Regulation of SIK1 abundance and stability is critical for myogenesis

Abstract

cAMP signaling can both promote and inhibit myogenic differentiation, but little is known about the mechanisms mediating promyogenic effects of cAMP. We previously demonstrated that the cAMP response element-binding protein (CREB) transcriptional target salt-inducible kinase 1 (SIK1) promotes MEF2 activity in myocytes via phosphorylation of class II histone deacetylase proteins (HDACs). However, it was unknown whether SIK1 couples cAMP signaling to the HDAC-MEF2 pathway during myogenesis and how this response could specifically occur in differentiating muscle cells. To address these questions, we explored SIK1 regulation and function in muscle precursor cells before and during myogenic differentiation. We found that in primary myogenic progenitor cells exposed to cAMP-inducing agents, Sik1 transcription is induced, but the protein is rapidly degraded by the proteasome. By contrast, sustained cAMP signaling extends the half-life of SIK1 in part by phosphorylation of Thr475, a previously uncharacterized site that we show can be phosphorylated by PKA in cell-free assays. We also identified a functional PEST domain near Thr475 that contributes to SIK1 degradation. During differentiation of primary myogenic progenitor cells, when PKA activity has been shown to increase, we observe elevated Sik1 transcripts as well as marked accumulation and stabilization of SIK1 protein. Depletion of Sik1 in primary muscle precursor cells profoundly impairs MEF2 protein accumulation and myogenic differentiation. Our findings support an emerging model in which SIK1 integrates cAMP signaling with the myogenic program to support appropriate timing of differentiation.

Myoblast differentiation is driven by a transcriptional cascade that is subject to precise temporal control during development as well as after acute muscle injury (1). The second messenger cAMP and its primary effector PKA are known to contribute to myoblast proliferation in the developing dermomyotome (2), to promote migration and fusion of muscle precursor cells (3, 4), and to exert potent anabolic effects on adult skeletal muscle (5). During differentiation of myoblasts ex vivo, intracellular cAMP peaks transiently before cell–cell fusion (6, 7), and PKA signaling is required for expression of differentiation markers in cultured myoblasts (8) and in mouse embryos (2). However, sustained cAMP-PKA signaling prevents myogenic differentiation, in part through inhibition of basic helix-loop-helix and MADS transcription factors (9–11). The mechanisms by which a transient peak of cAMP promotes myoblast differentiation are still poorly understood. Because ligands that induce cAMP production promote muscle regeneration (5), it is important to identify promyogenic cAMP signaling effectors.

The transcription factor CREB (cAMP response element-binding protein) is a key effector of cAMP-PKA in many cell types (12). CREB phosphorylation is induced during early stages of vertebrate myogenesis (2, 13), and CREB activity is required in mice for dermomyotome development (2). We previously showed that CREB activity is required for survival of adult skeletal myofibers, and salt-inducible kinase 1 (Sik1, also termed Snf1lk) is a critical CREB target gene for muscle maintenance (14). SIK1 is an AMPK-related serine/threonine kinase that shares structural features with other AMPK family members and requires phosphorylation by liver kinase B1 (LKB1) for activity (reviewed in ref. 15). However, SIK1 is unique among these enzymes by virtue of cAMP-dependent transcriptional regulation in several cell types including adrenocortical cells (16), hepatocytes (17), and skeletal myoblasts (14). In myoblasts, SIK1 directly phosphorylates and inhibits class II histone deacetylase proteins (HDACs), which leads to de-repression of MEF2 activity and full expression of terminal myogenic genes including desmin (14). Intriguingly, the SIK/AMPK-HDAC-MEF2 pathway is evolutionarily conserved in Caenorhabditis elegans sensory neurons (18) and mediates distinct physiologic functions in mammalian striatal neurons (19), chondrocytes (20, 21), and hepatocytes (22, 23).

We previously reported that amounts of Sik1 mRNA and SIK1 protein are very low in undifferentiated C2C12 myoblasts until the cells are exposed to agents that induce intracellular cAMP production (14). Paradoxically, sustained cAMP-PKA signaling inhibits SIK1-dependent substrate phosphorylation in cells by direct phosphorylation of SIK1 on Ser577 (24), and possibly additional mechanisms (14, 23), resulting in SIK1 cytoplasmic sequestration. Newly synthesized SIK1 protein is thus restricted from phosphorylating nuclear targets until intracellular cAMP levels decay. In myoblasts, after removal of the priming cAMP stimulus, SIK1 becomes dephosphorylated and potentiates class II HDAC phosphorylation (14), likely in the nucleus. Because cAMP levels (6, 7), class II HDAC phosphorylation (25, 26), and MEF2 activity (27) increase during myogenic differentiation, we tested the hypothesis that SIK1 links cAMP signaling with myogenic differentiation. We show that in myoblasts SIK1 is rapidly degraded by the proteasome, but SIK1 accumulates in differentiating myocytes. We provide evidence that SIK1 is required for myogenic differentiation and propose that transcriptional regulation and regulated degradation comprise a unified molecular mechanism to precisely tune SIK1 activity in muscle cells.

Results

SIK1 Is Degraded by the Proteasome in Myoblasts.

SIK1 protein expression is low in resting C2C12 myoblasts, but Sik1 mRNA and SIK1 protein rapidly accumulate 1–3 h after treatment with agents that induce cAMP [adenylyl cyclase agonist forskolin (FSK) plus phosphodiesterase inhibitor IBMX]. SIK1 protein declines gradually 4–6 h after continuous treatment (14). We prepared custom antibodies to further investigate the dynamics of SIK1 protein abundance and phosphorylation in C2C12 cells. In C2C12 cells treated with FSK/IBMX, the SIK1 antiserum recognizes endogenous SIK1 as an 85- to 90-kDa doublet (Fig. 1 and Fig. S1A), which is absent in C2C12 cells expressing SIK1-specific shRNA (Fig. S1A). The upper SIK1 band in FSK/IBMX-treated cells is likely PKA-phosphorylated SIK1, because phosphatase treatment collapses the bands to the lower molecular-weight species in both C2C12 and primary mouse skeletal myoblasts (Fig. S1 B and C). We noticed that after removal of the cAMP-inducing mixture and chase in complete medium, SIK1 becomes rapidly dephosphorylated and total SIK1 protein declines over time in both C2C12 myoblasts (Fig. S1 D and F) and primary mouse myoblasts (Fig. S1 E and F). The loss of SIK1 after removal of FSK/IBMX could result from termination of Sik1 transcription, termination of SIK1 translation, or turnover of newly synthesized SIK1 protein. To test the contribution of protein turnover, we included the protein synthesis inhibitor cycloheximide (CHX) in the chase medium. We observed that in both C2C12 and primary myoblasts the half-life of endogenous SIK1 is less than 30 min after removal of FSK/IBMX (Fig. 1 A and B). Similarly, stimulation of β-adrenergic receptors with isoproterenol/ IBMX induces SIK1 in C2C12 cells, but the protein is rapidly lost after agonist removal (Fig. S1 G and H). These data indicate that SIK1 protein abundance is regulated by a posttranslational mechanism that either directly affects SIK1 or another protein important for SIK1 stability. Based on the rapid posttranslational loss of SIK1 protein, we hypothesized that SIK1 is degraded by the proteasome. Consistent with this hypothesis, SIK1 protein was markedly stabilized when the proteasome inhibitor MG-132 was included in the chase medium compared with control medium (Fig. 1C). Moreover, we recovered ubiquitylated exogenous SIK1 protein from HEK293T cells by pull-down of his-tagged SIK1 (Fig. 1D) or of coexpressed his-tagged ubiquitin (Fig. S1I) under denaturing conditions. Thus, SIK1 protein is degraded in myoblasts by the proteasome after removal of a cAMP-inducing stimulus.

Fig. 1.

SIK1 protein is degraded by the proteasome in myoblasts. (A) Endogenous SIK1 and HSP90 proteins in C2C12 myoblasts treated with FSK/IBMX (3 h) to induce SIK1 protein, washed, and chased in medium containing CHX (n = 8). Compare Fig. S1D for wash only and no CHX controls. (B) Endogenous SIK1 protein in primary mouse skeletal myoblasts following 3 h of FSK/IBMX priming and CHX chase (n = 5). Fig. S1E shows vehicle control. (C) Endogenous SIK1 protein in C2C12 myoblasts treated with FSK/IBMX (3 h) and chased in medium containing the indicated agent (FSK, DMSO vehicle, or MG-132 but no CHX) for 30 min (n = 7). MG-132 was added 30 min before wash and chase. (D) SIK1 ubiquitylation visualized by pull-down of his-tagged SIK1 under denaturing conditions. Western blots show SIK1 (Left) and ubiquitin (Right) in input and pull-downs (His PD) (n = 3).

SIK1 Degradation Is Regulated by cAMP Signaling.

In the course of priming and chase studies, we found that inclusion of FSK/IBMX in the chase medium maintained SIK1 phosphorylation and nearly doubled the half-life of endogenous SIK1 protein in C2C12 myoblasts (Fig. 2 A and B), suggesting that cAMP signaling may protect SIK1 from degradation. However, because Sik1 mRNA transcription is also regulated by cAMP-PKA signaling (14), we evaluated Sik1 mRNA turnover. We performed the same type of cAMP priming and chase assays using the RNA polymerase inhibitor actinomycin D (ActD). As expected, Sik1 mRNA amounts declined following FSK/IBMX removal (Fig. 2C), but the half-life of Sik1 mRNA was insensitive to cAMP signaling and was longer than that of SIK1 protein. cAMP promotes SIK1 cytoplasmic accumulation in adrenocortical cells (24) and myoblasts (Fig. S2 A and B). This is thought to result primarily from direct PKA phosphorylation of SIK1 on Ser577 (24). To test whether PKA phosphorylation of SIK1(Ser577) affects SIK1 stability, we compared the half-lives of WT and SIK1(S577A) by CHX chase assays in C2C12 myoblasts. The half-life of SIK1(S577A) was similar to that of SIK1 WT (Fig. 2D and Table S1), suggesting that cAMP signaling regulates SIK1 by an alternate mechanism.

Fig. 2.

SIK1 degradation is regulated by cAMP signaling. (A) SIK1 protein in C2C12 cells primed with FSK/IBMX for 3 h and chased in medium with CHX ± FSK/IBMX (n = 4). (B) Quantification of data in A. Half-life of endogenous SIK1 (n = 3, mean percent SIK1 remaining normalized to HSP90 ± SD). (C) Sik1 mRNA degradation in C2C12 cells stimulated for 3 h with FSK/IBMX, chased with ActD ±FSK/IBMX, expressed as normalized percent of maximum ± SD (average of n = 2, representative of n = 4). (D) CHX chase assays of myc-SIK1 (WT or S577A) in C2C12 cells; see also Table S1 (n = 3).

Phosphorylation of SIK1(T475) Regulates SIK Stability and Localization.

We hypothesized that PKA may phosphorylate SIK1 at additional, uncharacterized residues and identified two putative PKA consensus phosphorylation sites in SIK1 (T268 and T475; Fig. 3A). Of these, T475 is conserved in all SIK orthologs (Fig. 3B). The analogous site in SIK3(T469) was recently shown to be phosphorylated by PKA in adipocytes (28), and SIK2(T484) has been implicated in degradation of SIK2 induced by calcium-dependent kinases (29). We mutated these sites alone or in combination in catalytically inactive SIK1(K56M) (kinase dead, or “KD”) to prevent autophosphorylation and tested phosphorylation of SIK1 by recombinant PKA in vitro. Mutation of T475 slightly reduced SIK1 phosphorylation by PKA and abrogated the PKA-dependent electrophoretic mobility shift of SIK1 (Fig. 3C and Fig. S2 C and D). The double SIK1(KD-T475A,S577A) and triple SIK1(KD-T268A,T475A,S577A, or “KD-3A”) phospho-site mutants were almost completely resistant to PKA phosphorylation (Fig. 3C and Fig. S2 C and D). We confirmed that phosphorylation of SIK1 in these assays was due to recombinant PKA; no phosphorylation was observed when PKA was omitted from the reactions (Fig S2E). We next verified that T475 is phosphorylated in cells by immunoprecipitating myc-tagged SIK1 (WT, T475A, T475E, and S577A) from HEK293T cells treated with FSK/IBMX for 30 min (Fig. 3D). An antibody specific for phosphorylated PKA consensus motifs revealed basal and FSK-induced SIK1 phosphorylation (Fig. 3D Upper), which was reduced by mutation of either T475 or S577. Mutation of T475 did not affect phosphorylation of S577 (Fig. 3D Lower) or the in vitro kinase activity of SIK1 on recombinant GST-HDAC5 (Fig. S2F). These results show that SIK1(T475) is phosphorylated in cells in response to cAMP signaling.

Fig. 3.

SIK1 phosphorylation by PKA on T475 contributes to stability. (A) Diagram of SIK1 protein with known (T182 by LKB1; S577 by PKA) and hypothesized (italic) phosphorylation sites. NLS, nuclear localization signal; UBA, ubiquitin-associated domain. (B) Sequence surrounding T475 in SIK orthologs. (C) In vitro kinase assay of recombinant PKA on purified SIK1 kinase-inactivated (KD) mutants. KD-3A, SIK1(KD-T268A,T475A,S577A). Upper, ³²P autoradiogram; Lower, silver stain. SIK1, IgG heavy chain (Hc), and PKA are indicated (n = 3). Compare Fig. S2 for quantification and no PKA control. (D) Western blots of immunoprecipitated myc-SIK1 mutants with phospho-PKA substrate, phospho-SIK1(S577), and myc antibodies. HEK293T cells were pretreated with vehicle or FSK/IBMX (30 min) (n = 3). (E) CHX chases of SIK1 mutants in C2C12 cells (n ≥ 3 per mutant). (F) Quantification of data in E (mean of n ≥ 3 ± SD); significance to WT shown for T475E, WT-NLS, and TE-NLS. See also Table S1. (G) Ubiquitylation of SIK1 mutants from MG-132 treated HEK293T cells purified by his-SIK pull-down under denaturing conditions, probed for ubiquitin or SIK1 (n = 3).

To determine whether T475 participates in SIK1 stability, we measured the half-lives of SIK1 mutants using CHX chase assays in C2C12 myoblasts. Mutation of T475 to glutamate (T475E, phosphomimetic) dramatically stabilized the protein compared with WT (Fig. 3 E and F and Table S1). Surprisingly, SIK1(T475A) had the same, extended half-life as SIK1(T475E) (Fig. 3 E and F and Table S1), suggesting that threonine at this site may be important for recognition by ubiquitin ligase machinery. However, when T475 was mutated to alanine in catalytically inactive SIK1(K56M), referred to as “KD-TA,” the double mutant had a much shorter half-life than SIK1 WT, KD, or T475 mutants (Fig. 3 E and F and Table S1). This finding suggests that T475A cooperates with the kinase activity to regulate SIK1 stability. We compared ubiquitylation of WT and mutant SIK1 proteins by ubiquitylation assays in HEK293T cells. Consistent with the observed differences in half-lives, ubiquitylation of SIK1(T475E) was decreased, whereas that of SIK1(KD-TA) was increased compared with WT SIK1 (Fig. 3G). Interestingly, SIK1(T475A) was ubiquitylated to the same or greater extent compared to SIK1-WT (Fig. 3G).

Because phosphorylation of SIK1 by PKA (24) and LKB1 (30) promote SIK1 nuclear export, we tested the subcellular localization of SIK1 mutant proteins in C2C12 cells. SIK1(T475E) and SIK1(T475A) were localized in both the cytoplasm and nucleus but were refractory to FSK-induced cytoplasmic sequestration, as were SIK1(S577A) and kinase-dead SIK1(K56M) (Fig. S2 A and B). By contrast, SIK1(KD-TA) was almost exclusively localized to the nucleus in 70% of the cells and was more resistant to cAMP-stimulated export than either SIK1(KD) or SIK1(S577A) (Fig. S2 A and B), suggesting that this mutant is destabilized in part due to constitutive nuclear localization. To test this, we measured the half-lives of WT SIK1 and SIK1(T475E) (“TE”) proteins fused to the SV40 nuclear localization signal (NLS), which resulted in constitutive nuclear localization in most cells (Fig. S2B). Strikingly, the constitutive NLS destabilized both SIK1-WT and SIK1(TE), the half-lives of which were similar to or shorter than that of SIK1(KD-TA) (Fig. 3 E and F and Table S1). These results suggest that after removal of the cAMP priming stimulus, SIK1(S577) and (T475) are dephosphorylated and SIK1 moves to the nucleus, where it is degraded by the proteasome.

The PEST Domain of SIK1 Promotes Degradation.

To better understand how SIK1 stability is regulated, we sought domains in the SIK1 primary sequence that may contribute to its short half-life. Like other AMPK-related kinases, SIK1 has a ubiquitin-associated (UBA) domain adjacent to its kinase domain (ref. 31 and Fig. 4A). However, UBA domains in AMPK-related kinases neither bind ubiquitin in vitro nor are they modified by ubiquitylation. Rather, the UBA domain is important for the conformation of the adjacent kinase domain and phosphorylation by the upstream activating kinase LKB1 (31). We further searched the primary sequence and identified a putative PEST domain (32) using the PESTfind algorithm (score of +8.9) (33). PEST domains are found in unstable proteins, but they do not confer instability by a uniform molecular mechanism (32). The SIK1 PEST domain (aa 451–472) is adjacent to T475 (Fig. 4A). Although this region is highly conserved between mouse, rat, and human SIK1, it lies in a region of SIK1 with unknown function and low overall similarity to other SIK kinases (15). To test the contribution of the PEST domain to SIK1 stability, we performed CHX chase assays on mutant SIK1 lacking this domain (SIK1ΔPEST). We observed a strong reduction of the basal turnover rate of SIK1ΔPEST compared with WT SIK1 (Fig. 4 B and C and Table S1). This internal deletion does not appear to affect the global folding or activity of SIK1, because SIK1ΔPEST exhibited catalytic activity comparable to that of WT SIK1 (Fig. S3). Surprisingly, SIK1ΔPEST was ubiquitylated to approximately the same extent as WT SIK1 in cells (Fig. 3G, lanes 2 and 3). We conclude that the PEST domain contributes to SIK1 degradation but is not required for ubiquitylation.

Fig. 4.

SIK1ΔPEST is a stabilized mutant. (A) Diagram of SIK1 protein showing the PEST domain (aa 451–472) and confirmed phosphorylation sites. PEST sequence from rat SIK1 is shown. (B) CHX chase of SIK1ΔPEST in C2C12 myoblasts (n = 5). (C) Quantification of data in B (mean of n ≥ 5 ± SD). See also Table S1.

SIK1 Is Required for Myogenic Differentiation.

We have shown that SIK1 is unstable in myoblasts, and we hypothesized that stabilized SIK1 might contribute to class II HDAC phosphorylation and MEF2 activity during myogenic differentiation. Consistent with our hypothesis and with the pattern of Sik1 expression in differentiating regions of mouse somites (34), we observed induction of both Sik1 mRNA and SIK1 protein during differentiation of primary mouse skeletal myoblasts cultured ex vivo (Fig. 5 A and B). We noticed that SIK1 protein induction precedes Sik1 mRNA and therefore hypothesized that increased SIK1 stability might contribute to the increased protein abundance in differentiating myocytes. To test this, we compared the half-life of endogenous SIK1 in primary undifferentiated myoblasts and fully differentiated myotubes. Because SIK1 is poorly abundant in undifferentiated cells, we treated the cells with a cAMP priming stimulus and performed chase assays as in Fig. 1. Consistent with our model, endogenous SIK1 is markedly stabilized in fully differentiated myotubes compared with undifferentiated myoblasts (Fig. 5 C and D). These data suggest that SIK1 accumulates as a result of both increased mRNA transcription and reduced protein degradation during differentiation of primary skeletal myoblasts.

Fig. 5.

SIK1 is required for differentiation of primary mouse myoblasts. (A) Sik1 mRNA in primary mouse muscle precursor cells cultured in differentiation medium (DM) for the indicated time in days. Sik1 mRNA normalized to Gapdh, expressed as fold change over day 0. Mean of n = 3 ± SD. (B) SIK1 protein in primary myoblasts incubated in DM for 0–5 d. (C) SIK1 half-life after FSK/IBMX priming and CHX chase in primary myoblasts (Upper) versus differentiated primary myotubes (>4 d, Lower). Cells were treated for 3 h with FSK/IBMX, washed, and incubated in medium containing CHX as indicated. (D) Quantification of data in C (mean of three experiments ± SD). (E) Western blots of differentiation markers during primary myoblast differentiation (hours in DM) after infection with Ad-USi or Ad-SIK1i. (F) Phase-contrast images of infected myoblasts 48 h after differentiation. A and B represent four or five experiments; C–F represent three experiments.

To directly test the importance of SIK1 for myogenic differentiation, we infected primary muscle precursor cells with adenovirus encoding unspecific shRNA (Ad-USi) or SIK1-specific shRNA (Ad-SIK1i) (14, 17). Forty-eight hours after infection, cells were either harvested or were induced to differentiate by incubation in differentiation medium (DM). Consistent with our prior results in C2C12 myoblasts (14), we observed that primary myoblasts with ∼20–50% of the normal amount of SIK1 (Fig. 5E) were prone to cell death. When plated at densities permissive for differentiation and incubated in DM, surviving Ad-SIK1i–infected cells failed to fully morphologically differentiate, whereas control Ad-USi–infected cells differentiated normally into elongated myotubes (Fig. 5F). We also monitored biochemical markers of differentiation. The transcription factor MEF2A increases during differentiation, and the contractile protein myosin heavy chain (MHC) is highly expressed at late stages of differentiation. Acute Sik1 knockdown completely blocked induction of MEF2A and blunted induction of MHC relative to control cells (Fig. 5E). In Ad-SIK1i–infected cells, class II HDAC phosphorylation was markedly reduced although not completely abrogated, possibly due to residual Sik1 expression (Fig. 5E, lane 6) or phosphorylation by other kinases. These data show that SIK1 is important for MEF2 expression and terminal differentiation of primary muscle precursor cells.

Discussion

In skeletal myoblasts, cascades of developmentally regulated transcription factors dictate cell fate decisions and myogenic differentiation (1). Although cAMP signaling and CREB-dependent transcription are dynamically regulated during myogenesis in vivo and in vitro (2, 13, 35–38), relatively little is known about molecular mechanisms by which cAMP-dependent transcription contributes to normal responses in myoblasts and adult skeletal muscle (5). We previously identified one cAMP effector pathway in myocytes mediated by the CREB target gene SIK1, which we showed inhibits class II HDACs and allows activation of MEF2-dependent transcription in skeletal muscle cells (14). The experiments presented here address several major outstanding questions about SIK1 function. How is SIK1 regulated in muscle cells? Does this change during myogenic differentiation? Is SIK1 required for regulation of MEF2 during myogenesis and for differentiation?

We found that cAMP signaling not only stimulates Sik1 transcription, but also promotes SIK1 protein stability by protecting SIK1 from degradation. In undifferentiated myoblasts, SIK1 is short-lived due to rapid proteasomal degradation when cAMP signaling declines. We identified a previously uncharacterized PKA phosphorylation site on SIK1 (T475) that affects both SIK1 stability and subcellular localization and provide evidence that cAMP signaling stabilizes SIK1 by a mechanism involving phosphorylation of T475. In support of this model, sustained phosphorylation of endogenous SIK1 correlates with an extended half-life. T475 phosphorylation could impede SIK1 recognition by a ubiquitin ligase or facilitate interaction with a deubiquitinating enzyme. Surprisingly, mutation of T475 to either alanine (phosphorylation-deficient) or glutamate (phosphorylation-mimetic) stabilized the protein, suggesting that threonine at this site is specifically important for SIK1 degradation. Mutation of T475 also rendered the protein insensitive to cAMP-induced nuclear export, which is thought to occur by phospho-Ser577–dependent masking of a neighboring nuclear localization sequence (24). Because dual mutations in the catalytic domain and T475 result in strong nuclear localization and destabilization, we propose that T475 cooperates with S577 and the kinase domain to direct subcellular SIK1 localization, possibly by cooperative binding to 14-3-3 proteins, which interact with the phosphorylated T-loop of SIK family kinases (30). The findings that forced nuclear localization of SIK1 results in rapid degradation and that sustained cAMP signaling (which causes SIK1 nuclear export) extends the half-life of endogenous SIK1 support the notion that cytoplasmic SIK1 is protected from degradation. We hypothesize that SIK1 is degraded in the nucleus by a mechanism involving the PEST domain, which may be masked by phosphorylation of T475. Although direct PKA phosphorylation of SIK1 is the most parsimonious mechanism to explain why SIK1 degradation is sensitive to cAMP signaling tone, our data do not exclude additional cAMP-dependent events, including transcriptional induction of an E3 ubiquitin ligase, modulation of a stability protein, or activation or inhibition of a distinct kinase or phosphatase. Indeed, the half-life of endogenous SIK1 in C2C12 or primary myoblasts pretreated with cAMP is dramatically shorter (15–30 min) than the half-life of exogenous SIK1 in unstimulated C2C12 myoblasts (4 h), suggesting that additional molecular mechanisms contribute to regulated SIK1 degradation.

The rapid, proteasome-dependent degradation of endogenous SIK1 we describe has not been previously reported. Our findings reveal a mechanism by which cAMP signaling exerts dual control of SIK1 accumulation: transcriptional induction and stabilization of the protein. SIK1 is rapidly degraded in undifferentiated myoblasts, but Sik1 mRNA and SIK1 protein accumulate during differentiation of primary muscle progenitors ex vivo, consistent with enrichment of Sik1 transcripts in differentiating regions of the somites in mouse embryos (34). We hypothesize that the observed accumulation of cAMP before muscle cell fusion exerts a priming effect (4) in part by stimulating CREB-dependent transcription of Sik1 mRNA (14, 39) and by stabilization of SIK1 protein by the mechanism we elucidated. Temporal control of SIK1 expression may affect differentiation, because SIK1 stability increases during myogenesis. However, cAMP-dependent stabilization is unlikely to fully account for elevated SIK1 abundance in differentiated myotubes, because intracellular cAMP decays later in differentiation (6, 7) and SIK1 protein appears dephosphorylated in differentiated myotubes. Depletion of SIK1 profoundly impairs muscle precursor cell survival and differentiation with concomitant loss of MEF2A expression, strongly reduced MHC accumulation, and failure of morphological differentiation. Class II HDAC phosphorylation is blunted but not eliminated in Sik1-depleted myocytes, so the lack of MEF2A induction may result from an HDAC-independent mechanism.

We propose a model in which SIK1 has a limited time in which to catalyze substrate phosphorylation in undifferentiated myoblasts exposed to a cAMP-inducing ligand. This would allow cells to rapidly return to a resting state after the cAMP signal decays. In vivo, exercise or activation of the “fight-or-flight response” causes catecholamine release from the adrenal medulla, resulting in cAMP accumulation in skeletal muscle and other tissues (5). We hypothesize that in this setting if muscle precursor cells responded with induction of Sik1, the protein would be rapidly degraded. Degradation could maintain low SIK1 levels and prevent aberrant differentiation of muscle progenitor cells. In response to muscle injury, muscle progenitors become activated and undergo myogenesis with concomitant increases in intramuscular cAMP (40), CREB activity, and Sik1 mRNA (39). SIK1 also exerts protective effects on degenerating mouse myofibers expressing dominant-negative CREB (14). It will therefore be interesting to test whether strategies to enhance SIK1 stability could promote myogenesis and muscle fiber health in degenerative conditions.

Materials and Methods

Cell Culture.

C2C12 and HEK293T cells (ATCC) were cultured in DMEM and 10% (vol/vol) FBS. Primary mouse myoblasts were isolated from neonatal ICR mice with approval by the University of Texas at Houston institutional animal care and use committee (AWC-08-125, -11-096, and -11-094) and purified by density-gradient centrifugation (39). High-density myoblasts were induced to differentiate by incubation for up to 5 d in differentiation medium [DM: DMEM and 2% (vol/vol) horse serum], when most cells have fused into myotubes.

Materials.

Lipofectamine, Plus Reagent, and DAPI (Invitrogen); PEI and anti–c-Myc affinity gel (Sigma); and Protein G-Sepharose and HisPur Cobalt Resin (Pierce) were used for experiments.

Antibodies.

Polyclonal antibodies to GST-rSIK1(aa 343–776) were raised in rabbits (approved by the Salk Institute animal care and use committee) and affinity-purified after preabsorbtion on GST. Phospho-specific antibodies to SIK1(S577) (peptide GRRA[pSer]DTSLTQGLKC) were raised and affinity-purified by Genscript. Commercial antibodies included myc (9E10) (Sigma); HSP90, MEF2(C21: pan-MEF2 antibody with highest affinity for MEF2A), and ubiquitin (Santa Cruz); phospho-HDAC(4/5/7/9) and phospho-PKA substrate (Cell Signaling); MHC(MF-20) (Developmental Studies Hybridoma Bank); and anti-mouse(DyLight-549) (Jackson Immunoresearch).

cDNA Constructs.

pGEX-HDAC5(189-573) has been described (14). pcDNA3-6xHis-Ub was a gift from Jeffrey Frost (University of Texas Medical School, Houston). cDNA encoding SIK1 from pSVL-rSIK1 (17) was cloned into the EcoRI/ XbaI sites of pcDNA3-mychisA with a stop codon mutation. Additional point mutations were introduced using site-directed mutagenesis. The SIK1 PEST domain (nucleotides 1,353–1,416) was deleted using a loop-out strategy with nested PCR primers. pACLA-RSV was created by replacing the CMV promoter in BamHI/ HindIII sites of pACLA (41) with the RSV-LTR promoter from pRc/RSV (Invitrogen). SIK1 cDNAs were cloned into AscI/ PmeI sites of pACLA-RSV. The SV40 NLS sequence (PPKKKRKV) was inserted in frame into XbaI-ApaI of pACLA-RSV-SIK1-mychis.

Cell Treatments.

Forskolin (10 µM; EMD Millipore), IBMX (1-isobutyl-3-methylxanthine, 18 µM; Sigma), CHX (cycloheximide, 50 µg/ mL; Calbiochem), ActD (actinomycin D, 10 µg/mL; Calbiochem), and isoproterenol (10 µM; Sigma) were used for cell treatments. For cAMP priming and chase, cells were pretreated with FSK/IBMX for 3 h, rinsed three times over 5 min [DMEM and 10% (vol/vol) FBS], incubated in growth medium with or without inhibitors, and harvested at time points by 2× cold PBS rinse and flash freezing in LN₂. Extracts were prepared in RIPA buffer [20 mM Hepes (pH 7.5), 137 mM NaCl, 1% (vol/vol) TX-100, 0.1% SDS, 0.5% Na-DOC, 0.5 mM EDTA, 5 mM Na₄P₂O₇, 20 mM β-glycerophosphate, and 50 mM NaF] or SDS buffer [40 mM Tris, 2% (wt/vol) SDS, 120 mM NaCl, 2 mM NaF, 1 mM Na₄P₂O₇, and 10 mM β-glycerophosphate] with 1 mM Na₃VO₄, protease inhibitors, and 10 µM MG-132.

Estimation of SIK1 Half-Life.

The intensities of SIK1 (both bands of doublet) and HSP90 loading control for each lane on the same reprobed blots were quantified in arbitrary units by densitometry on unsaturated Western blot films (ImageJ). Ratios of SIK1/ HSP90 at each time point were expressed as fold change from maximum (t = 0 for CHX chases of myc-SIK1; FSK/IBMX 3 h for endogenous) within each experiment, expressed as mean percent remaining ± SD among three or more experiments.

His Pull-Down.

Transfected HEK293T cells were treated with 10 µM MG-132 for 5 h, rinsed with cold PBS, and lysed in denaturing buffer [50 mM Na-phosphate (pH 7.4), 8 M urea, 250 mM NaCl, 1% (vol/vol) TX-100, 1 mM Na₃VO₄, 1× protease inhibitors, 10 mM NEM, and 10 µM MG-132]. Extracts were sonicated, centrifuged at 16,000 × g for 15 min at 4 °C, and tumbled with HisPur Cobalt resin for 1 h at 4 °C. Beads were washed in lysis buffer supplemented with 500 mM NaCl and 10 mM imidazole.

In Vitro Kinase Assay.

SIK1-myc-his proteins were purified from HEK293T cells by a tandem (His-myc) strategy. Recombinant GST-HDAC5 polypeptides were purified from Escherichia coli and used in kinase assays (30 °C, 20 min) with equivalent amounts of each SIK1 mutant as described (14). Recombinant PKA (NEB) was incubated with purified SIK1 in 50 mM Tris (pH 7.5), 10 mM MgCl₂ and 1 mM DTT (30 °C, 30 min). ³²P signals were quantified on autoradiograms using densitometry, corrected for background, normalized to total SIK1 (ImageJ), and expressed as mean percent labeling compared with SIK1-KD (set to 100%) ± SD.

Quantitative PCR.

Real-time PCR assays were performed on cDNA from total RNA as described (14). Amounts of Sik1 were normalized to Gapdh in arbitrary units, expressed as fold change.

Immunofluorescence.

Cells on glass coverslips were fixed [4% (wt/vol) paraformaldehyde], permeabilized (0.2% TX-100), stained (anti-myc, anti-mouse-549, DAPI), and imaged on a Nikon A1 confocal microscope (NIS-Elements software).

Statistics.

Differences between groups were evaluated using two-tailed Student t tests among replicate experiments or replicate samples, as indicated in figure legends (*P < 0.05, **P < 0.01, ***P < 0.001 for all panels).