In this study, we found that in vitro TNF-α treatment was able to significantly decrease the surface area of myotubes deriving from either the L6 or the C2C12 lines, with this effect being reproduced by addition of cell-permeating ceramides. Furthermore, both TNF-α and ceramide decreased the CK activity and MHC content of L6 myotubes. These observations suggested that the atrophic effects of TNF-α on muscle cells might rely on the production of ceramide triggered by the cytokine.
To verify this hypothesis, we used different ceramide-synthesis inhibitors together with TNF-α. Both a de novo
pathway inhibitor (myriocin) and two sphingomyelinase inhibitors (GW4869 and OMS) were able to suppress the effects of TNF-α on L6 myotube size, myosin heavy and light chain content, and CK activity, suggesting that ceramide formed by either of the pathways mediates the atrophic effects of TNF-α. The effects of inhibitors of the two different ceramide-synthesis pathways were not additive, suggesting that a mere reduction in ceramide formation, rather than complete suppression, is enough to prevent cell atrophy. Alternatively, this could result from the overall depletion of sphingolipids induced by the blockade of de novo
synthesis by myriocin [19
], which potentially also reduced sphingomyelinase-mediated ceramide synthesis. The sphingomyelinase inhibitors GW4869 and OMS also showed protective effects against TNF-α-induced atrophy in C2C12 myotubes, thereby confirming the involvement of ceramide formed by sphingomyelinase activation. However, in this cell line, myriocin was devoid of protective effects and in fact, showed a negative effect by itself on myotube size, contrary to the results in L6 myotubes. A possible explanation is that in the C2C12 line de novo
sphingolipid synthesis has to be maintained above a certain threshold, so as to supply cells with a compound that can be either a structural component such as sphingomyelin or a glycosylceramide, or a mediator such as S1P, and this would be essential to maintain cell homeostasis.
Quantification of the sphingolipids established that TNF-α markedly increased the ceramide content of L6 myotubes, consistent with a role for this sphingolipid mediator in the atrophic response. Ceramide accumulation was accompanied by a substantial decrease in sphingomyelin cell content, showing that ceramide resulted, at least in part, from sphingomyelin hydrolysis. However, among the molecular species of ceramide formed under TNF-α stimulation, the C18:0 species was found to be one of the most important (its increase representing 6.2% of the sum of ceramide species in the control), whereas there was no parallel hydrolysis of C18:0-sphingomyelin (Table ). It can thus be hypothesized that TNF-α was able to induce de novo synthesis of C18:0 ceramide, and to produce other species, mainly C16:0 and C24:1 ceramides, via sphingomyelinase activation. Myriocin totally prevented the TNF-α-induced rise in ceramide, and in parallel decreased the levels of sphingomyelin, supporting the above proposal that it is able to inhibit both the de novo and the sphingomyelinase pathways of ceramide synthesis. Similar to myriocin, the sphingomyelinase inhibitors were able to diminish TNF-α-induced ceramide rise, although to a lesser extent, and as expected, these inhibitors significantly restored the sphingomyelin cellular content. On the whole, these data are compatible with the interpretation that myotube atrophy induced by TNF-α involves accumulation of ceramide, and that the inhibitors counteracted the TNF-α effects by preventing ceramide synthesis.
Because ceramide can be rapidly metabolized and give rise to other mediators, of which S1P has particular importance in cell physiology [5
], we considered the possibility that this ceramide metabolite could interfere in the response of L6 myotubes to TNF-α stimulation. We found that S1P displayed a positive influence on myotube integrity, opposite to that of exogenous ceramide or TNF-α. A lowering of S1P availability might thus participate in the negative effects of TNF-α. Consistent with this idea, these effects were not amplified by S1P receptor inhibition, suggesting that S1P levels had already been reduced below an effective concentration by TNF-α treatment. These observations are in agreement with the reported attenuation of denervation-induced soleus atrophy by S1P perfusion in the rat [35
It is well known that muscle atrophy results from both decreased protein synthesis and accelerated proteolysis. In agreement with a pro-atrophic role of TNF-α, we found that the cytokine lowered protein synthesis and tended to increase proteolysis in L6 myotubes. These effects were reversed by the ceramide-synthesis inhibitors myriocin and OMS, indicating that ceramide is involved in the effects of TNF-α on protein metabolism. Concerning the elements of the proteolytic machinery that are regulated by ceramide, we could observe that ceramide-synthesis inhibition markedly lowered the expression of Atrogin-1 ubiquitin ligase, which was increased by TNF-α. As this atrogene plays a recognized role in the catabolism of muscle-specific proteins, at least part of the TNF-α/ceramide atrophic effects can be related to the modulation of this factor, and to the resulting decrease in eIF3f translation initiation factor levels. In addition, we found that ceramide-synthesis inhibitors significantly lowered the expression of LC3b, an essential component of the autophagic proteolytic system. The activation of autophagy by ceramide has already been described in other cell systems [13
]. It is thus possible that ceramide positively regulates both proteasome- and autophagy-dependent protein degradation in differentiated myotubes, possibly through coordinated changes in Foxo3 transcription factor phosphorylation. However, the transcriptional regulation of autophagy by ceramide in L6 myotubes remains speculative, in view of the absence of effect of ceramide-synthesis inhibitors on the expression of several other autophagy-related genes.
Protein homeostasis is, for a large part, under the control of signaling pathways of which the central elements are the kinases Akt and mTOR. These interconnected regulators integrate inputs from growth factors, nutrient availability, and energy levels so as to adapt protein synthesis and degradation to the physiological status of the cell. Akt is a major inhibitor of proteolysis through the control of Foxo transcription factors, which in turn regulate the expression of ubiquitin ligases involved in the specific degradation of muscle proteins by proteasome [16
]. mTOR kinase, in complex with the protein Raptor (mTORC1 complex), is indirectly activated by Akt, through phosphorylation of its inhibitor tuberous sclerosis complex [15
]. Activated mTORC1 is known to enhance protein translation, particularly through the activation of its substrate S6K1, and has been shown to be required for muscle hypertrophy [36
]. mTORC1 is also a major negative regulator of autophagy [15
]. Furthermore, mTOR also exists in complex with the protein Rictor to form the mTORC2 complex, which is able to phosphorylate and activate Akt. Both mTOR complexes are stimulated by the phospholipid messenger phosphatidic acid [17
], the product of the action of the signaling enzyme of cell membranes, PLD. Moreover, involvement of PLD in mTOR activation in response to exercise has been shown, suggesting its role in muscle-tissue hypertrophy [37
Ceramide is considered a general inhibitor of PLD, acting at the catalytic site, on the recruitment of activator proteins, and also at the transcriptional level [38
]. We previously found that ceramide selectively inhibits expression of the PLD1 isoform of the enzyme in L6 myoblasts [9
]. In the present study, we found that TNF-α markedly decreased expression of PLD1, and that ceramide-synthesis inhibitors rescued its expression, suggesting that PLD1 is one major target of ceramide in this signaling network. Because PLD is an activator of both mTOR complexes, we then considered the influence of these inhibitors on the mTORC1 substrate S6K1 and the mTORC2 substrate Akt. Ceramide inhibition in the presence of TNF-α increased the amounts of both S6K1 and Akt under the phosphorylated/activated state, possibly as a consequence of PLD1 upregulation. However, a discrepancy between PLD1 expression and S6K1/Akt phosphorylation state was apparent under the effect of TNF-α alone, which downregulated PLD1 without affecting, or even slightly enhancing, either S6K1 or Akt phosphorylation. A possible explanation for this is that the pleiotropic cytokine TNF-α may trigger other signaling pathways that are able to positively influence Akt and mTORC1, and thereby mask any negative effects of ceramide on S6K1 and Akt. By suppressing these negative effects, ceramide-synthesis inhibition would allow further activation of these mTOR effectors. Inhibition of TNF-α-induced ceramide accumulation could thus have positive trophic effects on muscle cells, at least partly through the upregulation of PLD1 and the resulting activation of S6K1 and Akt, which respectively enhance protein synthesis and reduce proteolysis. However, TNF-α by itself altered protein synthesis without having significant effects on S6K1 and Akt, thus we hypothesize that the cytokine triggered other undefined mTOR-independent pathways that negatively influenced proteosynthesis.
Another mechanism by which ceramide-synthesis inhibition might result in Akt stimulation and positive effects on myocyte size is related to the recognized ability of ceramide to hamper insulin/insulin-like growth factor (IGF) signaling in muscle tissue [6
]. Myriocin has thus been shown to decrease muscle ceramide levels and insulin resistance in mice placed on a high-fat diet [7
]. Because IGFs are involved in trophic effects on muscle tissue [40
], it is possible that in our study myriocin acted on myocyte size, both in vitro
and in vivo
, through the enhancement of signaling by endogenously produced IGF, and downstream Akt activation. However, we found no changes in IRS-1 tyrosine phosphorylation under TNF-α and myriocin treatments of L6 myotubes (not shown), which makes a role for IGF signaling in the effects of myriocin very unlikely. In addition, our in vitro
results suggest that myriocin does not modulate proteolysis by targeting the NFκB pathway, and thus that this pathway is not regulated by ceramide in muscle cells.
Finally, our study addressed the in vivo role of ceramide in a model of tumor-induced cachexia. The development of C26 adenocarcinoma induced a marked increase in ceramide levels in mouse muscle, together with severe atrophy. A low dose of myriocin (0.1 mg/kg) significantly limited muscle loss, reduced expression of some atrogenes, and partially restored myocyte size, confirming that ceramide accumulation participates in enhanced proteolysis and muscle atrophy. As for the small negative effect of myriocin alone on myocyte size, it might be attributable, similarly to the hypothesis for C2C12 cells, to a lowered supply in sphingolipid(s) involved in the maintenance of muscle-tissue homeostasis.