The IGF-PI3K-Akt signaling pathway regulates myogenin expression in normal myogenic cells but not in Rhabdomyosarcoma derived RD cells.

Transcription

1 The IGF-PI3K-Akt signaling pathway regulates myogenin expression in normal myogenic cells but not in Rhabdomyosarcoma derived RD cells. Qing Xu and Zhenguo Wu Department of Biochemistry, Hong Kong University of Science & Technology, Clear Water Bay, Kowloon, Hong Kong, P.R.China Running title: PI3K/Akt induce myogenin by targeting MyoD and MEF2 proteins Corresponding author: Zhenguo Wu, Ph.D. Phone: (852) Fax: (852)

2 Summary Insulin-like growth factors (IGFs) can stimulate skeletal muscle differentiation. One of the molecular mechanisms underlying IGF-stimulated myogenesis is transcriptional induction of myogenin. The current work is aimed to elucidate the signaling pathways mediating the IGF effect on myogenin promoter in mouse C2C12 myogenic cells. We show that PI3K/Akt and p70s6k are crucial signaling molecules mediating the stimulatory effect of IGFs on myogenin expression. We have identified three cis-elements, namely the E box, the MEF2 and the MEF3 sites, within the 133 bp mouse proximal myogenin promoter that are under the control of the IGF/PI3K/Akt pathway. Simultaneous mutation of all three elements completely abolishes activation of the myogenin promoter by PI3K/Akt. We demonstrate that PI3K/Akt can increase both the MyoD and the MEF2- dependent reporter activity by enhancing the transcriptional activity of MyoD and MEF2. Interestingly, IGF1 does not enhance myogenin expression in Rhabdomyosarcoma derived RD cells. Consistently, the constitutively active PI3K/Akt fail to activate the myogenic reporters, suggesting the IGF/PI3K/Akt pathway is defective in RD cells and the defect(s) is downstream to PI3K/Akt. This is the first time that a defect in the IGF/PI3K/Akt pathway has been revealed in RD cells which provides another clue to future therapeutic treatment of Rhabdomyosarcoma.

3 Introduction Mammalian skeletal muscle differentiation has been a model system in the past decade for studying the molecular mechanisms that switch the cellular program from proliferation to differentiation. Intensive studies have led to the discovery and characterization of two families of transcription factors that play pivotal roles during differentiation (1-4). One of them consists of MyoD family proteins (also called myogenic regulatory factors or MRFs) which include four members: Myf5, MyoD, myogenin, and MRF4-all members of the basic helix-loop-helix (bhlh) superfamily and exclusively expressed in skeletal muscles. Knock-out and knock-in data reveal the existence of a genetic hierarchy in which MyoD and Myf5 act upstream to determine the myogenic fate of muscle precursor cells, while myogenin and MRF4 act downstream of Myf5 and MyoD to control differentiation process (3,4). The other group of transcription factors important for muscle differentiation consists of MEF2 proteins which also include four members: MEF2A, 2B, 2C, and 2D (5). MRF and MEF2 members can physically interact with each other to synergistically activate many muscle-specific genes (2,6). Among four MRFs, myogenin is critically involved in executing the differentiation program. Although myoblasts from myogenin null mice are present, they do not differentiate in vivo leading to severe muscle deficiency and perinatal death of the homozygous mice (7,8). Expression of myogenin is considered one of the earliest molecular markers for cells committed to differentiation in vitro. The up-regulation of myogenin, in concomitant with induction of the cell cycle inhibitor p21cip1, indicates that cells have irreversibly withdrawn from the cell cycle and entered the differentiation program (1). Regulation of myogenin has been extensively studied by both transfection analyses in cell culture systems and transgenic studies (9-14). A 133 bp myogenin proximal promoter has been found to contain sufficient cis-elements to correctly target a lacz transgene to specific muscle-forming regions (10). Furthermore, an E box, a MEF2 site and a MEF3 site within this 133 bp proximal myogenin promoter have been identified as critical cis-elements regulating myogenin expression (9-15). MyoD, MEF2 and homeodomain-containing Six (see Discussion) proteins have been implicated

4 as the nuclear factors binding to these cis-elements. Despite the above progress, how exactly these nuclear factors are activated by various intracellular signaling pathways is less well understood at present. Nor is known about the relationships between these cis-elements/transacting factors and the IGF signaling pathway. Insulin-like growth factors (IGFs) have been shown to potently stimulate myogenesis in cultured myogenic cells and are required for normal skeletal muscle development during mouse embryogenesis (16,17). In rat L6 myogenic cells, IGFs display biphasic action profiles: initially they stimulate proliferation, then function as strong inducers of differentiation (17-19). One of the molecular mechanisms underlying the stimulatory myogenic effect of IGFs lies in their abilities to transcriptionally induce myogenin mrna (20). It remains obscure how exactly this is achieved. Several intracellular signaling pathways have been identified that are activated in response to IGF stimulation. One of them is mitogen-activated protein kinase (MAPK) mediated signaling pathway. It has been shown that the ERK subgroup of MAPKs can be activated by IGF treatment via the classical receptor tyrosine kinase (RTK)/Grb2-Sos/Ras/Raf mediated pathway (21). Activation of ERK may be partially responsible for the initial mitogenic effect of IGF (18,22). Another IGF activated intracellular signaling pathway that has attracted much attention is mediated by phosphatidylinositol-3 kinase (PI3K). In response to IGF stimulation, activated PI3K converts phosphatidylinositol 4, 5-bisphosohate to phosphatidylinositol 3, 4, 5-trisphosphate which results in subsequent activation of the pleckstrin homology domain (PH)-containing serine/threonine kinases PDK1and Akt/PKB (23-25). Activated PI3K also leads to activation of p70s6k, which is mediated mainly by mtor, PDK1 and atypical PKCs (26). It has been demonstrated that PI3K mediates the stimulatory effect of IGFs on muscle differentiation (22,27,28). Specific interference of endogenous PI3K activity abolishes myogenic differentiation. Akt is shown to mediate the PI3K effect during muscle differentiation (29). Deliberate activation of either PI3K or Akt greatly enhances muscle differentiation (28,29). However, the downstream targets that couple PI3K/Akt stimulatory signals to myogenic signaling pathways remain to be identified.

5 The present study sought to identify the signaling pathways mediating the effect of IGF on myogenin expression in C2C12 cells, and to define the IGF-responsive cis-elements on myogenin promoter and the nuclear signal receivers/transcription factors that bind these cis-elements. We showed PI3K, Akt and p70s6k are critical signaling molecules mediating the IGF effect. Using IGFmediated myogenin induction as a model, we demonstrated that the stimulatory effect of PI3K/Akt on myogenin promoter activation was mediated by the unique E box, the MEF2 and the MEF3 sites in the 133 bp proximal myogenin promoter. MyoD and MEF2 family proteins are implicated as downstream targets of PI3K/Akt. p70s6k was also shown to mediate part of the PI3K signal along with Akt. We also carried out similar experiments in RD cells, a cell line derived from Rhabdomyosarcoma (RMS)-a childhood malignant tumor expressing myogenic regulatory factors yet failing to undergo myogenic differentiation (30,31). We found that IGF1 failed to induce myogenin expression in RD cells. While IGF1 could activate PI3K leading to Akt phosphorylation and activation in RD, the constitutively active PI3K/Akt failed to activate the myogenic reporter genes suggesting the defect in the IGF/PI3K/Akt pathway in RD cells lies downstream to PI3K/Akt.

6 Experimental Procedures Plasmids, cell lines, and other reagents. The GBBS-Luc was generated by inserting a Sac I-Bgl II fragment from the GBBS-CAT into a Hind III-Bgl II digested pxp2 luciferase vector. G133-Luc, G133E-Luc, and G133MEF2-Luc were generated by inserting the Xba I Bgl II fragments from the corresponding CAT constructs into the Hind III Bgl II digested pxp2, respectively. Various binding site mutants in G133-Luc were generated by a PCR-mediated mutagenesis method (32). The mutations at the E box and the MEF2 site are the same as described in (10). The primers used to make mutations at the MEF3 site are: forward: 5 TAGAGGGGGGCTGAGCTCTCTGTGGCGTTG 3, reverse: 5 CAACGCCACAGAGAGCTCAGCCCCCCTCTA 3. C2C12 cells and L6 cells were purchased from ATCC and grown in Dulbecco s modified Eagle s medium (DMEM) containing 20% fetal bovine serum (growth medium, or GM) in a 37 C incubator with 5% CO 2. To induce differentiation, growth medium is substituted by differentiation medium (DMEM containing 2% horse serum, or DM) when cells are near confluent. 10T1/2 fibroblasts, HeLa and RD cells were grown in DMEM containing 10% FBS. IGF1 was purchased from R&D systems. The constitutively active Akt contains an in-frame myristoylation sequence. The dominant negative Akt contains a K179 to M mutation. The constitutively active PI3K is myc-tagged with a CAAX motif (p110α-caax). The constitutively active MKK6 contains T to E mutations at the two conserved Thr residues in the kinase subdomain VIII. 4RE-luc contains four copies of the E box fused to the luciferase gene. 3xMEF2-Luc contains three copies of the MEF2 site fused to the luciferase gene. Transfection and reporter assays. Lipofectamine Plus reagent (Gibco BRL) was used for all transfection experiments. Unless stated otherwise, cells were first grown in GM for 36 hours after transfection, then shifted to DM for another 24 hours prior to cell harvest. Luciferase units were determined in a Monolight 2010 luminometer (Analytical Luminescence laboratory) using luciferase reporter gene assay kits from Boehringer Mannheim and normalized against total protein amount present in each sample extract. Protein concentrations were determined using protein assay solution from Bio-Rad.

7 Northern Blot analysis. Total RNA was extracted from C2C12 cells using Trizol reagent from Gibco BRL following manufacturer s suggestion. 20 g of total RNA was separated on a 1% formaldehydeagarose gel, transferred to a Hybond-N + membrane (Amersham) and cross-linked to the membrane in a UV cross-linker (Stratagene). The mouse myogenin probe and the GAPDH control probe were labeled using random labeling kits (Stratagene). Hybridization was carried out at 60 C in the Church- Gilbert hybridization solution overnight and washed three times with 2 x SSC/0.1% SDS buffer before being subjected to autoradiography. Western blot analysis. Cells were lysed in the lysis buffer (50mM Hepes, ph7.6, 1% Triton X-100, 150mM NaCl, 1mM EGTA, 1.5mM MgCl 2, 10% glycerol, 100mM NaF, 20mM p- nitrophenylphosphate, 20mM β-glycerolphosphate, 50 µm sodium vanadate, 2mM dithiothreitol, 0.5mM phenylmethylsulphonyl fluoride, 2 µg/ml aprotinin, 0.5 µg/ml leupeptin, 0.7 µg/ml pepstatin) and whole cell lysates (WCL) were prepared after removing insoluble debris. 30 µg of WCL was separated by SDS-PAGE, transferred to an Immobilon-P (Millipore) membrane and probed with various antibodies. The protein bands were visualized using enhanced chemiluminescence detection kit (ECL, Amersham). Treatment of cells with protein kinase inhibitors. SB202190, PD98059, LY294002, and rapamycin (Calbiochem) were added to cells at the time of medium change from GM to DM at a final concentration of 10 µm, 25 µm, 25 µm and 20 nm, respectively. After 24 hour-treatment, cells were harvested for either Western blot analysis or luciferase assays. Results Myogenin induction by IGF1 displays a faster kinetics in C2C12 cells than in L6 cells. It has been demonstrated that IGF elicits dual effects in rat L6 myogenic cells (17-19). In the first 24 hours after IGF treatment, a mitogenic response takes place in L6 cells followed (around 24 hours after IGF1 treatment) by a potent myogenic response indicated by a sharp increase in myogenin mrna and protein levels (18,19). Indeed, in our experiment, no myogenin could be detected in L6 cells at either 4-hour or even 9-hour time point after IGF1 treatment (Fig.1B). However, myogenin was detected 24

8 hours after IGF1 treatment. In contrast to L6 cells, a much faster myogenic response to IGF1 was seen in C2C12 cells. As shown in Fig.1A, 50ng/ml IGF1 potently induced myogenin protein level as early as 4 hours after IGF1 treatment. 24 hours after IGF1 treatment, the myogenin expression reached the peak level. In agreement with previous finding (20), we showed IGF-mediated myogenin induction occurred at mrna level (Fig.1C). As expected, early induction of myogenin in C2C12 was accompanied by a faster appearance of multinucleated myotubes of increased size when the morphology of the cells was examined by microscopy (Fig.1D). The above results indicate that IGF1 can transcriptionally induce myogenin expression with a much faster kinetics in C2C12 cells than in L6 cells. PI3K/Akt mediates the stimulatory effect of IGF on myogenin induction. It has been well established that PI3K/Akt can serve as downstream mediators of IGF action in various cell types. Both PI3K and Akt have been implicated in muscle differentiation (22,27-29). Since IGF1 could transcriptionally up-regulate myogenin (Fig.1C) (20), we tested whether PI3K/Akt could mediate IGF s stimulatory effect on the myogenin promoter. A 1 kb mouse proximal myogenin promoter fused to a promoterless luciferase gene was constructed and used as a reporter (GBBS-Luc). This 1 kb myogenin promoter was shown to contain sufficient regulatory elements that are responsible for somite-restricted expression of a lacz transgene (9,10). When GBBS-Luc was cotransfected into C2C12 cells along with either an empty expression vector or various PI3K/Akt constructs, significant activation of the myogenin promoter was detected only with the constitutively active PI3K or Akt, but not with the empty vector or the dominant negative mutants of PI3K and Akt (Fig.2A). Furthermore, a promoterless luciferase vector pxp2 (into which the 1 kb myogenin promoter was inserted to make the GBBS-Luc) could not be activated by either the constitutively active PI3K or Akt, indicating it is the sequences in the myogenin promoter that specifically respond to PI3K/Akt signaling. Activation of the myogenin promoter reporter by either PI3K or Akt requires a functional kinase domain, as the kinase-dead PI3K/Akt mutants (PI3K-dn/Akt-dn) failed to activate the reporter (Fig.2A). To further delimit the region(s) in the myogenin promoter responsive to PI3K/Akt signaling, a 133 bp proximal myogenin promoter fragment fused to the luciferase gene was also constructed and used as a reporter

9 (G133-Luc). It was previously shown that this 133 bp myogenin promoter fragment could also specifically target a lacz transgene to somites (10). As shown in Fig.2A, the constitutively active PI3K/Akt activated G133-Luc to the similar extent as they did to GBBS-Luc, suggesting most if not all PI3K/Akt-responsive elements reside within the 133 bp proximal myogenin promoter. To further prove both PI3K and Akt were involved in regulating endogenous myogenin expression, we transiently transfected C2C12 cells with either the dominant negative or the constitutively active PI3K/Akt and analyzed the endogenous myogenin expression by Western blotting. We reasoned that, if PI3K/Akt indeed controlled myogenin expression in vivo and the effect of PI3K/Akt was strong enough, we might detect the difference in myogenin expression simply by assaying the level of the endogenous myogenin in the total cell population including both the transfected and untransfected cells. Indeed, compared with the empty vector transfected control cells, both the dominant negative PI3K (PI3K-dn) and Akt (Akt-dn) transfected cells had reduced level of myogenin expression in response to IGF1 stimulation (Fig.2B). In contrast, both the constitutively active PI3K (PI3K-ca) and Akt (Akt-ca) could recapitulate the stimulatory effect of IGF1 and resulted in significantly enhanced myogenin expression compared to the control cells (Fig.2B). Considering an average transfection efficiency of 30-40% in C2C12 cells, the effect of these transfected PI3K/Akt mutants on endogenous myogenin expression was quite significant. Localization of PI3K/Akt-responsive elements in the myogenin promoter. Previous studies on transcriptional regulation of myogenin revealed the single MEF2 site and the E box within the 133 bp proximal myogenin promoter are critical for induction of myogenin and correct targeting of a lacz transgene to somites (9-14). We set out to test whether one or both of these sites are also required to mediate the stimulatory effect of PI3K/Akt. The single E box and the MEF2 binding site in G133-Luc were mutated either individually or at the same time. These mutations have been shown to disrupt MyoD and MEF2 binding and correct somite-targeting by the lacz transgene (10). The responsiveness of these mutant myogenin promoter reporters to PI3K/Akt was tested by reporter assays. As shown in Fig.3A, mutation of either the E box or the MEF2 site partially inhibited the

10 reporter activation by PI3K/Akt. Simultaneous disruption of both sites did not significantly reduce the stimulatory effect of PI3K/Akt further. Since a residual 2-3 fold activation by PI3K/Akt could still be detected in the reporter harboring double mutations [G133(2+E)], we reasoned that another cis-element within the 133 bp proximal myogenin promoter might also be under the control of the PI3K/Akt signaling pathway. It was recently demonstrated that a conserved MEF3 site (Fig.3B), present in the 133 bp proximal myogenin promoters of chick, mouse and human origin, also plays an important role in regulating myogenin expression (15). Homeodomain-containing Six proteins were shown to bind the MEF3 site. To test whether MEF3 site was responsible for the residual activation of G133(2+E)-Luc by PI3K/Akt, we mutated the MEF3 site and generated a myogenin promoter reporter with triple mutations (G133TM-Luc). As shown in Fig.3A, this triple mutant myogenin promoter completely lost its ability to be activated by either the constitutively activated PI3K or Akt. We also mutated the MEF3 site (G133MEF3) either alone or in conjunction with the E box [G133(E+3)] or the MEF2 site [G133(2+3)]. We found that mutation of the MEF3 site alone had minimal effect on myogenin promoter activation in response to PI3K/Akt signaling (Fig.3A). In contrast, mutation of the MEF3 site together with either the E box or the MEF2 site severely compromised the ability of the myogenin promoter to be activated by PI3K/Akt. The above results indicate that the E box, the MEF2 and the MEF3 sites in the 133 bp proximal myogenin promoter are critical cis-elements under the control of the IGF/PI3K/Akt signaling pathway. p70s6k functions downstream of PI3K to regulate myogenin expression. It has been shown that p70s6k functions downstream of PI3K (26). p70s6k also plays an important role during muscle differentiation (22,27,33). Since the activation fold of G133-Luc by Akt was always less than that by PI3K (Fig.2A and 3A), this suggested effectors other than Akt could also mediate the effects of PI3K. p70s6k may be such an effector. We then tested whether p70s6k could fulfill this role to regulate myogenin expression. As p70s6k activation could be specifically blocked by the immunosuppressant rapamycin (26), we first checked whether rapamycin could block the stimulatory effect of IGF1 on myogenin induction. Kinase inhibitors PD98059, LY294002, and SB202190, which specifically

11 inhibit MEK1/2, PI3K, and p38 MAPK, respectively, were used as controls. Indeed, p70s6k was found to play a role in myogenin regulation, since rapamycin partially prevent myogenin induction by IGF1 (Fig.4A). As a positive control, LY and SB almost completely blocked myogenin induction by IGF1, since both PI3K and p38 MAPK were absolutely required for myogenin induction (33-35). In contrast, PD98059 further enhanced the stimulatory effect of IGF1, as inhibition of ERK activity at the early stage of differentiation was known to facilitate myogenic differentiation (22,33,36). We then carried out a reporter assay to see whether rapamycin could block the stimulatory effect of the constitutively active PI3K/Akt on a myogenin reporter (G133-Luc). PD98059 and SB were again used as controls. While SB almost completely abolished and PD98059 further enhanced the reporter activation by PI3K/Akt, rapamycin partially repressed the promoter activation suggesting that p70s6k indeed plays a role in regulating myogenin expression, in agreement with the data from Western blot analysis (Fig.4B). PI3K/Akt can activate either the MyoD or the MEF2-dependent reporter genes by enhancing the transcriptional activity of MyoD and MEF2. The above assays clearly establish that the E box, the MEF2 and the MEF3 sites in the proximal myogenin promoter are under the control of the IGF/PI3K/Akt/p70S6K signaling pathway. This suggests that MyoD, MEF2 and Six proteins are critical nuclear targets downstream to PI3K/Akt/p70S6K in the IGF signaling pathway. Since the 133 bp myogenin promoter is still long enough to accommodate binding sites for other unidentified nuclear factors which may complicate the interpretation of the data derived from G133-Luc and its derivatives, we decided to directly study the effect of PI3K/Akt on either the MyoD-dependent (4RE- Luc) or the MEF2-dependent (3xMEF2-Luc) reporter genes. When these two reporter genes were introduced into C2C12 cells along with the constitutively active PI3K or Akt, we showed that PI3K or Akt could significantly increase the reporter gene activity (Fig.5A). We also did the similar experiments in 10T1/2 fibroblasts. Without cotransfected MyoD, the constitutively active PI3K did not activate 4RE-Luc, suggesting the activation was mediated by exogenously transfected MyoD (Fig.5B). Interestingly, activation of 3xMEF2-Luc by PI3K could readily be detected even in the absence of exogenously transfected MEF2 gene (Fig.5C). This was because MEF2 proteins are

12 already present in 10T1/2 and other non-muscle cells (5,12,37). Co-transfection of the 10T1/2 cells with the constitutively active PI3K and a MEF2C expression vector further activated the MEF2 reporter activity. We then tested whether the transcriptional activity of MyoD or MEF2 could be activated by cotransfected PI3K in non-muscle cells. Full length MyoD and MEF2C were fused in frame to the gal4 DNA binding domain (aa 1-147) and the fusion genes were transfected into HeLa cells together with either an empty vector or PI3Kca. A significant increase in the transcriptional activity of both MyoD and MEF2C could be detected only in the presence of cotransfected PI3Kca (Fig.5D). These data confirmed that MyoD and MEF2 family proteins could serve as downstream nuclear targets of the IGF/PI3K/Akt signaling pathway and PI3K activated MyoD and MEF2C by enhancing their transcriptional activity. Akt does not directly phosphorylate MyoD or MEF2C. Since MyoD and MEF2C could be activated by Akt in the above reporter assays, we wanted to check whether MyoD and MEF2 could serve as direct substrates for Akt, as a consensus Akt phosphorylation site RXRXXS/T has been defined (23,24). Search of MyoD and other MRFs sequences found no such site present in MRFs. In contrast, a perfect match was found in all MEF2 members (Fig.6) within the highly conserved MADS domain. We carried out an in vitro kinase assay using IGF1 activated wild type HA-Akt or the constitutively activated Akt and either bacterially expressed or in vitro translated MEF2C as substrates. No phosphorylation could be detected either with MEF2C or recombinant MyoD (data not shown). Furthermore, a mutant MEF2C in which the conserved Thr20 in the Akt consensus phosphorylation site was mutated to Ala [MEF2C(T20A)], could not prevent the stimulatory effect of PI3K in reporter assays. These results suggested that MyoD or MEF2 are not direct targets of Akt. The IGF/PI3K/Akt signaling pathway is defective in RD cells. Since the IGF/PI3K/Akt pathway regulates myogenin expression in normal myogenic cells, we want to test whether this pathway also operates in Rhabdomyosarcoma (RMS) derived cell line RD. We first treated RD cells with or without IGF1 and examined myogenin expression by Western blotting. No up-regulation of myogenin was detected (Fig.7A), suggesting the signaling pathway originating from IGF1 to myogenin induction was defective in RD cells. To find out whether the signal relay from IGF1 to Akt was

13 normal, we checked the phosphorylation status of the endogenous Akt in RD cells upon IGF1 treatment using an antibody raised against phosphorylated serine 473, the residue known to be phosphorylated in activated Akt (24). As shown in Fig.7B, Akt was phosphorylated to similar extent upon IGF1 treatment in both C2C12 cells and RD cells, in agreement with the Akt kinase assay using Histon 2B as a substrate (data not shown). This suggested that the defect in RD lie downstream to PI3K/Akt. To further confirm this point, we carried out a reporter assay using either the MyoDresponsive (4RE-Luc), the MEF2-responsive (3xMEF2-Luc) or myogenin G133-Luc as reporters and the constitutively active PI3K, Akt or MKK6 as upstream activators. Compared to results in C2C12 cells (see Fig. 2A and 5A), activation of these myogenic promoters by PI3K was severely reduced and that by Akt was completely abolished. In contrast, activation of the reporters by the constitutively active MKK6 was not affected: similar extent of activation of the reporters by MKK6EE was detected in both RD and C2C12 cells (data not shown). This was in agreement with our previous results showing that deliberately activating the MKK6/p38 MAPK pathway could restore the activity of muscle-specific transcription factors and partially rescue the differentiation defect in RD cells (38). The above results suggested a defect (or defects) specifically occur downstream to PI3K/Akt in the IGF/PI3K/Akt pathway in RD cells. Discussion Faster myogenic effect of IGF1 in C2C12 than in L6 myogenic cells. Although clearly documented to exert a myogenic effect after the first 24 hours of mitogenic effect in rat L6 cells (17-19), IGF1 has a much faster myogenic effect (as early as 4 hours after IGF1 treatment) in mouse C2C12 cells. One of the explanations might lie in the fact that the two myogenic cell lines have very different expression profiles in terms of myogenic factors. MyoD is present in proliferating myoblasts and differentiated myotubes in C2C12 cells, while it is absent in L6 cells. C2C12 can also synthesize and secret IGF in an autocrine manner while L6 cells needs exogenous IGF for differentiation. Thus, the difference in protein expression profiles between the two cell types may partially account for the difference in kinetics of IGF1-mediated myogenin induction. A recent report may provide another

14 explanation. In C2C12 cells, it was shown that the activated Akt could directly phosphorylate c-raf and inhibit its activity (39). As activation of the Raf-MEK-ERK pathway at the early stage of differentiation is inhibitory (22,33,36), inhibition of raf and hence the ERK pathway by the IGF/PI3K/Akt pathway would be expected to reduce the mitogenic response and to increase the myogenic response in C2C12 cells, which was exactly what we observed in this study. It is possible that this cross-talk between the ERK and the Akt pathways is not operative in L6 cells, although further study is needed to formally test this hypothesis. IGF1 targets multiple myogenic factors to regulate myogenin expression. It was demonstrated previously that IGF could stimulate muscle differentiation by transcriptionally up-regulating myogenin mrna (20), yet the signaling components that relay the IGF signal and the nuclear targets that receive the IGF signal to regulate myogenin expression remained elusive. PI3K is known as a key intracellular mediator of the differentiation-promoting effect of IGFs, as interference with PI3K function by using either a specific pharmacological compound (e.g., LY294002) or a dominantnegative mutant of PI3K completely abrogates the myogenic effect of IGFs (22,27,28,40). One of the important downstream effectors of PI3K is protein serine/threonine kinase Akt/PKB (23,24). Akt has been implicated in a variety of biological processes, ranging from anti-apoptosis/cell survival, protein synthesis/translational regulation, to adipocyte differentiation (23). As for its role in myogenesis, it was recently shown that Akt could participate in regulating muscle differentiation downstream of PI3K (29). Based on the above results, it seems that IGF exerts its myogenic effect through an intracellular signaling pathway involving PI3K and Akt. Most likely, it is the same signaling pathway that functions in other cell types: IGF-IGFR I (IGF receptor I)-IRS (insulin receptor substrate)-pi3k-pdk1-akt. Depending on the specific downstream effectors/targets (e.g., Bad, 4E-BP1, etc.) and cellular contexts, the same pathway can elicit distinct biological responses. As the IGF-responsive cis-elements in the myogenin promoter and the downstream effectors/targets of Akt remained unknown during muscle differentiation, we decided to use myogenin induction by IGF as a model system to address these issues.

15 We provide evidence in this study showing that the constitutively active PI3K/Akt can recapitulate the effect of IGF1 in activating the myogenin promoter, confirming the utilization of the PI3K/Akt pathway by IGF1 to regulate myogenin expression during muscle differentiation. p70s6k is also required to mediate part of the PI3K effect, as specific inhibition of p70s6k activation by the immunosuppressant rapamycin partially blocks myogenin promoter activation by PI3K and myogenin protein induction by IGF1. Interestingly, rapamycin also partially inhibits myogenin promoter activation by Akt (Fig.4B). This can be explained by the current model in which p70s6k functions downstream of PI3K/PDK1 in a pathway parallel to Akt (26). If both signals emanating from Akt and p70s6k are required for optimal promoter activation, inhibition of p70s6k alone would be expected to partially inhibit the promoter activation by Akt. The exact targets for p70s6k in myogenin regulation remains to be defined. Three cis-elements, an E box, a MEF2 and a MEF3 sites, within a small stretch (133 bp) of the mouse myogenin proximal promoter are found to be responsible for IGF/PI3K/Akt mediated myogenin induction. Deletion of the E box and the MEF2 site either individually or together partially abolishes the promoter activation by PI3K/Akt, yet low level of activation of the reporter could still be detected (Fig.2A). Only when the MEF3 site is mutated together with the E box and the MEF2 site, is the activation of the myogenin promoter by PI3K/Akt completely abolished. This suggests that IGF/PI3K/Akt target multiple nuclear factors that bind these cis-elements to control myogenin expression. MEF2 proteins were shown to bind the MEF2 site in the myogenin promoter (11,12). Myogenic regulatory factors MyoD or Myf5 have been implicated in binding the E box to control myogenin expression (10,11). Interestingly, although MRFs are absolutely required, their binding to the E box is dispensable for myogenin promoter activation in cell culture system (11-13). As MRFs can physically interact with MEF2 proteins, the above result can be explained such that MRFs are recruited to the promoter by MEF2 proteins bound at the MEF2 site (6,11,41). Homeodomain-containing Six proteins were shown to bind the MEF3 site in the myogenin promoter to critically regulate myogenin expression (15). Six proteins are the mouse homologues of

16 Drosophila sine oculis (so) gene product. In Drosophila, So interacts with Eyes absent (Eya) which in turn interacts with Dachshund (Dac) to control eye development (42). Recently, it was demonstrated that the same regulatory network also operates during vertebrate skeletal muscle development (42,43). The physical interactions between mouse Six and Eya proteins could also be detected in non-muscle cells by transient transfection study (44). In the present study, we found that mutation of the MEF3 site alone only had marginal effect on myogenin promoter activation by PI3K/Akt in reporter assays. Mutation of the MEF3 site together with either the E box, the MEF2 site or both severely compromised the promoter activation by PI3K/Akt. This is reminiscent of the role of the E box/mrf in regulating myogenin expression (11,12,14). Just as in that case, binding of the Six/Eya/Dac complex to the MEF3 site may be required but not necessary as long as the Six complex can interact with the MEF2/MyoD/E complex either directly or indirectly. The following picture is thus emerging in which myogenin is transcriptionally regulated by two higher order nucleoprotein complexes: one consisting of MRFs/E proteins/mef2s bound at the MEF2 site and the E box, the other consisting of the Six/Eya/Dac complex bound at the MEF3 site, although the existence of such a complex at the MEF3 site remains to be experimentally confirmed. These two complexes may also be the final converging points for different intracellular signaling pathways controlling myogenin expression. Coordinated and cooperative action of both complexes may be required for proper myogenin induction during muscle differentiation. The molecular link between Akt and the downstream targets. It is known that Akt normally phosphorylates a serine/threonine in the RXRXXS/T motif (23,24). Search of primary protein sequences reveals no such motif in all four MRFs, Six1, Eya2, or Dac2. Interestingly, such a motif is present and conserved in all four MEF2 proteins (Fig.6). It is located inside the highly conserved MADS box involved in DNA binding, yet we failed to detect any phosphorylation on MEF2 by Akt in an immune-complex kinase assay (data not shown). At this point, we favor a model in which Akt does not directly phosphorylate the components in either the Six1/Eya2/Dac2 or the MEF2/MyoD/E12 complexes. How does Akt activate the myogenin promoter in a MyoD, MEF2, and Six- dependent manner? One possible scenario is that Akt may directly phosphorylate an intermediate coactivator

17 (e.g., p300/cbp) which in turn interacts with MyoD, MEF2, or Six1 complexes to facilitate transcriptional activation of the myogenin promoter. It has been shown that p300 can physically interact with MyoD and MEF2 and enhance their transcriptional activities (45,46). Alternatively, the effect of Akt on the myogenin promoter may be mediated by another unknown factor either at the transcriptional or post-translational level. Under this scenario, a factor will first be transcriptionally induced or post-translationally modified by the IGF/PI3K/Akt pathway, then it can cooperate with preexisting MRFs or MEF2s to activate the myogenin promoter. No solid data are available at present to either support or rule out these hypotheses. Further study is needed to address this issue. A defective IGF/PI3K/Akt pathway exists in RD cells. Like other RMS derived cell lines, RD cells contain several MRF transcripts (i.e., MyoD and myogenin) (47,48), yet they fail to differentiate upon mitogen removal. Previous work also showed that the expressed MRFs in RMS were normal in terms of DNA binding, however, they were transcriptionally inactive (48). Recently, we demonstrated that the p38 MAPK pathway was defective in RD and another RMS derived Rh30 cells (38). The defect was found to lie upstream to p38 MAPK, as re-introduction of a constitutively active MKK6, the upstream activator of p38 MAPK, into RD and Rh30 cells reactivated the transcriptional activity of MyoD and partially restored the defective differentiation program (38). In the current study, a defective IGF/PI3K/Akt pathway was revealed by Western blotting analysis and reporter assays (Fig.7). We tentatively mapped the defect somewhere downstream to PI3K/Akt, as the constitutively active form of both kinases failed to restore the activation of the MyoD-dependent, the MEF2-dependent, and the myogenin reporters in RD cells. We recently showed that p38 MAPK and IGF act on two separate and parallel pathways controlling muscle differentiation (33). Signals from both pathways are required as blocking of either pathway prevents differentiation. The fact that defects have been detected in both pathways in RD cells and that the rescue of the p38 MAPK pathway alone by the constitutively active MKK6 is sufficient to partially restore differentiation seems to be paradoxical. One possible explanation is that signals from both pathways converge on the same nuclear targets (e.g., MyoD and MEF2). Under normal physiological conditions, activation of one pathway alone in the absence of the other cannot

18 activate the targets to the threshold required for differentiation. The additive/synergistic effects on the targets from both pathways are required to initiate the differentiation. In the case of RD cells in which both pathways are defective, deliberate activation of one pathway by re-introducing a constitutively active signaling molecule into RD cells could conceivably make the signal more robust and durable than could normally be achieved. This might be sufficient to activate the targets to the desired threshold even in the absence of the second signal. RMS consists of two major subtypes based on cytogenetic and morphological criteria: embryonal RMS and alveolar RMS (30,31). While the IGF/PI3K/Akt pathway is defective in RD cells, an embryonal type of RMS, it would be interesting to test the functionality of the pathway in alveolar type of RMS (e.g., Rh30 cells). This will tell us whether the defective IGF/PI3K/Akt pathway is a common feature in RMS or only exists in certain subtype of RMS. Importantly, this newly identified defect in RD cells can conceivably help us in the future to design more accurate and effective molecular diagnostic and therapeutic tools. Since myogenin expression during muscle differentiation is influenced by a wide array of stimuli, we think myogenin promoter may serve as a differentiation sensor. Depending on the nature of the stimuli, myogenin is either induced or repressed, which eventually leads to an enhanced or aborted differentiation program. Acknowledgement We would like to thank Drs S.P. Yee, P.L. Puri, J. Han, C. Makris, G. Natoli, and M. Nemer, for providing various reagents. We also thank Ms. Carol Wong for technical assistance. The work was supported by a start-up fund from HKUST and a grant from Research Grant Council of HKSAR (HKUST6205/00M). References: 1. Lassar, A. B., Skapek, S. X., and Novitch, B. (1994) Curr Opin Cell Biol 6(6), Molkentin, J. D., and Olson, E. N. (1996) Curr Opin Genet Dev 6(4),

19 3. Yun, K., and Wold, B. (1996) Curr Opin Cell Biol 8(6), Arnold, H. H., and Winter, B. (1998) Curr Opin Genet Dev 8(5), Black, B. L., and Olson, E. N. (1998) Annu Rev Cell Dev Biol 14, Molkentin, J. D., Black, B. L., Martin, J. F., and Olson, E. N. (1995) Cell 83(7), Hasty, P., Bradley, A., Morris, J. H., Edmondson, D. G., Venuti, J. M., Olson, E. N., and Klein, W. H. (1993) Nature 364(6437), Nabeshima, Y., Hanaoka, K., Hayasaka, M., Esumi, E., Li, S., and Nonaka, I. (1993) Nature 364(6437), Cheng, T. C., Hanley, T. A., Mudd, J., Merlie, J. P., and Olson, E. N. (1992) J Cell Biol 119(6), Yee, S. P., and Rigby, P. W. (1993) Genes Dev 7(7A), Edmondson, D. G., Cheng, T. C., Cserjesi, P., Chakraborty, T., and Olson, E. N. (1992) Mol Cell Biol 12(9), Buchberger, A., Ragge, K., and Arnold, H. H. (1994) J Biol Chem 269(25), Malik, S., Huang, C. F., and Schmidt, J. (1995) Eur J Biochem 230(1), Johanson, M., Meents, H., Ragge, K., Buchberger, A., Arnold, H. H., and Sandmoller, A. (1999) Biochem Biophys Res Commun 265(1), Spitz, F., Demignon, J., Porteu, A., Kahn, A., Concordet, J. P., Daegelen, D., and Maire, P. (1998) Proc Natl Acad Sci U S A 95(24), Powell-Braxton, L., Hollingshead, P., Warburton, C., Dowd, M., Pitts-Meek, S., Dalton, D., Gillett, N., and Stewart, T. A. (1993) Genes Dev 7(12B), Florini, J. R., Ewton, D. Z., and Coolican, S. A. (1996) Endocr Rev 17(5), Rosenthal, S. M., and Cheng, Z. Q. (1995) Proc Natl Acad Sci U S A 92(22), Engert, J. C., Berglund, E. B., and Rosenthal, N. (1996) J Cell Biol 135(2), Florini, J. R., Ewton, D. Z., and Roof, S. L. (1991) Mol Endocrinol 5(5), Petley, T., Graff, K., Jiang, W., Yang, H., and Florini, J. (1999) Horm Metab Res 31(2-3), 70-6

20 22. Coolican, S. A., Samuel, D. S., Ewton, D. Z., McWade, F. J., and Florini, J. R. (1997) J Biol Chem 272(10), Marte, B. M., and Downward, J. (1997) Trends Biochem Sci 22(9), Alessi, D. R., and Cohen, P. (1998) Curr Opin Genet Dev 8(1), Leevers, S. J., Vanhaesebroeck, B., and Waterfield, M. D. (1999) Curr Opin Cell Biol 11(2), Dufner, A., and Thomas, G. (1999) Exp Cell Res 253(1), Kaliman, P., Vinals, F., Testar, X., Palacin, M., and Zorzano, A. (1996) J Biol Chem 271(32), Jiang, B. H., Zheng, J. Z., and Vogt, P. K. (1998) Proc Natl Acad Sci U S A 95(24), Jiang, B. H., Aoki, M., Zheng, J. Z., Li, J., and Vogt, P. K. (1999) Proc Natl Acad Sci U S A 96(5), Anderson, J., Gordon, A., Pritchard-Jones, K., and Shipley, J. (1999) Genes Chromosomes Cancer 26(4), Merlino, G., and Helman, L. J. (1999) Oncogene 18(38), Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K., and Pease, L. R. (1989) Gene 77(1), Wu, Z., Woodring, P. J., Bhakta, K. S., Tamura, K., Wen, F., Feramisco, J. R., Karin, M., Wang, J. Y., and Puri, P. L. (2000) Mol Cell Biol 20(11), Cuenda, A., and Cohen, P. (1999) J Biol Chem 274(7), Zetser, A., Gredinger, E., and Bengal, E. (1999) J Biol Chem 274(8), Bennett, A. M., and Tonks, N. K. (1997) Science 278(5341), Ornatsky, O. I., and McDermott, J. C. (1996) J Biol Chem 271(40), Puri, P. L., Wu, Z., Zhang, P., Wood, L. D., Bhakta, K. S., Han, J., Feramisco, J. R., Karin, M., and Wang, J. Y. (2000) Genes Dev 14(5), Rommel, C., Clarke, B. A., Zimmermann, S., Nunez, L., Rossman, R., Reid, K., Moelling, K., Yancopoulos, G. D., and Glass, D. J. (1999) Science 286(5445), Kaliman, P., Canicio, J., Shepherd, P. R., Beeton, C. A., Testar, X., Palacin, M., and Zorzano, A. (1998) Mol Endocrinol 12(1), 66-77

21 41. Black, B. L., Molkentin, J. D., and Olson, E. N. (1998) Mol Cell Biol 18(1), Relaix, F., and Buckingham, M. (1999) Genes Dev 13(24), Heanue, T. A., Reshef, R., Davis, R. J., Mardon, G., Oliver, G., Tomarev, S., Lassar, A. B., and Tabin, C. J. (1999) Genes Dev 13(24), Ohto, H., Kamada, S., Tago, K., Tominaga, S. I., Ozaki, H., Sato, S., and Kawakami, K. (1999) Mol Cell Biol 19(10), Puri, P. L., Avantaggiati, M. L., Balsano, C., Sang, N., Graessmann, A., Giordano, A., and Levrero, M. (1997) Embo J 16(2), Sartorelli, V., Huang, J., Hamamori, Y., and Kedes, L. (1997) Mol Cell Biol 17(2), Bouche, M., Senni, M. I., Grossi, A. M., Zappelli, F., Polimeni, M., Arnold, H. H., Cossu, G., and Molinaro, M. (1993) Exp Cell Res 208(1), Tapscott, S. J., Thayer, M. J., and Weintraub, H. (1993) Science 259(5100), Figure Legends Fig.1. IGF1 treatment leads to a faster myogenin induction in C2C12 cells than in L6 cells. When near confluent, C2C12 or L6 cells were shifted from GM to DM in the presence (50 ng/ml) or absence of IGF1. For A (C2C12) and B (L6), cells were harvested at the indicated time points and subjected to Western blot analysis using a monoclonal anti-myogenin antibody (F5D). The same blot was also re-probed with a monoclonal anti-βactin (Sigma) antibody to monitor the loading difference. C. Total RNA was extracted from C2C12 cells and subjected to Northern blot analysis using a myogenin cdna probe. The same blot was stripped and re-probed with a GAPDH (glyceraldehyde-3- phosphate dehydrogenase) probe to monitor the loading difference. D. C2C12 cells were fixed and subjected bright field microscopy analysis. GM: growth medium (20% FBS). SF: serum-free medium. Myog: myogenin. Fig.2. PI3K and Akt mediate IGF s stimulatory effect on myogenin expression. A. C2C12 cells were transfected with different luciferase reporters along with either an empty vector or various PI3K/Akt constructs (the ratio of the reporters to the effectors used was 1:1). Cells were first grown in

22 GM for 36 hours after transfection, then shifted to DM for another 24 hours before harvest. Luciferase activity was determined and normalized against total proteins present in the cell extracts. Fold activation is the ratio of the luciferase activity in cells transfected with various PI3K/Akt constructs to cells transfected with the empty vector. GBBS: the 1 kb mouse myogenin promoter fused to a luciferase gene. G133: the 133 bp proximal myogenin promoter fused to a luciferase gene. Pxp2: the promoterless luciferase construct. ca: constitutively active. dn: dominant negative. DM: differentiation medium (2% horse serum). Vec: empty vector. The data are a representative of 3 independent transfection experiments performed separately by both authors. Numerical fold activation was shown on the top of each bar. B. C2C12 cells were transfected with either an empty vector or PI3K/Akt mutant expression vectors. 24 hours after transfection, cells were shifted from GM to either DM or serum-free medium with or without IGF1 as indicated for another 24 hours before harvest. 30 µg of whole cell lysates was assayed for myogenin expression by immunoblotting. Fig.3. Localization of the IGF1/PI3K/Akt responsive cis-elements in the myogenin promoter. C2C12 cells were transfected with the myogenin reporter G133 and its derivatives along with either an empty vector or the constitutively active PI3K/AktA using a transfection scheme identical to that described in Fig.2A legend. A. G133E, G133MEF2: individual point mutations in the E box and the MEF2 site of G133, respectively. G133(E+2): double mutations in the E box and the MEF2 site of G133. G133TM: triple mutations at the E box, the MEF2 and the MEF3 sites of G133. G133MEF3: point mutations at the MEF3 site of G133. G133(E+3), G133(2+3): double mutations at the E box/mef3 sites and the MEF2/MEF3 sites of G133, respectively. The fold activation was determined the same way as described in Fig.2A. The data were presented as mean +/- standard deviations based on results from 4 independent transfection experiments. B. Sequence alignment of the proximal myogenin promoters from human, mouse and chick using the Clustal W program ( The conserved E box, MEF2 and MEF3 sites are underlined. Fig.4. p70s6k functions downstream of PI3K in regulating myogenin expression. A. Near confluent C2C12 cells were shifted from GM to serum-free medium (SF) with (50 ng/ml) or without IGF1. At the time of medium change, cells were either mock treated with DMSO or treated with

23 various protein kinase inhibitors for 24 hours before harvest. 30 µg of whole cell lysates was subjected to Western blot analysis for myogenin expression. The same blot was also re-probed with anti-βactin to monitor loading difference. B. C2C12 cells were co-transfected with the myogenin reporter G133 along with either an empty vector or the constitutively active PI3K/Akt. After growing for 36 hours in GM, cells were shifted to DM for another 24 hours in the presence of either DMSO (mock) or various kinase inhibitors. Cells were then harvested to determine the luciferase activity and the fold activation was determined the same way as described in Fig.2A. D: DMSO. Ra: rapamycin (20 nm). SB: SB (10 µm). PD: PD98059 (25 µm). LY: LY (25 µm). Fig.5. PI3K/Akt activates the MyoD- and the MEF2-dependent reporter activity and the transcriptional activity of MyoD and MEF2C. A. C2C12 cells were co-transfected with either 4RE- Luc or 3xMEF2-Luc along with either an empty vector or the constitutively active PI3K/Akt. The same 36-hour GM/24-hour DM culture scheme as in Fig.2A was used before cell harvest. B and C. 10T1/2 cells were co-transfected with either 4RE-Luc or 3xMEF2-Luc along with different effectors as indicated. D. HeLa cells were transfected with gal4-luc together with combinations of gal4dbd (aa 1-147), gal4mef2, gal4myod, an empty vector and PI3K-ca as indicated. Cells were grown in GM for 36 hours and shifted to DM for another 24 hours before harvest. Luciferase activity and fold activation were determined the same way as in Fig.2A. The data are a representative of 2-3 independent experiments. Fig.6. Sequence alignment of the first 30 amino acids in MEF2 family members. A hypothetical Akt consensus phosphorylation motif RXRXXT was shown in bold letter. Fig.7. RD cells are defective in the IGF/PI3K/Akt pathway. A. Near confluent RD cells were shifted from growth medium (10% FBS) to serum-free medium (SF) in either the absence or presence (50 ng/ml) of IGF1. Cells were harvested at the indicated time points and subjected to Western blot analysis for myogenin expression. The same blot was also re-probed with anti-βactin to control loading difference. B. Near confluent RD or C2C12 cells were shifted to SF in either the absence or presence (50 ng/ml) of IGF1 for 30 minutes. Whole cell lysates were subjected to Western blot analysis using a polyclonal antibody against Ser473 phosphorylated Akt (New England Biolabs). C.