Professional Documents
Culture Documents
com/
Neuroscientist
Published by:
http://www.sagepublications.com
Additional services and information for The Neuroscientist can be found at:
Subscriptions: http://nro.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://nro.sagepub.com/content/8/5/497.refs.html
What is This?
The Wnt signaling pathway is a highly conserved pathway critical for proper embryonic development.
However, recent evidence suggests that this pathway and one of its key enzymes, glycogen synthase
kinase 3β, may play important roles in regulating synaptic plasticity, cell survival, and circadian rhythms in
the mature CNS—all of which have been implicated in the pathophysiology and treatment of bipolar disor-
der. Furthermore, two structurally highly dissimilar medications used to treat bipolar disorder, lithium and
valproic acid, exert effects on components of the Wnt signaling pathway. Together, these data suggest that
the Wnt signaling pathway may play an important role in the treatment of bipolar disorder. Here, the authors
review the modulation of the Wnt/GSK-3β signaling pathway by mood-stabilizing agents, focusing on two
therapeutically relevant aspects: neuroprotection and modulation of circadian rhythms. The future devel-
opment of selective GSK-3β inhibitors may have considerable utility not only for the treatment of bipolar
disorder but also for a variety of classical neurodegenerative disorders. NEUROSCIENTIST 8(5):497–511,
2002: DOI: 10.1177/107385802237176
KEY WORDS Wnt signaling, Glycogen synthase kinase, Lithium, Valproic acid, Neuroprotection, Circadian rhythms, Bipolar disorder
Bipolar disorder (BD, also referred to as manic-depressive the United States alone (Begley and others 2001). It is
illness) is a common, chronic, and often life-threatening thus not altogether surprising that the Global Burden of
illness (Goodwin and Jamison 1990). The disease is Disease Study has identified BD as one of the leading
equally prevalent in men and women and can occur at causes of disability worldwide (Murray and Lopez
any age, with the early twenties being the median age of 1996).
onset (Goodwin and Jamison 1990; Angst and Sellaro Despite the devastating impact BD has on the lives of
2000). Two almost diametrically opposite mood states millions worldwide, little is known about its etiology or
characterize BD: mania and depression. Mania is char- pathophysiology. As is the case with most psychiatric ill-
acterized by a hyperaroused state (either euphoric or nesses, the etiology of BD was once thought to be due
dysphoric), increases in motor activity, racing thoughts, solely to environmental and social causes; however,
impaired judgment, decreased sleep, and an apparent there is now incontrovertible evidence that genetic fac-
decreased need for sleep. The depressive phases of the tors play a major role. Whereas BD has an overall life-
illness, which occur in the same individuals, are charac- time incidence of about 1%, first-degree relatives and
terized by depressed mood, suicidal ideation, anhedonia, monozygotic twin siblings of a patient with BD develop
impaired sleep, reduced psychomotor activity, and the disorder about 7% and 60% percent of the time,
impaired memory and concentration. Although exact respectively (Goodwin and Jamison 1990; Craddock and
figures are difficult to obtain, suicide is estimated to be Jones 1999; Potash and DePaulo 2000). Although bio-
the cause of death in upwards of 15% of individuals with logical factors portend a predisposition for the develop-
BD, and many other deleterious health-related effects are ment of BD, environmental factors also affect the out-
increasingly being recognized (Jamison 1986; Goodwin come. For example, sleep deprivation, drug use, and
and Jamison 1990). The lifetime cost estimate of dis- changes in hormonal levels (i.e., during the postpartum
ability and general medical care of persons with onset of period) are all documented precipitants of mania in sus-
bipolar disorder in 1998 is projected to be $24 billion in ceptible individuals (Goodwin and Jamison 1990). In a
consolidated view, current models for BD—and other
diseases with complex genetics—suggest that an inter-
Peter S. Klein, University of Pennsylvania, provided invaluable com- play between genetic, environmental, and epigenetic fac-
ments regarding the substance and style of this review. Sarah K. Tsou tors causes the disease (Craddock and Jones 2001;
provided excellent editorial assistance. The National Institute of
Mental Heath and the Theodore and Vada Stanley Foundation support Petronis 2001).
our research. Space limitations necessitated the citing of review papers The brain systems that have heretofore received the
in many cases; a full reference list upon which this article was based greatest attention in neurobiologic studies of BD have
is available upon request. been the monoaminergic neurotransmitter systems,
Address correspondence to: Husseini K. Manji, Laboratory of which were implicated by discoveries that effective anti-
Molecular Pathophysiology, Bldg 49, Room B1EE16, NIMH, NIH, 49 depressant drugs exerted their initial biochemical effects
Convent Dr., Bethesda, MD 20892 (email: manjih@intra.nimh.nih.gov). by regulating the intrasynaptic concentrations of sero-
that in Drosophila; however, although a true TIMELESS 2001). Furthermore, CKIε, which we have previously
ortholog has not been identified, orthologs of many of mentioned as an enzyme that interacts with GSK-3β, is
the proteins are present in humans, likely operate in a mutated in the autosomal dominant circadian mutant in
very similar manner, and may therefore play important Syrian hamsters, is the doubletime circadian mutant in
roles in the regulation of human circadian rhythms flies, and also interacts with GSK-3β and the Wnt sig-
(Wager-Smith and Kay 2000; Reppert and Weaver naling pathway, suggesting that these occurrences may
Lithium Valproate
Directly inhibits GSK-3β by competition with Magnesium Inhibits GSK-3β in some in vitro experiments, but not in
Inhibits the phosphorylation of GSK-3β substrates in cell others.
culture Increases back-phosphorylation of a putative GSK-3β
Mimics the effects of Wnt stimulation on development of substrate (GSK-3β enzyme is added to lysate after
diverse species incubation with valproate)
Increases β-catenin protein levels in cell culture Increases β-catenin protein levels in cell culture
Increases β-catenin protein levels in rat hippocampus Increases β-catenin mRNA expression in cell culture
Inhibits GSK-3β mediated potentiation of staurosporine- Increases β-catenin protein levels in rat hippocampus
induced apoptosis in cell culture Inhibits GSK-3β mediated potentiation of staurosporine-
Decreases platelet activating factor mediated increases induced apoptosis in cell culture
in GSK-3β activity in cell culture Decreases platelet activating factor mediated increases
Increases transcription from lef/tcf reporter constructs in GSK-3β activity in cell culture
Increases transcription from lef/tcf reporter constructs
Both lithium and valproic acid have effects on the Wnt pathway. See text and the following references: Maeda (1970); Kao and Elinson
(1988); McMahon and Moon (1989); Harwood and others (1995); Klein and Melton (1996); Stambolic and others (1996); Yost and oth-
ers (1996); Hedgepeth and others (1997); Lucas and Salinas (1997); Emily-Fenouil and others (1998); Lucas and others (1998);
Wikramanayake and others (1998); Yost and others (1998); Chen and others (1999); Hedgepeth and others (1999); Staal and others
(1999); Bijur and others (2000); Chen and others (2000); Vonica and others (2000); Grimes and Jope (2001); Phiel and Klein (2001);
Phiel and others (2001); Ryves and Harwood (2001); Tong and others (2001); Li and others (in press).
any observed biochemical finding. There are not clear recombinant GSK-3β mediated 32P incorporation into
phenotypic changes associated with treatment response, two putative GSK-3β substrates (Chen and others 1999).
and the available animal models are less than ideal. Furthermore, 1-day (and longer) treatment of SH-SY5Y
Indeed, numerous extracellular and intracellular effects cells with a therapeutically relevant concentration of
have been documented for lithium (Wang and others VPA (0.6 mM) resulted in a significant increase in both
1997; Manji and others 1999a). Whereas some of these nuclear and cytoplasmic β-catenin protein levels.
effects are likely related to the treatment efficacy of Additionally, similar to the effects of lithium, VPA also
these drugs, others may cause the numerous side effects, increased β-catenin levels in the rat hippocampus after
or simply be epiphenomenon, with no clinical signifi- chronic (3-week) VPA treatment at therapeutic concen-
cance. An approach some laboratories have undertaken trations (Guang Chen and H. K. Manji, unpublished
to locate potentially therapeutic targets involves identi- observation). Further supporting an action of VPA on
fying targets that are common to structurally dissimilar Wnt signaling, VPA has recently been demonstrated to
agents belonging to the same therapeutic class (e.g., increase lef/tcf-mediated reporter gene expression in
mood stabilizers) (Chen and others 1994; Lenox and Neuro 2A cells (Phiel and others 2001). Additionally,
others 1996; Manji and others 1996). Thus, although both VPA and lithium attenuate a platelet-activating-
chemically distinct agents likely do not work through factor-mediated increase in GSK-3β activity (Tong and
identical mechanisms, identification of targets that they others 2001) (Table 1).
regulate in concert in appropriate paradigms may great- Although the mechanism of lithium’s effect on the
ly serve to validate their therapeutic relevance. In this Wnt pathway appears to be due in large part (if not
context, it is noteworthy that the two most efficacious exclusively) to direct inhibition of GSK-3β, it is less
and most widely used agents in the treatment of BD, clear precisely how VPA inhibits this pathway—both
lithium and valproic acid (VPA), are structurally highly direct and indirect mechanisms may be operative. Thus,
dissimilar. It is thus not altogether surprising that recent Chen and others (1999) reported in vitro inhibition of
studies have begun to explore the possibility that the Wnt GSK-3β and –3α mediated phosphorylation of a CREB
signaling pathway may represent a convergent pathway peptide. Furthermore, a recent finding by Grimes and
for the treatment of BD by investigating the effects of Jope supports a direct inhibitory effect of VPA on GSK-
VPA on this signaling pathway. 3; they found that VPA inhibits the ability of immuno-
precipitated GSK-3β to phosphorylate recombinant
The Wnt Signaling Pathway Is Also a Target human tau in vitro (Grimes and Jope 2001). By contrast,
for the Actions of Valproic Acid Phiel and others found that VPA did not inhibit the in
vitro phosphorylation of glycogen synthase peptide or
Chen and others (1999) first described VPA as an acti- the phosphorylation of tau in cell culture (Phiel and oth-
vator of the Wnt pathway. They observed that incubation ers 2001). VPA also inhibits the phosphorylation of
of human neuroblastoma SH-SY5Y cells with 0.6 mM MAP-1B by GSK-3β in vivo, but not in vitro, suggest-
VPA caused an increase in the subsequent in vitro ing that VPA inhibits the GSK-3β-mediated phosphory-
Cultured Cells (human and rodent) Rodent Brain In Vivo Human Effects
Protects cells against a number of Reduces behavioral deficits and Increases N-acetyl asparate (NAA)
toxic stimuli including choline acetyltransferase activity levels in the brains of BD
induced by cholinergic lesions patients
Glutamate Protects against radiation injury Increases gray matter volume in the
NMDA Protects against cellular and behav- brains of BD patients
Calcium ioral changes after middle cereb- BD patients chronically
MPP+ ral artery occlusion treated with either lithium or val-
Thapsigarin Protects against quinolinic acid proate do not show reduced
β-amyloid toxicity PFCx volumes
Ouabain Enhances hippocampal neurogenesis
Age
Phenytoin
Carbamazepine
KCI deprivation
Serum deprivation
NGF deprivation
GSK-3β + staurosporin
or heat shock
Both preclinical (in cell culture and in intact animals) and clinical evidence suggests that lithium has neuroprotective effects. See Manji
and others (1999b, 2000a, 2000b) and the references found within.
lation of MAP-1B indirectly in intact neurons (Patricia ilar effects (at therapeutically relevant concentrations)
Salinas, personal communication). Thus, VPA inhibition on this important signaling pathway known to regulate
of GSK-3β may be substrate specific, or possibly due to synaptic plasticity, circadian rhythms, and neuronal sur-
differences in assay conditions (Phiel and Klein 2001). vival (Table 1).
At present, the degree to which VPA directly inhibits
GSK-3β, and its potential substrate specificity, requires Preclinical and Clinical Studies Suggest That
further delineation. Nevertheless, it is clear that VPA Mood Stabilizers Have Neurotrophic Effects
increases a downstream target of Wnt signaling, β-
catenin. Further recent data have identified an addition- The evidence reviewed above suggests that GSK-3β and
al pathway—isolated from GSK-3β—by which VPA may β-catenin are targets for the actions of both lithium and
activate the Wnt signaling cascade. VPA is an inhibitor VPA. Do these effects, in fact, have the potential to
of histone deacetylase (HDAC) (IC50 = 0.4 mM) and translate into therapeutic effects in patients? There is
increases β-catenin mRNA expression in Neuro 2A cell now a large body of data that shows that lithium, at con-
lines (Phiel and others 2001). It is noteworthy that other centrations similar to its IC50 for GSK-3 inhibition,
HDAC inhibitors—butyrate (like VPA, a small fatty exerts neurotrophic and neuroprotective effects in a vari-
acid) and trichostatin-A—also increase the transcription ety of preclinical and clinical paradigms (Table 2)
of tcf reporter constructs in cell culture (Bordonaro and (Chuang and others 2002; Jope and Bijur 2002). For
others 1999); thus, VPA’s effects on β-catenin mRNA, β- example, lithium protects cerebellar granule cells
catenin protein levels, and tcf transcriptional activity against glutamate-, NMDA receptor-, low potassium-,
may be mediated in part by HDAC inhibition. and toxic level of anticonvulsant-induced cell death
A final mechanism by which VPA may indirectly (Nonaka and others 1998a, 1998b; Manji and others
inhibit GSK-3β and regulate β-catenin is through the 2000b). In addition to the demonstration of protective
ERK/MAP kinase/RSK pathway. This pathway has effects in vitro, a number of studies have also investigat-
recently been demonstrated to be robustly activated by ed lithium’s neuroprotective effects in vivo. In this con-
VPA (Yuan and others 2001). Furthermore, GSK-3β can text, lithium has been shown to protect against a variety
be phosphorylated by Rsk, suggesting a mechanistic link of excitotoxic and ischemic insults (Pascual and
between the actions of VPA and the regulation of GSK- Gonzalez 1995; Nonaka and Chuang 1998; Arendt and
3β and β-catenin (Sutherland and others 1993; others 1999; Manji and others 1999b, 2000b). Please see
Stambolic and Woodgett 1994; Eldar-Finkelman and Table 2 and references contained within Manji and oth-
others 1995; Kleijn and Proud 2000; Desbois-Mouthon ers (2000b) and Manji and others (1999b) for a complete
and others 2001). review of lithium’s neuroprotective effects.
Overall, although they may regulate the Wnt signaling Although not as well studied, VPA also exerts neuro-
pathway by nonidentical mechanisms, it is striking that protective actions in several paradigms (Bruno and oth-
these two structurally highly dissimilar agents exert sim- ers 1995; Mark and others 1995; Mora and others 1999;
Manji and others 2000b). In a direct comparison, Manji Are the Neurotrophic and Neuroprotective Effects
and others (2000a) reported the neuroprotective effects of Mood Stabilizers Possibly Therapeutically
of lithium and VPA. Lithium and VPA protected Relevant in the Long-Term Treatment of BD?
human neuroblastoma SH-SY5Y cells from both thapsi-
garin- and MPP+-induced apoptosis at therapeutically Mood disorders have traditionally been conceptualized
relevant concentrations. Furthermore, at therapeutic as neurochemical disorders, but there is now evidence
concentrations, the degree of protection from both drugs from a variety of sources demonstrating regional reduc-
was similar (Fig. 3). VPA and lithium also protect SH- tions in CNS volume, as well as reductions in the num-
SY5Y cells from staurosporine-induced apoptosis in bers and/or size of glia and neurons in discrete brain
cells stably transfected with GSK-3β in a model that areas (Manji and others 2000b). Although the precise
has been shown to have the same results after treatment cellular mechanisms underlying these morphometric
with lithium, thus linking the neuroprotective effects of changes remain to be elucidated, the data suggest that
these medications with the Wnt/GSK-3β pathway mood disorders are associated with impairment of struc-
(Bijur and others 2000; Li and others 2002). Inhibition tural plasticity and cellular resilience. Structural imag-
of GSK-3β and/or the activation of the Wnt signaling ing studies have demonstrated reduced gray matter vol-
pathway is an attractive mechanism to explain these ume in areas of the orbital and medial prefrontal cortex
findings. (PFC), ventral striatum and hippocampus, and enlarge-
lithium’s inhibition of GSK-3β represents a putative changes and modulation of circadian rhythms. As dis-
cellular mechanism for the ability of this simple monova- cussed in this review, increasing preclinical and clinical
lent cation to alter circadian physiology in diverse organ- evidence implicates changes in neuroplasticity and cel-
isms. This also raises the possibility that GSK-3β inhibitors lular resiliency in both the pathophysiology and treat-
may have broad utility in the treatment of disorders ment of mood disorders (Jope 1999; Manji and others
associated with abnormalities of circadian rhythms. 2000b). Similarly, evidence accumulated over the past 3
decades suggests that changes in, or the modulation of,
Conclusions and Prospects circadian rhythms may underlie both the underlying
cause of BD and an explanation for why some treatments
Although some indirect in vivo effects should not be for the disorder are efficacious (Kasper and Wehr 1992).
excluded (Chalecka-Franaszek and Chuang 1999), lithi- Although both these areas of focus are certainly strong-
um does directly target GSK-3β—thereby stimulating ly correlated with BD, it remains to be seen with cer-
the Wnt pathway—at therapeutic concentrations (Klein tainty what the true underlying causes are.
and Melton 1996). Similarly, VPA targets the Wnt path- To date, although there is strong suggestive evidence
way and may also have more specific effects on GSK-3β that Wnt/GSK-3β regulation may be related to BD treat-
(Chen and others 1999). Although GSK-3β’s role as a ment efficacy, there is no direct evidence that abnormal-
mediator of many cellular processes leaves the quest for ities of Wnt signaling are present in BD. Although post-
putative downstream physiological targets wide open, mortem brain studies are fraught with methodological
we have focused on two targets that have been a focus of problems—including long postmortem delays, ante-
BD research in the past: regulation of neuroplastic mortem medication and drug use, causes of death that
have variable effects on brain tissue, and the inability to be alterations in other mediators of the Wnt pathway in
adequately study posttranslational modifications—two patients with BD.
studies have recently quantitated GSK-3β protein levels Recently, selective small molecule GSK inhibitors have
in the frontal cortex in BD patients, finding no differ- been developed (Coghlan and others 2000; Cross and oth-
ences from controls (Lesort and others 1999; Kozlovsky ers 2001; Smith and others 2001); these molecules have
and others 2000). However, studies have found altered been used successfully to protect cerebellar granule cells
levels of GSK-3—or other molecules in the Wnt from toxic stimuli (Cross and others 2001). Future research
pathway—in postmortem brains or peripheral cells from may examine these molecules for efficacy in the treatment
patients with another neuropsychiatric disorder, schizo- of BD. Even more promising may be the development of
phrenia (Yang and others 1995; Cotter and others 1998; drugs that, similar to FRAT1, inhibit GSK-3β-mediated
Miyaoka and others 1999; Kozlovsky and others 2000; phosphorylation of specific substrates (e.g., unprimed
Beasley and others 2001). In this regard, it is noteworthy substrates such as β-catenin and axin) but not substrates
that a leading hypothesis posits that early neurodevelop- requiring pre-phosphorylation, such as glycogen syn-
mental changes predispose to the development of schiz- thase, thus being specific for the Wnt-mediated effects
ophrenia in adulthood (Weinberger 1997). Clearly addi- of this multitalented enzyme (Thomas and others 1999).
tional studies are required to investigate these findings This approach may make it possible to discern which
further, as well as to pursue the possibility that there may GSK-3β pathway is relevant for the treatment of BD.