Akt phosphorylates CUGBP1 at Ser28 and increases interactions of CUGBP1 with cyclin D1 mRNA
The transition of proliferating myoblasts to differentiation status involves alterations of several signal transduction pathways which might control activities of CUGBP1. Our previous data and data from other groups suggested that binding activity of CUGBP1 and formation of CUGBP1-protein complexes are regulated by phosphorylation of CUGBP1 [3
]. Particularly, we have found that cyclinD3/cdk4/6 complexes phosphorylate CUGBP1 at Ser302 in vitro and in vivo [14
]. To identify other potential phosphorylation sites within CUGBP1, we have applied the Motif Scan Graphic Program. This program predicted multiple potential phosphorylation sites within the polylinker region of CUGBP1. These sites include consensuses for GSK3 (T277, T281, S285, S292, S300), three sites for PKC delta (Thr163, S268, S383) and a potential site for Akt kinase in the position 28 (Ser28) (). In addition, our previous work identified a site for cyclinD3/cdk4/6 in the position S302 [14
]. To examine if GSK3, PKC delta and Akt might phosphorylate CUGBP1, the purified fusion MBP-CUGBP1 was incubated with these kinases in the kinase buffer containing 32
P-γATP. This kinase assay indicated that Akt1 strongly phosphorylates CUGBP1 in vitro (), while GSK3 and PKC delta did not show detectable phosphorylation of CUGBP1 in this in vitro
assay (data not shown). These results showed that CUGBP1 is a substrate for Akt kinase in vitro. Failure of GSK3 and PKC delta to phosphorylate CUGBP1 in vitro suggests that these kinases might require additional factors which are absent in the in vitro
kinase assay. The Motif Scan Graphic Program predicted a potential Akt site at S28 (). To examine if Ser28 is a specific site for Akt kinase, we have mutated S28 to alanine and examined if Akt is able to phosphorylate this mutant. As seen in , replacement of S28 with alanine eliminates Akt-dependent phosphorylation of CUGBP1. Thus, these data demonstrate that Akt phosphorylates CUGBP1 at Ser28 in vitro.
Akt phosphorylates CUGBP1 at Ser28
Given the fact that CUGBP1 is the substrate for Akt, we have examined if Akt pathway is altered in cells from DM1 patients. Cytoplasmic and nuclear proteins from cultured DM1 myoblasts were loaded on the SDS gel and probed with antibodies to active ph-473-Akt. As one can see in , the amounts of active Akt are higher in cytoplasm of DM1 cells compared to those in control cells. Since one of functions of Akt is to promote cell cycle progression, we have examined expression of cell cycle proteins in skeletal muscle of DM1 patients, which contain elevated levels of CUGBP1 and have increased rate of proliferation [4
]. Total protein extracts were isolated from skeletal muscle of control and DM1 patients and examined by Western blotting with antibodies to cell cycle proteins. These studies showed that PCNA, cdc2 and cyclin D1 are up-regulated in DM1 muscle; while expression of the inhibitor of proliferation p27 is reduced (). This pattern of expression suggests that the a ctivation of Akt in DM1 cells might cause the increased levels of the cell cycle proteins. We have next tested this hypothesis by measuring the Akt-CUGBP1 pathway in DM1 cells. Since cyclin D1 is a strong promoter of cell proliferation, we have focused our further studies on the binding of CUGBP1 to the cyclin D1 mRNA. To test if CUGBP1 binds to the cyclin D1 mRNA, we have analyzed the nucleotide sequence of cyclin D1 mRNA and found that the 5′-region of cyclin D1 mRNA contains several GC-islands to which CUGBP1 binds (, upper). Using RNA probe containing this sequence, we have examined interactions of purified CUGBP1 with the 5′ region of cyclin D1 mRNA. In these experiments, we used the purified full-length CUGBP1 and purified CUGBP1 truncated proteins containing RNA-binding domains (RBD) 1, 2 and 3. As shown in , the full-length CUGBP1 binds to the 5′ region of cyclin D1 RNA. The purified proteins containing only RBD1 and RBD2 do not bind to this sequence, while RBD3 binds to the cyclin D1 riboprobe. Coomassie staining showed approximately identical loading of the proteins (, bottom image). Thus, these studies revealed that CUGBP1 binds to the 5′ region of cyclin D1 mRNA and that this binding is mediated by RBD3.
We have next examined if phosphorylation of CUGPB1 by Akt might regulate its interactions with cyclin D1 mRNA. In addition to S28A mutant, we also generated a phosphomimetic mutant in which Ser28 was replaced with aspartate (S28D), mimicking the phosphorylation status of CUGBP1. Wild type CUGBP1 and mutant S28A and S28D CUGBP1 proteins were purified to homogeneity, as determined by commassie staining, and their interactions with cyclin D1 RNA were examined by UV-cross link (). UV-cross link showed that un-phosphorylated CUGBP1 binds to cyclin D1 RNA with low affinity; however, phosphorylation of CUGBP1 by Akt significantly increases interaction of CUGBP1 with cyclin D1 mRNA (). Examination of S28A and S28D mutants showed that the mutation of Ser28 to alanine reduces interaction of CUGBP1 with cyclin D1 mRNA. In contrast, the S28D mutant binds stronger to cyclin D1 mRNA and this binding is identical to that of wild type CUGBP1 phosphorylated by Akt. Note that time of exposure of the membranes shown in were different: 16 hrs exposure is shown in the and hrs exposure is shown in . This suggest that, although the region of RBD3 mediates interaction of full-length CUGBP1 with cyclin D1 RNA, the binding of un-phosphorylated RDB3 to the cyclin D1 riboprobe is weaker than the binding of phosphorylated full-length CUGBP1. Taken together, these data show that phosphorylation of CUGBP1 by Akt at Ser28 significantly increases its binding to the cyclin D1 mRNA.
Cyclin D3-cdk4/6 increases binding of CUGBP1 to C/EBPβ and p21 mRNAs in differentiated myotubes
Interactions of CUGBP1 with Akt and cyclin D3/cdk4/6 are dependent on differentiation stages of myogenesis
We have previously found that cyclin D3-cdk4/6 regulate activity of CUGBP1 in differentiated settings such as quiescent livers [14
]. Therefore, we have next examined whether Akt1 and cdk4 are involved in the regulation of CUGBP1 during transition of myoblasts to myotubes. For this goal, we have used mouse C2C12 myoblasts and myotubes. CUGBP1 was precipitated from cytoplasmic extracts of proliferating mouse myoblasts and differentiated myotubes and the presence of ph-Akt and cyclin D3/cdk4/6 in the CUGBP1-IPs was examined by Western blotting. demonstrates that, in C2C12 myoblasts, the phosphorylated (“active”) Akt is present in the CUGBP1-IP, while cyclin D3/cdk4/6 are undetectable. In contrast, ph-Akt is not detectable in CUGBP1-IP from myotubes, but the association of CUGBP1 with cdk4, cdk6 and cyclin D3 is dramatically increased. The right panel of the (input) shows levels of the proteins in cytoplasmic extracts used for the IP. Although levels of CUGBP1, cyclin D3 and ph-Akt are increased during differentiation, the interactions of CUGBP1 with ph-Akt are reduced; while interaction of CUGBP1 with cyclin D3 is increased. These data show that different kinases are associated with CUGBP1 in proliferating myoblasts and in differentiated myotubes and that the association of CUGBP1 with ph-Akt does not depend on the levels of the proteins, but rather depends on the stage of cell differentiation.
To determine if the cyclin D3-cdk4 activates translational functions of CUGBP1 in differentiated C2C12 cells, we have examined expression of C/EBPβ since previous studies showed that the phosphorylation of CUGBP1 by cdk4-cyclin D3 enhances the ability of CUGBP1 to increase translation of C/EBPβ isoforms FL, LAP and LIP [14
]. We found that these C/EBPβ isoforms are increased in nuclear extracts of C2C12 myotubes (). The elevation of C/EBPβ isoforms correlates with the increased association of CUGBP1 with eIF2 and with increased phosphorylation of CUGBP1 by cyclin D3/cdk4/6 (see below).
Phosphorylation of CUGBP1 by cyclinD3/cdk4 increases interactions of CUGBP1 with C/EBPβ and p21 mRNAs
Since the phosphorylation of CUGBP1 by Akt increases its RNA-binding affinity to cyclin D1 (), we asked if phosphorylation of CUGBP1 by cyclinD3/cdk4/6 at Ser302 might also control its RNA-binding activity. To address this question, we have used purified electrophoretically homogenous wild type CUGBP1 and several CUGBP1 mutants. In one of the mutants, Ser302 was replaced with glycine (S302G) resulting in the expression of CUGBP1 un-phosphorylatable at Ser302 [14
]. We have also generated another mutant, S302D, which mimics cdk4-mediated phosphorylation of CUGBP1. We have first examined phosphorylation of WT CUGBP1 and CUGBP1 mutants by cyclin D3-cdk6 in in vitro
kinase assay. The complex of cyclin D3/cdk6 has been used because the complex of cyclin D3 with cdk4 is not formed efficiently in vitro. shows that cyclin D3-cdk6 phosphorylates WT, S28A and S28D mutants; while phosphorylation of S302G and S302D mutants is very weak or not detectable. These data confirmed our previous findings that S302 is a key amino acid for the phosphorylation by cdk4/cdk6 [14
]. We have next examined whether phosphorylation of CUGBP1 at Ser302 regulates its RNA-binding affinity toward C/EBPβ mRNA. CUGBP1 and CUGBP1 mutants were phosphorylated by cold ATP with baculovirus expressed cyclin D3-cdk6 as shown in and examined for the ability to interact with C/EBPβ. As seen in , the phosphorylation of WT CUGBP1, S28A and S28D mutants by cyclin D3/cdk6 complex significantly increases CUGBP1 binding to C/EBPβ mRNA, while the binding of un-phosphorylated CUGBP1 to C/EBPβ mRNA is weaker. A different result was observed for S302G and S302D mutants which are no longer under cdk4/cdk6 control and the pre-incubation with cyclin D3-cdk6 did not change their interactions with the C/EBPβprobe. However, the mutation of S302 to aspartate has increased the interactions with C/EBPβ mRNA. We have next examined the interaction of WT CUGBP1 and CUGBP1 mutants with p21 mRNA since our previous data showed that CUGBP1 regulates translation of p21 during differentiation of C2C12 cells. UV-cross linking studies showed that S302D mutant (mimicking CUGBP1 phosphorylation at Ser302) binds to the p21 mRNA much stronger than WT CUGBP1 and other CUGBP1 mutants (). Taken together, these investigations revealed that phosphorylation of CUGBP1 at Ser302 and the mutation of Ser302 to aspartate significantly increase interactions of CUGBP1 with C/EBPβ and p21 mRNAs.
Cyclin D3-cdk4-mediated phosphorylation of CUGBP1 increases formation of the translational CUGBP1-eIF2 complexes during normal muscle differentiation
In differentiated settings, the translational activity of CUGBP1 is mediated by its interactions with translation initiation factor eIF2 and by forming a high MW complex CUGBP1-eIF2 which contains eIF2 alpha and beta, Grp78 and CRT [14
]. Therefore, we have examined if differentiated myotubes utilize the same pathway. In contrasts to cyclins D1 and D2 which promote cell proliferation, cyclin D3 is highly expressed in certain differentiating settings such as differentiating myotubes and hepatocytes [14
] and supports differentiation and growth arrest of these cells. Previous studies of the cyclin D3 in differentiated myotubes have been focused on the nuclear accumulation of cyclin D3 and on nuclear functions of this protein [25
]. Since our data suggested that cyclin D3 also has biological functions in cytoplasm, we have first determined expression of cyclin D3 in cytoplasm of proliferating and differentiated C2C12 cells and compared these levels with levels of nuclear cyclin D3. Examination of two sets of differentiated cells showed that the levels of cyclin D3 are increased during differentiation in both cytoplasm and nuclei (). Although the amounts of cyclin D3 in cytoplasm are 2-fold lower than in nuclei (as ratios to β-actin), cyclin D3 is abundant and easy detectable in cytoplasm. Western blotting with antibodies to cdk4 also determined cdk4 in both cellular compartments with slightly higher concentrations in nuclei. Most important, the elevation of cyclin D3 during differentiation of myotubes is observed in cytoplasm suggesting that cyclin D3 plays a role in the regulation of cytoplasmic functions during differentiation. One of the possible functions of cyclin D3-cdk4 in cytoplasm of differentiated myotubes is the regulation of biological activities of CUGBP1.
The phosphorylation of CUGBP1 by cyclin D3/cdk4 during differentiation of C2C12 cells correlates with increased amounts of the CUGBP1-eIF2 complex
To determine if cyclin D3/cdk4 is able to phosphorylate CUGBP1 in differentiated C2C12 cells, cdk4 was immunoprecipitated from proliferating (day 0) and differentiated (day 5) C2C12 cells and used in a kinase assay with MBP-CUGBP1 substrate. shows that cdk4 phosphorylates CUGBP1. Interestingly, we have also detected the phosphorylation of an additional protein which is co-precipitated with cdk4. Since the size of this protein is 50 kDa, we suggested that this protein might be the endogenous CUGBP1. To test this suggestion and to determine amounts of cyclin D3 in cdk4 IPs, we have performed Western blotting of cdk4 IPs with antibodies to CUGBP1 and cyclin D3. As one can see in , both CUGBP1 and cyclin D3 are abundant in cdk4-IPs from differentiated myotubes; while only small amounts of cyclin D3 are detected in cdk4-IPs from proliferating myoblasts. Taken together, these studies showed that cyclin D3-cdk4 phosphorylates CUGBP1 in differentiated C2C12 cells.
To examine phosphorylation status of CUGBP1 during myogenesis we used 2D gel electrophoresis–Western blotting approach. Cytoplasmic proteins from proliferating (day 0) and differentiated (day 5) C2C12 were separated by 2D technique, transferred on the filter and probed with antibodies to CUGBP1. As can be seen in , differentiated C2C12 cells contained three additional phosphorylated isoforms of CUGBP1. These isoforms represent different phosphorylated forms of CUGBP1 because treatment of the protein extracts with alkaline phosphatase resulted in the shift of the signals to the alkaline region of the gel (data not shown). Thus, CUGBP1 phosphorylation is increased during normal differentiation.
We have next examined if the interaction of CUGBP1 with eIF2 is increased during muscle differentiation. For this goal, we have used two approaches: EMSA with the C/EBPβ probe and Co-immunoprecipitation analysis. Cytoplasmic proteins from myoblasts and myotubes were incubated with C/EBPβRNA probe containing the CUGBP1-binding site. The CUGBP1-eIF2 complex contains 8 protein components and, under conditions of EMSA, migrates close to the top of the gel because of high molecular weight of whole complex [14
]. shows that the RNA binding activity of the high molecular weight complex interacting with the C/EBPβ probe is significantly increased in myotubes compared with the activity of this complex observed in myoblasts. The high molecular weight complex binding with C/EBPβ in C2C12 myotubes migrates in the position typical for CUGBP1-eIF2 complex. However, the analysis of this complex by the traditional supershift assay is quite difficult, since the position of this complex in the EMSA gel is very close to the top of the gel. Therefore, we have determined if the interaction of CUGBP1 with eIF2 is increased during normal differentiation using a second approach: IP-Western analysis. CUGBP1 was immunoprecipitated from C2C12 cells and the CUGBP1-IPs were probed with antibodies to the known components of the CUGBP1-eIF2 complex. Upper image of the shows amounts of CUGBP1 in the lysates used for the Co-IP. As one can see, eIF2 alpha and beta, CRT and Grp78 are associated with CUGBP1 in myotubes; while these proteins do not bind or bind very weakly (CRT) to CUGBP1 in proliferating myoblasts. Consistent with the results of 2D-Western, the additional isoform of CUGBP1 is detectable in myotubes by regular Western blotting. Taken together, these studies showed that CUGBP1 interacts with cdk4 and cyclin D3 in differentiated myotubes and that the interactions of CUGBP1 with eIF2 complex are also increased in normal differentiated myotubes.
Misregulation of Akt and cyclinD3/cdk pathways in DM1 myogenesis
Given established pathways which regulate CUGBP1 activities during normal differentiation of mouse myotubes, we have examined these pathways in human normal myoblasts and in myoblasts from DM1 patients. In mouse myoblasts, CUGBP1 is phosphorylated by Akt kinase which increases interactions of CUGBP1 with cyclin D1 RNA. We have found that protein levels of PCNA, cdc2 and cyclin D1, which is a strong stimulator of proliferation, are increased in DM1 patients (). Examination of DM1 cells also revealed that Akt pathway is activated in cytoplasm of DM1 cells. These data suggested that DM1 cells might have misregulated Akt-CUGBP1 pathway. To test this hypothesis, we have applied primary myoblast cell lines derived from control and DM1 patients. Consistent with data shown in , Western blotting with ph-Akt revealed that Akt is activated in DM1 cell lines; while total levels of Akt are not changed (, upper image, input). We have next examined if Akt interacts with CUGBP1 in DM1 myoblasts. CUGBP1 was precipitated from control and DM1 myoblasts and ph-Akt was measured in CUGBP1-IPs. As shown in , the interaction of CUGBP1 with “active” Akt is significantly higher in DM1 myoblasts relatively to the interaction in control myoblasts. Thus, the elevation of ph-Akt in DM1 proliferating cells correlates with the increased interaction of active Akt with CUGBP1. These data are consistent with the hypothesis that the translation of cyclin D1 is increased in DM1 cells by Akt-mediated activation of interaction of CUGBP1 with cyclin D1 RNA. To test this suggestion, we have examined if the incubation of bacterially expressed CUGBP1 with proteins from DM1 cells might enhance interactions of CUGBP1 with cyclin D1 probe. For this goal, we incubated bacterially expressed, purified CUGBP1 with cytoplasmic extracts from control and DM1 cells at day 0 and 5 after initiation of differentiation and examined the RNA binding activity of the phosphorylated CUGBP1 towards cyclin D1 mRNA. shows that the incubation of CUGBP1 with proteins from DM1 myoblasts increases the interaction of CUGBP1 with cyclin D1 mRNA much stronger than incubation with proteins from myoblasts from normal patients. Protein extracts from differentiating cells have much lower capability to increase interactions of CUGBP1 with cyclin D1 mRNA. Nevertheless, densitometric calculations showed that the activation of CUGBP1 towards cyclin D1 by proteins from DM1 cells at day 5 is stronger than that observed in experiments with control cells at day 5 (, Bar graphs). This pattern of the activation of CUGBP1 towards cyclin D1 mRNA is consistent with the hypothesis that the increase of Akt-CUGBP1 pathway in DM1 cells supports high level of cyclin D1.
Mis-regulation of Akt-CUGBP1 and cyclinD3/cdk-CUGBP1 pathways in DM1 myogenesis
Since CUGBP1 displays its translational activities in differentiated cells via interactions with eIF2, we have examined if the formation of the CUGBP1-eIF2 complex is increased in control human myotubes and if the amounts of this complex are affected in DM1 patients. Myoblasts from normal and DM1 patients were subjected to differentiation, cytoplasmic proteins were isolated from myoblasts and myotubes and the formation of CUGBP1-eIF2 complex was examined by EMSA assay. Parallel plates were monitored to detect morphological changes typical for differentiation. In agreement with previous observations, DM1 cells show a delay of differentiation, while control cells are forming elongated myotubes (). Gelshift analysis indicated that the amounts of the high molecular weight CUGBP1-eIF2 complex are low in proliferating myoblasts from control patients and from patients with DM1; however, the amounts of this complex are significantly increased during differentiation of control myoblasts. In contrast, the CUGBP1-eIF2 complex is not increased in DM1 differentiating muscle cells (). Since cyclin D3-cdk4 phosphorylates CUGBP1 in vitro [14
] and because cyclin D3 and cdk4 are associated with CUGBP1 in differentiated mouse C2C12 cells (), we suggested that the lack of the CUGBP1-eIF2 complexes in DM1 differentiating cells might be associated with reduced levels of cyclin D3 and cdk4. To test this suggestion, we have examined the expression of cyclin D3, cdk4 and CUGBP1 in control and DM1 cells during differentiation. As can be seen in , cyclin D3 levels are increased in differentiated cells from control patients, but DM1 cells do not increase cyclin D3. Examination of cdk4 levels showed that there is also no increase of cdk4 in DM1 differentiating cells. The lack of cyclin D3/cdk4 elevation in DM1 differentiating myoblasts might be the cause for a failure to form the CUGBP1-eIF2 complex. Examination of ph-Akt showed that the amounts of active Akt are increased in normal cells during differentiation. In contrast, the activation of Akt was observed in DM1 cells before differentiation and it is not significantly changed during differentiation. Protein levels of CUGBP1 are elevated during differentiation of normal human myoblasts [4
]. Interestingly, DM1 cells already have elevated levels of CUGBP1 in myoblasts. In contrast to normal myotubes, cytoplasmic levels of CUGBP1 in DM1 differentiating cells are not increased but reduced compared to levels in proliferating DM1 cells ().
The lack of elevation of cyclin D3 in DM1 cells after initiation of differentiation suggests that the cyclin D3-CUGBP1 pathway is not activated in DM1 cells. To test this suggestion, CUGBP1 was immunoprecipitated from control and DM1 cells and cyclin D3 was examined in these IPs. As can be seen in , cyclin D3 is strongly associated with CUGBP1 in normal differentiated myotubes; however, no interaction of cyclin D3 with CUGBP1 is observed in DM1 cells. These data clearly show that cyclin D3-CUGBP1 pathway is inhibited in DM1 differentiating cells.
Ectopic expression of cyclin D3 corrects differentiation of DM1 myocytes
Based on our data described above, we hypothesized that the lack of cyclin D3 elevation might be a major cause of the impaired differentiation of DM1 cells. To test this hypothesis, we transfected pAdTrack-cyclin D3 plasmid (which, in addition to cyclin D3, also expresses GFP protein from an independent CMV promoter) into DM1 myoblasts and initiated differentiation of the cells by addition of fusion media. To determine if the transfected cyclin D3 activates CUGBP1-eIF2 complex, we have performed EMSA assay with cytoplasmic extracts from transfected cells. shows that transfection of cyclin D3 into DM1 cells increases amounts of the CUGBP1-eIF2 complex. Three approaches were used for the examination of differentiation of DM1 cells transfected with cyclin D3: immunostaining of desmin, examination of cell fusion and the quantification of myogenin by Western blot assay. First, the expression of desmin was determined using immunostaining of control and DM1 cells after two days of differentiation. shows that the levels of desmin are dramatically increased in normal cells in 2 days after initiation of differentiation. However, DM1 cells showed no or very weak increase of desmin at day 2 of differentiation. The lack of increase of desmin is consistent with a delay of differentiation in DM1 patients [4
]. DM1 cells were transfected with a pAdTrack-cyclin D3 plasmid and the transfected cells were analyzed by IF with antibodies to desmin. These studies showed that expression of desmin is dramatically increased in cells transfected with cyclin D3 at day 3 after initiation of differentiation, while un-transfected cells have very weak signal for desmin (). These un-transfected cells serve as a good control since these cells have been treated the same way as transfected cells, but they do not increase expression of desmin. The elevation of desmin was observed in each examined cyclinD3-positive cell. Thus, these data show that normalization of cyclin D3 levels during differentiation of DM1 cells rescues early steps of the differentiation.
Increase of the levels of cyclin D3 in DM1 myoblasts corrects early steps of differentiation
To further examine if ectopic expression of cyclin D3 corrects differentiation of DM1 cells, we have applied a second approach and have monitored the fusion of cells at day 3 after initiation of differentiation. shows that the fusion of DM1 cells is dramatically increased after transfection with cyclin D3; while GFP alone did not affect the fusion of DM1 cells. Calculation of the efficiency of fusion by counting multi-nucleated cells showed that only 11% of DM1 cells transfected with empty vector contain more than 6 nuclei per cell upon maintenance in fusion medium for 3 days. However, up to 61% of DM1 cells transfected with cyclin D3 contain high number of nuclei per cell (6–25). In these experiments, we found a significant portion of DM1-cyclin D3 transfected cells which had more than 20 nuclei in the fused cells (, right, field #3). This dramatic increase of the fusion of DM1 cells by the ectopic cyclin D3 is accompanied by the elevation of myogenin. As shown in , myogenin is almost undetectable in DM1 cells transfected with an empty vector; however, in the presence of cyclin D3 the levels of myogenin are significantly increased upon differentiation. In summary, these studies showed that ectopic expression of cyclin D3 in DM1 cells leads to the activation of CUGBP1-eIF2 pathway () and to correction of differentiation of DM1 cells (; ).
Figure 6 A. Ectopic expression of cyclin D3 dramatically increases fusion of DM1 cells during differentiation. Empty vector and pAdTrack-cyclin D3 plasmid were transfected into DM1 cells using Amaxa Transfector protocol. Fusion media was added on the next day (more ...)