Lithium Can Relieve Translational Repression of TOP mRNAs Elicited by Various Blocks along the Cell Cycle in a Glycogen Synthase Kinase-3- and S6-Kinase-independent Manner*

TOP mRNAs are translationally controlled by mitogenic, growth, and nutritional stimuli through a 5 (cid:1) -ter-minal oligopyrimidine tract. Here we show that LiCl can alleviate the translational repression of these mRNAs when progression through the cell cycle is blocked at G 0 , G 1 /S, or G 2 /M phases in different cell lines and by vari- ous physiological and chemical means. This derepressive effect of LiCl does not involve resumption of cell division. Unlike its efficient derepressive effect in mitotically arrested cells, LiCl alleviates inefficiently the repression of TOP mRNAs in amino acid-deprived cells and has no effect in lymphoblastoids whose TOP mRNAs are constitutively repressed even when they are proliferating. LiCl is widely used as a relatively selective inhibitor of glycogen synthase kinase-3. However, inhibition per se of this enzyme by more specific drugs failed to derepress the translation of TOP mRNAs, implying that relief of the translational repression of TOP mRNAs by LiCl is carried out in a glycogen synthase kinase-3-independent manner. Moreover, this effect is apparent, at least in some cell lines, in the absence of S6-kinase 1 activation and ribosomal protein S6 phosphorylation, thus further supporting the notion that translational control of TOP mRNAs does not rely on either of these variables. described from

TOP mRNAs are translationally controlled by mitogenic, growth, and nutritional stimuli through a 5-terminal oligopyrimidine tract. Here we show that LiCl can alleviate the translational repression of these mRNAs when progression through the cell cycle is blocked at G 0 , G 1 /S, or G 2 /M phases in different cell lines and by various physiological and chemical means. This derepressive effect of LiCl does not involve resumption of cell division. Unlike its efficient derepressive effect in mitotically arrested cells, LiCl alleviates inefficiently the repression of TOP mRNAs in amino acid-deprived cells and has no effect in lymphoblastoids whose TOP mRNAs are constitutively repressed even when they are proliferating. LiCl is widely used as a relatively selective inhibitor of glycogen synthase kinase-3. However, inhibition per se of this enzyme by more specific drugs failed to derepress the translation of TOP mRNAs, implying that relief of the translational repression of TOP mRNAs by LiCl is carried out in a glycogen synthase kinase-3independent manner. Moreover, this effect is apparent, at least in some cell lines, in the absence of S6-kinase 1 activation and ribosomal protein S6 phosphorylation, thus further supporting the notion that translational control of TOP mRNAs does not rely on either of these variables.
TOP mRNAs 1 encode ribosomal proteins, elongation factors, and several other proteins associated with the assembly or function of the translational apparatus (1). The translation of these mRNAs is selectively repressed when proliferation of vertebrate cells is blocked by a wide variety of treatments or when the cells are amino acid-starved (2). The shift of TOP mRNAs from polysomes into messenger ribonucleoprotein particles (subpolysomal fraction) under such circumstances clearly indicates that the translational repression results from a blockage at the translational initiation step (3). Translational activation of TOP mRNAs is apparent upon resumption of cell division when resting cells are induced to grow (increase in their mass) or when amino acid-starved cells are refed (4,5). It is noteworthy that human lymphoblastoid cell lines exhibit constitutive translational repression of TOP mRNAs, even when mitotically active (5)(6)(7).
A recent report has shown that transduction of growth and mitogenic signals into translational efficiency of TOP mRNAs involves the PI3-kinase-mediated pathway (5). Likewise, translational activation of these mRNAs by amino acids relies on the integrity of this pathway (4). Thus, recruitment of TOP mRNAs into polysomes by serum-or amino acid-refeeding can be abolished by pharmacological blocking of PI3-kinase with LY294002 as well as by overexpression of a dominantnegative mutant of PI3-kinase or of its effectors, 3-phosphoinositide-dependent kinase 1 and PKB (4,5).
PKB has been implicated through bioinformatic genetic and biochemical approaches in the phosphorylation of multiple cytoplasmic and nuclear substrates (8). One attractive substrate is GSK-3, which plays a key role in numerous signaling pathways involving various cellular processes ranging from glycogen and protein synthesis to cell cycle regulation and differentiation (9). GSK-3 appears in two closely related isoforms, GSK-3␣ and GSK-3␤, which are ubiquitously expressed in mammalian tissues (10). However, their activity is repressed when phosphorylated by PKB or upon LiCl treatment (for review see Ref. 11).
For decades, lithium salts have provided effective pharmacotherapy for bipolar disorder yet the underlying mechanism is still unknown (Ref. 12 and references therein). Lithium has been shown to be a direct and reversible inhibitor of GSK-3 (13) by acting as a competitive inhibitor of Mg 2ϩ (14) and by inducing the inhibitory phosphorylation of GSK-3 on either serine 21 of GSK-3␣ or serine 9 of GSK-3␤ (15). In addition, lithium has been shown to inhibit other targets including inositol monophosphatase, other structurally related phosphomonoesterases, phosphoglucomutase, and possibly other enzymes (Ref. 12 and references therein). Potent and more selective low molecular weight organic GSK-3 inhibitors have recently been developed in attempts to specifically interfere in cellular processes regulated by these enzymes (16,17).
The temporal relationship between the translational activation of TOP mRNAs as well as S6K activation and the phosphorylation of its substrate, rpS6, led to a model that proposed a causal relationship between these variables (18). However, using a wide variety of genetic manipulations and biochemical approaches, it has recently been shown that neither the activation of S6K nor rpS6 phosphorylation is necessary for translational activation of TOP mRNAs upon mitogenic, growth, or nutritional stimuli (4,5,19,20).
Here we show that LiCl can alleviate the translational repression of TOP mRNAs when cells are withdrawn from the cell cycle at various phases. This effect of LiCl does not involve resumption of cell division, activation of S6K, or phosphorylation of rpS6 and does not rely on inhibition of GSK-3. In contrast, translational repression elicited by amino acid deprivation is partially resistant to LiCl treatment, whereas that observed in dividing lymphoblastoids is completely resistant.

EXPERIMENTAL PROCEDURES
Cell Culture and DNA Transfection-HEK 293 cells were grown and transfected as well as starved for serum or amino acids as described previously (4,5,21). P1798.C7 mouse lymphosarcoma cells were grown as suspension cultures and mitotically arrested by treatment with 0.1 M dexamethasone (Sigma) for 24 h and hormonally withdrawn as described previously (7). NIH 3T3 mouse fibroblasts were grown as monolayer (7) and arrested by either 24 h treatment with 10 mM hydroxyurea (Sigma) or by splitting a confluent culture 1:2 and maintaining the cells for 5 days without further splitting (contact inhibition). The proliferation of contact-inhibited cells was quantified by the methylene blue-staining protocol (22). HeLa 229 cells were grown as monolayer as described previously (6) and arrested by a 24-h treatment with 15 M nocodazole (Sigma). WHI1660, a human lymphoblastoid cell line, was grown and manipulated as described previously (5). LY294002 (Cell Signaling Technology) was added at 50 M. CHIR 98014 and CHIR 99021 (Chiron) were added at 0.55 and 3.8 M, respectively (5-fold their respective EC 50 values) (17).
Polysomal Fractionation and RNA Analysis-These were performed as described previously (4). The isolated fragment probes used in the Northern blot analysis were a 0.97-kb fragment bearing the rpL32 processed gene 4A (23) and a 1.15-kb PstI fragment containing mouse ␣-actin cDNA (24).
Western Blot Analysis-Immunoblotting was performed as described previously (25). Anti-phospho rpS6 (Ser 240/244 ), anti-phospho-GSK-3␤ (Ser 9 ), and anti-phospho-S6K1 (Thr 389 ) antibodies were purchased from Cell Signaling Technology. The preparation and the specificity of the antibodies against rpS6 and its phosphorylated derivatives have been described previously (4). Polyclonal anti-rpS6 antibody was raised against a peptide corresponding to non-phosphorylatable residues 185-205 of mouse rpS6 (LQHKRRRIALKKQRTKKNKEE) and was kindly provided by Cell Signaling Technology. Monoclonal anti-␤-catenin was from BD Transduction Laboratories. AT180 (Pierce Endogen) is a monoclonal antibody that recognizes tau when it is phosphorylated at Thr 231 (Ref. 26 and references therein). Tau-5 is a phosphorylationindependent tau antibody (Ref. 26 and references therein).

LiCl Relieves the Translational Repression of TOP mRNAs in Cells Arrested at Various Phases of the Cell Cycle-
The identification of the PKB as a crucial mediator along the pathway transducing serum signals into translational efficiency of TOP mRNAs has prompted us to look for its effector. One such candidate is GSK-3, which exerts an inhibitory effect on the synthesis of both glycogen and proteins when unphosphorylated (27). Its activity is repressed when phosphorylated by PKB or upon LiCl treatment, and thereby its inhibitory effect is abolished (11). Therefore, we set out to examine whether LiCl can relieve the translational repression of TOP mRNAs in quiescent cells. First, we applied LiCl (20 mM) to HEK 293 cells that were arrested at G 0 by 40 h of serum starvation. Inhibition of GSK-3␤ by LiCl is associated with the phosphorylation of Ser 9 as can be judged by the gradual enhancement of the phosphorylation upon treatment of serumstarved HEK 293 cells with increasing concentrations of LiCl ( Fig. 1a) (15,28). Concomitantly, treatment of these cells with 1 mM LiCl for the last 18 h of starvation was sufficient to elevate polysomal association of rpL32 mRNA, whereas 20 mM LiCl elicited derepression of rpL32 mRNA nearly as efficiently as that elicited by serum refeeding (Fig. 1, b and c).
Similarly, the efficient translation of TOP mRNAs could be rescued by LiCl (60 mM) treatment of NIH 3T3 cells withdrawn from the cell cycle by contact inhibition (arrested at G 0 ) (Fig.  2a). The specificity of the effect of the lithium ion is exemplified by the failure of 60 mM KCl to relieve the translational repression in contact-inhibited cell (data not shown). Furthermore, FIG. 1. LiCl alleviates the translational repression of TOP mRNA in cells arrested at G 0 . a, HEK 293 cells were serum-starved for 40 h. LiCl was added at the indicated concentrations during the last 18 h of starvation. Cells were harvested, and the cytoplasmic proteins were subjected to Western blot analysis using the indicated antibodies. b, HEK 293 cells were treated as in a and harvested, and cytoplasmic extracts were prepared. These extracts were centrifuged through sucrose gradients and separated into polysomal and subpolysomal fractions. RNA from equivalent aliquots of these fractions was analyzed by Northern blot hybridization with cDNAs for rpL32 and actin. The radioactive signals were quantified by PhosphorImager, and the relative translational efficiency of these mRNAs is graphically presented as the percentage of the mRNA engaged in polysomes (each point represents an average of two independent measurements). c, HEK 293 cells were serum-starved for 40 h (ϪSerum), starved for 40 h with LiCl (20 mM) treatment for the last 18 h, or serum-starved for 40 h and then serum-refed for 3 h. All types of cells were harvested, and cytoplasmic extracts were prepared. These extracts were centrifuged through sucrose gradients and separated into polysomal (P) and subpolysomal (S) fractions. RNA from equivalent aliquots of these fractions was analyzed by Northern blot hybridization with cDNAs for actin and rpL32. The radioactive signals were quantified by PhosphorImager, and the relative translational efficiency of each mRNA is numerically presented beneath the autoradiograms as the percentage of the mRNA engaged in polysomes. These figures were expressed as the mean Ϯ S.E. of the number of determinations within parentheses or the mean Ϯ S.D. with the individual values within parentheses if only two determinations were available.
LiCl exerts its effect in these cells without inducing mitogenic activity as demonstrated in Fig. 2b.
We were intrigued by the question of whether the capacity of LiCl to rescue the translation efficiency of TOP mRNAs is confined to cells arrested at G 0 or whether LiCl functions similarly when cells are blocked at other phases of the cell cycle. To address this issue directly, we arrested NIH 3T3 cells at G 1 /S phase by applying hydroxyurea, which inhibits ribonucleotide reductase and thereby DNA synthesis (29). Likewise, we blocked HeLa cells at G 2 /M phase by nocodazole treatment, which inhibits the polymerization of microtubules and thereby the assembly of the mitotic spindle (30). Both these treatments led to selective translational repression of rpL32 mRNA as demonstrated by its unloading from polysomes, yet this repression could be relieved by LiCl (Fig. 2, a and c). Therefore, it appears that LiCl can rescue the translation of TOP mRNAs in resting cells, regardless of the cell type, the means used for mitotic arrest, and the phase along the cell cycle at which the block was set.
It has been shown previously that serum replenishment leads to inhibition of GSK-3␤ as judged by phosphorylation of Ser 9 (28,31). Nevertheless, our results clearly demonstrate indistinguishable levels of phosphorylation of these sites in untreated or serum-starved HEK 293 cells as well as in untreated NIH 3T3 and HeLa cells and their mitotically arrested counterparts (Fig. 3). These results might reflect the fact that GSK-3␤ is not significantly activated when these cells cease to proliferate, and consequently, the apparent translational repression does not result from GSK-3 activation. Alternatively, GSK-3␤ is activated upon mitotic arrest in a Ser 9 phosphorylation-independent fashion and thus still may be involved in translational repression of TOP mRNAs.

Inhibition of GSK-3 Is Not Sufficient to Relieve the Translational Repression of TOP mRNAs in Quiescent Cells-P1798
mouse lymphosarcoma cells can be arrested at G 0 by dexamethasone treatment, and even though serum was available, this block led to selective translational repression of TOP mRNAs ( Fig. 4) (32). Withdrawal of the hormone induced the translation of TOP mRNAs in a PI3-kinase-dependent manner as demonstrated by the ability of LY294002 (a PI3-kinase inhibitor) to completely suppress the translational activation of TOP mRNAs (Fig. 4). In contrast, LiCl could completely rescue TOP mRNAs from repression by dexamethasone when both these reagents were added together (compare the polysomal association of rpL32 mRNA in dexamethasone-treated cells with or without LiCl to that in hormonally withdrawn cells in Fig. 4). Interestingly, unlike the treatments leading to mitotic arrest in HEK 293, NIH 3T3, and HeLa cells, dexamethasone treatment by itself was able to enhance, at least partially, the phosphorylation of GSK-3␤ at Ser 9 (Fig. 3). Conceivably, if inhibition of GSK-3 is indeed necessary for derepression of TOP mRNAs, this enzyme remained sufficiently active in dexamethasonetreated P1798 cells. Alternatively, inactivation of GSK-3 plays no role in the relief of the translational repression of TOP mRNAs.
To further examine these two possibilities, we selected newly developed aminopyrimidine derivatives, CHIR 98014 and CHIR 99021, that potently and specifically inhibit GSK-3 (17). The mechanism underlying their inhibitory effect does not involve significant phosphorylation of GSK-3␤ at Ser 9 (Fig. 5, a  and c). Hence, to ascertain their capacity to inhibit this en- and LiCl (20 mM) or nocodazole-treated for 24 h, and then withdrawn from the drug for 24 h. The polysomal distribution of mRNAs encoding rpL32 and actin was analyzed and presented in a and c as described in the legend to Fig. 1. P, polysomal; S, subpolysomal. zyme, we employed an indirect assay of GSK-3 activity based on the apparent accumulation of cytoplasmic ␤-catenin protein in response to inactivation of GSK-3 (Ref. 15 and references therein). GSK-3 was indeed active in dexamethasone-treated P1798 cells (␤-catenin level was very low). However, when these cells were cotreated with CHIR 98014 for 24 h or with CHIR 99021 for 8 or 24 h, their GSK-3 was inactivated as judged by the robust accumulation of ␤-catenin (Fig. 5a). Nevertheless, this inhibition of GSK-3 failed to alleviate the translational repression of rpL32 mRNA, even though 3 h of hormonal withdrawal was sufficient for efficient recruitment of this mRNA into polysomes (Fig. 5b).
The failure of the two examined aminopyrimidine derivatives to relieve the translational repression of rpL32 mRNAs in P1798 cells might reflect an exceptional behavior of this cell type or that of the readout of the assay for GSK-3 activity. To address this issue directly, we also examined the effect of CHIR 99021 in HEK 293 cells. First, we set out to examine whether CHIR 99021 can inhibit GSK-3 in these cells. We took advantage of the fact that GSK-3 is known to phosphorylate the neuronal protein tau at Thr 231 (26). Hence, HEK 293 cells were cotransfected with tau and GSK-3␤ expression vectors. Fig. 5c shows that CHIR 99021 inhibited tau phosphorylation as efficiently as LiCl. We next demonstrated that CHIR 99021 failed to derepress the translation of rpL32 mRNA in serum-starved cells (Fig. 5d), unlike the apparent ability of LiCl to exert such an effect under similar conditions (Fig. 1a). Taken together, our results clearly show that translational activation of TOP mRNAs by LiCl cannot be attributed to the inhibition of GSK-3, regardless of the nature of the examined cell line, the means used for mitotic arrest, or the assay applied for monitoring GSK-3 activity.
LiCl Can Only Partially Rescue the Translation of TOP mRNAs in Amino Acid-starved Cells-The apparent ability of LiCl to alleviate the translational repression of TOP mRNAs in mitotically arrested cells and the fact that amino acids can transiently inactivate GSK-3 (33, 34) prompted us to extend this study to amino acid-starved cells. rpL32 mRNA is translationally repressed by 2 h of amino acid starvation (Fig. 6a). LiCl treatment starting 16 h before deprivation of amino acids inefficiently derepressed the translation of rpL32 mRNA, as most of this mRNA still remained translationally inactive (only 47% in the polysomal fraction). Conversely, refeeding for just 0.5 h led to efficient recruitment of rpL32 mRNA into polysomes (77% in the polysomal fraction). Notably, LiCl efficiently elicited the phosphorylation of GSK-3␤ on Ser 9 (Fig. 6b), indicating that cells did respond to LiCl treatment. Therefore, we have concluded that amino acid starvation represses the translation of TOP mRNAs through a pathway(s), which differs from that utilized by mitotic blockers, because it is only partially LiCl-sensitive.
LiCl Failed to Derepress the Translation of TOP mRNAs in Dividing Lymphoblastoids-We have previously shown that the translation of a wide variety of TOP mRNAs is constitutively repressed in Epstein-Barr virus-transformed human lymphoblastoids, even when proliferating (5-7). Hence, we set out to examine whether LiCl can alleviate this repression in one such cell line (WHI 1660). Fig. 7A shows that LiCl exerted its typical inhibitory effect on GSK-3␤ (inducing phosphorylation at Ser 9 ). Nonetheless, the translation of rpL32 mRNA remained repressed when these cells were LiCl-treated for 24 h at concentrations ranging from 20 (Fig. 6b) to 80 mM (data not shown). These results suggest that the apparent translational repression of TOP mRNAs in lymphoblastoids functions constitutively in a LiCl-resistant manner.
LiCl Can Relieve the Translational Repression of TOP mRNAs in the Absence of S6K1 Activation or rpS6 Phosphorylation-It has previously been argued that the translation of TOP mRNAs in serum-stimulated cells is induced through activation of S6K1 and phosphorylation of rpS6 (18). Hence, we set out to verify whether either or both of these variables mediate the effect of LiCl. Fig. 3 shows that mitotic arrest of HEK 293 and P1798 cells (at G 0 ), of NIH 3T3 cells (at G 1 /S), and of HeLa cells (at G 2 /M) is associated with inactivation of S6K1. This is exemplified by the dephosphorylation of Thr 389 , a critical site for its activity (35), and of its substrate rpS6 (see FIG. 5. Inhibition of GSK-3 is not sufficient for relieving the translational repression of TOP mRNAs. a, P1798 cells were dexamethasone (10 Ϫ7 M) treated for 24 h without (Ϫ) or with 0.55 M CHIR 98014 for 24 h or with 3.8 M CHIR 99021 for 8 or 24 h. Cells were harvested, and the cytoplasmic proteins were subjected to Western blot analysis as described in the legend to Fig. 2. b, P1798 cells were dexamethasone (10 Ϫ7 M)-treated for 24 h without or with 3.8 M CHIR 99021 (Dexa ϩ 99021) for the last 8 h or throughout the dexamethasone treatment or dexamethasone-treated for 24 h and then hormonally withdrawn for 3 h without any addition (Withdrawal). The polysomal distribution of mRNAs encoding rpL32 and actin was analyzed and presented as described in the legend to Fig. 1. c, HEK 293 cells in 60-mm plates were cotransfected with 200 ng of human tau expression vector and 400 ng of pFLAG-GSK3-␤ (51). 24 h later, cells were serumstarved for 41 h in the absence (Ϫ) or presence of 20 mM LiCl or 3.8 M CHIR 99021 for the last 18 h. Cells were harvested, and the cytoplasmic proteins were subjected to Western blot analysis using the following antibodies: anti-total tau (Tau-5); anti-phospho-tau Thr 231 (AT180), and anti-phospho-GSK-3␤ Ser 9. ␣ and ␤ Represent tau unphosphorylated and tau phosphorylated at Thr 231 , respectively. d, HEK 293 cells were serum-starved for 41 h without or with CHIR 99021 treatment (3.8 M) for the last 18 h. The polysomal distribution of rpL32 and actin mRNAs was analyzed and presented as described in the legend to Fig. 1.   FIG. 6. LiCl relieved only partly the translational repression of TOP mRNAs in amino acid-starved cells. a, HEK 293 cells were amino acid-starved for 2 h (Ϫ amino acids), treated for 24 h with 20 mM LiCl and amino acid-starved for the last 2 h (Ϫ amino acids ϩLiCl), or amino acid-starved for 2 h and then refed for 0.5 h (amino acid refeeding). Cells were harvested, and the polysomal distribution of the mRNAs encoding rpL32 and actin was analyzed and presented as described in the legend to Fig. 1. b, cells were either amino acid-starved (Ϫ) or refed (R) in the absence (Ϫ) or presence (ϩ) of LiCl and harvested, and the cytoplasmic proteins were subjected to Western blot analysis using the indicated antibodies. FIG. 7. LiCl was unable to alleviate the constitutive translational repression of TOP mRNAs in lymphoblastoids. a, cytoplasmic extracts were prepared from proliferating WHI 1660 lymphoblastoid cell line without or with 20 mM LiCl. The polysomal distribution of the mRNAs encoding rpL32 and actin was analyzed and presented as described in the legend to Fig. 1. b, cytoplasmic proteins from the WHI 1660 were subjected to Western blot analysis using the indicated antibodies. P, polysomal; S, subpolysomal. the phosphorylation status of Ser 240 and Ser 244 in Fig. 3). However, LiCl, which very potently rescued the translation of TOP mRNAs in similarly treated cells, failed to activate S6K1 or to induce rpS6 phosphorylation in serum-starved HEK 293 or dexamethasone-treated P1798 cells (Fig. 3). Clearly, the apparent lack of effect of LiCl on the phosphorylation of either of these proteins does not reflect a technical problem because LiCl induced the phosphorylation of Ser 9 in GSK-3␤. However, it should be noted that LiCl partially induced S6K1 activation and rpS6 phosphorylation in hydroxyurea-treated NIH 3T3 cells and nocodazole-treated HeLa cells (Fig. 3) yet the physiological significance of this cell type-specific effect is not clear. Taken together, these data suggest that LiCl can relieve the translational repression of TOP mRNAs in an S6K1 and rpS6 phosphorylation-independent fashion, at least in some cell types.

DISCUSSION
The data presented in this study suggest that inhibitory signals stemming from various blocks along the cell cycle converge into a pathway that leads to repression of TOP mRNA translation. Currently available information regarding the effect of the various mitotic inhibitory treatments used in this study on known signal transduction pathways is quite fragmentary or missing completely. Nevertheless, a tentative model depicting a pathway conveying signals from the mitogenic blocks to the translational repressor (denoted as Y) is presented in Fig. 8. The existence of a putative repressor has previously been inferred from circumstantial evidence (for review see Ref. 1). According to the proposed model, LiCl inhibits the activation of the repressor by blocking an upstream activator (denoted as X in Fig. 8). In contrast, amino acid starvation activates the repressor via a different mechanism, which is only partially lithium-sensitive (Fig. 6). This differential regulation of TOP mRNAs translation is further supported by a recent report, demonstrating that deprivation of both amino acids and serum has an additive repressive effect on the translation efficiency of TOP mRNAs (36). Likewise, the constitutively repressed translation of TOP mRNAs in Epstein-Barr virus-immortalized lymphoblastoids is resistant to LiCl treatment, implying a different mechanism.
We reasoned that GSK-3 is an attractive kinase to comprise the putative activator X in the tentative model. This was based on the following observations. (a) GSK-3 is a direct target of PKB, a critical component in the pathway transducing mitogenic signals into translational efficiency of TOP mRNAs (4, 5). (b) Several of the established GSK-3 targets are active in the cytoplasm (9). (c) GSK-3 has already been implicated in the regulation of protein synthesis through phosphorylation of eIF2B (37). Indeed, we have shown here that LiCl, a GSK-3 inhibitor, can relieve the translational repression of TOP mRNAs in mitotically arrested cells. Nevertheless, we also have demonstrated that translational repression of TOP mRNAs occurred despite inhibition of GSK-3 by CHIR 98014 or CHIR 99021 treatment of P1798 and HEK 293 cells (Fig. 5) or when proliferating lymphoblastoids were treated by lithium (Fig. 7). These two observations appear to be inconsistent with GSK-3 being an activator of translational repression of TOP mRNAs as initially hypothesized. However, it should be noted that overexpression of a constitutively active mutant of GSK-3␤ (GSK-3␤ S9A) (38) in mitotically induced (serumrefed) HEK 293 cells fully suppressed the translational activation of a chimeric TOP mRNA but had no inhibitory effect on a non-TOP chimeric mRNA. 2 Formally, we cannot exclude the possibility that translational activation of TOP mRNAs requires down-regulation of GSK-3. Nonetheless, in light of the other data presented in this report, it is conceivable that the apparent inhibitory effect of the constitutively active mutant simply reflects physiologically irrelevant high activity of GSK-3␤ obtained by its overexpression. Thus, it is possible that non-physiological substrates that are only fortuitously phosphorylated by endogenous GSK-3␤ were significantly phosphorylated to the extent that one or more of them affected the translational efficiency of TOP mRNAs. It should be mentioned that previously reported results derived from overexpression of various active and dominant negative kinases have been subjected to controversial interpretations (20, 39 -47).
An apparent correlation under some physiological circumstances between S6K activation and rpS6 phosphorylation on one hand and translational up-regulation of TOP mRNAs on the other hand has led to the assumption of a cause and effect relationship (18). In this study, we have demonstrated that LiCl can alleviate the translational repression of TOP mRNAs in HEK 293 and P1798 cells, even though S6K remains inactive and rpS6 is unphosphorylated in these cells (Figs. 1, 3, and 4). Therefore, these results contradict the prevailing dogma but are consistent with multiple other observations. (a) Hyperphosphorylation of rpS6 by overexpression of RSK2 (a non-physiological S6 kinase) or inhibition of rpS6 phosphatase by calyculin A failed to relieve the translational repression of TOP mRNAs in amino acid-starved cells (4).  TOP mRNAs is constitutively repressed in dividing lymphoblastoids, even though their S6K1 is active and rpS6 is phosphorylated ( Fig. 6) (5-7). (c) TOP mRNAs are translationally activated by serum refeeding or amino acid replenishment in S6K Ϫ/Ϫ ES cells (4,5) or in mouse erythroleukemia cells (19), even though their rpS6 is constitutively unphosphorylated or dephosphorylated, respectively, indicating that rpS6 phosphorylation is not necessary for efficient translation of TOP mRNAs. (d) Mistargeting of upstream signals by overexpression of a kinase-dead S6K1 mutant completely abolished any S6K activity without a concomitant inhibitory effect on translation of TOP mRNAs (4). (e) Rapamycin, a specific mammalian target of rapamycin inhibitor, completely blocks S6K activity and rpS6 phosphorylation, yet it exerts only a minor or no repressive effect on the translational activation of TOP mRNAs in several cell lines (Ref. 5 and references therein). (f) The translation of TOP mRNAs is normally controlled in mouse cells derived from double knock-out mice lacking both S6K1 and S6K2 (20).
Finally, lithium has been used to treat bipolar disorder for more than five decades and, in addition, it also exhibits neuroprotective qualities (Ref. 48 and 49 and references therein). Despite its clinical usefulness and multiple identified sites of action, the precise mechanism by which lithium exerts its beneficial effect is still unknown. The data presented in this study have extended the list of LiCl targets to include induced synthesis of the translational apparatus through activation of the translation of TOP mRNAs. Nevertheless, this activation cannot account for the neuroprotective or therapeutic properties of lithium because the concentration of LiCl that is required for efficient translational activation of TOP mRNAs is ϳ15-40fold higher than the upper limits of the serum concentration (1.5 mM) allowed in treated patients (www.mentalhealth.com/ drug/p30-l02.html) (50).