Neuroimaging and lesion studies have provided important information as to the neuroanatomical basis of bipolar illness (reviewed in Bolwig 1993;
Nasrallah et al. 1989). Examination of whole brain metabolic rates indicates significantly lower rates in drug-free patients with bipolar depression relative to rates in patients with bipolar mania or unipolar depression or subjects who are psychiatrically healthy (L. R. Baxter et al. 1985; Phelps et al. 1984 [but see Buchsbaum et al. 1986]), which normalized when subjects shifted to a euthymic or manic state (L. R. Baxter et al. 1985). Other studies indicate that patients with bipolar mania exhibit higher global glucose metabolism (H. Kishimoto et al. 1987) and significantly higher global cerebral blood flow (Rush et al. 1982 [but see Delvenne et al. 1990]) relative to that of psychiat-rically healthy subjects. Anteroposterior differences have not been observed (L. R. Baxter et al. 1985; Devous et al. 1984), although a relative hypo-frontality was found in subjects with bipolar disorder when a series of painful stimuli were administered during tracer uptake (Buchsbaum et al. 1986).
Moreover, no consistent hemispheric differences in metabolism have been observed in subjects with bipolar disorder (L. R. Baxter et al. 1985;
Buchsbaum et al. 1986; Phelps et al. 1984), but lesion studies indicate that damage to right hemisphere structures is more commonly associated with the development of secondary mania (M. R. Cohen and Niska 1980; Cummings and Mendez 1984; Forrest 1982; R. G. Robinson et al. 1988; Starkstein et al.
1990).
Several studies have demonstrated an enlargement of third and lateral ventricle volume in patients with bipolar disorder (Dewan et al. 1988;
Pearlson et al. 1984; Rieder et al. 1983; Schlegel and Kretzschmar 1987;
Strakowski et al. 1993b; Swayze et al. 1992), although the lateral ventricle enlargement is not consistently observed (Johnstone et al. 1986; Swayze et al. 1990). Ventricular enlargement would be expected to affect the func-tional integrity of structures that line the ventricles. Indeed, caudate and
thalamic nuclei, two paraventricular structures, exhibit an increased density bilaterally in bipolar patients (Dewan et al. 1988), and lesions of the right head of the caudate and right thalamus are associated with the development of secondary mania (Starkstein et al. 1991). However, other studies have found no difference in metabolic and blood flow rates in caudate or thalamus in patients with bipolar disorder (L. R. Baxter et al. 1985), and caudate and thalamus volume did not differ in patients with first-episode mania (Strakowski et al. 1993b) or in patients with bipolar disorder (Swayze et al.
1992).
Several converging lines of evidence indicate that abnormalities in the temporal lobe contribute to bipolar illness. First, temporal lobe epilepsy in the right hemisphere has been associated with the onset of bipolar symptomatology (Barczak et al. 1988; Drake 1988; Flor-Henry 1969; Gillig et al. 1988). Second, significantly lower regional cerebral blood flow is ob-served in the basal portion of the right temporal cortex of subjects with ma-nia (Migliorelli et al. 1993), whereas higher blood flow is observed in the left temporal lobes of subjects with bipolar disorder who are depressed (Devous et al. 1984) relative to control subjects. Third, temporal lobe volume is smaller bilaterally (Altshuler et al. 1991) or larger in the right hemisphere of subjects with bipolar disorder (Dewan et al. 1988), although another study found right temporal lobe volume to be larger in both psychiatrically healthy subjects and subjects with bipolar disorder (Swayze et al. 1992). Finally, le-sions of the basotemporal cortex are significantly correlated with the devel-opment of secondary mania (Jorge et al. 1993; Starkstein et al. 1991).
Importantly, the volume of the right hippocampus, a structure located within the temporal lobe, has been reported to be significantly reduced in pa-tients with bipolar disorder (Swayze et al. 1992).
Lithium-induced depletion of inositol via the uncompetitive inhibition of myo-inositol-1-phosphatase has been proposed to account for the physiologi-cal consequences of lithium and predicts that lithium will have the greatest impact on cells undergoing the greatest receptor-mediated PIP2 hydrolysis (e.g., Nahorski et al. 1991). Studies assessing regional changes in PI turnover indicate no consistent change in PI levels following long-term in vivo lithium treatment in either cortex or caudate/putamen (Honchar et al. 1989) or a comparable reduction of PI turnover in cortical, hippocampal, and striatal membranes (L. Song and Jope 1992). However, other studies have demon-strated that lithium-induced depletion of inositol is similar in magnitude among hypothalamus, hippocampus, and caudate, but unchanged in cerebel-lum (see W. R. Sherman et al. 1986); and K+- or agonist-stimulated [3H]IP accumulation is highest in cortex, thalamus, hippocampus, and striatum,
moderate in hypothalamus, pons, and medulla, but low to absent in cerebel-lum in the presence of lithium in vitro (5–10 mM) (R. D. Johnson and Minneman 1985; Rooney and Nahorski 1986). Hence, lithium appears to have its greatest impact on PI signaling in diencephalic and telencephalic re-gions, perhaps relating to its accumulation in these regions (see below). It is of note that PKC, which is activated by elevations of DAG in response to lithium-induced inositol depletion (reviewed in Manji and Lenox 1994), is selectively reduced in hippocampus, but not in other cortical and subcortical structures, following long-term lithium treatment (Manji et al. 1993).
Evidence from studies examining the effects of lithium on classic neuro-transmitter systems in different brain regions would also indicate that lith-ium’s actions have regional specificity. Following long-term lithium treatment, 3H-labeled serotonin binding is reduced in hippocampus and striatum but not in cortex or hypothalamus (Maggi and Enna 1980; Odagaki et al. 1990). Serotonin synthesis is reduced in hypothalamus and brain stem but not in cortex or cerebellum (Ho et al. 1970), and K+-stimulated seroto-nin release is enhanced in hypothalamus, hippocampus, and cortex, but spon-taneous release is reduced in hypothalamus and cortex but increased in hippocampus following long-term, but not short-term, lithium treatment (E. Friedman and Wang 1988). In a related study, basal and K+-induced sero-tonin release was increased in hippocampus but not in cortex following long-term lithium treatment (Treiser et al. 1981). Subchronic lithium ad-ministration reduces serotonin synaptosomal reuptake in striatum, hypothal-amus, hippocampus, and midbrain but not in pons or cortex (Ahluwalia and Singhal 1985). A competitive binding assay revealed that long-term lithium treatment reduced 5-HT1, 5-HT1C, and 5-HT2in frontal cortex, hippocam-pus, and choroid plexus, whereas 5-HT1A receptors were reduced in hippo-campus and choroid plexus, but not in frontal cortex (Mizuta and Segawa 1989). Short-term (5-day) lithium treatment was found to reduce tryptophan levels in limbic forebrain and striatum but reduced 5-HT synthe-sis only in striatum (Berggren 1987). DA content in brain following long-term lithium treatment is significantly reduced in pons and midbrain but not in hypothalamus, striatum, hippocampus, or cortex (Ahluwalia and Singhal 1980), whereas other studies indicate little variation in DA content across brain regions (Ho et al. 1970). NE binding in brain following long-term lithium treatment is significantly increased in the caudate nucleus but un-changed in brain stem, hypothalamus, or parietal cortex (O. G. Cameron and Smith 1980), and another study found no change inβ-adrenergic binding in cortex, hippocampus, or striatum (Maggi and Enna 1980). NE concentra-tions are reduced in pons, hypothalamus, and midbrain but not in striatum or
hippocampus (Ahluwalia and Singhal 1980), and NE reuptake is increased in pons, striatum, hippocampus, midbrain, and cortex but not in hypothalamus following long-term lithium treatment (Ahluwalia and Singhal 1985). ACh (muscarinic) binding in brain following long-term lithium treatment revealed no significant changes in cortex, hippocampus, or striatum (Maggi and Enna 1980), although the synthesis of ACh was observed to be increased in striatum, hippocampus, and cortex (Jope 1979). Finally, GABA receptor binding is reduced significantly in corpus striatum and hypothalamus but not in cortex, cerebellum, or hippocampus following long-term lithium treat-ment (Maggi and Enna 1980).
The premise that lithium exerts its therapeutic actions by acting at spe-cific neuroanatomical sites is supported by several lines of evidence. Lithium does not distribute evenly throughout the brain following either short- or long-term administration, but accumulates in specific regions where it may exert its greatest impact on cell signaling processes. Although there is some discrepancy between individual studies in terms of the regions showing the highest levels of lithium, a general trend has emerged. Studies using atomic absorption spectrophotometry to determine regional lithium concentrations indicate high levels in the striatum, intermediate levels in the hippocampus and hypothalamus, and low levels in the spinal cord and medulla in the rat following short-term in vivo administration (Ebadi et al. 1974). Following long-term in vivo administration (2–5 weeks, 30–45 mM lithium diet), high lithium levels are detected in diencephalic structures, particularly the hypo-thalamus, and in telencephalic structures including the striatum and hippo-campus, whereas low levels are detected in metencephalic structures such as the cerebellum (Bond et al. 1975; Edelfors 1975; Lam and Christensen 1992;
Savolainen et al. 1990; D. F. Smith and Amdisen 1981; Spirtes 1976).
Studies using radiographic dielectric track registration or nuclear magnetic resonance to localize lithium distribution also indicate forebrain accumula-tion (Heurteaux et al. 1986; S. C. Nelson et al. 1980; Ramaprasad et al.
1992; Thellier et al. 1980).
Conclusion
Lithium remains our most effective treatment for reducing the frequency and severity of recurrent affective episodes, but, despite extensive research, the underlying biological basis for the therapeutic efficacy of this drug remains unknown. Lithium is a monovalent cation with complex physiological and pharmacological effects within the brain. By virtue of the ionic properties it
shares with other important monovalent and divalent cations such as sodium, magnesium, and calcium, its transport into cells provides ready access to a host of intracellular enzymatic events affecting short- and long-term cell pro-cesses. It may be that, in part, the therapeutic efficacy of lithium in the treat-ment of both poles of bipolar disorder may rely on the “dirty” characteristics of its multiple sites of pharmacological interaction.
Strategic models to further delineate the mechanism(s) of action of lith-ium relevant to its therapeutic effects must account for a number of critical variables in experimental design. Lithium has a relatively low therapeutic in-dex, requiring careful attention to the tissue concentrations at which effects of the drug are being observed in light of the known toxicity of lithium within the central nervous system. Although such a poor therapeutic index may sug-gest a continuum between some of the biological processes underlying thera-peutic efficacy and toxicity, it may also account in part for the variability of effects of lithium observed in animal and in vitro studies. The therapeutic ac-tion of lithium is delayed, requiring long-term administraac-tion to establish ef-ficacy for both its treatment of acute mania and its prophylaxis of the recurrent affective episodes associated with bipolar disorder. Although its therapeutic effects are not reversed immediately upon abrupt discontinua-tion, there is accumulating evidence that abrupt withdrawal of lithium may sensitize the system to an episode of mania (E. Klein et al. 1992; Suppes et al.
1991).
In recent years, there has been dramatic progress in the identification of signal transduction pathways as targets for lithium’s actions. Regulation of signal transduction within critical regions of the brain by lithium affects the
“throughput” of multiple neurotransmitter systems; the ability of lithium to stabilize an underlying dysregulation of limbic and limbic-associated function is critical to our understanding of its mechanism of action. A general trend is beginning to emerge implicating right hemisphere structures, particularly the temporal lobe, caudate, thalamic nuclei, and, by implication, those regions that give and receive projections from these regions, including the hippocam-pus and hypothalamus, in the production of bipolar symptomatology. The biological processes in the brain responsible for the episodic clinical manifes-tation of mania and depression may be due to an inability to mount the appropriate compensatory responses necessary to maintain homeostatic reg-ulation, thereby resulting in sudden oscillations beyond immediate adaptive control (Depue et al. 1987; F. K. Goodwin and Jamison 1990b; Mandell et al. 1984). The resultant clinical picture is reflected in disruption of behavior, circadian rhythms, neurophysiology of sleep, and neuroendocrine and bio-chemical regulation within the brain. Lithium’s efficacy in treating these
symptoms may be the result of its ability to target and stabilize disruptive ac-tivity in critical regions of the brain. The behavioral and physiological mani-festations of the illness are complex and are mediated by a network of interconnected neurotransmitter pathways. The biogenic amines have been strongly implicated in the regulation of these physiological processes by vir-tue of their pharmacological actions and predominant neuroanatomical dis-tribution within limbic-related regions of the brain. Thus, lithium’s ability to modulate release of serotonin at presynaptic sites and affect DA-induced supersensitivity in the brain remains a relevant line of investigation into the respective action of lithium in altering the clinical manifestation of depres-sion and mania in the patient with bipolar disorder.
Some of the most exciting recent advances in our understanding of the long-term therapeutic action of lithium have been the identification of the effects on PKC-mediated events, particularly the posttranslational modifica-tion of important proteins responsible for regulamodifica-tion of signal transducmodifica-tion in the brain. Biochemical changes requiring prolonged administration of a drug such as lithium also suggest alterations at the genomic level, which may be mediated in large part by the activation and inactivation of subsets of genes with temporal specificity. In this context, the complex effect of lithium on PKC isozymes represents an attractive and heuristic mechanism by which the expression of various proteins involved in long-term neuronal plasticity and cellular response is modulated, thereby compensating for as yet genetically undefined physiological abnormalities in critical regions of the brain.
Many questions remain to be answered. For example, do the effects of chronic lithium on signal transduction pathways and gene expression stem exclusively from its demonstrated efficacy as an inhibitor of inositol monophosphatase and resultant changes in the DAG pathway, or does lith-ium under therapeutic conditions have other, more direct effects? Studies examining the effects of other inositol monophosphatase inhibitors (with no structural similarity to lithium) may provide important clues and offer in-sights into new drug development. In an interesting series of studies in the Xenopus, in which lithium exposure significantly altered the dorsal-ventral axis of the developing embryo, inhibition of inositol monophosphatase by an-other inhibitor did not result in similar alterations in morphogenesis, suggest-ing that lithium may be actsuggest-ing in an alternative pathway (Kao and Elinson 1989, 1998; P. S. Klein and Melton 1996). These studies have found that lithium inhibits glycogen synthase kinase-3β (GSK-3β) activity (Ki= 2.1 mM), which antagonizes the wnt signaling pathway associated with normal dorsal-ventral axis development in the Xenopus embryo. Studies that used an embryo expressing a dominant negative form of GSK-3β suggested that
myo-inositol reversal of dorsalization of the embryonic axis by lithium may be mediated by events independent of inositol monophosphatase inhibition (Hedgepeth et al. 1997). To what extent these findings relate to the thera-peutic action of lithium in the brain has not yet been determined. However, it is at the molecular level that some of the most exciting advances in the un-derstanding of the long-term therapeutic action of lithium will occur over the coming years. It would appear that the current studies of the long-term lith-ium-induced changes in PKC-mediated events, including gene expression and transcriptional modification of important phosphoproteins responsible for regulation of signal transduction in the brain, are a most promising avenue for future investigation. This is particularly of interest in light of converging data from studies of chronic lithium relating its therapeutic efficacy to the regulation of membrane-related events such as ion transport, neurotransmit-ter release, and the receptor-response complex, all of which involve some de-gree of cytoskeletal restructuring. Forthcoming research along these lines also may provide clues to a molecular basis for the pathophysiology of bipolar disorder.