Abstract
Many neuroactive steroids (NS) demonstrate neurotrophic and neuroprotective actions, including protection against apoptosis via Bcl-2 protein. NS are altered in post-mortem brain tissue from subjects with bipolar disorder, and several agents with efficacy in mania elevate NS in rodents. We therefore hypothesized that lithium and valproate may elevate NS, and compensatory NS increases may occur in Bcl-2 knockout mice. NS levels (allopregnanolone, pregnenolone) were determined in frontal cortex by negative ion chemical ionization gas chromatography/mass spectrometry in male Wistar Kyoto rats treated chronically with lithium, valproate, or vehicle. NS were also investigated in heterozygous Bcl-2 knockout mice. Allopregnanolone levels are significantly elevated in lithium-treated (p<0.05), but not in valproate-treated, rats. Pregnenolone levels also tend to be higher following lithium treatment (p=0.09). Knockout of Bcl-2 significantly increases pregnenolone levels in mice (p<0.01), while allopregnanolone levels are unaltered. NS induction may be relevant to mechanisms contributing to lithium therapeutic efficacy and neuroprotection.
Introduction
The neuroactive steroid (NS) allopregnanolone (3α-hydroxy-5α-pregnan-20-one) is synthesized de novo in the brain from cholesterol or from peripheral steroid precursors (Belelli and Lambert, 2005). A number of allopregnanolone actions are attributed to the fact that it potentiates GABAA receptor responses at nanomolar concentrations, doing so more potently than either benzodiazepines or barbiturates (Morrow et al., 1987). Allopregnanolone demonstrates anxiolytic (Wieland et al., 1991) and anticonvulsant effects (Kokate et al., 1996), and may be relevant to antipsychotic drug action (Barbaccia et al., 2001; Marx et al., 2003). More recently, neuroprotective roles for allopregnanolone have been demonstrated in a mouse model of Niemann–Pick type C disease (Griffin et al., 2004) and a rat model of traumatic brain injury (Djebaili et al., 2005). Allopregnanolone also protects against apoptosis via Bcl-2 protein in rat adrenal chromaffin and pheochromocytoma cells (Charalampopoulos et al., 2004) and protects against N-methyl-d-aspartate (NMDA)-induced apoptosis in mouse P19-derived neurons (Xilouri and Papazafiri, 2006).
NS may be relevant to bipolar disorder, as evidenced by the fact that NS concentrations are altered in post-mortem brain tissue from subjects with this illness (Marx et al., 2006b). Pregnenolone and dehydroepiandrosterone are significantly increased in both posterior cingulate and parietal cortex in patients with bipolar disorder, while allopregnanolone levels tend to be decreased in parietal cortex (Marx et al., 2006b). Given the presence of brain NS changes in patients with bipolar disorder compared to control subjects, it will be important to determine if medications utilized to treat this illness potentially act via mechanisms that include NS induction. Consistent with this possibility, our previous investigation revealed that intraperitoneal clozapine or olanzapine dose-dependently increases rat cerebral cortical allopregnanolone levels (Marx et al., 2003). Clozapine also produces marked elevations in the NS pregnenolone (Marx et al., 2006a). Both clozapine and olanzapine possess anti-manic properties (Suppes et al., 1999; Tohen et al., 2000), as well as antidepressant effects (Meltzer et al., 2003; Tollefson et al., 1998). The fact that these agents demonstrate efficacy in mania and also produce marked alterations in NS levels, combined with evidence for altered post-mortem brain NS concentrations in patients with bipolar disorder, logically raises the question as to whether pharmacological agents for bipolar disorder potentially achieve a degree of therapeutic efficacy through NS modulation. Conversely, a recent study has determined that lithium may prevent mirtazapine-induced elevations in allopregnanolone (Schule et al., 2007). More data addressing potential NS alterations following lithium are therefore required.
To our knowledge, the sole prior clinical examination of NS in patients with bipolar disorder reported serum increases in both allopregnanolone and progesterone in a group of female patients with bipolar I or bipolar II disorder in the luteal phase of the menstrual cycle compared to control subjects without psychiatric illness (Hardoy et al., 2006). Of 17 women with bipolar I or bipolar II disorder, two were drug-free (11.8%), 12 were receiving antipsychotics (70.6%), nine were receiving mood stabilizers (52.9%), and an unreported number of patients were receiving antidepressants. Specific drug regimens were not reported for individual subjects, and thus potential NS alterations following treatment with specific mood stabilizers are not discernible from the existing clinical literature. Furthermore, there are no reported animal studies investigating NS and lithium or valproate. The current study therefore determined potential alterations in allopregnanolone and pregnenolone in rat frontal cortex following lithium or valproate administered chronically at therapeutically relevant doses. Since allopregnanolone is protective against apoptosis, we also examined whether heterozygous knockout of Bcl-2 protein in mice produces compensatory increases in NS.
Method
Lithium and valproate administration
Male Wistar Kyoto rats purchased from Harlan (Indianapolis, IN, USA) were treated chronically with lithium or valproate for 4 wk at doses achieving therapeutic and physiologically relevant levels (0.74±0.31 mequiv/l for lithium, measured by atomic absorption; 41.6±3.2 µg/ml for valproate, measured by immunoassay), and compared to saline vehicle administration for 4 wk (n=9 per condition). Animals were fed with control rodent chow or chow containing a low dose of lithium (1.2 g Li2CO3/kg) or valproate (10 g/kg) for 1 wk to allow the animals to slowly acclimate to the drug, and were then fed with chow containing a full dose of lithium (2.4 g Li2CO3/kg) or valproate (20 g/kg) for 3 wk. On the last day of the treatment, animals were decapitated, and trunk blood was collected to measure lithium and valproate levels in the serum. Brain tissues were dissected on ice and frozen in dry ice immediately. Frontal cortex was defined as the cortical projection area of the mediodorsal thalamic nucleus. Animal use procedures were in accordance with the NIH Guide for the Care and Use of Laboratory Animals.
Bcl-2 knockout mice
Founder mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). Adult Bcl-2 knockout mice were crossbred with wild-type outbred mice to generate heterozygote mice. Bcl-2 heterozygote siblings were then interbred to generate homozygotes (−/−), heterozygotes (−/+), and wild-type (+/+) littermates. The genotypes of the offspring were determined by PCR analysis. Mice tail biopsy genomic DNA was isolated using DNeasy Tissue kit (Qiagen, Chatsworth, CA, USA). The Neo primers (5′-CTTGGGTGGAGAGGCTATTC-3′ and 5′-AGGTGAGATGACAGGAGATC-3′) yielded a 280-bp fragment, and the Bcl-2 primers (5′-CTTTGTGGAACTGTACGGCCCCAGCATGCG-3′ and 5′-ACAGCCTGCAGCTTTGTTTCATGGTACATC-3′) yielded a 215-bp fragment. Wild-type animals would have only the Bcl-2 fragment, heterozygote animals would have both fragments, and Bcl-2 −/− animals would have only the Neo fragment. The heterozygous offspring appeared entirely normal and were fertile. Bcl-2 −/+ mice and their wild-type littermates were used in this study because most of the null (−/−) offspring died prior to weaning. Mice (8-wk-old, weight ∼20–30 g) were sacrificed by neck dislocation without anaesthetization, and brain tissues were dissected on ice and frozen in dry ice immediately. Western blots performed and reported previously (Einat et al., 2005) showed lower Bcl-2 protein levels in heterozygotes than in wild-type littermates. Animal use procedures were in accordance with the NIH Guide for the Care and Use of Laboratory Animals (strain name: B6;129S2-Bcl2tm1Sjk/J; stock number: 002 265; wild type: B6;129SF2/J; stock number: 101 045).
NS analyses: gas chromatography/mass spectrometry (GC/MS), preceded by high-performance liquid chromatography (HPLC)
NS analyses in rat and mouse frontal cortex were performed as described previously (Marx et al., 2006a–c). Rodent brain tissue was homogenized in 5 vol distilled water containing ∼2200 cpm (4400 dpm/injection) of tritiated NS (PerkinElmer, Waltham, MA, USA) to detect the HPLC fraction containing the NS of interest. Deuterated allopregnanolone (D4-allopregnanolone, 400 pg) and deuterated pregnenolone (D4-pregnenolone, 400 pg) were used as internal standards. Supernatants were extracted three times with 3 vol of ethyl acetate and dried under nitrogen prior to HPLC purification. Samples were derivatized utilizing heptafluorobutyric acid anhydride, injected onto an Agilent 5973 Mass Spectrometer (MS) coupled to an Agilent 6890N Gas Chromatograph (GC), and analysed in the negative ion chemical ionization mode (NICI) utilizing methane as the reaction gas and helium as the carrier gas. Mass spectrometer single-ion monitoring (SIM) mode was utilized to focus on the most abundant ion fragment for each steroid derivative. Only peaks with a signal-to-noise ratio ⩾5:1 were integrated. The limit of NS detection with this method was 2 pg for allopregnanolone and 10 pg for pregnenolone. Mean intra-assay coefficients of variation were 3.9% for allopregnanolone and 3.0% for pregnenolone.
Statistical analysis
For rat experiments utilizing chronic lithium and valproate administration, data were analysed by ANOVA with post-hoc Dunnett tests. Data from Bcl-2 knockout mouse experiments were analysed by unpaired t tests.
Results
Allopregnanolone levels are significantly increased in the rat frontal cortex following chronic lithium administration compared to vehicle [Figure 1a; mean levels (±s.e.m.) 2.42±0.75 ng/g vs. 0.71±0.11 ng/g, respectively (ANOVA: p=0.017, F=4.85, d.f. 2, 24; post-hoc Dunnett test: p<0.05; n=9 per condition)]. Pregnenolone levels also tend to be higher in rat frontal cortex following chronic lithium administration compared to vehicle [Figure 1b; mean levels (±s.e.m.) 8.19±2.95 ng/g vs. 3.37±0.50 ng/g, respectively (ANOVA: p=0.069, F=3.00, d.f. 2, 24; post-hoc Dunnett test: p=0.09; n=9 per condition)]. In contrast, chronic valproate administration alters neither allopregnanolone nor pregnenolone levels (Figure 1a, b). Since pregnenolone can be metabolized to allopregnanolone, we examined the correlation between the levels of these two NS in rat frontal cortex and determined that allopregnanolone levels are positively correlated with pregnenolone levels (Figure 1c; Pearson correlation coefficient r=0.92, p<0.0001, n=27 XY pairs), with high degrees of correlation throughout this set of subjects, at low, medium, and high levels of these NS.
(a) Allopregnanolone levels in rat frontal cortex are significantly increased following chronic lithium administration compared to vehicle administration. (b) Pregnenolone levels in rat frontal cortex tend to be increased following chronic lithium administration compared to vehicle administration. (c) Allopregnanolone levels are positively correlated with pregnenolone levels in rat frontal cortex. * p<0.05, # p<0.1.
Pregnenolone levels in frontal cortex are significantly higher in heterozygous Bcl-2 knockout mice compared to wild-type control mice. * p<0.01.
Based on previous reports indicating that allopregnanolone protects against apoptosis (Charalampopoulos et al., 2004; Xilouri and Papazafiri, 2006), we hypothesized that compensatory up-regulation of allopregnanolone levels may occur in Bcl-2 knockout mice compared to wild-type mice. However, we found no differences in frontal cortex allopregnanolone levels between these two groups (3.96±0.33 ng/g in Bcl-2 knockout mice vs. 4.02±0.31 ng/g in wild-type control mice, unpaired t test p>0.05, n=10 per condition; data not shown). Pregnenolone levels, in contrast, are significantly higher in heterozygous Bcl-2 knockout mice compared to wild-type mice [Figure 2; mean levels (±s.e.m.) 4.03±0.78 ng/g vs. 1.65±0.23 ng/g, respectively, unpaired t test p=0.009, n=10 per condition). In these mice, pregnenolone and allopregnanolone levels are not significantly correlated (data not shown).
Discussion
In this pilot investigation, we find that chronic lithium administration produces significant increases in allopregnanolone levels (as well as a trend towards increases in pregnenolone) in rat frontal cortex, while administration of valproate does not alter NS. Additionally, we demonstrate a strong correlation between the concentrations of pregnenolone and its metabolite allopregnanolone in frontal cortex. Allopregnanolone concentrations are not altered in heterozygous Bcl-2 knockout mice, despite its reported role in protection against apoptosis; pregnenolone levels, however, are significantly increased in these animals. In this strain of mice, allopregnanolone and pregnenolone levels were not correlated. Possible therapeutic ramifications of these results are outlined below.
While chronic lithium administration, a primary treatment strategy for bipolar disorder, can result in neurotoxicity in patients with identifiable clinical risk factors (nephrogenic diabetes insipidus, old age, abnormal thyroid function, impaired renal function) (Oakley et al., 2001), 2-wk administrations of lithium at clinically relevant doses have been shown to enhance neurogenesis in rat hippocampus, increasing both Bcl-2 levels and the percentage of new cells that display a neuronal phenotype (Chen et al., 1999, 2000). Since the NS allopregnanolone dose-dependently increases proliferation of rat hippocampal neuroprogenitor cells and human cerebral cortical neural stem cells at physiologically relevant concentrations, and also increases expression of genes that promote progression through the cell cycle (Wang et al., 2005), we hypothesized that lithium treatment may produce elevations in brain allopregnanolone. Chronic lithium administration more than tripled allopregnanolone levels in rat frontal cortex (Figure 1a), raising the possibility that lithium-induced elevations in allopregnanolone may contribute to increased neurogenesis following lithium administration and potentially impact neuroplasticity. Based upon previous evidence that NS also modulate neurogenesis in other brain regions such as the hippocampus (Mayo et al., 2005), further examinations of multiple brain regions could reveal potential regional specificities of lithium effects.
Similar to lithium-induced allopregnanolone increases in the present study, previous reports indicate that second-generation antipsychotics that demonstrate efficacy in mania such as clozapine and olanzapine also elevate allopregnanolone levels in rodent brain (Barbaccia et al., 2001; Marx et al., 2003). Taken together, these data suggest that allopregnanolone elevations may contribute to the therapeutic efficacy exhibited by lithium. It should be noted, however, that mood stabilizers such as lithium may actually be more versatile in their actions on NS levels, as a previous augmentation study demonstrates that these compounds reverse antidepressant-induced NS increases (Schule et al., 2007).
Previous reports also indicate that allopregnanolone has pronounced neuroprotective effects in a mouse model of Niemann–Pick type C disease (Griffin et al., 2004) and a rat model of traumatic brain injury (Djebaili et al., 2005), suggesting a potential role for allopregnanolone induction in the neuroprotective effects of lithium as well (Gray et al., 2003). While allopregnanolone is not altered in frontal cortex of heterozygous Bcl-2 knockout mice compared to allopregnanolone levels in wild-type mice, pregnenolone levels are significantly higher in these animals (Figure 2). It is possible that pregnenolone elevations in heterozygous Bcl-2 knockout mice may reflect a compensatory mechanism that results in the normalization of downstream allopregnanolone metabolite levels in this Bcl-2 strain. Interestingly, although neuroprotective effects have also been attributed to valproate (Chuang, 2005), we find increases in neither allopregnanolone nor pregnenolone levels in response to chronic valproate administration. The discrepancies in the NS responses to these two mood stabilizers may represent another potential mechanistic difference in the neuroprotective properties of these two compounds (Hennion et al., 2002; Jin et al., 2005; Mora et al., 1999, 2002).
Finally, like the antipsychotic clozapine (Meltzer et al., 2003), lithium decreases suicidality (Baldessarini et al., 2006). We have determined previously that parietal cortex pregnenolone levels in patients with schizophrenia who died by suicide are significantly reduced in comparison to patients with schizophrenia who died of other causes (Bradford et al., 2006). In light of this association between reduced brain pregnenolone and increased suicidality, it is a logical possibility that increasing NS levels may contribute to the modulation of suicidal behaviours by lithium. Previous reports support this hypothesis, since clozapine elevates both pregnenolone (Marx et al., 2006a) and allopregnanolone (Barbaccia et al., 2001; Marx et al., 2003) in rat brain, and prior evidence suggests links between reduced allopregnanolone and depression (Uzunova et al., 2006).
In summary, these findings suggest that elevations in allopregnanolone following chronic lithium administration may be relevant to its therapeutic efficacy, as well its impact on neuroplasticity and neuroprotection. NS may also be implicated in Bcl-2 mechanisms relevant to lithium actions.
Acknowledgements
This work was supported by the following sources: K23 MH 65080 (C.E.M.), VA Advanced Research Career Development Award (C.E.M.), Durham VA REAP (R.D.M.), Mid-Atlantic MIRECC, and the National Institute of Mental Health, NIH, Bethesda, MD. We thank Gillian Parke for her excellent assistance with the manuscript.
Statement of Interest
Dr Marx is a co-applicant on a pending U.S. patent application for the use of neuroactive steroids to treat central nervous system disorders.
References
