Inhibition of Myoblast Differentiation by Tumor Necrosis Factor α Is Mediated by c-Jun N-terminal Kinase 1 and Leukemia Inhibitory Factor*

The proinflammatory cytokine, TNFα plays a major role in muscle wasting occurring in chronic diseases and muscular dystrophies. Among its other functions, TNFα perturbs muscle regeneration by preventing satellite cell differentiation. In the present study, the role of c-Jun N-terminal kinase (JNK), a mediator of TNFα, was investigated in differentiating myoblast cell lines. Addition of TNFα to C2 myoblasts induced immediate and delayed phases of JNK activity. The delayed phase is associated with myoblast proliferation. Inhibition of JNK activity prevented proliferation and restored differentiation to TNFα-treated myoblasts. Studies with cell lines expressing MyoD:ER chimera and lacking JNK1 or JNK2 genes indicate that JNK1 activity mediates the effects of TNFα on myoblast proliferation and differentiation. TNFα does not induce proliferation or inhibit differentiation of JNK1-null myoblasts. However, differentiation of JNK1-null myoblasts is inhibited when they are grown in conditioned medium derived from cell lines affected by TNFα. We investigated the induced synthesis of several candidate growth factors and cytokines following treatment with TNFα. Expression of IL-6 and leukemia inhibitory factor (LIF) was induced by TNFα in wild-type and JNK2-null myoblasts. However, LIF expression was not induced by TNFα in JNK1-null myoblasts. Addition of LIF to the growth medium of JNK1-null myoblasts prevented their differentiation. Moreover, LIF-neutralizing antibodies added to the medium of C2 myoblasts prevented inhibition of differentiation mediated by TNFα. Hence, TNFα promotes myoblast proliferation through JNK1 and prevents myoblast differentiation through JNK1-mediated secretion of LIF.

The proinflammatory cytokine, TNF␣ plays a major role in muscle wasting occurring in chronic diseases and muscular dystrophies. Among its other functions, TNF␣ perturbs muscle regeneration by preventing satellite cell differentiation. In the present study, the role of c-Jun N-terminal kinase (JNK), a mediator of TNF␣, was investigated in differentiating myoblast cell lines. Addition of TNF␣ to C2 myoblasts induced immediate and delayed phases of JNK activity. The delayed phase is associated with myoblast proliferation. Inhibition of JNK activity prevented proliferation and restored differentiation to TNF␣treated myoblasts. Studies with cell lines expressing MyoD:ER chimera and lacking JNK1 or JNK2 genes indicate that JNK1 activity mediates the effects of TNF␣ on myoblast proliferation and differentiation. TNF␣ does not induce proliferation or inhibit differentiation of JNK1-null myoblasts. However, differentiation of JNK1-null myoblasts is inhibited when they are grown in conditioned medium derived from cell lines affected by TNF␣. We investigated the induced synthesis of several candidate growth factors and cytokines following treatment with TNF␣. Expression of IL-6 and leukemia inhibitory factor (LIF) was induced by TNF␣ in wildtype and JNK2-null myoblasts. However, LIF expression was not induced by TNF␣ in JNK1-null myoblasts. Addition of LIF to the growth medium of JNK1-null myoblasts prevented their differentiation. Moreover, LIF-neutralizing antibodies added to the medium of C2 myoblasts prevented inhibition of differentiation mediated by TNF␣. Hence, TNF␣ promotes myoblast proliferation through JNK1 and prevents myoblast differentiation through JNK1-mediated secretion of LIF.
Massive loss of skeletal muscle occurs in a variety of disorders: chronic catabolic conditions such as AIDS, sepsis, and cancer and genetic disorders, collectively known as muscular dystrophies. In all cases of muscle wasting, including chronic diseases or genetic disorders, it is believed that inflammation contributes to the process (1). Neutrophiles and macrophages invade the damaged muscle tissue and secrete proinflammatory cytokines (2). Chronic secretion of proinflammatory cytokines, tumor necrosis factor ␣ (TNF␣) 2 in particular, has been found to actively damage the muscle tissue. In fact, blocking TNF␣ function significantly reduces the atrophy of skeletal muscle (3). TNF␣ is involved in many of the events leading to muscle wasting: inhibition of the differentiation process, inhibition of protein synthesis, selective proteolysis, and activation of caspase-mediated apoptosis (4,5). Thus, muscle loss is most probably a result of the perturbed balance between muscle degeneration and regeneration caused by proinflammatory cytokines.
Muscle regeneration occurs by the expansion and differentiation of satellite cells that lie beneath the basal lamina (6). Normally these cells are quiescent (resting), expressing the paired box transcription factor Pax7, but not the myogenic regulatory factors (MRFs). In response to trauma, such as injury, these cells are activated, first by initiating the expression of MyoD and then by proliferation. Activation of satellite cells is promoted by the release of growth factors such as hepatocyte growth factor (HGF), fibroblast growth factors (FGFs), and cytokines like IL-6 from the injured tissue and infiltrating macrophages. Finally, the expression of myogenin is induced, and cells differentiate and fuse with preexisting myofibers (reviewed in Ref. 7). Perturbation of muscle regeneration at any stage: activation, proliferation, or differentiation, causes fatal loss of muscle tissue. TNF␣ perturbs muscle regeneration in chronic and genetic disorders, though the mechanism is still obscure (8).
TNF␣ affects several intracellular signaling pathways leading to the activation of NFB, caspase 8, and stress-induced pathways like c-Jun N-terminal kinase (JNK) and p38 MAPK (9). Whereas the involvement of NFB and caspases in muscle wasting has been extensively studied, relatively little is known about the role of stress pathways. Several studies have indicated that NFB activity inhibits muscle differentiation by reducing the levels of the MyoD protein and activating the expression of cyclin D1 (10 -12). The role of JNK in the effect of TNF␣ on myogenesis has been less investigated. A recent study suggested that activation of JNK by TNF␣ blocks IGF-1 signaling necessary for the differentiation of myoblasts (13). Yet, the mechanism has not been fully explored; is JNK involved in myoblast proliferation, or alternatively is JNK involved in inhibition of myoblast differentiation but not in proliferation? Which isoforms of JNK participate in these processes? How does JNK affect these processes?
In the present work we approached these questions. We have shown that TNF␣ induces proliferation via JNK and indirectly prevents differentiation of myoblasts. The JNK1 isoform is involved in these processes. Interestingly, JNK1 mediates the expression of LIF, which prevents the differentiation process independently of proliferation. Therefore, our results indicate that TNF␣ affects myoblast proliferation directly via JNK1 and myoblast differentiation indirectly through the autocrine effect of LIF.

EXPERIMENTAL PROCEDURES
Materials-SP600125 (JNK inhibitor) was purchased from BIOMOL, dissolved in DMSO, and added to the culture medium to a final concentration of 20 M. TNF␣ was purchased from Cytolabs, was dissolved in water, and added to the culture medium to a final concentration of 20 ng/ml. LIF was purchased from Calbiochem and was added to cell culture to a final concentration of 20 ng/ml. Cytosine arabinoside (Ara-C) was purchased from Sigma-Aldrich, dissolved in PBS, and added to culture medium to a final concentration of 100 M. Thymidine was added to cell medium at a concentration of 2 mM. Neutralizing LIF antibody was from Chemicon and was added at 1 g/ml.
Cell Culture-C2 cells were a gift from Dr. David Yaffe (14). Cell lines were maintained in Dulbecco's modified Eagle's medium supplemented with 15% calf serum (HyClone), penicillin, and streptomycin (growth medium, GM). To induce differentiation, we used Dulbecco's modified Eagle's medium supplemented with 10 g of insulin per ml and 10 g of transferrin per ml (differentiation medium, DM). Fibroblasts from different mice strains (wild type, JNK1 Ϫ/Ϫ , JNK2 Ϫ/Ϫ , c-Jun Ϫ/Ϫ , and IKK␤ Ϫ/Ϫ ) were infected with a retrovirus encoding the MyoD protein and the hormone binding domain of estrogen receptor (pBABE puro MyoD:ER) (15). Myoblast cell lines were isolated following selection with puromycin (3 g/ml). These cells were grown in Dulbecco's modified Eagle's medium without phenol red supplemented with 15% calf serum (GM). Addition of 10 Ϫ7 M ␤-estradiol to DM induced translocation of the cytoplasmic chimera protein into the nucleus and initiation of the myogenic program (supplemental Fig. S1). In some experiments, conditioned media from certain cell lines (wild-type, JNK1 Ϫ/Ϫ , and JNK2 Ϫ/Ϫ MyoD:ER) were used to replace the medium of JNK1 Ϫ/Ϫ MyoD:ER cells grown in parallel. The medium was replaced every 6 h with fresh conditioned medium of growing cells.
Immunohistochemical Staining-Cells were fixed and immunostained. The primary antibodies used were monoclonal anti-MyHC (MF-20), anti-pc-Jun (Santa Cruz Biotechnology), monoclonal anti-phospho-histone H3 (Upstate Biotechnology), and anti-BrdU. Cells were exposed to secondary antibody and DAPI. The immunochemically stained cells were viewed at ϫ200 or ϫ650 magnification under a fluorescence microscope (Olympus, model BX50) and photographed with a digital camera.
Cell Cycle Analysis-BrdU was added to cell medium to a final concentration of 10 M. After 3 h, the cells were washed with PBS, fixed with methanol, and permeabilized in 0.25% Triton X-100. Following a PBS wash, the cells were incubated in 2 N HCL solution for 1 h. The solution was neutralized by washing the cells with 0.1 M borate buffer (pH 8.5). The cells were incubated for 90 min with 6 mg/ml anti-BrdU antibody (BMC9318, Roche Applied Science) in PBS containing 5% bovine serum albumin. Cells were exposed to a secondary antibody and DAPI and finally viewed at ϫ200 magnification under a fluorescence microscope (Olympus, model BX50).
Cells that were analyzed by fluorescence-activated cell sorting (FACS) were washed twice with PBS, fixed with 70% ethanol and treated with 2 mg/ml of RNase A (Worthington, Lakewood, NJ) for 30 min at 37°C. Ten minutes before analysis, propidium iodide (PI) was added at 100 g/l. DNA content was analyzed by using FACS and ModFit software (Becton Dickinson).
RT-PCR (Reverse Transcriptase PCR)-After removal of the culture medium, the cells were washed twice with ice-cold PBS. Total RNA extraction was performed using a commercial kit, TRI REAGENT TM (MRC). The procedure was conducted according to the manufacturer's protocol. RT reactions were performed following DNaseI treatment, phenol-chloroform extraction, and ammonium acetate precipitation. RNA was incubated with 1 l of a random hexamer (0.1 mg/ml), 4 min at 65°C. The RT reaction mixture was added including pairs of primers, which were incubated for 45 min at 42°C.

TNF␣ Induces Two Phases of JNK Activity Necessary to
Inhibit Myoblast Differentiation-TNF␣ was added to C2 cells during early differentiation stages, and JNK activity was measured by analyzing the phosphorylation state of c-Jun (Ser-63,73) (Fig. 1A). Two phases of c-Jun phosphorylation following the addition of TNF␣ were observed; an immediate phase at 30 -120 min and a late phase at 24 h (Fig. 1A). Similar results were obtained with an antibody to phospho-JNK, indicating that JNK is involved in c-Jun phosphorylation following TNF␣ treatment (Fig. 1A). Next, the expression of myogenic markers was analyzed (Fig. 1B). Addition of TNF␣ to DM reduced late expression of myogenin and myosin heavy chain (MyHC) (24 -48 h in DM). Because TNF␣ inhibits the expression of myogenic markers and myoblast differentiation (16), we asked whether JNK was involved in the process. TNF␣ was added to C2 cells during early differentiation stages, and differentiation was assessed 48 h later by analyzing myotubes formation (Fig. 1C). Differentiation was significantly prevented when TNF␣ was included in the medium. The addition of a pharma-cological inhibitor of JNK, SP600125, at a concentration preventing phosphorylation of c-Jun (Fig. 1C, lower panel) allowed the formation of myotubes in the presence of TNF␣ (Fig. 1C, upper panel). Partial rescue of differentiation in the presence of JNK inhibitor was also suggested by the expression of the early differentiation marker, myogenin and the late marker, MyHC (Fig. 1C). These results suggest that JNK is involved in TNFmediated inhibition of myoblast differentiation.
JNK1 but Not JNK2 Is Involved in Mediating the Effects of TNF␣ on Myoblasts-To further investigate the inhibitory role of JNK in myogenesis downstream of TNF␣ and to identify the JNK isoforms involved, myogenic lines were derived from JNK1-and JNK2-null fibroblasts (17). Expression vectors encoding MyoD-estrogen receptor chimera protein were stably integrated into wild-type, JNK1-, and JNK2-null fibroblasts (3T3) derived from knock-out mice. The inclusion of ␤-estradiol in the growth medium induces the translocation of the chimera protein (MyoD:ER) into the nucleus, where MyoD initiates the muscle program. The formation of myotubes in the three cell lines indicates that they acquired the myogenic phenotype ( Fig. 2A, upper panel). Proliferation of JNK1-deficient myoblasts was significantly slower than the other two cell lines (17). This cell line was also slower than the others in its differ- entiation rate (not shown). Interestingly, TNF␣ significantly inhibited differentiation of wild type as well as of JNK2-null cells, but did not affect JNK1-null cells ( Fig. 2A, lower panel).
The expression of myogenin and MyHC proteins analyzed by Western blotting indicated the same; the levels of these proteins were not affected following TNF␣ treatment in JNK1-null cells, whereas they were significantly reduced in wild-type and JNK2-null cells (Fig. 2B). Interestingly, this effect of TNF␣ on the expression of myogenic markers was observed late (48 h in DM) rather than early (24 h in DM) (Fig. 2B). Similarly, C2 myoblasts expressing shRNA directed to JNK1 partially restored MyHC expression to TNF␣-treated cells (supplemental Fig. S2). The results indicate that JNK1 but not JNK2 is involved in inhibition of differentiation by TNF␣. Next, we examined JNK kinase activity toward c-Jun following TNF␣ treatment of the three cell lines. Addition of TNF␣ to the wildtype cell line (3T3 MyoD:ER) induced two phases of c-Jun phosphorylation; the first at 30 min and the second at 24 h (Fig. 2C), a pattern similar to that observed in C2 myoblasts (see Fig. 1A). TNF␣ induced a much weaker phosphorylation of c-Jun in JNK1 Ϫ/Ϫ myoblasts relative to wild-type cells (Fig. 2C). Phosphorylation of c-Jun was not detectable 24 h following the addition of TNF␣ to JNK1 Ϫ/Ϫ myoblasts like in wild-type myoblasts. Addition of TNF␣ to JNK2-null cells induced a significant and sustained c-Jun phosphorylation that only disappeared later than 24 h (Fig. 2C). These results imply the involvement of JNK1 in the phosphorylation of c-Jun following TNF␣. The correlation between sustained and late phosphorylation of c-Jun and inhibition of myogenesis by TNF␣ indicates that the c-Jun protein itself may participate in a pathway downstream of TNF␣-inhibiting myogenesis. Therefore, we generated myoblasts from c-Jun Ϫ/Ϫ fibroblasts by expressing the MyoD:ER chimera protein. Similar to the other cell lines, the formation of myotubes indicated that these fibroblasts also acquired the myogenic phenotype, though their growth and differentiation rates were significantly slower than wild-type myoblasts (not shown). Interestingly, TNF treatment of these cells, did not reduce but rather induced expression of differentiation markers relative to control cells (Fig. 2D). Therefore, TNF␣ may have stimulated differentiation of this cell line in contrast to its inhibitory effect on wild-type and JNK2-null cells. Overall, TNF␣ did not inhibit differentiation of JNK1 and c-Jun-deficient myoblasts, indicating that these proteins mediate inhibition of differentiation by TNF␣. TNF␣ Induces a Wave of Myoblast Proliferation That Is Dependent on JNK1 and c-Jun Activities-The resistance of JNK1-and c-Jun-null cells to the inhibition of myoblast differentiation by TNF␣ was further investigated. JNK1-and c-Junnull fibroblasts share certain defects in cell cycle, significantly reducing their proliferation rates (17,18). Therefore, it is possible that TNF␣ induces proliferation of wild-type myoblasts and not of JNK1-and c-Jun-deficient myoblasts in a way that would explain its distinct effects on differentiation. We analyzed the percentage of C2 cells in S phase by BrdU incorporation (Fig. 3A). Twelve percent of C2 cells were in S phase following 24 h of growth in differentiation medium (DM; see "Experimental Procedures"). The number of C2 myoblasts in S phase was increased to 37% following treatment with TNF␣. knock-out and wild-type mice expressing the MyoD:ER chimera protein were differentiated in DM and ␤-estradiol in the absence or presence of TNF␣ (20 ng/ml). Forty-eight hours following TNF␣ treatment, cells were immunostained with primary anti-MyHC antibody. DAPI staining was used to detect cell nuclei. B, the above cell lines were grown in DM in the absence or presence of TNF␣ for the indicated time periods, and proteins were extracted and analyzed by Western blotting using antibodies to myogenin (F5D) and MyHC (MF20). Tubulin was used as a loading control. C, the above cell lines were grown in DM in the absence of presence of TNF␣ for the indicated time periods, and proteins were extracted and analyzed by Western blotting using antibodies to phosphorylated c-Jun (pc-Jun) and to total c-Jun protein (c-Jun). D, fibroblasts originated from c-Jun knock-out mouse and expressing the MyoD:ER chimera protein were differentiated in DM and ␤-estradiol in the absence or presence of TNF␣ (20 ng/ml). Proteins were extracted at different time periods after TNF␣ treatment and analyzed by Western blotting using antibodies to myogenin (F5D) and MyHC (MF20). Tubulin was used as a loading control.
The percentage of cells in S phase was decreased to 12% if the JNK inhibitor SP600125 was added to TNF␣-treated myoblasts (Fig. 3A). Hence, TNF␣ induced proliferation of differentiating C2 cells, which was likely mediated by JNK. We further analyzed the percentage of cells in S phase following the addition of TNF␣ to different MyoD:ER cell lines (Fig. 3B). The addition of TNF␣ to wild-type and JNK2-null cells significantly increased the number of cells in S phase after 24 h in DM relative to cells not treated with TNF␣ (Fig. 3B, upper panel). In contrast, TNF␣ did not increase the percentage of JNK1-or c-Jun-null cells in S phase after 24 h in DM (Fig. 3B, lower panel). FACS analysis of the different cell lines further substantiated the notion that TNF␣ did not induce proliferation of JNK1-and c-Jun-null cells, whereas it induced significant proliferation of wild-type and JNK2-null cells (Fig. 3C). This is suggested by the proportion of cells in S and G 2 /M stages relative to cells in G 1 stage of the cell cycle (Fig. 3C, left panel). In conclusion, TNF␣ induces the proliferation of myoblasts in a JNK1 and c-Jun-dependent manner.
The apparent correlation between proliferating myoblasts and c-Jun phosphorylation occurring 24 h following TNF␣ treatment may indicate that phosphorylated c-Jun was associated with proliferating cells. To further investigate this possibility, wild-type MyoD:ER myoblasts were grown for 24 h following their treatment with TNF␣ and were immunostained using anti-phospho-c-Jun, and anti-phospho-histone H3 (Fig.  3D). Phosphorylated histone H3 marks cells in mitosis. Interestingly, phospho-c-Jun staining was observed only in cells expressing phospho-histone H3 (Fig. 3D). Therefore, the second peak of c-Jun phosphorylation following TNF␣ treatment likely represents mitotic myoblasts. Interestingly, phospho-c-Jun appears to be excluded from DNA and is concentrated in what appears as centromeres of mitotic cells.
TNF␣ Inhibits Myoblast Differentiation Indirectly-The overt inhibition of differentiation by TNF␣ (see Figs. 1 and 3) could not be explained merely by the proliferating myoblasts, because not all myoblasts entered the cell cycle (see Fig. 3). Therefore, we hypothesized that the myoblasts that reentered the cell cycle could prevent differentiation of the remaining population of resting myoblasts. To investigate this possibility, we added Ara-C to differentiating cells treated with TNF␣. Ara-C is an analogue of cytosine used in chemotherapy, which upon incorporation to newly synthesized DNA kills cells in the S phase. The addition of TNF␣ to differentiation medium almost abolished the formation of myotubes (Fig. 4A, left  panel). However, when proliferating cells were eliminated by Ara-C treatment, most of the remaining cells formed myotubes despite the presence of TNF␣. Similarly, Ara-C treatment restored the expression of myogenic markers like myogenin and MyHC inhibited by addition of TNF␣ to differentiating myoblasts (Fig. 4A, right panel). A different approach was used to arrest cell cycle at the G 1 stage; thymidine was added to medium of myoblasts treated with TNF␣. Thymidine block allowed for myoblast differentiation in the presence of a TNF␣ concentration normally preventing differentiation (supplemental Fig. S3). The above results suggest that cell cycle arrest or removal of proliferating cells prevents the inhibitory effect of TNF␣, indicating that TNF␣ may not interfere directly with the differentiation of non-replicating myocytes. Consequently, we speculated that cells forced into the cell cycle by TNF␣ secrete growth factors/cytokines into the culture medium preventing the differentiation of the remaining myoblasts. To further investigate this possibility, we analyzed the differentiation of JNK1 Ϫ/Ϫ myoblasts grown in conditioned medium isolated from either JNK1 Ϫ/Ϫ , JNK2 Ϫ/Ϫ ,or wild-type myoblasts grown in the absence or presence of TNF␣ (Fig. 4B). Conditioned medium from JNK1 Ϫ/Ϫ cells grown without or with TNF␣ did not affect formation of myotubes (left panel) and muscle gene expression (right panel) of JNK1 Ϫ/Ϫ cells. However, conditioned medium originated from JNK2 Ϫ/Ϫ or wild-type cells treated with TNF␣ completely abolished muscle differentiation of JNK1 Ϫ/Ϫ cells. These results clearly indicate that wild-type and JNK2 Ϫ/Ϫ cells treated with TNF␣, but not JNK1 Ϫ/Ϫ cells secreted factor(s) into the growth medium that prevented the differentiation of JNK1 Ϫ/Ϫ cells.
Myoblasts Express the Proinflammatory Cytokines IL-6 and LIF in Response to TNF␣-Assuming that TNF␣ induced the secretion of growth factors or cytokines into the medium that inhibited the differentiation of myoblasts, we screened for the expression of certain growth factors and cytokines. The selected factors all involved cell proliferation, stemness, and antagonism of differentiation. C2 cells were treated with TNF␣, and 24 h later the expression of certain growth factors and cytokines was analyzed by semi-quantitative RT-PCR (Fig. 5A). The expression of some growth factors, known to inhibit myoblast differentiation, like bFGF, HGF, and PDGF was not affected by TNF␣ treatment. The expression of IGF-1, functioning both on proliferating as well as differentiating myoblasts, was down-regulated following TNF␣ treatment (Fig.  5A). The expression of two cytokines, IL-6 and LIF was substantially induced following TNF␣ treatment. We further analyzed the expression profile of these two cytokines in C2 cells treated with TNF␣ (Fig. 5B). Transcripts of IL-6 and LIF were identified in two peaks following the addition of TNF␣ to C2 cells; the first around 30 min to 2 h and the second around 24 h. This pattern is similar though not identical to that of c-Jun phosphorylation following the addition of TNF␣ to C2 cells (see Fig. 1A), suggesting that IL-6 and LIF expression may be mediated by JNK activity. The involvement of JNK1 in the expression of IL-6 and LIF was investigated using the chimera MyoD:ER-expressing cell lines. Addition of TNF␣ to wild-type and to JNK2 Ϫ/Ϫ myoblasts induced two phases of IL-6 and LIF expression, after 2 and after 24 h (Fig. 5C). In contrast, TNF␣ induced sustained expression of IL-6, but no expression of LIF in JNK1 Ϫ/Ϫ myoblasts (Fig. 5C). This result indicates that LIF but not IL-6 expression is dependent on JNK1 activity.
Because TNF␣ stimulates NFB activity, it was important to find out whether NFB was also involved in the expression of LIF. For analyzing NFB activation following TNF␣ treatments in the different cell lines, the cellular localization of p65, part of the NFB complex, was investigated (supplemental Fig. S4). P65 was translocated from the cytoplasm to cell nuclei following the addition of TNF␣ to wild-type MyoD:ER myoblasts, indicating that NFB was activated. Likewise, TNF␣ also induced the translocation of p65 into the nuclei of the other cell lines including JNK1-null myoblasts (supplemental Fig. S4). Therefore, it appears that NFB activity is induced by TNF␣ in JNK1 Ϫ/Ϫ myoblasts, which is not sufficient to induce the expression of LIF (see Fig. 5C). In another experiment, a myogenic cell line expressing MyoD:ER was derived from fibroblast cells originated from IKK␤-null mouse and therefore devoid of NFB activity (19). Massive apoptosis of these myoblasts occurred 48 h after the addition of TNF␣ to these cells (not shown). Still, TNF␣ induced a constitutive expression of LIF in the absence of NFB activity (supplemental Fig.  S5A). The constitutive expression of LIF correlated well with constitutive activity of JNK following TNF␣ treatment (supplemental Fig. S5B). Therefore, TNF␣-induced expression of LIF correlates with JNK activity and not with NFB activity.

TNF␣-induced Inhibition of Myoblast Differentiation Is Mediated by LIF-
The above results suggest that LIF may be the secreted factor mediating the inhibitory effect of TNF␣ on myoblast differentiation. To further explore this possibility, recombinant LIF protein was added to the medium of JNK1 Ϫ/Ϫ MyoD:ER cells. This cell line was selected because its differentiation process was not affected by TNF␣, and it did not synthesize LIF in response to TNF␣ (see Figs. 2, 3, and 5). Addition of LIF largely inhibited the differentiation of JNK1 Ϫ/Ϫ myoblasts judged by the expression of myogenin and MyHC in Western analysis and by immunostaining of myotubes with an antibody to MyHC (Fig. 6A). This inhibition is not a consequence of cell proliferation, because LIF did not increase the percentage of JNK1 Ϫ/Ϫ cells in S phase (data not shown). Therefore, the negative effect of LIF on JNK1 Ϫ/Ϫ MyoD:ER differentiation is reminiscent of conditioned medium originating from wild-type and JNK2 Ϫ/Ϫ cells treated with TNF␣ (see Fig. 4B). Consequently, our assumption is that LIF is the cytokine secreted by myoblasts in response to TNF␣, and it inhibits the ability of resting myoblasts to differentiate. To test this possibility, a LIF-neutralizing antibody was added to C2 myoblasts grown in DM in the absence or presence of TNF␣ (Fig. 6B). As observed before, TNF␣ abolished the differentiation of myoblasts. The neutralizing antibodies added to the differentiation medium (DM) restored differentiation of myoblasts, whereas control antibody did not reverse the effect of TNF␣ (Fig. 6B). Despite the partially restored differentiation, the inclusion of LIF-neutralizing antibody did not prevent proliferation of myoblasts by TNF␣ (BrdU staining, Fig. 6C). Interestingly, addition of recombinant LIF to C2 myoblasts at a concentration pre-α :

FIGURE 4. Conditioned medium from TNF␣-treated cells prevents the differentiation of JNK1 ؊/؊ myoblasts.
A, C2 myoblasts were grown in DM in the absence or presence of Ara-C. Two hours later, TNF␣ was added to cell medium and myoblasts were allowed to differentiate for 48 h. In one experiment (left panel), cells were fixed and immunostained using MyHC (MF20) as the primary antibody. DAPI staining was used to detect cell nuclei. In another experiment (right panel), proteins were extracted, and Western analysis was performed using antibodies to myogenin, MyHC, and tubulin. B, JNK1 Ϫ/Ϫ MyoD:ER cells were grown in medium originated from parallel growing cultures of JNK1 Ϫ/Ϫ , JNK2 Ϫ/Ϫ or wild-type myoblasts to which TNF␣ had been added or not. Cell medium was replaced every 6 h. Forty-eight hours later, the cells were fixed and immunostained using anti-MyHC (MF20) as the primary antibody. DAPI staining was used to detect total cell nuclei (left panel). In a second experiment, proteins were extracted 48 h after growth in conditioned medium, and Western analysis was performed using antibodies to myogenin, MyHC, and tubulin (right panel). venting their differentiation (not shown) did not induce their proliferation (Fig. 6C). In summary, we suggest a bimodal TNF␣ activity: it directly promotes myoblast proliferation through activation of JNK1, and it prevents the differentiation of non-dividing myoblasts through its induced secretion of LIF by the proliferating myoblasts.

DISCUSSION
TNF␣ affects many levels of muscle development and maintenance. Our study deals with the stages of myoblast growth and differentiation. The role of TNF␣ has been extensively studied because it is implicated both in the build up of muscle under normal healthy conditions and in its degeneration during many chronic diseases. Over the years in vitro studies have demonstrated that at high concentrations, TNF␣ induces myoblast proliferation, while low concentrations induce differentiation (20 -22). The opposite effects are likely a consequence of inducing different intracellular signaling pathways (21,22). Several mechanisms were suggested to describe inhibition of myoblast differentiation (11,12,(23)(24)(25). A few studies indicated the role of JNK in the inhibitory function of TNF␣ on myogenesis (13,26). These studies showed that JNK was involved in inhibiting IGF-I signaling, known to promote muscle growth and differentiation. Yet, none of these studies asked whether JNK was involved in myoblast proliferation, differentiation, or both. The results of the present study indicate two distinct activities of TNF␣ mediated by the JNK pathway, a pro-proliferative and an anti-differentiative pathway. Of note, all the experiments of this study were performed with cell lines, either fibroblasts expressing MyoD:ER chimera, or C2 myoblasts. Therefore, the conclusions drawn are limited to cell lines, and further studies with primary myoblasts should support more general conclusions.
TNF␣ Induces Proliferation of Myoblast through the JNK Pathway-Our experiments indicate that JNK activity is induced following treatment of myoblasts with TNF␣. JNK activity is apparent by a typical two-peak phosphorylation of c-Jun at Ser-63 and -73; the first after 30 min (immediate) and the second after 24 h (delayed). TNF␣ does not affect the initiation of myogenin expression (6 -12 h), but gradually reduces its later expression (24 -48 h). This pattern of myogenin expression is well correlated with the TNF␣-induced entry of myoblasts into the cell cycle with maximum cells in the S phase after 24 h. Proliferation is completely prevented if a pharmacological inhibitor of JNK is added to cells, suggesting the involvement of the pathway in the process. A more precise examination of JNK isoforms involved, using cell lines lacking the JNK1 or the JNK2 gene, indicates that JNK1 is necessary for the proliferation induced by TNF␣. This is not surprising because JNK1-null fibroblasts exhibit slow entry into the S phase and a delayed cell cycle progression relative to JNK2-null fibroblasts with an inverse phenotype of accelerated cell cycle (17). Moreover, addition of TNF␣ to fibroblasts was shown to induce JNK1 but not JNK2 activity in the phosphorylation of c-Jun (27). Therefore, JNK1 is also the major mediator of TNF-induced proliferation in myoblasts. Because JNK1 is the major and possibly the only kinase of c-Jun Ser-63,73 residues (17) and fibroblasts missing the c-Jun gene display a delayed entry into the S phase (18), we analyzed how TNF␣ affected c-Jun-deficient myoblasts. Interestingly, not only proliferation of c-Jun Ϫ/Ϫ myoblasts was slow but also their differentiation rate. Moreover, TNF␣ did not induce any proliferation of these cells and did not inhibit their differentiation. On the contrary, it appears that TNF␣ may have even accelerated differentiation. Presently, the basis for this observation is not clear, though recent studies may provide a plausible explanation (21,22). Low levels of autocrine TNF␣ induce the p38 MAPK pathway, which is necessary for myoblast differentiation in cell culture and for regeneration in injured muscle of mouse model (21). High levels of TNF␣ also induce, in addition to p38, the JNK pathway, which may be dominant over p38 and promote proliferation of myoblasts (28). Because the JNK pathway is not fully functional in JNK1-and c-Jun- defective myoblasts, these cells may respond to TNF␣ by sustained induction of p38, promoting cell cycle arrest and subsequent differentiation (29 -32). This model awaits further investigation. An interesting finding of the present study is the simultaneous phosphorylation of c-Jun and the wave of proliferating myoblasts occurring 24 h following the addition of TNF␣. The involvement of the JNK pathway and c-Jun in cell proliferation is well known, though the exact roles in cell cycle progression are poorly defined. c-Jun was suggested before to be involved in G 1 progression via the expression of cyclin D1 (33). However c-Jun may also be involved in mitosis. Nocodazole-released cells progressing through mitosis exhibit high phosphorylation of c-Jun at Ser-63 that disappears by the end of mitosis (33). Moreover, recent studies indicated that JNK is associated with the centrosome and is active from S phase through anaphase (34) and that JNK activity is necessary to complete mitosis in several cell lines (35). Here, we find that following treatment of myoblasts with TNF␣, phospho-c-Jun (Ser-63,73) is detected only in mitotic cells. Interestingly, expression of phospho-c-Jun is excluded from the chromosomes and is mostly concentrated at what appears as centrosomes. The association of phospho-c-Jun, like that of phospho-JNK (35) with centrosomes of mitotic cells, may be an indication of its role at this particular stage and location. Supporting this idea is the finding that expression of a c-Jun mutant in which serine residues 63 and 73 were replaced with non-phosphorable alanine resulted in a G 2 /M cell cycle block (36).
LIF Mediates the Anti-differentiation Function of TNF␣-The robust inhibition of differentiation by TNF␣ cannot be explained merely by its mitotic effect. Our results indicate that TNF␣ probably induced a burst of proliferation in a subset of myoblasts. Still, differentiation of myoblasts even after proliferation has ended is overtly prevented. We also find that myoblast cell lines resistant to induced proliferation by TNF␣, JNK1 Ϫ/Ϫ differentiate as well as untreated myoblasts or even better in the case of c-Jun Ϫ/Ϫ myoblasts. Therefore, a non-autonomous effect mediated by secreted factor(s) was considered. It was demonstrated before that incubation of differentiated C2C12 cells with TNF␣ induced the expression of proinflammatory cytokines like IL-6 and IFN-␥ (37). In this study we find that TNF␣ induces the expression of cytokines of the same family, IL-6 and LIF. Although, these two cytokines appear functionally redundant in many biological systems, our results indicate that they are differently regulated by TNF␣ and may exhibit separate activities on myoblasts. TNF␣ induced the expression of IL-6 but not of LIF in JNK1-deficient myoblast cells, indicating that JNK1 affects the expression of LIF in myoblasts. Because TNF␣ also induces the activity of NFB, which by itself is known to regulate the expression of many cytokines, we investigated the involvement of NFB in the expression of these cytokines. We took advantage of a myo- FIGURE 6. LIF mediates the inhibitory activity of TNF␣ on myoblast differentiation. A, JNK1 Ϫ/Ϫ MyoD:ER cells were grown in DM in the absence or presence of LIF (20 ng/ml). Forty-eight hours later, proteins were extracted and analyzed by Western blotting (left panel), and in a parallel experiment, cells were fixed and immunostained using anti MyHC (MF20) as the primary antibody. DAPI staining was used to detect total cell nuclei (right panel). B, C2 myoblasts were grown in DM to which either IgG from preimmune serum or anti LIF neutralizing antibody (1 g/ml) was added. TNF␣ was added to the medium 2 h later, and myoblasts were allowed to differentiate for 48 h. Proteins were extracted and analyzed by Western blotting (left panel), and in a parallel experiment, cells were fixed and immunostained using anti-MyHC (MF20) as the primary antibody. DAPI staining was used to detect total cell nuclei (right panel). C, C2 myoblasts were grown in DM to which anti LIF-neutralizing antibody (1 g/ml) was added. TNF␣ was added to the medium 2 h later, and myoblasts were allowed to differentiate for 24 h. To identify the nuclei involved in DNA synthesis, BrdU was added to DM 3 h before cells were fixed and immunostained, as described under "Experimental Procedures." Histograms represent an average of five different fields that were counted to calculate the percentage of cells in the S phase. The percentage of cells in the S phase was determined by dividing the number of BrdU-stained nuclei by the number of DAPI-stained nuclei in each microscopic field.
blast cell line devoid of the IKK␤ gene and thus exhibiting no NFB activity (19). In response to TNF␣, these cells expressed constitutive levels of LIF found in correlation with constitutive phosphorylation of c-Jun. Therefore, the expression of LIF is dependent on JNK and not on NFB activity. The effect of IL-6 was not further analyzed in the present study, because its expression in JNK1 Ϫ/Ϫ myoblasts treated with TNF␣ did not inhibit their differentiation. IL-6 was suggested as a myoblast differentiation-promoting cytokine in previous studies (38,39). Therefore, IL-6 is most likely not the cytokine mediating the inhibitory effects of TNF␣ on myoblast differentiation.
LIF is expressed following muscle injury and is necessary for muscle regeneration (40,41). LIF was shown to promote proliferation and inhibit differentiation of C2C12 and primary myoblasts (42,43). Signaling molecules like Stat3 (44) and ERK MAPK (45) function downstream of LIF in promoting proliferation of myoblasts. Interestingly, none of the previous studies distinguished the positive effect of LIF on proliferation from its negative effect on differentiation. The experiments presented here indicate a straightforward inhibitory function of LIF on differentiation without affecting proliferation. First, inclusion of a neutralizing antibody to LIF in myoblasts treated with TNF␣ restored partial differentiation without affecting the amount of proliferating cells. Second, addition of recombinant LIF to C2 myoblasts or to JNK1 Ϫ/Ϫ MyoD:ER cells growing in differentiation medium prevented differentiation without inducing further proliferation. These results do not necessarily contradict the results of previous studies but indicate that under certain conditions of either cell cycle-defect (JNK1 Ϫ/Ϫ cells) or during the process of myoblast differentiation, LIF cannot induce proliferation but can prevent differentiation. Under these conditions, one can analyze the net effect of LIF on the differentiation process and ask what are the intracellular events downstream of LIF preventing differentiation; how does LIF inhibit the function of myogenic regulatory factors? These questions are being investigated in our laboratory. Understanding the inhibitory effect of LIF on myoblast differentiation may extend beyond this particular process and lead to the understanding of how LIF can preserve multipotentiality of embryonic as well as adult stem cells.