Silencing the constitutive active transcription factor CREB by the LKB1-SIK signaling cascade Yoshiko Katoh 1 , Hiroshi Takemori 2 , Xing-zi Lin 1 , Mitsuhiro Tamura 1 , Masaaki Muraoka 3 , Tomohiro Satoh 3 , Yuko Tsuchiya 4 ,LiMin 1 , Junko Doi 5 , Akira Miyauchi 3 , Lee A. Witters 6 , Haruki Nakamura 4 and Mitsuhiro Okamoto 7 1 Molecular Physiological Chemistry, Osaka University Medical School, Japan 2 Laboratory of Cell Signaling and Metabolism, National Institute of Biomedical Innovation, Osaka, Japan 3 ProteinExpress Co. Ltd, Chiba, Japan 4 Institute for Protein Research, Osaka University, Japan 5 Food and Nutrition, Senri Kinran University, Osaka, Japan 6 Departments of Medicine and Biochemistry, Dartmouth College, Hanover, NH, USA 7 Department of Food and Nutrition, Tezukayama University, Nara, Japan Cyclic AMP-responsive element (CRE)-binding protein (CREB) is a transcription factor that plays an import- ant role in numerous physiological events, such as cell proliferation, survival, tumorigenesis, glucose metabo- lism and memory, in a phosphorylation-dependent manner [1,2]. Upstream signals arriving at CREB are Keywords cAMP responsive element; CRE-binding protein; LKB1; salt-inducible kinase; transducer of regulated CREB activity Correspondence H. Takemori, Laboratory of Cell Signaling and Metabolism, National Institute of Biomedical Innovation, 7-6-8, Asagi, Saito, Ibaraki, Osaka, 567-0085, Japan Fax: +81 72 641 9836 Tel: +81 72 641 9834 E-mail: takemori@nibio.go.jp (Received 16 January 2006, revised 19 April 2006, accepted 24 April 2006) doi:10.1111/j.1742-4658.2006.05291.x Cyclic AMP responsive element (CRE)-binding protein (CREB) is known to activate transcription when its Ser133 is phosphorylated. Two independ- ent investigations have suggested the presence of Ser133-independent acti- vation. One study identified a kinase, salt-inducible kinase (SIK), which repressed CREB; the other isolated a novel CREB-specific coactivator, transducer of regulated CREB activity (TORC), which upregulated CREB activity. These two opposing signals are connected by the fact that SIK phosphorylates TORC and induces its nuclear export. Because LKB1 has been reported to be an upstream kinase of SIK, we used LKB1-defective HeLa cells to further elucidate TORC-dependent CREB activation. In the absence of LKB1, SIK was unable to phosphorylate TORC, which led to constitutive activation of CRE activity. Overexpression of LKB1 in HeLa cells improved the CRE-dependent transcription in a regulated manner. The inactivation of kinase cascades by 10 nm staurosporine in LKB1-posit- ive HEK293 cells also induced unregulated, constitutively activated, CRE activity. Treatment with staurosporine completely inhibited SIK kinase activity without any significant effect on the phosphorylation level at the LKB1-phosphorylatable site in SIK or the activity of AMPK, another tar- get of LKB1. Constitutive activation of CREB in LKB1-defective cells or in staurosporine-treated cells was not accompanied by CREB phosphoryla- tion at Ser133. The results suggest that LKB1 and its downstream SIK play an important role in silencing CREB activity via the phosphorylation of TORC, and such silencing may be indispensable for the regulated activa- tion of CREB. Abbreviations A-loop, activation loop; AMPK, AMP-activated protein kinase; bZIP, basic leucine zipper domain; CRE, cAMP-response element; CREB, CRE-binding protein; DAPI, 4¢,6-diamidino-2-phenylindole; GFP, green fluorescent protein; GST, glutathione-S-transferase; HA, hemagglutinin; KID, kinase-inducible domain; moi, multiplicities of infection; PKA, protein kinase A; RT, reverse transcription; SIK, salt-inducible kinase; TORC, transducer of regulated CREB activity. 2730 FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS conveyed to transcriptional machineries via two distinct domains of CREB. The kinase-inducible domain (KID) is located in the N-terminal region, which contains an activating phosphoacceptor residue, Ser133. The other domain in the C-terminus is composed of a basic leu- cine zipper (bZIP) that is responsible for dimerization and binding to the CRE. Phosphorylation of Ser133 alters the affinity of KID to the KIX domain of CREB and p300, resulting in enhanced transcription of CRE- dependent genes [3]. Development of a specific anti- body that recognized phospho-Ser133 [4,5] enabled investigators to monitor the level of ‘activated CREB’, which has now significantly accelerated the studies of phosphorylation-dependent CREB activation. Possible involvement of the bZIP domain of CREB in the regulation of CRE-dependent gene expression has been suggested by the results of two lines of research. One was initiated by an mRNA subtraction study to isolate a specific molecule induced in the adre- nal gland under the stress of consuming a high-salt diet. The molecule isolated was a kinase, and, thus, we named it salt-inducible kinase (SIK) [6]. SIK is a mem- ber of the AMP-activated protein kinase (AMPK) fam- ily [7]. A gene database search found three isoforms of SIK, SIK1–3 [8,9]. In Y1 mouse adrenocortical tumor cells, levels of mRNA, protein and kinase activity of SIK1 were elevated within 30 min after the initiation of the cAMP–protein kinase A (PKA) cascade. Overex- pression of SIK1 inhibited gene expression(s) induced by cAMP [10]. Analyses of the promoter regions of such genes indicated that CREs in the promoters were the sites where SIK-mediated transcriptional repression occurred, and, thus, SIK1 was thought to repress CREB activity [11]. Although SIK seemed not to phos- phorylate CREB directly, it repressed CREB in a kinase-activity-dependent manner. Mapping the region where SIK1 exerted its repressive action suggested that SIK1 repressed CREB by acting on the bZIP domain [11]. In reporter gene assays, overexpression of the kin- ase domain of SIK1 repressed CRE-dependent tran- scription completely, even when CREB was supposed to be fully activated by overexpression of PKA. We therefore thought that the phosphorylation of CREB at Ser133 was not sufficient for making ‘activated CREB’. The second line of research began in an attempt to isolate novel factors that could modulate CREB activ- ity by using high-throughput transformation assays [12,13]. Expression vectors containing full-length cDNAs were cotransformed with reporter vectors in HEK293 cells, and a new family of coactivators was identified. They were named transducer of regulated CREB activity (TORC) 1–3. The N-terminal region of TORCs formed a coiled-coil structure, which interacted with the bZIP domain of CREB [12]. Once the TORCs had been overexpressed in HEK293 cells, CREB-dependent transcriptions were upregulated at, or beyond, the levels induced by cAMP. Activation of CREs by an overexpression of TORCs required CREB-, but not Ser133 phosphorylation, and, thus, TORCs were thought to be coactivators that did not require CREB phosphorylation. Cytochemical studies of TORCs, however, showed that the activating signals that could phosphorylate CREB, such as cAMP and Ca 2+ , also induced the nuclear import of TORCs [14,15], suggesting that combination of Ser133 phos- phorylation and the binding of TORCs to the bZIP domain produces the fully ‘activated CREB’. The above observations indicate that SIKs and TORCs share a common feature regarding the regula- tion of CREB activity, both acting on the bZIP domain of CREB in a phospho-Ser133-independent manner. Having examined this feature further, we found that SIK2 phosphorylated TORC2 at Ser171. The resulting phospho-TORC2 was exported from the nucleus to the cytoplasm, and this led to the apparent inactivation of the CREB activity [14]. We also showed that PKA phosphorylated SIK1 [16] and the phospho-SIK1 could not induce the nuclear export of TORC2 [17]. The importance of the phospho ⁄ dephospho regula- tion of TORC, by regulating CREB, was also shown as a physiological impact in hepatic gluconeogenesis [18]. However, it remains to be clarified as to whether the phospho ⁄ dephospho regulation of TORC is one of the regulatory mechanisms for the CREB activity or the cascade indispensable for the CREB action. AMPK family kinases, including SIK, have flexible activation loops (A-loops) near their substrate-binding pockets. Phosphorylation in the A-loop induces a structural change in the catalytic site, which turns on the kinase activities. Recently, the LKB1 tumor suppressor kinase [19] was reported to be a major upstream activator of AMPK family kinases [20]. LKB1 phosphorylates Thr residues in the A-loops of SIKs. By using LKB1-defect- ive HeLa cells and a compound inhibiting TORC- kinases, including SIKs, we tried to elucidate the importance of the Ser133-independent activation of CREB. The results suggested that the phospho ⁄ dephos- pho regulation of TORC plays an indispensable role in the regulated activation of CREB. Results All TORCs are substrates of SIK1 One of the downstream branches of the SIK-signaling cascade leads to the regulation of CRE-dependent Y. Katoh et al. Silencing of CREB by LKB1-SIK FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS 2731 transcription, and we recently succeeded in identifying TORC2, a CREB-specific coactivator, as an endog- enous substrate of SIK2. Because mammals have three TORC isoforms, we decided to clarify which isoform could act as the endogenous substrate of SIK1. Figure 1A shows that the SIK-phosphorylation motif is highly conserved among the three isoforms. SIK1 was able to phosphorylate all the TORC peptides except for the S171A-TORC2 (Fig. 1B). The levels of SIK1-dependent phosphorylation of TORC isoforms were also examined in cultured cells. By using COS-7 cells, glutathione-S-transferase (GST)- tagged full-length TORCs were coexpressed with con- trasting SIK1 mutants; one was a kinase-defective mutant (K56M) and the other was a mutant constitu- tively phosphorylating TORC [17] (S577A mutant). As shown in Fig. 1C, the levels of phosphorylation at Ser171 of TORC2 and Ser163 of TORC3 were strik- ingly elevated in the presence of SIK1 (S577A) (see the lanes indicated by 577). However, the corresponding residue, Ser167 of TORC1, seemed to be phosphor- ylated even in cells expressing inactive SIK1 (the lanes indicated by 56), and its phosphorylation level was enhanced slightly in cells expressing SIK1 (S577A). In contrast to the phosphorylation at Ser167, binding of 14-3-3 to TORC1 was significantly enhanced by SIK1 (S577A), suggesting that SIK1 could phosphorylate TORC1, but some as yet unidentified kinases, other than SIK1, might phosphorylate at Ser167. To evaluate the direct action of SIK1 on the trans- activation activity of TORCs, assays were performed using Gal4-fused TORCs (Fig. 1D). SIK1 was able to completely inhibit the transactivation activities derived from all TORCs. Together, these results suggested that SIK1 could phosphorylate all TORCs and thereby repress their transactivation activities. SIK1 is unable to induce the nuclear export of TORC in HeLa cells The nucleo-cytoplasmic redistribution of TORC2 is important for both the stimuli-induced CRE activation AB DC Fig. 1. SIK affects the functions of all TORCs. (A) Amino acid alignment of the SIK phosphorylation motif (box) of TORC1–3. SIK2 was shown to phosphorylate Ser171 of TORC2 [14]. The corresponding Ser residues, Ser167 (TORC1), Ser171 (TORC2) and Ser163 (TORC3) are indicated in bold. (B) GST–TORC peptides were prepared in E. coli and used as substrates for an in vitro kinase assay ([ 32 P]dATP[aP]). GST- tagged SIK1(1–354) prepared in COS-7 cells was used as an enzyme. (Upper) Incorporation of 32 P into GST–TORC peptides. (Lower) Coomassie Brilliant Blue (CBB) staining of the substrates. GST–Syntide2 was used as a positive control substrate. (C) COS-7 cells were co- transformed with pEBG-TORC1 (1.5 lg), pEBG-TORC2 (4 lg) or pEBG-TORC3 (6 lg) and pTarget-SIK1s (2 lg each). pEBG is a mammalian expression plasmid for the GST-fusion protein. After 48 h, GST–TORCs were purified by glutathione columns (CP:) and then subjected to western blot analyses (WB:) using anti-GST (upper), anti-(phospho-Ser171 TORC2) (middle) and anti-(14-3-3) (lower) sera. 577 indicates the SIK1 mutant (S577A) which represses the CRE activity constitutively. 56 indicates the SIK1 mutant (K56M) with no kinase activity. (D) HEK293 cells were cotransformed with the expression plasmids for Gal4-fusion TORC1–3 (0.05 lg) and a 5xGAL4-luciferase reporter plasmid (0.2 lg) with an internal reporter phRL-TK(Int – ) (0.03 lg) in the presence or absence of the SIK1 (S577A) plasmid (0.1 lg). The specific trans- activation activities of TORCs were expressed as the fold-activation of the empty Gal4 vector, pM. Means and SD are indicated (n ¼ 4). Silencing of CREB by LKB1-SIK Y. Katoh et al. 2732 FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS and the SIK-mediated CRE repression. Interestingly, Bittinger et al. found that TORC2 and TORC3 never moved out of the nucleus in HeLa cells [15]. Because HeLa cells lacked LKB1, which had been reported to phosphorylate SIKs and activate them [20], we thought that the impaired nuclear export of TORC2 in HeLa cells was a result of the loss of SIK activity [20]. To test this, we compared the behaviors of TORCs in HeLa and COS-7 cells using a green fluorescent pro- tein (GFP)-fusion technique. Because GFP–TORC1 was present in the cytoplasm of COS-7 cells (Fig. 2A), we were unable to see the SIK1-induced intracellular redistribution of GFP–TORC1 [compare SIK1(–) with SIK1(+)]. In contrast to TORC1, GFP–TORC2 clearly showed SIK1-dependent nuclear export. GFP– TORC3 also moved out of the nucleus in a slightly lower level. As expected, overexpression of SIK1 did not induce the intracellular redistribution of TORCs in HeLa cells (Fig. 2B). Overexpression of LKB1 in HeLa cells restores the nucleo-cytoplasmic shuttling of TORC2 To elucidate the mechanism underlying the impaired nucleo-cytoplasmic shuttling of TORC in HeLa cells, we first tested whether overexpression of LKB1 in this cell line could restore the SIK1-induced nuclear export of TORC2 (Fig. 3A). As shown in the third panel of the ‘upper’ set, a small population of GFP–TORC2 moved to the cytoplasm in LKB1-overexpressing HeLa cells. Furthermore, expression of LKB1 and SIK1 in combination could completely induce the nuclear export of GFP–TORC2 (final panel). As expected, dis- tribution of GFP–TORC2 was not influenced by the overexpression of LKB1 in LKB1-positive COS-7 cells (lower set). The Thr182 of SIK1 is phosphorylated by LKB1, resulting in conversion from inactive SIK1 to the act- ive form. The importance of phospho-Thr182 was also supported by the fact that substitution of the Thr with a negatively charged residue produced a constitutive active enzyme [20]; hence we prepared the T182E mutant. As shown in Fig. 3B, however, neither SIK1 (T182E) nor SIK1 (T182A) could enhance LKB1-sup- ported nuclear export of GFP–TORC2 in HeLa cells. Differential properties of the A-loops of the individual isoforms of SIK1–3 Because the SIK1 (T182E) mutant did not induce the nuclear export of TORC2, we assayed the kinase activ- ity of this mutant. The T182E mutant, prepared as a GST-fusion protein using COS-7 cells, was much less active than wild-type SIK1 (Fig. 4A). As expected, A B Fig. 2. SIK1 alone is unable to induce the nuclear export TORC2 in HeLa cells. (A) COS-7 cells cultured on cover slips were cotrans- formed with expression vectors for GFP-tagged TORC1–3 with (+) or without (–) the SIK1 (S577A) plasmid. After 24 h, the cells were fixed for cytochemical analyses as described in Experimental proce- dures. Green fluorescent signals of GFP–TORC1–3 (upper) and blue fluorescent signals of nuclear staining with DAPI (lower) are shown. (B) The same experiments were performed using HeLa cells. More than 80% of GFP-positive cells had similar patterns as shown in each representative panel. A B Fig. 3. LKB1 is essential for the SIK1-induced nuclear export of TORC2 in HeLa cells. (A) LKB1-defective HeLa cells (upper) and LKB1-positive COS-7 cells (lower) were cotransformed with the GFP–TORC2 expression plasmid and SIK1 expression plasmid in the presence or absence of the LKB1 expression plasmid (pEBG- LKB1). (B) Thr182, the LKB1-dependent phosphorylation site, of SIK1 was substituted with Glu or Ala, and the resultant mutants were subjected to cytochemical analyses of GFP–TORC2. HeLa cells were transformed with plasmids as in Fig. 2. Y. Katoh et al. Silencing of CREB by LKB1-SIK FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS 2733 neither SIK1 (T182A) nor a negative control mutant, K56M, showed kinase activities. The discrepancy between our T182E mutant and the mutant in previous reports [20,21] might be caused by the different sources, Escherichia coli or cultured cells. However, similar discrepancies have also arisen in studies of AMPK [22]. There are three isoforms of SIK in the AMPK-rela- ted kinase family. Although the overall sequence of A-loops in SIKs is highly conserved (Fig. 4B), some variety is found in the N-terminal side of the LKB1- phosphorylatable Thr in SIK3. To see whether the SIK2 and SIK3 isoforms behave similarly to SIK1 with regard to LKB1-dependent phosphorylation of Thr, we prepared several mutants in which the corres- ponding Thr residues were substituted. Kinase assays of GST–SIK2s (Fig. 4C) produced results similar to those of SIK1s. However, GST–SIK3s (Fig. 4D) provi- ded results quite different from the others. SIK3 (T163A) had a little peptide phosphorylation activity, and SIK3 (T163E) had activity as high as that of the wild-type enzyme. SIK kinase activity is sufficient to induce the nuclear export of TORC2 in HeLa cells The finding of the constitutive active SIK3 mutant, SIK3 (T163E), prompted us to investigate whether the kinase activity of SIK was sufficient to export TORC2 even under LKB1-defective conditions. Expression plasmids for GFP–TORC2 and SIK3s were cotrans- formed into HeLa cells. In LKB1-overexpressing HeLa cells, GFP–TORC2 was exported from the nucleus to the cytoplasm by either wild-type SIK3 or SIK3 (T163E) mutant (Fig. 5, upper). SIK3 (T163A) mutant was unable to enhance the nuclear export of GFP– TORC2. As expected, even in LKB1-nonexpressing HeLa cells (Fig. 5, lower), SIK3 (T163E) could induce the nuclear export of GFP–TORC2, although wild- type SIK3, induced little export. These results suggested A B DC Fig. 4. The effect of the Thr to Glu substitu- tion in the A-loop on the kinase activities of SIKs. (A) GST–SIK1 (1–354) and its mutants were prepared using COS-7 cells and were subjected to an in vitro kinase assay. K56M is kinase-defective SIK1, a negative control mutant. GST–Syntide2 was used as a sub- strate. (B) Alignment of amino acid sequences of the A-loops of SIK1–3 and AMPKa1. Conserved residues are marked by black boxes, and the Thr residues repor- ted to be phosphorylated by LKB1 are indi- cated by a phosphor symbol. (C) Thr175 of GST–SIK2 (full-length), corresponding to Thr182 of SIK1, was replaced with Glu or Gly, and the resultant mutants were subjec- ted to an in vitro kinase assay. K49M is kin- ase-defective SIK2. (D) Thr163 of the SIK3 kinase domain (1–340) was substituted with Ala or Glu. K37M is kinase-defective SIK3. These panels were one of representative sets using SIK enzymes that had been pre- pared using COS-7 cells at least three times. Fig. 5. The constitutive active SIK3 exports TORC2 without LKB1 in HeLa cells. The effects of substitutions at Thr163 of SIK3 on the intracellular localization of GFP–TORC2 in HeLa cells in the pres- ence (upper, +LKB1) or absence (lower, –LKB1) of LKB1 as des- cribed in Figs 2 and 3. pEBG–SIK3 (1–340) was used for the overexpression of SIK3. Silencing of CREB by LKB1-SIK Y. Katoh et al. 2734 FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS that LKB1 could regulate the intracellular distribution of TORC2 through SIKs, and that SIK kinase activity might be sufficient to induce the nuclear export of TORC2. Environments of CRE-dependent transcription in HeLa cells Next, we examined the expression of an endogenous target of CREB, NR4A2 (Nurr1) gene (Fig. 6A). The level of NR4A2 mRNA was significantly induced by forskolin treatment in HEK293 cells. In HeLa cells, however, it had already been expressed moderately, and its level was not enhanced strongly by forskolin treatment. The level of 36B4 RNA, generally used as an internal standard, showed no change. To find out why HeLa cells expressed NR4A2 mRNA constitutively, we compared the expression level and the status of components in the TORC– CREB system between HeLa and HEK293 cells. The mRNA levels of TORCs in HeLa cells did not differ substantially from those in HEK293 cells (Fig. 6B). However, protein levels of TORCs seemed to be much lower in HeLa cells than in HEK293 cells (Fig. 6C,D). TORC2 proteins in HEK293 cells are known to migrate as two bands on SDS⁄ PAGE (Fig. 6C). A slowly moving form was the major form in nonstim- ulated HEK293 cells and was shown to be the phos- phorylated form. However, the rapidly moving form was the major form in forskolin-treated cells and was shown to be the dephosphorylated form [14,15]. In HeLa cells, however, only the rapidly moving AB DC E Fig. 6. Impaired CRE-dependent transcrip- tion in HeLa cells. (A) Quantification of the mRNA levels for NR4A2 and 36B4 in HEK293 and HeLa cells using real-time PCR analyses. Forskolin (20 l M) was added to the culture medium 4 h prior to the harvest for total RNA extraction. One unit is equival- ent to 6 pg of the standard plasmids con- taining respective amplicons. To quantify 36B4 RNA, a reverse-transcribed mixture was further diluted at 1:100. Means and SD are indicated (n ¼ 3). (B) Quantification of the mRNA levels for TORC1–3 in HEK293 and HeLa cells by real-time PCR analyses. Forskolin (20 l M) was added to the culture medium 4 h prior to the harvest for total RNA extraction. (C) Analyses of the level and modification of the TORC2 protein in HEK293 and HeLa cells. Forskolin (20 l M) was added to the culture medium 2 h prior to the harvest for immunoprecipitation (IP) followed by western blot analyses (WB). (D) Analyses of the levels and modifications of TORC1 and TORC3 in HEK293 and the HeLa cells. (E) The phosphorylation of CREB at Ser133 is impaired in the HeLa cells. Forskolin (20 l M) was added to the culture medium 30 min prior to the harvest for western blotting. Y. Katoh et al. Silencing of CREB by LKB1-SIK FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS 2735 dephospho form was seen. Because our antibody raised against TORC1 was able to detect both TORC1 and TORC3 with equal efficiency, analyses of these two TORCs were carried in a single blot (Fig. 6D). Simi- larly to the case of TORC2, TORC1 and TORC3 also formed two bands in the gel. The cases of TORC1 and TORC3 were apparently the same as TORC2, but might be substantially different, because all of the bands responded to forskolin treatment in HEK293 cells and shifted to lower positions. However, in HeLa cells the bands remained at the same positions irres- pective of treatment. It should be mentioned here that Ser133 phosphorylation of CREB might not be strong in forskolin-treated HeLa cells (Fig. 6E), suggesting that more than one step in the regulatory pathway of CREB might be impaired in HeLa cells. LKB1 restores the accentual regulation of CRE-dependent transcription To examine whether the constitutive expression of NR4A2 mRNA in HeLa cells suggested the impaired regulation of CREs, and if so, to test whether LKB1 could restore the forskolin-induced activation of CRE- dependent transcription, we tried to perform reporter assays. Because plasmid-based reporters could not pro- vide a high enough level of reporter activities in HeLa cells (not shown), we prepared an adenovirus-mediated reporter system. As shown in Fig. 7A, weak enhance- ment of CRE activity by forskolin was observed in the control HeLa cells (LacZ). Coinfection with the LKB1- adenovirus (LKB1) repressed basal CRE activity to one-tenth of its original level within 24 h. At this time point, forskolin was unable to substantially induce CRE activity. At 48–72 h post infection, however, a large induction of the CRE activity by forskolin was observed. In addition, forskolin induced NR4A2 mRNA in LKB1 expressing HeLa cells (Fig. 7B). To investigate whether the impaired regulation of CRE-dependent transcription in HeLa cells resulted from the dysfunction of the overall phosphorylation cascades in the SIK–TORC system or the particular combination of SIK- and TORC isoforms, the levels of proteins and the phosphorylation of individual isoforms were examined. As shown in Fig. 7C, the level of SIK1 protein was elevated slightly by forskolin in control cells (LacZ). When the LKB1-adenovirus was infected, the basal level of SIK1 decreased significantly, and the level was elevated prominently by forskolin. These results agreed with the fact that SIK1 gene expression depended on its own CREs [18]. In vitro kinase assays using the SIK1 protein purified by immunoprecipitation indicated that overexpression of LKB1 restored SIK1 kinase activity in HeLa cells (see the panel indicated by 32 P-ATP. As expected, Thr182 was not phosphory- lated in the LKB1 nonexpressing HeLa cells. Because sensitivity of the anti-(phospho-Thr182 IgG) was less than that of the anti-(SIK1 IgG), we could not dis- cuss whether Thr182 was phosphorylated in the LKB1- expressing cells when cells were not stimulated with forskolin (third lane from the left). After forskolin treatment, however, the level of SIK1 protein had risen sufficiently so that we were able to detect phospho- Thr182 (the final lane). In the case of SIK2 (Fig. 7D), the protein level was not influenced by overexpression of LKB1. The levels of kinase activity and phospho-Thr175 were elevated in the LKB1-expressing HeLa cells. By contrast, the protein level of SIK3 increased in LKB1-expressing HeLa cells, and restoration of the kinase activity and phosphorylation at Thr163 also occurred (Fig. 7E). Next, we examined the phosphorylation of TORCs in the same way (Fig. 7F). Similarly to the cases for HEK293 cells, overexpression of LKB1 restored phos- pho- ⁄ dephospho regulation of TORCs in HeLa cells. We describe briefly here the results of TORC1. A small part of the TORC1 population had already been phos- phorylated in control HeLa cells (lanes indicated by LacZ), and LKB1 enhanced the phosphorylation of TORC1 (third lane from the left). Forskolin treatment stimulated the dephosphorylation of TORC1 a little (final lane). Results in Figs 7F and 1C suggest that multiple cascades might be operating differentially in the phosphorylation of TORC1, and these cascades may be classified into three categories, LKB independ- ent, SIK independent and SIK dependent. Finally, we assayed the level of CREB phosphoryla- tion (Fig. 7G). Forskolin-induced phosphorylation of CREB at Ser133 was evident in LKB1-overexpressing HeLa cells. These results suggested that LKB1 could modulate the actions of all participants in the SIK– TORC–CREB cascade. SIK activity restores the regulation of CRE-dependent transcription in HeLa cells To obtain direct evidence of SIK-mediated phosphory- lation of TORC in HeLa cells, the constitutive active SIK3 mutant was overexpressed in this cell line. As shown in Fig. 8A, the constitutive active SIK3 mutant, T163E, could phosphorylate TORC2 even in the absence of LKB1. Moreover, the T163E mutant could recover the forskolin-dependent induction of CRE activity without LKB1 (Fig. 8B). Finally, it should be noted that the forskolin- induced CREB phosphorylation at Ser133 was restored Silencing of CREB by LKB1-SIK Y. Katoh et al. 2736 FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS by overexpression of SIK3 (T163E) (Fig. 8C). These observations suggested that restoration of SIK activity might be sufficient for repairing impaired CREB- dependent transcription in HeLa cells. LKB1-mediated phosphorylation of SIK1 at Thr182 enhances phosphorylation at Ser577 We noticed a large discrepancy between the increase in total SIK1 activity and the decrease in the level of TORC phosphorylation in forskolin-treated LKB1- expressing HeLa cells (Fig. 7C,F). In this regard, we found a similar case when CREB was activated by PKA; PKA also phosphorylated SIK1 at Ser577, which diminished SIK1-mediated cytoplasmic retention of TORC2 [17]. Interestingly, phosphorylation at Ser587 of SIK2 (corresponding to Ser577 of SIK1) was not obvious in LKB1 nontransformed HeLa cells (Fig. 7D, lower). To compare the specific level of SIK1 phosphorylation at Ser577 with that at Thr182 and the AB DE C FG Fig. 7. LKB1 restores forskolin-induced CRE activity in HeLa cells. (A) HeLa cells were cotransfected with adenovirus reporters, Ad-CRE-fLuc or Ad-TK-rLuc (an internal standard), at moi 3 and an LKB1-adenovirus or a lacZ-adenovirus at moi 30. After the indicated periods, cells were harvested for the luciferase assays. Forskolin (20 l M) was added to the culture medium 8 h prior to the harvest. (B) HeLa cells were trans- fected with the adenovirus of LKB1 or lacZ. After 72 h, total RNA was purified from the cells and the levels of mRNA for NR4A2 (Nurr1) and 36B4 were quantified using real-time PCR analysis as described in the Experimental procedures. Forskolin (20 l M) was added to the culture medium 8 h prior to the harvest. n ¼ 3. (C–G) HeLa cells were transfected with the LKB1-adenovirus or the lacZ-adenovirus. After 72 h, cells were treated with forskolin (20 l M) for 1 h and then subjected to immunoprecipitation (IP:) using anti-SIK1 (C), anti-SIK2 (D), anti-SIK3 (E), anti-TORC2 or anti-TORC1 ⁄ 3 (F) sera followed by western blotting. Immunopurified SIK enzymes were also subjected to in vitro kinase assays ([ 32 P]dATP[aP]) using GST–Syntide2 as a substrate. The panels represent one of duplicate experiments. Anti-(phospho-Thr182 SIK1) was able to detect phospho-Thr175 of SIK2 and phospho-Thr163 of SIK3. Anti-(phospho-Ser577 SIK1) cross-reacted with phospho-Ser587 of SIK2. To detect overexpressed LKB1 and total- ⁄ phospho-CREB, cell lysate was subjected to western blotting (G). Y. Katoh et al. Silencing of CREB by LKB1-SIK FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS 2737 kinase activity, GST-fusion SIK1 was overexpressed in HeLa cells. As shown in Fig. 9A, Ser577 was not phosphorylated in control HeLa cells and was apparently less phosphorylated by forskolin treatment (indicated by LacZ). Overexpression of LKB1 induced phosphorylation at Ser577, and its level was signifi- cantly elevated after forskolin treatment (indicated by LKB1), suggesting that phospho-Ser577 might be the result of an autophosphorylation of SIK1. Other indi- cators, such as phospho-Thr182 and kinase activities, depended on LKB1, but not on the forskolin treat- ment. When Ser577 is phosphorylated, the phospho-SIK1 moves to the cytoplasm. Using this property, we exam- ined the LKB1-initiated autophosphorylation of SIK1 at Ser577. As shown in Fig. 9B, in control HeLa cells (–), GFP–SIK1 was localized only in the nucleus. When GFP–SIK1 was coexpressed with LKB1 or PKA, part of the SIK1 population moved to the cyto- plasm. LKB1-induced nuclear export of SIK1 was abolished by the T182A substitution (Fig. 9C). Substi- tution at Ser577 completely inhibited both LKB1- and PKA-induced nuclear export of SIK1. Finally, we tested the level of TORC phosphoryla- tion using wild-type and S577A-SIK1 (Fig. 9D). In COS-7 cells, the Ser577 mutant SIK1 phosphorylated GST-fusion TORC2 more efficiently than the wild- type. These observations suggested that the PKA-phos- phorylatable Ser577 also acted as the autophosphory- lation site, and that phospho-Ser577 might be a critical modulator of the TORC phosphorylation activity of SIK1. In this context, LKB1 might also play important roles in the attenuation step of the phosphorylation of TORC. The SIK3 T163E-mutant having the additional mutation at Ser493, equivalent to S577A of SIK1, also suggested the importance of the Ser phosphorylation (Fig. 8B). Inhibition of kinase cascades activates CRE-dependent transcription constitutively The constitutive activation of CRE-dependent tran- scription in HeLa cells (Fig. 6A,B) might be due to inactivation of the phosphorylation cascades from LKB1 to TORCs, suggesting, paradoxically, that inhi- bition of the kinase cascades could mimic impaired CREB regulation even in LKB1-positive cells. To test this possibility, we performed CRE-reporter assays in HEK293 cells in the presence of various kinase inhibi- tors. As shown in Fig. 10A, no specific kinase inhibitor could activate the CRE. However, staurosporine (STS; 10 nm), a nonspecific kinase inhibitor [23], induced CRE reporter activity to levels as high as forskolin. Moreover, staurosporine upregulated transcription of the NR4A2 gene to a level higher than that elevated by forskolin treatment (Fig. 10B). A B C Fig. 8. The constitutive active SIK3 restores forskolin-induced CRE activity without LKB1 in HeLa cells. (A) HeLa cells were infected with the SIK3 (full-length) adenoviruses. After 72 h incubation, TORC2 was analyzed by immunoprecipitation, and SIK3 and LKB1 were detected by western blotting using cell lysates. The lower band in the SIK3 panel might be degraded products. (B) HeLa cells were cotransformed with SIK3-adenoviruses and reporter-adeno- viruses, and CRE activity was measured as described in Fig. 1. S493A of SIK3 is equivalent to S577A of SIK1. (C) HeLa cells were infected with the SIK3 T163E adenovirus or its control virus, LacZ. After 72 h incubation, cells were harvested to analyze the level of phospho-CREB using western blotting. These panels represent experiments performed at least in duplicate. Silencing of CREB by LKB1-SIK Y. Katoh et al. 2738 FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS Staurosporine has been classified as a PKC inhibitor (Fig. S1A). However, another PKC-specific inhibitor, bisindolylmaleimide I (Bis) did not induce any CRE reporter activity in our assay system (Fig. 10A), sug- gesting that PKC might not be the kinase responsible for staurosporine-induced CRE activity. To investigate the phosphorylation status of TORC and CREB in staurosporine-treated cells, GST-tagged TORC2 and endogenous CREB were examined in COS-7 cells (Fig. 10C). Forskolin induced both the dephosphorylation at Ser171 and the decrease in the level of bound 14-3-3. It enhanced the phosphoryla- tion of CREB at Ser133, of course. As expected, sta- urosporine completely inhibited the phosphorylation of TORC2 and did not enhance the phosphorylation of CREB. Because staurosporine significantly blocked SIK1- mediated CRE repression (not shown), the efficacy of staurosporine on SIK1 was estimated by measuring its IC 50 as regards the kinase activity. The in vitro IC 50 was  0.15 nm (Fig. S1B). To evaluate SIK1 inhibition in vivo, the difference between forskolin-induced CRE- reporter activity and its activity in the presence of SIK1 S577A was used. An in vivo IC 50 of staurospo- rine against the exogenously expressed SIK1 S577A mutant was  5.0 nm (Fig. S1C). These results sugges- ted that the kinase activity of endogenous SIK might be inhibited by staurosporine at a dose as low as that against PKC. To examine whether staurosporine-induced dephos- phorylation of TORC2 was accompanied by its nuclear accumulation, GFP–TORC2 was expressed in HeLa cells in the presence or absence of the LKB1–SIK cas- cades, and the cells were treated with staurosporine (Fig. 10D). Staurosporine inhibited the nuclear export of GFP–TORC2 in all the cases tested. These results, however, might indicate two possibilities, namely that staurosporine either inhibited SIKs directly or blocked the upstream cascades of SIKs, including LKB1. To clarify this point, COS-7 cells that had been expressing GST-tagged SIK1 were treated with sta- urosporine and GST–SIK1 protein was purified (Fig. 10E). SIK1 enzyme purified from the staurospo- rine-treated cells was phosphorylated at Thr182 but did not show any kinase activities. Because Thr172 of AMPKa1 (corresponding to Thr182 of SIK1) is also phosphorylated by LKB1, we performed the same experiment using GST–AMPKa1 (lower left). Neither the phosphorylation level at Thr172 nor the kinase activity of AMPKa1 was affected by staurosporine treatment. Forskolin treatment did not alter the levels of Thr phosphorylation or the kinase activity of SIK1 AB C D Fig. 9. Ser577 is an autophosphorylation site of SIK1. (A) HeLa cells were transformed with the expression plasmid, pEBG-SIK1 for GST- fusion SIK1 (full-length), then transfected with the lacZ- or LKB1-adenovirus. After 48 h incubation, the cells were treated with forskolin (20 l M) for 30 min, and the GST–SIK1 protein was purified using a glutathione column (CP). The SIK1 protein was subjected to western blot analyses as well as kinase assays as described in Fig. 7. (B) HeLa cells were transformed with the GFP–SIK1 expression plasmid with the LKB1- or the PKA-expression plasmid as described in Fig. 2. (C) Mutant GFP–SIK1, T182A or S577A, was expressed in LKB1-expressing HeLa cells with or without PKA. (D) COS-7 cells were transformed with the GST–TORC2 expression plasmid in the presence of the SIK1 expression plasmid, wild-type or S577A. After 48 h incubation, the cells were treated with forskolin (20 l M) for 1 h, and then the GST– TORC2 protein was purified. Y. Katoh et al. Silencing of CREB by LKB1-SIK FEBS Journal 273 (2006) 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS 2739 [...]... either by the action of PKA or by autophosphorylation, might cause the cytoplasmic localization of SIK1 (Fig 9B,C) In this context, phosphorylation at Thr182 in the A-loop might be a prerequisite step for the cytoplasmic localization of SIK1 To summarize, although CREB is believed to be active when it is phosphorylated by kinases cascades, this study gives a new insight that CREB has a constitutive active. .. strategies using the combination of RNAi But when the total number of the TORC kinases, their cross-talks and feedback regulations are considered, these strategies would require a quite formidable effort At present, therefore, our strategy to use staurosporine or LKB1-defective cells is the only practical method to extract the constitutive active potency of CREB without classical agonists Further elucidation... 2730–2748 ª 2006 The Authors Journal compilation ª 2006 FEBS Y Katoh et al Silencing of CREB by LKB1-SIK A B C D E Fig 12 SIKs and MARK4 have the potential to downregulate CRE-dependent transcription (A) COS-7 cells were transformed with the CREreporter plasmid (pTAL-CRE: 0.2 lg) in the presence or absence of the PKA expression plasmid (pIRES-PKA: 0.1 lg) and expression plasmids for the kinase domains... complex, the structure of which had already been determined (PDBID 1NVR) [37], and the staurosporine was placed at the corresponding position of the SIK1 to the staurosporine-binding site on the Chk1 After energy minimization of the complex model, we obtained a final model of the SIK2–staurosporine complex We next calculated the electrostatic potential on the molecular surface [38] of SIK2 using the program... [27] The genomic rearrangement of t(11;19), which is often associated with mucoepidermoid carcinoma, produces a fusion protein that has the N-terminal CREB binding region (1–42 amino acids) of TORC1 ⁄ MECT1 and the transcriptional activation domain of other transcription factor, Mastermind-like gene family 2 (MAML2) The resultant chimeric protein, MECT1– MAML2, binds to CREB, activates CRE-mediated transcription. .. complete loss of the LKB1 gene [33] However, neither knockdown of the LKB1 protein, using an siRNA technique, nor overexpression of LKB1 affected CREB activity in HEK293 cells (not shown) This was also supported by the result that overexpression of LKB1 in COS cells did not alter the intracellular distribution of GFP–TORC2 (Fig 3A) Therefore, we suppose that the constitutive activation of CREB seen in.. .Silencing of CREB by LKB1-SIK Y Katoh et al A D B C E Fig 10 Inhibition of kinase cascades including SIKs constitutively activates CRE-dependent transcription (A) The screening of protein kinase inhibitors having a potency to induce CRE-dependent transcription HEK293 cells that had been transformed with the pTAL-CRE reporter (f Luc) and an internal standard... CRE-dependent transcription might be lower than those of SIKs in living cells Finally, we would like to briefly mention MARK4 Because the kinase domains of SIKs are highly conserved with those of four other kinases that belong to the MARK subfamily, we tested the possibility of MARKs as CREB regulators by using one MARK, MARK4 The kinase domain of MARK4 also showed CREB- repression activity In addition, the kinase... the b ⁄ c subunits The AMPKa subunits were expressed as GST-fusion proteins, and the b ⁄ c subunits were HA-tagged proteins (E) Levels of expressed AMPKa and b ⁄ c subunits in the cell lysate were examined by western blot analysis The b ⁄ c subunits were detected by an anti-(HA-tag IgG) The active AMPKa subunits were monitored by antiphospho-T172 IgG wild-type enzymes phosphorylated by LKB1 [20,21] A... shows the case of SIK2) Thus mutants SIK1 (E110D), SIK2 (E103D) and SIK3 (E91D) were prepared The kinase activity of SIK1 (E110) and SIK3 (E91) was lower than those of their parents (not shown), and therefore these mutants did not seem suitable for the analysis of staurosporine action Mutant SIK2 (E103D), however, showed kinase activity comparable with the parent and the activity was not influenced by the . Silencing the constitutive active transcription factor CREB by the LKB1-SIK signaling cascade Yoshiko Katoh 1 , Hiroshi. to whether the phospho ⁄ dephospho regulation of TORC is one of the regulatory mechanisms for the CREB activity or the cascade indispensable for the CREB