Mammalian Target of Rapamycin (mTOR) Signaling Is Required for a Late-stage Fusion Process during Skeletal Myotube Maturation*

Skeletal myogenesis is a well orchestrated cascade of events regulated by multiple signaling pathways, one of which is recently characterized by its sensitivity to the bacterial macrolide rapamycin. Previously we reported that the mammalian target of rapamycin (mTOR) regulates the initiation of the differentiation program in mouse C2C12 myoblasts by controlling the expression of insulin-like growth factor-II in a kinase-independent manner. Here we provide experimental evidence suggesting that a different mode of mTOR signaling regulates skeletal myogenesis at a later stage. In the absence of endogenous mTOR function in C2C12 cells treated with rapamycin, a kinase-inactive mTOR fully supports myogenin expression, but causes a delay in contractile protein expression. Myoblasts fuse to form nascent myotubes in the absence of kinase-active mTOR, whereas the formation of mature myotubes by further fusion requires the catalytic activity of mTOR. Therefore, the two stages of myocyte fusion are molecularly separable at the level of mTOR signaling. In addition, our data suggest that a factor secreted into the culture medium is responsible for mediating the function of mTOR in regulating the late-stage fusion leading to mature myotubes. Furthermore, taking advantage of the unique fea-tures of cells stably expressing a mutant mTOR, we have performed cDNA microarray analysis to compare global gene expression profiles between mature and nascent myotubes, the results of which have implicated classes of genes and revealed candidate regulators in myotube maturation or functions of mature myotubes.

Ordered and temporally separable events in skeletal muscle differentiation have been defined in the in vitro system of myoblast cell cultures (1,2), consisting of myogenin expression, cell cycle withdrawal, phenotypic differentiation as marked by contractile protein expression, and cell fusion to form multinucleated myotubes (3). Growth and regeneration of adult skeletal muscle in vivo require satellite cell activation, proliferation, and differentiation (4 -6). The myofiber size increase is accompanied by a proportional increase of the number of nuclei within the growing myofiber, indicating the requirement of myoblast fusion for such processes (4, 6 -8). Whereas myoblast fusion has been recognized as a tightly regulated process important for mammalian muscle development, the regulatory mechanisms are not well understood at the molecular level. Recently, studies of calcium signaling in mammalian muscle growth have provided insights into the regulation of myoblast fusion, and revealed two distinct stages of fusion controlled by separate molecular pathways, the initial myoblast-myoblast fusion to form nascent myotubes with limited number of myonuclei and subsequent myoblast-myotube fusion resulting in myotube size increase (7,9,10). Calcium-dependent NFATC2 activity is specifically required for the second-stage fusion by regulating the expression of IL-4, 1 an unexpected myoblast recruitment factor (7,10).
The cellular target of the bacterial macrolide rapamycin, mTOR, regulates mammalian cell growth and proliferation by mediating signal transduction of multiple pathways (11). mTOR governs a nutrient-sensing pathway through a recently identified protein complex containing Raptor and G␤L (12)(13)(14), and at the same time directly participates in the cytoplasmic relay of mitogenic signals via the lipid second messenger phosphatidic acid (15,16). Two best characterized downstream effectors for mTOR are ribosomal S6 kinase 1 and the initiation factor eIF4E-binding protein 1; both are involved in regulation of protein synthesis (17). Like the other members of the phosphatidylinositol kinase-related kinase family (18), mTOR is a large protein with Ser/Thr protein kinase activity. Both S6 kinase 1 and eIF4E-binding protein 1 are regulated through phosphorylation by multiple kinases, one of which is most likely mTOR (19 -23).
Recently, mTOR has also been recognized as an important player in skeletal myocyte differentiation. Rapamycin inhibits differentiation of various myoblast cell cultures (24 -26), and a rapamycin-resistant recombinant mTOR fully rescues C2C12 differentiation in the presence of rapamycin (27,28). Unexpectedly, we have found that neither the kinase activity of mTOR, nor S6 kinase 1, are required for the initiation of myoblast differentiation (27). We have further identified the nutrient-dependent expression of the insulin-like growth factor IGF-II as a key mediator of this unusual mTOR signaling in myogenesis (29). However, in the course of our investigation it became evident that rapamycin inhibited the multistep myogenesis at more than one stage. Here we present evidence that a distinct mTOR function controls myogenesis at the myotube maturation stage, which is mechanistically separable from the role of mTOR in the earlier events of differentiation. In contrast to mTOR regulation of the initiation of differentiation as marked by myogenin expression and the formation of nascent myotubes, which is independent of mTOR kinase activity, a secondstage myocyte fusion that leads to mature myotubes requires kinase activity of mTOR. Our observations also suggest that a secreted factor is responsible for mediating the function of mTOR in the second-stage fusion. Furthermore, gene expression profile analyses have led to the identification of genes differentially expressed during myotube maturation, which potentially define the molecular signatures of mature myotubes and provide candidates for the target of mTOR regulation during myotube maturation.

EXPERIMENTAL PROCEDURES
Antibodies and Other Reagents-Anti-FLAG M2, anti-troponin T antibodies, and gelatin solution were from Sigma. All secondary antibodies were from Jackson ImunoResearch Labs. The MF20 anti-sarcomeric myosin heavy chain (MHC) and F5D anti-myogenin antibodies were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD, National Institutes of Health and maintained by The University of Iowa, Department of Biological Sciences. The antitubulin antibody was a gift from Dr. V. Gelfand. Serum and G418 were from Invitrogen. Rapamycin was from Calbiochem.
Cell Culture and Establishment of Stable Cell Lines-C2C12 myoblasts were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum at 37°C with 7% CO 2 . For differentiation, cells were plated on tissue culture plates coated with 0.2% gelatin and grown to 100% confluence, changed into DM (Dulbecco's modified Eagle's medium with 2% horse serum), and replenished with fresh DM daily for 3-4 days.
For stable expression in cells, various mTOR cDNAs were inserted into the pIRESneo vector (Clontech) at the NotI site, containing a sequence encoding the FLAG epitope at the 5Ј end. C2C12 cells were transfected by pIRESneo-mTOR using FuGENE 6 (Roche). After 48 h, the cells were trypsinized and seeded into 10-cm plates at various densities in growth medium containing 0.5 mg/ml G418. After 7-10 days, colonies were picked and expanded, and the expression of FLAG-mTOR protein was examined by Western analysis. Upon confirming the recombinant protein expression, each cell line was examined for its ability to differentiate in the absence and presence of rapamycin. Approximately 60% of all clones for both RR-and RR/KI-mTOR retained the ability to differentiate, and among them most differentiated in the presence of rapamycin.
Immunofluorescence Microscopy and Quantitative Analyses of Myocytes-C2C12 myoblasts or myotubes grown on tissue culture plates were fixed in 3.7% formaldehyde (in phosphate-buffered saline), permeabilized in 0.1% Triton X-100, and then incubated with MF-20 antibody (in 3% bovine serum albumin/phosphate-buffered saline) followed by fluorescein isothiocyanate anti-mouse IgG antibody (in 3% bovine serum albumin/phosphate-buffered saline), with 4,5-diamidino-2-phenylindole (4 g/ml) whenever applicable. The stained cells were examined with a Leica inverted fluorescence microscope. Fluorescent images were recorded with a CCD camera using SPOT version 3.2.4 software (Diagnostic Instruments, Inc.). The average number of nuclei per myotube was generated from 100 or so randomly chosen MHC-positive cells containing two or more nuclei. The percentage of myotubes containing 5 or more nuclei was calculated from similar sampling as above. The fusion index was calculated from the ratio of nuclei number in myocytes with two or more nuclei versus the total number of nuclei. All data are expressed as the mean Ϯ S.D. (n Ն 3). Student's t tests were performed for comparisons, with p Ͻ 0.05 considered statistically significant.
cDNA Microarray-Microarray analysis was carried out using the NIA 15K cDNA chips (containing ϳ15,000 mouse cDNAs) developed by the National Institute on Aging (30) and manufactured by the Keck Biotechnology Center, University of Illinois at Urbana-Champaign. Total RNA was isolated from RR/KI44 cells grown on 60-mm plates using the RNeasy minikit (Qiagen) following the manual. Probes for hybridization were generated according to the TIGR protocols (www.tigr.org/ tdb/microarray/protocolsTIGR.shtml). Briefly, 10 g of total RNA was used to generate aminoallyl-modified cRNA, which was then labeled with Cy3 or Cy5. More than 200 pmol of dye incorporation per sample and less than 50 of nucleotides/dye ratio were used as criteria for acceptable quality of the probes. The cRNA from 0 h was labeled with Cy3 and those from other time points were labeled with Cy5, and vice versa (dye swap). Microarray hybridizations were also carried out according to the TIGR protocols. For each experimental point, 6 hybridizations were performed, which included three independent RNA samples and dye swap for each sample.
Microarray Data Acquisition and Analysis-Hybridization images were acquired by using a GenePix 4000B scanner with GenePix Pro 6.0 software. After analyzing each spot in the slides for quality control, Cy3 and Cy5 intensities with foreground and background values formed GenePix Result (gpx) data files, which were used for statistical analysis. Data were imported to the R software environment (www.R-project.org) and analyzed. Once background subtracted Cy3 or Cy5 intensities were acquired, spots with median intensity less than 2 ϫ S.D. of background of each block were filtered out. Regional Lowess (locally weighted polynomial regression) normalization was performed using the R/maanova package. Analysis of variance test was performed to obtain the genes that displayed differential expression at least at one time point (24, 48, or 72 h) compared with 0 h (p Ͻ 0.01). Hierarchical clustering of the data were performed using the self-organizing map algorithm in the GeneSpring software package (Silicon Genetics, Inc.).
Real Time RT-PCR-Total RNA was isolated as described for the microarray procedures. cDNA was synthesized from 1 g of total RNA with Superscript II reverse transcriptase (Invitrogen) using oligo(dT) primer (Invitrogen). qRT-PCR was performed on an ABI Prism 7700 sequence detection system (Applied Biosystems) using SYBR green chemistry in a MicroAmp 96-well reaction plate following the manufacturer's protocols. Actin ␤, which was unchanged (0.9 to 1.1-fold) in all the microarray data, was used as the reference to obtain the relative -fold change for target samples using the comparative C T method. In short, C T difference (⌬C T ) between target sample and Actin ␤ at each time point was obtained and the difference of ⌬C T (⌬⌬C T ) between 0 h and another time point was calculated. The relative -fold change using 0 h as control was obtained using the formula 2 Ϫ⌬⌬CT . All the PCR primers used in this study (Table I) were confirmed to have equal efficiency as the actin ␤ primers, thus validating the comparative C T method.

C2C12 Cells Supported by a Kinase-inactive mTOR Form
Smaller Myotubes-Previously we showed that a rapamycinresistant and kinase-inactive mTOR was able to support C2C12 differentiation in the presence of rapamycin by inducing normal IGF-II expression and phenotypic differentiation (27,29). To facilitate further studies we established C2C12 cell lines stably expressing the rapamycin-resistant mTOR mutant (S2035T (19); designated "RR"), and separately, the RR mutant also carrying a kinase-inactive mutation (D2357E) (designated "RR/KI"). Although the mechanism of rapamycin-resistance is not yet clear and rapamycin is unlikely to directly inhibit the kinase activity of mTOR (12-15, 31), rapamycin effectively eliminates the functions of endogenous wild-type mTOR as confirmed by the complete lack of any differentiation events in rapamycin-treated C2C12 cells, thus creating an mTOR functionally null background to allow the assessment of any recombinant mTOR carrying the RR mutation. Most of the experiments reported hereafter were performed in the presence of rapamycin, taking advantage of the "chemical knock-out" of mTOR by rapamycin. To eliminate the influence of clonal variation, several independent C2C12 cell clones expressing either RR-mTOR or RR/KI-mTOR were used in most experiments. Multiple clones of RR-and RR/KI-mTOR cells differentiated normally as monitored by the formation of MHC-positive myotubes by 72 h of serum deprivation (representative results from two independent clones for each mTOR mutant are shown in Fig. 1A). FLAG-tagged recombinant mTOR was expressed at similar levels in various stable clones (Fig. 1B). Expression of the recombinant mTOR had no detectable positive or dominant negative effect on differentiation. This was expected because the recombinant mTOR proteins were at a level comparable or lower than that of the endogenous mTOR (data not shown). Both the RR-and RR/KI-mTOR cells differentiated in the presence of rapamycin, confirming our previous observations with transient expression of the same mTOR proteins (27). It was apparent, however, that the RR/KI-mTOR cells did not behave identically to the RR-mTOR cells: rapamycin treatment reduced the size of RR/KI-mTOR myotubes (Fig. 1A). Quantification of the myotube size by the average number of nuclei per myotube is shown in Fig. 1C. We did not notice this difference earlier in the transiently transfected cells (27), likely because of a combination of lower transfection efficiency, slower rate of differentiation in those cells, and insufficient time to allow RR-mTOR myotubes to fully mature and to be distinguished from RR/KI-mTOR myotubes (see Fig. 4, below).
The Early Differentiation Marker Myogenin Is Expressed Normally in Kinase-inactive mTOR Cells-In search for a mechanistic explanation for the aforementioned morphological difference, we studied the differentiation program closely in these stable cells. We first examined the expression of the myogenic regulatory factor myogenin. Myogenin is considered one of the early markers of myoblast differentiation. As shown in Fig. 2, myogenin protein expression was induced in differentiating C2C12 cells and reached its maximal level by 48 h of differentiation, and rapamycin treatment abolished its expression. The myogenin expression profile in the RR-mTOR cells was similar to that in the parental cells, and rapamycin treatment had little effect, confirming the rapamycin-resistant nature of this mutant. Likewise, the RR/KI cells displayed a similar myogenin expression pattern, with or without rapamycin. Because myogenin expression signifies the beginning of the differentiation program, this observation suggests that the mTOR kinase-defective cells entered the differentiation program normally.
We next examined the expression of the contractile protein MHC, which typically appeared around 48 h during differentiation. Again, both in the presence and absence of rapamycin, the RR-mTOR cells displayed normal MHC expression patterns (Fig. 3), suggesting that rapamycin-resistant mTOR fully rescued the differentiation program from rapamycin inhibition. On the other hand, the expression of MHC was delayed in the RR/KI-mTOR cells upon treatment by rapamycin, and it eventually (24 h later) reached a maximal level similar to that in the absence of rapamycin (Fig. 3). We also assessed the expression of another contractile protein, troponin T. Similarly, the RR/ KI-mTOR cells displayed a delayed troponin T expression profile in the presence of rapamycin, whereas the RR cells expressed troponin T within a normal time frame with or without rapamycin treatment (Fig. 3). All the Western blots shown in Figs. 2 and 3 were from one clone for each mTOR construct; several other clones were examined and yielded similar results (data not shown).
C2C12 Cells Supported by a Kinase-inactive mTOR Form Nascent Myotubes, but Not Mature Myotubes-We next took a closer look at the myotube formation process in the mTORexpressing cells, and asked whether the smaller myotube size in RR/KI-mTOR cells merely reflected a delay in differentiation, as the delayed expression of contractile proteins might have suggested. This appeared not to be the case, as clearly shown in Fig. 4. When the two different types of cultures were compared in the presence of rapamycin (and therefore in the absence of endogenous mTOR function), the RR/KI-mTOR myotubes were permanently arrested at a smaller size; they did not grow significantly even after two more days in differentiation medium and never reached the normal size (Fig. 4A) as quantified by the average number of nuclei per myotube (Fig. 4B). In the absence of rapamycin, the rate and extent of myotube formation were indistinguishable between RR/KI-mTOR and RR-mTOR cells (data not shown).
It was possible that differentiation of the RR/KI cells in the presence of rapamycin was overall impaired, resulting in less cells fusing to form myotubes. We therefore measured the fusion index, defined as the percentage of total nuclei in myo-cytes that contained two or more nuclei. The formation of MHC-positive myotubes in RR/KI-mTOR cultures occurred slightly slower than in RR-mTOR cultures; this is reflected by the lower fusion index of RR/KI-mTOR cells at 48 h (Fig. 4C), which is consistent with the observation that contractile protein expression was delayed in the same cells (Fig. 3). However, by 72 h the fusion index became indistinguishable between the two types of cells (Fig. 4C). We examined the fusion index for several independent clones of RR-mTOR and RR/KI-mTOR cells, and the results are summarized in Table II. Typically, both RR and RR/KI cells reached ϳ40% fusion around 72 h, with variations of 6% or less among different cell lines; rapamycin treatment had no obvious effect on the degree of fusion in either type of cells. Apparently, in RR/KI cells under rapamycin treatment the total number of myotubes was higher, which compensated for the smaller size of these myotubes. Taken together, it is evident that cells supported by RR/KI-mTOR can enter the differentiation program normally and form nascent myotubes, but further fusion that gives rise to mature myotube size is impaired in these cells. Hence, these two stages of myocyte fusion are separable, and regulated by distinct mTOR signaling mechanisms.
mTOR Regulates a Secreted Factor(s) Required for Myotube Maturation-Horsley et al. (10) recently identified IL-4 as the first "myoblast recruitment factor" responsible for the fusion process leading to myocyte maturation. To probe the mechanism by which mTOR regulates myotube maturation, we tested the possibility that a secreted factor might be responsible for mTOR-mediated myotube growth. Conditioned media were collected from differentiating parental C2C12 cells or RR-mTOR cells in the presence of rapamycin at 48 and 72 h, and added to RR/KI-mTOR myotubes (two different clones) differentiating in the presence of rapamycin at 72 and 96 h, respectively. The treated RR/KI-mTOR myotubes were fixed and stained at 120 h, and the percentage of large myotubes (with 5 or more nuclei) was calculated. Conditioned media from both C2C12 and RR-mTOR cells (two different clones) rescued the nascent myotube arrest in RR/KI-mTOR cells by promoting the formation of mature myotubes, comparable in size to those in the absence of rapamycin (Fig. 5A). To rule out the possibility that rapamycin induces the secretion of an inhibitor, conditioned medium from RR/KI-mTOR cells differentiating in the presence of rapamycin were supplied to RR-mTOR cells, and no inhibitory effect was observed on RR-mTOR myotube maturation (data not shown). Together, these observations suggest that mTOR regulates the secretion of a factor(s) required for myoblast-myotube fusion during myotube maturation, and that this function is dependent on the catalytic activity of mTOR, in contrast to the kinase-independent regulation of mTOR in IGF-II expression during initiation of differentiation (Fig. 5B). The identity of this putative factor is currently under investigation.
Global Gene Profiling of Myotube Maturation-The RR/KI-mTOR cells can represent two stages of myogenesis distinctively: mature myotubes form at 72 h differentiation in the absence of rapamycin, whereas rapamycin treatment effectively "freezes" the cells at the nascent myotube stage. This provides a relatively clean experimental system for the investigation of global gene expression during myotube maturation. Taking advantage of this unique system, we performed cDNA microarray analyses using the NIA mouse 15K cDNA chips, to compare gene expression profiles in RR/KI-mTOR cells undergoing differentiation at 24, 48, and 72 h in the absence or presence of 100 nM rapamycin. For each hybridization experiment, total RNA from cells at the start of differentiation (0 h) was used as the reference. The experiments were carried out 6

FIG. 2. Expression of myogenin in rapamycin-resistant C2C12 cells.
Parental or stable C2C12 cells expressing recombinant mTOR proteins were induced to differentiate in DM with or without 100 nM rapamycin (Rap). The cells were lysed at the indicated times, and the lysates were subjected to Western analyses using an anti-myogenin antibody. Anti-tubulin blots served as loading controls. The experiments were repeated two to six times in several different clones, and representative results are shown.

FIG. 3. Expression of contractile proteins in rapamycin-resistant C2C12 cells.
Parental or stable C2C12 cells expressing recombinant mTOR proteins were induced to differentiate in DM with or without 100 nM rapamycin (Rap). The cells were lysed at the indicated times, and the lysates were subjected to Western analyses using anti-MHC and anti-troponin T antibodies. The experiments were repeated two to six times in several different clones, and representative results are shown. Anti-tubulin blots as loading controls were the same as described in the legend to Fig. 2. times for each experimental point using 3 biological replicates and Cy3/Cy5 dye swapping. The average of the data at each point after the initial data screening was used for further analysis (see "Experimental Procedures" for details).
Of ϳ15,000 elements (ϳ12,000 unique mouse genes) screened, 2983 genes displayed statistically significant change in expression at one or more time points, of which 483 genes underwent changes of 2-fold or more. It should be pointed out that these changes may be due directly to the absence of kinase-active mTOR (in the presence of rapamycin), or they may be a consequence of myotube arrest at the nascent stage. The microarray data do not tell us whether any genes are directly regulated by mTOR signaling, but they provide potential candidates. The self-organizing map assigned these 483 genes into 9 clusters based on their expression patterns (Fig. 6). A complete list of all these genes can be found in supplemental data Table S1. Cluster (1, 1) contains 153 genes that are downregulated in both nascent and mature myotubes; they may be suppressors of myocyte differentiation. Among them the Id proteins (Id1, Id2, and Id3) are inhibitors of E box proteins and known to be down-regulated during myogenesis. Cluster (3, 3) contains 133 genes that are up-regulated in both nascent and mature myotubes. These genes may play positive roles in the early differentiation process. And indeed, IGF-II is found within this group, which confirms our previous observations that the RR/KI-mTOR cells produced normal levels of IGF-II in the presence of rapamycin and that IGF-II likely mediates the FIG. 5. mTOR regulates myotube maturation through a secreted factor. A, conditioned media from differentiating parental or RR-mTOR cells (clones 2 and 9) were added to RR/KI-mTOR cells (clones 21 and 44) during differentiation (see text for details). The percentage of myotubes containing 5 or more myonuclei was calculated. Data shown are the average results of three independent experiments with error bars representing S.D. Black and gray bars represent RR/ KI21 and RR/KI44 cells, respectively. Data comparison was made within each clone against "no Rap no CM," and only those of "ϩRap no CM" were found significantly different (*, p Ͻ 0.01). B, a model for the dual roles of mTOR in skeletal myocyte differentiation. See text for details.

TABLE II Fusion index for C2C12 cells stably expressing RR-mTOR and RR/
KI-mTOR The cells were differentiated and immunostained as described in Fig.  1 legend. The percentage of fusion was calculated from the ratio of nuclei number in myocytes with two or more nuclei vs. the total number of nuclei in all cells. Three independent clones for RR-mTOR and four clones for RR/KI-mTOR were analyzed. The average results of at least three independent experiments and standard deviations are shown. kinase-independent function of mTOR in the initiation of differentiation (29). Contractile proteins, such as myosin light chain, dystroglycan, troponin T1, tropomyosin ␤2, and actinin ␣2 are also found in this cluster as expected. However, it should be noted that cluster (3,3) may also include genes that are up-regulated to a different extent in nascent versus mature myotubes. We will discuss the identification and functional implication of those genes later.
Genes in clusters (3,1) and (3,2), which are up-regulated in mature myotubes and down-regulated or unchanged in nascent myotubes (compared with pre-differentiation), likely play positive roles in skeletal myotube maturation, although the expression of these genes may or may not be directly regulated by the kinase activity of mTOR. Some genes have been previously indicated by microarray data to increase during C2C12 differentiation (32), such as SerpinB1A, encoding a serine/cysteine protease inhibitor; BoP, encoding a transcriptional repressor; SudD in epigenetic regulation; and stearoly-coenzyme A desaturase 2 (Scd2) in fatty acid metabolism, to name a few. Our data may pinpoint a specific stage, myotube maturation, during myogenesis when these proteins are functional. Genes in clusters (1,2) and (1,3) are down-regulated during maturation of myotubes, but not during the formation of nascent myotubes, suggesting that they may play negative roles in either myotube maturation or functions of mature myotubes.
Genes Differentially Expressed in Mature Myotubes or during Myotube Maturation-To identify genes differentially ex-pressed in mature myotubes, gene expression profiles were compared between RR/KI-mTOR myotubes formed in the absence and presence of rapamycin at 72 h. -Fold change of 1.7 and 2.0 were used as cutoffs to identify genes up-regulated and down-regulated in mature myotubes, respectively. The lower threshold of 1.7-fold change (nevertheless, statistically significant by analysis of variance test) was used because we were particularly interested in genes up-regulated in mature myotubes. A total of 223 genes were found, of which 97 genes displayed higher expression levels in mature myotubes (supplemental data Table S2) and 126 genes lower expression levels (supplemental data Table S3). 15 of the 97 up-regulated genes belong to clusters (3, 1) or (3, 2) (Fig. 6), and the implication of their functions was discussed earlier. However, 14 of the 97 genes belong to cluster (3,3), which means they were upregulated in both nascent and mature myotubes when compared with myoblasts, but more highly expressed in mature than nascent myotubes. All 97 genes may be involved in myotube maturation, or function in mature myotubes. The 126 down-regulated genes may negatively impact myotube maturation and therefore may be suppressed during maturation.
Functional grouping of those two classes of genes is summarized by the pie graphs shown in Fig. 7. In both classes almost half of the genes are unknown. Notably, 19% of the genes (35% of the known genes) up-regulated in mature myotubes are involved in energy metabolism, which most likely reflects the functional maturation of skeletal myotubes. Interestingly, a FIG. 6. Microarray data clustering by self-organizing map. 487 genes displaying 2-fold or more change at least at one time point compared with time 0 were used for the clustering analysis. The horizontal axis represents time (h) and rapamycin treatment (100 nM). The vertical axis represents -fold change in log scale. The genes with similar temporal expression patterns are clustered and the number of genes in each cluster is indicated. A complete list of those genes can be found in supplemental data Table S1. significant number of genes either up-regulated or down-regulated (9 and 11%, respectively) are involved in signal transduction. A large number of genes encoding matrix/structural proteins are up-regulated (13%), and even more genes in that category down-regulated (18%), in mature myotubes. This is consistent with the structural remodeling of the myotube as it grows and matures.
It is conceivable that certain proteins important for myotube maturation may be present only transiently during differentiation. Indeed, IL-4 has been reported to express temporarily during myogenesis and down-regulated in mature myotubes (10). Therefore, we also looked for genes expressed at higher levels in RR/KI-mTOR cells in the absence of rapamycin than in the presence of rapamycin at 24 or 48 h but not at 72 h. These genes are listed in supplemental data Table S4. Of particular interest is Adamts1 (a distintegrin and metalloprotease with thrombospondin motifs), which encodes a secreted factor thought to be involved in extracellular matrix remodeling.
Among all the genes preferentially up-regulated during myotube maturation, three encode for secreted factors. They are the IGF-binding protein, Igfbp-5, diazepam binding inhibitor, and Adamts1. Because of the limitation of available reagents, we were only able to examine IGFBP-5 in mTOR-regulated myotube maturation. So far we did not find IGFBP-5 to reverse the small myotube phenotype in rapamycin-treated RR/KI-mTOR cells (data not shown).
Confirmation of Microarray Data by Quantitative RT-PCR-To validate our microarray experiments, we selected 7 genes from various categories described above, and analyzed their expression patterns by real-time quantitative RT-PCR. The selected genes were Igf-2, Gpx3, Bop, Cyclin G, Scd2, SerpinB1A, and Adamts1. Independent RNA samples prepared similarly as those used in the microarray hybridization experiments were subjected to qRT-PCR. As shown in Fig. 8, for all the genes tested, the expression profiles over the time course and in response to rapamycin treatment are consistent between the microarray and qRT-PCR experiments. This confirmation gives us high confidence in the microarray data. DISCUSSION Studies of rapamycin-sensitive signaling in skeletal muscle differentiation so far have mostly relied on the inhibitor rapa-mycin. Whereas an essential role for mTOR in myoblast differentiation has become apparent (24 -28), it is not known whether mTOR simply controls one critical molecular event or regulates multiple stages of myogenesis. In the current investigation we took advantage of cells stably expressing a rapamycin-resistant mTOR or its kinase-inactive version to probe the function of mTOR in distinct events leading to the formation of mature myotubes. Our results reveal the role of mTOR as a master regulator of skeletal myogenesis by controlling multiple processes through different mechanisms (see the model in Fig. 5B). In cells relying on kinase-inactive mTOR, the entry into differentiation is intact, as indicated by the normal expression of myogenin (Fig. 2) and IGF-II (29). The expression of contractile proteins and formation of nascent myotubes also proceed to the full extent, although suffering a delay compared with cells with intact mTOR kinase activity (Figs. 3 and 4, Table II). On the other hand, the formation of mature myotubes is absolutely dependent on wild-type mTOR kinase activity ( Figs. 1 and 4). Although we cannot definitively rule out the possibility of kinase-inactive mTOR acting as a dominant negative mutant, this scenario is unlikely considering that the expression level of recombinant protein in those cells was not in excess compared with the endogenous level, and that in the absence of rapamycin the cells behaved identically to parental cells. Regardless of the mode of action by kinase-inactive mTOR, our data clearly indicate differential regulatory mechanisms by mTOR utilizing different levels of its kinase activity. The physiological significance of our findings awaits confirmation by future in vivo studies, such as the generation and examination of skeletal muscle-specific mTOR-null animals, because conventional mTOR knock-out leads to early embryonic lethality in mice (33,34).
The concept that there are two distinct stages of myocyte fusion in mammalian muscle cells has been proposed by Pavlath and colleagues (7), who have demonstrated that NfatC2Ϫ/Ϫ muscle cells are able to fuse and form myotubes with a limited number of nuclei, but that subsequent myotubemyoblast fusion is abolished in those cells (7). Here, we identify another molecular pathway, most likely independent of the NFATC2 pathway, that specifically regulates myotube maturation. Our observations further validate the idea that myoblast-myoblast fusion and subsequent myoblast-myotube fu-  Tables S2 and S3. sion are separable events controlled by distinct molecular mechanisms. It is noteworthy that Wilson et al. (35) has recently reported the involvement of PI3K in skeletal myocyte maturation (35). They observed in cultured myocytes that inhibition of differentiation by the PI3K inhibitor LY294002 was partially rescued by the expression of an active Akt, resulting in smaller myotubes with fewer myonuclei. It is therefore possible that PI3K utilizes Akt to regulate nascent myotube formation, whereas a different effector mediates function of PI3K in the second-stage myocyte fusion leading to myotube maturation, which presents a striking analogy to the dual roles of mTOR signaling. Alternatively, inhibition of mTOR may be directly accountable for the defect in myotube maturation in the presence of LY294002, because LY294002 is also an effective inhibitor of the mTOR kinase activity at the concentrations required to inhibit PI3K (36).
Pavlath and colleagues (10) have identified the cytokine IL-4, acting downstream of NFATC2, as a myoblast recruitment factor for myoblast-myotube fusion. Interestingly, mTOR also regulates the secretion of a factor that is essential for the second stage myotube fusion (Fig. 5), and this factor does not appear to be IL-4 (data not shown). It is feasible that the mTOR pathway may act in parallel with NFATC2 to regulate muscle fusion. IL-4 has been known to regulate macrophage fusion (37). The new revelation of the role of IL-4 in muscle cells suggests that fusion mechanisms may be conserved among different cell types. Another type of cell fusion occurs during osteoclast differentiation, when monocyte/macrophage precursor cells fuse at the bone surface. Two secreted factors are both necessary and sufficient for this fusion process: the tumor necrosis factor-related cytokine RANKL (receptor activator of NF-B ligand), and the growth factor CSF-1 (colony-stimulating factor-1) (38). Our preliminary experiments indicated that neither of those factors (or combined) was sufficient to rescue the rapamycin-induced myotube fusion defect (data not shown). It is possible that the secreted factor in question has not been previously linked to any type of cell fusion.
Gene profiling comparing myoblast and myotube stages of C2C12 cells was reported previously (32). Our microarray data on RR/KI-mTOR C2C12 cells differentiating under normal conditions generally agreed with the previous report, even though a complete overlap was not expected because of the different experimental conditions and different coverage of the genome, our chip was a 15,000 cDNA array and theirs an unrelated 11,000 oligo array (32). Our microarray analysis was aimed at gathering more specific information: the comparison of RR/KI- mTOR cells differentiating in the absence and presence of rapamycin revealed potential targets of mTOR kinase-dependent signaling during myocyte differentiation and maturation, and furthermore, the data contributed to our general understanding of the gene expression programs specifically underlying myotube maturation. It is not surprising that most genes found to associate with myotube maturation fall into one of these functional categories: energy/metabolism, extracellular matrix/structure protein, and signal transduction, but many of them have not previously been linked to myotube maturation or even skeletal muscle differentiation. Our data provide a starting point from which the molecular wiring of skeletal muscle growth can be potentially delineated.
Of particular interest are three genes encoding secreted factors: Igfbp-5, Adamts1, and diazepam binding inhibitor, which appear to be up-regulated during differentiation and inhibited upon rapamycin-induced arrest of nascent myotubes. IGFBP-5 is the predominant IGFBP in C2C12 cells as well as mouse muscle tissues, and its expression positively correlates with differentiation in vitro (39,40). Paradoxically, exogenous IGFBP-5 is known to inhibit myoblast differentiation (41). Like most IGF-binding proteins, IGFBP-5 can modulate IGF actions either positively or negatively, depending on a number of factors in a specific cellular context (42). Inhibition of Igfbp-5 expression by rapamycin in C2 cells was previously reported (43). Our observation not only confirms the rapamycin sensitivity but also suggests that the kinase activity of mTOR is required for Igfbp-5 expression. Although exogenous IGFBP-5 alone does not seem to reverse the rapamycin inhibition of myotube size in RR/KI-mTOR cells (data not shown), further investigation is warranted to examine the involvement of IGFBP-5 in muscle maturation and its regulation by mTOR signaling. ADAMTS1 belongs to a novel family of extracellular matrix proteases that are related to the ADAM (a disintegrin and metalloprotease) family (44). One ADAM member, ADAM12 (meltrin ␣), is thought to participate in myoblast differentiation and myotube formation (45). Unlike the ADAM proteins that are transmembrane proteins, ADAMTS1 is a secreted protein. Implicated in inflammatory response, angiogenesis, and organ morphogenesis (44), the involvement of ADAMTS1 in skeletal muscle development or function has never been reported. Although a recombinant ADAMTS1 expressed in COS-7 cells was found associated with the extracellumar matrix and not secreted into the medium (46), the possibility of ADAMTS1 acting as a "fusion factor" in C2C12 cells cannot be dismissed until further investigation. The involvement of diazepam binding inhibitor in myogenesis is also a novel concept that should be tested. Finally, one needs to keep in mind that the regulation of the fusion factor by mTOR could occur at a post-transcriptional level, such as the well established translational regulation, and thus would not be revealed by gene profiling analysis.