In Vitro Effects of Mood Stabilizers
To examine the effects of mood stabilizing drugs on primary neurosphere formation, cells from the SVZ of 8-week-old male mice were cultured at low cell density (five cells/μL) in the presence of the drugs. LiCl at 0.5–1.0 mM (within the therapeutically relevant concentrations in the CSF) (Table 1), increased the number of primary neurospheres, whereas LiCl at concentrations greater than 3.0 mM showed toxic effects on primary neurospheres (F(5, 61) = 24.69, p < .001) (Fig. 1A). Similarly, at doses equivalent to therapeutically relevant CSF concentrations, both VPA (F(5, 57) = 35.85, p < .001) and CBZ (F(4, 56) = 11.96, p < .001) increased the number of primary neurospheres from the adult SVZ (Fig. 1B, 1C).
Table Table 1.. Therapeutic concentrations of mood stabilizers
Figure Figure 1.. Effects of mood stabilizers on neurosphere formation in vitro. (A–C): Cells from the adult SVZ were cultured in the presence of LiCl (A), VPA (B), or CBZ (HBC complex, Sigma) (C). All drugs increased the numbers of primary neurospheres when added at therapeutic concentrations and decreased the numbers at high concentrations as compared to control (n ≥ 6 for each concentration from at least three independent experiments). (D–F): The resultant primary neurospheres were subcloned to generate secondary and tertiary neurospheres in the absence of the drugs. (D): Single primary neurospheres were triturated and one tenth of the cells were cultured to form secondary neurospheres. The numbers of resultant secondary neurospheres are expressed as per single primary neurospheres. (n = 5 animals, 4–8 neurospheres from each animal). (E): Neurospheres were serially passaged in bulk and cultured in the absence of the drugs at a clonal density (two cells/μL). Cells from the primary neurospheres treated with 0.5 mM LiCl, 60 μM VPA or 5 μM CBZ, produced more secondary and tertiary neurospheres than their control counterpart (n = 5). (F): Cells from the primary neurospheres grown in the presence of 0.5 mM LiCl, 60 μM VPA or 5 μM CBZ were plated at 0.5 cells/μL and cultured in the absence of the drugs. Total cell numbers after 4 or 7 days in vitro were counted (n = 4). (G): Effects of mood stabilizers at the therapeutic concentrations (0.5 mM LiCl, 60 μM VPA and 5 μM CBZ) were examined using cells from the E14.5 medial ganglionic eminence. Mood stabilizer treatment increased primary neurosphere formation, whereas 10 μM PHT or 1.0 μM FLX showed no effect (n ≥ 4). Data represent means ± SEM. *, p < .05. Abbreviations: CBZ, carbamazepine; FLX, fluoxetine; PHT, phenytoin; VPA, valproic acid.
Download figure to PowerPoint
These results could be interpreted in two ways: mood stabilizers could enhance the survival of cultured NSCs,  or mood stabilizers could increase the self-renewal capability of neural stem/progenitor cells  and thereby increase the number of neurosphere-initiating cells, although these possibilities are not mutually exclusive. To distinguish between those possibilities, we first investigated whether or not mood stabilizers prevent cell death by means of counting dead cells by trypan blue exclusion. The number of dead cells 24 hours after the plating was comparable between drug-treated and non-treated cells (14.71 ± 4.07% for control, 15.29 ± 4.18% for LiCl, 13.47 ± 3.81% for VPA, 13.88 ± 3.60% for CBZ, mean ± SEM, F(3, 64) = 0.75, p = .53). We then subcloned primary neurospheres that had been grown in the presence of 0.5 mM LiCl, 60 μM VPA, or 5 μM CBZ and cultured the cells in the absence of the drugs to generate secondary neurospheres. Because primary neurospheres are clonally derived from single NSCs when the adult SVZ cells are cultured at low cell density  (see Discussion), the number of secondary neurospheres derived from each primary neurosphere represents the number of symmetric expansive divisions of the original NSCs during the culture. We cultured one-tenth of the cells from each primary neurosphere to keep the cell densities below two cells/μL and found that a greater number of secondary neurospheres were generated from single primary neurospheres treated with mood stabilizers than those without treatment (F(3, 16) = 7.024, p = .003) (Fig. 1D). It is unlikely that these effects are mediated by histone deacetylase inhibitor activities because trichostatin A, another histone deacetylase inhibitor, neither increased the number of primary neurosphere nor enhanced the self-renewal but, inversely, alleviated the primary neurosphere formation at 10 nM or higher concentrations (F(4, 11) = 35.46, p < .001) (supplemental online Fig. 1).
The self-renewal capacity of NSCs was also analyzed through serial bulk passaging of neurospheres, in which drug-treated or control neurospheres were dissociated in bulk and cultured at a constant low cell density (two cells/μL) in the absence of the drugs (Fig. 1E). Bulk passaging demonstrated that NSCs treated with mood stabilizers in the primary culture exhibited enhanced self-renewal, with increased numbers of both secondary (F(3, 16) = 10.36, p < .001) and tertiary neurospheres (F(3, 16) = 5.46, p = .009) as compared to controls (Fig. 1E). The proliferation of cells from drug-treated primary neurospheres was examined by plating cells at a very low density (0.5 cells/μL) and counting cell numbers 4 or 7 days later (Fig. 1F). Cells obtained from primary neurospheres grown in the presence of mood stabilizers proliferated more rapidly than those from control neurospheres (F(3, 12) = 10.45, p = .001) (Fig. 1F). These results suggest that in vitro treatment with therapeutically relevant doses of mood stabilizers enhances the self-renewal of NSCs from the adult brain.
The effect of mood stabilizers on NSCs derived from the MGE ventricular zone of E14.5 embryos was also evident (F(4, 64) = 10.41, p < .001), whereas another anti-convulsant, phenytoin, or anti-depressant, fluoxetine, had no such effect (Fig. 1G), suggesting that the effect was specific to mood stabilizing drugs. Individual neurospheres treated with mood stabilizers produced βIII tubulin+ neurons, GFAP+ astrocytes, and O4-antigen+ oligodendrocytes under differentiation conditions (online supplemental Fig. 2), suggesting that the multipotentiality of NSCs was not altered by the drug treatment.
In Vivo Effects of Mood Stabilizers
Next, we used the neurosphere assay to determine whether lithium treatment increased the number of neurosphere-forming NSCs in the SVZ of the adult brain. Because each primary neurosphere is clonally derived from a single NSC when cultured at low cell densities, the total number of NSCs in the SVZ can be estimated by counting primary neurospheres [34, 36, 45]. When LiCl was administered to mice for 3 weeks (3 g/L in drinking water), serum lithium concentrations reached 0.86 ± 0.23 mEq/L (mean ± SD), a value close to that observed during therapeutic management of bipolar patients . The total number of neurosphere-forming NSCs in the lateral portion of the SVZ was increased in lithium-treated animals compared to controls, and this increase was statistically significant only after 3-week treatment but not after 1- or 2-week treatment (F(3, 42) = 8.36, p < .001) (Fig. 2A). In contrast, treatment with higher doses of LiCl, which subsequently resulted in higher serum concentrations of 4.36–6.59 mEq/L, reduced the number of neurosphere-forming NSCs in the adult brain (F(1, 6) = 8.12, p = .036) (Fig. 2B).
Figure Figure 2.. Number of neural stem cells in the adult brain subventricular zone was determined by the neurosphere assay. (A): The total number of neural stem cells in the lateral portion of the SVZ was increased in mice treated with LiCl (0.86 ± 0.23 mEq/l, mean ± SD) for 3 weeks (n = 10) but not for 1 week (n = 9) or 2 weeks (n = 6) as compared to controls (n = 21). (B): The total numbers of neural stem cells in the mice, which were treated with higher doses of LiCl and subsequently showed toxic levels of lithium in their sera (4.36–6.59 mEq/l) (n = 3), were reduced relative to control (n = 4). (C, D): Mice were given VPA or CBZ for 3 weeks and were subjected to the neurosphere assay. The total numbers of neural stem cells in the lateral portions of the SVZ were increased in mice treated with VPA (n = 6) as compared to controls (n = 8) (C), and in mice treated with CBZ (n = 10) as compared to DMSO-injected controls (n = 8) (D). (E): Serial bulk passages of neurospheres were performed at a clonal density (two cells/μL) every 6 days. The primary neurospheres derived from mice given LiCl (n = 9, serum Li+ concentrations at 0.22–1.19 mEq/L) or VPA (n = 9) or CBZ (n = 3) produced more secondary and tertiary neurospheres than their control counterpart. Data represent means ± SEM. *, p < .05. Abbreviations: CBZ, carbamazepine; DMSO, dimethyl sulfoxide; VPA, valproic acid.
Download figure to PowerPoint
We asked if a similar stimulatory effect occurred following treatment with VPA or CBZ. Mice given VPA (0.5 g/L in drinking water) or CBZ (1 mg/kg body weight by intraperitoneal injection, twice daily) achieved much lower drug concentrations in their sera (∼6.0 μg/mL for VPA and 0.5 μg/mL or lower for CBZ) than observed during treatment of bipolar patients . Nevertheless, administration of VPA and CBZ produced similar increases in the numbers of neurosphere-forming NSCs. Compared to controls, the total number of NSCs in the SVZ was increased in both VPA-treated mice and in CBZ-treated mice (F(1, 12) = 11.10, p = .006 and F(1, 16) = 9.38, p = .007, respectively) (Fig. 2C, 2D). In contrast, the number of NSCs in the SVZ from mice treated with phenytoin (200 mg/L drinking water, 3 weeks), an anticonvulsant lacking mood stabilizing action, was comparable to control (95.1 ± 7.6% of control, n = 3).
To determine the effect of mood stabilizer administration on the self-renewal capacity of NSCs, primary neurospheres were serially passaged in bulk and cells were cultured at low cell density (two cells/μL) in the absence of the drugs. NSCs derived from the SVZ of mice which had been given mood stabilizers and showed serum drug concentrations at or below the therapeutic doses, exhibited enhanced self-renewal. That is, we observed increased numbers of both secondary (F(2, 21) = 10.36, p < .001 for LiCl and VPA and t(4) = 5.85, p = .004 for CBZ) and tertiary neurospheres (F(2, 20) = 9.74, p = .001 for LiCl and VPA and t(4) = 9.10, p < .001 for CBZ) as compared to appropriate controls (Fig. 2E). Thus, these results suggest that administration of mood stabilizers enhances the self-renewal of NSCs in the adult brain and expands the NSC pool in the SVZ.
At this point, we focused on lithium treatment for further in vivo analyses, because administration of LiCl by drinking water for 3 weeks or longer consistently resulted in serum Li+ concentrations in the range 0.4–1.2 mEq/L, whereas 3-week treatments with VPA or CBZ produced low and variable drug concentrations. In these trials, the number of NSCs in the SVZ of treated mice was assessed by a different method. After 3 weeks of treatment, mice received 5 BrdU injections every 3 hours and were sacrificed 4 weeks after the last injection. During the 4-week period following the last injection, transit-amplifying neural progenitor cells, having incorporated BrdU, diluted the label by further divisions, differentiated and migrated away from the SVZ, or died . BrdU+ cells that remained in the SVZ for 4 weeks are considered relatively quiescent NSCs. BrdU+ cells in the SVZ were increased in number in the mice given LiCl as compared to control (t(10) = −3.97, p = .003) (Fig. 3A).
Figure Figure 3.. Lithium increases BrdU+ cells in the adult brain. Mice were administered LiCl for 3 weeks, received 5 BrdU injections (65 mg/kg) every 3 hours, and then sacrificed 4 weeks (A), 1 hour (B), or 7 days (C) after the last injection. Cryosections were immunostained for BrdU incorporation, which overlapped the nuclear staining with Hoechst 33,342. (A): More BrdU+ cells which remained in the SVZ for 4 weeks after the labeling were detected in the mice given LiCl (n = 6) than in control (n = 6). (B): BrdU labeling demonstrated more dividing cells in the SVZ of lithium-treated mice (n = 10) than control (n = 7). (C): BrdU+ cells that migrated to the GCL of the olfactory bulb 7 days after the labeling were counted. Representative higher magnification images for BrdU positivity (red, right panels) show a similar region as indicated by a box in a lower magnification image for Hoechst 33,342 nuclear staining (blue, left panel). More BrdU+ cells were observed in the GCL of lithium-treated mice (n = 4) than in that of control mice (n = 6). Scale bars = 100 μm. Data represent means ± SEM. *, p < .05. Abbreviations: RMS, rostral migratory stream; BrdU, 5-bromo-2′-deoxyuridine; GCL, granular cell layer.
Download figure to PowerPoint
Next, we asked if the lithium-induced increase of NSCs in the SVZ also increased the number of neural progenitors and subsequent neurogenesis. After 3 weeks of lithium treatment, mice received 5 BrdU injections every 3 hours and were sacrificed 1 hour (total 13 hours, Fig. 3B) or 7 days (Fig. 3C) after the last injection. BrdU labeling over 13 hours that encompasses estimated cell cycle times of neural progenitor cells  demonstrated more dividing cells in the SVZ of lithium-treated mice than control (F(1, 15) = 5.13, p = .039 for SVZ) (Fig. 3B), suggesting that neural progenitor cells were also increased after chronic lithium treatment. As a result of this increase in proliferating cells in the SVZ, more BrdU+ cells survived and migrated to the olfactory bulb (t(8) = −2.82, p = .022) of LiCl-treated mice than that of control (Fig. 3C). These findings demonstrate that lithium administration expanded the size of the NSC pool and increased neural progenitor proliferation in the SVZ, resulting in increased cell supply to the olfactory bulb in the adult brain.
Notch Signaling Mediates the Effects of Mood Stabilizers.
Recent studies suggest that Notch signaling plays a critical role in the maintenance of NSCs [13, 47, 48]. Therefore, we investigated whether mood stabilizers activate Notch signaling in neurospheres. RT-PCR analyses revealed that two downstream genes of the Notch pathway, Hes1 and Hes5, were upregulated in all three-drug-treated neurosphere cultures relative to control (Fig. 4A), which was confirmed by quantitative RT-PCR (F(3, 20) = 4.94, p = .01) (supplemental Fig. 3A). We next performed a reporter assay using a luciferase gene plasmid, in which the firefly luciferase gene was placed downstream of either the Hes1 or Hes5 gene promoter . Treatment with 0.5 mM LiCl, or 60 μM VPA, or 5 μM CBZ enhanced both Hes1 (F(3, 44) = 68.34, p < .001) and Hes5 (F(3, 44) = 23.48, p < .001) gene promoter activities (Fig. 4B and online supplemental Fig. 3B). Western blotting revealed that all three mood stabilizers increased Hes1 protein in neurosphere cultures (F(3, 8) = 10.78, p = .004) (Fig. 4C and online supplemental Fig. 3C). Thus, mood stabilizing drugs at therapeutic concentrations upregulate the expression of Hes1/5 genes, suggesting activation of the Notch signaling pathway.
Figure Figure 4.. Mood stabilizers activate Notch signaling in neurospheres. (A, C, D): Primary neurospheres were grown in the presence of 0.5 mM LiCl, 60 μM VPA, or 5 μM CBZ. (A): RNA from the neurospheres was subjected to RT-PCR analysis. RT ± indicates the presence/absence of reverse transcriptase in the assay. (B): Firefly luciferase reporter assay was performed using secondary neurospheres. Treatment with 0.5 mM LiCl, 60 μM VPA, or 5 μM CBZ enhanced Hes1 gene promoter activities. (C, D): Protein extracts from the primary neurospheres were separated on SDS-PAGE, and then immunoblotted with anti-Hes1 (C) and anti-cleaved Notch1 (D) antibody. (E): Primary neurospheres were formed in the presence of therapeutic doses of mood stabilizers together with 1.0 μM γ-secretase inhibitor, L-685,458, which inhibits the activation of Notch signaling. (F): The medial ganglionic eminence cells from E14.5 presenilin1−/− (n = 8) and their wild-type littermates (n = 8) were cultured. Mood stabilizers enhanced the formation of primary neurospheres from wild-type cells but showed few effects on those from presenilin1−/− cells. (G, H): γ-Secretase assay was performed in the presence of 0.5 mM LiCl, 60 μM VPA, 5 μM CBZ, or 10 μM L-685,458. Myc-tagged cleaved Notch1 intracellular domain was immunoblotted with anti-c-myc antibody (G). (H): The bands were quantified and relative levels of myc-tagged cleaved Notch1 to control as 100% are shown. All experiments were carried out four or more times. Data represent means ± SEM. *, p < .05. Abbreviations: CBZ, carbamazepine; RT, reverse transcription; VPA, valproic acid.
Download figure to PowerPoint
To obtain direct evidence of Notch signaling activation, we quantified amounts of an active form of Notch1 receptor in neurospheres by Western blotting. Anti-cleaved Notch1 antibody recognizes the Notch intracellular domain, which is a direct measure of the amount of Notch signaling activation. Primary neurospheres exposed to LiCl, VPA, or CBZ contained significantly more cleaved Notch1 than the control neurospheres (F(3, 12) = 11.63, p < .001) (Fig. 4D and online supplemental Fig. 3D).
If mood stabilizers enhance NSC expansion through Notch signaling, then inhibition of Notch signaling should block this enhancement. Indeed, addition of 1 μM γ-secretase inhibitor, L-685,458, to cultures abolished the effects of LiCl, VPA, and CBZ on primary neurosphere formation (Fig. 4E). Presenilin1 is a major component of the γ-secretase complex or γ-secretase itself  and homozygous mutation for the presenilin1 gene results in the depletion of NSCs due to the impaired activation of Notch signaling . We examined the effects of mood stabilizers on neurosphere formation from MGE cells of E14.5 presenilin1−/− embryos because those embryos die around the time of birth . In contrast to wild-type controls, where mood stabilizers increased the number of primary neurospheres, those drugs exhibited few effects, if any, on primary neurosphere numbers from presenilin1−/− NSCs (Fig. 4F). Finally, we examined whether or not mood stabilizers modify γ-secretase activities using γ-secretase source from neurospheres and myc-tagged full-length Notch1 as a substrate. Indeed, γ-secretase activities were enhanced in the presence of therapeutic doses of LiCl and VPA by ∼60%, but not in the presence of CBZ (Fig. 4G, 4H). These results suggest that mood stabilizers enhance the self-renewal capability of NSCs by activating Notch signaling and that these effects are mediated by, at least in part, the augmentation of γ-secretase activities.
Both lithium and VPA have been shown either directly or indirectly to inhibit GSK-3β [20, , –23], a key mediator of Wnt and other signal pathways, and inhibit inositol monophosphatase (IMPase), leading to inositol depletion [16, , –19]. Therefore, we determined whether or not effects of mood stabilizers on neurospheres were mediated by the inhibition of GSK-3β signaling or inositol depletion. We examined levels of phosphorylated GSK-3β at Ser9, an inactive form of GSK-3β, in the neurospheres by Western blotting following drug treatment. Treatment with therapeutic doses of mood stabilizers did not alter the ratio of phosphorylated to total GSK-3β (Fig. 5A, 5B), whereas high doses of mood stabilizers increased the phosphorylation of GSK-3β (F(3, 16) = 4.86, p = .014) (Fig. 5C, 5D).
Figure Figure 5.. Effects of mood stabilizers on GSK-3β signaling and the inositol pathway. (A): Primary neurospheres from the adult SVZ were generated in the presence of 0.5 mM LiCl, 60 μM VPA, or 5 μM CBZ. Protein extracts of the neurospheres were separated on SDS-PAGE and then immunoblotted with anti-GSK-3β or anti-phospho-GSK-3β (Ser9) antibodies. (B): The bands were quantified and the ratio of phosphorylated to total GSK-3β was calculated. (C): Primary neurospheres were grown and toxic doses of mood stabilizers were added to the culture at 5 days in vitro. After 15-minute treatment, protein extracts of the neurospheres were subject to Western blotting. (D): The bands were quantified and the ratio of phosphorylated to total GSK-3β was calculated. (E, F): Total RNA was extracted from primary adult neurospheres cultured with mood stabilizers at therapeutic (E) or higher concentrations (F) and subject to RT-PCR analysis. Mood stabilizers at the therapeutic concentrations showed few effects on the expression levels of IMPA1/2 in the neurospheres (E), whereas the drugs at the higher concentrations downregulated gene expression (F). RT ± indicates the presence/absence of reverse transcriptase in the assay. (G, H): Effects of inositol rescue were determined by adding 1 mM myo-inositol in cultures of the neurosphere assay. Myo-inositol did not alter the effects of 0.5 mM LiCl, 60 μM VPA, or 5 μM CBZ, which increased the number of primary neurospheres (G), but resumed the toxic effects of 5 mM LiCl, 300 μM VPA or 50 μM CBZ (H). Data represent means ± SEM. from 3–5 independent experiments. *, p < .05. Abbreviations: CBZ, carbamazepine; VPA, valproic acid.
Download figure to PowerPoint
Similarly, therapeutic concentrations of these drugs did not change the expression of IMPase 1/2 genes (Fig. 5E). By contrast, coculture with higher concentrations of mood stabilizers lowered the expression levels of those genes (Fig. 5F). Addition of myo-inositol to cultured cells rescue the depletion of inositol . Adult SVZ cells were cultured to generate primary neurospheres in the presence of 1.0 mM myo-inositol together with lithium, VPA and CBZ. Myo-inositol itself showed virtually no ability to increase the number of neurospheres and did not reverse the effects of mood stabilizers at therapeutic concentrations (Fig. 5G). By contrast, coculture with 1.0 mM myo-inositol reversed the toxic effects of mood stabilizers, observed in the presence of 5.0 mM LiCl, 300 μM of VPA, and 50 μM of CBZ (Fig. 5H). Together, these results suggest that Notch signaling activation in the neurospheres by therapeutic doses of mood stabilizers is not mediated by the alteration of GSK-3β or inositol signaling.
Notch Signaling Activation In Vivo
Activation of Notch signaling in NSCs by mood stabilizers was evident not only in vitro but also in vivo. In situ hybridization assessment of Hes1 and Hes5 gene expression showed that Hes1/5-expressing cells were increased in the SVZ of mice treated with LiCl for 3 weeks (Fig. 6A, 6B). The lateral portion of the SVZ was excised and then homogenized in lysis buffer. Comparable amounts of protein were obtained from each mouse (control, 1.79 ± 0.42 mg (n = 12) vs. LiCl, 1.77 ± 0.30 mg (n = 12), mean ± SD), verifying equivalent tissue isolation. Samples of 20 μg protein were separated on SDS-PAGE and immunoblotted with anti-cleaved and anti-total Notch1 antibody (Fig. 6C, 6D). Higher amounts of cleaved Notch1 were detected in the SVZ of the mice treated with lithium for 3 weeks relative to control mice (F(1, 18) = 75.38, p < .001) (Fig. 6E) and an insignificant increase of total Notch1 protein was also observed (Fig. 6F). Increase of cleaved Notch1 protein level was only detected after 3-week, but not 1-week, treatment by lithium (online supplemental Fig. 4A, 4B). These results indicate that lithium treatment results in either increased amounts of activated Notch in individual neural stem/progenitor cells or increased numbers of neural stem/progenitor cells with equivalent levels of activated Notch. In contrast, GSK-3β signaling was not altered by lithium treatment in vivo (online supplemental Fig. 4C, 4D).
Figure Figure 6.. Lithium administration activates Notch signaling in the SVZ of the adult brain. (A, B): Mice were given LiCl for 3 weeks and received 5 BrdU injections. Cryosections were analyzed by in situ hybridization for Hes1(A) and Hes5(B) gene expression (purple) and double labeled for BrdU incorporation (brown). Lithium treatment for 3 weeks increased the number of Hes1/5-expressing cells in and close to the SVZ, where BrdU+ neural stem/progenitor cells resided. Scale bar = 50 μm (lower magnification) or 10 μm (higher magnification). (C–F): The SVZ tissue was subjected to Western blotting. Treatment with lithium for 3 weeks resulted in a significant increase of cleaved Notch1 (n = 8) (C, D) and an insignificant increase of total Notch1 protein (n = 4) (E, F). Data represent means ± SEM. *, p < .05. Abbreviations: CC, corpus callosum; STR, striatum.
Download figure to PowerPoint