Information

What is a myotube?

What is a myotube?


We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

If I understand correctly, the following images show the main components in a human skeletal muscle:

From Life: The Science of Biology:

From Human Physiology/The Muscular System in wikibooks:

However, both resources don't mention the word "myotube" even once.

Wikipedia's Myogenesis page says:

Muscle fibers generally form the fusion of myoblasts into multi-nucleated fibers called myotubes.

But I don't know how to parse this sentence.

To Add to my confusion, searching for "myotube" in Wikipedia redirects me to Wikipedia's Muscle page.

So my question is:
What exactly is a myotube?
Could you point it out in one of the images above?


A myotube is a type of cell which will develop into a muscle fiber. It is formed by the fusion of multiple myoblasts, and thereby acquires multiple cell nuclei. The nuclei are in the middle of the myotube cell, unlike the mature muscle fiber where the nuclei are at the periphery of the cell. A myotube has a tubular form instead of the more typical rounded cell form of the myoblast. The myotube starts with few or no myofibrils, and goes through a long, gradual development which assembles filaments, myofibrils and other structures of a muscle fiber. The transition into a mature muscle fiber is not distinct, but might be defined as the point where all nuclei have migrated to the periphery of the cell.

More concisely, the Dictionary of Cell and Molecular Biology entry for myotube says:

(biology) Elongated multinucleate cells (three or more nuclei) that contain some peripherally located myofibrils.

They are formed in vivo or in vitro by the fusion of myoblasts and eventually develop into mature muscle fibres that have peripherally located nuclei and most of their cytoplasm filled with myofibrils. In fact, there is no very clear distinction between myotubes and muscle fibres proper.

Since the diagrams given in the question are for mature muscle, they do not show myotubes but instead show their result: muscle fibers.


Myotube is A skeletal muscle fiber formed by the fusion of myoblasts during a developmental stage;

a few myofibrils occur at the periphery, and the central core is occupied by nuclei and sarcoplasm so that the fiber has a tubular appearance

and u have the diagrams.


FGF signaling directs myotube guidance by regulating Rac activity

Nascent myotubes undergo a dramatic morphological transformation during myogenesis, in which the myotubes elongate over several cell diameters and are directed to the correct muscle attachment sites. Although this process of myotube guidance is essential to pattern the musculoskeletal system, the mechanisms that control myotube guidance remain poorly understood. Using transcriptomics, we found that components of the Fibroblast Growth Factor (FGF) signaling pathway were enriched in nascent myotubes in Drosophila embryos. Nullmutations in the FGF receptor heartless (htl), or its ligands, caused significant myotube guidance defects. The FGF ligand Pyramus is expressed broadly in the ectoderm, and ectopic Pyramus expression disrupted muscle patterning. Mechanistically, Htl regulates the activity of Rho/Rac GTPases in nascent myotubes and effects changes in the actin cytoskeleton. FGF signals are thus essential regulators of myotube guidance that act through cytoskeletal regulatory proteins to pattern the musculoskeletal system.


Abstract

Hydrogel-based three-dimensional (3D) cellular models are attractive for bioengineering and pharmaceutical development as they can more closely resemble the cellular function of native tissue outside of the body. In general, these models are composed of tissue specific cells embedded within a support material, such as a hydrogel. As hydrogel properties directly affect cell function, hydrogel composition is often tailored to the cell type(s) of interest and the functional objective of the model. Here, we develop a parametric analysis and screening method to identify suitable encapsulation conditions for the formation of myotubes from primary murine myoblasts in methacryloyl gelatin (GelMA) hydrogels. The effect of the matrix properties on the myotube formation was investigated by varying GelMA weight percent (wt%, which controls gel modulus), cell density, and Matrigel concentration. Contractile myotubes form via myoblast fusion and are characterized by myosin heavy chain (MyHC) expression. To efficiently screen the gel formulations, we developed a fluorescence-based plate reader assay to quantify MyHC staining in the gel samples, as a metric of myotube formation. We observed that lower GelMA wt% resulted in increased MyHC staining (myotube formation). The cell density did not significantly affect MyHC staining, while the inclusion of Matrigel increased MyHC staining, however, a concentration dependent effect was not observed. These findings were supported by the observation of spontaneously contracting myotubes in samples selected in the initial screen. This work provides a method to rapidly screen hydrogel formulations for the development of 3D cellular models and provides specific guidance on the formulation of gels for myotube formation from primary murine myoblasts in 3D.


Results

Expression of netrins and netrin receptors in cultured myoblasts

Our interest in Ig/FNIII repeat proteins led us to examine expression of netrin receptors of this class during myogenic differentiation in vitro. Western blot analyses of C2C12 myoblasts revealed a uniform level of neogenin in cells cultured in growth medium (GM) and cells transferred into DM over a period of 4 d (Fig. 1 A). This contrasts with expression of CDO, which was transiently up-regulated during the differentiation time course (Fig. 1 A Kang et al., 1998). Duplicate blots probed with an antibody against DCC revealed only trace levels of this protein (unpublished data). F3, a myoblast line derived by treatment of 10T1/2 fibroblasts with 5-azacytidine, also expressed neogenin under both GM and DM conditions (Fig. 1 B). 10T1/2 fibroblasts and 10T1/2 cells converted to myoblasts by stable expression of MyoD expressed indistinguishable levels of neogenin (Fig. 1 C) again, this contrasts with CDO expression, which was induced by MyoD (Fig. 1 C Kang et al., 1998). Finally, expression of oncogenic Ras in C2C12 cells, which results in reduction of MyoD and CDO and blocks differentiation (Kang et al., 1998), had no effect on neogenin levels (unpublished data). Together, these results indicate that neogenin is expressed in the myogenic lineage, but its expression is not dependent on myogenic factors.

Specific netrins were examined in a similar fashion. Netrin-3 was expressed in both C2C12 and F3 cells under GM conditions and levels increased when cells were transferred into DM (Fig. 1, A and B). This increase was also apparent at the mRNA level, with C2C12 cells displaying a single netrin-3 transcript of ∼7 kb (Fig. 1 D). In contrast, duplicate Western and Northern blots probed for expression of netrin-1 protein and mRNA, respectively, did not reveal a signal, despite the ability to detect netrin-1 produced transiently (unpublished data). Netrin-3 produced by C2C12 cells was not observed in the culture medium supernatant (unpublished data), suggesting that, like endogenous netrins-1 and -2 in chick floor plate and netrin-1 in adult rat spinal cord, it is likely to be associated with cell membranes and/or ECM (Serafini et al., 1994 Manitt et al., 2001). 10T1/2 and 10T1/2-MyoD cells also expressed netrin-3 (Fig. 1 C). It is concluded that myoblast cell lines and 10T1/2 cells express both netrin-3 and one of its receptors, neogenin, raising the possibility that an autocrine signaling pathway may exist in these cells.

Neogenin levels regulate myotube formation

To assess a role for neogenin in myogenesis, we transfected different “strains” of C2C12 cells and F3 cells with a vector engineered to drive expression of a human neogenin cDNA and a puromycin resistance gene. Transfected cultures were selected and analyzed for neogenin expression. Stable overexpression of neogenin was achieved only in one strain of C2C12 cells. These cells, previously described by us and designated as C2C12(E) cells (Kang et al., 1998), are highly differentiation-proficient in that they: (a) expressed detectable myogenin before a shift into DM (b) were less sensitive to inhibition of differentiation by moderate levels of serum (e.g., 5% FBS) and (c) completed the differentiation process over a 2-d time period. These cells were used for all subsequent experiments and are simply referred to as C2C12 cells. Neogenin vector transfectants (C2C12/neogenin cells) produced two- to threefold more neogenin than control vector transfectants (C2C12/puro cells Fig. 2 A). Overproduction of neogenin modestly decreased proliferation of C2C12 cells in GM, relative to vector controls (Fig. S1), and did not alter their morphology in GM (not depicted). When observed 24 h after shifting the cultures to medium containing 5% FBS (a time at which the differentiation process was in mid-progression), C2C12/neogenin cells had formed larger myotubes with more nuclei than did C2C12/puro cells (Fig. 2 B). C2C12/neogenin cultures displayed an increase in the total number of nuclei present in myosin heavy chain (MHC)–positive cells and in the average number of nuclei/myotube (Fig. 2 C). When analyzed for expression of muscle-specific proteins, including MyoD, myogenin, MHC, and troponin T (TnT), C2C12/neogenin cells showed accelerated expression of TnT, but otherwise were similar to C2C12/puro cells (Fig. 2 D). Overexpression of neogenin therefore resulted in enhanced formation of myotubes without dramatic changes in several biochemical markers of differentiation.

To assess the effect of reducing neogenin levels on myotube formation, an RNAi approach was taken. A sequence from the mouse neogenin coding region was inserted into the pSilencer vector, and was cotransfected with a GFP expression vector into C2C12 cells the pSilencer vector without an insert was used as a control. Transfected cultures were sorted for the presence of GFP and assessed for reduction of neogenin levels. A representative Western blot is shown in Fig. 3 A. GFP-sorted cells that received the neogenin RNAi vector produced fewer and smaller myotubes than sorted control transfectants (Fig. S2). To more precisely quantify this effect, C2C12 cells were cotransfected with the neogenin RNAi vector or the vector lacking an insert, plus a plasmid directing expression of nlacZ (encoding nuclear-localized β-galactosidase β-gal). 2 d later, the cultures were transferred to 5% FBS for 24 h, and then fixed and double stained for MHC and β-gal activity. When control vector-transfectants fused with nontransfected cells, many (often most) of the nuclei in the myotube became positive for β-gal activity, presumably because the cytoplasmically translated protein diffused within the myotube (Fig. 3 B). When the number of nuclei in β-gal + cells was scored, ∼56% had more than five nuclei, ∼21% had two to four nuclei and ∼23% had a single nucleus (Fig. 3 C). In contrast, the distribution of β-gal + cells that received the neogenin RNAi vector was strongly skewed toward single nucleus-containing cells, with only about ∼5% displaying more than five nuclei (Fig. 3 C). Cells that received an RNAi vector containing a neogenin sequence that was ineffective at reducing neogenin protein levels behaved equivalently to those that received the vector lacking an insert (unpublished data).

As a further test of specificity, we sought to rescue the cell fusion defect produced by neogenin RNAi via ectopic expression of the human neogenin cDNA. The sequence of the RNAi used to target endogenous mouse neogenin is not fully conserved in the human cDNA, and therefore the human neogenin expression vector should be impervious to RNAi-mediated “knockdown”. Three conditions were studied: (1) C2C12 cells cotransfected with pSilencer lacking RNAi sequences and the control vector for human neogenin expression (pSil/pBP) (2) cells cotransfected with pSilencer containing neogenin RNAi sequences and the control vector for human neogenin expression (RNAi/pBP) and (3) cells cotransfected with pSilencer containing RNAi sequences and the human neogenin expression vector (RNAi/hNeogenin). For unknown reasons, the pSil/pBP transfectants reproducibly produced fewer myotubes with more than five nuclei than cells that received only the control pSilencer vector (∼30% vs. ∼56%, respectively compare Fig. 3 D with Fig. 3, C and E). Importantly, however, RNAi/pBP transfectants were strongly blocked from forming myotubes, whereas the RNAi/hNeogenin transfectants closely resembled the control pSil/pBP cultures (Fig. 3, D and E). Therefore, forced expression of neogenin rescued the effects of RNAi to neogenin, further indicating that the effects of the RNAi were specific. It is concluded that reduction of neogenin levels reduced myoblasts' ability to participate in the formation of myotubes. Together, with the overexpression data in Fig. 2, these results strongly suggest that neogenin levels are rate limiting for this aspect of C2C12 myogenesis.

Recombinant netrin promotes myotube formation

To assess the effects of netrin-3 on myoblast cell lines, we initially attempted to synthesize recombinant mouse protein in COS and 293T cells but, similar to the findings of Puschel (1999), were unable to produce sufficient quantities of secreted, soluble material. Mouse netrin-3 is orthologous to human netrin-2L and, although not orthologous to chicken netrin-2, is grouped with both these proteins by dendrogram analyses in a sub-category designated “netrin-2–like”, distinguishable from a “netrin-1–like” group (Puschel, 1999). Therefore, we chose to use recombinant chicken netrin-2 for this work as it is commercially available. C2C12 cells were cultured in GM, then transferred into 5% FBS plus or minus chicken netrin-2, and observed 24 h later. C2C12 cell proliferation was not altered by netrin-2 treatment (Fig. S3). However, although the control cultures displayed small MHC + myotubes under these conditions, the netrin-2–treated cultures showed distinctly larger myotubes with increases in the total number of nuclei in MHC + cells and in the average number of nuclei/myotube (Fig. 4, A and B). When analyzed for expression of muscle-specific proteins, the cells that received netrin-2 produced significantly more TnT than control cells, but the levels of MyoD, myogenin, MHC, and CDO were unchanged (Fig. 4 C). Thus, similar to overexpression of neogenin, treatment of C2C12 cells with a neogenin ligand resulted in enhanced myotube formation and alterations in expression of TnT, but not several other muscle markers.

To assess whether netrin-2 exerted its effects on myotube formation in a manner dependent on neogenin, a C2C12 cell derivative that stably expressed the neogenin RNAi vector was generated these cells expressed considerably less neogenin than control transfectants and produced fewer and smaller myotubes than control transfectants when shifted to DME/5% FBS for 24 h (Fig. 5, A–C). When the control cells were treated with netrin-2 under these conditions they behaved similarly to parental C2C12 cells, forming larger myotubes with more nuclei. In contrast, the neogenin RNAi-expressing cells were not affected by netrin-2 (Fig. 5, B and C). Therefore, it is concluded that the ability of netrin-2 to promote myotube formation by C2C12 cells depends on the netrin receptor, neogenin.

Neogenin activates myogenic bHLH factor– and NFAT-dependent reporter constructs

CDO signals to enhance myogenic bHLH factor–dependent transcription (Cole et al., 2004), and netrin-1 signals through DCC to stimulate NFAT-mediated transcription (Graef et al., 2003). Therefore, we assessed whether neogenin could stimulate reporter constructs specific for myogenic bHLH factors and NFAT. In transient assays of C2C12 cells, cotransfection of the neogenin expression vector enhanced activity of a reporter construct driven by four reiterated myogenic E-boxes (4Rtk-luc), approximately threefold above that seen with a control vector lacking a cDNA insert (Fig. 6 A). Cotransfection of 10T1/2 fibroblasts with expression vectors for MyoD and its heterodimeric partner E12 led to activation of 4Rtk-luc, and this activity was also enhanced approximately threefold by coexpression of neogenin (Fig. 6 B). However, neogenin could not activate the reporter in the absence of MyoD (unpublished data).

Cotransfection of the neogenin expression vector into either C2C12 or 10T1/2 cells enhanced NFAT-luciferase reporter activity approximately twofold in contrast, a CDO expression vector had no effect on NFAT-dependent transcription (Fig. 6, C and D). Of the various NFATc isoforms implicated in myogenesis, only NFATc3 is activated in differentiating C2C12 myoblasts (Delling et al., 2000). To assess whether netrin-neogenin signaling could activate NFATc3, C2C12 and 10T1/2 cells were treated with recombinant chicken netrin-2 for 6 or 30 h. Immunoblots probed with an antibody specific to NFATc3 revealed increased levels of NFATc3 after 6 h of netrin-2 treatment, including a more quickly migrating form that presumably represents dephosphorylated, activated NFATc3 (Fig. 6 E). This response returned to baseline, or lower, within 30 h. Consistent with this relatively short duration of action, addition of netrin-2 to control or neogenin-transfected cultures enhanced NFAT-luciferase reporter activity by only ∼40% in each case (unpublished data).

To assess whether the effects of netrin-2 on NFATc3 were dependent on neogenin, C2C12 cells were transiently transfected with neogenin RNAi or control vectors plus a GFP expression vector, sorted, and treated with medium containing 5% FBS, plus or minus netrin-2. Netrin-2 treatment of control transfectants resulted in increased levels of NFATc3, similar to parental C2C12 cells in contrast, neogenin RNAi transfectants failed to respond to netrin-2 (Fig. 6 F). The C2C12 derivative that stably expressed neogenin RNAi (Fig. 5) displayed a similar lack of response (Fig. 6 G). Thus, as seen with enhancement of myotube formation, netrin-2 required neogenin to exert its effects on NFATc3 levels.

Neogenin forms a complex with CDO

Ig/FNIII proteins can bind in a cis fashion to additional members of this family to form complexes that regulate their function. For example, Robo receptors bind to DCC to silence netrin-1–mediated attraction (Stein and Tessier-Lavigne, 2001), and BOC binds CDO to stimulate myogenesis (Kang et al., 2002). To test whether neogenin might also interact with CDO, lysates from C2C12 and F3 cells were immunoprecipitated with antibodies to CDO or neogenin and blotted with the reciprocal antibody. Neogenin was present in CDO immunoprecipitates, and CDO in neogenin immunoprecipitates, from both cell lines, suggesting that these two proteins do indeed form a complex (Fig. 7, A and B). More neogenin co-immunoprecipitated with CDO under DM than GM conditions, although this enhanced association in DM was not evident in the reciprocal co-immunoprecipitation (Fig. 7, A and B). Co-immunoprecipitation of neogenin and CDO was not diminished when C2C12 cells were collected as a single-cell suspension in the presence of EDTA (which blocks cadherin-mediated adhesion), suggesting that the interaction occurs in a cis fashion (Fig. 7 C). CDO associates with N- and M-cadherins in myoblasts (Kang et al., 2003), and cadherins also co-immunoprecipitated with neogenin (Fig. 7 B), suggesting these proteins may interact together in a higher order structure. Furthermore, consistent with netrin-3 functioning as a neogenin ligand in C2C12 cells, it too was observed in neogenin immunoprecipitates (Fig. 7 B).

To investigate complex formation between CDO and neogenin in more detail, transient transfections in 293T cells were performed. Neogenin was brought down by CDO antibodies when CDO was coexpressed, but not when CDO was omitted analogously, CDO was brought down by neogenin antibodies in a neogenin-dependent fashion (Fig. 7, D and E). As seen with C2C12 cells, collection of transiently transfected 293T cells as a single-cell suspension in EDTA did not diminish the association between neogenin and CDO (Fig. 7 F). Additionally, when 293T transfectants that expressed only CDO were cocultured with transfectants that expressed only neogenin, immunoprecipitation of either protein failed to coprecipitate the other, an interaction expected to be observed if the binding occurred in a trans manner (unpublished data). Together, the transient expression data: (a) establish the specificity of the antibodies (b) confirm the cis nature of the CDO-neogenin association and (c) indicate that additional, cell type–specific factors are not likely to be required for this interaction.

CDO contains an ectodomain comprised of five Ig and three FNIII repeats, a single pass transmembrane region, and a 270–amino acid cytoplasmic tail (Kang et al., 1997). To identify regions of CDO involved in complex formation with neogenin, a series of CDO deletion mutants that lack each individual Ig and FNIII repeat were tested for their ability to co-immunoprecipitate neogenin. 293T cells were transiently transfected with expression vectors for neogenin and either wild-type CDO or individual CDO mutants (designated by the symbol “Δ” followed by the deleted domain). Cell lysates were precipitated with antibodies to the CDO intracellular region, and the immunoprecipitates blotted and probed with antibodies to neogenin or CDO (Fig. 7 G). The CDO mutants ΔIg1 and ΔFN2 were deficient (though not completely defective) in their ability to associate with neogenin, relative to full-length CDO. In contrast, ΔIg2, ΔIg4, ΔIg5, and ΔFN1 displayed no significant reduction in this property. ΔFN3 gave somewhat variable results over multiple experiments, suggesting it may be involved in neogenin binding but is not as important as Ig1 or FN2. Deletion of Ig repeat 3 apparently resulted in destabilization of CDO, as only very weak expression of this protein was observed. A construct in which the CDO signal sequence was linked directly to its transmembrane and cytoplasmic regions (designated CDO(TMintra) Kang et al., 2002) was brought down in neogenin immunoprecipitates, albeit inefficiently, suggesting that the intracellular regions of CDO and neogenin may also associate (unpublished data). Together, the results in Fig. 7 are consistent with the conclusion that neogenin and CDO form complexes in cis and that this interaction is dependent on the presence of specific repeats in the CDO ectodomain. Despite its sequence similarity to, and ability to interact with, neogenin, CDO seems not to be an independent netrin receptor, in that it was unable to bind a netrin-1-Fc fusion protein under conditions where neogenin did. Furthermore, BOC was also unable to bind netrin-1-Fc, and coexpression of CDO and/or BOC did not appear to alter netrin-1-Fc binding to neogenin (unpublished data).

To gain insight into whether neogenin's ability to bind to CDO is important for netrin-neogenin signaling, primary myoblasts derived from wild-type mice and mice homozygous for a targeted mutation of Cdo were analyzed for their response to recombinant chicken netrin-2. Cdo + / + and Cdo/ − myoblasts express similar levels of MyoD, but during differentiation Cdo/ − cells produce lower levels of myogenin and form myotubes very inefficiently (Cole et al., 2004). Like myoblast cell lines, primary myoblasts expressed netrin-3 and neogenin, regardless of their Cdo genotype (Fig. 8 A). Before treatment with netrin-2, Cdo + / + and Cdo/ − myoblasts were similar in that each expressed somewhat higher levels of the more slowly migrating form of NFATc3 than the more quickly migrating, presumably activated, form (Fig. 8 B). After 3 h of exposure to netrin-2, the Cdo + / + cells showed a pronounced shift to the more quickly migrating form, consistent with activation of NFAT signaling in contrast, the Cdo/ − myoblasts showed no response and the ratio of the two forms of NFATc3 resembled that seen in untreated cells (Fig. 8 B). Therefore, loss of CDO resulted in loss of responsiveness to netrin-2 despite the presence of normal levels of its receptor, suggesting that neogenin's interaction with CDO is important for signaling by this pathway.


Model behavior? Myotubes and the quest for an ex vivo model of muscle

Understanding how muscles work is important, but dissecting muscle fibers from humans is difficult, could an alternative option be myotubes? Research published in Skeletal Muscle investigates this, and here Jennifer Levy explains more.

Like many things in life, biology experiments require balance. The researcher must balance the ease of the experiment with the relevance of the result. As a researcher becomes more reductionist in their experimental design, the experiment becomes more feasible.

For example, a nephrologist could design a study of kidney function within a model organism, in isolated kidneys, or in a kidney cell line grown in a petri dish. In fact, she can even study certain reactions in purified proteins in vitro.

With each of these reductions, however, as ease increases (smaller sizes, fewer complicating factors, less variability), the physiological impact of the result becomes more convoluted. Do cultured kidney cells behave exactly the same as kidney cells within a functioning kidney? Surely not.

A recent Skeletal Muscle manuscript by Olsson et al. confronts the issue of ease versus relevance in muscle fibers. The functional unit of human skeletal muscle is the muscle fiber. Researchers can learn valuable lessons about muscle biology and diseases by studying muscle fibers isolated from patients with musculoskeletal diseases, as well as healthy controls.

Differentiated myotubes

Intact muscle fibers can be dissected from humans, but this technique is difficult, often resulting in a very small numbers of viable fibers. While just a few viable fibers may be usable for certain techniques (imaging or patch clamping studies), others (such as biochemistry) require larger numbers of cells.

Therefore, researchers will isolate muscle precursor cells from biopsies, and then differentiate these into myotubes in culture dishes. These myotubes serve as a model for human adult skeletal muscle. Myotubes appear very similar to isolated muscle fibers – both are elongated, multinucleate and contract upon electrical stimulation. For these reasons, differentiated myotubes have become a commonly used tool for studying many of the molecular and physiological properties of muscle.

The importance of calcium

Olsson et al. chose to directly compare the calcium transients and subsequent contractions observed in myotubes to those observed in muscle fibers.

Regulation of intracellular calcium is crucial to skeletal muscle function. For one, the global wave of calcium that floods the muscle cell in response to membrane depolarization (aka the ‘calcium transient’) acts to couple surface membrane depolarization to muscle contraction.

And the intrinsic speed of the calcium transient is in turn modulated by the presence and localization of calcium-handling proteins. Olsson et al. chose to directly compare the calcium transients and subsequent contractions observed in myotubes to those observed in muscle fibers.

They found marked differences between the two. Upon electrical stimulation, nearly all of the intact muscle fibers showed transient increases in calcium and force-producing contractions whereas only about 50% of myotubes showed similar transient increases in calcium upon stimulation.

Further, the kinetics of these increases were altered, and they were not always accompanied by myotube contractions. The reason for this triggering misfire becomes clear as the authors go on to show that myotubes lack the arrangements of actin and myosin filaments necessary for contraction to occur. Additionally, a number of important calcium-handling proteins are downregulated or mislocalized in myotubes, compared to muscle fibers.

What can researchers do?

Is there anything a researcher can do to encourage a myotube to be more similar to a fiber? The authors of the current study do not address this experimentally, but previous work from Tanaka et al. show that human muscle cells that were innervated by being co-cultured with rat spinal cord showed continuous contraction and extensive cross-striations similar to intact muscle fibers. Innervation is likely the key to making a myotube more like a muscle fiber.

Overall, these results provide a cautionary tale for muscle researchers.

For researchers interested in the functional studies of human muscle fibers, innovation will be required. Olsson et al. demonstrate reasonable success in isolating fibers from intercostal muscle biopsies obtained during thoracotomies. The intercostal muscles are particularly amenable to fiber isolation due to the relatively short lengths of the fibers (0.5 – 1 cm). Biopsies from muscles containing longer fibers (such as limb muscles, which can be up to 30 cm) will likely prove difficult to adapt to this technique.

Overall, these results provide a cautionary tale for muscle researchers: myotubes are not an adequate model of adult muscle fibers when studying the details of calcium dependent processes. Future studies could likely show marked differences between myotubes and muscle fibers in additional cellular processes, as well.


Engel, A. G. & Fanzine-Armstrong, C. Myology 1937 (McGraw-Hill, New York, 1994).

Renganathan, M., Messi, M. L., Schwartz, R. & Delbono, O. FEBS Lett. 417, 13–16 (1997).

Adams, G. R. & Haddad, F. J. Appl. Physiol. 81, 2509–2516 (1996).

Goldspink, G. J. Anat. 194, 323–334 (1999).

Florini, J. R., Ewton, D. Z. & Coolican, S. A. Endocr. Rev. 17, 481–517 (1996).

Musaro, A., McCullagh, K. J., Naya, F. J., Olson, E. N. & Rosenthal, N. Nature 400, 581–585 (1999).

Semsarian, C., Sutrave, P., Richmond, D. R. & Graham, R. M. Biochem. J. 339, 443–451 (1999).

Rommel, C. et al. Science 286, 1738–1741 (1999).

Friday, B. B., Horsley, V. & Pavlath, G. K. J. Cell Biol. 149, 657–666 (2000).

Dudek, H. et al. Science 275, 661–665 (1997).

Svegliati-Baroni, G. et al. Hepatology 29, 1743–1751 (1999).

Yu, H. & Berkel, H. J. LA State Med. Soc. 151, 218–223 (1999).

Olson, E. N. & Williams, R. S. Cell 101, 689–692 (2000).

Wesselborg, S., Fruman, D. A., Sagoo, J. K., Bierer, B. E. & Burakoff, S. J. J. Biol. Chem. 271, 1274–1277 (1996).

Eves, E. M. et al. Mol. Cell. Biol. 18, 2143–2152 (1998).

Aman, M. J., Lamkin, T. D., Okada, H., Kurosaki, T. & Ravichandran, K. S. J. Biol. Chem. 273, 33922–33928 (1998).

Carver, D. J., Aman, M. J. & Ravichandran, K. S. Blood 96, 1449–1456 (2000).

Liu, Q. et al. Genes Dev. 13, 786–791 (1999).

Jacob, A., Cooney, D., Tridandapani, S., Kelley, T. & Coggeshall, K. M. J. Biol. Chem. 274, 13704–13710 (1999).

Bohni, R. et al. Cell 97, 865–875 (1999).

Huang, H. et al. Development 126, 5365–5372 (1999).

Leevers, S. J., Weinkove, D., MacDougall, L. K., Hafen, E. & Waterfield, M. D. EMBO J. 15, 6584–6594 (1996).

Montagne, J. et al. Science 285, 2126–2129 (1999).

Bodine, S. C. et al. Nature Cell Biol. 3, 1014–1019 (2001).

Glass, D. J. et al. Cold Spring Harb. Symp. Quant. Biol. 57, 53–62 (1992).

Azpiazu, I., Saltiel, A. R., DePaoli-Roach, A. A. & Lawrence, J. C. J. Biol. Chem. 271, 5033–5039 (1996).


ASJC Scopus subject areas

  • APA
  • Author
  • BIBTEX
  • Harvard
  • Standard
  • RIS
  • Vancouver

In: Journal of Cell Biology , Vol. 169, No. 2, 25.04.2005, p. 257-268.

Research output : Contribution to journal › Article › peer-review

T1 - Transcriptional regulation of myotube fate specification and intrafusal muscle fiber morphogenesis

AU - Tourtellotte, Warren G.

N2 - Vertebrate muscle spindle stretch receptors are important for limb position sensation (proprioception) and stretch reflexes. The structurally complex stretch receptor arises from a single myotube, which is transformed into multiple intrafusal muscle fibers by sensory axon-dependent signal transduction that alters gene expression in the contacted myotubes. The sensory-derived signal transduction pathways that specify the fate of myotubes are very poorly understood. The zinc finger transcription factor, early growth response gene 3 (Egr3), is selectively expressed in sensory axon-contacted myotubes, and it is required for normal intrafusal muscle fiber differentiation and spindle development. Here, we show that overexpression of Egr3 in primary myotubes in vitro leads to the expression of a particular repertoire of genes, some of which we demonstrate are also regulated by Egr3 in developing intrafusal muscle fibers within spindles. Thus, our results identify a network of genes that are regulated by Egr3 and are involved in intrafusal muscle fiber differentiation. Moreover, we show that Egr3 mediates myotube fate specification that is induced by sensory innervation because skeletal myotubes that express Egr3 independent of other sensory axon regulation are transformed into muscle fibers with structural and molecular similarities to intrafusal muscle fibers. Hence, Egr3 is a target gene that is regulated by sensory innervation and that mediates gene expression involved in myotube fate specification and intrafusal muscle fiber morphogenesis.

AB - Vertebrate muscle spindle stretch receptors are important for limb position sensation (proprioception) and stretch reflexes. The structurally complex stretch receptor arises from a single myotube, which is transformed into multiple intrafusal muscle fibers by sensory axon-dependent signal transduction that alters gene expression in the contacted myotubes. The sensory-derived signal transduction pathways that specify the fate of myotubes are very poorly understood. The zinc finger transcription factor, early growth response gene 3 (Egr3), is selectively expressed in sensory axon-contacted myotubes, and it is required for normal intrafusal muscle fiber differentiation and spindle development. Here, we show that overexpression of Egr3 in primary myotubes in vitro leads to the expression of a particular repertoire of genes, some of which we demonstrate are also regulated by Egr3 in developing intrafusal muscle fibers within spindles. Thus, our results identify a network of genes that are regulated by Egr3 and are involved in intrafusal muscle fiber differentiation. Moreover, we show that Egr3 mediates myotube fate specification that is induced by sensory innervation because skeletal myotubes that express Egr3 independent of other sensory axon regulation are transformed into muscle fibers with structural and molecular similarities to intrafusal muscle fibers. Hence, Egr3 is a target gene that is regulated by sensory innervation and that mediates gene expression involved in myotube fate specification and intrafusal muscle fiber morphogenesis.


What is a myotube? - Biology

Department of Computational Biology and Medical Sciences, Graduate School of Frontier Sciences, University of Tokyo

Laboratory of Computational Biology, Graduate School of Biological Sciences, Nara Institute of Science and Technology Department of Biological Sciences, Graduate School of Science, University of Tokyo

Department of Biological Sciences, Graduate School of Science, University of Tokyo

Department of Biological Sciences, Graduate School of Science, University of Tokyo Department of Engineering Science, Graduate School of Informatics and Engineering, University of Electro-Communications

Department of Biological Sciences, Graduate School of Science, University of Tokyo

Department of Functional Biology, Graduate School of Biostudies, Kyoto University

Department of Applied Chemistry, Graduate School of Engineering, University of Tokyo

Department of Computational Biology and Medical Sciences, Graduate School of Frontier Sciences, University of Tokyo Department of Biological Sciences, Graduate School of Science, University of Tokyo CREST, Japan Science and Technology Corporation

2018 Volume 43 Issue 2 Pages 153-169

  • Published: 2018 Received: April 24, 2018 Released on J-STAGE: August 31, 2018 Accepted: July 06, 2018 Advance online publication: July 26, 2018 Revised: -

(compatible with EndNote, Reference Manager, ProCite, RefWorks)

(compatible with BibDesk, LaTeX)

Automatic cell segmentation is a powerful method for quantifying signaling dynamics at single-cell resolution in live cell fluorescence imaging. Segmentation methods for mononuclear and round shape cells have been developed extensively. However, a segmentation method for elongated polynuclear cells, such as differentiated C2C12 myotubes, has yet to be developed. In addition, myotubes are surrounded by undifferentiated reserve cells, making it difficult to identify background regions and subsequent quantification. Here we developed an automatic quantitative segmentation method for myotubes using watershed segmentation of summed binary images and a two-component Gaussian mixture model. We used time-lapse fluorescence images of differentiated C2C12 cells stably expressing Eevee-S6K, a fluorescence resonance energy transfer (FRET) biosensor of S6 kinase (S6K). Summation of binary images enhanced the contrast between myotubes and reserve cells, permitting detection of a myotube and a myotube center. Using a myotube center instead of a nucleus, individual myotubes could be detected automatically by watershed segmentation. In addition, a background correction using the two-component Gaussian mixture model permitted automatic signal intensity quantification in individual myotubes. Thus, we provide an automatic quantitative segmentation method by combining automatic myotube detection and background correction. Furthermore, this method allowed us to quantify S6K activity in individual myotubes, demonstrating that some of the temporal properties of S6K activity such as peak time and half-life of adaptation show different dose-dependent changes of insulin between cell population and individuals.

Key words: time lapse images, cell segmentation, fluorescence resonance energy transfer, C2C12, myotube

Live-cell fluorescence imaging provides the advantage of observing a biological system function with spatiotemporal resolution. With recent advances in imaging techniques and fluorescence biosensors, it has become clear that variability in intracellular states induces heterogeneity in the dynamic behavior of individual cells exposed to a uniform stimulus (Hughey et al., 2015 Snijder and Pelkmans, 2011 Tomida et al., 2015 Yao et al., 2016). Therefore, single-cell studies have uncovered the importance of understanding cellular dynamics and heterogeneity.

For single-cell studies, there is considerable need to develop algorithms that enable automatic segmentation and signal intensity quantification in individual cells from a large number of time-lapse images. Automatic segmentation methods have thus far been developed mainly for round shape and mononuclear cells (Bajcsy et al., 2015 Kodiha et al., 2011). For example, the marker based watershed segmentation is widely studied and used for efficient cell segmentation (Chalfoun et al., 2014 Hodneland et al., 2016 Wong et al., 2010). Typical use of the watershed segmentation is to flood a region from the stained nuclei as an initial flooding source until it touches the neighbors or cellular boundaries. However, such a method that requires a nucleus as a marker cannot be used directly for polynuclear cells, such as differentiated C2C12 myotubes.

C2C12 cells, derived from mice myoblasts, have been used as model cell lines for the study of cell differentiation and muscle functions in vitro (Fodor et al., 2017 Rahar et al., 2018 Rapizzi et al., 2008 Soltow et al., 2013). By differentiation induction of C2C12 cells, a large subpopulation of myoblasts forms elongated polynuclear myotubes, while a subpopulation of the cells, called reserve cells, remains undifferentiated (Fig. 1) (Yoshida et al., 1998). When combined with heterogeneity in the cell population, the elongated polynuclear form of the myotubes makes it difficult to perform automatic quantitative segmentation of individual myotubes using conventional methods developed for round shapes and mononuclear cells without a heterogeneous cell population. The elongated polynuclear form property of myotubes leads to over-segmentation by direct use of watershed segmentation. In addition, reserve cells that occupy spaces between the myotubes, make it difficult to automatically identify background regions.

Differentiated C2C12 myotubes stably expressing Eevee-S6K. Differentiated C2C12 cells stably expressing Eevee-S6K (Yellow). Nuclei were stained with DAPI (Blue). Because myotubes are much thicker than reserved cells, myotubes (red arrows) were much brighter than reserve cells (green arrows). Because of the relatively weaker fluorescence signal in reserve cells, only nuclear signals, but not cell body signals, were visible.

Here we developed an automatic quantitative segmentation method of a myotube using a watershed segmentation of summed binary images and two-component Gaussian mixture model (GMM). We established a C2C12 cell line that stably expressed Eevee-S6K, a fluorescence resonance energy transfer (FRET) biosensor, which monitor S6 kinase (S6K) activity, and acquired time-lapse fluorescence images both of cyan fluorescent protein (CFP) and FRET-induced yellow fluorescent protein (FRET-YFP) in differentiated C2C12 myotubes (Aoki and Matsuda, 2009). We made each image of FRET-YFP binary and summed the images. This procedure resulted in an enhanced contrast between myotubes and reserve cells, thereby permitting selective detection of myotubes. In addition, to identify individual myotubes, we used a myotube center, rather than a nucleus, as a marker of individual myotubes. To identify the myotube center, we applied a distance transform to the binary images of FRET-YFP. A distance transform is an operation that converts a binary image to a distance map image where all pixels have a value corresponding to the Euclidean distance to the nearest boundary pixel (Leymarie and Levine, 1992). We generated binary distance map images and summed the images, permitting enhancement of myotube centers. Using a myotube center instead of a nucleus, individual myotubes were segmented automatically by watershed segmentation.

Removing the noise signal from background regions is an essential prerequisite for obtaining quantitative measurements. However, it is difficult to identify background regions because of the presence of reserve cells that occupy the spaces between the myotubes. We used a two-component GMM that fit a fluorescence intensity histogram and estimated background intensity without manual selection of the background regions. The use of the two-component GMM ensures quantitative and objective measurement of signal intensity.

Thus, in this study, we provide a framework for automatic quantitative segmentation of individual myotubes that combines watershed segmentation using automatic myotube segmentation and background correction. Furthermore, this method allowed us to quantify S6K activity in individual myotubes, demonstrating that some of the temporal properties of S6K activity such as peak time and half-life of adaptation show different dose-dependent changes of insulin between cell population and individuals.

Automatic quantitative segmentation of elongated polynuclear myotubes using time lapse images

We developed an automatic quantitative segmentation method of a myotube from fluorescence time lapse images of differentiated C2C12 cells stably expressing Eevee-S6K (Aoki and Matsuda, 2009). The automatic quantitative segmentation method consists of three steps, Step I, II and III. Step I is pre-processing, where the use of median filter and white top-hat filter reduce variation of fluorescence intensity of the images. Step II is myotube segmentation, where the use of summed binary time-lapse images enhance myotubes and myotube centers, rather than reserve cells. Step III is background correction where we used a two-component GMM that fit a histogram of fluorescence intensity and estimated mean background intensity without manual selection of the background regions. The parameters used in each Steps were summarized in Table I.

Parameter Value (pixel) Description
median filter a 5 side length of filter window
white top hat b 35 side length of filter window
white top hat c 30 side length of filter window
threshold size d 10000 threshold area of myotube

(a) is the common value in detection of myotubes and detection of myotube centers. (b) is used for detection of myotubes. (c) is used for detection of myotube centers. (d) is used after watershed segmentation.

Step I: Pre-processing of time lapse images

Differentiated C2C12 cells are polynuclear and elongated in shape. Therefore, a myotube region in an obtained image shows a non-uniform distribution of fluorescence. For proper segmentation of myotubes, image smoothing of obtained images is essential. We first applied a median filter to remove outliers (Huang et al., 1979) and then applied a white top-hat filter to reduce variation of fluorescence intensity (Fig. 2).

Pre-processing of fluorescence images. In Step I-i, we applied median filter with 5×5 square window to the raw FRET-YEP time-lapse images. In Step I-ii, we applied white top-hat to images processed by the median filter. Filter window of the white top-hat is 35×35 and 30×30 in detection of myotubes and detection of myotube centers, respectively.

Step II: Segmentation method of myotubes

In Step II, we detected myotubes and myotube centers using thresholding methods including the triangle method and Otsu’s method (Otsu, 1979 Zack et al., 1977) (Fig. 3, Fig. 4), and the individual myotubes were segmented by watershed segmentation using the detected myotubes and myotube centers (Bajcsy et al., 2015) (Fig. 5). Step II consisted of three groups that contained 11 substeps in total Step II-i to II-iii, detection of myotubes Step II-iv to II-x, detection of myotube centers Step II-xi, watershed segmentation.

Detection of myotubes. (A) Detection of myotubes consisted of three substeps Step II-i, first binarization Step II-ii, summation Step II-iii, second binarization. In step II-ii, the intensity increases from blue to red. (B) FRET-YFP intensity histogram at frame 0. Red line indicates the threshold value, which was determined by the triangle method. (C) Intensity histogram of the summed binary image. The red line indicates the threshold value, as determined by Otsu’s method. (D) Labeled images of detected myotubes. One color corresponds to one continuous myotube region. Black denotes a region of either reserve cells or background.

Detection of myotube centers. (A) Detection of myotube centers consists of seven substeps Step II-iv, first binarization of fluorescence time-lapse images Step II-v, distance transform of the first binarized images Step II-vi, second binarization of the distance map images Step II-vii, Summation of the second binary images Step II-viii, third binarization of the summed images Step II-ix, labeling of the third binary image Step II-x, denoising of the labeled image. Note that in Step II-v and II-vii, colors denote intensity in a pixel, whereas, in Step II-ix and II-x, one color corresponds to one continuous region. Black indicates either reserve cells or background regions. (B) Skeletonized image of a labeled image. The red arrow head indicates short length debris. (C) Length histogram of a skeletonaized image. We assumed that length is a number of pixels in a continuous region of a skeletonized image. The red line indicates threshold length determined by the triangle method.

Watershed segmentation. (A) Myotubes and myotube centers were detected from time-lapse FRET-YFP images. A detected myotube was used as a boundary for watershed segmentation. A detected myotube center was used as a marker for watershed segmentation. Watershed segmentation was performed using detected myotubes and myotube centers, areas that were less than 10000 were removed. (B) Overlays of raw FRET-YFP time-lapse image at frame 0 and the result of watershed segmentation. Yellow lines indicate contours of segmented myotube regions.

Step II-i to II-iii: Detection of myotubes. Detection of myotubes consists of three substeps, Step II-i to II-iii (Fig. 3A) Step II-i, first binarization of FRET-YFP time-lapse images Step II-ii, summation of the first binarized images Step II-iii, second binarization of the summed image.

In Step II-i, we detected candidate regions of myotubes in each image by making them binary (Step II-i first binarization in Fig. 3A). Since the fluorescence intensity histogram of the FRET-YFP was skewed to the left, the triangle method was used to determine the threshold for the first binarization (Fig. 3B) (Zack et al., 1977). In addition, in order to prevent fluctuation of the threshold value due to the number of pixels, we standardized the total area of the intensity histogram to be one prior to using the triangle method. However, the binary images included myotubes and fragmented regions of reserve cells. To restrict this to only myotubes, enhanced contrast between myotubes and reserve cells will be needed.

In Step II-ii, we summed the binary images (Step II-ii Summation in Fig. 3A). Because myotubes were thicker and brighter than reserve cells, the summation of binary images allowed the contrast between myotubes and reserve cells to be enhanced.

In Step II-iii, we made the summed image binary to detect myotubes (Step II-iii Second binarization in Fig. 3A). Since the intensity histogram of the summed image was bimodal, Otsu’s method was used to determine the threshold for the second binarization (Fig. 3C). Otsu’s method is a statistical thresholding method to find a threshold that minimizes the within-class variance (Otsu, 1979).

Thus, we detected myotubes by successive thresholding methods, including the triangle method and Otsu’s method, using the summed binary image of FRET-YFP time-lapse images (Fig 3D). However, we still detected some of the myotubes as one continuous region rather than as individual myotubes. Further segmentation will be needed to detect individual myotubes.

Step II-iv to II-x: Detection of myotube centers. For segmentation of individual cells, watershed segmentation has been used most frequently in fluorescence imaging (Bajcsy et al., 2015). In watershed segmentation, a stained cell nucleus often is used as a marker of a cell because the majority of the cell type is mononuclear and round. However, conventional watershed segmentation methods using a nucleus as a marker of an individual cell are not designed to identify polynuclear and elongated shape cells such as myotubes. Therefore, we used a myotube center, instead of a nucleus, as a marker of a myotube for watershed segmentation.

Detection of myotube centers consists of seven substeps, Step II-iv to II-x (Fig. 4A) Step II-iv, first binarization of FRET-YFP time-lapse images Step II-v, distance transform of the first binary images Step II-vi, second binarization of the distance map images Step II-vii, summation of the second binary images Step II-viii, thrird binarization of the summed images Step II-ix, labeling of the third binary image Step II-x, denoising of the labeled image.

In Step II-iv, we made binary images using the triangle method for detection of candidate regions of myotubes.

In Step II-v, we transformed the binary images to distance map images that emphasized myotube centers (Step II-v Distance transform in Fig. 4A). However, some regions of the myotube centers were failed to be emphasized due to the presence of nuclei, and therefore were over-segmented.

In Step II-vi, the distance map images were made binary using the triangle method (Step II-vi Second binarization in Fig. 4A), and then summed to emphasize myotube centers (Step II-vii Summation in Fig. 4A). Because the nuclear positions in a myotube were changed frequently, the summed image attenuated the influence of nuclear existence.

In Step II‑viii, the summed image was made binary using Otsu’s method for detection of myotube centers (Step II-viii Third binarization in Fig. 4A).

In Step II-ix, we labeled the binary image with one continuous region as one myotube center (Step II-ix Labeling in Fig. 4A). Although the labeled image seemed to show a successful detection of myotube centers, it contained lots of small debris ranging in length from one to dozens of pixels (Fig. 4B and C). Therefore, we removed debris by the short length, where a length was defined as a number of pixels of a skeletonized region. Skeletonization is a thinning algorithm that removes outer pixels of a region to find its medial axis. For accurate myotube segmentation, these debris need to be properly removed.

Since the length histogram of the myotube centers was skewed to the left, Step II-x used the triangle method to determine the threshold (Fig. 4C). We detected individual myotube centers by removing regions smaller than the threshold (Step II-x Denoising in Fig. 4A, Table I).

Throughout the first ten substeps (Step II-i to II-x), we obtained an individual myotube region as a boundary for watershed segmentation, and an individual myotube center as a marker for watershed segmentation. Thus, the method using a myotube center instead of a nucleus enabled us to segment individual myotubes by the watershed segmentation.

Step II-xi: Watershed segmentation. In Step II-xi, we applied watershed segmentation to the myotubes detected in Step II-iii using the detected myotube center in Step II-x as a marker. Individual myotubes were segmented by removing regions smaller than a threshold (Fig. 5).

Evaluation of segmentation

In this study, implementation of the triangle method and Otsu’s method enabled an automatic segmentation of myotubes. However, because both the triangle method and Otsu’s method are based on the intensity histogram of an image, the segmentation results could be affected by excitation intensity and number of time-lapse images. Therefore, we compared the segmentation performance of our method to other traditional segmentation methods (Otsu’s method and triangle method) with various transmittance of excitation light and the various number of time-lapse images (Fig. 6). Ground truth of myotubes were generated from the MIP image by manual segmentation (Fig. S1), and Jaccard index (Levandowsky and Winter, 1971) was calculated between ground truth and each method. The Jaccard index is an indicator of the similarity between two regions (Bajcsy et al., 2015). The higher value of the index, the more similarity. Note that, like the ground truth, the Otsu’s method and the triangle method performed segmentation using MIP images. It was an advantageous performance comparison to Otsu method and triangle method. Also note that it is difficult to prepare perfect ground truth for cells with ambiguous boundaries between cells.

Accuracy of segmentation. (A) Left panel: Representative segmentation results with the indicated transmittance of excitation light (25%, 12%, 6% and 3%). Yellow lines are contours of individual myotubes. Right panel: Representative segmentation results when changing the number of time-lapse images by resampling every two to ten images. Full shows segmentation results using the full set of the images. Every 2, Every 5 and Every 10 indicate segmentation results obtained by resampling the images at every two, five and ten time points, respectively. Transmittance was set to 25%. (B) Jaccard indices in each condition in (A) (median±iqr, n=8 stage positions). Segmentation result of ground truth was used as the reference to calculate the Jaccard index. N. S. (Not significant), p>0.05 (Welch’s t-test). p values were corrected by Bonferroni correction.

Our method showed significantly higher Jaccard index than the traditional segmentation methods with 12 % or less transmittance of excitation light (Fig. 6A left panel, 6B left panel). This result indicates that our method is more robust to change in excitation intensity than the traditional segmentation methods. For reduction of number of images, there is no significant difference under every conditions. (Fig. 6A right panel, 6B right panel). These results suggest that the segmentation performance of our method is comparable to the traditional segmentation method against the number of images used. In addition, our method has an advantage that adjacent myotubes can be separated by implementing detection of myotube center as a marker for watershed segmentation.

Step III: Background correction method

Due to various factors, including the setup of the optical system, properties of the detector, and the fluorescent probe, a fluorescence image often contains background intensity variation in the same field. In addition, live-cell fluorescence imaging often is performed with weak excitation light to prevent photo-bleaching and photo-toxicity, which generates an image with a low signal to noise ratio. In ratiometric data, such as the FRET ratio, failure to identify of background regions leads to significant artifacts in signal intensity quantification. Therefore, proper background correction is needed for quantitative signal acquisition (Zimmermann et al., 2003).

Manual selection of background regions near the regions of interest (ROIs) has been used widely for background correction (Ceelen et al., 2007 Horie et al., 2015 Rahar et al., 2018). However, in the case of differentiated C2C12 cells, it is difficult to manually identify background regions, because reserve cells occupy spaces between myotubes (Fig. 1). Here, we used a two-component Gaussian mixture model (GMM) to estimate background regions.

Background correction consisted of five substeps (Fig. 7) Step III-i, maximum intensity projection (MIP) of FRET-YFP time-lapse images, Step III-ii, binarization of the MIP image Step III-iii, NOT AND (NAND) operation on the raw time lapse fluorescence images and binary MIP image Step III-iv, estimation of background intensity using two-component Gaussian mixture model (GMM). Step III-v, signal intensity quantification of individual myotubes.

Background correction. Background correction consists of five substeps Step III-i, maximum intensity projection (MIP) Step III-iii, NOT AND (NAND) operation Step III-iv, two-component Gaussian mixture model (GMM). Step III-v, signal intensity quantification. In Step III-iii, the intensity increases from blue to red. In Step III-iv, a red line indicates estimated background intensity distribution and a red dashed line indicates an average of the distribution as the estimated background intensity. A yellow line indicates the estimated intensity distribution of a region that included reserve cells and dead cells. In Step III-v, each blue line corresponds to the time series of each individual myotube. A red line indicates the mean time series of the FRET ratio of myotubes in each background correction.

MIP in Step III-i is a processing technique that keeps only the pixels of maximum intensity along the z-axis of stack images.

In Step III-ii, binarization of MIP images derived from FRET-YFP time-lapse images enabled reliable separation of myotubes from other regions composed of reserve cells and background regions (Step III-ii in Fig 7).

In Step III-iii, by performing the NAND operation on the raw time-lapse fluorescence images and the binary MIP image, we extracted a region composed of reserve cells and background regions from each time lapse image (Step III-iii in Fig. 7).

In Step III-iv, we estimated background intensity both of CFP and FRET-YFP time-lapse images using two-component GMM (Step III-iv in Fig. 7). In the two-component GMM, the histogram was divided into two components of Gaussian mixture distributions which corresponded to signal distribution of a region that included reserve cells and background regions, and background intensity was estimated as an average of the lower component.

In Step III‑v, we subtracted the estimated background intensity from each time point of the CFP and FRET-YFP time-lapse images, quantified signal intensities of CFP and FRET-YFP in individual myotubes (Fig. S2, right panel), and acquired a time series of the FRET ratio (FRET-YFP/CFP) (Step III-v in Fig. 7).

Evaluation of background correction

For evaluating accuracy of background correction using two-component GMM, we compared the accuracy of background correction using two-component GMM with other automatic background correction methods using raw histogram (RAW) and kernel density estimation (KDE) (Fig. 8). In a background correction using RAW, background intensity was estimated as the mode intensity of the histogram. In background correction using KDE, background intensity was estimated as the mode intensity of the histogram, approximated by kernel density estimation.

Comparison of accuracy of background corrections using RAW, KDE and two-component GMM. (A) Upper panels: time series of the FRET-ratio (FRET-YFP/CFP) of individual myotubes in RAW, KDE, and GMM, respectively (n=96, eight stage positions). Here, we used the same time lapse fluorescence images for all background corrections using RAW, KDE, and two-component GMM. One blue line corresponds to the time series of one myotube. The red line indicates the mean time series of FRET ratio of myotubes in each background correction. Lower panels: absolute first order differential time-series of the FRET ratio of individual myotubes in each background correction. The red line indicates the mean time series of the absolute first order differential time series. (B) AUC distributions of the absolute first order differential time-series of myotubes in each background correction. *, p<0.05 (Steel-Dwass test). In each violin plot, box plots are shown in the inset, and a white dot denotes the median.

Steps III‑i to III‑iii are common steps for all background corrections using RAW, KDE and two-component GMM. In Step III-iv, background intensity was estimated using RAW, KDE and two-component GMM, respectively. In Step III-v, time series of the FRET ratio (FRET-YFP/CFP) was acquired using each background correction method (Fig. 8A).

For evaluating accuracy of background corrections using RAW, KDE and two-component GMM, we calculated a quantification noise of the time series of FRET ratio in a myotube at a frame index i as di and a total quantification noise of a myotube as AUC, described by

where < y i >i = 1 N is the time series of the FRET ratio at a frame index i, N is the total number of frames, Δt is the time interval between frames, < d i >i = 1 N - 1 is the absolute first order differential time series between yi and yi–1 (Fig. 8A, lower panel), and AUC is the area under the curve of di in Eq (1).

The median of the AUC in background correction using two-component GMM was significantly smaller than those using RAW or KDE (Fig. 8B), indicating that the two-component GMM gives the lowest quantification noise among the three methods. In addition, GMM has advantages, in that the estimated distribution was continuous and the parameters including the mean and the variance in the Gaussian component were determined automatically.

We compared the accuracy of background corrections for manual estimation and two-component GMM (Fig. S1 and Fig. 9). In manual estimation, we generated the MIP image from FRET-YFP time-lapse images and selected the regions of myotubes and background visually (Fig. S1). For each selected region, we quantified signal intensities of CFP and FRET-YFP, and calculated a time series of the FRET ratio (Manual in Fig. 9A). In two-component GMM, the FRET ratio was calculated from intensities of CFP and FRET-YFP in the manually selected regions (Manual+two-component GMM in Fig. 9A). There was no significant difference in distributions of the AUCs among the three background correction methods, indicating that our proposed method using two-component GMM enabled provided objective and reasonable quantification compared with manual estimation (Fig 9B).

Comparison of accuracy of background correction using Manual and Manual+two-component GMM. (A) Upper panels: time series of the FRET ratio (FRET-YFP/CFP) of individual myotubes in Manual and Manual+two-component GMM, respectively (n=80, eight stage positions). The result of background correction using automatic myotube segmentation and two-component GMM in Fig. 6 was shown (two-component GMM). One blue line corresponds to the time series of one myotube. The red line indicates the mean time series of the FRET ratio of myotubes in each background correction. Lower panels: absolute first order differential time series of the FRET ratio of individual myotubes in each background correction. The red line indicates the mean time series of the absolute 1st order differential time series. (B) AUC distributions of the absolute first order differential time-series of myotubes in each background correction. N. S. (Not significant), p>0.05 (Steel-Dwass test). In each violin plot, box plots are shown in the inset, and a white dot denotes the median.

In an intramolecular FRET biosensor such as Eevee-S6K, CFP and YFP are linked by a linker domain, and the total noises of the time series of CFP and FRET-YFP should show a high correlation in a steady state. Therefore, we examined the coefficient of determination of the AUCs between CFP and FRET-YFP (Fig. 10). The coefficients of determination of the AUCs in background correction using RAW, KDE and two-component GMM were 0.822, 0.854, and 0.924, respectively (Fig. 10, upper panels). Similarly, the coefficients of determination of AUCs in Manual and Manual+two-component GMM were 0.899 and 0.892, respectively (Fig. 10, lower panels). Thus, background correction using two-component GMM shows the highest coefficient of determination of the AUCs of the absolute first order differential time series between CFP and FRET-YFP, indicating that two-component GMM is the most reasonable method for background correction.

Coefficients of determination between time-lapse images of CFP and FRET-YFP in each quantification method. One dot corresponds one myotube. The black line is the regression of AUCs of CFP and FRET-YFP.

We next examined whether our proposed method using two-component GMM can be used for myotubes under other conditions, such as myotubes stimulated with insulin or myotubes expressing an ATP probe stimulated with an electrical pulse stimulation (EPS) (Fig. 11). First, we used insulin-stimulated differentiated C2C12 cells stably expressed Eevee-S6K (Fig. 11A, upper panels). Individual myotubes were identified and quantified by manual estimation or by our proposed method using GMM (Fig. 11B, upper panels). We calculated the correlation coefficient between the time series of the FRET ratio quantified by manual estimation and two-component GMM in each region where the Jaccard index was larger than 0.5 (Fig. 11C, upper panel). Most of the correlation coefficients were larger than 0.98, indicating that our proposed method using two-component GMM is comparable to the manual estimation. In addition, we used EPS-stimulated differentiated C2C12 cells stably expressed mitAT1.03 which is FRET biosensor for monitoring ATP concentration in a mitochondrion (Imamura et al., 2009) (Fig. 11A, lower panel), and individual myotubes were quantified (Fig. 11B, lower panels). Similar to Eevee-S6K, all of the correlation coefficients were larger than 0.98 (Fig. 11C, lower panel), indicating that our proposed method using two-component GMM is comparable to manual estimation.

Quantification of signal activity of S6K and ATP concentration by applying our proposed method. (A) Time lapse of the FRET ratio image. Differentiated C2C12 cells stably expressing Eevee-S6K were stimulated with 100 nM of insulin at 50 min. Differentiated C2C12 cells stably expressing mitAT1.03 were stimulated with EPS (10 ms with 50 V, 1Hz interval) at 10 min and continued for 15 min. (B) A time series of the FRET-ratio (FRET-YFP/CFP) in response to insulin (n=90, eight stage positions) and EPS (n=10, one stage position) quantified by two-component GMM and quantified by Manual (n=80, eight stage positions, insulin and n=10, one stage position, EPS). One blue line corresponds one myotube. The red dashed line indicates time points of insulin stimulation. The red filled area indicates a period of EPS. (C) Histograms of Pearson’s correlation coefficients of the time series of the FRET ratio between two-component GMM and Manual in response to insulin (Upper panels) and EPS (Lower panels).

The differences between population and individual myotubes

Our proposed framework aims to quantify time series of fluorescence time lapse images in living individual myotube objectively. Using the framework, we tried to quantify insulin-dependent S6K activation and looked for different properties between cell population and individual myotube.

Myotubes were stimulated with various doses of insulin (0 nM to 100 nM) and time series of S6K activity in individual myotubes were quantified from the obtained fluorescence time lapse images (Fig. 12A). We defined properties of the time series such as Peak, Peak time, AUC, etc., from the time series (Fig. 12A lower right panel, B). For cell population, Peak, AUC, Half-life of adaptation and Intensity at half-life of adaptation increased in a dose-dependent manner of insulin. Variance in Peak time decreased, and Adaptation precision slightly decreased in a dose-dependent manner of insulin.

Insulin-dependent S6K activation in single myotube and cell population. (A) Time series of the FRET-ratio (FRET-YFP/CFP) in response to various doses of insulin in individual single myotubes. From Control to 100 nM, number of myotubes were n=98, 102, 100, 97, 86, 103 and 98, respectively. One blue line corresponds to the time series of one myotube. The red line indicates the mean series of each time series. Lower right panel is definition of the properties of time series. (B) The cell population response of time series in (A). In each violin plot, box plots are shown in the inset, and a white dot denotes the median. The red line indicates the median series of each properties.

In order to investigate the difference in properties between the cell population and the individuals, we performed correlation analysis with bootstrap subset (Bootstrap) as the cell population and all data (All) as the individuals (Fig. 13). The correlation of Peak time with all other properties were significantly different between the cell population and the individuals. Also, the correlation of Half-life of adaptation with all other properties were significantly different between the cell population and the individuals. These results suggest that Peak time and Half-life of adaptation show different responses between the cell population and individuals. The reason why the Peak time was significantly different in the combinations with all properties may be because the variance of Peak time decreased in a dose-dependent manner of insulin (Fig. 12B). Consistent with this, the correlations between Peak time with all other properties were decreased in a dose-dependent manner of insulin (Fig. S3).

Correlation between the properties in individual myotubes and population. (A) Spearman’s rank correlation coefficients between the properties. All (n=684) and Bootstrap (n=700) indicate the correlations in all individuals and bootstrap subsets of population of myotubes, respectively. The bootstrap subsets in each dose data were generated by iterating 100 times to randomly sample 10 points and calculate the median. Cyan rectangles the correlations higher in the cell population than in the individuals, magenta rectangles the correlation higher in the individuals than in the cell population, yellow rectangles the reversed correlation between the cell population and individuals. (B) The difference in correlation coefficients between All and Bootstrap. N. S. (Not significant), *p<10 –3 , **p<10 –5 , ***p<10 –7 . z score is a test static of the difference between two correlation coefficients (see Materials and methods).

We further analyzed Half-life of adaptation that shows significant differences from all other properties except Peak time. Differences in properties between the cell population and individuals can be classified into three groups that show higher correlation in the cell population than the individuals, higher correlation in the individuals than the cell population, and reversed relationship between the cell population and individuals (Fig. 13). In the combination of Half-life of adaptation, Peak and AUC showed stronger correlation with Half-life of adaptation in the cell population than in the individuals (Fig. 13A, B). Consistent with this, the distributions of Peak and AUC were more separated in the cell population than the individuals (Fig. S4A). Adaptation precision showed higher correlation with Half-life of adaptation in the individuals than in the cell population (Fig. 13A, B). In addition, the correlation of Half-life of adaptation with Adaptation precision in each dose decreased in a dose-dependent manner of insulin (Fig. S3 bottom center). This suggests that Half-life of adaptation and Adaptation precision in individuals differently responded in a dose-dependent manner of insulin. The correlation between intensity at Half-life of adaptation with Half-life of adaptation in the cell population was positive, whereas that in the individuals was negative (Fig. 13A, B). The distribution of Half -life of adaptation and intensity at Half -life of adaptation for each dose varied in the negative correlation direction in individuals, but not in the cell population (Fig. S4C). Furthermore, the correlations of Half-life of adaptation with intensity at Half-life of adaptation in each dose of insulin were negative in the individuals (Fig. S4 bottom right). These results indicate that individuals possess hidden properties that can not be seen in cell population. Thus, some of the properties of insulin dose-dependent S6K activation differ between the cell population and individuals.

Advantages of automatic myotube segmentation

Automatic segmentation of cells in fluorescence images has been developed for mononuclear and round shape cells (Bajcsy et al., 2015 Kodiha et al., 2011). However, an automatic segmentation method for elongated polynuclear cells, such as differentiated C2C12 myotubes has yet to be developed. Indeed, intracellular signal activity and reactive oxygen species in myotubes have been measured in vitro only by manual ROI selection of myotubes (Ceelen et al., 2007 Horie et al., 2015 Rahar et al., 2018). Since our method does not make assumptions on the number of nuclei and cell shape, our method can be used for any type of polynuclear and deformed cells. However, since our method does not implement cell tracking, it is difficult to apply it to moving cells. Using our method for images every fixed period and implementing cell tracking, there is a possibility that our method can be applied to moving cells.

Some recent studies using deep learning showed improvements of accuracy and throughput of segmentation of cells (Van Valen et al., 2016). Deep learning is a powerful tool for segmentation of cells with high accuracy and high throughput, and may be used effectively for elongated polynulcear cells, such as C2C12 myotubes. Despite versatility of deep learning for any type of cells, deep learning requires an enormous training dataset and computational resources in general. Especially, in cells with ambiguous cell boundary such as myotube, deep learning may learn bias of manual segmentation. By contrast, our proposed method does not require any training and few computational resources.

The calculation time of our method was less than 15 min even when processing 131 images including from segmentation to quantification, using AMD Ryzen 7 1800X (Table II). When only the segmentation was performed, our method took about 30 sec for one dataset (131 frames). This is fast enough compared with manual segmentation which took about 15–20 min using Roi tool in Fiji. In background correction, two-component GMM took about 3 sec per image. This is practical time, because our method was fully automated after setting the segmentation parameters.

Step Mean±S.D.
(sec)
Max
(sec)
Detection of myotubes (Step II-i to II-iii) a 6.13±0.52 6.82
Detection of myotube centers (Step II-iv to II-x) b 24.97±3.72 32.33
Watershed (Step II-xi) c
Background correction (Step III) d 2.94±0.84 4.73

Because (a), (b) and (c) are batch processes, the computation times were measured for each dataset (1344×1024 pixels, 131 frames) with eight stage positions (n=8). (c) was too fast to measure the computation time. Because of (d) is a sequential process for each image, the computation time was measured in each image (131 frames, eight stage positions, n=1048).

Heterogeneity of cell response

We quantified S6K activity in individual myotubes and demonstrated that some of the properties of S6K activity in a dose-dependent manner of insulin differed between cell population and individuals. This raises the possibility that some of the properties may be underestimated or incorrectly estimated by the cell population analysis. Moreover, single-cell analysis will address whether variation of cell population is derived from intra-cellular variation or inter-cellular variation.

The activity of other signaling molecules is also known to differ between individual cells. For example, in a mitogen-activated protein kinase (MAPK), Asynchronous oscillation of p38 MAPK response to external stimulation was revealed by single-cell analysis (Tomida et al., 2015). In the case of calcium response to ATP stimulation, the variance of calcium response was revealed that caused by internal state of the cells (Yao et al., 2016).

Taken together, single-cell analysis is important to reveal hidden properties of cell system that can not be observed in cell population. Signal transduction in skeletal muscle has thus far been studied by conventional biochemical and molecular biological methods such as western blot, that reflect the activity at cell population level (Areta et al., 2014 Baar and Esser, 1999 Dalle Pezze et al., 2016 Deldicque et al., 2008 West et al., 2016). In contrast, our proposed method can automatically detect and quantify signal activity of single myotube, and will open the door to single-cell analysis in signal transduction of skeletal muscle.

Construction of C2C12 cell line stably expressing fluorescent biosensors

C2C12 cells (kindly provided by Takeaki Ozawa, University of Tokyo, Tokyo, Japan) were cultured at 37°C under 5% CO2 in Dulbecco’s modified Eagle’s medium (DMEM), (High Glucose) with L-Glutamine and Phenol Red (Wako Pure Chemical Industries Limited, Osaka, Japan) supplemented with 10% fetal bovine serum (Nichirei Bioscience Incorporated, Japan). C2C12 cells stably expressing FRET biosensors, Eevee-S6K (plasmid was kindly provided by Kazuhiro Aoki, National Institute for Basic Biology, Aichi, Japan) (Aoki and Matsuda, 2009 Komatsu et al., 2011) and mitAT1.03 (plasmid was kindly provided by Hiromi Imamura, Kyoto University, Kyoto, Japan) (Imamura et al., 2009), were constructed using the PiggyBac Transposase System (System Biosciences, U.S.A.), respectively. Three hundred μL of Opti-MEM (Life technologies, U.S.A.), 4 μL of Lipofectamin 2000 (Invitrogen, U.S.A.), 1.0 μg of PiggyBac transposon vector clone (kindly provided by Kazuhiro Aoki, National Institute for Basic Biology, Aichi, Japan) and 0.2 μg of PiggyBac transposase expression vector (PB210PA-1, Funakoshi, Japan) were mixed and let stand for 5 min. Thereafter, 70% confluent C2C12 cells were plated on a 35 mm dish and transfected with the mixture and incubated for 6 h. For selection of infected cells, the cells were cultured with DMEM (High glucose) with L-Glutamine and Phenol Red containing 10% fetal bovine serum and 20 μg/mL of Blasticidin S Hydrochloride (Wako, Japan). Selected cells were seeded on a Cell Culture Dish 430167 (Corning Incorporated, U.S.A.) and cultured until forming the colonies. The colonies were picked and seeded on a Cell Culture Dish 430167. After seeding and proliferation, the cells were stored at a concentration of 1.0×10 5 cells/mL with Bambanker (NIPPON Genetics, Japan). Eevee-S6K is localized in the cytosol and mitAT1.03 is localized in the mitochondria.

Differentiation induction of C2C12 cells

For differentiation induction, C2C12 cells were plated on 35 mm/Collagen Coated Dish Collagen type I (IWAKI, Japan) at a concentration of 1.0×10 5 cells/dish and cultured for two days under conditions described above until confluent, and then confluent cells were cultured in DMEM with L-Glutamine and Phenol Red supplemented with 2% horse serum (Nichirei Bioscience Incorporated) for seven days (Odelberg et al., 2000).

Fluorescence time lapse imaging

C2C12 myotubes were starved in 2 mL of Medium199, Hanks’ Balanced Salts (Life technologies, U.S.A.) overnight and then mineral oil (Sigma Aldrich) was stratified to prevent vaporization of the medium prior to the fluorescence time-lapse imaging. Fluorescence time-lapse imaging was performed with an inverted fluorescence microscope, IX 83 (Olympus, Tokyo, Japan) equipped with a UPLSAPO10X2 objective lens (Olympus), a ORCA-R2 C10600-10B CCD camera (Hamamatsu Photonics), a U-HGLGPS mercury lamp (Molecular Devices, Sunnyvale, CA), XF3075 and XF3079 emission filter for CFP and YFP (Omega Optical), respectively, and an MD-XY30100T-META automatically programmable XY stage (Molecular Devices).

All the images of myotubes were 1344×1024 pixels and 0.645 μm/pixel. Time-lapse images of myotubes expressing Eevee-S6K were acquired for 650 min every 5 min (total 131 frames) with eight stage positions. Time-lapse images of myotubes expressing mitAT1.03 were acquired for 45 min every 1 min (total 46 frames) with one stage position. After background subtraction, FRET-YFP/CFP ratio images were created with Meta-Morph software (Universal Imaging, West Chester, PA).

Our proposed method was developed using Python and Python modules, including Mahotas (Coelho, 2013), Scikit-image (van der Walt et al., 2014) and Scikit-learn (Pedregosa et al., 2011) for automatic quantification. Fiji (Schindelin et al., 2012) was used for manual quantification of fluorescence intensity of FRET-YFP and CFP. The FRET-YFP and CFP intensities were averaged over the whole myotube area.

Significance of the difference between two correlation coefficients

Fisher’s z-transformation of correlation coefficients r is given by

where i is an index of correlation coefficient. A z score of the difference between two correlation coefficients, described by

where n1 and n2 are numbers of data corresponding to r1 and r2. As the z score follows normal distribution, significance test was performed with significance level α=10 –3 , 10 –5 , 10 –7 .

Availability of data and materials

The datasets of time-lapse images, segmentation results and segmentation programs are available from our database (http://kurodalab.org/info/Inoue).

We thank Kazuhiro Aoki for kindly providing FRET biosensor, Eevee-S6K.

We thank Shinsuke Uda for helpful discussions and technical advice. We thank our laboratory members Masashi Fuji and Atsushi Hatano for their critical reading of this manuscript, helpful discussions, and technical assistance with the experiments.

This work was supported by the Creation of Fundamental Technologies for Understanding and Control of Biosystem Dynamics, CREST, of the Japan Science and Technology Agency (JST)(#JPMJCR12W3) and by the Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number (17H06300, 17H06299, 18H03979). K. Kunida receives funding from a Grant-in-Aid for Young Scientists (B) (#16K19028).


South Carolina Junior Academy of Science

Cachexia is a muscle wasting syndrome found in patients with late stages of cancer. It leads to rapid loss of body fat and skeletal muscle tissue and has a significant impact on the mortality rates of patients. Cachexia is multifactorial, but studies have shown that high levels of inflammatory cytokines such as IL-6 and LIF induce symptoms that promote this muscle wasting. However, it is unknown whether this is from an overexpression of protein synthesis or an increased protein degradation. LIF is a cytokine in the IL-6 family, and its impact on cachexia needed to be measured. C2C12 cells were treated with LIF for 24 hours prior to collection and the rates of myotube atrophy and protein synthesis were measured in order to determine the role of LIF on cachexia. Myotubes were photographed and the average diameters of the groups grown with LIF and without LIF were calculated. Puromycin, which was introduced to the cell cultures 30 minutes prior to collection, and p6070SK content was measured for the control and experimental groups using a western blot. Cell groups that were introduced to LIF had smaller myotube diameter and were less organized than the control group. Cell groups with LIF had significantly lower amounts of puromycin and p6070SK in their cells. Based upon these results, LIF induces symptoms myotube atrophy symptoms, as well as the rate of protein synthesis. However, further research is required to understand LIF’s signaling pathways and the affect other cytokines have on cachexia.


Human myotube differentiation in vitro in different culture conditions

Human muscle cells derived from satellite cells, maintained in standard tissue culture conditions, do not differentiate as rapidly or as completely as myoblasts from other species (chicken, rat, mouse). In an attempt to improve myogenesis, we studied the effects of modifying the culture media and of coculturing muscle with nerve cells, using myoblasts grown in standard culture media as the basis for comparison. Myogenesis was measured by fusion index, creatine kinase (CK) activity acetylcholinesterase (AChE) activity (total and molecular forms) and the number of acetylcholine receptors (AChR). Modification of culture media accelerated fusion of myoblasts, but the cell density decreased and myotubes were unable to survive for long periods. In contrast, coculturing muscle with nerve cells increased both cell density and the number of myotubes. CK, AChE and AChR increased in the presence of defined media. In the nerve-muscle cocultures the increase was less marked. Manipulating culture conditions modified the molecular forms of AChE. Only a (4 + 6.5) S peak was present in control cultures, but a 10S peak appeared in defined media. The 16S form was detected only in nerve-muscle cocultures. This study shows that fusion of human myoblasts and differentiation of myotubes in tissue culture can be accelerated by removal of serum macromolecules. Further differentiation of myotubes was achieved only in the nerve-muscle cocultures.


Watch the video: Sliding Filament Theory Of Muscle Contraction Explained (September 2022).


Comments:

  1. Lichas

    They are similar to the expert)))

  2. Eilig

    It is remarkable, it is the valuable information

  3. Ahmadou

    remarkably, the very useful phrase



Write a message