The AMPK-related kinase SIK2 is regulated by cAMP via phosphorylation at Ser358 in adipocytes

SIK2 (salt-inducible kinase 2) is a member of the AMPK (AMP-activated protein kinase) family of kinases and is highly expressed in adipocytes. We investigated the regulation of SIK2 in adipocytes in response to cellular stimuli with relevance for adipocyte function and/or AMPK signalling. None of the treatments, including insulin, cAMP inducers or AICAR (5-amino-4-imidazolecarboxamide riboside), affected SIK2 activity towards peptide or protein substrates in vitro. However, stimulation with the cAMP-elevating agent forskolin and the β-adrenergic receptor agonist CL 316,243 resulted in a PKA (protein kinase A)-dependent phosphorylation and 14-3-3 binding of SIK2. Phosphopeptide mapping of SIK2 revealed several sites phosphorylated in response to cAMP induction, including Ser358. Site-directed mutagenesis demonstrated that phosphorylation of Ser358, but not the previously reported PKA site Ser587, was required for 14-3-3 binding. Immunocytochemistry illustrated that the localization of exogenously expressed SIK2 in HEK (human embryonic kidney)-293 cells was exclusively cytosolic and remained unchanged after cAMP elevation. Fractionation of adipocytes, however, revealed a significant increase of wild-type, but not Ser358Ala, HA (haemagglutinin)–SIK2 in the cytosol and a concomitant decrease in a particulate fraction after CL 316,243 treatment. This supports a phosphorylation-dependent relocalization in adipocytes. We hypothesize that regulation of SIK2 by cAMP could play a role for the critical effects of this second messenger on lipid metabolism in adipocytes.

An important step in elucidating the biological function of SIK2 and its related kinases is to determine how they are affected by different cellular signals. The most well-characterized mode of regulation for SIK2 so far discovered is the phosphorylation of its activation (T-) loop Thr 175 by LKB1, which is required for SIK2 catalytic activity, as demonstrated by the complete lack of activity in the Thr175Ala mutant, or in LKB1-deficient cells [8]. The Tloop phosphorylation site of SIK1 and SIK3, but not SIK2, binds to 14-3-3 scaffolding proteins, which in turn regulates the activity and localization of these kinases [9]. Agents that increase the AMP/ATP ratio have been shown not to influence SIK2 activity in various systems [8,10]. On the other hand, one study reported that AICAR (5-amino-4-imidazolecarboxamide riboside), an AMP mimetic, as well as glucose starvation, stimulated SIK2 activity in 3T3-L1 adipocytes [2]; however, the underlying mechanism for this activation was not described. Another report proposes that insulin activates SIK2 in hepatocytes via phosphorylation of Ser 358 by PK (protein kinase) B, and that this regulation in part mediates the ability of insulin to inhibit gluconeogenesis [11]. More recently, SIK2 was suggested to be regulated by CaMK (Ca 2 + /calmodulin-dependent kinase) I/IV in neuronal cells, via a phosphorylation of Thr 484 [12]. The cAMP/PKA pathway has been shown to regulate SIK isoforms, in particular SIK1. Treatment of Y1 cells and 3T3-L1 fibroblasts with cAMP-elevating agents was demonstrated to induce the phosphorylation of SIK1 on Ser 577 (human Ser 575 ), resulting in its nuclear export [7,13]. The homologous site of SIK2, Ser 587 , was similarly phosphorylated; however, the effect on SIK2 localization was not as clear as for SIK1, probably owing to the lack of an NLS (nuclear localization signal) in SIK2 [3,14]. The phosphorylation of SIK1 and 2 in response to cAMP induction is thought to restrict their inhibitory action on different transcriptional regulators, including CREB (cAMP-response-element-binding protein) co-activator CRTC2 (CREB-regulated transcription co-activator-2) [previously called TORC2 (transducer of regulated CREB activity 2)] [14,15].
As mentioned above, the expression of SIK2 is many-fold higher in adipose tissue than elsewhere, and is induced during adipocyte differentiation [3]. In addition, SIK2 protein expression and activity were also shown to be up-regulated in adipose tissue of db/db mice. This, along with its described role in other tissues and relationship to AMPK, a known target of anti-diabetic drugs, prompted us to carefully investigate the regulation of SIK2 in adipocytes. Various cellular signals with important roles in the regulation of AMPK and/or adipocyte function were studied and we present evidence that cAMP, a critical second messenger in the control of lipid metabolism, regulates SIK2 in adipocytes at many levels.
[γ -32 P]ATP was from PerkinElmer and phosphocellulose P81 paper was from Whatman. Hydromount was from National Diagnostics. AMARA peptide, HDAC5tide (where HDAC is histone deacetylase) and IRS1tide (where IRS1 is insulin receptor substrate1) were synthesized by Dr Grahame Bloomberg (University of Bristol, Bristol, U.K.) and Sakamototide was synthesized by Pepceuticals. Adenoviral vectors for HA-SIK2 wild-type, Ser358Ala and Ser587Ala were produced by Vector BioLabs.

Subcellular fractionation of primary rat adipocytes
Primary rat adipocytes expressing wild-type or Ser358Ala HA-SIK2 were isolated, stimulated and homogenized in ice-cold homogenization buffer containing 30 mM Tris/HCl (pH 7.5), 1 mM EGTA, 1 mM EDTA, 0.25 mM sodium orthovanadate, and then further centrifuged at 60 000 g for 45 min at 4 • C. The supernatant (referred to as 'cytosol') was supplemented with 1 % Nonidet P40, after which insoluble particles were removed by centrifugation at 13 000 g for 15 min at 4 • C. The resulting pellet (60 000 g, referred to as the 'particulate fraction') was resuspended in homogenization buffer containing a final concentration of 1 % Nonidet P40, after which the insoluble particles were removed by centrifugation as described above.

Generation of stable cell lines
Flp-in TM T-REx TM -293 cells cultured in 10-cm dishes were transfected with 0.5 μg of pcDNA5/FRT/TO-FLAG constructs encoding N-or C-terminally FLAG-tagged wild-type or Ser358Ala SIK2 and 4.5 μg of cDNA encoding Flp recombinase. After 3 days, DMEM selection medium containing 1 % penicillin/streptavidin, 10 % FBS, 15 μg/ml blasticidin B and 100 μg/ml hygromycin was added. The medium was changed every 24 h until only cells expressing wild-type or Ser358Ala FLAG-SIK2 remained. Selection medium was kept throughout experiments. Cells were stimulated and lysed (described below) 12 h after induction with 0.1 μg/ml tetracycline.

Culture, transfection and stimulation of HEK (human embryonic kidney)-293 cells
HEK-293 cells cultured in 10-cm dishes were transfected with 2 μg of cDNA construct encoding wild-type or the indicated mutant forms of HA-or GST (glutathione transferase)-SIK2, using the polyethylenimine method [17]. At 24 h posttransfection, the cells were starved for 16 h, stimulated, rinsed twice with PBS and lysed in 0.5 ml of ice-cold lysis buffer described above. Lysates were centrifuged at 13 000 g for 15 min to remove cell debris. GST fusion proteins were affinitypurified on glutathione-Sepharose and eluted in 50 mM Tris/HCl (pH 7.5), 0.1 mM EGTA, 1 mM DTT, 0.27 M sucrose and 20 mM glutathione.

Western and far-Western blot analysis
Cell lysates or IPs prepared from 50-500 μg of total protein, as described above, or pull downs prepared by the addition of 5 μg of Escherichia coli-expressed GST-14-3-3ζ bound to glutathione-Sepharose (described below), were heated in SDS sample buffer, before loading on to a precast 4-12 % Novex gradient gel and transferred on to a nitrocellulose membrane as described previously [16]. Detection was performed using HRP-conjugated secondary antibodies and ECL (enhanced chemiluminescence) reagent. Far-Western membranes were incubated overnight with 5 μg/ml recombinant yeast His-14-3-3 (BMH1/2) produced in E. coli, in TBS-T (Tris-buffered saline containing 0.2% Tween 20) containing 5 % (w/v) BSA and 0.5 M NaCl and then washed in TBS-T containing 0.5 M NaCl. An anti-His antibody was used to detect His-14-3-3. Films were developed and quantified as described previously [16].

Phosphopeptide mapping of SIK2 and identification of 14-3-3 isoforms by mass fingerprinting
Wild-type and mutant forms of HA-SIK2 were expressed in HEK-293 cells and primary rat adipocytes as described above. SIK2 immunoprecipitated from 10 mg of total protein was loaded on to a polyacrylamide gel, which was subsequently stained with colloidal Coomassie Blue. The bands representing HA-SIK2 and 14-3-3 were excised, incubated with 50 mM iodoacetamide to alkylate cysteine residues, washed and digested with trypsin for 16 h. Tryptic digests were analysed by LC (liquid chromatography)-MS with precursor of 79 scanning on an Applied Biosystems 4000 Q-Trap system as described in [22]. SIK2 tryptic digests were also analysed by LC-MS on either an LTQ-orbitrap classic mass spectrometer system (HEK-293 cells) or an LTQ-orbitrap Velos mass spectrometer system (adipocytes) each coupled to a Proxeon Easy-LC HPLC system. The peptide mixtures were separated on either an LC-Packings PepMap C 18 column (0.075 mm×150 mm) (HEK-293 cells) or a Proxeon EASY-Column (0.075 mm×100 mm) (adipocytes) equilibrated in 0.1 % formic acid/water and eluted with a discontinuous gradient of acetonitrile/0.1 % formic acid at a flow rate of 300 nl/min. The orbitrap was set to analyse the survey scans at 60 000 resolution and top 5 ions in each duty cycle, were selected for MS/MS (tandem MS) in the LTQ linear ion trap with multistage activation.
The MS/MS spectra were searched against the SwissProt database using the Mascot search engine (Matrix Science) run on an in-house server using the following criteria: peptide tolerance = 10 p.p.m., trypsin as the enzyme, and carboxyamidomethylation of cysteine as a fixed modification with oxidation of methionine and phosphorylation of serine, threonine and tyrosine as a variable modification. Any MS/MS spectra that could be assigned to a phosphopeptide were inspected manually using QualBrowser software.

Confocal microscopy of HEK-293 cells expressing HA-or GFP-tagged SIK1 and SIK2
HEK-293 cells were cultured on cover slips in six-well plates and transfected as described above using 0.5 μg of cDNA. At 16 h post-transfection, cells were stimulated, washed with PBS and fixed using 4 % formaldehyde. For the HA constructs, the cells were permeabilized using PBS containing 0.25 % Triton X-100 and then blocked in fish skin gelatin (3 %) for 30 min before primary antibody was added. Secondary antibody (Alexa Fluor TM 594) was added after three washes with PBS containing 0.1 % Tween. Hoechst nuclear stain was added in one of the final washes before mounting the cover slips on slides using Hydromount. For confocal imaging, a Zeiss LSM 510 META microscope was used with excitation wavelengths 405 (Hoechst nuclear stain), 488 (GFP) or 561 (Alexa Fluro TM 594) nm. A Plan-Apochromat 63×/1.4 oil DIC (differential interference contrast) objective was used and a frame size of 488 pixels×488 pixels×1 pixels. and includes S.E.M. The statistic significance of differences was defined as *P < 0.05, **P < 0.01 and ***P < 0.001 and determined using a paired two-tailed Student's t test.

RESULTS
Increased cAMP levels in adipocytes results in a PKA-dependent phosphorylation of SIK2 with no effect on its intrinsic kinase activity The kinase domain of SIK isoforms and AMPK are well conserved and SIK has been shown to phosphorylate AMPKtargeted sequences in both intact cells and in cell-free assays.
HDAC5tide and Sakamototide were selected as peptide substrates for SIK2 kinase activity measurements. In order to investigate the role of different cellular stimuli in the regulation of SIK2, 3T3-L1 adipocytes were treated with a panel of agents, including AICAR and phenformin, known AMPK activators that affect the AMP (or ZMP)/ATP ratio, the cAMP-elevating agent forskolin, the Ca 2 + ionophore ionomycin and insulin (Figure 2A). These stimuli were selected on the basis of their suggested role in the regulation of SIK2, AMPK or adipocyte function. The cellular stimuli were validated by monitoring the phosphorylation of target proteins. AMPK was phosphorylated on its T-loop Thr 172 site and activated in response to AICAR, phenformin, forskolin and ionomycin ( Figure 2B). PKB Ser 473 phosphorylation was used as a positive control for activation of the insulin signalling pathway and HSL Ser 563 phosphorylation for increased cAMP levels and the activation of PKA (Figure 2A). None of the treatments affected total SIK2 levels. The kinase activity of immunoprecipitated SIK2 was measured in vitro towards the peptide substrates HDAC5tide (Figure 2A) and Sakamototide (results not shown), as well as HDAC5 (1-550) recombinant protein ( Figure 2C). Activity measurements did not show any significant change in response to the different treatments compared with the basal activity. However, when monitoring SIK2 phosphorylation, employing phosphoPKA-and phosphoPKB-substrate antibodies, which recognize consensus sites for PKA and PKB respectively, we detected a phosphorylation of SIK2 in response to forskolin with both antibodies (Figure 2A). Phosphorylation of SIK2 was also observed in primary rat adipocytes in response to the β 3 -adrenergic receptor agonist CL 316,243 ( Figure 3B). None of the other treatments induced phosphorylation of SIK2.
To investigate the requirement of PKA for cAMP-induced phosphorylation of SIK2, we used the PKA inhibitor H89. Lysates were analysed with regard to total SIK2 and phosphorylation by PKA (pPKA substrate antibodies). HSL phosphoSer 563 was used as a control. Quantified phosphorylation data are presented as means + − S.D. for three to five individual experiments. **P < 0.01 and ***P < 0.001. 3T3-L1 (C) or rat (D) adipocytes were stimulated with forskolin (50 μM, 15 min) or CL 316,243 (100 nM, 30 min). 14-3-3 binding to SIK2 was demonstrated by the presence of SIK2 in 14-3-3 pull downs using GST-14-3-3ζ coupled to glutathione-Sepharose (C, upper panel and D), or by the direct interaction of recombinant yeast 14-3-3 (BMH1/BMH2) to immunoprecipitated SIK2 on a far-Western blot (C, lower panel). (E) Primary rat adipocytes were pre-incubated for 20 min with the PKA inhibitor H89, followed by stimulation with forskolin (50 μM, 15 min). The binding of 14-3-3 to SIK2 was demonstrated by the presence of SIK2 in 14-3-3 pull downs. Quantified 14-3-3-binding data from pull downs and far-Western blots are presented as means + − S.D. for duplicate or single samples and are representative for two individual experiments. The blots shown are representative.

As shown in Figures 3(A) and 3(B)
, pretreatment with increasing concentrations of H89 completely, or in the case of rat adipocytes partially, prevented the cAMP-induced phosphorylation of SIK2 on PKA consensus sites. To evaluate a potential role of PKB and/or the MAPK (mitogen-activated protein kinase) signalling pathway, we used a PKB inhibitor (Akti1/2) [16,23], a MEK1 [MAPK/ERK (extracellular-signalregulated kinase) kinase 1] inhibitor (PD0325901), as well as a MAPK activator (PMA). Neither inhibition of PKB, nor activation or inhibition of the MAPK pathway (as confirmed by phosphoPKB and phosphoERK1/2 blotting) affected basal or cAMP-induced phosphorylation of SIK2 (results not shown). These results confirm an important role for PKA in the cAMPinduced phosphorylation of SIK2, although other cAMP-activated pathways, such as those regulated by Epacs (exchange protein directly activated by cAMPs), may also play a role in rat adipocytes.

The adaptor protein 14-3-3 binds SIK2 in response to cAMP elevation in adipocytes
The T-loops of SIK1 and SIK3 have previously been shown to bind 14-3-3 proteins [9]. In the present study we demonstrated that phosphorylation of SIK2 in response to increasing levels of cAMP coincides with the binding of SIK2 to 14-3-3 proteins (Figures 3C and 3D). Endogenous SIK2 was detected in GST-14-3-3 pull downs from both 3T3-L1 adipocytes and primary rat adipocytes in response to forskolin and the β 3 -adrenergic receptor agonist CL 316,243 respectively ( Figures 3C, upper panel, and Figure 3D). Additionally, the direct interaction between SIK2 and 14-3-3 was illustrated by the ability of recombinant 14-3-3 to bind SIK2 in a far-Western blot experiment ( Figure 3C, lower panel). In the presence of H89, the binding of SIK2 to 14-3-3 proteins was decreased to the same extent as the phosphorylation ( Figure 3E). This further suggests that the cAMP-induced phosphorylation of SIK2 induces its binding to 14-3-3.

Identification of SIK2 phosphorylation sites induced by cAMP
In order to determine cAMP-induced phosphorylation sites in SIK2 of importance for 14-3-3 binding, a phosphopeptide mapping of SIK2 was performed ( Figures 4A and 4B). Tryptic digest and analysis of phosphopeptides from wild-type HA-SIK2 expressed in HEK-293 cells revealed five sites whose phosphorylation was induced following forskolin stimulation: Ser 576 , Ser 358 , Ser 343 , Thr 484 and Ser 587 ( Figure 4A). The same sites were detected in wild-type HA-SIK2 adenovirally expressed in primary rat adipocytes treated with CL 316,243, confirming the relevance of these sites in a more physiological setting ( Figure 4B). A total of 70-90 % of the protein sequence was obtained in the phosphopeptide mapping and, in addition to sites induced by cAMP elevation, we also detected peptides phosphorylated on Ser 534 , Ser 512 (primary adipocytes) and Ser 90 (HEK-293 cells). These phosphopeptides, however, displayed low abundancy and were therefore not studied further. The region containing potential T-loop phosphorylations, for example Thr 175 , was not covered in our analysis, probably owing to the large size of this tryptic peptide (65 amino acids). Colloidal Coomassie Blue staining, excision and mass fingerprinting of the relevant bands in the HA-IP revealed that SIK2 bound to all seven of the 14-3-3 isoforms (β, ε, η, γ , τ , ζ and σ ) ( Figures 4A  and 4B).

Ser 358 of endogenous SIK2 is phosphorylated in primary rat adipocytes and is critical for phosphorylation and 14-3-3 binding of SIK2 in response to cAMP
Selected sites that emerged from the phosphopeptide mapping (Ser 343 , Ser 358 , Thr 484 and Ser 587 ) were subjected to site-directed mutagenesis analysis in HEK-293 cells, in order to investigate their role in the cAMP-induced phosphorylation and 14-3-3 binding of SIK2. Mutation of Thr 484 and Ser 587 to alanine did not affect the overall PKA phosphorylation or the 14-3-3 binding of HA-SIK2, as judged by the presence of endogenous 14-3-3 in the anti-HA-SIK2 IPs ( Figure 4C). However, mutation of Ser 343 or Ser 358 resulted in a strong reduction of forskolininduced phosphorylation of PKA consensus sites, and complete loss of 14-3-3 binding to SIK2 ( Figure 4C). A phosphopeptide mapping of these two mutants ( Figure 4D) revealed that Ser 358 was not phosphorylated in the Ser343Ala mutant, whereas Ser 343 was normally phosphorylated in the Ser358Ala mutant, implying that Ser 343 is required for Ser 358 phosphorylation and that Ser 358 is the site directly mediating 14-3-3 binding. The results described above (Figure 2) suggest that phosphorylation and 14-3-3 binding of SIK2 does not affect its kinase activity. To further evaluate this, we examined the activity of stably expressed wild-type and Ser358Ala mutant SIK2 in HEK-293 cells. As shown in Figure 4(E), no difference in kinase activity or T-loop (Thr 175 ) phosphorylation was observed between the wild-type and Ser358Ala SIK2, or following forskolin stimulation.
To study site-specific phosphorylation of endogenous SIK2, we generated antibodies that specifically recognize SIK2 only when phosphorylated at Ser 358 ( Figure 4C). Using these antibodies we were able to demonstrate that Ser 358 of SIK2 was indeed phosphorylated in response to forskolin in 3T3-L1 adipocytes ( Figure 5A) and CL 316,243 in primary rat adipocytes (results not shown). Although PKB was activated, as measured by Ser 473 PKB phosphorylation, no phosphorylation of Ser 358 could be detected in response to insulin, in contrast with previous findings in the liver [11]. In addition, using adenoviral expression of wild-type, Ser358Ala and Ser587Ala HA-SIK2, Ser 358 was confirmed as the major site responsible for the PKA-dependent phosphorylation of SIK2 in adipocytes ( Figure 5B, upper panel). Ser587Ala HA-SIK2 was phosphorylated almost to the same extent as the wildtype HA-SIK2 upon stimulation with CL 316,243, suggesting that Ser 587 is not a major site induced by cAMP in adipocytes. Furthermore, wild-type and Ser587Ala HA-SIK2, but not the Ser358Ala mutant, was demonstrated to directly bind 14-3-3 in CL 316,243-stimulated rat adipocytes, as detected using coimmunoprecipitation and far-Western blotting ( Figure 5B, lower panel).

No evidence of nucleo-cytoplasmic shuttling of SIK2 exogenously expressed in HEK-293 cells
HEK-293 cells were used to study the localization of HA-or GFPtagged SIK2 in the absence or presence of the cAMP-elevating agent forskolin. The localization of wild-type and Ser575Ala (Ser 577 in mouse and rat) SIK1 was monitored as a control. Confocal imaging revealed that the subcellular localization of wild-type and Ser575Ala SIK1 was nuclear in non-treated cells ( Figure 6A). Upon stimulation with forskolin, wild-type SIK1 translocated to the cytosol, whereas Ser575Ala SIK1 remained in the nucleus ( Figure 6A). This is in agreement with previous findings describing an NLS in the C-terminal region of SIK1 and a nucleo-cytoplasmic shuttling regulated by cAMP via phosphorylation of Ser 577 /Ser 575 [7,13,24]. In contrast with SIK1, wild-type SIK2 and the mutant variants Ser358Ala and Ser587Ala were all cytoplasmic rather than nuclear in non-treated cells ( Figure 6B). Furthermore, all three SIK2 variants remained cytoplasmic after forskolin treatment ( Figure 6B). The results of the present study therefore suggest SIK2 to be cytoplasmic, with no obvious nucleo-cytoplasmic shuttling in response to cAMP elevation, when overexpressed in HEK-293 cells. We could not confirm earlier findings suggesting that Ser587Ala is of importance for a nuclear export of SIK2 [3]. Our findings were consistent, independent of the epitope tag used for detection of SIK1 and SIK2.

Phosphorylation-dependent relocalization of SIK2 in response to β-adrenergic stimulation of adipocytes
A common function of 14-3-3 proteins is to trap its binding partners in the cytosol [25], and we therefore more specifically investigated the localization of exogenously expressed SIK2 in adipocytes, using a fractionation protocol. Primary rat adipocytes expressing wild-type or Ser358Ala HA-SIK2 were stimulated, homogenized and centrifuged (60 000 g) to obtain a cytosol and crude particulate fractions. The fractions were characterized using cytosolic (GAPDH), nucleic (H3), mitochondrial (COX IV) and plasma membrane (Na/K ATPase) markers. The particulate fraction contained nuclei, mitochondria and plasma membrane ( Figure 6C), but no GAPDH (results not shown). GAPDH was the only detectable marker in the cytosol fraction. As shown in Figure 6(C), a significant increase in exogenously expressed wild-type HA-SIK2, but not the Ser358Ala mutant, was found in the cytosolic fraction upon treatment of rat adipocytes with CL 316,243 (P = 0.0121). This coincided with a significant decrease in wild-type HA-SIK2 in the crude particulate fraction (P = 0.0027), suggesting a relocalization of SIK2 upon cAMP stimulation. These data demonstrate a functional importance of Ser 358 phosphorylation for the regulation of SIK2 localization in response to cAMP in adipocytes.

DISCUSSION
The present study is the first to investigate the regulation of SIK2 in response to cAMP in adipocytes, a highly relevant cell type for SIK2 [2] and, in agreement with previous findings, we report a cAMP-induced phosphorylation of SIK2 [3,14]. An important conclusion from the present study is that, in adipocytes, Ser 358 , rather than Ser 587 , is the main site phosphorylated and responsible for the binding of SIK2 to 14-3-3 proteins. In contrast with SIK1, SIK3 and the Par-1/MARK family of AMPK-related kinases, which bind 14-3-3 in resting cells [9,26], binding of SIK2 to 14-3-3 proteins is dynamic, taking place only after stimulation of the cAMP/PKA pathway, re-enforcing a physiological role for this interaction.
In previous studies performed in Y1 cells and 3T3-L1 fibroblasts, Ser 577 of SIK1, and the corresponding site in SIK2 Ser 587 , was suggested to be the phosphorylation site targeted by cAMP regulation [3,14]. These conclusions were drawn on the basis of sequence motif analysis and site-directed mutagenesis, and it is therefore not clear whether other sites are also phosphorylated in these systems. We used a more direct strategy and identified all cAMP-induced SIK2 phosphorylation sites in both HEK-293 cells and adipocytes. Although we found that Ser 587 of SIK2 is indeed phosphorylated in response to cAMP in adipocytes, mutation of this residue only resulted in a moderate reduction in total PKA phosphorylation, as measured by consensus blotting of PKA phosphorylation sites on SIK2, and did not affect the 14-3-3 binding of SIK2. Mutation of Ser 358 on the other hand, resulted in a complete loss of cAMPinduced phosphorylation and 14-3-3 binding in both HEK-293 cells and adipocytes. We therefore conclude that Ser 358 is the main site phosphorylated in response to cAMP, and clearly the most important one for the binding of SIK2 to 14-3-3. We also show that this site is phosphorylated at an endogenous level. Another cAMP-induced site identified was Ser 343 , the mutant of which was also not able to bind 14-3-3. When analysing phosphopeptides in the Ser358Ala and Ser343Ala mutants, we noted that Ser 358 was not phosphorylated in the Ser343Ala mutant, a finding that was confirmed using anti-phosphoSer 358 antibodies. This suggests that Ser 343 may be required for Ser 358 phosphorylation to occur. As Ser 343 was normally phosphorylated in the Ser358Ala mutant, which was unable to bind 14-3-3, we propose that Ser 358 , and not Ser 343 , is the site with which 14-3-3 interacts. With an arginine residue positioned at − 5 and − 3 from the phosphorylated serine, Ser 358 is situated in a sequence that is one of the most common motifs for 14-3-3 binding in mammalian proteins [27].
The effect of cAMP-induced phosphorylation on the intrinsic kinase activity of SIK isoforms was previously poorly investigated. We show that neither cAMP elevation or mutation of PKA-targeted sites affect the in vitro kinase activity of SIK2 towards peptide or protein substrates, implicating another role for the phosphorylation we observed. A reported consequence of phosphorylation-induced binding of proteins to 14-3-3 is a change in their subcellular localization, commonly an accumulation in the cytosol [25]. Ser 577 phosphorylation of exogenously expressed SIK1 has been shown to result in its export from the nucleus of Y1 cells and 3T3-L1 fibroblasts in response to cAMP elevation,  The adipocytes were then homogenized and centrifuged at 60 000 g to obtain cytosol and crude particulate fractions. The fractions were identified using markers for cytosol (GAPDH), nuclei (H3), plasma membrane (Na/K ATPase) and mitochondria (COX IV). HA-SIK2 protein was analysed. Since the expression level of the Ser358Ala mutant was often lower than the wild-type, the non-stimulated sample for each construct was set to 100 %. Quantified data are presented as means + − S.D. for seven to nine independent experiments. Blots shown are representative. *P < 0.05, **P < 0.01.
In previous studies of SIK2 in 3T3-L1 fibroblasts, the wild-type protein was, in contrast with SIK1, virtually absent from the nucleus of resting cells, and did not change in response to cAMP [3], similar to our results in HEK-293 cells. However, on the basis of an increased presence of the Ser587Ala SIK2 mutant in the nucleus, compared with the wild-type, as well as the ability of SIK2 to repress gene expression from a Cre reporter, it was concluded by previous investigators that SIK2 is most likely also subject to cAMP-induced nucleo-cytoplasmic shuttling [3]. In our HEK-293 cell experiments, using confocal microscopy and two differently tagged versions of the kinases, we found no evidence of a difference in localization between wild-type, Ser587Ala or Ser358Ala SIK2, or following forskolin treatment. On the basis of the present and previous studies, the subcellular localization and its potential regulation by cAMP, appears to be distinctly different for SIK1 and SIK2, at least when these kinases are exogenously expressed in 3T3-L1 fibroblasts or HEK-293 cells. This is probably due to the presence of an NLS, located close to Ser 577 in SIK1, whereas no such sequence is apparent in SIK2 [14]. Exactly how phosphorylation of SIK1 induces its relocalization to the cytoplasm has not been determined, but 14-3-3 binding to Ser 577 , thereby masking the NLS, could be a potential mechanism, a possibility that to our knowledge has not been investigated. Adipocytes represent a more physiologically relevant model for studies of SIK2 than HEK-293 cells, and we therefore also wanted to investigate potential regulation of SIK2 localization in these cells. Primary adipocytes are not suitable for use in immunocytochemistry experiments, due to their large lipid droplet and thin cytoplasm, and we therefore instead used a fractionation approach. Since our studies in HEK-293 cells did not support a relocalization of SIK2 from the nucleus, we prepared a crude particulate fraction in order to also monitor non-nuclear membranes. In these experiments, we detected a significant increase in the wild-type HA-SIK2 protein in the cytosol following cAMP elevation, and this was paralleled by a reduced amount of SIK2 in the particulate fraction. This relocalization was mediated by phosphorylation of Ser 358 , as the PKA-insensitive Ser358Ala mutant did not display any change in localization. From the present study we cannot point out which membrane compartment, nuclear or extra-nuclear, that SIK2 relocates from in adipocytes. Nevertheless, we hypothesize that phosphorylation-induced relocalization of SIK2, mediated by 14-3-3 binding, is important for its cellular function, possibly restricting its access to downstream substrates.
In agreement with previous studies in which SIK2 activity was not significantly affected by AICAR or phenformin, in mouse embryonic fibroblasts, HeLa cells [8] or in skeletal muscle [10], the present study does not support an activation of SIK2 in response to changes in cellular AMP/ATP levels in 3T3-L1 adipocytes. AMPK activation upon AICAR treatment or other agents that increase the AMP/ATP ratio [28] requires the binding of AMP to the γ -subunit of AMPK [29,30]. AMP binding or binding to any protein with the characteristics of a γ -subunit has not yet been demonstrated for SIK2 or any of the other AMPKrelated kinases [9].
In the liver, insulin has previously been proposed to induce an activation of SIK2 via PKB phosphorylation of Ser 358 [11]. This regulation was also studied in HEK-293 cells, in which a phosphorylation site mutant (Ser358Ala) was shown not to be phosphorylated. Another study reported SIK2 to be phosphorylated at Ser 587 , rather than Ser 358 , upon insulin stimulation of brown adipocytes [31]. In the present study using 3T3-L1 adipocytes, although PKB was clearly activated, we did not observe an insulin-induced phosphorylation of endogenous SIK2. Our results do not support a role for insulin in the regulation of Ser 358 or Ser 587 phosphorylation of SIK2 in adipocytes, but rather suggest that these sites are induced by cAMP. The fact that PKB was not activated in response to forskolin suggests that the SIK2 signal observed with an antibody recognizing PKB consensus motifs following forskolin treatment is due to the overlap between PKA and PKB consensus sequences. We also used a PKB inhibitor, as well as a MAPK activator and inhibitor, to rule out that these signalling pathways were involved in the forskolin-induced phosphorylation observed with the phosphoPKB and phosphoPKA substrate antibodies (results not shown).
A key question that remains to be addressed is the biological function of SIK2 in adipocytes, and how it is affected by the regulation we describe. Recently, a total body SIK2 knockout mouse model was described, with no major effects on body mass [38]. One could speculate that other SIK isoforms are upregulated as a compensatory mechanism in these mice, and a more precise examination of the adipose tissue is needed to reveal the role of SIK2 in adipose tissue. In other cell types, including β-cells, melanocytes and hepatocytes, SIKs are suggested to repress gene expression by phosphorylating, and thereby reducing the nuclear presence of, transcriptional regulators, like CRTCs [11,15,38,39]. cAMP/PKA-mediated phosphorylation has been suggested to inhibit SIK2 function, leading to CRTC activation. However, no precise mechanism underlying this inhibition was presented in these studies. In the present study we have shown that, in adipocytes, cAMP signals may restrict, or potentially stimulate, SIK2 action, not by affecting its intrinsic kinase activity, but rather by inducing its phosphorylation of Ser 358 , binding to 14-3-3 and a subsequent movement away from potential substrates. We hypothesize that SIK2 in this manner could mediate effects of cAMP on adipocyte gene expression or participate in the acute regulation of lipid storage by cAMP, such as catecholamine-induced lipolysis.

AUTHOR CONTRIBUTION
Emma Henriksson designed and performed most of the experiments and wrote the paper; Olga Göransson was responsible for the conception and design of the study, and reviewed and edited the paper prior to submission; Kei Sakamoto provided essential reagents and scientific advice, and reviewed and edited the paper prior to submission; Nicholas Morrice provided scientific advice, performed the phosphopeptide mapping and MS analysis, and reviewed and edited the paper prior to submission; Helena Jones, Kashyap Patel and Mark Peggie performed confocal microscopy analysis, labelling of peptide for Edman sequencing and cloning respectively, and reviewed the paper prior to submission.