INTRODUCTION

Glycogen synthase kinase-3 (GSK3) has recently become recognized as a broadly influential enzyme that modulates many aspects of neuronal function (Grimes and Jope, 2001). GSK3 is now known to exert profound influences on cellular architecture and plasticity, to regulate multiple transcription factors and consequently the expression of many genes, and to contribute to the regulation of apoptosis, and thus cell survival (Frame and Cohen, 2001; Grimes and Jope, 2001). In accordance with these many important actions, the activity of GSK3 is tightly regulated. The phosphorylation state of GSK3 is the most well-characterized mechanism that regulates its activity, although regulation of its subcellular distribution and interactions with GSK3-binding proteins also contribute to controlling its actions (Jope and Johnson, 2004). Both of the two known isoforms of GSK3 are inhibited by phosphorylation of an N-terminal serine, which involves serine-9 in GSK3β and serine-21 in GSK3α. This site can be phosphorylated by several different kinases, including Akt (also known as protein kinase B) (Cross et al, 1995), protein kinase A (Fang et al, 2000; Li et al, 2000), protein kinase C (Goode et al, 1992), and others, indicating that many signaling cascades converge on GSK3 to regulate its activity. Impairments of this inhibitory control of GSK3 can result in abnormally high GSK3 activity, a condition that can have detrimental effects on neural plasticity, structure, and survival (Jope and Johnson, 2004). In opposition to the inhibitory N-terminal serine phosphorylation, tyrosine phosphorylation (tyrosine-216 in GSK3β and tyrosine-279 in GSK3α) increases the activity of GSK3, but this appears to contribute a lesser regulatory effect than serine phosphorylation (Frame and Cohen, 2001) and the mechanisms regulating tyrosine phosphorylation remain unclear. Recently, GSK3β (the isoform most often investigated) has received much attention from investigators studying mood disorders because the mood stabilizer lithium was found to be a direct inhibitor of GSK3β, a finding that raised the possibility that GSK3β may not be adequately controlled in mood disorders (Klein and Melton, 1996; Phiel and Klein, 2001; Jope, 2003).

Impaired monoaminergic neurotransmission has long been linked to depression. This is indicated in part because the majority of antidepressant drugs increase levels of the monoamines serotonin (5HT) and/or norepinephrine, and drugs selectively affecting 5HT are especially in wide use. Thus, pharmacological augmentation of 5HT with monoamine oxidase (MAO) inhibitors, tricyclic antidepressants, or selective serotonin reuptake inhibitors (SSRIs), can be therapeutic for depression (Duman et al, 1997). Furthermore, dysregulation of 5HT neurotransmission in major depression is substantiated by results using a wide variety of experimental approaches, including studies of 5HT receptors and transporters (Middlemiss et al, 2002; Stockmeier, 2003), brain imaging and post-mortem studies (Drevets et al, 1999; Ichimiya et al, 2002; Stockmeier, 2003), 5HT metabolism (Owens and Nemeroff, 1994), and neuroendocrine actions (Yatham and Steiner, 1993). Regardless of whether mood disorders involve impairments in 5HT synthesis, release, or receptor-stimulated actions, any of these can result in deficient downstream signaling activity. Therefore, clarifying the signals generated by stimulation of 5HT receptors is an important goal for understanding the neuropathological underpinnings of mood disorders. Considering the potential importance of regulating GSK3β in mood disorders suggested by the action of lithium, the present study was undertaken to determine if serotonergic activity regulates the phosphorylation of GSK3β. For this purpose, the in vivo phosphorylation of GSK3β was examined in three regions of mouse brain that are known to express serotonergic receptors, the prefrontal cortex, hippocampus, and striatum, following administration of agents that alter serotonergic activity.

MATERIALS AND METHODS

Animals and Treatments

Adult, male C57BL/6 mice (Charles River Laboratories, Inc., Wilmington, MA) were treated with the following agents by intraperitoneal injections: d-fenfluramine (40 mg/kg body weight), clorgyline (2 mg/kg), 8-hydroxy-2-(dipropylamino)tetralin (8-OH-DPAT; 0.3–10 mg/kg), WAY100635 (1 mg/kg), LY53857 (0.3–5 mg/kg), 2,5-dimethoxy-4-iodoprenyl-2-aminopropane (DOI; 20 mg/kg), haloperidol (0.2 mg/kg), imipramine (30 mg/kg; all obtained from Sigma), and fluoxetine (20 mg/kg; obtained from the National Institute of Mental Health's Chemical Synthesis and Drug Supply Program). For intraperitoneal injection of mice, all these drugs were dissolved in saline, and saline was used as vehicle controls.

Tissue Preparation

Mice were anesthetized with carbon dioxide for 15 s, decapitated, and brain regions were rapidly dissected in ice-cold saline. Brain regions were homogenized in ice-cold lysis buffer containing 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.5% NP-40, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 5 μg/ml pepstatin, 0.1 mM β-glycerophosphate, 1 mM phenylmethanesulfonyl fluoride, 1 mM sodium vanadate, and 100 nM okadaic acid. The lysates were centrifuged at 20 800 g for 10 min. Protein concentrations in the supernatants were determined using the Bradford protein assay (Bradford, 1976).

Immunoblotting

Samples were mixed with Laemmli sample buffer (2% SDS) and placed in a boiling water bath for 5 min. Proteins were resolved in 10% SDS-polyacrylamide gels, and transferred to nitrocellulose. Blots were probed with antibodies to phospho-Ser9-GSK3β and total GSK3β (Cell Signaling Technology, Beverly, MA). Immunoblots were developed using horseradish peroxidase-conjugated goat anti-mouse, or goat anti-rabbit IgG, followed by detection with enhanced chemiluminescence, and the protein bands were quantitated with a densitometer.

RESULTS

Serotonergic Activity Regulates GSK3β in Mouse Brain Regions In Vivo

To test if serotonergic signaling activity regulates the phosphorylation of GSK3β in mouse brain in vivo, d-fenfluramine was administered to mice to stimulate 5HT release and reduce its reuptake, and 5HT catabolism was inhibited by administration of the MAO inhibitor clorgyline. Treatment with d-fenfluramine (40 mg/kg, 1 h) alone caused a modest increase in the levels of phospho-Ser9-GSK3β in mouse prefrontal cortex, hippocampus, and striatum (Figure 1a). While treatment with clorgyline alone did not change phospho-Ser9-GSK3β levels, pretreatment with clorgyline (2 mg/kg, 3 h prior to d-fenfluramine) enhanced the increases in phospho-Ser9-GSK3β levels induced by d-fenfluramine treatment in all three brain regions. Following combined treatment with clorgyline plus d-fenfluramine these increases amounted to 410±107% of the control level in the prefrontal cortex (n=11; p<0.01), 425±109% in the hippocampus (p<0.01), and 313±63% in the striatum (p<0.05) (Table 1). These results reveal for the first time that serotonergic signaling activity regulates the phosphorylation of GSK3β.

Figure 1
figure 1

Serotonergic stimulation increases phospho-Ser9-GSK3β in vivo. (a) Mice were treated with d-fenfluramine (d-FEN; 40 mg/kg, 1 h), clorgyline (CLG; 2 mg/kg; 3 h prior to d-FEN), or CLG alone (4 h), and phospho-Ser9-GSK3β levels were measured by immunoblot analysis in prefrontal cortex (PFC), hippocampus (HIP), and striatum (STR). (b) Representative immunoblots of phospho-Ser9-GSK3β (left) and total levels of GSK3β (right), and (c) quantitation of phospho-Ser9-GSK3β levels, showing the time dependence of d-FEN-induced increases in phospho-Ser9-GSK3β in CLG-pretreated mice. Values from a representative experiment are shown as the percent of control values from vehicle (saline)-treated mice, and values at 0 h were obtained in samples from mice treated with CLG alone.

Table 1 Effects of Serotonergic Agents on Phospho-Ser9-GSK3β

The time dependence of d-fenfluramine-induced increases in phospho-Ser9-GSK3β was measured in clorgyline-pretreated mice. Administration of d-fenfluramine rapidly and robustly increased the levels of phospho-Ser9-GSK3β in the prefrontal cortex, hippocampus, and striatum, while the total level of GSK3β remained unaltered (Figures 1b and c). Maximal increases in phospho-Ser9-GSK3β were evident in all three brain regions between 0.5 and 1 h after treatment with d-fenfluramine, followed by a gradual decline towards control levels over the next several hours.

To test if inhibition of monoamine reuptake alone was sufficient to modulate the phosphorylation of GSK3, measurements were made in prefrontal cortex 1 h following administration of fluoxetine (20 mg/kg), an SSRI, or imipramine (30 mg/kg), a tricyclic antidepressant that inhibits the reuptake of both 5HT and norepinephrine (Figure 2). Both treatments caused significant increases in the level of phospho-Ser9-GSK3β, which reached 229±39% of the control level (n=10, p<0.05) after fluoxetine treatment and 268±47% of the control level (n=4, p<0.05) after imipramine treatment, whereas the total levels of GSK3β were not changed. In contrast to fluoxetine and imipramine, administration of the antipsychotic haloperidol (0.2 mg/kg) had no effect on the level of phospho-Ser9-GSK3β.

Figure 2
figure 2

Monoamine reuptake inhibitors increase phospho-Ser9-GSK3β in vivo. (a) Representative immunoblots of phospho-Ser9-GSK3β and total GSK3β, and (b) quantitative values of phospho-Ser9-GSK3β level (percent of saline control), in PFC 1 h after treatment with saline (Ctr), fluoxetine (FLX; 20 mg/kg; n=10), imipramine (IMI; 30 mg/kg; n=4), or haloperidol (HAL; 0.2 mg/kg). Means±SEM; *p<0.05 compared with control (saline) values (ANOVA).

5HT1A Receptor Activation Increases Phospho-Ser9-GSK3β

5HT1A receptor-selective agents were used to begin to investigate the 5HT receptor subtypes that mediate the in vivo 5HT-induced increase in phospho-Ser9-GSK3β. Administration of the 5HT1A receptor antagonist WAY100635 (1 mg/kg; 2 h) alone had no significant effect on the level of phospho-Ser9-GSK3β (Figure 3a, Table 1). However, pretreatment with WAY100635 greatly attenuated the increase in phospho-Ser9-GSK3β levels caused by administration of clorgyline plus d-fenfluramine in prefrontal cortex, hippocampus, and striatum (Figures 3a and b). Conversely, treatment with the 5HT1A receptor agonist 8-OH-DPAT (10 mg/kg; 1 h) alone significantly increased the level of phospho-Ser9-GSK3β in all three brain regions (Figure 3a). These increases amounted to 240±55% of the control level in the prefrontal cortex (p<0.05; n=7), 239±35% in the hippocampus (p<0.05), and 362±94% in the striatum (p<0.05) (Table 1). Examination of the dose–response to 8-OH-DPAT revealed that a dose as low as 0.3 mg/kg (hippocampus) or 1 mg/kg (prefrontal cortex and striatum) caused increased phospho-Ser9-GSK3β, while the total level of GSK3β was unaltered (Figure 3c). These results indicate that activation of 5HT1A receptors increases phospho-Ser9-GSK3β in mouse brain in vivo.

Figure 3
figure 3

Phospho-Ser9-GSK3β is increased by stimulation of 5HT1A receptors. (a) Phospho-Ser9-GSK3β was measured in PFC, HIP, and STR following administration of the 5HT1A receptor antagonist WAY100635 (1 mg/kg; 2 h), CLG (2 mg/kg) 3 h prior to d-FEN (40 mg/kg; 1 h), CLG 3 h and WAY 100635 1 h prior to d-FEN, or the 5HT1A receptor agonist 8-OH-DPAT (10 mg/kg; 1 h). The total level of GSK3β also was measured in each sample and was not changed by any of these treatments (data not shown). (b) Quantitative values (given as the percent of saline controls) of the stimulation of phospho-Ser9-GSK3β following treatment with CLG prior to d-FEN (shaded bars) and the attenuation of that stimulation provided by pretreatment with WAY100635 (black bars) (treatment protocols as described in (a)) in PFC, HIP, and STR. Means±SEM. *p<0.05 compared with control (saline) values (ANOVA). (c) Immunoblots and quantitative values of concentration-dependent effects of treatment with 8-OH-DPAT (0, 0.3, 1.0, and 3.0 mg/kg; 1 h) on phospho-Ser9-GSK3β (closed circles) and total GSK3β (open circles) in PFC, HIP, and STR. Values from a representative experiment are shown as the percent of values from control (saline-treated) mice.

Phospho-Ser9-GSK3β is Increased by Blocking 5HT2 Receptors

To examine if regulation of 5HT2 receptors influences the phosphorylation of GSK3β, we utilized the 5HT2 receptor agonist DOI and antagonist LY53857. Whereas activation of 5HT2 receptors with DOI (20 mg/kg; 1 h) did not change phosphorylation of GSK3β (Figure 4a, Table 1), administration of LY53857 (5 mg/kg; 2 h) alone caused a large increase in phospho-Ser9-GSK3β in prefrontal cortex (346±97% of control level, n=7, p=0.055), hippocampus (193±30%; p<0.05), and striatum (378±60; p<0.01) without changing the total level of GSK3β (Figure 4a, Table 1). The duration of the increased phospho-Ser9-GSK3β following blockade of 5HT2 receptors was examined by making measurements 1, 2, 4, 6, and 24 h after a single administration of LY53857 (5 mg/kg) (Figure 4b). Treatment with LY53857 caused a rapid and prolonged increase in phospho-Ser9-GSK3β in all three brain regions followed by a return to control levels after 24 h. Examination of the dose–response to LY53857 demonstrated that increased phosphorylation of GSK3β was dose dependent with the effects more prominent in the prefrontal cortex and striatum than in the hippocampus (Figure 4c). The lack of an effect of DOI and the increases in phospho-Ser9-GSK3β caused by LY53857 indicate that endogenous activity of 5HT2 receptors contributes to the maintenance of dephosphorylated GSK3β and that maximal signaling of 5HT2 receptors to reduce phospho-Ser9-GSK3β was achieved by endogenous 5HT in the absence of further stimulation.

Figure 4
figure 4

Phospho-Ser9-GSK3β is increased by blocking 5HT2 receptors. (a) Phospho-Ser9-GSK3β was measured by immunoblots in PFC, HIP, and STR following administration of the 5HT2 receptor agonist DOI (20 mg/kg; 1 h) or the 5HT2 receptor antagonist LY53857 (5 mg/kg; 2 h). Immunoblots of total GSK3β show that it was unaltered by treatment with LY53857. (b) Quantitation of phospho-Ser9-GSK3β levels, showing the duration of increases in phospho-Ser9-GSK3β induced by LY53857 (5 mg/kg) in three mouse brain regions. (c) Concentration-dependent effects of treatment with LY53857 (0, 0.3, 1, 3, and 5 mg/kg; 2 h) on phospho-Ser9-GSK3β. Values in (b) and (c) are shown as the percent of values from control (saline-treated) mice. (d) LY53857 was administered 1 h prior to treatment with d-FEN (40 mg/kg, 1 h) in CLG-pretreated mice. (e) 8-OH-DPAT (10 mg/kg; 1 h) was administered alone or 1 h after LY53857 (5 mg/kg; 2 h).

These findings were used to examine the effect of blocking 5HT2 receptors, which normally mediate dephosphorylation of GSK3β, on the phosphorylation of GSK3β caused by stimulating serotonergic activity. Pretreatment with LY53857 to block 5HT2 receptors followed by administration of clorgyline plus d-fenfluramine resulted in a very large increase in phospho-Ser9-GSK3β (Figure 4d). This supports the conclusion that activation of 5HT2 receptors following administration of clorgyline plus d-fenfluramine was inhibitory towards the phosphorylation of GSK3β, an effect opposite to that of 5HT1A receptors. To further examine these opposing actions of the 5HT1A and 5HT2 receptors, 5HT1A receptors were stimulated by administration of 8-OH-DPAT and 5HT2 receptors were blocked by treatment with LY53857. These results showed that blocking 5HT2 receptors allowed a greater 8-OH-DPAT-induced increase in phospho-Ser9-GSK3β than was caused by administration of the 5HT1A agonist alone (Figure 4e). Taken together, these results demonstrate that serotonergic regulation of phospho-Ser9-GSK3β is achieved by a balance between the opposing actions of 5HT1A and 5HT2 receptors (Figure 5).

Figure 5
figure 5

Schematic depiction of the regulation of GSK3β activity by 5HT1A and 5HT2 receptors.

DISCUSSION

The results of this study show that enhanced serotonergic activity in several regions of mouse brain increases the inhibitory Ser9-phosphorylation of GSK3β. It is of interest that this effect also is caused in vivo by a therapeutically relevant concentration of lithium (De Sarno et al, 2002). Examination of receptor subtypes involved in the serotonergic regulation of phospho-Ser9-GSK3β revealed that stimulation of 5HT1A receptors causes phosphorylation of GSK3β, whereas 5HT2 receptor stimulation mediates dephosphorylation of GSK3β. Thus, the balance between the activities of these 5HT receptors can modulate GSK3β in vivo.

Previous studies have provided evidence that 5HT1A receptors are deficient in major depressive disorder, although this remains a subject of intense investigation (reviewed in Stockmeier, 2003). For example, positron emission tomography measurements of brain 5HT1A receptors identified widespread reductions associated with major depressive disorder (Drevets et al, 1999; Sargent et al, 2000). Furthermore, a recent report showed that 5HT1A receptor-coupled activation of signal transduction systems was reduced in post-mortem brain samples from depressed suicide victims compared with matched controls (Hsiung et al, 2003). Also relevant was their finding that Akt activity is substantially lower in post-mortem brain samples from depressed suicide victims than in control samples (Hsiung et al, 2003). This is of interest because Akt normally causes inhibitory Ser9-phosphorylation of GSK3β (Cross et al, 1995), so its impairment also could contribute to abnormally active GSK3β. Our finding that stimulation of 5HT1A causes Ser9-phosphorylation (inactivation) of GSK3β, raises the possibility that low 5HT1A activation in depression may result in deficient inhibitory control of GSK3β through impaired phosphorylation, resulting in abnormally active GSK3β. However, this postulate is difficult to address directly because we previously found that phospho-Ser9-GSK3β is dephosphorylated post mortem and large variations were observed among post-mortem samples from human subjects (Lesort et al, 1999, and unpublished observations). Only with the development of new methodologies will experiments be able to accurately measure the activity or phosphorylation state of GSK3β in human brain samples.

Opposite to the downregulated 5HT1A receptors, there is substantial evidence that 5HT2 receptors are upregulated in depression (reviewed in Stockmeier, 2003). The present study found that stimulation of 5HT2 receptors supports dephosphorylation (activation) of GSK3β. Thus, it can be surmised that upregulated 5HT2 receptors in depression may be associated with abnormal increases in GSK3β activity in depression. The results also showed that endogenous serotonergic activity, in the absence of any stimulant, acts predominantly through 5HT2 receptors to keep GSK3β dephosphorylated because treatment with the 5HT2 antagonist LY53857 increased phospho-Ser9-GSK3β, whereas treatment with the 5HT1A receptor antagonist WAY100635 had no effect without additional serotonergic stimulation. The basal 5HT2 receptor-mediated dephosphorylation of GSK3β was maximal, because it was unaltered by administration of the 5HT2 agonist DOI. In contrast, the 5HT1A receptor effect could be activated further by stimulation, because administration of the 5HT1A agonist 8-OH-DPAT increased phospho-Ser9-GSK3β, and serotonergic stimulation (with administration of clorgyline plus d-fenfluramine or 5HT reuptake inhibitors) increased phospho-Ser9-GSK3β, an increase that was greatly attenuated by the 5HT1A receptor antagonist WAY100635. Taken together, reduced signaling capacity of 5HT1A receptors which would disable inhibitory regulation of GSK3β, and increased signaling capacity of 5HT2 receptors which activate GSK3β, by acting together could cause a substantial increase in GSK3β activity in main depression. It is not yet known whether these modulatory effects of serotonergic activity on GSK3β emanate directly from 5HT receptor-coupled signaling pathways or if they reflect changes in the activities of neuronal circuits influenced by 5HT, a question that is currently under investigation.

These findings suggest the possibility that reduced serotonergic activity, as appears to occur in depression, is associated with deficient inhibitory control of GSK3β, raising the question of how this condition would affect neuronal function. The answer to this rests with what is known about the actions of GSK3β. This enzyme has extensive effects on neuronal function, including influences on gene expression through the regulation of at least 15 different transcription factors (Jope and Johnson, 2004). Equally important, GSK3β extensively influences neuronal architecture, plasticity, intracellular transport mechanisms, and even survival, generally impairing these events (Grimes and Jope, 2001), so modulation of GSK3 activity can underlie structural dynamics and neural plasticity over time. This may go so far as to include neuronal survival, a relevant possibility because imaging and post-mortem studies indicate that depression is associated with cell loss (Rajkowska, 2002) and GSK3 is well known to promote cell death (Grimes and Jope, 2001). Thus, hyperactive GSK3β associated with dysregulated serotonergic activity could detrimentally affect neuronal structure, plasticity, and survival, and may be amenable to regulation by antidepressant treatments, which is currently under investigation.