ACTH-induced Nucleocytoplasmic Translocation of Salt-inducible Kinase

Salt-inducible kinase (SIK), a serine/threonine protein kinase expressed at an early stage of adrenocorticotropic hormone (ACTH) stimulation in Y1 mouse adrenocortical tumor cells, repressed the cAMP-responsive element (CRE)-dependent gene transcription by acting on the basic leucine zipper domain of the CRE-binding protein (Doi, J., Takemori, H., Lin, X.-z., Horike, N., Katoh, Y., and Okamoto, M. (2002)J. Biol. Chem. 277, 15629–15637). The mechanism of SIK-mediated gene regulation has been further explored. Here we show that SIK changes its subcellular location after the addition of ACTH. The immunocytochemical and fluorocytochemical analyses showed that SIK was present both in the nuclear and cytoplasmic compartments of resting cells; when the cells were stimulated with ACTH the nuclear SIK moved into the cytoplasm within 15 min; the level of SIK in the nuclear compartment gradually returned to the initial level after 12 h. SIK translocation was blocked by pretreatment with leptomycin B. A mutant SIK whose Ser-577, the cAMP-dependent protein kinase (PKA)-dependent phosphorylation site, was replaced with Ala could not move out of the nucleus under stimulation by ACTH. As expected, the degree of repression exerted by SIK on CRE reporter activity was weak as long as SIK was present in the cytoplasmic compartment. The same was true for the SIK-mediated repression of a steroidogenic acute regulatory (StAR) protein-gene promoter, which contained a CRE-like sequence at −95 to −85 bp. These results suggest that in the ACTH-stimulated Y1 cells the nuclear SIK was PKA-dependently phosphorylated, and the phosphorylated SIK was then translocated out of the nuclei. This intracellular translocation of SIK, a CRE-repressor, may account for the time-dependent change in the level of ACTH-activated expression of the StAR protein gene.

Extracellular hormonal stimuli, received at the cell surface by the corresponding receptors, are transduced into several chemical messengers in the cytoplasm and regulate the transcription of genes in the nuclei. Adrenocorticotropic hormone (ACTH), 1 binding to its receptor on the adrenocortical cell surface, activates adenylate cyclase and increases intracellular cAMP, which in turn activates the cAMP-dependent protein kinase (PKA) (1)(2)(3). The activated PKA then promptly stimulates the biosynthesis and secretion of steroid hormone. This is achieved through stimulation of several intracellular events; PKA activates cholesterol esterases (4 -7), thus increasing the intracellular cholesterol pool, and elevates the levels of transcription (8,9), mRNA stability (10), translation (8), and posttranslational modification (11)(12)(13)(14)(15) of the steroidogenic acute regulatory (StAR) protein (16). This accelerates the transport of cholesterol from the outer to inner mitochondrial membranes, which activates the side chain cleavage cytochrome P450 (CYP11A) reaction (17)(18)(19)(20)(21). The resultant elevation of the plasma glucocorticoid level acts as a negative signal on the ACTH secretion from the pituitary gland, thus completing the feedback regulatory loop of the hypothalamic-pituitary-adrenocortical axis (22,23). Several signaling molecules (24 -30) other than those mentioned above, such as phosphodiesterase (31) and protein phosphatase (32) have also been implicated in regulating steroidogenesis, indicating that the steroidogenic cells are equipped with finely tuned signal transducing systems. However, the precise mechanism underlying the acute regulation of steroidogenic gene expression, including the molecular events occurring during the initiation, maintenance, and termination of the steroidogenic gene transcription, is as yet not well understood.
Salt-inducible kinase (SIK) was identified as a serine/threonine protein kinase induced in the adrenal glands of high-salt diet-fed rats (33,34). When Y1 mouse adrenocortical tumor cells were incubated with ACTH, the cellular levels of mRNA, protein, and kinase activity of SIK were elevated within 1 h through the cAMP/PKA signaling mechanism, and then after a few hours the levels declined (35). In contrast, the transcription of StAR and CYP11A genes begin later, at a time coinciding with the decline of the SIK level. A finding that the ACTHinduced transcription of the CYP11A gene failed to occur in SIK-overexpressing cells (35) prompted us to explore further the mechanism of action of SIK. Subsequent studies (36) demonstrated that SIK repressed the PKA-induced activation of the human CYP11A gene promoter by acting on cAMP-responsive element (CRE)-binding protein (CREB) bound to the CRE (36).
The suppression of ACTH-induced transcription of the StAR gene in SIK-overexpressing cells was also noted, but the time course of its manifestation seemed different from that of the CYP11A gene. Thus, 2 h after the addition of ACTH, the level of StAR mRNA in the SIK-overexpressing cells was elevated to a similar level as that in the control cells, but the level was markedly suppressed after 12 h (35). This may imply that the time course of StAR gene expression during ACTH activation might be divided into two consecutive phases; the first in which SIK was not capable of exerting a repressive effect on the StAR gene, and the second in which SIK exerted repressive activity.
During the course of this study to explore the mechanism underlying SIK-mediated steroidogenic gene regulation, we noticed that SIK changed its subcellular location during ACTH stimulation. Before stimulation it was present both in the nuclear and cytoplasmic compartments; however, within 5 min after ACTH treatment, the nuclear SIK moved into the cytoplasmic compartment. The level of SIK in the nuclei was not recovered until 6 h post-ACTH treatment. The nuclear export of SIK occurred concomitantly with PKA-dependent phosphorylation at Ser-577 of SIK. When the phosphorylation was inhibited, SIK was retained in the nuclei. During SIK's absence from the nuclear compartment the SIK-mediated repression of CRE-reporter activity was not obvious. Experiments conducted on the human StAR gene promoter suggested that SIK repressed the StAR promoter via CREs coupled with a cAMPresponsive SF-1 site. The two consecutive phases found in the time course of StAR gene expression in SIK-overexpressing Y1 cells could be explained by SIK intracellular translocation during ACTH stimulation.
DNA fragments containing 150 and 84 bp of the human StAR promoter were prepared from pGL2 by digesting with KpnI/HindIII and ligating into the KpnI/HindIII site of pGL3. A StAR promoter fragment (Ϫ2.0 to Ϫ1.3 kbp) was amplified by PCR using the Ϫ2.0F primer (5Ј-AAAGGTACCGTGGTGGCAGGACACAGAACT) and the Ϫ1.3R primer (5Ј-GGGGTCTTGGTATGTGCAGAA) (the sequences from Gen-Bank TM accession no. AP000065), digested with KpnI/HindIII and ligated into the KpnI/HindIII site of pGL3. The DNA fragment of 1.3 kbp upstream of the transcriptional start site was prepared from pGL2 by HindIII digestion and introduced into the HindIII site of pGL3 reporter with the Ϫ2.0 to Ϫ1.3-kbp fragment. To prepare a Ϫ1.8-kbp StAR promoter construct, a Ϫ2.0 to Ϫ1.8-kbp fragment was removed from the Ϫ2.0-kbp StAR promoter construct by KpnI/AatI digestion. The mutant StAR promoter (Ϫ150 bp) with a non-functional SF-1/CRE site at Ϫ95 to Ϫ85 bp was created by site-directed mutagenesis with a mutagenic primer (5Ј-GAGGCAATCGCTCTATCTAGAACCCCTTCCTTTGC) using a site-directed mutagenesis kit, GeneEditor (Promega).
To prepare the expression vectors for GFP-SIK fusion proteins (GFPs fused to the NH 2 termini of SIKs), pEGFP-C plasmid was constructed as follows. First, pEGFP-N1 plasmid was digested with BglII/BamHI (in the multicloning site) and self-ligated, and the resultant EcoRI and BamHI sites were eliminated. Second, oligonucleotides (5Ј-GTACGGA-TCCCTGCGGCCGCTGAATTCTAA and 5Ј-GGCCTTAGAATTCAGCG-GCCGCAGGGATCC) were annealed and ligated into the BsrGI/NotI site of the above vector, and thus a new BamHI-NotI-EcoRI site with a stop codon in the 3Ј-end of the EGFP-coding region was created. cDNA fragments of full-length SIK and its mutants were cloned into the BamHI/EcoRI site of the pEGFP-C plasmid.
An overexpression vector for hemagglutinin (HA)-tagged SIK protein was prepared by modification of the pSVL vector (Amersham Biosciences). Oligonucleotides corresponding to a coding sequence of the HA tag linked to BamHI-EcoRI sites (5Ј-TCGAATGGCTTACCCATA-CGACGTCCCAGACTACGCGGGATCCCTCGAGGAATTC and 5Ј-GATCGAATTCCTCGAGGGATCCCGCGTAGTGTGGGACGTCGTAT-GGGTAAGCCTA) were ligated into the XhoI/BamHI site of pSVL, and the resultant plasmid was named pSVL(HA). SIK cDNAs were inserted into the BamHI-EcoRI site of pSVL(HA).
To construct mammalian expression vectors (pEBG-SIK peptides) of GST-fused SIK peptides, the DNA fragments encoding the SIK peptides were prepared from the Escherichia coli GST expression vector, pGEX-6P3, by BamHI/NotI digestion followed by ligation into the BamHI/NotI of pEBG.
To create CREB-bZIP expression vector, a cDNA fragment encoding the CREB bZIP domain, similar to an inhibitory CRE modulator (CREM) isoform ICRE, was amplified by RT-PCR using primers (5Ј-GGGAATTCATGGCTTCCTCCCCAGCACT and 5Ј-GGGATCCATTT-TCCACCTTAACAGGTGA) and pM-CREB(F) plasmid as a template, and it was digested with EcoRI/BamHI and ligated into the EcoRI/ BamHI of pIRES1neo vector.
Cell Culture and Reporter Assay-Y1 and COS-7 cells were maintained in Dulbecco's Modified Eagle's medium (Sigma) containing 10% fetal calf serum and antibiotics at 37 C under an atmosphere of 5% CO 2 , 95% air. The method of reporter assays was described previously (35). To introduce DNAs into cells, LipofectAMINE 2000 was used in this study.
For fluorescent microscopical observation, cells were cultured on poly-L-lysine coated coverslips (18-mm) (Matsunami Co. LTD, Tokyo, Japan) using a 12-well dish. Cells were fixed with 1 ml of 4% paraformaldehyde dissolved in PBS and washed with PBS three times. For immunostaining, cells were incubated with 1 ml of 1% bovine serum albumin in PBS for 30 min at room temperature and then reacted with anti-SIK antibody (1:1000 dilution) (35) for 1 h at 4 C. After the cells were washed with PBS four times, the antigen-antibody complexes were reacted with anti-rabbit IgG-Cy2 conjugate IgG (1:300 dilution) (Amersham Biosciences) for 1 h at 4 C. Cells on the coverslip were embedded onto a slide glass using 50% glycerol.
GST-SIK Peptide-The expression plasmids for GST-SIK peptide (pGEX-6P3-SIK (573-581)) and its mutant peptides were transformed into bacterial strain BL21 (codon plus) (Stratagene). The resultant transformants were grown in 300 ml of Terrific Broth (Sigma) containing 8% glycerol and 50 mM potassium phosphate (pH 7.4) at 28°C. When the optical density at 600 nm of the culture solution reached 1.0, recombinant proteins were induced by the addition of 0.2 mM iso-1-thio-␤-D-galactoside. The bacteria were harvested by centrifugation 3 h after the induction, and the cell pellet was suspended in 50 ml of sonication buffer (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, and 1 mM phenylmethylsulfonyl fluoride). The cells were broken by sonication, and the resultant cell suspension was mixed with Triton X-100 (final 1%), incubated at 4 C for 30 min, and subjected to centrifugation at 100,000 ϫ g for 45 min. The GST-SIK peptides were recovered in the supernatant. For purification of the peptides, glutathione-Sepharose and HiTrap Q column chromatography (Amersham Biosciences) were performed as described in Ref. 35.
In Vitro Kinase Assay-The procedure for in vivo kinase assay was described in Ref. 36. To purify GST-SIK peptide, a glutathione column (MicroSpin TM GST Purification Module, Amersham Biosciences) was used. The catalytic subunit of PKA from bovine heart was purchased from Sigma. PKA (0.5 units) was mixed with 0.5 Ci (18.5 kBq) of [␥-32 P]ATP and substrates (5 g of GST or GST-SIK peptides) in 20 l of PKA-reaction buffer (50 mM Tris-HCl (pH 7.4), 1 mM dithiothreitol, and 5 mM MgCl 2 ). Reactions were performed at 30°C for 10 min and stopped by the addition of 10 l of 3ϫ SDS sample buffer. The reaction mixture was boiled for 5 min, and proteins in 10-l aliquots were subjected to SDS-PAGE in a 15% gel. The gel was dried and exposed to an x-ray film at room temperature for 10 min. To examine the kinase activity of immunopurified HA-tagged SIK, 30 l of the samples were mixed with 10 l of 4ϫ SIK-reaction buffer (200 mM Tris-HCl (pH 7.4), 4 mM dithiothreitol, and 40 mM MnCl 2 ) containing 20 g of GST-Syntide2 and 1.0 Ci (37 kBq) [␥-32 P]ATP, and the mixture was incubated at 30°C for 1 h. After the addition of 20 l of 3ϫ SDS sample buffer, the reaction mixture was heated at 100°C for 5 min, and an aliquot (10 l) was subjected to SDS-PAGE (15%) followed by autoradiography as described above.
Preparation of Antiphospho-Ser-577-specific Antibody and Western Blotting-Phospho-SIK peptides (CQEGRRApSDTSLT, the 571-582residue peptide with an extra Cys at the NH 2 terminus: pS indicates phosphoserine), conjugated with keyhole limpet hemocyanin (KLH) or biotin utilizing the NH 2 -terminal Cys, were obtained from Sigma Genosys Co. LTD. 0.4 mg of KLH-phospho-SIK peptide was emulsified with one volume of TiterMax Gold (TiterMax USA Inc., Norcross, GA) and used to immunize Japanese white rabbits (females, 2.0 kg-bodyweight). To purify the specific immunoglobulin G (IgG), 0.5 mg of biotin-linked phospho-SIK peptide was loaded onto a 1-ml streptavidin column (HiTrap-streptoavidine, Amersham Biosciences), and the column was washed with 10 ml of PBS containing 1 M NaCl and then equilibrated with 5 ml of PBS. The antiserum (5 ml) was diluted with 5 ml of PBS, filtered with a disk filter (0.45 m), and applied onto the phospho-SIK peptide column. The column was washed with 5 ml of PBS, 5 ml of PBS containing 1 M NaCl, and the antiphospho-Ser-577 (anti-pS577) IgG was eluted with 5 ml of 150 mM glycine-HCl (pH 2.5). Almost 80% of IgG was eluted in the first 1-ml fraction; this fraction neutralized with 0.15 ml of 1 M Tris-HCl (pH 9.5).
To detect SIK protein or peptide with pS577, the samples were subjected to SDS-PAGE (10 or 15% gel, respectively) and immunoblotted onto polyvinylidene difluoride membranes, as described in Ref. 35. The membrane was first incubated with 1% bovine serum albumin for 30 min, and then incubated with the anti-pS577 antibody (diluted at 1:1000 with 4 ml of PBS) for 1 h. The antigen-antibody complexes on the membranes were reacted with goat anti-rabbit IgG-horseradish peroxidase conjugate (Cappel, Durham, NC). The peroxidase conjugate was visualized with a Konica immunostaining kit (Konica, Tokyo, Japan).

ACTH-dependent Intracellular Translocation of SIK-By us-
ing an anti-SIK antibody we examined the subcellular location of SIK in resting and ACTH-stimulated Y1 cells. In the resting cells a weak signal for the presence of SIK (stained green) was found in both the nuclear and cytoplasmic compartments (Fig.  1A). After ACTH treatment the cells looked circular and had an intense signal only in the cytoplasmic compartment. No significant staining occurred with preimmune serum (data not shown). Because ACTH enhanced the biosynthesis of SIK protein in Y1 cells (35), the immunochemically stained signal after the stimulation must have resulted from the sum total of the SIK present before ACTH treatment and that newly synthesized in the cytoplasm after the treatment. Therefore, when the intracellular translocation of SIK is considered, the immunocytochemical approach may cause difficulty in the interpretation of the results. To overcome this difficulty, GFP-tagged SIK protein was expressed in Y1 cells, and the subcellular translocation of green fluorescence signal was followed after the addition of ACTH. As shown in Fig. 1B, the fluorescence signal was present in both the nuclear and cytoplasmic compartments of resting cells. The nuclear signal formed speckles, the size of respective speckle appearing to be negatively correlated with the number of speckles in a nucleus (data not shown). When the cells were treated with ACTH, the green fluorescence signal seemed to move out of the nuclei and was diffusely distributed in the cytoplasm. When the cells were pretreated with LMB, an inhibitor of CRM1-dependent nuclear export (39 -41), the fluorescence signal in the resting cell seemed to be more concentrated in the nucleus, and this localization did not change after ACTH treatment (Fig. 1C). When GFP tag alone was introduced into Y1 cells, the fluorescence signal did not localize at any specific subcellular area, and its distribution did not change after the ACTH addition (data not shown). To answer whether the cAMP/PKA signaling was involved in the intracellular redistribution of SIK, Y1 cells and PKA-less mutant Y1, Kin-7, cells (38) were treated with 8-Br-cAMP (Fig. 1D). cAMPstimulated the nuclear export of GFP-SIK in Y1, but not in Kin-7, cells. Moreover, in PKA-overexpressing Kin-7 cells GFP-SIK was exclusively present in the cytoplasm. These results indicated that ACTH induced the nuclear export of SIK through the activation of cAMP/PKA signaling.
PKA-dependent Phosphorylation of Ser-577 in SIK and Its Implication in Subcellular Localization of SIK-There are three consensus PKA-dependent phosphorylation motifs (R/ K)(R/K)X(S/T) in SIK protein; phosphorylatable amino acid residues, Thr-268, Thr-475, and Ser-577, were found in the respective motifs ( Fig. 2A). Thr-268 exists in the NH 2 -terminal kinase domain, SIK(N), a peptide from the NH 2 terminus to residue 343, that possessed the kinase activity as well as the activity to repress CRE-reporter transactivation. We previously demonstrated that SIK(N) was not phosphorylated by PKA during ACTH stimulation (36), and therefore, Thr-268 might not be the target residue of PKA. To examine whether or not the ACTH-induced nuclear export of full-length SIK protein was somehow related to PKA-dependent phosphorylation, GFP-linked SIK and GFP-linked SIK-mutants harboring mutations in the PKA phosphorylation motifs, SIK(T268A), SIK(T475A), and SIK(S577A), were expressed in Y1 cells. SIK(T268A) and SIK(T475A), both present in the nuclei of resting cells, moved out of the nuclei after stimulation by ACTH in a similar manner as the wild-type SIK. On the other hand, SIK(S577A) was present only in the nuclei of resting cells and did not move following ACTH treatment (Fig. 2B). These results indicated that Ser-577 was important for the nuclear export of SIK protein, and this amino acid residue might be phosphorylated by PKA in the ACTH-stimulated cells.
In order to evaluate the extent of phosphorylation of Ser-577 in SIK, we raised a polyclonal antibody directed against a phospho-Ser-577(pS577)-SIK-peptide (571-582-amino acid residue peptide). The immunospecificity of the antibody was tested by immunoblot analyses of a GST-fused SIK-peptide and its mutant (Fig. 2C). Anti-pS577 antibody reacted with the GST-SIK-peptide (573-581) only when it had been preincubated with ATP in the presence of PKA. GST and GST-linkedmutant SIK-peptide(S577A) were not recognized by the antibody.
By using the anti-pS577 antibody, we designed experiments to test whether Ser-577 in SIK was PKA-dependently phosphorylated in cells, and, if so, what level of kinase activity the phosphorylated SIK possessed. COS-7 cells were used because their efficiency for gene transfection was generally high, and because the green fluorescence signal in GFP-SIK-overexpressing COS-7 cells, like that in GFP-SIK-overexpressing Y1 cells, was localized both in the nuclear and cytoplasmic compartments. The fluorescence signal moved out of the nuclei when the cells were treated with 8-Br-cAMP (shown in Supplemental Materials, Fig. 1). COS-7 cells were transformed with HAtagged SIK expression vector and treated with or without 8-Br-cAMP, and SIK in the cell lysates was purified by anti-HA antibody and subjected to immunoblot analysis and in vitro kinase assay (Fig. 2D). The anti-pS577 antibody slightly reacted with SIK prepared from the pretreated cells, suggesting that an endogenously active protein kinase(s) present in COS-7 cells could phosphorylate a small portion of SIK. The signal for the phosphorylated SIK was enhanced by the 8-Br-cAMP treatment, indicating that SIK was PKA-dependent phosphorylated at Ser-577. As expected, the anti-pS577 antibody did not react with SIK(S577A). When evaluated by the autophosphorylation activity or GST-Syntide2 phosphorylation activity, the kinase activities of the wild-type and mutant SIKs were similar to one another, irrespective of their state of phosphorylation (shown in the lower panel of Fig. 2D).
The above findings strongly suggested that the PKA-mediated phosphorylation of Ser-577 in SIK was important for the ACTH-stimulated SIK nuclear export. This was confirmed by using other SIK mutants whose PKA recognition motifs had been disrupted by replacing the Arg residues upstream of Ser-577 with Ala (Fig. 3A). First, we asked whether or not the Ser-577 in the Arg-mutated peptides could be phosphorylated by PKA. When GST-fused SIK-peptide (573-581) and the mutant peptides were subjected to in vitro kinase assays, not only SIK-peptide(S577A) but also SIK peptides with mutations at Arg residues, though these contained the intact Ser-577, were not phosphorylated (Fig. 3B). Second, we attempted to conduct the in vivo kinase assays on the Arg-mutated full-length SIK proteins to see whether Ser-577 in the Arg-mutated proteins could be phosphorylated by PKA in Y1 cells. However, multiple phosphorylation sites present in the full-length SIK made it difficult to discern the level of specific phosphorylation at Ser-577 (data not shown). Therefore, Y1 cells that had been transformed with expression vectors for the above-mentioned GSTfused SIK-peptides were prelabeled with 32 PO 4 and stimulated with or without ACTH (10 Ϫ6 M) for 15 min. The levels of phosphorylation of the peptides were visualized by autoradiography (Fig. 3C). The wild-type peptide was phosphorylated in resting Y1 cells, and the level of phosphorylation was elevated by ACTH treatment. As expected, the phosphorylation could not be seen for the S577A peptide. The R575A peptide could be phosphorylated in the presence and absence of ACTH to the similar extent. The weak phosphorylation could be seen for the mutant peptides R574A and R574A/R575A. Finally, the Argmutated full-length SIK proteins were fused to GFP, and they were expressed in Y1 cells (Fig. 3D). In contrast to GFP-SIK(S577A), the Arg-mutated SIKs, GFP-SIK(R574A), GFP-SIK(R575A), and GFP-SIK(R574A/R575A), were localized both in the nuclear and cytoplasmic compartments of resting cells; rather, the fluorescence signal for GFP-SIK(R575A) in the cytoplasm seemed more intense than that in the nuclei. The ACTH treatment of cells expressing these Arg-mutated SIKs did not change the subcellular distribution of fluorescence signal. Taken together, these results suggested that the PKAmediated phosphorylation of Ser-577 was indeed important for SIK to move out of the nuclei to the cytoplasm in the ACTHstimulated cells. However, considering that the Arg-mutated SIKs, the phosphorylation of which could not be detected after the incubation with PKA, were nevertheless present outside the nuclei, we surmised that Ser-577 in these mutant SIKs might be phosphorylated by the other protein kinase in the resting cells, and the phosphorylated SIKs were exported out of the nuclei. (This possibility remains to be tested in the future.) Nuclear SIK Represses CRE Reporter Activity-Considering the above finding that SIK changed its subcellular location during ACTH stimulation, together with our prior report that the overexpression of SIK in Y1 cells repressed the PKA and CREB-mediated CRE-activation (36), we surmised that SIK intracellular location must be crucial for its capability to suppress the transcriptional activation of CRE-dependent genes.
To ascertain this point, we asked whether the temporal level of nuclear SIK during ACTH stimulation could well account for the temporal level of transcription repression exerted by SIK; hence, we designed the following experiments. First, population of cells with GFP-SIK in their nuclei was determined at several time points following the addition of ACTH (Fig. 4A). The number of cells containing nuclear SIK rapidly decreased and within 5 min reached 10% of that at the start, confirming the finding in Fig. 1 that most SIK in the stimulated cells was localized in the cytoplasm. SIK seemed to be retained in the cytoplasm for a few hours, after which it gradually reentered the nuclei; hence the number of cells with a nuclear fluorescence signal reached 30% after 4 h, and then 70% after 12 h. Secondly, HA-SIK-expressing Y1 cells were treated with ACTH, and at several time points the levels of pS577-SIK in cell lysates were estimated by immunoblot analyses (Fig 4B). A low level of phosphorylated SIK was present even before ACTH treatment, suggesting that the basal PKA or other protein kinase(s) could phosphorylate Ser-577 in the resting cells. ACTH stimulated the phosphorylation of Ser-577, the level of pS577-SIK being clearly elevated within 15 min, reaching the maximum after 30 min. Then, the level began to gradually decline until the 4-h point. The results in Fig. 4, A and B, taken together, suggested that, while most SIK in the nucleus seemed not phosphorylated in the resting cells, it would be PKA-dependent phosphorylated after ACTH stimulation, and the phosphorylated SIK would move out of the nucleus. Then, the phosphorylated SIK would be dephosphorylated in the cytoplasm after several hours and reenter the nucleus.
Based on these results, we reasoned that SIK capability to suppress ACTH-dependent transcriptional activation should appear more strongly during the period when most SIK existed in the nucleus than the period when it was present in the cytoplasm. To ascertain this point, Y1 cells, transfected with the CRE reporter plasmid and SIK expresssion vector or empty vector, were stimulated with ACTH, and the time course of reporter expression was determined. In the absence of SIK, the reporter activity was significantly activated within 3 h after ACTH treatment, reached the maximum at the 6-h point, and then decreased at the 12-h point (white bars in Fig. 4C). In the SIK-co-expressing cells, the ACTH-induced CRE activity (gray bars in Fig. 4C) seemed to be only slightly repressed 3-6 h post-treatment, the time when most of the expressed SIK should be present in the cytoplasm; thus, the base line-subtracted CRE activities in SIK-co-expressing cells was about 80 -90% of that without SIK. And then, at 12 h after the ACTH addition, the time when SIK should have reentered the nucleus (see Fig. 4A), the extent of SIK-mediated repression became more prominent, so the CRE activity in SIK-co-expressing cells was only 20% of that without SIK. It should be noted in particular that when the nuclear resident mutant SIK, SIK(S577A) (see Fig. 2), was co-expressed instead of wild-type SIK, the CRE activity was completely suppressed at any time of incubation with ACTH (black bars in Fig. 4B), suggesting that SIK presence in the nucleus was necessary for its transcription repression activity. The importance of the nuclear presence of unphosphorylated SIK for its capability of suppressing the PKA-dependent CRE activation was further verified by using the Arg-mutated SIKs (Fig. 4D). As expected, the PKA-insensitive mutants, SIK(S577A), SIK(R574A), and SIK(R574A/ R575A), repressed PKA-induced CRE activity, probably because these SIKs could not be phosphorylated and therefore stayed in the nuclei. SIK(R575A), whose Ser-577 seemed constitutively phosphorylated and therefore predominantly present in the cytosol (Fig. 3C), also strongly repressed PKA-induced CRE activity. This suggests that a small amount of SIK(R575A) seen in the nucleus with or without ACTH treatment (Fig. 3D) was enough to exert the suppression of gene expression.
In light of the above findings, we conceived that the following events consecutively took place in the stimulated cells: SIK, present in the nuclei of resting cells and acting as a repressor of the CRE-dependent gene transcription, would be phosphorylated at Ser-577 by PKA immediately after the addition of ACTH, the resultant pS577-SIK would translocate to the cytoplasm. The diminishing level of the repressor in the nuclei would trigger the initiation of CRE-dependent gene transcription. After a certain time pS577-SIK in the cytoplasm would be dephosphorylated, and SIK would reenter the nuclei and block transcription.
The Basic Leucine Zipper (bZIP) Domain of CREB Is the Target of Nuclear SIK-By using several Gal4-CREB chimeric reporter systems, we previously demonstrated that an intact bZIP domain of CREB was important for SIK(N), the NH 2 -terminal kinase domain, to exert its repressive action on the PKA-induced CRE activity (36). In this study we confirmed the importance of the bZIP domain for the repressive effect of full-length SIK. As shown in Fig. 5A, the full-length SIK seemed to only slightly repress the PKA-dependent elevated CREB activity. That the extent of repression by the full-length SIK was weaker than that by SIK(N) (35) could be explained by SIK translocation to the cytoplasm after PKA-dependent phosphorylation of the full-length SIK. Thus, when the effect of the nuclear resident SIK, SIK(S577A), was tested, it exerted prominent repression. Furthermore, this repression was abolished when the kinase activity of SIK was disrupted (see the result given by SIK(K56M/S577A). When a leucine zipper-mutated CREB, CREB(L311A/L318A), was used for the chimeric reporter assays, even the overexpression of nuclear resident SIK(S577A) failed to repress the PKA-dependent elevated CREB(L311A/L318A) activity (Fig. 5B). Similar results were obtained when the bZIP-less Gal4-CREB was used for reporter systems (data not shown). These results suggested that the nuclear SIK repressed PKA-induced activation of CREB by acting on the CREB bZIP domain.

Time Course of ACTH-stimulated StAR Gene Expression Could Be Explained by Nucleocytoplasmic Translocation of SIK-
We previously demonstrated that the transcription response of StAR and CYP11A genes to ACTH stimulation was weaker in the full-length SIK-overexpressing Y1 cells than in the control Y1 cells (35). A particular interest of note was the time-dependent change in the level of StAR mRNA (Fig. 6A). At 3 h after cAMP stimulation the level of StAR mRNA in the SIK-overexpressing cells appeared to be similar to that in the control cells. However, after 12 h the level in the SIK-expressing cells was clearly repressed compared with that in the control cells. We asked whether the time-dependent change in the StAR mRNA level of cAMP-stimulated SIK-expressing cells could well reflect the time-dependent change in the intracellular location of SIK. This point was tested by experiments designed similarly to those described in Fig. 4C.
Before doing that, we had to determine the CRE site in the StAR promoter that might mediate the SIK-dependent repression, because the human StAR promoter seems to have two CRE-containing regions (GenBank TM accession no. AP000065). One is a region near Ϫ2 kbp from the initiation site, where an SF-1 binding site (AAGGTCA) and a CRE-like site (TTC-CGTCA) are present in tandem, with a 16-nucleotide stretch between them (Fig. 6B). This CRE-like site could bind CREB as shown in Supplementary Fig. 2. The tandem alignment of SF-1 and CRE-like sequences well resembles that in the CYP11A distal cAMP enhancer responsible for both PKA and SIK actions (36). On the other hand, a recent report (42) showed that mouse StAR gene expression was regulated by CREB/ATF family transcription factors that acted on three half-CRE (TGAC) sequences at Ϫ110 to Ϫ67 bp. Among the three half-CREs the one at the 5Ј-side shares a T, corresponding to the Ϫ88th of the human StAR promoter, with an SF-1 binding site (TATCCTT) (see Fig. 6C). To determine which CRE site, the distal CRE at Ϫ2 kbp or the proximal CREs at Ϫ110 to Ϫ67 bp, is indeed responsible for SIK action, various sizes of the promoter were constructed, linked to the luciferase reporter, and introduced into Y1 cells with or without the SIK(N) and PKA expression vectors (Fig. 6B). The response of the 1.8-kbp-long promoter to the addition of PKA was one-third less than that of the 2.0-kbp-long promoter. However, the addition of SIK(N) appeared to repress the promoter activities to the similar extent, suggesting that, although the region between Ϫ2.0 and Ϫ1.8 kbp contained an important element(s) for full activation by PKA, it might not contain the site for SIK action. When the 150-bp-long promoter was used, the SIK-mediated repression appeared most prominently, and this SIK effect was abolished by removing a region between Ϫ150 and Ϫ84 bp. These results suggested that the site responsible for SIK action was the  4). C, Y1 cells were transformed with reporters (wild-type (on the left) and SF-1/CRE mutant (on the right), StAR promoter plasmids (0.2 g) and pRL-SV40, internal control, (0.03 g)), several SIK-expression vectors (0.2 g) (wild-type SIK (WT), nuclear resident SIK(S577A), kinase defective mutant SIK(K56M/ S577A) or empty expression vectors (pIRES)) and PKA expression vector (0.1 g). After the 15-h incubation, the cells were harvested for measurements of luciferase activities (n ϭ 3). D, Y1 cells were transformed with 0.2 g of reporter constructs (pGL3-StAR(Ϫ150) or pGL3), 0.2 g of SIK expression vectors (SIK WT) or empty expression vectors (IRES) and 0.03 g of pRL-SV40, internal control. After 12 h, ACTH (10 Ϫ6 M) was added to the cells, and they were harvested after 3 or 12 h to measure the luciferase activities by the Dual-Luciferace Reporter Assay System. Transformation efficiencies were corrected by Renilla luciferase activities. The specific promoter activities of StAR were expressed as foldexpression of the reporter activity of empty vector, pGL3. Means and S.D. were indicated (n ϭ 4). E, Y1 cells were transformed with reporter constructs (pGL3-StAR(Ϫ150) or pGL3] (0.2 g), nuclear resident SIK-expression vector (0.1 g), PKA expression vector (0.2 g), dominant negative CREB expression vectors (indicated amounts) and pRL-SV40, internal control, (0.03 g). After the 15-h incubation, the cells were harvested for measurements of luciferase activities (n ϭ 3). proximal CRE site. To further confirm that SIK exerts its repressive action on the proximal CRE site, we tested the effect of mutated SIKs on the transcriptional activities of the 150-bplong StAR promoter and its SF-1/CRE-disrupted mutant (Fig.  6C). Although the wild-type SIK weakly repressed the PKAinduced StAR promoter activity, the nuclear resident SIK(S577A) strongly repressed the promoter activity, indicating that SIK presence in the nuclei was essential for transcription repression. This was confirmed by the disappearance of repression by the inactive nuclear resident SIK, SIK(K56M/ S577A). When the SF-1/CRE site in the StAR promoter was disrupted, the basal StAR promoter activity significantly decreased, and the addition of PKA could not raise the activity. Thus, the SIK repressive effect could not precisely be determined for the SF-1/CRE-disrupted promoter. Now that the region responsible for SIK action was identified, we examined the time course of ACTH-dependent elevation of the 150-bp-long StAR promoter activity in the SIKcotransfected, or the control, Y1 cells (Fig. 6D). At 3-h post-ACTH, the time when most of the expressed SIK should be present in the cytoplasm, no difference in the promoter activity was found between the SIK-transfected cells and control cells. However, at 12-h post-ACTH, when SIK should have reentered the nucleus, the StAR promoter activity in the SIK-expressing Y1 cells was about 40% lower than that in the control cells. Similar results were obtained with a longer StAR gene promoter (2-kb-long) (data not shown). Taken together, these results suggest that the nucleocytoplasmic translocation of SIK and its action on the CRE site in the StAR gene promoter could explain the time course of StAR gene expression at the early phase of ACTH stimulation.
As to the mechanisms underlying SIK action on the CRE site, we previously suggested that SIK, by acting on the CREB bZIP domain, might inhibit the transcriptional activity of CREB on the CRE (36). To verify this in the case of the StAR gene promoter, we constructed two dominant negative CREBs and tested their effects on the PKA-stimulated StAR promoter activity in the presence of the nuclear resident SIK(S577A) (Fig. 6E). CREB-AD is a molecule composed of only the transcription activation domain of CREB, and thus, when expressed in the reporter assay systems, would compete with endogenous CREB, and, therefore, inhibit the formation of functional transcription complex. The other dominant negative CREB, CREB-bZIP, contains only the CREB bZIP domain, and, similarly to a CREM isoform ICER, it would occupy the CRE site on the promoter and possibly inhibit the SIK effect on the bZIP domain by absorbing bZIP domain binding factors. When 500 ng of CREB-AD plasmid was co-transfected with the reporter assays systems, CREB-AD, by dominant-negatively acting on the transcription complex, inhibited the PKA-dependent activation by 40%, and the SIK(S577A)-dependent repression occurred to a similar extent in the presence or absence of CREB-AD. In contrast, 50 ng of CREB-bZIP plasmid seemed enough to acquire about 30% inhibition of PKA-dependent activity. Under this condition, CREB-bZIP also reduced the SIK(S577A)-mediated repression of PKA-induced StAR promoter activity. When 500 ng of CREB-bZIP plasmid was added, the StAR promoter activity was completely abolished irrespective of the presence of PKA or SIK(S577A), as in the case for the SF-1/CRE-disrupted promoter. These results suggested that the bZIP domain was indeed an important part of CREB for SIK action on the StAR promoter. DISCUSSION The present study demonstrated that SIK was localized both in the nuclear and cytoplasmic compartments of Y1 cells, but when the cells were stimulated by ACTH the nuclear SIK rapidly moved to the cytoplasmic compartment through the action of cAMP/PKA signaling. The results of site-directed mutagenesis of SIK protein and immunochemical analyses using pS577-specific IgG showed that the PKA-dependent phosphorylation of Ser-577 in SIK occurred during ACTH stimulation (Fig. 2D). The Ser-577-disrupted mutant SIK, SIK(S577A), was present in the nuclei of Y1 cells, irrespective of the ACTH stimulation (Fig. 2B). SIK with mutations at one of the two Arg residues upstream of Ser-577 (Ser-577 being resistant to PKAdependent phosphorylation perhaps because of the steric hindrance introduced in their PKA recognition motifs) could not move out of the nuclei even under ACTH stimulation. This suggested that the PKA-dependent phosphorylation of Ser-577 was necessary for SIK nuclear export. However, it is worth noticing that in the unstimulated Y1 cells a substantial amount of SIK was present in the cytoplasmic compartment (Fig. 1B); in the PKA-less mutant Kin-7 cells the cytoplasmic presence of SIK was noted (Fig. 1D). Taking into account that the immunoblot analysis revealed the presence of a substantial amount of phosphorylated SIK in the unstimulated Y1 cells (Fig. 4B), we conjecture that in the resting cells a portion of SIK had been phosphorylated at Ser-577 in a PKA-independent manner, the resultant phosphorylated SIK being pumped out of the nucleus into the cytoplasm.
Numerous proteins are phosphorylation-dependently exported from the nuclei. Phospho-Ser-containing motifs, R(S/ Ar)XpSXP or RX(S/Ar)XpSXP (Ar: aromatic), in the nuclear proteins, such as class II histone deacetylases (43)(44)(45), Cdc25C (46,47), and transcription factor FKHRL1 (48,49), were thought to be recognized by dimers of the 14-3-3 family proteins and form a complex; the resultant complex composed of the protein and 14-3-3 would move out of the nucleus into the cytoplasm (50). Hence, the nature of protein kinases that phosphorylate the Ser residue in the 14-3-3 binding motif of the nuclear protein has been a recent research interest. AMPactivated protein kinase (AMPK) phosphorylate a Ser residue, Ser-89, in the 14Ϫ3Ϫ3 binding motif of p300 and inhibited p300-mediated transcriptional co-activation (51). c-TAK1, a member of AMPK family, phosphorylated Ser-216 in the 14-3-3 binding motif of Cdc25C, initiating the nuclear export of Cdc25C (52). WPK4, a wheat protein kinase that also belongs to the AMPK family, phosphorylated Ser-388 and Ser-418 in its own COOH-terminal domain; the phosphorylated WPK4 then formed a complex with TaWIN1, wheat 14-3-3 protein. When WPK4 phosphorylated its substrate nitrate reductase, TaWIN1 bound to WPK4 was transferred to the nitrate reductase; accordingly, the nitrate reductase became inactive (53). Viewing these previous reports, we asked whether or not SIK, a member of the AMPK family, could phosphorylate its own Ser-577. When SIK(N) was incubated with a peptide with a sequence around Ser-577 in the presence of ATP, the peptide was not phosphorylated (data not shown). In addition, the level of autophosphorylation of the full-length SIK seemed similar to that of SIK(S577A) (Fig. 2). These results suggested that SIK could not phosphorylate its own Ser-577. Therefore, we at present surmise that SIK, phosphorylated at Ser-577 by PKA, or possibly by a serine/threonine protein kinase(s) other than SIK, would be recognized by a carrier protein(s) like the 14-3-3 protein and form a complex, the resultant complex moving out of the nucleus.
A protein destined for the nuclear localization contains a unipartite, or bipartite, basic amino acid cluster, such as KKKRK in SV40 T-antigen (54) and RKR-Xn-RKRKR in Tcell protein tyrosine phosphatase (55), which is recognized by an importin ␣/␤ heterodimer and acts as the nuclear localization signal (56,57). Such a basic amino acid cluster was not found in the primary structure of SIK, suggesting that a nonclassical form of nuclear localization signal is functioning for the nuclear import of SIK. On the other hand, CRM1 (importin ␤ homolog), the target of LMB, recognizes a leucine-rich nuclear export signal of a nuclear protein, and carries the protein into the cytoplasm (58,59). A hydrophobic stretch, LGLNKIKGL (residues 602-610), which resembles one of the best characterized nuclear export signals in human immunodeficiency virus-1 Rev proteins (60), is present near Ser-577 of SIK. SIKs harboring mutations at Leu and Ile residues in this stretch, when introduced into Y1 cells, were present in the nuclei at low levels, and they could move out of the nuclei normally. 2 These results suggest that the motifs involved in the nuclear and cytoplasmic localization of SIK still remain to be characterized.
The transcriptional repression activity of exogenous SIK in ACTH-stimulated cells changed in a time-dependent manner (Fig. 4C); SIK could repress the ACTH-induced CRE activity after 12 h, but not before. The time-dependent change in SIK repressive activity seemed to well reflect that in the nuclear level of SIK (Fig. 4, A and B). Consistent with this point was a fact that the nuclear resident SIK(S577A) was always able to strongly repress the CRE activation (Fig. 4D). The level of phosphorylation of proteins in ACTH-treated adrenocortical cells is known to rapidly rise and then decline concomitantly with activation of protein phosphatases by PKA (32,(61)(62)(63). When the PKA expression vector instead of ACTH was used as the stimulant, the wild-type SIK could not efficaciously repress the CRE and CREB-mediated activation (Figs. 4D and 5A). This suggested that PKA might activate the cytosolic protein phosphatase(s), which in turn dephosphorylated the phospho-Ser-577 of cytoplasmic SIK and accelerated the reentry of SIK into the nucleus.
The SIK-overexpressing pIRES-SIK1 cells, in comparison with the parent Y1 cells, responded to ACTH by raising StAR mRNA to the similar level at the 2-h point, but after 12 h the level in the SIK-overexpressing cells was clearly repressed (35). The same was true when the cells were treated with 8-Br-cAMP (Fig. 6A). The similar results were obtained by the cotransfection assays with a human StAR promoter-linked reporter and SIK-expression vector; thus, the SIK repression activity was non-significant at the 3-h point, but significant at the 12-h point, after ACTH stimulation (Fig. 6D). As for the site conferring cAMP responsiveness on the StAR gene, several investigations have been reported; accordingly, co-operative interaction between SF-1 bound to the proximal SF-1 site (9,37) and other transcriptional factors bound to neighboring sites, such as the CCAAT/enhancer site (64,65) and GATA-4 site (66), have been thought to give the full activation of cAMPdependent as well as basal StAR promoter activity (67,68). Recently, SF-1 and CREB, or transcriptional factors belonging to its family, were implicated in the cAMP-activated mouse StAR promoter activity (42). Considering that the proximal SF-1 sequence and a half of the consensus CRE sequence overlapped one another, we prepared human StAR promoter reporters with the wild-type and mutant SF-1/CRE sequences (Fig. 6C). The PKA-activated StAR promoter activity was repressed only by the nuclear resident SIK(S577A) but not by the wild-type SIK. When the SF-1/CRE sequence in the promoter was disrupted, the StAR promoter did not respond to PKA or SIK. Based on these results we may conclude that StAR mRNA in the ACTH-stimulated SIK-overexpressing Y1 cells was induced by CREB, or one of its family factors; bound to the CRE site; and the extent of the mRNA expression was influenced by the nucleocytoplasmic translocation of SIK. Consistent with our previous report (36), the assays using Gal4-CREB chimeric reporters indicated that the nuclear resident SIK(S577A) could down-regulate CREB transactivation function by acting on the bZIP domain (Fig. 5). Overexpression of the bZIP domain of CREB suppressed not only PKA-induced activation, but also SIK-mediated repression, of StAR promoter activity (Fig. 6E). The role played by the bZIP domain in the transcription complex and SIK action on the complex could be conceived based on the recent findings by Hong et al. (69) that a component of the large subunit of replication factor C, RFC140, interacted with the bZIP domain of CCAAT/enhancerbinding protein ␣ and up-regulated the transcriptional activity. Thus, it is possible that SIK might modulate the interaction between the CREB bZIP domain and the other bZIP binding factors. This possibility must not exclude another possibility that SIK-mediated repression of CREs might also depend on a neighbor sequence, because SIK failed to repress the PKA-dependent enhancer activity of the CRE site at Ϫ2.0 kbp.
Contrasting with the StAR gene, the CYP11A gene was not expressed for the first 6 -8 h following ACTH treatment (35). The difference in the time course of ACTH response between the two genes could be explained by the fact that, for robust activation of most natural promoters, binding of not a single, but multiple, transcription factors would be necessary. Furthermore, the time courses of steroidogenic gene expression after ACTH stimulation differ among the cell types. For instance, bovine adrenocortical primary culture cells (70) and human adrenocortical carcinoma cell line H295R (71) showed quite different time courses in their steroidogenic gene expression. This suggests that in the early phase of ACTH/cAMP signal transduction several regulators, in different combinations depending on the cell types, are involved in gene expression. Therefore, these regulators, one of them being SIK, should be further characterized before the full understanding of steroidogenesis.