Lithium salts are clinically important drugs used to treat bipolar mood disorder. The mechanisms accounting for the clinical efficacy are not completely understood. Chronic treatment with lithium is required to establish mood stabilization, suggesting the involvement of neuronal plasticity processes. CREB (cAMP-response-element-binding protein) is a transcription factor known to mediate neuronal adaptation. Recently, the CREB-co-activator TORC (transducer of regulated CREB) has been identified as a novel target of lithium and shown to confer an enhancement of cAMP-induced CREB-directed gene transcription by lithium. TORC is sequestered in the cytoplasm and its nuclear translocation controls CREB activity. In the present study, the effect of lithium on TORC function was investigated. Lithium affected neither the nuclear translocation of TORC nor TORC1 transcriptional activity, but increased the promoter occupancy by TORC1 as revealed by chromatin immunoprecipitation assay. In a mammalian two-hybrid assay, as well as in a cell-free GST (glutathione transferase) pull-down assay, lithium enhanced the CREB–TORC1 interaction. Magnesium ions strongly inhibited the interaction between GST–CREB and TORC1 and this effect was reversed by lithium. Thus our results suggest that, once TORC has entered the nucleus, lithium as a cation stimulates directly the binding of TORC to CREB, leading to an increase in cAMP-induced CREB target-gene transcription. This novel mechanism of lithium action is likely to contribute to the clinical mood-stabilizing effect of lithium salts.
Lithium salts are clinically important drugs used to treat bipolar mood disorder and other mental diseases. Although some biochemical effects of lithium are known, such as inhibition of GSK3 (glycogen synthase kinase 3), IMPase (inositol monophosphatase) and AC (adenylate cyclase), the molecular mechanisms that underlie the clinical effects of lithium have remained unclear [1,2]. It is noteworthy that the mood-stabilizing effect of lithium develops slowly, over weeks and months of lithium treatment , suggesting that the clinical effects of lithium are secondary to neuronal adaptation and based on a lithium-induced new pattern of gene expression [3,4]. The transcription factor CREB (cAMP-response-element-binding protein) has been implicated in adaptive responses of the brain to various stimuli  and has been suggested to mediate also lithium-induced changes in gene expression [6,7]. Lithium can influence CREB-directed gene transcription at multiple levels. On the one hand, lithium may affect upstream signalling pathways that converge on CREB, including cAMP-, calcium- and GSK3-mediated signalling pathways . On the other hand, a newly discovered CREB co-activator called TORC (transducer of regulated CREB) has been identified recently as a novel lithium target . Lithium was found to enhance cAMP-induced gene transcription mediated by CREB and its DNA binding site, the CRE (cAMP-response element). This stimulation by lithium was not conveyed by protein kinase A, IMPase or GSK3, but was conferred to CREB by TORC .
Cyclic AMP is well-known to promote the phosphorylation of CREB at Ser119 (of the CREB-327 isoform) within the CREB-transactivation domain through protein kinase A, thereby enhancing its association with the histone acetylase paralogues p300 and CBP (CREB-binding protein) . The newly discovered TORC is another essential CREB co-activator [10–14]. Under basal conditions, TORC is sequestered in the cytoplasm through phosphorylation by SIK (salt-inducible kinase) 2 and other members of the AMPK (AMP-activated protein kinase) family of serine/threonine kinases [15–17]. Following exposure to cAMP, TORC is dephosphorylated and translocated to the nucleus, where it associates with the CREB bZip (basic leucine zipper) domain [10–14]. In co-operation with CBP, TORC then increases CREB target-gene expression .
Although TORC has been identified as a novel target of lithium action to regulate gene expression , it was unknown how lithium, in the presence of cAMP, enhances the transcriptional activity conferred by TORC to CREB. Therefore, in the present study, the effect of lithium on TORC function was investigated.
Expression vectors for GST (glutathione transferase)–CREB and GST–CREB R300A were generated by PCR amplification. GAL4–CREB and GAL4–CREB R300A  were used as template DNA. After PCR, the fragments were subcloned into pGEX2T (GE Healthcare) via the BamHI and EcoRI restriction sites. The expression vector for human TORC1 under the control of a T7 promoter in pcDNA3 has been described previously . For ChIP (chromatin immunoprecipitation) assays, an expression vector for N-terminally FLAG-tagged human TORC1 was generated by PCR amplification using TORC1 as template DNA, followed by subcloning into the pcDNA3 vector (Invitrogen). The plasmid 4xSomCRET81Luc has been described previously . The reporter gene construct 5xGal4E1B-Luc  contains the luciferase gene under the control of five repeats of the responsive element for the yeast transcription factor GAL4. For PCR amplification of GAL4–TORC1 and GAL4–TORC11–44 (amino acids 1–44 of TORC1), human TORC1 was used as template DNA. The resulting DNA was subcloned into the GAL4 DNA-binding-domain-coding vector pSG424 . These constructs encode full-length TORC1 or the first 44 amino acids of TORC1 respectively, C-terminally fused to the DNA-binding domain (amino acids 1–147) of GAL4. The expression plasmids for VP16 (virus protein 16)–bZip and VP16–bZip R300A, encoding the bZIP domain of CREB (amino acids 262–327) fused C-terminally to the transactivation domain of VP16, were generated by PCR amplification. GAL4–bZip or GAL4–bZip R300A served as template DNA, followed by subcloning into the VP16 domain-coding vector pHKnt-VP16 . Supplementary Table S1 (http://www.bioscirep.org/bsr/029/bsr0290077add.htm) gives an overview of all primer sequences. The correct sequence of all constructs was verified by DNA sequencing.
Cell culture and transfection
HIT cells (hamster insulinoma tumour cells) (HIT-T15)  were cultured in RPMI 1640 medium supplemented with 10% (v/v) fetal calf serum, 5% (v/v) horse serum, 100 units/ml penicillin and 100 μg/ml streptomycin. For ChIP assays, HIT cells were transiently transfected with 4 μg of expression vector per 6-cm-diameter dish using Metafectene (Biontex) according to the manufacturer's instructions. For mammalian two-hybrid assays, the DEAE–Dextran method  was used to transfect the cells with 2 μg of expression vector and 1 μg of cytomegalovirus green fluorescent protein expression vector per 6-cm-diameter dish to control for transfection efficiency. Co-transfections were carried out with a constant amount of DNA, which was maintained by adding the empty pBluescript vector (Stratagene).
Cells for the GAL4 assay were treated with 2 mM 8-Br-cAMP (8-bromo-cAMP) for 6 h and with 20 mM LiCl for 7 h prior to harvest. For the two-hybrid assay, cells were treated with 1 mM 8-Br-cAMP for 6 h and with 20 mM LiCl 30 h prior to harvest. Luciferase activity was determined 48 h after transfection as described previously .
HIT cells were grown on coverslips in 6-well plates. Treatment with 1 mM or 2 mM 8-Br-cAMP and with 20 mM LiCl was performed for 30 min and 90 min prior to fixation respectively. Fixation and antibody incubations were performed as described previously . An anti-pan-TORC antibody (Calbiochem) was used as a primary antibody against TORC proteins. The secondary antibody used was an AlexaFluor® 488-conjugated anti-rabbit IgG (Invitrogen). The coverslips were mounted on to slides using Vectashield mounting medium with DAPI (4′,6-diamidino-2-phenylindole) (Linaris). The slides were analysed using a Zeiss Axiovert 200 microscope using OpenLab™ 3.1 software for Macintosh OS9.2 (Improvision).
RNA extraction and reverse transcription
Total RNA was extracted from HIT cells using the RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. cDNA was synthesized from total RNA by reverse transcription using oligo(dT)15 primer and M-MLV reverse transcriptase (Promega) according to the manufacturer's instructions. Analysis of the mRNA level of TORC proteins in HIT cells was performed by PCR using primers for human TORC1, TORC2 and TORC3, and hamster β-actin was used as an internal standard (for primer sequences see Supplementary Table S2 at http://www.bioscirep.org/bsr/029/bsr0290077add.htm), followed by agarose-gel electrophoresis and densitometric analysis of the bands.
ChIP and qRT-PCR (quantitative real-time PCR)
ChIP was performed as described previously  with some modifications. Transfected HIT cells were treated with 2 mM 8-Br-cAMP and with 20 mM LiCl for 30 min and 90 min respectively prior to cross-linking with 1% formaldehyde. Cells were harvested, resuspended in cell lysis buffer [10 mM Tris/HCl (pH 8.0), 10 mM NaCl, 0.2% Nonidet P40 and protease inhibitors], and homogenized using a 1 ml Dounce homogenizer. After centrifugation, the pellet was resuspended in nuclei lysis buffer [50 mM Tris/HCl (pH 8.1), 10 mM EDTA, 1% SDS and protease inhibitors]. The DNA was sheared by sonication and the cell debris was pelleted by centrifugation at 18000 g for 10 min at 4°C. The supernatant was collected and 10% was used as an input control. Immunoclearing, immunoprecipitation and purification of DNA were performed as described previously . qRT-PCR was performed using the fluorescent reporter probe method [24a]. The four repeats of the CRE in the 4xSomCRET81Luc plasmid were chosen as the DNA sample used to measure the efficiency of the ChIP assay. The following primers were used: 5′-GCAATAGCATCACAAATTTCACAAA-3′ (forward primer) and 5′-CCGCCCCGACTGCAT-3′ (reverse primer) combined with the fluorescent reporter probe 5′-CGAATTCGCCGGATCTCGAGCTC-3′ [with 5′-Fluorescein and 3′-TAMRA modifications (6-carboxytetramethylrhodamine)]. For qRT-PCR, the reaction mixture from Eurogentec, containing the AmpliTaq® DNA polymerase (Applied Biosystems), was used and prepared according to the manufacturer's instructions. Each sample was measured in triplicate on a 384-well plate (in 18 μl volumes) using the ABI Prism 7900HT Sequence Detection System and was analysed using SDS 2.1 Software (Applied Biosystems).
Expression and purification of GST-fusion proteins from bacteria
Competent cells from Escherichia coli DH5α strain were transformed by heat-shock with expression vectors coding for GST–CREB and GST–CREB R300A. Bacteria were grown in 1 litre of LB (Luria–Bertani) medium at 37°C. The expression of the GST-fusion proteins was induced by incubation with 1 mM IPTG (isopropyl β-D-thiogalactoside) for 4 h. The bacteria were harvested in sonication buffer [20 mM Hepes (pH 7.5), 1 M NaCl, 1 mM DTT (dithiothreitol) and 1 mM PMSF] and stored at −80°C. The suspension was thawed on ice and the bacteria were disrupted by sonication. The cell debris was pelleted by centrifugation at 12000 g for 10 min at 4°C and the GST-fusion proteins were extracted from the supernatant by affinity chromatography using glutathione–agarose beads. The isolated proteins immobilized on agarose were stored on ice.
Labelling of proteins with [35S]
Human TORC1 labelled with [35S] was generated by in vitro transcription/translation using the TNT T7-coupled reticulocyte lysate system (Promega) according to the manufacturer's instructions, including T7 polymerase and [35S]methionine (1000 Ci/mmol) (GE Healthcare). The human TORC1 coding sequence under the control of the T7 promoter in pcDNA3 was used as a template.
GST pull-down assay
The GST pull-down assay using GST–CREB [wt (wild-type) or R300A] and 35S-labelled TORC1 was performed as described previously . Briefly, the agarose-bound proteins were washed with reaction buffer [20 mM Tris/HCl (pH 7.5), 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P40, 1 mM DTT and 1 mM PMSF]. Considering the valency of lithium and magnesium, the salt concentration was balanced with NaCl to 100 mM to prevent any effects occuring which may be the result of changes in ionic strength of the buffer (for further details see Supplementary Table S3 at http://www.bioscirep.org/bsr/029/bsr0290077add.htm). For the binding reaction, 3.5 μl of 35S-labelled TORC1 was added to equal amounts of GST-fusion protein in 250 μl of reaction buffer and incubated overnight at 4°C. The samples were analysed by SDS/PAGE. The gel was dried and the radioactively labelled proteins were visualized using a phosphorimager (Fuji Film BAS 1800 II) and densitometrically analysed using AIDA 4.15 software (Raytest Isotopenmessgeräte).
Statistical analysis (two-way ANOVA with Student's t test) was performed using the software STATISTICA 7 (Statsoft).
Nuclear translocation of TORC proteins
Nuclear translocation of TORC has been shown to be a key regulator of CREB transcriptional activity [10,11,14,25,26]. Using immunocytochemistry, the effect of lithium and cAMP on the cellular localization of endogenous TORC proteins in HIT cells was investigated. HIT cells, like neurons, are electrically excitable  and share with neurons the expression of proteins, such as the scaffold protein JIP (c-Jun N-terminal kinase-interacting protein)-1b/IB (islet brain) 1 , or the transcription factors Pax6 [29,30] and BETA2 (β-cell E box transactivator 2)/NeuroD [29,31]. CREB-directed transcription has been well characterized in HIT cells [8,18,21,27,32,33], including the biology of TORC . Endogenous TORC was mainly retained in the cytoplasm under resting conditions (Figure 1A, top panel). Treatment of HIT cells with 8-Br-cAMP induced the translocation of TORC into the nucleus in a concentration-dependent manner (Figure 1A, middle panel, and Figure 1B). These results are consistent with other reports in which the nuclear translocation of TORC in different cell lines and also in primary hippocampal neurons was inducible by treatment with forskolin, a potent activator of AC [10,14,26,34]. Furthermore, the nuclear translocation of TORC was induced in HIT cells by membrane depolarization induced by high potassium concentrations and blocked by the calcineurin inhibitor cyclosporine A (results not shown), consistent with a previous report . In HIT cells, LiCl did not affect the translocation of TORC (Figure 1A, bottom panel), neither alone nor in combination with 8-Br-cAMP (Figure 1B).
Enhancement of the nuclear accumulation of TORC in HIT cells by cAMP but not by LiCl, as revealed by immunocytochemistry
The pan-TORC antibody used for immunocytochemistry can detect all three isoforms of TORC, as it is directed against the first 42 amino acids which are highly conserved among the TORCs . To identify the isoforms expressed in HIT cells, total RNA was extracted and the expression levels of the different isoforms were examined. This analysis revealed an 8-fold higher expression of TORC1 compared with TORC2 and the absence of TORC3 (see Supplementary Figure S1 at http://www.bioscirep.org/bsr/029/bsr0290077add.htm). It is noteworthy that TORC1 is the isoform which is particularly copious in brain tissues, such as the hippocampus, cerebral cortex and cerebellum [11,26,34]. Therefore we focused on TORC1 in our further investigations.
Transactivation by TORC1
To investigate the effect of lithium on TORC1 transcriptional activity, the GAL4 system was employed. In a luciferase reporter gene assay, HIT cells were transfected with an expression vector coding for TORC1 fused to the DNA-binding domain of the yeast transcription factor GAL4. The promoter of the reporter gene contained five repeats of the GAL4-binding site (Figure 2A). Thereby TORC1 is tethered to the promoter and its own transcriptional activity can be examined independently of its binding to CREB. The GAL4 DNA-binding domain alone exhibited virtually no activity and did not show any response to the stimuli (results not shown). In contrast, GAL4–TORC1 conferred strong transcriptional activity (Figure 2), indicating the presence of a transactivation domain in TORC1 as suggested previously [11,12]. Treatment with 8-Br-cAMP increased the activity of TORC1 3-fold (Figure 2B), and LiCl did not affect the transcriptional activity of TORC1 either alone or in combination with 8-Br-cAMP (Figure 2B). In view of these results, an influence of lithium salts on either the translocation of TORC1 into the nucleus or the transactivation potential of TORC1 seems to be unlikely.
Lack of effect of LiCl on transactivation by TORC1
Effect of lithium on the recruitment of TORC1 to the promoter
To examine whether lithium influences the recruitment of TORC1 to the promoter, we employed a ChIP assay. HIT cells were transfected with an expression plasmid containing a promoter controlled by four repeats of the rat somatostatin CRE. It contains the consensus sequence 5′-TGACGTCA-3′, which constitutes a high-affinity binding site for endogenous CREB . An expression vector coding for TORC1 tagged with a FLAG epitope was also co-transfected (Figure 3A). After treatment, DNA and proteins were cross-linked and precipitated with an antibody against the FLAG epitope. The amount of precipitated DNA was determined by qRT-PCR. In control experiments, pBluescript (as control DNA) was co-transfected with FLAG–TORC1 and did not result in a specific signal in real-time PCR (results not shown). Treatment of HIT cells with LiCl alone or with 8-Br-cAMP alone did not induce a detectable increase in promoter occupancy (Figure 3B). An increase has been shown previously for TORC2 upon forskolin stimulation [13,16]. This system might be not sensitive enough to detect the effect elicited by 8-Br-cAMP alone. In fact, stimulation with forskolin is in general approx. 10-fold more potent when compared with 8-Br-cAMP [18,32]. However, the combination of cAMP and lithium enhanced the promoter occupancy of TORC1 2-fold (P<0.025, n=5; Figure 3B).
Lithium increased the recruitment of TORC1 to the promoter in the presence of cAMP as revealed by ChIP assay in HIT cells
Effect of lithium on the interaction between CREB and TORC1: mammalian two-hybrid assay
A mammalian two-hybrid assay was employed to investigate the effect of lithium and cAMP directly on the interaction between CREB and TORC1. Only the interaction domains of the proteins of interest were used. HIT cells were transiently transfected with the luciferase reporter gene under the control of five repeats of the GAL4-binding site and an expression vector coding for the first 44 amino acids of TORC1 as a GAL4-fusion construct. Either CREB bZip wt (bZip wt) or mutant CREB bZip R300A (bZip R300A) constructs were also co-transfected, each fused to the potent viral protein VP16, or VP16 alone (control) (Figure 4A). Compared with the VP16 control, VP16–bZip wt showed a 2-fold higher basal activity (Figure 4B), demonstrating the interaction between bZip wt and TORC11–44. LiCl strongly increased the interaction between bZip wt and TORC11–44 (Figure 4B). Treatment with 8-Br-cAMP alone had no effect on luciferase activity, whereas treatment with LiCl plus 8-Br-cAMP increased the interaction between TORC11–44 and bZip wt to the same extent as LiCl alone (Figure 4B). The bZip R300A mutant (VP16–bZip R300A) did not show a significant difference from the VP16 control (Figure 4B), indicating a lack of interaction between bZip R300A and TORC11–44. This confirms previous reports in which this mutation disrupted complex formation  and the enhancement by lithium of cAMP-induced CREB target-gene transcription . These results show an enhancement by lithium of the protein–protein interaction between bZip and TORC11–44 that is independent of cAMP. The two-hybrid assay could not be applied to investigate the interaction of the full-length proteins as a result of very high basal activity of GAL4–TORC1 (results not shown).
Enhancement by lithium of the interaction between CREB bZip and TORC11–44, as revealed by the mammalian two-hybrid assay
Effect of lithium on the interaction between CREB and TORC1:
in vitro GST pull-down assay
Recombinant GST–CREB wt and GST–CREB R300A mutant constructs were expressed in bacteria and TORC1 was labelled by in vitro transcription and translation using [35S]methionine. The specificity of the interaction was examined by use of GST alone, which exhibited no remarkable interaction with 35S-labelled TORC1 (6.6% of GST–CREB wt, P<0.002; Figure 5A). In the presence of 20 mM LiCl, the amount of 35S-labelled TORC1 recovered from GST–CREB wt was clearly increased (P<0.035; Figure 5). In contrast, no specific binding of 35S-labelled TORC1 to the mutant GST–CREB R300A was detectable, and there was no effect of treatment with lithium (Figure 5A). A concentration–response curve of LiCl revealed significantly increased binding of 35S-labelled TORC1 to GST–CREB wt from concentrations of 5 mM LiCl (Figure 5B). These results indicate that lithium also enhances directly the interaction between CREB and TORC1 in this cell-free system.
Enhancement by lithium of the protein–protein interaction between CREB and TORC1 in the
in vitro GST pull-down assays
Lithium is known to inhibit GSK3 by competition with magnesium at the non-Mg–ATP magnesium-binding site [35,36]. Therefore the effect of increasing concentrations of magnesium on the interaction between CREB and TORC1 was investigated in the GST pull-down assay. This analysis revealed that magnesium strongly inhibited the interaction in a concentration-dependent manner (Figure 6A): 0.5 mM MgCl2 reduced the interaction by 40% (P<0.03), and in the presence of 10 mM and 20 mM MgCl2 only 25% and 14% of the interaction between CREB and TORC1 remained respectively (Figure 6A). This strong inhibition by high magnesium concentrations was attenuated by 5 mM LiCl (Figure 6B). LiCl thus enhanced the interaction by approx. 3-fold in the presence of 10 mM or 20 mM MgCl2 (Figure 6C). These results indicate an inhibitory effect of magnesium on the interaction between CREB and TORC1, which is reversed by lithium.
Effect of magnesium on the interaction between CREB and TORC1 in the
in vitro GST pull-down assays
CREB activity has been known for years to be induced by Ser119 phosphorylation and CBP/p300 cofactor recruitment [9,37]. In contrast with CBP/p300, the more recently recognized other essential co-activator, TORC, binds to the CREB bZip domain [11,12]. This interaction is independent of CREB Ser119 phosphorylation and extends the functional role of the bZip domain of CREB to include not only DNA binding and dimerization, but also co-activator recruitment. The regulation of TORC by cAMP-activated and other signalling pathways is thus as important for CREB activity as the regulation of CREB Ser119 phosphorylation [10,11,14,38]. This crucial role of the TORC family of co-activators was emphasized further by the recent identification of TORC as the target of a clinically important drug when the mood stabilizer lithium was shown to stimulate CREB target-gene expression through TORC . The present study now defines what step in TORC function is influenced by lithium, and demonstrates that lithium does not affect TORC nuclear translocation and transcriptional transactivation, but increases as a cation the binding of TORC to the CREB bZip domain.
Cyclic AMP- and Ca2+-activated signalling pathways control the nucleocytoplasmic shuttling of TORC [14,39]. Under resting conditions, TORC1 and TORC2 are sequestered in the cytoplasm bound by 14-3-3 proteins . This binding occurs in response to TORC phosphorylation by SIK. Elevated levels of cAMP inhibit SIK kinase activity via the protein kinase A-mediated phosphorylation of SIK at Ser587 (SIK2). In turn, the phosphorylation of TORC decreases and TORC accumulates in the nucleus [14,39]. Our present results confirm a previous report that TORC accumulates in the nucleus upon increased cAMP levels in HIT cells . However, no effect of lithium was detected on the translocation of TORC into the nucleus. Likewise, no effect of lithium on transactivation by TORC was found in the present study. The TORC proteins contain a strong transactivation domain at their C-terminus [11,12]. This can be studied when TORC is fused to the heterologous GAL4 DNA-binding domain and tested for induction of a promoter with GAL4-binding sites, as has been shown previously . The transactivation potential of TORC may not entirely be constitutive, since cAMP stimulated GAL4–TORC1 transcriptional activity (Figure 2B). A possible explanation is offered by NONO (non-POU-domain-containing octamer-binding protein) [p54nrb (nuclear RNA-binding protein, 54 kDa)], as cAMP has been shown to stimulate the binding of TORC to NONO, which then acts as a bridge between the CREB/TORC complex and RNA polymerase II .
Several lines of evidence support the view that, once TORC has reached the nucleus, lithium enhances the binding of TORC to CREB. As revealed by the ChIP assay, the stimulation by lithium of cAMP-induced CREB target-gene expression  is accompanied by an increase in CRE promoter occupancy by TORC. Since lithium does not alter the nuclear translocation of TORC (see above), the increase in promoter occupancy suggests that lithium stimulates TORC–CREB binding. This was then demonstrated directly by the results of the mammalian two-hybrid assay and GST pull-down assay. It is notable that lithium stimulates TORC–CREB binding independently of cAMP and protein kinase A activity, since in the mammalian two-hybrid assay, lithium induced a robust increase in the binding of the TORC N-terminus to the CREB bZip domain in the absence of cAMP, and cAMP had no effect. This conclusion is further supported by the results of the GST pull-down assay. Lithium thus depends on cAMP to stimulate CRE/CREB-directed transcription , because cAMP provides CREB Ser119 phosphorylation/CBP recruitment, as well as nuclear translocation of TORC. Consistent with the previous finding that the stimulation by lithium of CRE/CREB-directed transcription is not mediated via protein kinase A, GSK3 or IMPase , the effect of lithium in the cell-free GST pull-down assay indicates that lithium increases TORC–CREB binding independently of protein kinase A, and also of other enzymes and cellular proteins. Therefore, when taken together, the results of the present study strongly suggest that once TORC has entered the nucleus, lithium as a cation enhances directly the binding of the TORC N-terminus to the CREB bZip domain, resulting in an increase in CREB target-gene expression (Figure 7).
A new mechanism of action for lithium
Both the CREB bZip domain and the TORC N-terminus are highly structured, forming a continuous α-helix and coiled-coil structure respectively [12,41]. On the basis of glutaraldehyde crosslinking experiments, it has been suggested that TORC binds as a tetramer to the CREB dimer . The binding of a cation such as lithium to such a composite structure, we speculate, may induce a conformational change that increases complex stability. The best-studied example of lithium acting as a cation is the competition of lithium with magnesium for one of the two magnesium-binding sites of GSK3, the non-Mg–ATP magnesium-binding site, leading to inhibition of the kinase activity of GSK3 [35,36,42]. Li+ and Mg2+ have similar ionic radii , and it has been suggested that this enables lithium to physically invade the magnesium-binding site within GSK3, where, as a result of its lower charge density, it disrupts catalytic function . A similar model could hold true also for TORC–CREB binding, since in the present study magnesium was found to inhibit the binding of TORC to CREB over a similar range of concentrations that it stimulates GSK3 activity , and lithium antagonized the effect of magnesium, thus increasing TORC–CREB binding through relief from magnesium inhibition. Interestingly, the crystal structure of the CREB bZip domain revealed the presence of a hexahydrated magnesium ion in the cavity between the bifurcating basic regions . The role of this magnesium binding to the CREB bZip domain in the stimulation by lithium of TORC–CREB binding remains to be investigated.
Clinical studies have demonstrated neuronal atrophy, loss of glial cells/neurons and impairments in neuroplasticity and resilience in the brains of patients that suffer from bipolar mood disorder and suggest that chronic treatment with lithium may attain mood stabilization via neuroprotective actions and induction of neuronal adaptation [43–46]. Since CREB plays major roles in mediating adaptive responses and neuroprotection in the brain , the discovery of lithium enhancement of TORC–CREB binding thus suggests its therapeutic relevance. In cortical neuronal cells, for example, the addition of both glutamate and glycine increases the free intracellular magnesium ion concentration of approx. 300 μM to levels that are almost 10-fold higher  and contributes to glutamate-induced excitotoxicity . In view of the results of the present study, it is not unreasonable to suggest that such fluxes in the concentration of magnesium ions could disrupt the TORC–CREB interaction and that lithium could reverse this effect and increase the binding of TORC to CREB, thus increasing CREB activity and the expression of CREB target genes, such as BDNF (brain-derived neurotrophic factor) and Bcl-2 [6,46] that promote cell survival. Furthermore, TORC has been shown to be required for long-term synaptic plasticity in the hippocampus [26,34]. Thus the stimulation by lithium of the binding of TORC to CREB, as demonstrated in the present study, represents a novel mechanism of lithium action that may contribute to the clinical mood-stabilizing effect of lithium salts.
basic leucine zipper
glycogen synthase kinase 3
- HIT cell
hamster insulinoma tumour cell
quantitative real-time PCR
transducer of regulated CREB
virus protein 16
We thank Dr Ralph Krätzner for methodological discussions regarding GST pull-down assays. We also thank Doris Krause for excellent technical assistance.
The work was supported by the Deutsche Forschungsgemeinschaft [grant number SFB 403/A3] (to W.K.); and by the University of Göttingen (Heidenreich von Siebold Programm) (to E.O.). A.H. receives a Georg-Christoph-Lichtenberg thesis grant from the Government of Lower Saxony and is enrolled in the MSc/PhD Program Neurosciences at the University of Göttingen.
These authors contributed equally to this work.