HuR promotes myogenesis by stabilizing the MyoD, myogenin and p21 mRNAs during the fusion of muscle cells to form myotubes. Here we show that HuR, via a novel mRNA destabilizing activity, promotes the early steps of myogenesis by reducing the expression of the cell cycle promoter nucleophosmin (NPM). Depletion of HuR stabilizes the NPM mRNA, increases NPM protein levels and inhibits myogenesis, while its overexpression elicits the opposite effects. NPM mRNA destabilization involves the association of HuR with the decay factor KSRP as well as the ribonuclease PARN and the exosome. The C terminus of HuR mediates the formation of the HuR–KSRP complex and is sufficient for maintaining a low level of the NPM mRNA as well as promoting the commitment of muscle cells to myogenesis. We therefore propose a model whereby the downregulation of the NPM mRNA, mediated by HuR, KSRP and its associated ribonucleases, is required for proper myogenesis.
Muscle differentiation, also known as myogenesis, represents a vital process that is activated during embryogenesis and in response to injury to promote the formation of muscle fibres1,2. Myogenesis requires the activation of muscle-specific promyogenic factors that are expressed at specific steps of the myogenic process and act in a sequential manner. We and others have demonstrated that the expression of genes encoding some of these promyogenic factors such as MyoD, myogenin and the cyclin-dependent kinase p21 are not only regulated at the transcriptional level but are also modulated posttranscriptionally3,4,5,6,7,8. Indeed, modulating the half-lives of these mRNAs plays an important role in their expression. The RNA-binding protein HuR, via its ability to bind specific AU-rich elements (AREs) in the 3′-untranslated regions (3′-UTRs) of these mRNAs, protects them from the AU-rich-mediated decay machinery4,5,7,8,9. This HuR-mediated stabilization represents a key regulatory step that is required for the expression of these promyogenic factors and proper myogenesis.
HuR, a member of the ELAV family of RNA-binding proteins, specifically binds to AREs located in the 3′-UTRs of its target transcripts8,9,10,11,12 leading to their stability that in turn enhances the expression of the encoded proteins5,7. In addition to mRNA stability, HuR modulates the nucleocytoplasmic movement and the translation of target transcripts13,14,15,16. Our previous data have indicated that HuR associates with MyoD, myogenin and p21 transcripts only during the fusion step of myoblasts to form myotubes5. This finding led to the conclusion that HuR promotes myogenesis by stabilizing these mRNAs specifically at this step. In the same study, however, we showed that depleting HuR from proliferating myoblasts prevented their initial commitment to the differentiation process. These observations indicated that HuR promotes muscle fibre formation by also regulating the expression of target mRNAs during the early steps of myogenesis.
Recently, we discovered that HuR promotes myogenesis through a novel regulatory mechanism involving its caspase-mediated cleavage4. As muscle cells are engaged in the myogenic process, a progressive accumulation of HuR in the cytosol is triggered. In the cytoplasm, HuR is cleaved by caspase 3 at its 226th residue, an Asp (D), generating two cleavage products (HuR-CPs: -CP1, 24 kDa and -CP2, 8 kDa). These HuR-CPs, generated from ~50% of cytoplasmic HuR, are required for muscle fibre formation4,8. Indeed, while wt-HuR can rescue myogenesis in cells depleted of endogenous HuR, the non-cleavable HuRD226A mutant failed to do so4,8. In addition, HuR-CP1, by associating with import factor transportin 2, prevents HuR nuclear import promoting its cytoplasmic accumulation. While these data clearly establish that HuR-CP1 modulates the cellular movement of HuR during myogenesis, the role of HuR-CP2 remains unclear.
HuR is not the only RNA-binding protein involved in the post-transcriptional regulation of promyogenic factors. The KH-type splicing regulatory protein (KSRP) is known to associate, in proliferating myoblasts, with the AREs of the myogenin and p21 mRNAs leading to their rapid decay3. By doing so, KSRP participates in ensuring the proliferation of myoblasts and prevents their premature commitment to the myogenic process. KSRP promotes mRNA decay in muscle cells by recruiting ribonucleases such as PARN and members of the exosome complex (for example, EXOSC5) to ARE-containing mRNAs such as myogenin and p21 (refs 3, 17). It was also suggested that when myoblasts become competent for differentiation, KSRP releases myogenin and p21 mRNAs leading to their stabilization. As a consequence, myoblasts enter myogenesis and fuse to form myotubes3. Since at this same step HuR associates with and stabilizes these ARE-bearing mRNAs5,7, we concluded that the induction of myogenesis involves both KSRP and HuR that modulate the expression of the same mRNAs in an opposite way but at different myogenic steps.
Surprisingly, however, here we report that in undifferentiated muscle cells, HuR and KSRP do not compete but rather collaborate to downregulate the expression of a common target, the Nucleophosmin (NPM, also known as B23) mRNA. HuR forms a complex with KSRP that is recruited to a U-rich element in the 3′-UTR of NPM mRNA. The HuR/KSRP complex, in collaboration with PARN and the exosome, then destabilizes the NPM mRNA leading to a significant reduction in NPM protein levels. Our data also provide evidence supporting the idea that the HuR/KSRP-mediated decrease of NPM expression represents one of the main events that helps myoblasts commit to the myogenic process.
NPM is a HuR-mRNA target in undifferentiated muscle cells
We first identified the mRNAs that depend on HuR for their expression in undifferentiated, C2C12 cells. Endogenous HuR was depleted (siHuR) or not (siCtr) from these cells and total RNA was then prepared and hybridized to mouse arrays containing 17,000 probe sets of known and unknown expressed sequence tags. Consistent with the fact that HuR acts as a stabilizer for many of its mRNA targets10,12, we observed a significant decrease in the steady-state levels of 18 mRNAs in siHuR-treated cells. Surprisingly, however, we found that 12 mRNAs are significantly upregulated in these cells (Supplementary Table 1). While many of these transcripts do not have typical AREs and are not known to modulate myogenesis, these observations suggested that in muscle cells HuR, in addition to stabilizing some of its target mRNAs, could also be involved in the destabilization of other transcripts. Therefore, we investigated further this unexpected possibility and assessed whether this HuR destabilizing activity is part of its promyogenic function. To this end, we chose to study, among these upregulated HuR-mRNA targets, the NPM mRNA. NPM was identified as a target of HuR in other cell systems18 and its downregulation has been shown to be required for the differentiation of other cell models19,20. Hence, moderating NPM expression in muscle cells could be one of the early events through which HuR promotes myogenesis.
First, we validated the results of our microarray data using northern blot analysis. The knockdown of HuR in C2C12 cells increased the level of NPM mRNA by twofold while, as previously described5, it decreased the level of MyoD mRNA by >50% (Fig. 1a). Knocking down HuR in these cells also increased NPM protein levels, while myogenin protein levels were reduced (Fig. 1b). Conversely, overexpressing GFP-HuR protein in C2C12 cells significantly decreased the levels of both NPM mRNA and protein, compared with cells transfected with green fluorescent protein (GFP) only (Supplementary Fig. 1). Together, these observations establish that in undifferentiated muscle cells, HuR prevents the overexpression of NPM mRNA and protein.
HuR-mediated NPM mRNA destabilization promotes myogenesis
If downregulating NPM levels represent an important event during the early steps of myogenesis, decreasing its expression levels in C2C12 cells should promote their ability to enter this process. To investigate this possibility, we assessed NPM mRNA and protein levels during muscle cell differentiation. We observed that, although, as expected21, MyoD mRNA levels increased during this process, the abundance of the NPM mRNA and protein significantly decreased as soon as myogenesis was initiated (Fig. 2a–c). To determine whether decreasing NPM expression is required for the initiation of myogenesis, we tested the impact that NPM knockdown (Fig. 2d) could have on the ability of C2C12 cells, expressing or not HuR, to enter this process. We observed that the depletion of endogenous NPM by small interfering RNA (siRNA) not only triggered the formation of more and larger muscle fibres when compared with siCtr-treated cells, but also re-established myogenesis in HuR-knockdown C2C12 cells (Fig. 2d–h and Supplementary Fig. 2). These observations, together with the fact that overexpressing GFP-NPM prevented myogenesis (Supplementary Fig. 3), clearly demonstrate that one way by which HuR promotes the early steps of muscle fibre formation is by downregulating NPM expression.
We have previously shown that HuR is cleaved during muscle cell differentiation into two cleavage products HuR-CP1 (24 kDa) and HuR-CP2 (8 kDa). Our data also indicated that HuR cleavage is a key regulatory event required for proper muscle fibre formation4. Since our data (Fig. 2c) have shown that the cleavage of HuR during myogenesis correlates with a decrease in NPM expression, we decided to investigate the effect of HuR-CPs on NPM expression in muscle cells. As expected, we observed that in muscle cells depleted of endogenous HuR, the wild-type (wt-HuR) but not the non-cleavable isoform of HuR (HuRD226A) was able to re-establish myogenesis as well as the downregulation of the steady-state level of the NPM mRNA. Surprisingly, however, HuR-CP2 but not HuR-CP1 was also able to re-establish, in these cells, the downregulation of NPM mRNA levels and myogenesis similarly to wt-HuR (Fig. 2i–l). While these results suggest a correlation between HuR cleavage and the HuR-mediated downregulation of the NPM transcript, they clearly indicate that the C terminus of HuR could play an important role in this effect.
To identify the mechanisms responsible for the decreased expression of NPM mRNA during myogenesis (Fig. 2a), we first performed a nuclear run-on assay22 to assess whether this decrease is due to a change in transcription. Since we did not observe any change in the transcription rate during myogenesis (Supplementary Fig. 4), we concluded that the HuR-mediated downregulation of NPM mRNA expression is likely to occur at the level of mRNA stability. Indeed, a pulse-chase experiment23 using the RNA polymerase II inhibitor actinomycin D (ActD) showed that depleting HuR in C2C12 cells significantly increased, from ~5 to >9 h, the half-life of the NPM mRNA (Fig. 3a,b). Of note, treatment of cells with ActD did not affect their viability (Supplementary Fig. 5). These results, together with the fact that HuR does not affect the localization or the translation of the NPM mRNA (Fig. 3c,d), indicate that HuR promotes the formation of muscle fibres by destabilizing the NPM mRNA.
HuR destabilizes NPM mRNA via U-rich elements in the 3′-UTR
HuR is known to regulate its mRNA targets by interacting with AU-/U-rich elements in their 3′-UTRs9,24. Therefore, to define the molecular mechanism by which HuR destabilizes the NPM mRNA, we first assessed whether HuR and NPM mRNA associate in muscle cells. Immunoprecipitation (IP) (Fig. 4a) followed by reverse transcription PCR (RT-PCR; upper panel) or RT-quantitative PCR (RT-qPCR; lower panel) analyses23,25 showed that HuR and the NPM mRNA coexist in the same complex in muscle cells. To determine whether this association is direct or indirect, we performed RNA electromobility shift assays (REMSAs)23 using recombinant glutathione S-transferase (GST) or GST-HuR proteins and radiolabelled RNA probes corresponding to the entire 5′- or 3′-UTRs of NPM mRNA (Fig. 4b). We observed that GST-HuR forms a complex only with the 3′ but not the 5′-UTR of NPM mRNA (Fig. 4c). To further define the binding site(s) of HuR, we divided the NPM 3′-UTR into two probes P1 and P2 (Fig. 4b) and observed a strong association of GST-HuR with only the P1 probe (Fig. 4d). Using surface plasmon resonance analysis, we confirmed these associations and showed that GST-HuR is bound to the NPM-3′-UTR and P1 probe with higher affinity (dissociation constants (KD) of 86 × 10−9 and 132 × 10−9 M, respectively) than the P2 probe (KD=324 × 10−9 M) (Fig. 4e). In addition, further subdivision of P1 and 2 probes into three smaller regions, P1/2-1, -2 and -3, showed that HuR only binds to the P1-1 region of NPM-3′-UTR (Fig. 4f and Supplementary Fig. 6a,b). Analysis of the secondary structure of the P1-1 region using the mfold prediction software for mRNA folding26 indicated the existence of two U-rich elements (E1 and E2) similar to HuR cis-binding motifs previously identified in other mRNA targets11,23,26 (Supplementary Fig. 6c). To determine whether E1 and/or E2 elements mediate the binding of HuR to the NPM 3′-UTR, we mutated each one of their uracils (Us; shown as Ts) as well as those of the element separating E1 and E2 to cytosine (C) (Fig. 4g) and performed REMSA experiments as described above. Since mutating each site separately (P1-1-mut1 or -mut2), but not those of the hinge element (P1-1-mut3), equally disrupted the P1-1/HuR complex (Fig. 4h,i), we concluded that the E1 and E2 elements together constitute the minimum U-rich element required for the direct binding of HuR to the NPM-3′-UTR.
Next, we tested whether these U-rich elements are involved in the HuR-mediated destabilization of the NPM mRNA in muscle cells. We generated reporter complementary DNA (cDNA) plasmids expressing the Renilla luciferase (Rluc) in which we inserted, at the 3′ end, either the wild-type NPM 3′-UTR (Rluc-NPM-3′) or the NPM 3′-UTR mutant 1 (Rluc-NPM-3′-mut1, deficient in its ability to bind HuR) (Figs 4h and 5a). An IP/RT-qPCR experiment showed that the association between HuR and the Rluc-NPM-3′ mRNA is significantly higher than its association with the control reporters Rluc or Rluc-NPM-3′-mut1 (Fig. 5b,c). We next assessed the impact of depleting endogenous HuR on the expression of the Rluc-NPM-3′ mRNA. siCtr- or siHuR-treated C2C12 cells were transfected with the Rluc, Rluc-NPM-3′ or Rluc-NPM-3′-mut1 plasmids (Supplementary Fig. 7a) and the steady-state levels of these mRNAs were determined by RT-qPCR analysis. While we observed a 3.5-fold increase in the level of the Rluc-NPM-3′ mRNA in cells depleted of HuR, but not in those treated with siCtr, the levels of Rluc-NPM-3′-mut1 remained high in both cell types (Fig. 5d). On the other hand, overexpressing GFP-HuR in C2C12 cells decreased the level of the Rluc-NPM-3′ mRNA by >65% but had no significant effect on the levels of Rluc or Rluc-NPM-3′-mut1 mRNAs (Fig. 5e, Supplementary Fig. 7c). The effect of knocking down or overexpressing HuR on luciferase activity (which is proportional to Rluc protein levels) was similar to those seen for the mRNA levels (Supplementary Fig. 7b,d). Next, we investigated the functional relevance of these HuR-binding sites on NPM mRNA stability. We performed ActD pulse-chase experiments on C2C12 cells expressing the various Rluc reporters described above and treated or not with 50 nM of recombinant HuR conjugated to the cell-permeable peptide Antenapedia (AP)14 (Fig. 5f). We observed that the half-life of the Rluc-NPM-3′ mRNA was significantly reduced (<2 h) only in cells treated with AP-HuR-GST but not in those treated with AP-GST (>6 h). However, the Rluc-NPM-3′-mut1 mRNA remained stable under all treatments (half-life >6 h). While these results clearly demonstrate that an intact HuR-binding site is required for HuR-induced destabilization of the NPM mRNA, it does not provide any indication on their relevance during myogenesis. To address this question, we generated GFP-conjugated full-length wild-type (GFP-NPM-3′) or mutated (GFP-NPM-3′-mut1) isoforms of NPM (Fig. 5g). These NPM isoforms were then overexpressed in C2C12 cells and 48 h later these cells were induced for differentiation for 3 days. Our experiments showed that overexpressing GFP-NPM-3′ reduced the efficiency of muscle fibre formation by ~50% when compared with GFP alone (Fig. 5h–j). Interestingly, the overexpression of the GFP-NPM-3′-mut1 isoform, which does not bind HuR, provided a much higher inhibition of myogenesis efficiency than GFP-NPM-3′ (>70%). Together, these results show that the recruitment of HuR to the E1 and E2 U-rich elements of NPM 3′-UTR is required for NPM downregulation and the promotion of myogenesis.
HuR destabilizes the NPM mRNA in a KSRP-dependent manner
It has been shown that during myogenesis, transcripts such as myogenin and p21 but not MyoD are destabilized by KSRP in proliferating myoblasts3. However, at later stages (fusion step), these mRNAs are stabilized by HuR, leading to an increase in their expression levels and to the promotion of myotube formation5,7. Therefore, since KSRP and HuR target common transcripts in muscle cells3,5,7 and our observation showed that depleting the expression of either of these two proteins in C2C12 cells equally prevented myogenesis (Supplementary Fig. 8), we tested whether KSRP could be implicated in the HuR-mediated destabilization of the NPM mRNA in proliferating C2C12 cells. We first determined the effect KSRP depletion could have on the expression level of the NPM mRNA and protein. We showed that, similarly to HuR depletion, the knockdown of KSRP significantly increased the expression levels of NPM mRNA (>2-fold) and protein (>1.5-fold) (Fig. 6a,b). In addition, ActD pulse-chase experiments indicated that the knocking down KSRP increased the half-life of NPM mRNA to a similar extent than HuR (compare Fig. 3b with Fig. 6c). However, as expected3, KSRP knockdown had no effect on the stability of the MyoD mRNA, while it increased the half-life of p21 mRNA from ~3.6 to >6 h (Fig. 6c). Altogether, these data show that both HuR and KSRP decrease the expression level of the NPM protein by shortening the half-life of the NPM mRNA. Surprisingly, we additionally observed that overexpressing KSRP alone in C2C12 cells, unlike HuR, does not affect NPM mRNA or protein levels (Fig. 6d,e) despite promoting the decreased expression of p21. Since knocking down KSRP by >90% prevented the HuR-mediated decrease of the NPM mRNA and protein (Fig. 6f–h), we concluded that, although KSRP alone is not able to destabilize NPM mRNA, KSRP is required for the HuR-mediated downregulation of NPM level in muscle cells.
To gain insight into the molecular mechanisms by which HuR and KSRP affect the fate of NPM mRNA in muscle cells, we first determined whether KSRP also associates with this transcript. IP experiments using the anti-KSRP antibody17 followed by RT–PCR and -qPCR analyses showed that KSRP binds to the NPM mRNA similarly to HuR (Figs 4a and 7a). REMSA using total extracts prepared from C2C12 cells that were incubated with radiolabelled NPM-3′-UTR or the P1 or P2 probes (Fig. 4b) and with the anti-KSRP antibody showed that, similarly to HuR (Fig. 4), KSRP associates with the 3′-UTR of NPM through the P1 but not the P2 region (Fig. 7b). To determine the KSRP-binding site(s) in the NPM 3′-UTR, we incubated the recombinant KSRP with intact or mutated radiolabelled P1 probes and performed the same REMSA experiments described in Fig. 4. KSRP formed complexes with P1-1 and P1-1-mut2 but not with P1-1-mut1 (mut-E1) probes (Fig. 7c). In addition, IP experiments with anti-KSRP antibody were performed on cells expressing the Rluc reporters described in Fig. 5a. We observed that, while KSRP associated with the Rluc-NPM-3′ mRNA and Rluc-NPM-3′-mut2, it failed to interact with the Rluc-NPM-3′-mut1 message (Fig. 7d,e and Supplementary Fig. 9). Therefore, together these results show that KSRP directly associates with the NPM mRNA through the E1 element that is also one of the sites that mediates HuR binding.
Next, we investigated whether HuR and KSRP can form a complex. IP experiments on cell extracts treated or not with 100 μg ml−1 RNAse A indicated that HuR and KSRP associate in C2C12 cells in an RNA-independent manner (Fig. 7f,g). However, this RNA-independent interaction between HuR and KSRP was not seen using a lower dose of RNAse A (12.5 μg ml−1). Moreover, GST pull-down experiments, using recombinant His-KSRP and GST-HuR, further confirmed this direct interaction (Fig. 7h). We then assessed whether the association between the NPM mRNA and HuR or KSRP requires an intact HuR/KSRP complex. IP/RT-qPCR experiments showed that depleting HuR from C2C12 cells significantly decreased (>40%) the amount of NPM mRNA that binds to KSRP (Fig. 7i). Similarly, although knocking down KSRP in C2C12 cells did not affect the binding of HuR to the myogenin and p21 mRNAs (Supplementary Fig. 10), it decreased its association with the NPM mRNA (Fig. 7j). Taken together, these data suggest that in muscle cells the HuR/KSRP complex assembles in an RNA-independent manner and is recruited to the P1-1 element, thus promoting the rapid destabilization of the NPM mRNA.
To gain insight into the mechanism by which HuR executes these functions, we first determined the HuR domain responsible for this activity. We generated plasmids expressing GFP-tagged wild-type HuR or various deletion mutants of HuR (Fig. 8a). Each one of these isoforms was then expressed in C2C12 cells that were subsequently used for IP experiments with the anti-KSRP antibody. Western blot analysis with the anti-GFP antibody has shown that, similarly to wt-HuR, polypeptides harbouring the RRM3 motif (GFP-HNS-RRM3 and GFP-RRM3), but not those harbouring RRM1-2 or HuR nucleocytoplasmic shuttling (HNS) alone, were able to associate with KSRP (Fig. 8b) and promote a significant decrease in NPM mRNA level (Fig. 8c). In addition, an IP/RT-qPCR experiment on cells expressing GFP-HuR, -RRM1-2 or -RRM3 indicated that, while RRM1-2 did not show a strong association with the NPM mRNA, RRM3 associated more with NPM message than wt-HuR (Fig. 8d,e). Since, during myogenesis, cytoplasmic HuR undergoes caspase-mediated cleavage generating HuR-CP1 (24 kDa, RRM1-RRM2-ΔHNS1) and HuR-CP2 (8 kDa, ΔHNS2-RRM3)4 (Fig. 2C and Supplementary Fig. 11a), we next verified the effect of these CPs on NPM expression. Our data confirmed that HuR-CP2, which harbours the RRM3 motif, but not HuR-CP1, associated with KSRP and the NPM mRNA and was also able to form a complex with the HuR/KSRP-binding element in the NPM 3′-UTR (the P1-1 probe) (Supplementary Fig. 11b–g). In addition, similarly to GFP-RRM3 (Fig. 8c), we observed that the overexpression of HuR-CP2 promoted a significant decrease in NPM mRNA steady-state levels in a KSRP dependent manner (Supplementary Fig. 11h,i). Together, these results strongly indicate that the RRM3 motif of HuR plays a key role in mediating the formation of the HuR/KSRP in differentiating muscle cells.
Since KSRP has been shown to promote mRNA decay by recruiting the ribonuclease PARN and components of the exosome such as EXOSC5 (refs 3, 17), we investigated whether enzymes could play a role in the HuR/KSRP-mediated destabilization of NPM mRNA. Our IP experiments showed that HuR associates with both PARN and EXOSC5 in an RNA-independent manner (Fig. 9a and Supplementary Fig. 12). Interestingly, although both PARN and EXOSC5 interact with the NPM mRNA in C2C12 cells, this association is lost in HuR- or KSRP-depleted cells (Fig. 9b). To assess whether PARN and/or EXOSC5 play a role in the HuR/KSRP-mediated effect on NPM expression, we expressed GFP-HuR or GFP-HuR-CP2 in C2C12 cells depleted or not of these two ribonucleases (Fig. 9c), and the expression level of the NPM mRNA was determined using RT-qPCR. Our data showed that, while knocking down PARN (Fig. 9d) or EXOSC5 (Fig. 9e) in C2C12 cells increased the expression of the NPM mRNA, the absence of these ribonucleases prevented the HuR- or HuR-CP2-mediated NPM mRNA decay (Fig. 9f). The fact that the depletion of XRN1 (a 5′–3′ exoribonuclease component of the decapping complex)27 did not have a major effect on NPM mRNA expression (Supplementary Fig. 13) suggests that only ribonucleases that are recruited by KSRP such as PARN and EXOSC5 take part in this destabilization activity. Interestingly, we also demonstrate that knocking down PARN and EXOSC5 totally prevented C2C12 cells from entering the differentiation process (Fig. 9g–i) and the depletion of NPM in these cells partially re-established their myogenic potential (Supplementary Fig. 14). Of note, the weak rescue of myogenesis in cells depleted of both EXOSC5 and NPM suggests that EXOSC5 targets other messages in muscle cells. Together, our data indicate that during early myogenesis the formation of the HuR/KSRP complex leads to the recruitment of ribonucleases such as PARN and EXOSC5, which in turn destabilize the NPM mRNA, thus promoting muscle fibre formation.
In this work, we delineate the molecular mechanisms by which muscle cells downregulate NPM expression and show that this event represents a key regulatory step required for their commitment to myogenesis. Although previous studies have indicated that lowering the expression of NPM is required for the differentiation of a variety of cells such as HL60 and K562 (refs 19, 20), the molecular mechanism and the players behind this reduction remained elusive. Here we show that the decrease in NPM expression involves the destabilization of its mRNA via a mechanism mediated by the RNA-binding protein HuR. This observation was unexpected since HuR has been previously shown to stabilize the NPM mRNA in intestinal cells undergoing stress18. In undifferentiated muscle cells, the depletion of endogenous HuR results in the stabilization of the NPM mRNA and its increased expression, while HuR overexpression elicits the opposite effects. This mRNA destabilization activity of HuR involves its association, via the HuR-RRM3 motif, with KSRP and the ribonucleases PARN and EXOSC5 (ref. 17)17. Together, our data support a model whereby in response to a myogenic signal, HuR forms a complex with KSRP, PARN and the exosome, which in turn is recruited to a U-rich element in the 3′-UTR, leading to the destabilization of the NPM mRNA and the promotion of the early steps of myogenesis.
Although a NPM overexpression was previously linked to the inhibition of the differentiation of human promyelocytic leukaemia HL60 cells and megakaryocytic K562 cells19,20, our work provides the first demonstration that NPM is implicated in muscle fibre formation. NPM is highly expressed in a variety of cell lines and tissues and has been involved in many cellular processes such as cell proliferation and stress response28,29. In fact, manipulating the abundance of NPM has a dramatic effect on cell fate and on embryogenesis. Npm−/− mice do not complete their embryonic development and die between embryonic days E11.5 and E16.5 due to a severe anaemia caused by a defect in primitive hematopoiesis30. On the other hand, a high expression level of NPM is involved in cell transformation and tumour progression29,31,32,33. NPM levels significantly decrease in response to a variety of stimuli, among them those known to trigger cell cycle arrest and cell differentiation19,20,29. These data and the fact that NPM and HuR are both essential for the development of numerous cell lines and tissues12,28,29,34,35,36 highlight the importance of delineating the mechanism(s) by which HuR downregulates NPM expression.
Previously, others and we have demonstrated that HuR plays a prominent role during muscle cell differentiation by increasing the stability of promyogenic mRNAs at the later stages of this process5,7. Here we show that HuR is also involved in the early stages of myogenesis by degrading the NPM mRNA in a KSRP-dependent manner. Since previous reports have indicated that during myogenesis HuR and KSRP affect the half-lives of the same promyogenic transcripts, but in opposite ways, their collaboration to degrade a common target mRNA in undifferentiated muscle cells was unexpected. Indeed, it has been shown that KSRP plays a key role in maintaining low levels of p21 and myogenin mRNAs in undifferentiated muscle cells by promoting their rapid decay3. The fact that HuR does not associate with these mRNAs in these cells5 could explain why these transcripts became available to KSRP and the KSRP-associated decay machinery. However, during the step where muscle cells fuse to form myotubes, p38MAPK phosphorylates KSRP leading to the release of p21 and myogenin mRNAs from this decay machinery3. These findings, together with the data outlined in this manuscript, confirm the importance of HuR during muscle fibre formation and indicate that the functional consequences and the binding specificity of HuR to its target mRNA could change from one myogenic step to another depending on the nature of its protein ligand.
It is not surprising that HuR exerts different/opposite functions on its target transcripts. Numerous reports have indicated that depending on extra- or intracellular signals and/or the trans- or the cis-binding partners, HuR either activates or represses the turnover or translation of its mRNA targets10,16,37,38. Indeed, the Steitz laboratory has shown that HuR mediates the decay of the viral small nuclear RNA Herpesvirus saimiri U RNA 1 (HSUR 1) in an ARE-dependent manner39,40. Recently, HuR has been associated with the destabilization of the p16 mRNA both in HeLa and in IDH4 cells38. HuR participates in the decay of this message by forming a complex with AUF1 and Argonaute, two well-known mRNA decay factors38,41,42,43. Our data uncover a novel mechanism by which HuR mediates the decay of its mRNA target leading to muscle fibre formation. This mRNA decay-promoting activity of HuR depends on its association with KSRP, a factor known to promote the decay of ARE-containing mRNAs17,44. Interestingly, besides a difference in the way they form, the HuR/AUF1 and HuR/KSRP complexes also show a difference in their functional outcome. While in muscle cells HuR/KSRP assemble in a RNA-independent manner before their recruitment to the same U-rich element in the NPM 3′-UTR, the formation of the HuR/AUF1 complex in HeLa cells is RNA dependent and occurs via the binding of each one of these two proteins with a different element in the p16 3′-UTR38. Likewise, unlike the HuR/AUF1 complex that promotes the proliferation of HeLa and IDH4 cells38, the HuR/KSRP complex triggers the differentiation of muscle cells, an event linked to a cell cycle withdrawal21.
Besides the association with different protein ligands, we still do not know whether other processes such as post-translational modifications are also involved in the functional switch of HuR during myogenesis. It has been shown that phosphorylation on various serine residues could impact the association of HuR with some of its target mRNAs such as p21 (refs 45, 46, 47). While the implication of phosphorylation in HuR-mediated effects in muscle cells is still elusive, we recently showed that caspase-mediated cleavage plays a key role in modulating the promyogenic function of HuR in refs 4 and 8. Although our data indicated that HuR-CP1 promotes the cytoplasmic accumulation of HuR during the myoblast-to-myotube fusion4, the function of HuR-CP2 remained unclear. Here we demonstrate that HuR-CP2 associates with KSRP and is sufficient to downregulate the expression of NPM mRNA in muscle cells (Supplementary Fig. 11). Interestingly, we also show that, unlike the non-cleavable isoform of HuR (HuRD226A), HuR-CP2 can re-establish the myogenic potential of C2C12 cells depleted of endogenous HuR (Fig. 2i–l). While additional experiments are needed to confirm the role of this cleavage product in the promyogenic function of HuR, these observations raise the possibility of its implication in maintaining a low expression of levels of NPM during myogenesis. While these and likely other modifications modulate HuR function in muscle cells, the data described above establish that HuR, through its ability to promote the decay of target mRNAs, acquires more flexibility to use different ways to impact important processes such as muscle fibre formation. Therefore, since a change in HuR function and associated partners has been linked to chronic and deadly diseases such as cancer and muscle wasting48,49,50, delineating the molecular mechanisms behind the HuR-mediated mRNA decay activity could identify novel targets that help design effective strategies to treat these patients.
The pYX-Asc plasmid containing the full-length mouse NPM cDNA (Accession Number: BC054755) was purchased from Open Biosystems (Catalogue Number: MMM1013-9497870). To generate the GFP-NPM plasmid, full-length mouse NPM was amplified by PCR using the pYX-Asc plasmid as template and the following primers: forward 5′-GGCAAGCTTCGTCTGTTCTGTGGAACAGGA-3′ and reverse 5′-CGGGATCCGGGAAAGTTCTCACTTTGCATT-3′. The pAcGFP1-C1 (Clontech) and the PCR products were digested by HindIII and BamHI restriction enzymes (NE Biolabs). To generate the pRL-SV40-NPM plasmid, the full-length 3′-UTR of mouse NPM was amplified by PCR using the pYX-Asc plasmid as template and the following primers: forward 5′-GCTCTAGAGAAAAGGGTTTAAACAGTTTGA-3′ and reverse 5′-CCGGCGGCCGCACTTTATTAAAATACTGAGTTTATT-3′. The pRL-SV40 vector (Promega) and the PCR products were digested by XbaI and NotI restriction enzymes (NE Biolabs). PCR inserts were ligated into the plasmids using the T4 DNA ligase (NE Biolabs) according to the manufacturer’s instructions. pRL-SV40-NPMmut1 and mut2 plasmids were generated by Norclone Biotech Laboratories, Kingstone, ON, Canada (as described in text). The GFP and GFP-HuR plasmids were generated and used as described in ref. 51.
Cell culture and transfection
C2C12 muscle cells (ATCC, Manassas, VA, USA) were grown in a media containing 20% fetal bovine serum (Invitrogen) in DMEM (Dulbecco’s modified Eagle medium from Invitrogen). To induce muscle cell differentiation, cells were switched to a media containing DMEM, 2% horse serum, penicillin/streptomycin antibiotics (Invitrogen) and 50 mM HEPES, pH 7.4 (Invitrogen), when their confluency reached 100%5,6.
For the generation of stable cell lines, C2C12 cells were transfected with GFP, GFP-HuR and GFP-NPM plasmids, and 24 h post-transfection G418 (1 mg ml−1; Sigma-Aldrich) was added to the media. The stable cell lines were selected with G418 for 2 weeks and then sorted by flow cytometry (fluorescence-activated cell sorting) to select clones with equal expression levels. Transfections with siRNAs specific for HuR were performed using jetPEI (Polyplus Transfection) according to the manufacturer’s instructions. For DNA plasmid transfections, C2C12 cells at 70% confluency were transfected into six-well plates with 1.5 μg of plasmid DNA using Lipofectamine and Plus reagents (Invitrogen) according to the manufacturer’s instructions. Cells were subsequently incubated at 37 °C for 24 h before collecting and analysis.
siRNA oligonucleotides against HuR (5′-AAGCCUGUUCAGCAGCAUUGG-3′ (ref. 5)), NPM (5′-CAUCAACACCGAGAUCAAAdTdT-3′) (Dharmacon, USA), KSRP (5′-GGACAGUUUCACGACAACGdTdT-3′ (ref. 3)) and the siRNA Ctr (5′-AAGCCAAUUCAUCAGCAAUGG-3′) ref. 5 were synthesized by Dharmacon, USA. siRNA-PARN (5′-GGAUGUCAUGCAUACGAUUtt-3′), EXOSC5 (5′-UCUUCAAGGUGAUACCUCUtt-3′) and XRN1 (5′-GAGGUGUUGUUUCGAAUUAtt-3′) were acquired from Ambion, USA.
Preparation of cell extracts and immunoblotting
Total cell extracts were prepared52 and western blot was performed using antibodies against HuR (3A2, 1:10,000)53, NPM (anti-B23 clone FC82291, Sigma-Aldrich, 1:30,000), My-HC (Developmental studies Hybridoma5, 1:1,000), Myoglobin (DAKO, 1:500), α−tubulin (Developmental studies Hybridoma Bank, 1:1,000), GFP (Clontech, Mountain View, CA, USA, 1:1,000), and KSRP (affinity-purified rabbit serum17, 1:3,000), anti-myogenin (F5D, obtained from Developmental Studies Hybridoma Bank, 1:250) and caspase 3 cleavage product (Cell Signaling Technology, 1:1,000). Scans of the most important blots are shown in Supplementary Fig. 15.
Immunofluorescence5 was performed to detect GFP expression as well as to visualize myotubes using antibodies against My-HC (1:1,000) and Myoglobin (1:500). Staining with 4′,6-diamidino-2-phenylindole (DAPI) was employed to visualize nuclei.
The fusion index5 indicating the efficiency of C2C12 differentiation was determined by calculating the number of nuclei in each microscopic field in relation to the number of nuclei in myotubes in the same field.
Northern blot analysis and actinomycin D pulse-chase experiments
The extraction of total RNA from C2C12 cells was performed using Trizol reagent (Invitrogen) according to the manufacturer’s instructions. Northern blot analysis23 was performed using probes specific for MyoD, HuR, 18S48 and NPM (Supplementary Table 2). These probe RNAs were generated using the PCR Purification Kit (GE Healthcare) and radiolabelled with α32P deoxycytidine triphosphate (dCTP) using Ready-to-Go DNA labelling beads (GE Healthcare) according to the manufacturer’s instructions. The stability of NPM mRNA was assessed by the addition of the transcriptional inhibitor actinomycin D (5 μg ml−1; Sigma-Aldrich)48 for the indicated periods of time.
1 μg of total RNA was reverse transcribed using the M-MuLV RT system. A 1/80 dilution of cDNA was used to detect the mRNAs using SsoFast EvaGreen Supermix (Bio-Rad). Expression of NPM, MyoD, p21, myogenin and RLuc was standardized using GAPDH as a reference (see primers in Supplementary Table 2), and relative levels of expression were quantified by calculating 2−ΔΔCT, where ΔΔCT is the difference in CT (cycle number at which the amount of amplified target reaches a fixed threshold) between target and reference.
Genomic DNA extracts and analysis by RT-qPCR
Cell pellets are resuspended in the digestion buffer (100 mM NaCl; 10 mM Tris–HCl, pH 8; 25 mM EDTA, pH 8; 0.5% SDS; 0.1 mg ml−1 proteinase K) and incubated at 50 °C overnight. Genomic DNA (gDNA) is extracted with phenol chloroform (Invitrogen) according to the manufacturer’s instructions. RT-qPCR experiment was performed as described above starting from 16 pg of extracted gDNA.
Cell extracts were prepared in the lysis buffer (50 mM Tris, pH 8; 0.5% Triton X-100; 150 mM NaCl; complete protease inhibitor (Roche)). When indicated, cell extracts were digested for 30 min at 37 °C with RNase A (100 μg ml−1). Anti-KSRP serum (2 μl) (affinity-purified rabbit serum17) was incubated with 50 μl of proteinA-sepharose slurry beads (washed and equilibrated in cell lysis buffer) for 1 h at 4 °C. Beads were washed three times with cell lysis buffer and incubated with 300 μg of cell extracts overnight at 4 °C. Beads were washed three times with cell lysis buffer and co-immunoprecipitated proteins were analysed by western blot. HuR, PARN and EXOSC5 IP were performed3,5,54 using antibodies against HuR (3A2), PARN (Cell Signaling, USA), EXOSC5 (Abcam, USA) and IgG (Jackson Immunoresearch Laboratories, USA).
Preparation of mRNA (mRNP) complexes and analysis by RT-PCR
Purified RNA from mRNP complexes5,54 was resuspended in 10 μl of water and 4 μl was reverse transcribed using the M-MuLV RT system (NE Biolabs) according to the manufacturer’s protocol. Subsequently, 2 μl of cDNA was amplified by PCR or qPCR using NPM-specific primers (Supplementary Table 2). When analysed by qPCR, a 1/20 dilution of cDNA was used to detect the mRNAs using SybrGreen (SsoFast EvaGreen Supermix). The mRNA levels associated with these mRNP complexes were then standardized against RPL32 mRNA levels (used as a reference) and compared with mRNA levels in the IgG control. In experiments transfected with Rluc reporter constructs, the Rluc mRNA associated with immunoprecipiated HuR was determined by RT-qPCR, standardized against RPL32 mRNA levels and then the steady-state levels for each treatment. These standardized Rluc mRNA levels were then compared with Rluc mRNA levels in the IgG IP.
cDNA array analysis
Microarray experiments were performed using mouse array, which contain probe sets of characterized and unknown mouse expressed sequence tags (ESTs) from 17,000 genes55. RNAs from siHuR- and siCtr-treated C2C12 cell extracts were prepared56, processed and hybridized on the arrays. The data were processed using the Array Pro software (Media Cybernetics), then normalized by Z-score transformation55 and used to calculate differences in signal intensities. Significant values were tested using a two-tailed Z-test and a P-value of ≤0.01. The data were calculated from two independent experiments.
Fluorescence in situ hybridization
The fluorescence in situ hybridization experiments were performed57 using a DNA fragment of ~500 bp corresponding to the coding region of mouse NPM. The fragment was amplified by PCR using the following primers fused to either a T7 or T3 minimal promoter sequence: NPM forward, 5′-TAATACGACTCACTATAGGGACGGTTGAAGTGTGGTTCAG-3′ and NPM reverse, 5′-AATTAACCCTCACTAAAGGAACTTGGCTTCCACTTTGG-3′. The PCR product was used as the template for in vitro transcription of the NPM probe needed for fluorescence in situ hybridization. The antisense (T3) and sense (T7) probes were prepared using digoxigenin-RNA labelling mix (Roche Diagnostics). The RNA probes were quantified, denatured and incubated with permeabilized cells57. After the hybridization, the cells were used for immunofluorescence to detect HuR57.
A total of 4 × 107 myoblasts were grown and treated with siRNAs as described above. Briefly, the cytoplasmic extracts obtained from lysed myoblast cells were centrifuged at 130,000 g for 2 h on a sucrose gradient (10–50% w/v)58. RNA was extracted using Trizol LS (Invitrogen), and was then analysed on an agarose gel. The levels of NPM and GAPDH mRNAs were determined by RT-qPCR.
RNA electrophoretic mobility shift assays
The NPM cRNA probes were produced by in vitro transcription59. The accession number in the NCBI database of the NPM mRNA sequence used to generate these probes is NM_008722. The NPM probes 5′-UTR, 3′-UTR, P1 and P2 were generated by PCR amplification using a forward primer fused to the T7 promoter (Supplementary Table 2) as well as pYX-Asc-NPM expression vector as the template. For smaller probes (P1-1 to P2-3, P1-1-mut1, P1-1-mut2, P1-1-mut1-2), oligonucleotide sense and antisense (see Supplementary Materials and Methods) were directly annealed and used for in vitro transcription. The RNA-binding assays23 were performed using 500 ng purified recombinant protein (GST or GST-HuR) incubated with 50,000 c.p.m. of 32P-labelled cRNAs. Supershift experiments were also performed with 10 μg total C2C12 cell extract incubated with 50,000 c.p.m. of 32P-labelled cRNAs. An anti-KSRP antibody was then added to the reaction to supershift the RNP/cRNA complex containing KSRP.
The activity of Renilla luciferase was measured using a Renilla luciferase assay system (Promega) with a luminometer following the manufacturer’s instructions.
In vitro pull-down experiments
GST pull-down assay4 was performed using GST-HuR and recombinant His-KSRP. The interaction of the His-KSRP with the pulled-down GST-HuR was analysed by western blot using the GST and KSRP antibodies.
Accession Codes: Microarray data has been deposited in Gene Expression Omnibus under accession number GSE57113.
How to cite this article: Cammas, A. et al. Destabilization of nucleophosmin mRNA by the HuR/KSRP complex is required for muscle fibre formation. Nat. Commun. 5:4190 doi: 10.1038/ncomms5190 (2014).
Gene Expression Omnibus
We thank Dr C. von Roretz for critical review of this manuscript and insightful discussion. We are grateful to Veronique Sauve and Dr Kalle Gehring for their help with the BIACORE experiments. A.C. is a recipient of the Canderal scholarship for PDFs (2010–2011) as well as the Peter Quinlan Fellowship in Oncolgy (2011–2012). V.D.-R. is a Research Fellow of the Terry Fox foundation through an award from the Canadian Cancer Society (former National Cancer Institute of Canada). This work was supported by a CIHR (MOP-89798) operating grant to I.-E.G. M.G. is supported by the NIA-IRP, NIH. R.G. is supported by grants from AIRC (IG 2010–2012) and AICR (No. 10-0527). I.-E.G. is a recipient of a TierII Canada Research Chair.
Supplementary Figures 1-15 and Supplementary Table 1-2