is a highly ordered, structurally stable tissue composed of differentiated, contractile myofibers arrayed within concentric sheaths of extracellular matrix and connective tissue. Very little turnover of muscle tissue occurs in vivo, except when muscle repair and/or regeneration is required to respond to overwork, trauma or disease. In such cases, a population of adult myogenic precursor cells, satellite cells, is required to provide a source of new myonuclei. In undamaged muscle, satellite cells are a rare, highly dispersed population of mitotically quiescent single cells, maintained in a niche outside the plasmalemma but beneath the basal lamina of host myofibers. This niche insulates the satellite cell from a majority of extracellular stimuli, allowing them to become ''activated'' only in specific conditions associated with damage or disease; known activating factors are currently limited to hepatocyte growth factor (HGF), nitric oxide (NO), and possibly tumor necrosis factor-alpha (TNF-a) (reviewed in [1,2]).
Once activated, satellite cells will emerge from beneath the basal lamina and proliferate extensively, establishing a large population of differentiation-competent myoblasts. These myoblasts will eventually differentiate and fuse either to one another or to existing myofibers to replace or repair damaged muscle. Interest in satellite and other muscle-derived stem cells has increased dramatically over the past several years, reflecting both the advent of new molecular and technical tools and increased realization of their clinical potential as vectors for cell and gene therapy. While myoblast transplantation as a therapy for muscular dystrophy was initially shown to be feasible in mice in 1989, albeit under conditions that would not be applicable to muscle disease therapy [3], subsequent efforts to engraft satellite cells or other myoblasts into host muscle have met with extremely limited success. Three major problems contributing to this lack of success are: 1) the majority of engrafted cells die within 3 days of injection; 2) if immunosuppression is not adequate, any surviving myoblasts are rejected within 2 weeks; and 3) those myoblasts that do survive do not migrate more than 200 lm away from the injection site (reviewed in [4]). This last hurdle, in particular, has proven difficult to overcome even as significant progress has been made on the other two. While current protocols can result in up to 30% dystrophin-positive fibers after treatment [5,6], they require a very large number of injections: the cited studies utilized 100 injections per square centimeter of muscle surface (up to 4,000 injections in a single individual) to treat highly restricted anatomical areas. These highly concentrated myoblast injections also lead to necrosis secondary to oxygen deprivation and toxic metabolites caused by such a large cell bolus [7].
It remains unclear to what extent cellular migration is involved in satellite cell-mediated regeneration. However, it seems plausible that given the isolation and relatively sparse distribution of satellite cells in uninjured tissue, accumulation of a large population of activated myoblasts at a site of focal injury would require directional motility. Early in vivo data is consistent with this hypothesis, suggesting that activated satellite cells may not only traverse the entire length of a myofiber, but will also migrate between fibers [8][9][10][11]. However, these studies were primarily descriptive rather than mechanistic, and few studies examining satellite cell motility have appeared in the literature since.
Asking specific questions about satellite cell migration in vivo is, like most other aspects of satellite cell physiology, complicated by the rarity and dispersion of satellite cells within the tissue. Conversely, in vitro experiments on large, purified satellite cell populations in adherent tissue culture remove any potential contribution of endogenous stimuli and substrates, and may introduce artifactual results, particularly when examining adhesion-based signals. In the most extreme case, immortalized myoblast cell lines are convenient and consistent, but are the furthest removed from an in vivo system (reviewed in [12]). To attempt to bridge the gap between these systems, we have developed a method to analyze primary satellite cells on their native migration substrate (the surface of a myofiber) by timelapse microscopy, in a programmable three-dimensional matrix. While it is not a true in vivo system, in that soluble and matrix-associated factors and forces that do not originate with either the satellite cells themselves or the single myofibers are absent, the relative advantages of this in vitro system are that it: 1) permits retention of the native adhesion and migration substrate for endogenous primary satellite cells, and 2) allows the establishment of reproducible, defined experimental conditions to test hypotheses and define genetic, protein, and morphologic changes in a temporally coordinated fashion. We can manipulate the availability of soluble factors and/or adhesion/guidance factors by supplementing the medium with exogenous factors or by adding blocking antibodies or pharmaceutical inhibitors. Multiple means of assessing the role of cell-intrinsic factors are also possible, including: 1) comparing cells from wild-type versus mutant mouse strains; 2) comparing cells from the same strain carrying conditional mutations in either the recombined or unrecombined state; 3) overexpressing factors of interest via viral transduction; or 4) knocking down factors of interest with siRNAs encoded in viral vectors.
We have used this system to establish that satellite cell motility is regulated by soluble factors such as HGF, as well as adhesion-based signaling between the myofiber external lamina and a7b1 integrin on the satellite plasma membrane. Additionally, we describe unexpected ''behaviors'' that suggest intriguing possibilities for satellite cell interactions with one another and their environments. In particular, the duration and closeness of associations between either sister cells or unrelated cells suggests that it is difficult to accurately interpret the cell lineage relationship of adjacent cells in static preparations. We also present novel data showing that primary satellite cells express a wide array of guidance receptors and ligands that are well known in the context of neuronal and endothelial cell migration, but which have not to date been examined in satellite cells. This suggests that active pathfinding by satellite cells may play a role in muscle regeneration and repair, and thus may constitute a new area of focus for both basic and therapeutic research.
Reverse Transcriptase Polymerase Chain Reaction. Intronspanning primers were written using PrimerSelect (DNAStar, Madison, WI, www.dnastar.com). Primer sequences and product lengths are listed in Supporting Table 1. Total RNA was harvested from primary adult myoblasts after 4 days in monoculture, and from proliferating MM14 cells or proliferating C2C12 cells, then reverse-transcribed to cDNA (SuperScript II, Invitrogen). 100 ng of each sample was used as template for simultaneous polymerase chain reactions (PCR).
Briefly, muscle is dissected from the hind limbs, carefully separated from associated tissues, and digested in 400 U/mL collagenase type I (Worthington) diluted in Ham's F-12 medium (Invitrogen). When single fibers are liberated, they are manually picked with a pipette and cultured at 37 C and 5% CO 2 in growth medium [Ham's F-12 (Gibco), 15% horse serum (Equitech), and penicillin/streptomycin (Gibco) supplemented with 0.5 nM rhFGF-2]. For standard 24-48 hour data collections, fibers are cultured for 24 hours before being repicked into 48well plates for timelapse analysis.
Collagen Gel Culture and Timelapse Capture. 3 to 5 myofibers are added to 200 ll of acid-extracted rat tail type I collagen (made in-house [16]; 2 mg/ml in growth medium) per well in 48well plates (Corning, Lowell, PA, www.corning.com), and the collagen is allowed to rapidly polymerize at 37 C. The wells are then overlaid with growth medium containing supplements as appropriate [0.5 nM FGF-2, 80 ng/ml HGF (R&D Systems Inc., Minneapolis, MN, http://www.rndsystems.com), 200 ng/ml SDF-1 (R&D Systems), 1 uM lysophosphatidic acid (LPA) (Avanti Polar Lipids), 1 uM S1P (Avanti Polar Lipids, Alabaster, AL, www. avantilipids.com), or antibodies as listed below]. All conditions are represented by duplicate wells in every experiment. Multiple 10x fields are identified per well and marked for return; images are automatically collected from each field every 10 minutes using IPLab (Scanalytics, Rockville, MD, www.scanalytics.com).
Postimaging Analysis. Stacked images generated by IPLab (Scanalytics) are imported into MetaMorph (Axon Instruments/ Molecular Devices Corp., Union City, CA, http://www.moleculardevices.com) and arranged in sequential order. Distance of migration is measured using digital pixel trace measurements. If a cell selected for tracking proliferates during the 24-hour collection period, one daughter cell is selected at random to continue the trace. Tortuosity of tracks was determined using Fractal5 [17,18], and is defined as the deviation from a correlated random walk, where 1 ¼ a straight line and 2 ¼ a path so tortuous as to cover the entire two-dimensional plane.

Myogenic Cell Adhesion and Motility on Purified Matrix Substrates Is Quantitatively and Qualitatively Distinct
The external lamina surrounding individual myofibers in vivo is primarily composed of type IV collagen and laminin [28], while the interstitial connective tissue contains type I, III, and V collagen and fibronectin [29]. In tissue culture, the choice of substrate can have dramatic effects on cell activity. Hauschka and Konigsberg reported in 1966 that collagen deposited by fibroblasts on tissue culture dishes promoted enhanced proliferation and differentiation of embryonic chick myoblasts [30]. Fibronectin and laminin have also been shown to have distinct effects on myoblast adhesion, morphology, and motility: while both will support adhesion, laminin specifically promotes myoblast locomotion while cells on fibronectin tend to organize vinculin-containing focal contacts and assemble a-actinin stress fibers [31]. To establish baseline measures for substrate effects on primary adult myoblasts, we adhered primary satellite cells to purified collagen, fibronectin, or laminin, then stained them with fluorescently-labeled phalloidin to visualize the actin cytoskeleton. We also included two commonly-used satellite cell-derived lines: C2C12 cells [32], which are the most commonly used myoblast cell line in adhesion/migration studies (as well as in most other applications), and MM14 myoblasts [33], which are less widely used but retain more of the morphological and molecular characteristics of primary satellite cells.
We find that C2C12 cells maintained their characteristic flattened morphology on all three substrates with extensive f-actin networks and extensions, while MM14 cells and primary satellite cells are rounded when adhered to either collagen or fibronectin, but become polarized and elongated when adherent to laminin (Fig. 1A). Satellite cells and myogenic cell lines seeded on purified extracellular matrix (ECM) components also display differential motility, consistent with previous descriptions of MM14 locomotion in vitro [31]. Cells plated on collagen I adhere and retain a characteristic round morphology, but are not motile (Fig. 1B). Cells plated on fibronectin adhere minimally and remain very transiently attached (Fig. 1B). However, cells plated on laminin are strongly adherent, take on a bipolar, flattened morphology, extend cellular processes, and in the case of MM14 cells and primary satellite cells, have significantly greater mobility ( Fig.  1B; movies 1b1, 1b2, and 1b3). We observed clear differences in motility between primary cells and cell lines: on laminin, primary cells are more than sixfold more motile than C2C12 cells, and 1.6-fold more motile than MM14s (Fig. 1C).
The cellular receptors for ECM components including laminin and fibronectin are integrins, a family of type I transmembrane glycoproteins found only in metazoans (reviewed in [34]) that combine one of 18 a chains and one of 8 b chains into one of 24 permitted pairings to form functional heterodimers with unique substrate affinities. Embryonic/fetal myoblasts and cell lines derived from embryonic skeletal muscle have previously been shown to express a4, a5, a6, a7, aV, and b1, while adult skeletal muscle expresses a5b1, a6b1, and a7b1, primarily at myotendinous and neuromuscular junctions (reviewed in [35]). a5b1 integrin is a fibronectin receptor, whereas a6b1 and a7b1 show strong specificity for laminin (reviewed in [36]). The deletion phenotypes of a5 and a7 integrin both include muscular dystrophy, suggesting a critical role for signaling through these subunits (reviewed in [34]), and a6 integrin is upregulated in regenerating dy/dy dystrophic mice [37]. To assess the role of specific integrin chains in 2D motility, we treated each cell type/substrate with neutralizing antibodies directed against integrins a4, a5, a6, a7, b1, or b2. Blocking a7 integrin, b1 integrin, or both together additively and significantly decreased the velocity of MM14 cells and primary satellite cells (Fig. 1D); no other treatment/condition produced a significant change in motility (data not shown). This is consistent with a role for binding to laminin through the integrin a7b1 receptor in satellite cell motility.
To determine if the collagen matrix used in the 3D assays described below could affect satellite cell motility independently of supplementation, primary cells were plated on laminin as described above and half of the wells were overlaid with collagen; cells were then tracked for 24 hours. We noted Siegel, Atchison, Fisher et al. 2529 www.StemCells.com no significant differences in cell morphology or motility (data not shown).

Primary Satellite Cells and Myogenic Cell Lines Differentially Express a Wide Variety of Integrin Chains
The contribution of satellite cells to studies of adult muscle integrin expression is likely to be undetectable, since they make up such a small fraction of the total muscle mass; no comprehensive survey of adult satellite cell integrin expression has yet been performed. To determine the repertoire of integrin chains present in adult myoblasts, we assayed primary satellite cells, MM14 cells, and C2C12 cells for integrin chain expression. By RT-PCR we can detect surprisingly broad integrin expression in satellite cells: all or most known integrins (aE and/or aL are not detected in some samples) can be detected in primary satellite cells, while a smaller subset are detected in myogenic cell lines ( Fig. 2A). We extended this result with Western blotting using chain-specific antibodies: while primary cells express slightly more integrin protein than MM14 cells, C2C12 cells possess significantly less than would be expected, and both cell lines express more of a higher molecular weight isoform of integrin b1 (potentially integrin b1D, an isoform associated with the formation and stabilization of focal adhesions [38]) than primary cells (Fig.  2B). There are also apparent discrepancies between the relative amounts of integrin chain mRNA expression and protein expression, especially in C2C12 cells: while some chains, such as a2 integrin, show lower levels of expression of both transcript and protein, several others show robust bands by RT-PCR and minimal or no signal by Western blot. We speculate that this could be due to the standard practice of culturing C2C12 cells on plastic, with no available physiological substrate, however this is untested. These results suggest that satellite cells may have greater flexibility in their potential for attachment and interaction with the ECM than was previously appreciated. Quiescent satellite cells reside between the sarcolemma and the exterior lamina of the host myofiber; after activation, they move from beneath the lamina to the surface of the myofiber. We observed that satellite cells in 3D fiber culture exit their sublaminal niche as early as 12 hours after synchronous activation by fiber harvest (Fig. 3 and movie 3a). We confirmed that satellite cell movement after activation is on the exterior of the myofiber by costaining isolated fibers before and after satellite cells have become visible outside the basal lamina with anti-laminin (to delineate the myofiber's exterior surface) and anti-syndecan-4 (to identify the satellite cell plasma membrane). While satellite cells are still beneath the laminin sheath at 4 hours after harvest, at 24 or 48 hours, they are instead adhered to the exterior of the myofiber (Fig. 3 and movie 3b).
After activation, satellite cells begin to proliferate and move along the myofiber. While making qualitatitive observations, we noted several activities and interactions that had not previously been described. These include prolonged association and co-migration of sister cells after division ( Fig. 4 and movie 4a1 and 2), as well as of unrelated cells (Fig. 4 and movie 4b). Some cells participate in extended, dynamic interactions ( Fig. 4 and movie 4c). Extension of pseudopodia in opposing directions produces the characteristic bipolar spindle formations of muscle cells, and we observe both bipolar and unipolar membrane extensions in migrating satellite cells. The cell body remains stable until one pseudopod releases, then the cell body moves rapidly toward the still attached    Table 2, supporting movie Information, and additional Quicktime movies for more detailed information. pseudopod, resulting in incremental movement along the fiber. Some cells also move through the collagen matrix, in which case we frequently observed ruffled lamellae at the distal edges of the leading pseudopods that appeared to be sampling the surrounding environment ( Fig. 4 and movies 4d and 4e). While no labeling has been done to confirm the identity of these cells as satellite cells, and we therefore cannot rule out the possibility that they represent a different cell type, this would be consistent with in vivo data suggesting that satellite cells are capable of moving from one fiber to another [39].
We tabulated the frequency of each activity among 270 tracked cells in fiber cultures treated with FGF-2, and find that they occur with varying but relatively high frequency (Fig. 4F). An expanded version of this analysis may be found in Supporting Table 2, and full-length movies used to generate this table may be accessed at http://cornelisond.biology.missouri.edu/Cornelison_lab/movies/index. html. See the Supporting Movie Information file for complete descriptions of criteria used to score cell activities.

Addition of Hepatocyte Growth Factor to the 3D Collagen Matrix Increases Directional Motility Along the Myofiber
Motility requires sequentially establishing, then releasing, adhesive contacts between areas of the cell membrane to matrix substrates; soluble motogens are frequently the stimulus for adhesion receptor and cytoskeletal modifications that allow such transient adhesions. Soluble factors affecting myoblast motility include numerous well-studied cytokines, chemokines, and growth factors that would be expected to be available in the context of acute muscle damage. Hepatocyte growth factor (HGF), originally referred to as ''scatter factor'' because of its pro-motogenic effect on epithelial cells [40,41], has been shown to be essential for migration of somitic myoblasts to the developing limb bud [42,43]. It has been shown to be a potent motogenic factor for numerous primary cell types [44][45][46] including satellite cells [39,47]. HGF signaling through its high-affinity receptor, c-met, results in activation of the Ras-ERK MAP kinase cascade and the phosphatidylinositol 3-kinase (PI3K) pathway, both of which are independently required for cell migration [48][49][50]. Recent data suggest that these pathways converge in myoblasts to promote formation of both filamentous and branched actin structures (in filopodia and lamellopodia, respectively) by the actin-related protein 2/3 (Arp2/3) complex [51]. The small chemokine CXCL12, also referred to as stromal cell derived factor-1 (SDF-1), is the sole ligand for CXCR4, a seven-transmembrane pass receptor which has also been reported as a marker for quiescent and activated satellite cells [52]. During development, SDF-1 is a chemoattractant for emigrating myoblasts [53]. SDF-1 is secreted by resident fibroblasts in skeletal muscle and increases during injury [52] and would therefore be available to influence satellite cell activity. In addition to growth factors and cytokines, signaling lipids have been reported to stimulate migration, survival, and proliferation in different cell types [54]; significant effects have been observed using both S1P (sphingosine 1-phosphate) and LPA (lysophosphatidic acid).
FGF-2, HGF, SDF-1, S1P, and LPA were added at uniform concentrations across the entire well to the collagen matrix, singly or in combination, at bioactive concentrations [39,52,54,55]. In all conditions tested, all cells moved a minimum distance of 50 lm from their site of origin. The great majority were significantly more motile (traveling up to 2,500 lm in 24 hours), which is significantly longer than the maximum reported distance traveled by exogenous satellite cells 6 weeks after experimental engraftment (200 lm) [56] but consistent with reports of extensive endogenous satellite cell migration in vivo [8,10]. Figure 5A shows average motility from 24 to 48 hours after harvest for 255 individual fiberassociated satellite cells treated with FGF-2. When exogenous HGF or other potential motogens were included in the polymerized collagen matrix, they did not significantly affect the mean distance traveled (Fig. 5B), although longer-migrating cells are more common with HGF treatment (Fig. 5C). However, when tracings of individual satellite cell movement are compared, HGF treatment leads to striking increases in linear displacement compared to FGF-treated cells (Fig. 5D and  movies 5d1 and 5d2). When the tortuosity of cell tracks (rate and angle of turning) is compared between FGF2-and HGFtreated cells, HGF was found to produce significantly straighter paths as well (Fig. 5E). In the format used here (uniform addition to the culture), no other conditions led to significant changes in motility, although there were significant differential effects on cell proliferation (data not shown). While it is possible that a gradient or point source of cytokines would elicit directional or proximity-based responses, such experiments are not currently feasible to analyze in this system.

a7b1 Integrin Neutralization Specifically Inhibits Satellite Cell Motility on the Surface of Myofibers
We demonstrated in Figure 1 that binding to laminin via a7b1 integrin appears to be critical for satellite cell motility in monoculture. To determine whether this is an accurate reflection of motility on the host myofiber, we added neutralizing antibodies to several integrin a chains, individually and in combination with a neutralizing antibody directed against b1 integrin. Of the conditions tested, only neutralization of a7 ( Fig. 6 and movie 6a), b1 ( Fig. 6 and movie 6b), or both together ( Fig. 6 and movie 6c) had a negative effect on satellite cell motility in 3D myofiber culture (Fig. 6D); the effects of blocking both are also additive. This is in contrast to blockade of either a6 (also a specific laminin receptor), a5 (a fibronectin receptor), or aV, all of which actually increased satellite cell velocity. Blocking antibodies to a2, a4, and b2 also do not decrease satellite motility on myofibers, although all antibodies tested successfully blocked cell adhesion in monoculture as appropriate (data not shown). In addition, treatment with cyclic RGD peptides, which would block multiple integrins including aVb3, aVb1, aVb5, aVb6, aVb8, and aIIbb3, and which has been shown to block human satellite cell motility in monoculture [57], does not inhibit satellite cell migration on the myofiber. One could speculate that the cell-matrix interactions mediated by integrins other than a7b1 may be involved in adhesive interactions with the myofiber matrix, such that their inhibition permits an increase in cell velocity along the myofiber. Although motility is dramatically decreased, neither cell adhesion to the myofiber, cell proliferation, or cell-cell interactions appear to be affected by a7b1integrin blockade. These data support the conclusion that the primary in vivo substrate for satellite cell migration is the exterior lamina of the host myofiber, engaged by the a7b1 laminin receptor on satellite cells. as regulating guided cell migration, including semaphorins and plexins, Ephs and ephrins, netrins and Dcc/Unc5 family members, and Robos and Slits (reviewed in [58][59][60][61][62]). To assess whether satellite cell motility might involve responses to specific migration cues, we used RT-PCR and immunohistochemistry to evaluate their expression of guidance receptors and ligands. Surprisingly, we found that multiple guidance factors are strongly represented in activated satellite cells: mRNAs encoding members of all four major classes of guidance receptors were detected, as well as ephrin and semaphorin ligands (Fig. 7A). Immunostaining of fiber-associated satellite cells confirmed that guidance receptor expression is specific to satellite cells: Figure 7B shows localization of Robo1 on fiber-associated satellite cells, but not on the fiber itself.

DISCUSSION
As the resident stem cell of skeletal muscle, satellite cells are of interest both as activatable adult stem cells and as potential vectors for cell or gene therapy. Current work on the ''life cycle'' of satellite cells after injury is generally focused on the molecular mechanisms mediating activation from the quiescent state, proliferation to form a pool of myoblasts sufficient to replace myonuclei lost to damage or disease, and dif-ferentiation and fusion to regenerate the injured myofibers. The generation (and re-generation) of quiescent satellite cells remains the focus of significant research interest as well. However, inquiry into a potential role for satellite cell motility or migration in vivo is comparatively scarce. This is largely due to technical difficulties in visualizing satellite cells dynamically within the muscle tissue, but may also be influenced by the observations that individual satellite cells are sufficiently proliferative to be able to fully replace damaged muscle locally [63] as well as the comparatively minimal motility observed in vitro, particularly in experiments using C2C12 cells as a model. In this report, we used a 3D timelapse culture system to assess the movement of primary satellite cells on their host myofibers and to begin to characterize the molecular and cellular mediators of satellite cell motility.
When we compared primary mouse satellite cells with the two most frequently used myogenic cell lines, C2C12 and MM14, in monoculture on purified matrix components including collagen, fibronectin, and laminin, we observed cell typespecific differential morphology, adhesion, and motility on different substrates. Most striking were the differences in motility between primary cells and C2C12 cells, which are frequently used as a satellite cell model in migration and adhesion assays. Both qualitative differences in integrin chain expression and quantitative differences in the relative amounts of expression were observed between primary adult myoblasts and C2C12 cells. This leads us to hypothesize that satellite cells may have the ability to interact with a much broader range of substrates than was previously considered, potentially including unfolded proteins, cell adhesion molecules such as ICAM, VCAM, and MadCAM, as well as ECM components not commonly studied in the context of skeletal muscle such as fibrinogen, von Willebrand's factor, vitronectin, thrombospondin, and osteopontin. This may have important implications for understanding satellite cell adhesion and motility in the context of damaged and regenerating muscle in vivo. The lack of many of these more uncommon integrins in muscle cell lines, together with the increased motility of primary cells, suggests that studies performed on cell lines may underestimate aspects of satellite cell migration, particularly in vivo.
In the novel context of this 3D timelapse system, we made qualitative observations of previously unreported satellite cell behaviors. In particular, the tendency of some sister cells to maintain physical contact for long periods of time after division and across long migration distances raises a caveat for interpretations of differential staining among adjacent cells in fixed fiber preparations. Maintenance of stem/progenitor cells by asymmetric cell division is an important and wellestablished paradigm in adult stem cells, including satellite cells. Asymmetric distribution of cell-surface proteins [64], expression of nuclear transcription factors [65], or segregation of parental DNA strands [66,67] in muscle satellite cells can be used to identify cells generating daughters with different identities, such as a progenitor cell and a more committed myogenic cell. When observed in the context of the host myofiber in vivo, these asymmetric divisions also appear to occur in a planar orientation [65]. Particularly given that such methods exist to validate bona fide asymmetric cell divisions, our data suggest that more rigorous criteria than the appearance of differently staining, adjacent cells in static Quantitative representation of satellite cell motility in 3D myofiber culture after treatment with neutralizing antibody to a6 integrin, a7 integrin, aV integrin, b1 integrin, cyclic RGD, and pairwise combinations. Neutralization of a7 and b1 integrin, but neither neutralization of a6 or aV nor cyclic RGD, decreases satellite cell motility. * ¼ p < .05; ** ¼ p < .01; *** ¼ p < .001. Number of cells tracked per condition ranged from 6 to 35. preparations should be applied to verify such an event. We were also intrigued to note that some cells detach from their host fiber and migrate through the collagen matrix, in many cases moving in the direction of other myofibers embedded in the same well. These cells can be detected re-joining a fiber, after which they change their morphology from an elongated, polar appearance to the more rounded shape more often seen in fiber-associated cells. It is worth noting that these behaviors have all been observed during a relatively short timeframe: in our hands, satellite cells are activated as a consequence of fiber harvest, and most cultures were analyzed during the 24-48 hour period after fiber isolation.
It is important to note that while we have preserved one critical aspect of the in vivo environment (the interaction of a primary cell with a physiological substrate) numerous other known and unknown components of a true in vivo system are absent. The influence of extracellular matrix molecules associated with the areas between myofibers, of adjacent myofibers themselves, and of soluble factors that are not replicated in our system represent a significant caveat in contemplating whether these activities and behaviors occur in vivo, with what frequency, and whether they differ between, for example, healthy and dystrophic muscle regeneration. In addition, it would be intriguing to assess whether the observed behav-ioral heterogeneity in vitro is correlated to functional and phenotypical heterogeneity in vivo. As in vivo imaging capabilities increase, ideally it will eventually be possible to replicate these studies in situ.
When we stimulated fiber-associated satellite cells with soluble factors that have previously been described as motogenic factors for myoblasts in other contexts, we were surprised to observe that there were no significant effects on cell velocity, although there were changes in cell proliferation (data not shown). This may be due to the release of saturating amounts of cytokines from the myofibers and/or satellite cells present in the culture, the stage of the satellite cells being examined, or an effect of adhesion to the fiber rather than a tissue culture substrate. In particular, the potential contribution of soluble factors released from the myofibers or changes in their adhesive properties due to the culture conditions or changes in myofiber integrity (including fiber death or semihypercontraction) should not be discounted. However, while it did not significantly increase total velocity, HGF alone did produce a striking increase in cell displacement: cells treated with HGF traveled greater distances from their origin and moved more directly, compared to cells stimulated with FGF or other cytokines. In light of the critical roles in satellite cell activation and proliferation already associated with HGF  Table 1. (B): Immunohistochemistry for Robo1, a transmembrane guidance molecule receptor (Abcam); expression is exclusive to the satellite cells and appears localized to the satellite cell membrane contacting the myofiber surface. Scale bar ¼ 100 lm. Abbreviation: HGF, hepatocyte growth factor. signaling, it is intriguing to postulate a role for HGF in facilitating long-distance migration of activated satellite cells as well.
Distinct from simple motility, directed migration and pathfinding are mediated by the activity of cell surface guidance receptors. Signaling through these diverse receptors results in either attraction or, more frequently, repulsion, organizing cytoskeletal changes at the leading edge of motile cells. While a large and detailed body of literature is available in the context of other cell types, particularly neural crest, neurons, and endothelial cells, comparatively little has been published in muscle and there has been no suggestion that satellite cells are capable of directed guidance. While they have not been examined in the context of myoblast migration, several members of this class have been shown to be necessary for myogenic differentiation in vitro [68,69]. This may reflect a requirement for cell migration to aggregate and fuse to form myotubes [69,70], or a role in contact-mediated signaling distinct from migration guidance. After observing the morphological and directional changes that satellite cells display in single fiber culture, including apparent sampling and pseudopodial movement grossly similar to neuronal growth cones, we assayed primary satellite cells for expression of cell guidance receptors. Surprisingly, we found that all major classes of guidance receptors are represented, as well as two classes of guidance ligand, supporting the possibility that satellite cells may in fact actively migrate, presumably toward a site of local injury. The majority of the guidance receptors whose expression we detected on primary satellite cells usually elicit repulsive interactions when they bind their ligand (reviewed in [71]), suggesting that satellite cells might be maintained in a motile, transiently adherent state through their interactions with immobilized guidance cues.
We propose a model in which soluble motogens such as HGF, released by damaged areas of the myofiber, promote satellite cell motility, while repulsive interactions with comparatively intact areas of the myofiber maintain cell motility in the absence of injury. Injured myofibers, and/or immune cells recruited to the site of the injury, could be the source of an ad hoc chemoattractant gradient, while repulsive guidance cues would be lost on damaged areas of the fiber lamina. Future experiments in programmed matrices will test this model. In addition to suggesting an intriguing new functional process in these adult stem cells, these results may have important clinical implications for the design and execution of satellite cell-based therapies.