CHAPTER 17: Schizophrenia and Bipolar Disorder

• Schizophrenia is characterized by psychotic symptoms such as hallucinations and delusions (often referred to as “positive” symptoms), negative or deficit symptoms such as lack of motivation and social withdrawal, and cognitive impairments affecting attention, working memory, declarative memory, language fluency, and diverse aspects of social cognition.

• Bipolar disorder is characterized by extreme mood swings between mania (characterized by euphoria or irritability, grandiosity, increased energy, and decreased need for sleep) and depression (characterized by sadness, loss of interest, decreased energy, disturbed sleep and appetite, ideas of worthlessness, guilt, and suicidal ideation). Affected individuals may return to normal mood states (euthymia) between acute episodes, but many have residual symptoms. Psychotic symptoms may occur in either manic or depressed episodes, most often congruent with mood. Bipolar patients with psychosis may have cognitive deficits similar to those seen in individuals with schizophrenia.

• A milder form of the illness, bipolar II disorder is also recognized in which hypomania (mild mania) alternates with depression. Such patients do not generally suffer psychotic symptoms.

• Some individuals experience a mixture of symptoms characteristic of schizophrenia and bipolar disorder and are currently given the diagnosis of schizoaffective disorder.

• The heritability of schizophrenia and bipolar disorder is estimated to be very high, in the range of 65% to 80%. In contrast, the heritability of unipolar depression is less, approximately 35%, with greater environmental influence.

• Despite their high heritabilities, the genetic risk of schizophrenia and bipolar disorder is highly polygenic, with many hundreds of genetic variants each contributing small increments of risk in the population, with different combinations of loci acting in different individuals. Genetic studies have found significant sharing of genetic risk variants between schizophrenia and bipolar disorder.

• Schizophrenia is characterized by loss of gray matter in frontal and temporal regions of the cerebral cortex. Unaffected first-degree relatives may have milder gray matter loss with similar spatial patterns.

• First-degree relatives of individuals with schizophrenia may have attenuated symptoms such as suspiciousness, social isolation, eccentric beliefs, and mild cognitive impairments without psychosis. This condition is currently diagnosed as schizotypal disorder.

• Antipsychotic drugs are effective for the positive (psychotic) symptoms of schizophrenia, for acute manic episodes, and when combined with an antidepressant for psychotic depression. They are also efficacious in preventing acute relapses in schizophrenia, and often in combination with mood-stabilizing drugs in bipolar disorder. They lack efficacy for the negative and cognitive symptoms of schizophrenia.

• The therapeutic action of antipsychotic drugs depends on their ability to antagonize D₂ dopamine receptors. This mechanism of action can also cause significant Parkinson-like motor side effects. Second-generation antipsychotic drugs partly mitigate motor side effects by blocking other neurotransmitter receptors (most significantly 5HT_2A receptors) but have their own significant side effects, in particular, weight gain and elevation of serum glucose and lipids.

• Lithium is effective both for acute mania and for mood stabilization in bipolar disorder. Certain anticonvulsant drugs, such as valproate, are also effective. However, which of these drugs’ many pharmacological actions is responsible for their therapeutic efficacy remains unknown.

Schizophrenia, bipolar disorder, and schizoaffective disorder, like all psychiatric disorders, are still diagnosed solely by clinical observation and history of illness. Despite increasing knowledge of these conditions, objective diagnostic tests have not yet been identified that can be applied with great enough sensitivity (avoidance of false-negatives) and specificity (avoidance of false-positives) to be acceptable for clinical or research use. Thus, neither of the two major diagnostic manuals, the Diagnostic and Statistic Manual of Mental Disorders, 5th ed. (DSM-5), published by the American Psychiatric Association, nor the International Classification of Diseases, 10th ed. (ICD-10), published by the World Health Organization, lists any objective medical diagnostic tests for these disorders. In fact, the difficulty in finding objective tests or biomarkers may partly result from reliance on current diagnostic systems as the “gold standards.” Even for schizophrenia, bipolar disorder, and major depression, which have been closely studied for more than a century, it is clear that the DSM-5 errs in drawing them as discrete, discontinuous categories. Such a categorical model works well for most infectious diseases, for example, but most common, chronic human illnesses tend to be highly heterogeneous with respect to symptoms, severity, disease course, and treatment responsiveness. Psychiatric disorders, including those discussed in this chapter, would be better captured as quantitative deviations from health on multiple dimensions (similar to type 2 diabetes or coronary artery disease). One artifact of the excessively narrowly drawn categories of the DSM-5 is that a substantial fraction of patients with any psychiatric diagnosis have two or more or, as is common with the disorders discussed in this chapter, shifting diagnoses over time. Genetics is now providing some insight into this complex picture as both shared and unshared genetic and environmental risk factors have been discovered for schizophrenia and bipolar disorder. (Overlapping genetic risk factors are well documented for many disorders in medicine including rheumatoid arthritis and psoriasis as well as Crohn disease and ulcerative collitis.) Ultimately, as science progresses, psychiatric diagnosis is likely to look very different from the DSM-5.

Despite the serious shortcomings of DSM-5 diagnoses, well-studied disorders such schizophrenia, bipolar disorder, and autism are not arbitrary fictions; these clinical diagnoses yield very high heritabilities (0.65–0.80) across different countries and cultures. That “rough and ready” utility of these diagnoses for some areas of research such as genetics, however, does not mean that they are useful for others, or that they represent valid “natural kinds.”

Schizophrenia is associated with very severe symptoms and significant disability. It typically begins in late adolescent years or the early twenties; it may rarely begin in childhood, or as late as the early forties. The global mean prevalence of schizophrenia is between 0.5% and 1.0% with local variation. The male to female ratio is approximately 1.4 to 1, with males also tending to have earlier onset. The lifetime prevalence of psychoses of all causes has been estimated to range between 3% and 4% of populations worldwide. Because of its early age of onset, chronic course, and severely impairing effects, schizophrenia ranks among the top 20 global causes of years of healthy life lost to disability, despite its relatively low prevalence.

Symptom Clusters in Schizophrenia

Schizophrenia is characterized by three major symptom clusters denoted as positive (or psychotic), negative (or deficit), and cognitive 17–1. In addition, many patients with schizophrenia have disturbances of mood, most often depression.

The positive (psychotic) symptoms of schizophrenia include hallucinations and delusions 17–2. Negative (deficit) symptoms represent significant impairments of normal functions including loss of motivation, a marked decrease in adaptive social interactions, blunting of affect, and impoverishment of the content of thought and speech. The cognitive symptoms of schizophrenia impair many measurable functions including attention, working memory, declarative memory, verbal memory and fluency, and diverse aspects of social cognition. Most significant as a cause of inability of patients to function at school or work is impaired cognitive control: the reduced ability to regulate thought, emotion, and behavior in accordance with goals. Long before the advent of modern cognitive science, the German psychiatrist Emile Kraepelin described the mind in schizophrenia as an orchestra without a conductor. Because of cognitive impairments and negative symptoms, people with schizophrenia continue to be disabled even when their psychotic symptoms are successfully controlled with available medications.

It is hypothesized that the different symptom clusters characteristic of schizophrenia result from abnormalities in different, although partially overlapping, neural circuits. For example, psychotic symptoms are thought to reflect abnormalities in subcortical dopamine systems, while impairment in cognitive control is thought to reflect abnormalities in circuits based in the prefrontal cortex. The involvement of diverse neural circuits is thought to result from pathological processes that unfold during development and affect diverse cells, synapses, and circuits in the brain.

The Schizophrenia Prodrome

A diagnosis of schizophrenia is typically made only with the first episode of florid psychotic symptoms, but there is generally—perhaps always—a prodromal period in mid to later teen years. The schizophrenia prodrome may first manifest with declining school or vocational performance, increasing social isolation, unstable mood, irritability, anxiety, and increasingly odd and eccentric thoughts. It may initially be challenging to distinguish the schizophrenia prodrome from depression, drug use, or other pathological conditions of adolescence. As the schizophrenia prodrome progresses, however, the affected individual begins to have mild or attenuated psychotic symptoms (see 17–2), at which point it becomes easier to distinguish from other forms of adolescent psychopathology.

When individuals with the schizophrenia prodrome are carefully examined, they exhibit broad declines in cognitive functioning from their earlier baselines as well as the emergence of negative symptoms 17–1. These cognitive impairments and negative symptoms are not typically as severe as they will become after a first psychotic episode. Nonetheless, a significant fraction of the impairment that characterizes schizophrenia may be established before the first psychotic symptoms emerge. It has been hypothesized that cognitive and negative symptoms may, in fact, be a risk factor for psychosis. In some studies using structural magnetic resonance imaging (MRI), individuals with prodromal schizophrenia show cerebral cortical thinning, most severe in frontal and temporal cortices 17–2, that occurs more rapidly and to a greater extent than that observed in typically developing adolescents. However, measures of cortical thickness cannot currently be used as biomarkers of the schizophrenia prodrome since adolescence is a period in which synaptic pruning normally occurs 17–3. If measures of change in cortical thickness are to yield useful diagnostic information, they will have to meet the challenging test of reliably showing a faster rate of change or a distinctly more significant degree or spatial pattern of loss than control subjects, a difficult proposition against a normal background of dynamic cortical remodeling.

The schizophrenia prodrome creates the possibility of early intervention, ideally before severe pathophysiological changes have occurred, such as excessive and inappropriate dendritic pruning and synaptic loss. In multiple clinical trials conducted in prodromal subjects, antipsychotic drugs have convincingly failed to delay the onset of psychosis (but do produce serious side effects), suggesting that entirely different mechanisms of action will be needed. Indeed, the therapeutic goals should be to alter pathological developmental processes, and thus to move beyond the palliation of late-stage psychotic symptoms. Such efforts will require an improved understanding of the molecular and cellular basis of the underlying abnormalities. Cognitive remediation designed for prodromal subjects has shown early promise in delaying conversion to psychosis. As many as 20% of apparently prodromal individuals at high risk of conversion to psychosis make a durable recovery independently of treatment for reasons that are not understood.

Once a first psychotic episode has occurred, the course of schizophrenia is chronic, with periods of milder “residual” psychotic symptoms punctuated by relapses of florid psychosis. Treatment of a first episode with antipsychotic drugs generally brings a full remission of the positive symptoms. Unfortunately, while antipsychotic drugs extend the period between relapses, relapses do occur and over time respond less completely to treatment. In addition, negative and cognitive symptoms, which are generally present before the onset of psychosis, worsen over time, eventually reaching a plateau well below the affected person’s baseline, and are unresponsive to current pharmacological treatments. The side effect burden of antipsychotic drugs contributes to their discontinuation by patients, which can lead to more frequent relapses.

Genetic and Nongenetic Risk Factors for Schizophrenia

Based on twin studies, it has long been recognized that genes play a major role in risk of schizophrenia and bipolar disorder. Monozygotic (MZ) twin pairs, whose DNA sequences are 100% identical, have a concordance for schizophrenia of nearly 50%. Given a population prevalence of 0.5% to 1.0%, this is a 50- to 100-fold increase in risk for the MZ cotwin of an affected member of the twin pair. In contrast, dizygotic twin pairs, who share, on average, 50% of their DNA sequences, have a concordance for schizophrenia of approximately 15%. In addition, adoption studies found that risk of schizophrenia is determined by the biological rather than adoptive families. Overall, twin methodologies give heritabilities of 0.65 to 0.8 for schizophrenia and bipolar disorder, meaning that 65% to 80% of the phenotypic variance is explained by genes. This is among the highest rates of heritability for relatively common human diseases. (Diseases caused by highly penetrant single gene mutations—so-called Mendelian disorders—are rare because they more strongly diminish the likelihood of having offspring.)

Family and twin studies have also found that nonpsychotic first-degree relatives of individuals who meet criteria for schizophrenia often exhibit cognitive impairment and cortical thinning intermediate between population norms and their relatives with schizophrenia. An observation first made in adoption studies, but highly replicated in family studies thereafter, is that individuals genetically related to persons with schizophrenia may exhibit social isolation, suspiciousness, eccentric beliefs, and magical thinking, but lack frank psychotic symptoms such as hallucinations and delusions. When such symptoms produce impairment in functioning, the individual is diagnosed with schizotypal disorder (or in some older nomenclatures, by the misnomer, schizotypal personality disorder).

Despite the high heritabilities of schizophrenia and bipolar disorder, success in discovering specific disease risk–associated genetic variants was unsuccessful until recent advances in genomic technologies made it possible to search for genetic risk factors at the scale necessary to investigate a highly polygenic phenotype, that is, one in which each of a very large number of genes contributes small increments of risk. While the aggregate risk conferred by genes is large for schizophrenia and bipolar disorder, it is partitioned among numerous genes—likely many hundreds, perhaps more than 1000—of which no single gene is necessary or sufficient.

At the time of this writing, a global consortium has used DNA microarrays to genotype more than 40,000 patients with schizophrenia and 40,000 control subjects for common genetic variants and has determined the complete DNA sequence for the exomes (all protein-coding genes) of several thousand schizophrenia patients and control subjects. The microarrays (“gene chips”) permit the determination of the version of a DNA segment a person has at approximately 750,000 genomic loci and then enables the investigator to ask whether each variant studied increases the risk of schizophrenia in the population studied. The result has been the discovery of more than 100 genetic loci that influence the risk of schizophrenia and a prediction that there are likely to be more than 1000 genes (5% of the human genome) in which variation influences the risk of this syndrome. Some of the risk variants that have been identified influence such well-known neural proteins as L-type and T-type Ca²⁺ channel subunits 17–3 as well as components of the postsynaptic specializations of excitatory synapses (see 5–1 in Chapter 5). Given the goal of understanding disease pathogenesis and developing disease-altering treatments, efforts are under way to discover the majority of risk-associated loci. This will permit the identification of pathways that can yield targets for treatment development. The genetic analysis of bipolar disorder is lagging somewhat behind schizophrenia, based on the identification of populations and collection of samples, but is likely to catch up soon.

Syndromal forms of autism (ie, autism associated with such features as low IQ, seizures, and dysmorphic facies), and less commonly schizophrenia, can be associated with chromosomal deletions, duplications, or more complex rearrangements, all referred to as copy number variants (CNVs). CNVs may affect DNA segments ranging from several hundred kilobases to several megabases in length and may affect multiple genes. Several CNVs are among the most penetrant genetic risk factors for schizophrenia, for example, the 22q11.2 syndrome, which results from a deletion of variable length affecting approximately 3 megabases of DNA and up to 44 genes in the middle of chromosome 22. The 22q11.2 syndrome, previously called velocardiofacial syndrome (for its facial and cardiac defects) or DiGeorge syndrome (for its discoverer), is characterized by cleft palate, congenital heart disease, facial dysmorphology, hypoparathyroidism, and other abnormalities. Approximately 25% of individuals with the 22q11.2 syndrome develop schizophrenia; however, approximately 20% receive a diagnosis of autism and the majority warrant neither psychiatric diagnosis. As is the case with most CNVs that have been studied, the most penetrant central nervous system (CNS) phenotype is intellectual disability (mental retardation). To summarize an evolving story, CNVs that cause neuropsychiatric symptoms primarily produce impairments across multiple cognitive domains (intellectual disability), but may also significantly increase the risk of autism or schizophrenia. The resulting psychiatric phenotypes, if any, depend not only on the presence of a particular CNV but also on the polygenic background in which the CNV occurs. The incomplete penetrance and variable expressivity (multiple phenotypes) exhibited by the 22q11.2 deletion and most other CNVs illustrate the genetic complexity of the risk for neuropsychiatric disorders.

One chromosomal rearrangement that deserves mention, not for its scientific significance, but for the attention it has garnered, resulted from a balanced exchange between chromosomes 1 and 11 in a Scottish family. The family member who first came to medical attention had schizophrenia, and thus a gene damaged by the translocation was named disrupted in schizophrenia (DISC-1). DISC-1 plays interesting roles in brain development and plasticity and thus appears to be a plausible risk gene. However, examination of the Scottish family over multiple generations reveals that affected individuals with the translocation predominantly have severe depression; several have schizophrenia, a small number have bipolar disorder; and several family members with the translocation have no psychiatric diagnosis. Moreover, well-powered population studies have not found an association between DISC-1 and schizophrenia. Despite the name given to DISC-1 and its interesting biology, it does not at present appear to figure in the pathogenesis of schizophrenia.

In addition to genetic risk factors for schizophrenia, epidemiological studies have identified several environmental risk factors that have been replicated by different groups. These include urban birth, increasing paternal age, season of birth, intrauterine exposure to viral infection, and adolescent use of cannabis. Maternal starvation during pregnancy has been identified as a risk factor for schizophrenia based on famine in the Netherlands following World War II and in China. Risk factors such as urban birth and migration are clearly proxies for causal factors that are not yet identified. Perhaps greater knowledge of genetic risk factors will be helpful in better understanding the causal mechanisms of environmental risk factors.

Pathogenesis

Given the still early stage of the genetic analysis of schizophrenia, it is only possible to speculate about molecular mechanisms of pathogenesis. One very general hypothesis is that genetic risk involves some combination of synapses vulnerable to inappropriate pruning and perhaps other developmental mechanisms that eliminate far too many synapses. This hypothesis is consistent with microscopic postmortem findings, which do not find cell loss in the cerebral cortex of patients with schizophrenia, but rather significant loss of dendritic arbors and synaptic spines from cortical neurons. The loss of neural processes would be consistent with the observation of smaller cell bodies (which would require less volume to support the metabolic needs of diminished processes) and thus denser cell packing in cortical layers. The most consistent gross anatomic finding, based on structural MRI and postmortem brain examinations, is thinning of cerebral cortical gray matter, with the greatest tissue loss in prefrontal and temporal regions (see 17–2), and a compensatory enlargement of the lateral and third cerebral ventricles. Loss of dendritic spines and thus synapses in prefrontal and temporal cortex could explain the cognitive deficits of schizophrenia as well as the milder cognitive impairments of first-degree relatives who have cortical thickness intermediate between schizophrenia and healthy individuals. Consistent with intermediate performance on cognitive tasks, first-degree relatives show resting-state connectivity intermediate between healthy controls and patients with schizophrenia 17–4.

The excessive elimination of cerebral cortical synapses in schizophrenia implicates the major excitatory and inhibitory neurotransmitter systems glutamate and γ-aminobutyric acid (GABA), respectively. A role for excitatory synapses in the pathogenesis of schizophrenia is supported further by early genetic findings, which include genes encoding several subunits of NMDA, AMPA, and metabotropic glutamate receptors (eg, GRIA1, GRIN2A, GRIN2B, GRM3) or regulatory proteins (NRGN).

Historically, it was hypothesized that NMDA glutamate receptors are involved in the pathogenesis of schizophrenia because phencyclidine, ketamine, and related drugs (see chemical structures in Chapter 16), which act as open channel blockers of NMDA receptors (Chapter 5), produce psychotic-like symptoms and cognitive deficits reminiscent of some symptoms of schizophrenia. In fact, phencyclidine or ketamine administration in animals, and ketamine administration in human volunteers, continues to be used as a model in pharmaceutical development for schizophrenia-like effects. These observations notwithstanding, it is not clear whether phencyclidine or ketamine administration to healthy animals or human volunteers recapitulates molecular mechanisms of schizophrenia or produces phenocopies of the syndrome. This is a significant issue in the development of therapeutics, since the reversal of pharmacologically induced deficits in an animal model (eg, ketamine administration) will only benefit patients with schizophrenia if the underlying pathogenic mechanism in the human patients has been appropriately modeled in the animals.

Both phencyclidine and ketamine were developed as dissociative anesthetics. Both have significant abuse potential; both release dopamine, may be experienced as rewarding, and can prove highly addictive (Chapter 16). Phencyclidine is not legally marketed for human use, and is known on the street as PCP or angel dust. Ketamine, although also abusable (with a street name of special K), has continued to be used in anesthesia and analgesia and more recently has gained attention for exerting rapid, even if transient, antidepressant effects in humans (Chapter 15). Both drugs produce a sense of depersonalization and dissociation of subjective experience from sensory stimuli. Thus, in addition to having antinociceptive properties, they also decrease the aversiveness of normally painful stimuli.

Dopamine

This monoamine neurotransmitter has long been implicated in the production of psychotic symptoms based on the fact that all effective antipsychotic drugs (described below) are D₂ dopamine receptor antagonists (Chapter 6). Moreover, high and repeated doses of amphetamines or cocaine, which increase synaptic levels of dopamine, or L-DOPA, which is converted to dopamine by L-aromatic amino acid decarboxylase, can cause delusions and hallucinations. Historically, these findings led to an early hypothesis that schizophrenia is caused by excessive levels of dopaminergic neurotransmission. With the recognition that schizophrenia results from abnormal developmental processes and that psychotic symptoms are a late manifestation of the disease process (coming at the end of the prodrome), it would be more fruitful to investigate dopamine for its contributions to psychotic symptoms rather than to the schizophrenia syndrome as a whole.

Interestingly, the gene encoding the dopamine D₂ receptor has recently emerged as a genome-wide significant risk factor for schizophrenia. Nevertheless, the mechanism by which blockade of D₂ receptors reduces psychotic symptoms remains unclear. Antipsychotic drugs block D₂ receptors on initial drug exposure, but their antipsychotic effects (as opposed to sedative effects) emerge only after days or weeks of drug treatment, suggesting that significant molecular and cellular adaptations must first occur.

Animal and cellular models

Despite the evolutionary distance between mice and humans (most recent common ancestor ~90 million years ago) and thus the vast differences between human and mouse cerebral cortex, among many other differences, it is possible to produce useful mouse models of monogenic brain disorders such as fragile X syndrome, Rett syndrome, and rare monogenic forms of autism (Chapter 14). In contrast, there is no animal model for schizophrenia. No monogenic forms of the illness have been found from which to construct genetic mouse models. Even the most highly penetrant genetic risk factor known, the 22q11.2 deletion syndrome described earlier, depends in humans on a polygenic background to determine whether affected individuals develop schizophrenia or autism, or have no mental illness. Given that all 22q11.2 patients have intellectual disability, it is difficult to interpret the psychiatric significance of any behavioral deficits in mice with a knockout of the orthologous chromosomal region. Moreover, the hypothesized pathophysiological basis of schizophrenia, excessive cortical pruning resulting in loss of neuropil and associated with cognitive decline, has not been described in such mice. Perhaps as the genetic analysis of the syndrome progresses and as new genome engineering technologies advance, it will be possible to “knock-in” large numbers of human risk alleles in mice or other animals in order to study molecular and cellular aspects of the pathogenesis of schizophrenia. In the mean time, attempts have been made to model specific symptom clusters.

Currently, amphetamine administration in mice and rats is used to screen for the in vivo efficacy of D₂ dopamine receptor antagonists, but this is not a model of psychotic symptoms. How can one document hallucinations or delusions in a rodent or even a nonhuman primate? Modeling the negative and cognitive symptoms is more plausible, but abnormalities in these behavioral domains are common across several psychiatric syndromes and hence lack specificity. Another limitation in the use of rodents is that prefrontal cortical regions that are thought to be impaired in schizophrenia are highly rudimentary in rodents compared with primates (Chapter 14). Given the lack of genetic animal models and the challenge of studying specific symptom clusters, many groups use pharmacological challenges, such as NMDA receptor antagonists or psychostimulants, as noted above, measuring end points such as locomotor activity or prepulse inhibition—the ability of a prior stimulus (eg, noise) to suppress startle responses to a subsequent noise. However, it is far from certain that these drugs cause neurobiological abnormalities that recapitulate those seen in schizophrenia, or that the behaviors measured reflect core symptoms of the illness. Likewise, groups have demonstrated behavioral abnormalities in rodents that are treated during development with drugs that inhibit cell division or with drugs that induce an inflammatory response, but until we learn more about the pathophysiology of schizophrenia in humans, it is impossible to know the extent to which these approaches in animals are relevant to the human syndrome.

Based on these challenges in developing bona fide animal models of schizophrenia, investigators are turning increasingly to the study of induced neurons. While cellular models cannot be used to study circuit abnormalities of behavior, they can be used to study molecular and cellular abnormalities produced by individual risk alleles or, using patient-derived cells, abnormalities produced by a complex genetic liability to schizophrenia (or bipolar disorder or any other genetically influenced illness). In the most commonly used methods, fibroblasts from patients and control subjects are reprogrammed into induced pluripotent stem cells (iPS cells) that can then be differentiated into diverse cell types including neurons (see 1–1 in Chapter 1). Alternatively, fibroblasts can be reprogrammed directly into neurons. These technologies are very recent, and it is still not possible to reliably generate mature neuronal cell types of interest—however, progress is this regard is occurring rapidly. Abnormalities in morphology, synapse formation, and migration have been reported for neuron-like cells derived from patients with schizophrenia; however, this work is difficult to interpret at this stage because it has not yet been possible to determine whether the phenotypic differences are caused by schizophrenia risk genes, as opposed to other genetic differences (not associated with disease) in the cells. Using modern genome engineering technologies, it is becoming increasingly feasible to swap risk-associated alleles into and out of iPS lines, making it likely that these technical problems can be solved. In psychiatry and neurology, and indeed across all fields of medicine, the shared goal is to produce cellular models of genetically influenced disorders that can be used to screen for novel therapeutics in vitro.

Mood disorders are heterogeneous, but are generally divided into two major classes based on symptoms and patterns of familial transmission. As discussed in Chapter 15, individuals with unipolar depression experience episodes of depression only and are more likely to come from families with both anxiety disorders and depression. Individuals with bipolar disorder experience mania as well as depression. It has been observed that bipolar disorder and schizophrenia may occur in the same families along with intermediate conditions that often go by the name of schizoaffective disorder. Based on genome-wide association studies, they share approximately 70% of the common genetic variants associated with each disease phenotype. Recent reports also describe cognitive deficits and cortical thinning in bipolar patients who suffer psychotic episodes, albeit less severe than those observed in schizophrenia. Finally, as noted, there are poorly delineated intermediate states lumped as schizoaffective disorder that involve full symptoms of bipolar disorder as well as psychotic symptoms that extend in time after the resolution of acute episodes of mania or depression. The similarities between schizophrenia and bipolar disorder notwithstanding, there are also important differences. In bipolar disorder, even in the presence of psychosis, the abnormalities of mood are prominent; some patients return close to baseline functioning between episodes; and bipolar patients benefit from Li⁺, whereas patients with schizophrenia do not.

When bipolar disorder includes full episodes of mania, it is often described as type 1 bipolar disorder. Individuals with milder episodes of mania (hypomania) and full episodes of depression are said to have type 2 bipolar disorder. Individuals affected by bipolar I disorder generally have a far more severe and disabling course. The prevalence of bipolar I disorder is approximately 1% of the population worldwide and affects men and women equally.

Mania is characterized by a severe disturbance in mood, either euphoria or irritability, or the two in combination. Mania is typically accompanied by dramatic increases in energy, decreased need for sleep and nourishment, impulsiveness, and excessive engagement in reward-directed behaviors such as spending sprees, use of alcohol or other drugs of abuse, and engagement in excessive sexual behavior. During manic episodes individuals may become grandiose, often delusionally so. Affected individuals may think of themselves as gods, prophets, or global leaders, or as controlling world events. Mania also may be characterized by thoughts that are moving so fast as to be confusing to the affected individual (flight of ideas), and by rapid, pressured (difficult to interrupt) speech. As noted, psychotic symptoms occur frequently during manic episodes. The content of delusions and hallucinations is generally consistent with the person’s elevated mood. For example, people with mania may have delusions that they possess special powers. The symptoms that characterize depressive episodes in bipolar disorder are also mood congruent, but negative in valence.

Individuals affected by bipolar disorder have recurrent episodes of the illness, both manias and depression. However, the rate of cycling between mania, depression, and normal mood (euthymia) varies widely. Individuals with rapid cycling (in which mood states can change several times per week) tend to respond poorly to treatment. Between periods of mania and depression, a fraction of people with bipolar disorder may be relatively free of symptoms, but a larger fraction have residual symptoms, most often depression.

Research on the neurobiology of bipolar disorder has proceeded slowly, in part reflecting the lack of convincing animal models. The symptoms of bipolar disorder would seem to implicate the brain’s reward circuitry and prefrontal cortical circuits in mania, as well as circuits similar or identical to those involved in unipolar depression during bipolar depressed episodes. Circadian clock-related mechanisms (discussed in Chapter 13) have also been implicated, based on the rapid cycling between mania, depression, and euthymia, and the ability of sleep deprivation to induce manic episodes, in some patients with bipolar disorder. Interestingly, mice with genetic disruption of the Clock gene display many symptoms—increased activity, enhanced reward, reduced fear, and decreased sleep—reminiscent of a manic episode. It is hoped that progress in the genetic analysis of bipolar disorder, which is proceeding effectively at the time of this writing, will help with pathophysiological understanding of the syndrome and ultimately better treatments.

Chlorpromazine and reserpine, the first antipsychotic drugs, were introduced in the early 1950s. Chlorpromazine, in particular, revolutionized the treatment of psychotic disorders including schizophrenia, bipolar disorder, and psychotic depression. Chlorpromazine 17–5 was initially developed in 1951 as an antihistamine for use as a preanesthetic. The drug was observed to be highly sedating both in animals (where it was said to induce “artificial hibernation” in part because it lowered body temperature) and in surgical patients. Its effectiveness in psychotic patients was later demonstrated. Beyond the simple sedation produced by barbiturates, and widely used to manage patients with schizophrenia or mania, chlorpromazine produced remission of psychotic symptoms such as hallucinations and delusions over a period of weeks. Especially in higher doses chlorpromazine also produced extrapyramidal motor side effects reminiscent of Parkinson disease such as tremor and rigidity. In the early 1960s, chlorpromazine and the subsequently discovered antipsychotic drug haloperidol 17–5 were found to increase the turnover of catecholamines (dopamine and norepinephrine) in rodent brain, an observation which led to the discovery that antipsychotic drugs act by blocking dopamine receptors.

The recognition that antipsychotic drugs block dopamine receptors tied their therapeutic effects to their Parkinson-like side effects, which were caused by the same mechanism. (Thus, extrapyramidal side effects represent “on-target” toxicity of antipsychotic drugs.) Soon motor side effects such as catalepsy became a screening tool for new antipsychotic drugs in laboratory animals. The term neuroleptic, Greek for neuron clasping, was applied to the first generation of antipsychotic drugs, referring to the motor side effects of these medications. As receptor binding studies were developed in the 1970s, the therapeutic effects of antipsychotic drugs were clearly associated with blockade of D₂ dopamine receptors (see below).

Studies of chlorpromazine roughly coincided with the discovery of the antipsychotic properties of reserpine. Reserpine, the active component of snakeroot or Rauvolfia serpentina, had been used as a traditional medicine in India. It was initially used as an antihypertensive agent in the United States until clinical trials demonstrated its effectiveness as an antipsychotic drug. At therapeutic doses reserpine caused motor side effects similar to those associated with the use of chlorpromazine. It was later discovered that reserpine acts by depleting dopamine and other monoamine stores and thereby decreasing monoamine neurotransmission, an action mediated via blockade of the vesicular monoamine transporter (Chapter 6). Thus, despite very different molecular targets, both reserpine and chlorpromazine were shown to interfere with dopamine neurotransmission.

Mechanism of Antipsychotic Drug Action

Effects of antipsychotic drugs on D₂ dopamine receptors

Receptor binding studies, primarily in the rat striatum—a rich source of such receptors—identified two major dopamine binding sites, termed D₁ and D₂. Although some older antipsychotic drugs, such as chlorpromazine, are potent antagonists of both receptor types, antagonism of the D₂ receptor strongly correlates with antipsychotic efficacy 17–6. Haloperidol, for example, a butyrophenone compound that is a potent antipsychotic medication, has a high affinity for the D₂ but not the D₁ receptor. Molecular cloning studies later revealed that the D₁ binding site consists of D₁ and D₅ receptors and that the D₂ binding site consists of D₂, D₃, and D₄ receptors (Chapter 6). Among these the D₂ receptor is the critical molecular target of antipsychotic drug action.

The mechanism by which antagonism of D₂ receptors produces antipsychotic effects is not fully understood. Antipsychotic drugs usually take several weeks to exert their full antipsychotic effects; thus, it is not direct antagonism of D₂ receptors that decreases psychosis; rather it is slowly developing adaptations in cells, synapses, and neural circuits resulting from sustained blockade of D₂ receptors. In addition, D₂ dopamine receptors are both presynaptic and postsynaptic (Chapter 6) and are expressed on neurons in the striatum at very high levels, and at lower levels in the prefrontal cortex, amygdala, and hippocampus (where D₁ dopamine receptors predominate). Pyramidal neurons in the prefrontal cortex receive glutamatergic inputs both from other cortical regions and from the thalamus. These corticocortical and thalamocortical glutamatergic synapses are modulated by local GABAergic interneurons and by numerous afferent projections, which include not only dopamine projections from the ventral tegmental area but also serotonergic, noradrenergic, cholinergic, and multiple peptidergic inputs. How sustained D₂ receptor blockade reorders the functioning of this circuitry remains an intense area of investigation.

Side effects of antipsychotic drugs

Although antipsychotic drugs revolutionized the treatment of schizophrenia and other psychotic disorders, their efficacy is limited to psychotic (positive) symptoms and is incomplete. Most patients have residual psychotic symptoms and continue to have florid psychotic relapses. Moreover, the drugs have serious side effects. Side effects of the first generation of antipsychotic drugs resulted from both “on-target” effects mediated by D₂ receptor blockade and “off-target effects” mediated by actions at many other receptors 17–1.

Side effects caused by D₂ dopamine receptor antagonism include extrapyramidal motor side effects and hyperprolactinemia. The synthesis and release of prolactin by anterior pituitary lactotrophs are inhibited by dopamine, which is released from the arcuate nucleus of the hypothalamus acting on D₂ receptors (Chapter 10). Thus, antipsychotic drugs with high affinity for D₂ receptors cause elevated levels of prolactin that can result in sexual dysfunction and rarely galactorrhea.

Extrapyramidal side effects can be divided into four major categories. The first consists of a syndrome that resembles Parkinson disease. This syndrome is characterized by rigidity, difficulty in initiating movements, mask-like facial expressions, and a resting tremor. A second type is acute dystonia, which is characterized by a sudden and spastic contraction of muscles—often those of the face and neck. Dystonic reactions may be dramatic, painful, and disturbing to individuals who experience them. A third category of extrapyramidal side effects consists of akathisia, characterized by a subjective sense of anxiety and intense restlessness. The fourth category is tardive dyskinesia, late-onset involuntary abnormal choreiform movements. Tardive dyskinesia occurs only with long-term drug exposure and is not readily reversible after cessation of drug treatment. An often debilitating side effect, risk of tardive dyskinesia has limited the use of first-generation antipsychotic drugs in countries where there is access to second-generation drugs.

Because D₂ receptor antagonism is both the therapeutic mechanism of antipsychotic drugs and the cause of both elevated prolactin levels and extrapyramidal side effects, all antipsychotic drugs have some liability for these side effects. However, the risk correlates with the affinity of a drug for the D₂ dopamine receptor and the dose of the drug used. High-potency antipsychotic drugs, such as haloperidol, have a high affinity for D₂ receptors, and thus have a greater liability for these side effects than lower-potency drugs, such as chlorpromazine. In addition, many low-potency first-generation antipsychotic drugs possess anticholinergic (antimuscarinic) properties that tend to ameliorate extrapyramidal side effects, although not hyperprolactinemia (Chapters 6 and 17). In fact, when high-potency first-generation antipsychotic drugs are used, they are often coadministered with drugs that have antimuscarinic properties such as benztropine or diphenhydramine.

“Off-target side effects” of antipsychotic drugs include sedation, which results from the blockade of H₁ histamine receptors, muscarinic cholinergic receptors, and possibly α₁-adrenergic receptors in the CNS. Hypotension results from the antagonism of α₁-adrenergic receptors in the periphery. Autonomic side effects such as dry mouth, blurred vision, urinary retention, and constipation (Chapter 9) are caused by the antagonism of muscarinic cholinergic receptors again in peripheral tissues. Delirium, which may occur in the elderly or in patients receiving multiple anticholinergic drugs, is most often due to blockade of muscarinic cholinergic receptors in the CNS (Chapter 14). Even among the first-generation drugs there is a wide diversity of affinities for different neurotransmitter receptors (see 17–1) permitting some ability to trade off side effects depending on the situation of a particular patient.

Second-generation antipsychotic drugs

First-generation antipsychotic drugs were very successful in their ability to treat the psychotic symptoms of schizophrenia and other disorders. They had significant limitations, however, because of side effects, as noted above. The prototype for all second-generation antipsychotic drugs is clozapine, which was discovered in the 1960s 17–7. Clozapine was initially believed to be an ordinary low-potency antipsychotic drug. However, it produced far less severe extrapyramidal side effects than expected. Despite this apparent benefit, its wide use was delayed for many years because approximately 1% of patients treated with clozapine develop agranulocytosis (a significant decrease in white blood cells), a potentially fatal but usually reversible side effect. Clozapine had other problematic side effects as well, among them, sedation, weight gain, and a decrease in seizure threshold. However, a clinical trial in the 1980s demonstrated that clozapine could produce significant benefit to schizophrenic patients who had failed to respond to other antipsychotic drugs. The combination of unique efficacy and diminished risk of motor side effects stimulated the development of new drugs, using some of the properties of clozapine as a prototype, while trying to avoid its liabilities.

Pharmacology of second-generation antipsychotic drugs

Detailed examination of the receptor binding properties of clozapine revealed that it has antagonist or possibly inverse agonist properties at many neurotransmitter receptors 17–8. An important characteristic that explains its low risk of extrapyramidal side effects and lack of effect on prolactin levels appears to be its relatively low affinity for D₂ receptors. Serotonergic mechanisms had once elicited interest as a possible contributor to psychotic disorders because lysergic acid diethylamide (LSD) and related psychedelic drugs produce hallucinations as partial agonists at the 5HT_2A serotonin receptor (Chapter 6). It was recognized, however, that the effects of LSD and related hallucinogens were quite unlike the symptoms of schizophrenia. Nevertheless, clozapine rekindled interest in serotonin because it is a high-affinity antagonist of multiple serotonin receptors including the 5HT_2A receptor. While there is no compelling evidence that 5HT_2A receptor blockade produces antipsychotic effects—for example, highly selective 5HT_2A antagonists do not exert antipsychotic effects by themselves or potentiate the actions of D₂ antagonists—several lines of evidence suggest that blockade of 5HT_2A receptors limits the extrapyramidal side effect liability of antipsychotic medications. As a result, a high ratio of 5HT_2A to D₂ receptor affinity became a design principle for many second-generation drugs. Examples 17–7 include risperidone, olanzapine (a close chemical analog of clozapine), quetiapine, ziprasidone, sertindole, and asenapine. Aripiprazole 17–7 is similar to the aforementioned drugs by virtue of being a high-affinity antagonist at 5HT_2A receptors. However, it differs in that the other second-generation drugs are D₂ receptor antagonists (and some, perhaps inverse agonists), whereas aripiprazole is a high-affinity D₂ receptor partial agonist. As a partial agonist, aripiprazole can act as an agonist or antagonist depending on the strength of endogenous dopamine neurotransmission. It would act as an agonist in low dopaminergic states and as an antagonist in hyperdopaminergic states. In practice, it has low extrapyramidal side effect liability and does not elevate prolactin.

A recent finding with potential implications for the mechanism of action of antipsychotic drugs is that D₂ receptor signaling bifurcates intracellularly (Chapter 6). Depending on the particular agonist or antagonist, receptor actions are biased toward either the classical G_i/G_o signaling pathway or a β-arrestin–dependent pathway that activates AKT (previously called protein kinase B) and glycogen synthase kinase-3β (GSK3β), both of which are serine/threonine protein kinases discussed in Chapter 4. This is an example of ligand-directed signaling, which could be exploited in the development of novel antipsychotic drugs, based on emerging evidence that the antipsychotic and extrapyramidal side effects of D₂ antagonists might be differentially mediated via these two signaling cascades. GSK3β inhibition has been implicated in the actions of Li⁺ as will be discussed in later sections of this chapter.

Comparative clinical trial data and clinical experience suggest that second-generation drugs are not generally more efficacious than first-generation drugs, and that both classes are less efficacious than clozapine. Like first-generation drugs, the second-generation drugs lack any significant efficacy for cognitive and negative symptoms. While second-generation drugs have a lower liability for causing extrapyramidal side effects than first-generation drugs, they are not free of such side effects. Moreover, they have a significant risk of causing weight gain and metabolic derangements such as hyperglycemia and hyperlipidemia, termed metabolic syndrome. The latter side effects of the newer agents can be severe, with patients gaining as much as 10% to 20% of body weight with chronic treatment, and are believed to be mediated at least in part via antagonism of 5HT_2C receptors, which are known to be important in the control of feeding behavior (Chapter 10).

Despite decades of investigation, the mechanism responsible for clozapine’s unique clinical properties remains to be identified. Clozapine binds many receptors (see 17–8). For example, it has relatively high affinity for the 5HT_2C, 5HT₃, and 5HT₆ receptors as well as the 5HT_2A receptor. In addition, clozapine and its metabolites bind many additional neurotransmitter receptors. One view is that it is the mixture of receptor actions that is important, but this remains hypothetical.

Other strategies for treating schizophrenia

As described, the NMDA glutamate receptor has been proposed as a target for antipsychotic drug action based on the psychotomimetic properties of NMDA open channel blockers such as ketamine and phencyclidine (see 16–4 in Chapter 16). Perhaps more convincing is the observation that several ionotropic and metabotropic glutamate receptor subunits and other components of postsynaptic excitatory synapses are implicated in schizophrenia pathogenesis by unbiased, large-scale genome-wide association studies, as stated above.

If NMDA receptor hypofunction contributes to schizophrenia symptoms, agents that augment NMDA receptor function might be useful treatments. Such reasoning led to clinical trials involving the use of glycine in the treatment of schizophrenia. As discussed in Chapter 5, glycine promotes NMDA receptor function by binding to a positive allosteric regulatory site on the receptor complex. D-Cycloserine, which is a partial agonist at this site and may lead to the in vivo formation of glycine, has been administered as an adjunct to antipsychotic drugs in clinical trials. The outcomes of these trials have been mixed, some positive and others negative, but overall the effects appear to be modest. Based on binding properties it is not clear that trials of D-cycloserine fully test the hypothesis that the positive allosteric regulatory site might be a valid target for schizophrenia treatment. There remains interest in several alternative approaches, such as: (1) the use of D-serine—thought to be an endogenous agonist at the glycine site; (2) the activation of serine racemase that generates D-serine from L-serine; or (3) the inhibition of the plasma membrane glycine transporter-1 (GlyT1) that might increase levels of glycine at excitatory synapses (Chapter 5). However, none of these approaches has yet been validated in clinical studies.

Partial agonists of the type II metabotropic glutamate receptor subtypes, mGlu2 and mGlu3, have been studied as possible treatments for schizophrenia. These are presynaptic G protein–coupled autoreceptors (Chapter 5) that, when stimulated, decrease glutamate release 17–10. The logic of using these drugs for schizophrenia was based on the observation that the psychotomimetic drugs ketamine and phencyclidine, which act as postsynaptic NMDA receptor channel blockers, lead to compensatory excessive presynaptic glutamate release, which could then overstimulate AMPA glutamate receptors that would be unblocked (see 15–10 in Chapter 15). A nonselective mGlu2/3 agonist (pomaglumetad or LY2140023) was successful in decreasing behavioral abnormalities in putative animal models of schizophrenia, but failed in human clinical trials. Given uncertainties as to whether LY2140023 engaged its molecular targets (mGlu2/3 receptors), the hypothesis that agonists, partial agonists, or positive allosteric modulators (PAMs) of these receptors might have therapeutic utility in treating psychosis remains to be definitively tested. Interest in the hypothesis has been rekindled by the emergence of GRM3, the gene encoding mGlu3, as genome-wide significant in schizophrenia. That said, there is well-justified disillusionment, as outlined earlier, with animal models of schizophrenia and animal-based assays meant to predict truly novel antipsychotic drugs in humans. Decades of failure to identify new mechanisms of action in schizophrenia (and indeed all psychiatric disorders) have raised the important question of whether, where ethically possible, compounds shown to be safe in animals would be better tested for efficacy in humans than in animals.

Treatments are also under development with the aim of enhancing cognition in schizophrenia, with many similar drugs also under development for the palliation of memory loss in neurodegenerative disorders such as Alzheimer disease (Chapter 18). Much is known about the pharmacology of attention, working memory, and declarative memory (Chapter 14). However, clinical trials to date have been disappointing. One approach is the development of partial agonists of the α₇ nicotinic acetylcholine receptor (nAChR7) or PAMs of certain muscarinic cholinergic receptors, all of which are enriched in memory circuits in the brain. Interest in nicotinic agonists is driven by observations that cigarette smoking is extremely common in people with schizophrenia, suggesting the hypothesis that nicotine might ameliorate some of the symptoms experienced by these patients. Other compounds in development target the cyclic AMP pathway, which has long been implicated in memory function. These include inhibitors of certain subtypes of phosphodiesterases (PDEs), the enzymes that metabolize cyclic AMP (Chapter 4). Inhibitors of PDE4 have failed in clinical trials because they cause nausea; thus, efficacy has not been well tested. Trials are now under way for inhibitors of PDE10, which is highly enriched in striatum. Nausea has also made it difficult to develop D₁ dopamine receptor agonists, which are known from animal models to promote attention, working memory, and memory consolidation via actions in prefrontal cortex and hippocampus. In addition to pharmacology, there is much promise in early studies of cognitive psychotherapies both for prodromal schizophrenia and for the cognitive impairments of the full-blown disorder.

Overall, the first- and second-generation antipsychotic drugs remain the mainstays of treatment for the psychotic symptoms of schizophrenia, bipolar disorder, and psychotic depression, and for the agitation and psychosis seen in some patients with neurodegenerative disorders (Chapter 18). There remains significant unmet medical need for less toxic and more efficacious antipsychotic drugs and especially for treatments of the disabling negative and cognitive symptoms of schizophrenia that are untreated by medications available today.

The lightest solid element in the periodic table, lithium, circulates in blood as a monovalent cation. It is highly effective, at a concentration of 0.5 to 1.2 mM, in the treatment of acute mania and as a mood stabilizer, decreasing the risk of recurrence of manic and depressive episodes in individuals with bipolar disorder. As with antidepressants and antipsychotic drugs, the beneficial clinical effects of Li⁺ emerge after weeks of treatment.

Li⁺ has many protein targets, which has made it difficult to identify its mechanism of action with certainty. At therapeutic blood levels, Li⁺ affects: (1) the coupling of some neurotransmitter receptors to G proteins, (2) the expression of G_αi and subtypes of adenylyl cyclase, (3) the modification of G proteins by ADP-ribosylation, (4) cAMP-dependent and Ca²⁺-dependent protein phosphorylation, (5) GSK3β, and (6) the phosphatidylinositol second messenger pathway in the brain. Some of these effects are brain region specific. Which, if any, of these actions mediates Li⁺’s antimanic and mood-stabilizing properties remains unknown.

Inositol depletion hypothesis

Over two decades ago, Li⁺ was found to inhibit several inositol phosphatases in the phosphatidylinositol signaling cascade (see 4–9 in Chapter 4), which led to the inositol depletion hypothesis. Many neurotransmitter receptors, including α₁-adrenergic, 5HT_2A,2C-serotonergic, and M_1,3,5 muscarinic cholinergic receptors, are linked to the enzyme phospholipase C by means of the G protein G_q. Phospholipase C hydrolyzes phosphatidylinositol bisphosphate (PIP₂), a plasma membrane phospholipid, to yield two second messengers: diacylglycerol and inositol triphosphate (IP₃). Diacylglycerol activates protein kinase C, and IP₃ binds its receptor on the endoplasmic reticulum to release intracellular Ca²⁺ (Chapter 4).

Phosphatidylinositol is synthesized from free inositol and a lipid moiety. Most cells can obtain free inositol directly from plasma, but neurons cannot because inositol does not cross the blood–brain barrier. Consequently, neurons must either recycle inositol by dephosphorylating inositol phosphates after they are generated from the hydrolysis of phosphatidylinositols or synthesize it de novo from glucose-6-phosphate, a product of glycolysis. At therapeutic concentrations, Li⁺ inhibits several inositol phosphatases, most significantly inositol monophosphatase (IMPase). Such inhibition blocks the ability of neurons to generate free inositol from recycled inositol phosphates or glucose-6-phosphate. Thus, Li⁺-exposed neurons have a diminished ability to resynthesize PIP₂ after it is hydrolyzed in response to neurotransmitter receptor activation. Accordingly, it has been hypothesized that when firing rates of neurons are abnormally high, Li⁺-treated neurons are depleted of PIP₂, and neurotransmission dependent on this second messenger system is dampened.

This hypothesis is intriguing because it suggests that the effects of Li⁺ may be evident only in cells with abnormally high firing rates and that Li⁺ may be capable of treating both manic and depressive states because of its effects on multiple neurotransmitter systems. However, the story is more complex. Chronic inhibition of inositol phosphatases may lead to a buildup of active inositol phosphates, including IP₃, and thus may facilitate rather than dampen the actions of neurotransmitters on this pathway. Indeed, whether long-term Li⁺ administration dampens or facilitates phosphatidylinositol-dependent signal transduction in the brain is not clear. Moreover, the critical cells in the brain that are targets of Li⁺’s therapeutic action remain unknown, and it is not known which of the many phosphatidylinositol-dependent neurotransmitter systems must be dampened (or facilitated) to produce Li⁺’s clinical effects. More importantly, this hypothesis does not explain why several weeks must elapse before Li⁺ exhibits a therapeutic effect.

Regulation of the Wnt pathway and glycogen synthase kinase-3β

Currently, the leading hypothesis of Li⁺ action in bipolar disorder is inhibition of GSK3β 17–4. In addition to its therapeutic properties in bipolar disorder, Li⁺ has teratogenic effects in several animal models, for example, embryos of Xenopus laevis, an African clawed toad. Li⁺ causes abnormal dorsalization of the embryo (ie, the production of two spines). This effect was definitively traced to inhibition of GSK3β, whereas highly selective blockers of IMPase do not produce teratogenic effects in Xenopus. Moreover, Li⁺ is capable of altering cell fate in mutants of the yeast Dictyostelium discoideum that lack phospholipase C and thus cannot generate IP₃. As well, certain other mood-stabilizing drugs also inhibit GSK3β, among many other actions (see below). However, while preclinical experiments have implicated GSK3β as Li⁺’s therapeutic target, the lack of a convincing animal model of bipolar disorder has made it difficult to establish this hypothesis with confidence. What is needed is a selective inhibitor of GSK3β that is safe to administer to bipolar patients and to test its clinical efficacy. Perhaps then this long-standing question of the target of Li⁺ will be answered, opening avenues to new therapeutic approaches.

Other drugs for bipolar disorder

Several drugs initially developed as anticonvulsants (Chapter 19) are widely used in place of Li⁺ or as adjuncts to Li⁺ in the treatment of bipolar disorder. Valproate is effective at ameliorating the symptoms of manic episodes and reducing the likelihood of recurrent episodes, and is better tolerated by some patients than Li⁺. As discussed in Chapter 19, valproate, like Li⁺, has numerous actions, making it difficult to identify which action is primarily responsible for its mood-stabilizing properties. Thus, valproate is a use-dependent Na⁺ channel blocker, a Ca²⁺ channel blocker, and increases synaptic levels of GABA via unknown mechanisms. Valproate is also a nonselective inhibitor of histone deacetylases (Chapter 4). There is no evidence to connect any of these specific actions of valproate to its therapeutic effects in bipolar disorder. Several other anticonvulsants have been used to treat bipolar disorder, most notably carbamazepine, but the number of well-designed and convincing clinical trials remains small.