Live cell imaging reveals marked variability in myoblast proliferation and fate
© Gross and Rotwein; licensee BioMed Central Ltd. 2013
Received: 22 January 2013
Accepted: 28 March 2013
Published: 2 May 2013
Skip to main content
© Gross and Rotwein; licensee BioMed Central Ltd. 2013
Received: 22 January 2013
Accepted: 28 March 2013
Published: 2 May 2013
During the process of muscle regeneration, activated stem cells termed satellite cells proliferate, and then differentiate to form new myofibers that restore the injured area. Yet not all satellite cells contribute to muscle repair. Some continue to proliferate, others die, and others become quiescent and are available for regeneration following subsequent injury. The mechanisms that regulate the adoption of different cell fates in a muscle cell precursor population remain unclear.
We have used live cell imaging and lineage tracing to study cell fate in the C2 myoblast line.
Analyzing the behavior of individual myoblasts revealed marked variability in both cell cycle duration and viability, but similarities between cells derived from the same parental lineage. As a consequence, lineage sizes and outcomes differed dramatically, and individual lineages made uneven contributions toward the terminally differentiated population. Thus, the cohort of myoblasts undergoing differentiation at the end of an experiment differed dramatically from the lineages present at the beginning. Treatment with IGF-I increased myoblast number by maintaining viability and by stimulating a fraction of cells to complete one additional cell cycle in differentiation medium, and as a consequence reduced the variability of the terminal population compared with controls.
Our results reveal that heterogeneity of responses to external cues is an intrinsic property of cultured myoblasts that may be explained in part by parental lineage, and demonstrate the power of live cell imaging for understanding how muscle differentiation is regulated.
Muscle regeneration following injury occurs through stimulation of muscle stem cells, termed satellite cells . Once activated, satellite cells proliferate to repopulate the injured area, and then exit the cell cycle to differentiate and eventually fuse to form new myofibers [1, 2]. A similar series of steps occurs during muscle differentiation in culture. Yet, in both situations not all cells exposed to the same milieu have the same outcome. Some myoblasts continue to proliferate, others die, and another fraction becomes quiescent [3–6]. Because proliferation and death can occur simultaneously within a population, and can skew the fraction of cells that ultimately differentiate, it has been challenging to determine why some cells adopt one fate rather than another.
Muscle differentiation in culture has been studied primarily using endpoint assays that average cellular responses across the entire population. These assays require analyzing different cohorts of cells at different times and have inherently low temporal resolution. Furthermore, most endpoint assays assume homogeneity across the entire population. This assumption has been increasingly questioned by single cell measurements in other systems that find extensive variability within a population with regard to several critical parameters, including levels of gene or protein expression, responses to growth factor-activated signaling pathways, cell-cycle progression, and viability [7–11].
Live cell imaging resolves several limitations inherent in endpoint assays by allowing the same cells to be tracked with high temporal and spatial fidelity. This significantly improves the amount and quality of acquired data [12, 13]. Furthermore, when combined with lineage tracing, live cell imaging can lead to insights regarding how cell fate decisions occur . These approaches are especially important when identifying mechanisms controlling differentiation, in which a decision regarding the outcome of individual cells could be based on a niche signal, but could also be heritable or stochastic [14, 15].
Here we have used live cell imaging and lineage tracing to assess both proliferation and the early phases of differentiation in the C2 muscle cell line. Our results reveal marked variability in both lineage size and fractional survival, but remarkable homogeneity within individual lineages in terms of cell fate and behavior. We also assessed the impact of IGF-I treatment, and found that although myoblast proliferation and survival increased, cell fate remained similar within lineages. These experiments suggest that myoblast fate is not stochastic, and provide an approach for discerning how various treatments might alter satellite cell behavior and function.
Fetal and newborn calf serum was purchased from Hyclone (Logan, UT, USA). Horse serum, goat serum, Dulbecco’s modified Eagle’s medium (DMEM), and PBS were from Life Technologies (Carlsbad, CA, USA). Porcine gelatin was from Sigma (St. Louis, MO, USA), Hoechst 33258 nuclear dye, from Polysciences (Warrington, PA, USA), and R3-IGF-I from GroPep (Adelaide, Australia). The primary antibody to troponin-T (CT3 from J. J-C. Lin) was purchased from the Developmental Studies Hybridoma Bank (Iowa City, IA, USA), and the secondary antibody, AlexaFluor 594-conjugated-goat anti-mouse IgG, was from Life Technologies. C2 myoblasts were obtained from Yaffe and Saxel , and HEK293FT cells were from Life Technologies. Other chemicals were reagent grade and were purchased from commercial suppliers.
A recombinant lentivirus was generated to express enhanced green fluorescent protein (EGFP) under control of the EF-1α promoter using as a base Addgene plasmid #12258 (Cambridge, MA, USA). Lentivirus was prepared in HEK293FT cells and purified as described [17, 18]. Prior to use the virus was diluted in DMEM plus 2% fetal calf serum, and filtered through a 0.45 μM Gelman syringe filter (Pall Life Science, Ann Arbor, MI, USA).
C2 myoblasts were grown and expanded on tissue culture plates coated with 0.2% gelatin in growth medium (DMEM, 10% heat-inactivated fetal calf serum, and 10% heat-inactivated newborn calf serum), as described [19, 20]. For generation of EGFP-expressing C2 cells, myoblasts were transduced with the EGFP lentivirus as indicated . Over 90% of cells expressed the recombinant protein, and EGFP expression persisted at comparable levels for more than five additional passages. EGFP-positive and control C2 myoblasts were grown separately and mixed at a 1:4 ratio prior to plating for live cell imaging. Using a mixed population of myoblasts at this ratio makes it possible to track labeled cells in dense populations, and is also amenable to automated tracking, which was not possible from bright field images. For live cell imaging experiments, cells were plated on 6-well plates and then immediately placed in the IncuCyte FLR (Essen Biosciences, Ann Arbor, MI, USA), a microscopy system located inside a standard tissue culture incubator. The incubator was maintained at 37°C in humidified air with 5% CO2. Bright field and EGFP images were acquired at 10× magnification from four locations per well at 15-min intervals in order to accurately and completely track all labeled cells. The four locations, which were predefined by the imaging system and consistent across all experiments, were arranged as a square, with each point equidistant from the midpoint of the well. After 24 h in growth medium cells were washed with PBS, and differentiation medium (DM, DMEM with 2% horse serum) was added. For selected wells, R3-IGF-I [2 nM] was added with DM.
To quantify cell number a module was created using the Cell Profiler software program  that loaded EGFP images as a batch, converted images to grayscale, and performed an illumination correction. EGFP-positive cells in each field were identified by applying a background adaptive threshold, which separates primary objects from the background by setting a threshold at twice the value of the mode of the histogram for pixel intensity. Objects identified by the algorithm that were below 15 microns in diameter were discarded from the automated count. Cell confluence was calculated for each field using an IncuCyte algorithm, which quantifies the relative cell area in each bright field image.
To analyze muscle differentiation, plates that had been imaged for 90 h (24 h in growth medium plus 66 h in DM) were washed with PBS, fixed with paraformaldehyde for 10 min, and washed again with PBS followed by a 90-s treatment with 50% acetone - 50% methanol, as described [19, 20]. After three additional PBS washes wells were incubated with 0.25% goat serum in PBS for ≥2 h to block non-specific antibody binding, followed by incubation overnight at 4°C with troponin-T primary antibody (1:100 dilution), washing with PBS, and incubation for 90 min at 20°C with AlexaFluor 594-conjugated-goat anti-mouse IgG (1:3,000 dilution) and Hoescht nuclear dye. Cells were visualized with a Nikon Eclipse Ti-U inverted microscope and a Nikon DS-Qi1Mc camera using the NIS elements 3.1 software.
To assess observed cell viability data, we first calculated the number of living and dead cells among sibling pairs. Cells that failed to divide, or that had a sibling that underwent a second division, were not entered into this analysis since they had no comparable sibling. This excluded 16 cells, which had a percent survival of 25%. Assuming random death between pairs of cells, we calculated the expected number of pairs composed of two living myoblasts, a living and a dead cell, or two dead cells. Expected frequencies were then compared to observed data using a χ-squared test with two degrees of freedom. To test for a relationship between a variable in sibling pairs (cell cycle duration or time to death), we calculated the Pearson correlation coefficient. To test for a correlation between random cells, we randomized the pairings and performed the same test. Results were considered statistically significant when P ≤0.01.
Tracking cells beyond 60 h revealed that EGFP-positive myoblasts fused with both EGFP-expressing and non-labeled cells to form multinucleated myotubes (Additional file 2: Figure S2). These results were confirmed by identifying troponin-expressing cells by immunocytochemistry (Additional file 2: Figure S2). Thus, neither EGFP expression nor live cell imaging compromised muscle differentiation.
Since confluence is frequently used to establish when DM is added, we tracked confluence and compared it to measures of cell number. Immediately upon plating, confluence was approximately 20% and cell number was approximately 25% of its maximum value (Figure 2C). Cells soon spread out and began to divide so that by 10 h in culture when the EGFP-positive myoblast number per field was approximately 30% to 40% of maximal, confluence had reached approximately 85% to 95% (Figure 2C). By 24 h in growth medium when the cell number was approximately 70% to 80% of maximal, confluence was approximately 99%, and it remained constant despite a further rise in myoblast number (Figure 2C). Thus, confluence and cell number are poorly correlated.
Our automated counting algorithm measured changes in cell number, but was unable to quantify individual instances of cell death or division. In order to quantify death and division, we manually tracked myoblasts and their progeny over a 60-h incubation period. Both cell division and death could be readily detected and monitored (Figure 1 and Additional files 3 and 4: Movie). During cell division, cells condensed into a circular shape, which was followed by mitosis and emergence of two progeny (Figure 1A). Cell death was detected by shrinkage, blebbing, lysis, and the ultimate loss of EGFP fluorescence (Figure 1B). Comparing manual and automated measures of the total cell number revealed similar kinetics, thus validating the automated cell counting algorithm (Additional file 1: Figure S1A, B).
Despite the range of cell cycle durations in the population, there was a remarkably close correlation between siblings (Figure 5C). This relationship was not detected between cells paired randomly (Figure 5D). Between siblings the Pearson correlation coefficient for cell cycle duration was 0.85 (P = 1.834e-13), but between random pairs of cells it was -0.07 (P = 0.631).
Similarly, even though incubation with IGF-I reduced the percentage of cells that died (Figure 3), concordance among siblings was 75% (50% both living and 25% both dying, Additional file 6: Figure S4). This bias toward concordant sibling fates was nearly identical to that observed in cells incubated with DM alone (Figure 6A), despite the percentages of both myoblasts living and both dying being reversed (P <0.0001). These results indicate that survival was not purely stochastic, but instead was biased by parental lineage.
When the time from last division to death was tracked between concordant siblings (Figure 6B), we found a close correlation similar to that seen with cell cycle duration, further reinforcing the importance of parental lineage. The Pearson correlation coefficient for time to death between siblings was 0.72 (P = 2.247e-12), while by contrast between random cells the value was 0.12 (P = 0.3271) (Figure 6C).
To assess how heterogeneity in lineage size or survival might be reflected in the total population after a differentiation time course, we plotted the number of living myoblasts in each lineage over time, grouping lineages according to outcome. We found that the population was evenly represented by each of the founder cell lineages during incubation in growth medium, but not after addition of DM. One group of myoblasts, comprising approximately 40% of the initial population (Figure 7C, green tracing), maintained a similar representation for the entire culture period, while another equivalently sized group of founders failed to have one cell survive after incubation in DM (Figure 7C, red). In contrast, a third group significantly expanded from approximately 20% of the initial population to approximately 60% of the final cohort (Figure 7C, red). Thus, the overall population at the end of the experiment differed substantially from the population at the start.
In myoblasts incubated in DM plus IGF-I the relative number of lineages in each group was different. IGF-I treatment resulted in only 20% of founders not being represented in the final population, and 57% of founders comprised 85% of the final group (Figure 7D). Thus, addition of IGF-I in DM maintained the myoblast lineage distribution so that it more closely resembled the population at the start.
Here we have used live cell imaging and lineage tracing to address the dynamics of muscle cell proliferation and survival in the C2 myoblast cell line. We find a wide variation in the rate and extent of both proliferation and viability of myoblasts derived from different parental cells, but concordant behavior in cells arising from the same parents. As a consequence, the population of myoblasts undergoing differentiation varied substantially from the cells present at the start of an experiment. Addition of IGF-I to DM reduced population heterogeneity primarily by sustaining myoblast viability, and thus increased the number and sizes of surviving lineages. As a result, the terminal population more closely resembled the cohort of myoblasts at the start than it did in untreated cells. Our observations reveal that under standard treatment protocols extensive heterogeneity is an intrinsic property of cultured myoblasts, and that an effect of IGF-I is to decrease this variability.
We found that cell cycle durations were heterogeneous across the population and that cell division continued after DM was added. The average cell cycle duration of the population in growth medium was 14.2 h. This value matches doubling times of C2 myoblasts obtained by other approaches, including labeling of DNA synthesis and direct cell counting , but was easier to acquire and potentially more accurate, since individual myoblasts were tracked rather than averaging multiple time points from different groups of cells. Despite agreement of our data with previous observations, we found that the cell cycle duration of individual myoblasts varied extensively (range, 8–26 h), and that approximately 10% of the founder cells failed to divide once over the entire 60-h tracking period. Remarkably, these observations are similar to results obtained with satellite cells derived from dissociated single muscle fibers , where both the onset of proliferation and individual cell cycle durations varied in the population (range, 5.1-17.8 h), and approximately 16% of cells failed to divide once .
By assessing myoblast lineage, we observed that there was a close correlation between the cell cycle durations of siblings (Figure 5C), in remarkable agreement with an observation noted more than 25 years ago in primary quail myoblasts . These similarities between cells of shared parentage in both primary myoblasts and a muscle cell line point to the potential importance of heritability and the immediate environment in regulating cell fate.
Myoblast differentiation requires exit from the cell cycle in G1 [30–32]. Since our data showed that cell division continued well after addition of DM in a fraction of cells, this indicates that the onset of muscle differentiation is heterogeneous. Furthermore, since IGF-I treatment prolonged the time of cell division, it is likely to increase the duration over which cells exit the cell cycle (Figure 3B). This problem of variability is further compounded by methods that rely on confluence to mark the time when DM should be added [33, 34], since confluence is relatively difficult to visually quantify, and as seen here, small changes in confluence can equate to large differences in cell numbers (Figure 2B).
It is well known that a fraction of cultured myoblasts succumb to apoptotic cell death during incubation in DM [23–25, 35]. Similarly, it has been reported that in response to muscle injury in vivo, satellite cell proliferation is followed by a period of satellite cell death . We found that the net decline in cell number following 36 h in DM was approximately 35%. This is in line with the 20% to 30% value previously reported with endpoint methods, including TUNEL assays, cell counting, and ‘live-dead’ staining [25, 35]. Yet, by tracking both the division and death of individual cells, we found that over 60% of the population died during 36 h in DM, with the majority of death occurring during the initial 15 h (Figure 3C). This value differed substantially from cell number calculations because of concurrent proliferation of other EGFP-labeled myoblasts during incubation in DM. Thus, we suggest that traditional apoptosis assays underestimate the extent of myoblast death by failing to account for ongoing proliferation of other cells in the culture.
By identifying and tracking individual myoblasts and their offspring, we found that cell death was not random. Rather, siblings were biased toward adopting concordant fates after incubation in DM. The tendency toward common outcomes was maintained even in cells exposed to IGF-I, despite enhanced myoblast survival from IGF-I treatment (Figure 3C, Additional file 4: Figure S4). The mechanisms responsible for these biases are unclear, but could arise from genetic or epigenetic differences, or from environmental influences. We suspect that a fraction of the bias may be explained by similar levels or activity of the IGF-I - PI3-kinase - Akt signaling pathway between related cells, since exposure to IGF-I reduced myoblast death, but maintained concordant fates between siblings. It is therefore possible that cells of shared parentage inherit similar amounts of signaling components, and/or share epigenetic or genetic alterations that affect regulation of this pathway. This is consistent with observations that cell siblings adopt concordant fates in response to apoptosis-inducing agents because of a common inheritance of proteins from their mother [7, 8]. Alternatively, as siblings share a similar microenvironment, we cannot exclude the possibility that paracrine factors also contribute to the regulation of cell survival.
The main impact of cell death not being random was a dramatic change in the composition of the myoblast population by the end of the culture period. This was not apparent during the initial 24 h of incubation in growth medium because myoblast viability was complete and most of the cells underwent at least one cell division (Figure 5). Variability arose, however, during the subsequent 36 h in DM, as distinct subpopulations developed rapidly from heterogeneous cell division coupled with variable survival. This led to substantial differences in the contributions of different lineages to the final myoblast population (Figure 6A, 7, Additional file 6: Figure S4). Our results suggest that measurements that average cellular characteristics during a differentiation time course, such as immunoblots or gene expression assays, can obscure the properties of subpopulations.
IGF-I exerts potentially contradictory effects on muscle cells, including promoting both proliferation and differentiation . Our observations suggest one resolution to this problem. Analysis of the onset of the last division revealed that IGF-I led to an average delay of approximately 5 h compared with untreated controls (Figure 3B, Additional file 6: Figure S4). As this delay did not lead to more than one additional cell division, our interpretation is that the main action of IGF-I is to maintain myoblast survival so that otherwise vulnerable cells are able to complete a single final round of replication. These effects of IGF-I complicate comparisons with untreated cells, as both fractional myoblast survival and the starting points for differentiation are different. Future applications of reporters for different aspects of differentiation are needed to improve our understanding of the kinetics and regulation of muscle differentiation by separating out these confounding factors.
An important question in skeletal muscle biology is how satellite cell fate is regulated during muscle regeneration. Following injury, satellite cells must divide sufficiently to insure adequate numbers of differentiating myoblasts for immediate muscle repair, but also must maintain a reserve population for regeneration after subsequent injury [4, 6, 36]. Thus, multiple cell fate decisions are necessary to ensure adequate current and future muscle repair. Live cell imaging has begun to be applied to this question, and has revealed the importance of asymmetric and symmetric satellite cell divisions [14, 37, 38]. Other studies have shown that the satellite cell population is very heterogeneous, not only in terms of its behavior in response to proliferative or migratory cues [28, 39], but also with respect to cell of origin , and surface protein expression [6, 40]. Thus both environmental and genetic factors potentially play influential regulatory roles in muscle regeneration through effects on satellite cells. Elucidating how this heterogeneity impacts the decisions that lead to satellite cell differentiation will be critical to understanding the dynamics of muscle regeneration and how under certain circumstances such as aging this process can go awry.
We have used live cell imaging and lineage tracing to assess both proliferation and the early events of differentiation in C2 myoblasts. Our results reveal marked variability in lineage size and fractional survival, but remarkable homogeneity within individual lineages in terms of cell fate. Treatment with IGF-I increased myoblast number by maintaining viability and by stimulating a fraction of cells to complete one additional cell cycle in DM, and as a consequence reduced the variability of the terminal population compared with controls. Our results reveal that heterogeneity is an intrinsic property of cultured myoblasts, and demonstrate the power of live cell imaging to provide insights into the regulation of muscle differentiation.
Dulbecco’s modified Eagle’s medium
Enhanced green fluorescent protein
Insulin like growth factor-I
Phosphate buffered saline.
We thank Trey Knollman of Essen Bioscience for the use of IncuCyte imaging system. This work was supported by NIH research grant 5R01 DK042748 (to PR) and NIH Training Grant T32 CA106195 (SMG).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.