Evidence for a Role of Phospholipase C- 1 in the Pathogenesis of Bipolar

cated in the pathogenesis of BD or in the mechanism of ORIGINAL RESEARCH ARTICLE lithium response remains to be determined. Bipolar disorder (BD) is a major psychiatric con- Evidence for a role of dition characterized by episodes of mania and ␥ depression, affecting up to 1% of the general popu- phospholipase C- 1 in the lation.1 Lithium has been used in the prophylaxis and treatment of BD for almost half a century, and remains pathogenesis of bipolar the first-choice therapy for preventing recurrences.2 disorder Although lithium is considered specific for the treatment of BD, with no comparable effect in other psychi- G Turecki1, P Grof2, P Cavazzoni2, A Duffy2, atric disorders, its effectiveness varies widely. There is E Grof2, B Ahrens3, A Bergho¨fer3,BMu¨ller- compelling evidence that lithium is more effective in 3 4 4 forms of BD characterized by typical symptomatology Oerlinghausen , M Dvora´kova´ , E Libigerova´ , 3–5 M Vojtechovsky´4, P Zvolsky´4, R Joober1, and the absence of comorbidity. There is also evi- A Nilsson5, H Prochazka5, RW Licht6, dence that responders and nonresponders to lithium 6 6 6 treatment differ in certain neuroendocrine responses NA Rasmussen , M Schou , P Vestergaard , 6 7 7 7 involving the serotonergic and endorphin systems. In A Holzinger , C Schumann , K Thau , addition, family studies indicate a higher recurrence 1 2 GA Rouleau and M Alda risk for bipolar disorder among relatives of patients who respond well to lithium treatment.5,7,8 Taken 1Centre for Research in Neuroscience, The Montreal together, these findings suggest that response to lith- General Hospital, McGill University, Canada; 2Department ium prophylaxis may help define a distinct bipolar of Psychiatry, University of Ottawa, Canada; 3Department phenotype with less genetic heterogeneity. of Psychiatry, Free University, Berlin, Germany; The mechanism by which lithium acts is not exactly 4Department of Psychiatry, Charles University, Prague and known. Recent findings indicate that lithium may sta- Hradec Kra´love´, Czech Republic; 5Karsudden Hospital, bilize mood by acting at the phosphoinositide second Katrineholm, Sweden; 6Psychiatric Hospital, University of messenger system.9 Cellular responses mediated by Århus, Risskov, Denmark; 7University Clinic of Vienna, inositol phospholipids are involved in many brain pro- Department of Psychiatry, Austria cesses.10,11 They are initiated by a phospholipase C (PLC) isozyme after activation by a membrane receptor Keywords: bipolar disorder; phospholipase C; genetics; that can be coupled to a G protein, protein tyrosine lithium; signal transduction; linkage; association kinase or several lipid-derived second messengers such as arachidonic acid.12 Lithium is thought to inhibit the Several studies have indicated that patients with bipolar enzyme inositol monophosphatase, leading to a disorder (BD) who respond well to lithium prophylaxis reduction in the availability of inositol.9 constitute a biologically distinct subgroup. Lithium is thought to stablize mood by acting at the phosphoinosi- In this study we present preliminary evidence sug- tide cycle. We have investigated a polymorphism gesting that patients who have an excellent response located in the gene (PLCG1) that codes for a ␥-1 iso- to lithium prophylaxis have a higher frequency of a zyme of phospholipase (PLC), an enzyme that plays an polymorphism located in the gene coding for the ␥-1 important role in the phosphoinositide second messen- isozyme of phospholipase C (PLCG1) on chromosome ger system. A population-based association study and 20. This result was further explored in a family link- a family-based linkage study were carried out on age study. patients who were considered excellent responders to Patients with bipolar disorder were recruited from lithium prophylaxis. Response to lithium was evaluated six centers that collaborate in the International Group ± prospectively with an average follow-up of 14.4 6.8 for the Study of Lithium (IGSLI) (see methods). These years. The PLCG1 polymorphism was investigated in 136 excellent lithium responders and 163 controls. In patients have been followed in specialized lithium addition, the segregation of this marker was studied in clinics and their response to lithium prophylaxis has 32 families ascertained through lithium-responsive been assessed systematically and prospectively. In bipolar probands. The allele distributions between lith- order to be included in the study, all patients had to ium-responsive bipolar patients and controls were dif- meet stringent criteria of excellent lithium response, ferent, with a higher frequency of one of the PLCG1 described previously5 and summarized in Table 1. A polymorphisms in patients (␹2 = 8.09; empirical total of 136 patients were included in the association P = 0.033). This polymorphism, however, confers only a study. The mean (± standard deviation) age of onset = small risk (OR 1.88, CI 1.19–3.00). Linkage studies with was 27.6 (± 9.9) years. The number of the illness epi- the same marker yielded modest support for the sodes prior to lithium treatment was 8.2 (± 10.1). involvement of this gene in the pathogenesis of BD Patients have been stabilized on lithium monotherapy when unilineal families were considered (Max ± LOD = 1.45; empirical P = 0.004), but not in the whole for 14.4 ( 6.8) years. Control subjects for this study sample. Our results provide preliminary evidence that a were 128 psychiatrically unaffected individuals who PLC isozyme may confer susceptibility to bipolar dis- were collected in similar fashion by participating order, probably accounting for a fraction of the total centres, and consisted of healthy married-in individ- genetic variance. Whether this polymorphism is impli- uals from the linkage study, hospital staff and normal Phospholipase C-␥1 in the pathogenesis of BD G Turecki et al 535 Table 1 Criteria used to diagnose bipolar patients as excel- together (PLCG1/5 ± 1: ␹2 = 10.35; d.f. = 1, P = 0.007* lent lithium responders and PLCG1/5 ± 2: ␹2 = 7.59; d.f. = 1, P = 0.019*). Fur- thermore, patients and controls also differed when Each patient must meet criteria A, B and C. alleles were grouped according to size, with patients presenting an excess of long (CA) repeats (PLCG/1– A. Diagnosis of primary episodic bipolar disorder based n on the SADS-L (lifetime version) interview and PLCG/7), whereas controls have predominantly short ␹2 = = Research Diagnostic Criteria (RDC). (PLCG1/8–PLCG1/18) variants ( 8.30; d.f. 1, P = 0.007*) (see Figure 1). B. High Recurrence Risk Parametric linkage results using the PLCG1 marker either B1: five or more episodes prior to lithium gave non significant maximum lod scores when results treatment from all families were considered (see Table 3). Indi- or B2: four episodes prior to lithium; of these two or viduals were considered affected when they met RDC more during the 2 years preceding lithium criteria for bipolar disorder, schizoaffective disorder or treatment recurrent major disorder. Three major genetic models or B3: three episodes prior to lithium, plus one more were explored in order to maximize the evidence for with 12 months after lithium discontinuation linkage. These models were: (a) dominant (allele frequency (q) 0.012, male penetrance (fM) 0.4, female pen- C. Unequivocal Lithium Response etrance (fF) 0.7, and normal penetrance 0.005 for males C1: No recurrence requiring additional biological (fM0) and 0.009 for females (fF0)); (b) intermediate intervention (ECT, antidepressants, (q = 0.024, fM = 0.4/0.2, fF = 0.7/0.35, fM0 = 0.005, neuroleptics) during the entire observation fF0 = 0.009); (c) recessive (q = 0.16, fM = 0.35, fF = 0.65, time on lithium monotherapy fM0 = 0.005, fF0 = 0.009). As previous findings sug- and C2: Minimum period of observation of 3 years gested that the study of unilineal pedigrees may provide an advantage to overcome part of the complexity and C3: Average plasma lithium concentration over 0.6 13–14 mEq L−1 of BD, we further analyzed the data according to lineality. Among all pedigrees, 13 were clearly unilineal (six of paternal and seven of maternal origin). The largest lod score observed among unilineal families = volunteers. Thirty-five more control subjects were was 1.45 (P 0.004) under the dominant model. Simi- included without assessment of psychiatric status. All lar results in these families were found using a nar- rower diagnostic definition, which excluded recurrent cases and controls were of similar ethnic background = = and were, whenever possible, matched for geographi- major depression (Zmax 1.02, P 0.006). There was no cal origin. The mean age (± standard deviation) and sex difference between maternal and paternal pedigrees. ratio (M : F) were 50.0 (± 14.4) years, 0.87 for patients Nonparametric linkage analysis provided similar and 51.45 (± 14.8), 0.86 for controls. results (see Table 3). Thirty-two of the Canadian probands had families An increasing body of evidence supports the hypoth- available for linkage analysis. The family sample con- esis that alterations in the phosphoinositide signal transduction system may be implicated in the patho- sisted of 224 interviewed and genotyped individuals, 9,15 of whom 95 were affected. physiology of bipolar disorder. Studies of postmor- Allele distributions in individuals from different tem brains from patients with BD and suicide victims centers were similar, both for cases (␹2 = 11.95; with major depression have shown a marked reduction d.f. = 12, P Ͼ0.4), and controls (␹2 = 19.61; d.f. = 13, in phosphatidylinositol hydrolysis by G protein Ͼ coupled to phospholipase C stimulation, when com- P 0.1). This allowed pooling of the Canadian and Eur- 15,16 opean samples (patients as well as controls) for all pared to normal controls. Furthermore, recent mag- comparisons. netic resonance spectroscopy studies have indicated The distributions of alleles in patients and controls that lithium reduces myoinositol levels in critical brain regions in bipolar patients.17 The phosphoinositide- are shown in Table 2. A significant difference was ␥ observed in the overall allele frequency distribution specific PLC -1 isozyme is present in neurons; studies between patients and controls (␹2 = 23.22: d.f. = 13, in rats have found it highly localized in the hippocam- 11 ␥ P = 0.029*) with allele PLCG1/5 being considerably pus and basal ganglia. PLC -1 is activated by a num- more frequent among patients than controls (␹2 = 8.09; ber of receptors that promote phosphorylation of the d.f. = 1, P = 0.033* and OR = 1.88, CI 1.19–3.00). In enzyme, usually via protein tyrosine kinase, resulting order to take into account possible instability of the in the hydrolysis of phosphatidylinositol 4,5-biphos- phate. This produces two intracellular messengers: dinucleotide repeat, we examined whether the differ- 12 ence between groups remained within the bounds of diacylglycerol and inositol 1,4,5-triphosphate. It is alleles with ± 1 and ± 2 repeats. Indeed, patients dif- therefore possible that PLCG1 variants account for a fered from controls when these alleles were pooled proportion of the total genetic variability predisposing to BD. The finding of an association between PLCG1 and BD *Indicates empirical P values observed using Monte Carlo simul- in our study suggests that this polymorphism is in link- ations (see methods). age disequilibrium with the sequence/gene that confers Phospholipase C-␥1 in the pathogenesis of BD G Turecki et al 536 Table 2 Allele frequency distribution in lithium responder bipolar patients and controls

Allele frequency (%)

N PLC PLC PLC PLC PLC PLC PLC PLC PLC PLC PLC PLC PLC PLC G1/1 G1/2 G1/4 G1/5* G1/6 G1/7 G1/8 G1/10 G1/11 G1/12 G1/13 G1/16 G1/17 G1/18

Controls 163 0.6 0.9 5.0 12.6 6.3 5.0 4.4 51.9 0.6 1.3 6.0 0.3 4.1 0.9 Patients 136 1.5 0.4 7.4 21.3 7.4 4.0 2.9 43.0 0.7 0.0 2.9 1.1 5.1 2.2 Total 299 1.0 0.7 5.8 16.6 6.8 4.6 0.8 47.8 0.7 0.7 4.4 0.7 4.6 1.5

(*) indicates the associated allele.

However, there is increasing evidence that certain tandem repeat sequences, such as VNTRs do inﬂuence gene transcription, thereby inﬂuencing phenotype expression.18,19 In addition, comparative genomic studies have indicated that repeat arrays and genomic locations of dinucleotide repeats are highly conserved,

suggesting that (CA)n repeats may have a functional sig- niﬁcance.20 Their role has been hypothesized to be, among others, related to the regulation of gene transcription. Therefore, it is possible that the microsatellite marker tested in this study by itself confers certain susceptibility to BD. Based on the observation that

patients present longer (CA)n repeats than controls, one possible hypothesis would be that long alleles determine different gene transcription rates, predisposing individuals to BD. Finally, we cannot exclude the possibility that our Figure 1 Allele frequency distribution in patients (JbJ) finding is spurious. It is conceivable that our results ̅ and controls (J J). reflect problems such as admixture or population stratification effects, rather than a functional genetic association.21 Although possible, it is unlikely that our Table 3 Summary of parametric and nonparametric linkage findings result from admixture effect because samples results in the whole sample of 32 pedigrees and in unilineal were drawn from panmictic, non-isolated populations pedigrees only of similar ethnic origin. Moreover, when data from dif- Parametric Nonparametric ferent centers were compared, both stratified and not stratified for clinical status, no differences in allele fre- Z (␪)WZ P quency distributions were observed. max obs The linkage results obtained for unilineal families in Whole sample 66.84 0.13 the parametric analysis, as well as those observed for Dominant 0.01 (0.2) the nonparametric linkage analysis were not significant Intermediate 0.00 (0.5) given the standard criteria suggested for linkage stud- Recessive 0.00 (0.5) ies.22 However, these results are congruent with those obtained in the association study and warrant further Unilineal families 40.08 0.04 Dominant 1.45 (0.01)§ consideration. Defining thresholds of significance in Intermediate 1.20 (0.01) linkage studies of complex traits has been rather diffi- 22–25 Recessive 0.25 (0.05) cult. It is clear that in order to avoid false positive results, stringent criteria should be used.22 This may §P = 0.004. lead, however, to situations in which samples of ‘realistic’ size will not have sufficient power to detect loci that account for a small proportion of the total gen- susceptibility to BD. This interpretation is supported etic variability.23 Considering the genetic and pheno- by the observation that, compared to controls, patients typic complexity of BD, as well as the modest risk attri- have a higher frequency of alleles within the bounds buted to the PLCG1 locus, it is likely that our family of ± 2 repeats from the associated allele. Furthermore, sample was not large enough to achieve conventional microsatellite repeats are usually not transcribed, and significance levels. Hence, the results of the linkage are therefore commonly regarded as non-functional, study observed in unilineal families should be further unlikely to confer susceptibility to BD. explored as they may also be indicating that variation Phospholipase C-␥1 in the pathogenesis of BD G Turecki et al 537 of PLCG1 may play a role in the pathogenesis of lith- countries, all subjects were subsequently assessed by ium-responsive BD. the same senior clinician (PG). All patients were Cau- Further evidence supporting the involvement of this casians of Western and Central European descent. All locus in the etiology of BD will have to come through individuals included in the association and linkage independent replication. Consistency between studies studies provided written informed consent. is an important way to validate findings. In this con- For linkage analyses, two phenotypic schedules were text, it is important to note that the patient cohort used used. Under the first one, relatives were considered in our study was carefully selected according to a sys- affected if they met RDC criteria for bipolar disorder, tematic and prospective evaluation of lithium schizoaffective disorder or recurrent major depression response. This locus could therefore be related to the with the additional criterion of functional impairment responsiveness to lithium treatment rather than to during depressive episodes. Under the second pheno- bipolar disorder itself. However, the design of the typic schedule, only subjects with bipolar and schi- present study does not allow us to further explore this zoaffective disorders were considered affected while issue. Thus, the validity of future attempts at inde- the phenotype of unipolar depression was treated as pendent replication of these findings will accordingly unknown. Pedigrees were considered unilineal if the be contingent on lithium response being evaluated trait segregated exclusively through one of the proband using similar methodology and criteria. parent’s family. More specifically, the parent, this par- In conclusion, we have obtained preliminary evi- ent’s parents and siblings had to be unaffected. This is dence that a polymorphism at the locus that codes for consistent with published criteria for lineality assess- a PLC isozyme is associated with BD in patients with ment. an excellent response to lithium. Whether the locus that confers susceptibility to BD is the (CA)n, another PLCGI genotyping polymorphism/mutation in the PLCG1 gene or another Genomic DNA was extracted by a standard method29 neighbouring gene remains to be clarified. We are cur- from venous blood samples. The PLCG1 polymorphism rently conducting studies investigating this gene for was originally identified by searching a cosmid iso- additional polymorphisms which may be similarly lated using a clone containing the PLCG1 cDNA. This associated with this disorder. In addition, we are in the is an untranslated (CA)n repeat with 14 observed alleles process of collecting a group of lithium non-responders ranging in size from 150 to 184 bp.30 PCR was carried which may help to address important points related to out in a total volume of 12.5 ␮l containing 40 ng gen- the specificity of this finding. omic DNA; 125 ng of primers PLCpr1 (5′-AAC- CAGTCTGCTCTTCCGGTG-3′) and PLCpr2 (5′- CTGCCTTCAACTGATCTCAATGG-3′); 200 ␮M each of Methods dGTP, dCTP, and dTTP; 25 ␮M dATP; Patient ascertainment and diagnostic assessment 1.5 ␮Ci[35S]DATP; 0.5 units of Taq DNA polymerase Patients for the association study were recruited as fol- (Bio/Can Scientific, Toronto, Canada); and 2.0 ␮lof10× lows: 68 from Canada, 21 from Germany, 20 from the buffer (Bio/Can Scientific) with MgCl2 included in the Czech Republic, 17 from Sweden, seven from final concentration of 1.5 mM. Samples were over-laid Denmark, and three from Austria. Families for the link- with mineral oil and processed throughout 35 cycles age study were collected only among Canadian pro- of denaturation at 94°C, annealing at 56°C, and elong- bands. Thirty-two had positive family history and were ation at 72°C, followed by a final elongation period cooperative. Families were considered when they met of 72°C. PCR products were analyzed on a 6% dena- the following criteria: first, at least four family mem- turating polyacrylamide gel (38 : 2 acrylamide : bers (including the proband) must have been available bisacrylamide): Samples were run for a period of 2 h for interview and blood sample. Of these, at least two in a vertical electrophoresis gel apparatus (Life Tech- subjects must have been affected. Second, at least one nologies, Gaithersburg, USA). Gels were dried and of the proband’s parents must have been unaffected. exposed to X-ray films for 48–72 h at room tempera- Finally, only subjects older than 15 years were asked ture. All marker determinations were made blind to the to participate. The pedigrees were extended in such a clinical diagnoses. Autoradiographs were read and way as to include all first-degree relatives of all affected interpreted independently by two different readers, subjects. As the penetrance is presumably low, all first- and both readings yielded identical results. degree relatives of all subjects whose first-degree relatives were affected were included. The procedure for Statistical analysis an extended sampling of families for linkage analysis The presence of an association was investigated using was based on the rule proposed by Cannings and ␹2 and odds ratio tests by standard methods using the Thompson.26 BMDP statistical package.31 In addition, we analyzed All diagnoses were based on the Schedule for Affect- the data using the method proposed by Curtis and ive Disorders and Schizophrenia27 (SADS-L) interview Sham that estimates empirical P values using a Monte and Research Diagnostic Criteria28 (RDC). All diagnos- Carlo approach32 (implemented in the computer pro- tic and treatment information was reviewed by a panel gram CLUMP). This method is particularly suitable for of experienced psychiatrists in a blind fashion. To the analysis of tables with a large number of columns further ensure uniformity of diagnoses in six different and sparse cells, and therefore has large application in Phospholipase C-␥1 in the pathogenesis of BD G Turecki et al 538 association studies using highly polymorphic markers, 14 Stine OC, Xu J, Koskela R, McMahon FJ, Gschwend M, Friddle C such as this one. The advantage of this method is that et al. Evidence for linkage of bipolar disorder to chromosome 18 with a parent-of-origin effect. Am J Hum Genet 1995; 57: 1384– it provides a robust and useful alternative to the 1394. excessively conservative Bonferroni correction in cases 15 Jope RS, Song L, Li PP et al. The phosphoinositide signal transduc- where the observed contingency tables have a large tion system is impaired in bipolar affective disorder brain. J Neuro- number of columns.32 Parametric linkage analysis was chem 1996; 66: 2402–2409. carried out using the MLINK33 program of the 16 Pacheco MA, Stockmeier C, Meltzer HY, Overholser JC, Dilley GE, FASTLINK34 package. To determine the empirical level Jope RS. Alterations in phosphoinositide signaling and G-protein levels in depressed suicide brains. Brain Res 1996; 723: 37–45. of significance of the observed linkage results, simula- 17 Bebchuk J, Moore G, Manji H. Lithium regulation of brain myoinos- 35 tions were carried out using the SIMULATE and itol in bipolar affective disorder. Second International Conference SLINK36,37 programs. Nonparametric linkage analysis on Bipolar Disorder, Pittsburgh, 1997. was conducted using the SimIBD program which cal- 18 Bennett ST, Todd JA. Human type 1 diabetes and the insulin gene: culates a simulation-based nonparametric statistic that principles of mapping polygenes. Annu Rev Genet 1996; 30: 38 343–370. provides a powerful test for linkage. Empirical P 19 Lesch KP, Bengel D, Heils A et al. Association of anxiety-related values both for parametric and nonparametric linkage traits with a polymorphism in the serotonin transporter gene regu- analysis were estimated by running 1000 simulations. latory region. Science 1996; 274: 1527–1531. For the contingency tables, 10 000 simulations were 20 Deka R, Shriver MD, Yu LM et al. Conservation of human chromo- used. some 13 polymorphic microsatellite (CA)n repeats in chimpanzees. Genomics 1994; 22: 226–230. 21 Owen MJ, Holmans P, McGuffin P. Association studies in psychi- Acknowledgements atric genetics. Mol Psychiatry 1997; 2: 270–273. 22 Lander E, Kruglyak L. Genetic dissection of complex traits: guide- The authors wish to thank Dr Kenneth Morgan for lines for interpreting and reporting linkage results. Nat Genet 1995; helpful comments on the manuscript and Carrie Rob- 11: 241–247. ertson for technical assistance. GT has a CAPES Fel- 23 Risch N, Botstein D. A manic depressive history. Nat Genet 1996; lowship. MA holds an Intermediate Fellowship from 12: 351–353. the Ontario Mental Health Foundation. This research 24 Witte JS, Elston RC, Schork NJ. Genetic dissection of complex traits. Nat Genet 1996; 12: 355–356. was supported in part by grants from the Medical 25 Curtis D. Genetic dissection of complex traits. Nat Genet 1996; 12: Research Council of Canada and from the Ontario Men- 356–358. tal Health Foundation. 26 Cannings C, Thompson EA. Ascertainment in the sequential sampling of pedigrees. Clin Genet 1997; 12: 208–212. 27 Endicott J, Spitzer RL. A diagnostic interview: the schedule for References affective disorders and schizophrenia. Arch Gen Psychiatry 1978; 1 Goodwin D, Jamison K. Manic-Depressive Illness. Oxford Univer- 35: 773–782. sity Press: New York, 1990. 28 Spitzer RL, Endicott J, Robins E. Research diagnostic criteria: 2 Schou M. Forty years of lithium treatment. Arch Gen Psychiatry rationale and reliability. Arch Gen Psychiatry 1978; 35: 773–779. 1997; 54: 9–13. 29 Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning. A Labora- 3 Gitlin MJ, Alshuler LL. Unanswered questions, unknown future for tory Manual. Cold Spring Harbor Laboratory Press, NY, 1989. one of our oldest medications. Arch Gen Psychiatry 1997; 54: 30 Rothschild CB, Akots G, Fajans SS, Bowden DW. A microsatellite 21–23. polymorphism associated with the PLC1 (phospholipase C) locus: 4 Gershon S, Soares JC. Current therapeutic profile of lithium. Arch identification, mapping, and linkage to the MODY locus on chro- Gen Psychiatry 1997; 54: 16–20. mosome 20. Genomics 1992; 13: 560–564. 5 Grof P, Alda M, Grof E, Zvolsky P, Walsh M. Lithium response and 31 Dixon WJ, Brown MB, Engelman L, Jennrich RI. BMDP Statistical genetics of affective disorders. J Affect Disord 1994; 32: 85–95. Software Manual. University of California Press: Berkeley, LA, 6 Grof E, Grof P, Brown GM, Downie S. Lithium effects on neuroen- Oxford, 1990. docrine function. Prog Neuro-Psychopharmacol & Biol Psychiat 32 Sham PC, Curtis D. Monte Carlo test for association between dis- 1984; 8: 541–546. ease and alleles at highly polymorphic loci. Ann Hum Genet 1995; 7 Mendlewicz J. Prediction of treatment outcome: family and twin 59: 97–105. studies in lithium prophylaxis and the question of lithium red 33 Lathrop GM, Lalouel JM, Julier C, Ott J. Strategies for multilocus blood cell/plasma ratios. In: Cooper TB, Gershon S, Kline NS, linkage analysis in humans. Proc Natl Acad Sci USA 1984; 81: Schou M (eds). Lithium. Controversies and Unresolved Issues. 3443–3446. Excerpta Medica: Amsterdam, 1979. 34 Cottingham RWJ, Idury RM, Schaffer AA. Faster sequential genetic 8 Smeraldi E, Petroccione A, Gasperini M, Macciardi F, Orsini A, Kidd KK. Outcomes on lithium treatment as a tool for genetic stud- linkage computations. Am J Hum Genet 1993; 53: 252–263. ies in affective disorders. J Affect Disord 1984; 6: 139–151. 35 Terwilliger JD, Ott J. Handbook of Human Genetic Linkage. The 9 Manji HK, Potter WZ, Lenox RH. Signal transduction pathways. Johns Hopkins University Press: Baltimore and London, 1994. Molecular targets for lithium’s actions. Arch Gen Psychiatry 1995; 36 Ott J. Computer-simulation methods in human linkage analysis. 52: 531–543. Proc Natl Acad Sci USA 1989; 86: 4175–4178. 10 Gerfen CR, Choi WC, Suh PG, Rhee SG. Phospholipase C I and 37 Weeks DE, Ott J, Lathrop GM. SLINK: a general simulation program II brain isozymes: immunohistochemical localization in neuronal for linkage analysis. Am J Hum Genet 1990; 47: A204. systems in rat brain. Proc Natl Acad Sci USA 1988; 85: 3208–3212. 38 Davis S, Schroeder M, Goldin LR, Weeks DE. Nonparametric simul- 11 Rhee SG, Suh PG, Ryu SH, Lee SY. Studies of inositol phospholi- ation-based statistics for detecting linkage in general pedigrees. Am pid-specific phospholipase C. Science 1989; 244: 546–550. J Hum Genet 1996; 58: 867–880. 12 Rhee SG, Bae YS. Regulation of phosphoinositide-specific phospholipase C isozymes. J Biol Chem 1997; 272: 15045–15048. 13 McMahon FJ, Stine OC, Chase GA, Meyers DA, Simpson SG, Correspondence: M Alda, MD, Department of Psychiatry, University DePaulo JR. Influence of clinical subtype, sex, and lineality on age of Ottawa, Royal Ottawa Hospital, 1145 Carling Avenue, Ottawa, at onset of major affective disorder in a family sample. Am J Psy- Ontario K1Z 7K4, Canada. E-mail: [email protected] chiatry 1994; 151: 210–215. Received 6 February 1998; revised and accepted 10 June 1998