A molecular model for bipolar affective disorder

A molecular model for bipolar affective disorder

Medical Hypotheses Medical Hypotheses (1995) 45, 255-264 © PearsonProfessionalLtd 1995 A Molecular Model For Bipolar Affective Disorder H. M. LACHMA...

927KB Sizes 0 Downloads 53 Views

Medical Hypotheses Medical Hypotheses (1995) 45, 255-264

© PearsonProfessionalLtd 1995

A Molecular Model For Bipolar Affective Disorder H. M. LACHMAN* and D. F. PAPOLOS* *Department of Psychiatry and tDepartment of Medicine, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, New York 10461, USA. Tel: (+001) 718-430-2428; Fax: (+001) 718-430-8772 (Reprint requests to HML)

Abstract 1 The biological basis of bipolar disorder is not known. Models for the illness have been proposed that were based on the neurobiological effects of pharmacological agents that affect mood. Although of great interest, these models have not adequately explained the striking clinical pattern of illness in which patients may experience either unipolar episodes or bipolar cycles of mania and depression. We now present a new model suggesting that the unique clinical heterogeneity found in patients with bipolar disorder could be explained by a defect in a 'downstream' portion of a signal transduction pathway that can regulate two or more neurotransmitter systems that have opposite effects on neuronal activity. This model may target specific candidate genes for involvement in bipolar disorder.

Introduction Manic--depressive illness or bipolar disorder (BPD) is characterized by recurrent oscillations in mood, anxiety and energy states as well as disturbances in the sleep-wake cycle. It is a common illness affecting approximately 1% of the population (1,2). One of the distinguishing features of BPD is the tendency to remit and recur spontaneously with an increasing frequency over time. In the majority of cases, the disease-free interval between successive mood states is extremely variable, ranging from weeks to years. During the manic phase of the illness, patients are extremely energetic, grandiose, elated and irritable, and have a reduced need for sleep. The impulsiveness and poor judgement associated with untreated hypomania and mania commonly result in inappropriate actions that adversely affect work and family life and can lead to unemployment, legal difficulties and divorce. The depressive episodes are characterized by

feelings of sadness or irritability, emptiness, despair, and an inability to experience joy (anhedonia). Perhaps not surprisingly, 15-20% of individuals with BPD commit suicide (3). Although adequate treatment is available in the form of lithium salts and anticonvulsants such as sodium valproate and carbamazepine, approximately one third of patients do not respond or experience unacceptable side-effects. Many patients fail to seek medical attention or are misdiagnosed. A better understanding of the biological basis of the illness is essential. A number of twin, family and adoption studies indicate that genetic factors underlie BP vulnerability (4-8). However, despite years of investigation, the precise mode of inheritance has not been determined. Although in some families, single gene transmission has been suggested, a simple mendelian pattern is not usually observed (6,9,10). Consistent with a twogene concept is the finding that monozygotic twins generally show a 4-fold higher concordance compared

Date received 8 M a r c h 1995 Date accepted 5 April 1995

255

256 with that found in dizygotic twins (2). For a single major locus, one would only expect a 2-fold difference. Understanding the precise mode of inheritance is an important consideration in inherited disorders that are being mapped by genetic linkage analysis. In linkage analysis, one attempts to identify the chromosomal position of disease causing alleles by virtue of their co-inheritance with markers known to be located on a particular chromosomal segment (11, 12). Thereafter, the gene of interest can often be identified following additional mapping and extensive cloning, as shown in the identification of the genes for cystic fibrosis and Huntington's disease (13-15). Initially, studies by Egeland et al and Baron et al provided strong support for linkage of BPD to chromosome 11 in an Old Order Amish pedigree, and chromosome X in several Israeli pedigrees, respectively (16,17). However, these data have never been replicated in independent sets of families and a reanalysis of the original studies has resulted in diminished evidence for linkage to chromosomes 11 and X (18-20). Two recent reports suggesting linkage to genes on chromosomes 18 and 21 in some BP families have appeared, but these await replication (21,22). A number of factors may have played a role in the difficulties encountered so far in the BP linkage studies. These include diagnostic uncertainties, the absence of mendelian inheritance and genetic heterogeneity, the latter of which can lead to the false rejection of linkage (23). It has been suggested that other methods of statistical analysis employing model-free, nonparametric techniques, such as the sib pair and affected pedigree member methods, may be more appropriate for complex traits (24-26). However, nonparametric methods are statistically not as powerful as a classical linkage study that applies the correct genetic model. Because of the difficulties encountered so far with linkage analysis, some investigators have turned to a candidate-gene approach. The main advantage of using candidate genes is that the expected low frequency of recombination between a candidate-gene polymorphism and a disease causing mutation would increase the probability of establishing linkage. The primary disadvantage is that a relatively small percentage of human genes have been fully characterized so far. Furthermore, the complexity of BPD and the absence of an acceptable model make it difficult to choose feasible candidates for analysis. Earlier models for BPD

Some investigators have analyzed candidate genes in BPD based on the catecholamine hypothesis. Ac-

MEDICALHYPOTHESES cording to this model, mania is caused by an increase in central catecholamine neurotransmission, primarily dopaminergic, and depression by a decrease in noradrenergic neurotransmission (27,28). The model was based largely on pharmacological observations. For instance, many hypertensive patients treated with reserpine, which depletes central catecholamine stores, developed clinical depression, and drugs that increase catecholamine availability, like monoamine oxidase inhibitors, have antidepressant properties. The catecholamine hypothesis was supported in some studies which generally show an increase in catecholamine turnover occurring in mania and a decrease in depression (29). However, these are state-dependent phenomena so the relevance to the underlying basis of mood disorders is not clear. Also, some effective antidepressants do not appear to affect catecholamines. So far linkage to the 13-adrenergic and dopamine D1D4 receptor genes has been rejected (30-33). Another model, proposed by Janowsky and Dilsaver, is the cholinergic hypothesis (34,35). This model was based on the observed dysphoric and mood-altering effects of cholinesterase inhibitors and centrally acting cholinergic agonists. The model suggests that mania is due to a reduction in central cholinergic drive. Some investigators have found an enhancement of muscarinic cholinergic mediated responses by lithium thereby lending some support to the model (36). So far, there are no reports of linkage studies based on candidate genes derived from this model. More recent models of the illness have been based on findings related to the effects of lithium on signal transduction in neuronal cells. Lithium is a potent inhibitor of the enzyme myoinositol-l-phosphatase (M1P) which catalyzes the dephosphorylation of myoinositol-1-phosphate to inositol (37). Inositol is a precursor of phosphatidyl inositol 4,5-bisphosphate (PIPE), which is the major substrate of phospholipase C (PLC), a membrane bound enzyme that is linked to a variety of growth factors and neurotransmitters. Ligand-binding leads to PLC activation which catalyzes the cleavage of PIP 2 into the second messengers inositol 1,4,5 trisphosphate (IP3), which mobilizes intracellular Ca +z stores, and diacylglycerol (DAG) which activates protein kinase C (PKC) (38,39). Inositol used for PIP 2 synthesis is derived from diet, biosynthesis and a resalvage pathway from IP 3 that requires M1P. The brain depends more heavily on the latter sources because the brain-blood barrier prevents efficient uptake of plasma inositol. Since lithium inhibits M1P, the brain may be more sensitive to a potential inositol-lowering effect, an idea referred to as the inositol-depletion hypothesis (38). Some support for this model has been obtained by demonstrating lithium induced PIP2 and inositol depletion,

A MOLECULAR MODEL FOR BIPOLAR AFFECTIVE DISORDER

but this is not a consistent finding (40). Also, more recent studies have shown that lithium's effect on PLC-coupled signal transduction is complex, since an enhancing effect on PKC mediated induction of fos gene expression has been described (36,41). Another view is that lithium's therapeutic effect is due to an inhibition of G-proteins (42). These are ubiquitous membrane-associated heterotrimeric proteins, composed of t~, [3 and y subunits, that regulate signal transduction (43-45). Each subunit gene is a component of a multigene family (46,47). Upon ligand activation, the G-protein dissociates into the tx-subunit, which exchanges a molecule of GTP for GDP, and a complex made up of [3 and 7-subunits. The t~-subunit is the primary functional moiety and different subtypes are capable of regulating a number of second messenger and effector systems including PLC, adenylate cyclase, phospholipase A, and ion channels (43-45). Its action on effector targets is terminated by the induction of an intrinsic GTPase activity that leads to GTP hydrolysis, and ultimately to [3"t-subunit reassociation and receptor coupling (48). Although the ~lycomplex was initially viewed as a negative regulator of the o~-subunit, it has been found to have a direct effect on some signal transduction targets (48-50). Studies that have measured lithium's effect on Gproteins have focused primarily on its inhibition of agonist-stimulated GTP binding in isolated membrane preparations, which occurs at therapeutic concentrations (42,51). However, in studies using intact cells, lithium did not inhibit G-protein coupled stimulation offos gene expression (36,40).

A n e w m o d e l for B P D

As a result of the reported inhibitory effects of lithium on signal transduction, we and others previously proposed that BPD could be due to a primary defect in a brain-signal transduction pathway that leads to overactivation of downstream effector targets (38,42, 52). Our original model provided a plausible explanation for the cycling phenomenon and the swings in mood and energy that characterize BPD, but failed to adequately address phenotypic heterogeneity, an important feature of BPD. Although the classic cycling pattern characterized by a switch from one mood extreme to the other is eventually experienced by the majority of patients, episodes of mania and depression can occur as isolated phenomena and can be interspersed by long periods of euthymia (2). One course of illness followed by many patients is to experience several episodes of major depression prior to the development of

257 classic bipolar cycling. These findings suggest that the inheritance of the same set of vulnerability alleles can lead to the development of either isolated episodes of unipolar depression or mania, as well as classic bipolar cycles. This view is consistent with a number of observations. First, severe unipolar symptoms are observed in the postpartum period in approximately 10-15% of women with underlying BPD (53). Second, in some identical twin pairs concordant for psychiatric illness, one may have BP disorder while the other has unipolar symptoms. Third, in some pedigrees a parent with unipolar illness has children who develop BPD. Fourth, recurrent major depression, without mania or hypomania, is commonly found in individuals whose first degree relatives have BPD. An altemative explanation for this last observation is that several genes are segregating within bipolar families and individuals with recurrent depression have inherited fewer abnormal alleles than their more seriously affected relatives. However, the existence of a latent BP vulnerability in these individuals is supported by the emergence of mania or a rapid cycling phenomenon in approximately 10-15% of patients with recurrent depression who are treated with tricyclic antidepressants (54,55). Models for BPD that focus on individual neurotransmitter systems do not adequately explain these phenomena. For example, assume that mania is caused by a sustained increase in a particular neurotransmitter or neuropeptide that results in an increase in postsynaptic neuronal activity. Following this increase, a compensatory downregulation of postsynaptic receptor function might ensue, leading to a diminution in responsiveness, and consequently to depression. This simple model is not very consistent with clinical observations. It is difficult to see how depression could occur as an isolated phenomenon, in the absence of mania, since decreased postsynaptic function (depression) results from the downregulation that is initiated by a surge of neurotransmitter release (mania). Similarly, a defect in a specific neurotransmitter or neuropeptide receptor that leads to increased or decreased stimulation of intracellular signal transduction pathways is also not very consistent with the clinical pattern of illness since these defects would be expected to stimulate a fixed pattern of cellular responses. For example, a gain of function mutation in a receptor that is coupled to an increase in neuronal conduction would be expected to lead to a unipolar episode, such as mania, or to a cycle of mania followed by depression (increased conduction followed by negative feedback), but not to unipolar depression. A cellular abnormality that does appear to be more consistent with a clinical spectrum that includes unipolar and bipolar symptoms is a defect in a 'down-

258

MEDICALHYPOTHESES

stream' portion of a signal transduction pathway that can interact with two or more signalling systems in a single neuron. If these signalling pathways have opposite effects on neuronal activity, a defect in a regulatory protein that interacts with both pathways could conceivably lead to either an increase or a decrease in conduction within the same neuron.

Phosphatases One illustrative example of this type o f 'downstream regulator' is a brain-expressed phosphatase that can dephosphorylate proteins activated by more than one signal transduction pathway. Calcineurin is an example since phosphoproteins that have been induced by cAMP, PKC and Ca*2/calmodulin have been found to be substrates for this phosphatase (56-59). Consider a neuronal pathway in the limbic system that is regulated by different neuromodulators which exert either a positive or negative effect on neuronal activity. Assume that these effects are mediated by signal transduction pathways that affect protein phosphorylation and that the same phosphatase can deactivate some of these proteins (Fig. 1, left). With a defective

phosphatase, intense exposure to 'positively' acting neuromodulators would read to over-phosphorylation of target proteins that would cause a sustained increase in neuronal activity, leading to one mood extreme, mania for example, whereas activation of 'negatively' acting neuromodulators would lead to overphosphorylation of other proteins that have the opposite effect on neuronal activity, resulting in the opposite mood extreme, depression. The specific mood

experienced would depend on which neurotransmitter or neuropeptide receptor system is being stimulated, presumably by a particular stressor (Fig. 1, right). The term 'stressor' incorporates any behavioural, endocrine, pharmacological or environmental factor that can precipitate a mood disturbance, such as seasonal changes, REM sleep deprivation, the postpartum period and major life events. Euthymia could be explained in this model by assuming that under basal conditions, the phosphatase activity is sufficient to handle a subthreshold number of substrate molecules thereby preventing overphosphorylation from occurring. One could also explain subsyndromal episodes and persistent temperament disturbance that are considered part of the spectrum of mood disorder, such as dysthymia

A

l

"["

..~/.

B

C

Transducers & Effectors

Neuronal Activ~y

--c~increased ,S~gnal Transduction & Eflector Activation

==

=~

=[aim

Depro~don

Mania

Bipolar

Neuronal Activity

Cycle

Fig. 1 (Left set) Downstream regulator (phosphatase) affecting different signal transduction pathways. Signal transduction initiated by different ligands leads to phosphorylation of proteins that have opposite effects on neuronal activity (closed triangle and square). One pathway (closed circle) can cause an increase in conduction as well as initiate feedback downregulation by either, 1) downregulation of receptor that stimulated the pathway, 2) downregulation of other positively acting receptors, 3) signal transduction cross talk. Critical phosphoproteins (solid and striped rectangular symbols) within each of these pathways are substrates for the same phosphatase and have different effects on neuronal activity (-, +). Additional components of signal transduction and effector pathways are depicted by arrows. (Right set) Induction of mood disturbance. Stressors (A, B or C) increase ligand activation of receptors (bold receptors and upper arrows) resulting in over-stimulationof different signal transduction pathways. A defective phosphatase, depicted by loss of substrate binding site, leads to increase in functional half-life of critical phosphoproteinsthat prolongs respective effects on neuronal activity (bold lower arrows). Specific mood disturbance depends on which receptor system is being stimulated.

259

A MOLECULAR MODEL FOR BIPOLAR AFFECTIVE DISORDER

and cyclothymia, by assuming that intermediate levels of activation cause an increase in phosphoproteins that is sufficient to minimally affect mood but not enough to cause a major mood disturbance. This model illustrates how either manic or depressive episodes could occur independently. What about a bipolar cycle? One plausible hypothesis is that one of the pathways regulated by the phosphatase is capable of both stimulating and inhibiting neuronal conduction. This could be accomplished by a second messenger system that activates effector targets that exert a positive effect on conduction, and also initiates negative feedback inhibition. An example is cAMP which may initially enhance neuronal responsiveness by phosphorylating proteins, through the effects of cAMP dependent kinase, but which may also activate receptor specific kinases, such as [3ARK1 and [3ARK2, that inhibit further incoming signals by phosphorylating the [3-adrenergic and other receptors (60,61). An alternative explanation is that the pathway stimulated by the stressor that causes cycling may activate proteins that have an enhancing effect on conduction through one effector arm, but inhibits another effector system through signal transduction 'cross talk'. This model is compatible with the putative inhibitory effect of lithium on G-protein and/or PLC mediated signal transduction. Lithium is used in the treatment of acute mania but is less effective as an antidepressant. It also provides effective prophylaxis in the prevention of bipolar mood swings. This could be explained by differences in the sensitivities to therapeutic concentrations of lithium of the different G-protein or PLG-coupled pathways that we postulate are activated in mania, depression and cycling. In other words, the signal transduction pathway activated in the induction of mania may be more susceptible to the inhibitory effect of lithium on G-proteins or PLC than the one activated in depression. The delay in therapeutic response observed during lithium treatment could be due to the presence of biologically active phosphoproteins that have a long half-life because of a phosphatase defect. Consequently, lithium would block incoming signals but would not influence effector molecules that were activated prior to treatment.

is the component that exhibits the greatest functional diversity. If this is correct, it would suggest that different ~ subunits may share common pools of [3 and T subunits. Indeed, this has been demonstrated by some investigators (62). Additionally, the same 13-subunit could be shared by more than one T-subunit resulting in 13T complexes with different functional capacities (63). Therefore, it is conceivable that a defect in a single subunit gene could introduce a common abnormality into several different G-protein coupled receptor systems by disrupting the function of either ct-subunits or ~T-complexes, or both. Either a gain or loss of function defect could conceivably disrupt pathways that have opposite effects on neuronal activity. However, for simplicity, a gain of function abnormality in a 13-subunit that leads to an increase in the functional half life of several different t~-subunits and/or different [3T complexes is shown in Figure 2. If these abnormal heterotrimeric G-proteins are coupled to receptors and effector targets that have opposite effects on neuronal activity, then opposite mood extremes could be envisioned depending on which pathway is being stimulated by a particular stressor. Bipolar mood swings cycles could occur if the abnormal G-protein exerts one effect on conduction by stimulating downstream effector targets, as well as the opposite effect by activating negative feedback pathways (receptor downregulation or cross talk inhibition). A defective subunit that only affects the function of a single heterotrimeric G-protein would not be a feasible candidate in our model since a restricted effect on neuronal activity would occur leading to a more homogeneous set of symptoms.

Transcription factors

The final example, that will be mentioned, of a distinctive downstream regulator that may influence oppositely acting neuromodulators is a transcription factor. It is not within the scope of this paper to review the control of gene expression since other useful reviews exist (64-66). Rather, a few general principles will be considered in the context of this model for BPD. A number of transcription factors are induced in the central nervous system by neuromodulators. These include the so-called immediate early response G-proteins genes like the leucine zipper transcription factors cG-proteins are other signal transducers that have fos and c-jun. These are constituents of multigene the capacity to interact with more than one receptor gene families that include fral, fra2, junB and junD, system, in particular the ~ and T subunits. So far, more that are rapidly induced by depolarizing current and subunit genes have been identified than [3 and T by ligands that activate PKC, cAMP and other second genes (45). Although this could reflect the greater em- messenger systems (64-68). Post-translational alteraphasis that has been placed on the ~-subunits, it is tions can also activate transcription factors as seen in likely that more c~ genes exist in the genome since this the serine phosphorylation of c-jun protein.

260

MEDICAL HYPOTHESES

A

Stressor

B

C eel

Ligand

• •

lib

Receptor

I-y']

P-I*-

Inactive I ~, I G-pmteinslGDP I [~A I

Io ,l

I

Active (z and/or

subunits

-

+

Depression

Mania

+/-

,uror,a, Activity

Bipolar Cycle

Fig. 2 Abnormal G-protein. A single abnormal [~-subunitis a common component of three different heterotrimeric G-proteins that are made up of different ct and y subunits. Each G-protein exerts a different effect on neuronal activity. As a result of the defect in the common subunit, ¢t and/or ~ subunits exert an abnormal effect on downstream effector targets. Stressor A induces a pathway that leads to inhibition of neuronal activity whereas stressor B causes an increase. Stressor C activates a pathway that causes an increase in activity followed by a decrease as a result of negative feedback by mechanisms shown in Figure 1. The subscripts shown with the ¢t and y subunits are arbitrary designations.

The leucine zipper transcription factors form dimers that bind to specific D N A sequences located in regulatory regions usually located immediately upstream of the transcription start site. The fos and jun-like proteins dimerize to form AP-1 transcription factors that bind to a common D N A element. Combinations

of AP-1 proteins that share common components can generate complexes with opposite effects on transcription. For example, Fos/Jun and Fra2/Jun both bind to AP-1 recognition sequences but may induce opposite effects on gene expression (69). The brain also expresses intracellular receptors for steroid hormones that bind to specific DNA sequences following ligand activation (70). These include the receptors for glucocorticoids (Types I and II), thyroid hormone and sex steroids. Some of these may be particularly interesting in the context of mood disorders since glucocorticoids and thyroxine have significant

mood altering effects (71,72), and abnormalities in HPA-axis regulation and cortisol release are consistently found in depression (73). Furthermore, the influence of sex hormones is suggested by the association of premenstrual, puerperal and menopausal periods with exacerbations of mood disorder (74). An interesting level of complexity is introduced by the interactions of different classes of transcription factors. For example, members of the AP-1 family may bind ligand activated glucocorticoid receptors thereby causing mutual inhibition of steroid and AP-1 induced gene expression (75,76). Fos and Jun can also interact with the cAMP responsive element binding protein (CREB) at the protein and DNA levels (77,78). Other transcription modulators do not directly bind D N A but rather influence gene expression by inhibiting transcription factors through protein-protein interaction. The yeast transcription inhibitors Ssn6

A MOLECULARMODELFORBIPOLARAFFECTIVEDISORDER

261

and tupl (79) and a human equivalent called tuple may be examples o f this type o f regulation (80). The downstream position o f transcription regulators in the signal transduction scheme and their capacity to both activate and inhibit gene expression is consistent with their consideration as potential candidate genes in BPD according to our model. A number of different scenarios are possible but one simple model is an abnormality in a single transcription modulator that attenuates the effects o f positive and negative regulators of transcription through protein-protein interaction. Consider a set o f genes

A)

Signal Transduction

Signal Transduction

~]~r~

[

B)

+

expressed in the limbic system that regulate the exp r e s s i o n o f proteins, such as neuropeptides, involved in the control o f sleep, appetite, libido and energy level, behaviours that are affected in mood disorder. In response to environmental factors that activate different signal transduction pathways, an appropriate increase or decrease in gene expression would occur depending on whether activation o f positive or negative transcription factors occurs (Fig. 3A). However, in response to a stressor on either o f these pathways, and a 'loss o f function' defect in a negative regulator of both transcription factors, either superinduction o f

l

Stressor

"

TranscriptionFactors

, Oooe xore sion I

~

I

or

Stressor

Increased Increased T Signal _ Signal ransauction ~ransauction

Transcription Factors

i

I.,.

i

I.

le°°x0r°ss'°°

t or Mania Depression

Fig. 3 Defective transcription regulator. (A) The expression of critical genes is regulated by positively and negatively acting transcription factors that are induced by different signal transduction pathways. The activity of both transcription factors is regulated by the same negative regulator. Appropriate increases or decreases in gene expression occur depending on which signal transduction system is being stimulated. The + and - rectangles depict cis-acting regulatory regions that bind to the transcription factors and the term 'gene expression' refers to the gene or genes that are regulated by them. 0B) The positive and negative transcriptionfactors are over-activated by different stressors. Combined with reduced activity of the common negative regulator, either a marked increase or a marked decrease in gene expression occurs leading to the emergence of opposite mood states.

262

gene expression, leading to one mood extreme, or a drastic reduction leading to the opposite mood, may occur (Fig. 3B). Bipolarity is more difficult to explain but could be due to the sequential activation of positive and negative transcription factors by certain stressors. Alternatively, a different spectrum of gene products may be activated by a 'bipolar stressor' that may themselves cause opposite changes in neuronal activity. A primary defect in gene expression in BPD is consistent with the mood-altering effects of glucocorticoids and sex steroids as well as the ability of lithium, antidepressants and electroconvulsant shock to significantly affectfos gene expression (36,40,41). Abnormal regulation of gene expression is also consistent with the long duration of the mood disturbance and the delay that occurs in the therapeutic response to lithium and antidepressants. It should be noted that altered gene expression could be an important contributor to the bipolar phenotype even in the absence of specific inherited abnormalities in transcription regulators. Primary abnormalities that occur further upstream, such as those described in the sections on G-proteins and phosphatases, could easily interfere with gene expression secondarily since transcription factors are effector targets of most second messenger systems that are activated in the soma.

Discussion Despite almost a decade of investigation, genetic linkage analysis has so far failed to unequivocally identify linked chromosomal markers in BPD. It is too early to evaluate the validity of the chromosomes 18 and 21 findings. The overall failure to establish linkage is due to the complexity of the illness at both the clinical and genetic levels. However, with improvements in the human genome map to a resolution of 2-3 centimorgans, there is renewed hope that linkage analysis can ultimately be used to identify markers (81). At this level of resolution, it is possible to establish linkage of genes that have modest effects on phenotype. However, to map a disease-causing gene at this level will require the use of approximately 1000 markers and on the order of 105 genotype determinations. This level of analysis was required to map loci that may be involved in the development of Type I diabetes (82). Although this can be accomplished using robotic work stations and fluorescein-based automated genotyping, the task is still quite daunting. Equally formidable is the work required to identify the disease causing allele after linkage has been established. Typically, linkage studies can initially map a gene of interest to approximately 5-10 centimorgans,

MEDICAL HYPOTHESES

a region that may contain several hundred genes. Once linkage is demonstrated, further mapping is accomplished by analyzing additional markers and meiotic recombinants in informative families until the allele can be localized to a relatively small chromosomal segment, usually less than 500 kilobases. Eventually the gene of interest may be identified through a variety of cloning strategies. It required several years of work by three large labs to clone the gene responsible for cystic fibrosis after linkage had been established to chromosome 7 and ten years to isolate the Huntington's disease gene following the linkage study that first identified the responsible locus on chromosome 4. From a genetic perspective, these are relatively simple diseases since they are single gene defects with virtually complete penetrance. For a complex, inherited disorder like BPD, confirmation of linkage and further mapping of a potential linked locus will be quite difficult to accomplish (83). These difficulties demonstrate that other approaches to the problem, such as a candidate gene analysis, may be helpful. As the number of genes identified in the human genome increases in the next few years, candidate gene analysis becomes more feasible. As more cDNAs are placed on the genetic map, a plausible model for the illness could be used to identify candidate genes for analysis at relatively early stages of the mapping process. The feasibility of analyzing specific candidate genes has been strengthened by new techniques for mutation screening. These include single strand conformation polymorphism analysis, heteroduplex analysis, and chemical cleavage (84, 85). Automated sequencing capability using PCR generated templates is also fairly rapid and is the most accurate method available for detecting allelic differences. The hypothesis that BPD is caused by a defect in a shared component of signalling pathways that have opposite effects on neuronal activity is compatible with the clinical heterogeneity experienced by patients with the illness. By applying the general principles of this model, a more selective candidate gene analysis may be possible.

Acknowledgements HML is supported by the Carmel Hill Fund and is a recipient of the Stanley Foundation Research Award from NAMI. DFP is a recipient of an NIMH Physician Scientist Career Development Award.

References 1. Nurnberger J I, Goldin L R, Gershon E S. In: Winokur G, Clayton P, eds. The Medical Basis of Psychiatry, 2nd edn. New York: W B Saunders, 1994: 459-492.

263

A MOLECULARMODEL FOR BIPOLAR AFFECTIVEDISORDER 2. Goodwin F K, Jamison J R. In: Manic Depressive Illness. New York: Oxford University Press, 1990. 3. Dilsaver S C, Chen Y W, Swann A C, Shoaib A M, Krajewski K J. Suicidality in patients with pure and depressive mania. Am J Psychiatry 1994; 151: 1312-1315. 4. Bertelsen A, Harvald B, Hauge M. A Danish twin study of manic depressive disorders. Br J Psychiatr 1977; 130:330-351. 5. Spence M A, Amali H, Sadovick A e t al. A single major locus is the best explanation for bipolar family data: result of complex segregation analysis. Psych Genet 1993; 3: 143. 6. Rice J P, Reich T, Andreason N C et al. The familial transmission of bipolar illness. Arch Gen Psychiatr 1987; 44: 441-447. 7. Sham P C, Morton N E, Rice J P. Segregation analysis of the NIMH collaborative study: family data on bipolar disorder. Psychiatr Genet 1992; 2: 175-184. 8. Nurnberger J I, Gershon E S. Genetics of affective disorders. In: Post R, Ballenger J, eds. Neurobiology of Mood Disorders. Baltimore: Williams and Wilkins, 1984: 76--101. 9. Tsuang M J, Bucher K D, Fleming J A, Faravone S V. Transmission of affective disorders: an application of segregation analysis to blind family study data. J Psychiatr Res 1985; 19: 23-29. 10. Bucher K D, Elston R C. The transmission of manic depressive illness. I. Theory, description of the model and summary of results. J Psychiatr Res 1981; 16: 53-63. 11. Botstein D, White R, Skolnick M, Davis R W. Construction of a genetic linkage map in man using restriction fragment length polymorphisms. Am J Hum Genet 1980; 32: 314-331. 12. Edwards A, Civitello A, Hammond H A, Caskey C T. DNA typing and genetic mapping with trimeric and tetrameric tandem repeats. Am J Hum Genet 1991; 49: 746-756. 13. Kerem B, Rommens J M, Buchanan J A e t al. Identification of the cystic fibrosis gene: genetic analysis. Science 1989; 245: 1073-1080. 14. Riordan J R, Rommens J M, Kerem B et al. Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA. Science 1989; 245: 1066-1073. 15. The Huntington's Collaborative Research Group. A novel gene containing a trinucleotide repeat that is expanded and unstable on Huntington's disease chromosomes. Cell 1993; 72: 971-983. 16. Egeland J A, Gerhard D S, Pauls D L e t al. Bipolar affective disorder linked to DNA markers on chromosome 11. Nature 1987; 325: 783-787. 17. Baron M, Risch N, Hamburger R et al. Genetic linkage between X-chromosome markers and bipolar affective illness. Nature 1987; 326: 289-292. 18. Kelsoe J R, Ginns E I, Egeland J A et al. Re-evaluation of the linkage relationship between chromosome l i p loci and the gene for bipolar affective disorder in the old order Amish. Nature 1989; 342: 238-243. 19. Berretini W H, Goldin L R, Gelemter, Gejman P V, Gershon E S, Detera-Wadleigh S. X-chromosome markers and manicdepressive illness: rejection of linkage of Xq28 in nine polar pedigrees. Arch Gen Psychiatr 1990; 47: 366-373. 20. Gershon E S. Marker genotyping errors in old data on X-linkage in bipolar illness. Biol Psychiatr 1991; 29: 721-729. 21. Berritini W H, Ferraro T N, Goldin L R et al. Chromosome 18 DNA markers and manic-depressive illness: evidence for a susceptibility gene. Proc Natl Acad Sci 1994; 91: 5918-5921. 22. Staub R E, Lehner T, Luo Y et al. A possible vulnerability locus for BP affective disorder on chromosome 21. Nature Genet 1994; 8: 291-296. 23. Merikangas K R, Spence M A, Kupfer D J. Linkage studies of bipolar disorder: methodologic and analytic issues. Arch Gen Psychiatr 1989; 46:1137-1141. 24. Weeks D E, Lange K. The affected pedigree member method of linkage analysis. Am J Hum Genet 1988; 42: 315-326. 25. Blackwelder W C, Elston R C. A comparison of sib-pair

26.

27.

28. 29. 30.

31.

32.

33.

34. 35. 36.

37.

38. 39.

40.

41.

42.

43. 44. 45. 46.

47.

linkage tests for disease susceptibility loci. Genet Epidemiol 1985; 2: 85-97. Risch N. Linkage strategies for genetically complex traits. The power affected relative pairs. Am J Hum Genet 1990; 46: 229-241. Schildkraut J J. The catecholamine hypothesis of affective disorder: a review of supporting evidence. Am J Psychiatr 1965; 143: 509-522. Bunney W E, Davis J M. Norepinephrine in depressive reactions. Arch Gen Psychiatr 1965; 13: 483-494. Potter W Z, Manji H K. Catecholamines in depression: an update. Clin Chem 1994; 40: 279-287. Holmes D, Byrnjolfsson J, Brett P e t al. No evidence for a susceptibility locus predisposing to manic depession in the region of the dopamine D2 receptor gene. Br J Psychiatr 1991; 158: 635-641. Mitchell P, Selbie L, Waters B e t al. Exclusion of close linkage of bipolar disorder to D1 and D2 receptor gene markers. J Affect Disord 1992; 25 1-12. De bruyn A, Mendelbaum K, Sandkuijl L A e t al. Nonlinkage of bipolar illness to tyrosine hydroxylase, tyrosinase, and D2 and D4 dopamine receptor genes on chromosome 11. Am J Psychiatr 1994; 151: 102-106. Michell P, Waters B, Morrison N, Shine J, Donald J, Eisman J. Close linkage of bipolar disorder to chromosome 11 markers is excluded in two large Australian pedigrees. J Affect Disord 1991; 21: 23-32. Janowsky D S, E1-Yousef K, Davis J M, Sekerke H J. Parasympathetic suppression of manic symptoms by physostigmine. Arch Gen Psychiatr 1973; 28: 542-547. Dilsaver S C, Coffman J A. Cholinergic hypothesis: a reappraisal. J Clin Psychopharmacol 1989; 9: 173-179. Kalasapudi V, Sheftel G, Divish M, Papolos D, Lachman H. Lithium augments fos gene expression in PC12 pheochromocytoma cells: implications for therapeutic efficacy in mood disorders. Brain Res 1990; 521: 47-56. Hallcher L, Sherman W R. The effects of lithium ion and other agents on the activity of myoinositol-l-phosphatase from bovine brain. J Biol Chem 1980; 261: 8100-8130. Berridge M J. Inositol trisphosphate, calcium, lithium and cell signalling. JAMA 1989; 262: 1834-1841. Nishizuka Y. Intracellular signaling by hydrolysis of phospholipids and activation of protein kinase C. Science 1992; 258: 607~514. Kennedy E D, Challiss R A J, Ragan I, Nahorski S R. Reduced inositol polyphosphate accumulation and inositol supply induced by lithium in stimulated cerebral cortex slices. Biochem J 1990; 267: 781-786. Divish M, Sheftel G, Kalasapudi V, Boyle A, Papolos D, Lachman H. Differential effect of lithium on PKC and cAMP activatedfos gene expression. J Neurosci Res 1991; 28: 40--48. Avissar S, Schreiber G, Danon A, Belmaker R H. Lithium inhibits adrenergic and cholinergic increases in GTP binding in rat cortex. Nature 1988; 331: 440-442. Gilman A G. G-proteins and regulation of adenylate cyclase. JAMA 1989; 262: 1819-1825. Birnbaumer L, Abramowitz J, Brown A M. Receptor-effector coupling by G proteins. Biochim Biophys Acta 1990; 1031: 163-224. Simon M I, Strathmann M P, Gautam N. Diversity of G proteins in signal transduction. Science 1991; 252: 802-808. Cali J J, Balcueva E A, Rybalkin I, Robishaw J D. Selective tissue distribution of G-protein ~/subunits including a new form of the 7 subunit identified by cDNA cloning. J Biol Chem 1992; 267: 24023-24027. Simon M I, Strathmann M P, Gautam N. Diversity of Gproteins in signal transduction. Science 1991; 252: 802-808.

264 48. Conklin B R, Bourne H R. The structural elements of G a subunits that interact with G[3"t, receptors and effectors. Cell 1993; 73: 631-641. 49. Moriarty T M, Gillo B, Carty B J, Premont R T, Landau E M, Iyengar R. Beta-gamma subunit of GTP binding proteins inhibits muscarinic receptor stimulation of phospholipase C. Proc Natl Acad Sci 1988; 85: 8865-8869. 50. Pitcher J A, Inglese J, Higgins J B e t al. Role of [3T subunits of G-proteins in targetting the [3-adrenergic receptor kinase to membrane bound receptors. Science 1992; 257: 1264-1267. 51. Avissar S, Schreiber G. The involvement of gnanine nucleotide binding proteins in the pathogenesis of affective disorder. Biol Psychiatry 1992; 31: 435-459. 52. Lachman H M, Papolos D F. Abnormal signal transduction: a hypothetical model for bipolar affective disorder. Life Sci 1989; 45: 1413-1426. 53. Reich T, Winokur G. Post partum psychosis in patients with manic depressive illness. J Nerv Ment Dis 1970; 151: 60458. 54. Kukopulus A, Caliari B, Tundo A e t al. Rapid cyclers, temperament and antidepressants. Compr Psychiatr 1983; 24: 249-258. 55. Dunner D L, Fleiss J L, Fieve R R. The cause of development of mania in patients with recurrent depression. Am J Psychiatr 1976; 133: 905-908. 56. Sim A. The regulation and function of protein phosphatases in the brain. Mol Neurobiol 1992; 5: 229-246. 57. Halpain S, Girault J A, Greenberg P. Activation of NMDA receptor induces dephosphorylation of DARPP32 in rat striatal slices. Nature 1990; 343: 369-372. 58. Liu Y, Storm D R. Dephosphorylation of neuromodulin by calcineurin. J Biol Chem 1991; 264: 12800-12804. 59. Walaas S I, Greengard P. Protein phosphorylation and neuronal function. Pharm Rev 1991; 43: 299-349. 60. Arriza J L, Dawson T M, Simerly R B e t al. The G-protein coupled receptor kinases ~ARK1 and ~IARK2 are widely distributed at synapses in rat brain. J Neurosci 1992; 12: 4045-4055. 61. Benovic J, Onorato J J, Arriza J L e t al. Cloning, expression and chromosomal localization of ~-adrenergic receptor kinase 2. A new member of the receptor kinase family. J Biol Chem 1991; 266: 14939-14946. 62. von Weizsacher E, Strathmann M P, Simon M I. Diversity among the [3 subuhits of heterotrimeric GTP-binding proteins: characterization of a novel ~-subunit cDNA. Biochem Biophys Res Comm 1992; 183: 350-356. 63. Tang W J, Gilman, A G. Type specific regulation of adenylyl cyclase by G protein beta gamma subunits. Science 1991; 252: 1500-1503. 64. Karin M. Signal transduction from the cell surface to the nucleus through the phosphorylation of transcription factors. Curt Opin Cell Biol 1994; 6: 415-424. 65. Abate C, Curran T. Encounters with Fos and Jun on the road to AP-1. Sem Cancer Biol 1990; 1: 19-26. 66. Edwards D R. Cell signalling and the control of gene expression. Trends Pharmacol Sci 1994; 15: 239-244.

MEDICAL HYPOTHESES 67. Morgan J I, Curran T. Role of ion flux in the control of c-fos expression. Nature 1986; 322: 552-555. 68. Pennypacker K R, Hong J S, McMillian M K. Pharmacological regulation of AP-1 transcription factor DNA binding activity. FASEB J 1994; 8: 475--478. 69. Suzuki T, Okuno H, Yoshida T, Endo T, Nishina H, Iba H. Difference in transcriptional regulatory function between c-Fos and Fra-2. Nuc Acids Res 1991; 19: 5537-5542. 70. Yamamato K R. Steroid receptor regulated transcription of specific genes and gene networks. Ann Rev Genet 1985; 19: 209-252. 71. Wolkowitz O M, Rubinow D, Doran A R et al. Prednisone effects on neurochemistry and behavior. Arch Gen Psychiatry 1990; 47: 963-968. 72. Bauer M S, Whybrow P C, Winokur A. Rapid cycling bipolar affective disorder. I. Association with grade I hypothyroidism. Arch Gen Psychiatr 1990; 47: 427--432. 73. Greenberg L, Edwards E, Henn F A. Dexamethasone suppression test in helpless rats. Biol Psychiatr 1989; 26: 530-532. 74. Pariser S F. Women and mood disorders. Menarche to menopause. Ann Clin Psychiatr 1993; 5: 249-254. 75. Yang-Yen H F, Chambard J C, Sun Y L e t al. Transcriptional interference between c-jun and the glucocorticoid receptor: mutual inhibition of DNA binding due to direct protein-protein interaction. Cell 1990; 62: 1205-1215. 76. Diamond M I, Miner J N, Yoshinaga S K, Yamamoto K R. Transcription factor interactions: selectors of positive or negative regulation from a single DNA element. Science 1990; 249: 1266-1272. 77. Cook S J, McCormick F. Inhibition by cAMP of Ras-dependent activation of Raf. Science 1993; 262: 1069-1072. 78. Mellon P L, Clegg C, Correll L A, McKnight G S. Regulation of transcription by cAMP-dependent protein kinase. Proc Natl Acad Sci 1989; 86: 488-491. 79. Keleher C A, Redd M J, Schultz J, Carlson M, Johnson A D. Ssn6/Tup 1 is a general repressor of transcription in yeast. Cell 1992; 68: 709-719. 80. Halford S, Wadey R, Roberts C et al. Isolation of a putative transcriptional regulator from the region of 22qll deleted in DiGeorge syndrome, Shprintzen syndrome and familial congenital heart disease. Hum Mol Genet 1993; 2: 2099-2107. 81. Gyapay DG, Morissette J, Vignal A e t al. The 1993-94 Genethon human genetic linkage map. Nature Genetics 1994; 7: 246. 82. Davies J L, Kawaguchi Y, Bennett S T et al. A genome wide search for human type 1 diabetes susceptibility genes. Nature 1994; 371: 130-136. 83. Suarez B K, Hampe C I, Van Eerdewegh P. In: Gershon E S, Cloninger C R, eds. New Genetic Approaches to Mental Disorders. Washington, DC: American Psychiatric Press, 1994: 23--46. 84. Orita M, Suzuki Y, Sekiya T, Hayashi K. Rapid and sensitive detection of point mutations and DNA polymorphisms using the polymerase chain reaction. Genomics 1989; 5: 874-879. 85. Cotton R G. Current methods of mutation detection. Mut Res 1993: 285: 125-144.