Brief Research Communication

Molecular Genetic Studies of Bipolar Disorder andPuerperal Psychosis at Two Polymorphisms in theEstrogen Receptor a Gene (ESR 1)

Ian Jones,* Fiona Middle, Fiona McCandless, Natasha Coyle, Emma Robertson, Ian Brockington,Corinne Lendon, and Nick CraddockDivision of Neuroscience, University of Birmingham, Birmingham B15 2QZ, United Kingdom

A number of lines of evidence point to thepossible involvement of estrogen pathwaysin the pathophysiology of bipolar disorderin general and puerperal psychosis in par-ticular. There is strong evidence from clini-cal, follow-up, and genetic studies to sup-port the hypothesis that most cases ofpuerperal psychosis are manifestations ofan affective disorder diathesis with a puer-peral trigger and that genes influence sus-ceptibility to both diathesis and trigger. Thenature of the trigger is unknown but in viewof the abrupt onset at a time of major physi-ological change it is widely believed thatbiological, probably hormonal, mechanismsare of paramount importance, with estrogenreceiving the most attention to date. Wehave undertaken a case control associationstudy of bipolar disorder and puerperal psy-chosis at two known polymorphisms withinthe estrogen receptor a gene (ESR 1) in asample of 219 unrelated bipolar probandsand 219 controls. We could exclude thesepolymorphisms from an important contri-bution to susceptibility to bipolar disorderwith a high level of confidence. We found nosupport for the hypothesis that they con-tribute specific susceptibility to the puer-peral trigger, but due to the small numbersof puerperal probands (n = 26) no firm con-clusions can be drawn regarding their in-volvement in puerperal psychosis. Am. J.Med. Genet. (Neuropsychiatr. Genet.) 96:850–853, 2000. © 2000 Wiley-Liss, Inc.

KEY WORDS: bipolar disorder; puerperalpsychosis; estrogen

INTRODUCTION

Family, twin, and adoption studies provide compel-ling evidence that genes influence susceptibility to bi-polar disorder [Craddock & Jones, 1999]. The mode ofinheritance is complex, and the (epistatic) interactionof several susceptibility genes represents a plausiblemodel for transmission [Craddock et al., 1995]. Casecontrol association studies represent a paradigm forinvestigating the role of specific candidate genes in dis-ease susceptibility but depend critically on the avail-ability of good candidates [Craddock & Owen, 1996a,1996b]. For complex disorders there are marked ben-efits in focusing on a homogenous subtype that allowsa subset of hypotheses to be tested [Lander & Schork,1994]. Molecular genetic studies of puerperal psychosis(PP) provide an excellent example of this strategy.

Puerperal Psychosis

Puerperal Psychosis, a psychotic episode precipi-tated by birth, follows approximately one in 1,000 de-liveries. Parturition is the most potent known precipi-tant of severe psychiatric disorder, with a 22-foldincreased risk of affective psychosis in the 4 weeks fol-lowing childbirth [Kendell et al., 1987]. Despite diversepresentations the clinical and genetic evidence sup-ports the hypothesis that the majority of postnatal psy-chotic episodes are manifestations of an affective dis-order diathesis with a puerperal trigger [Brockington,1996]. The majority of PP episodes are affective, withmania particularly common in the 2 weeks followingchildbirth [Protheroe, 1969; Thuwe, 1974; Dean & Ken-dell, 1981; Meltzer & Kumar, 1985; Schopf et al., 1985;McNeil, 1986; Kendell et al., 1987; Dean et al., 1989].Index PP probands frequently develop further puer-peral and nonpuerperal affective episodes [Davidson &Robertson, 1985] and women with bipolar disorder de-velop puerperal episodes following 20 to 30% of births[Bratfos & Haug, 1966; Reich & Winokur, 1970; Ken-dell et al., 1987].

Although there are no twin or adoption studies of PP,family studies of PP consistently demonstrate familial

Contract grant sponsor: Wellcome Trust; Contract grant spon-sor: South Birmingham Mental Health Trust; Contract grantsponsor: West Midlands Regional Health Authority; Contractgrant sponsor: Women’s Mental Health Trust

*Correspondence to: Dr. Ian Jones, Division of Neuroscience,University of Birmingham, Queen Elizabeth Psychiatric Hospi-tal, Birmingham B15 2QZ, United Kingdom.E-mail: I.R.Jones@bham.ac.uk

Received 13 October 1999; Accepted 16 June 2000

American Journal of Medical Genetics (Neuropsychiatric Genetics) 96:850–853 (2000)

© 2000 Wiley-Liss, Inc.

aggregation of psychiatric (particularly affective) dis-order, with morbidity risks for first degree relatives inthe range 10 to 50% [Protheroe, 1969; Thuwe, 1974;Kadramas et al., 1979; Whalley et al., 1982; Schopf etal., 1985; Platz & Kendell, 1988; Dean et al., 1989]. Thesingle large study using direct interview of relativesfound a significantly higher risk of affective illness inrelatives of PP women compared with parous bipolarwomen without puerperal illness [Dean et al., 1989].The family data, therefore, suggests a major overlap inthe genetic factors predisposing to PP and bipolar dis-order.

There is also evidence supporting familial factorsplaying a role in susceptibility to the postpartum trig-ger. The study of Dean et al. [1989] suggested a higherrisk of puerperal illness in relatives of PP probandscompared with those of nonpuerperal BP probands. Wehave examined the occurrence of episodes of PP infamilies multiply affected with bipolar disorder partici-pating in our ongoing sib-pair study of bipolar disorder[Bennett, 1998] and find conclusive evidence that fa-milial factors are implicated in susceptibility to the pu-erperal trigger, with puerperal episodes shown to clus-ter in families (k 4 0.57, P 4 0.001) [Jones &Craddock, unpublished data]. The importance of ge-netic factors in determining the puerperal trigger isfurther supported by several striking examples of fa-milial clustering of puerperal illness [e.g., Craddock etal., 1994] and concordance in reports of identical twinpairs [Kallmann 1938; Kane, 1968].

There is no evidence that the psychosocial context inwhich the delivery occurs influences susceptibility toPP [McNeil, 1988; Marks et al., 1991] and consider-ation of the abrupt onset during a time of major physi-ological changes (e.g., hormonal fluctuations) convincesmost researchers that biological factors are of funda-mental importance [Brockington, 1996]. The possiblerole of several hormones (e.g., prolactin, follicle stimu-lating hormone, luteinizing hormone) has been consid-ered but evidence for the involvement of estrogen—with its dramatic rise in levels during pregnancy andprecipitous fall following parturition—is most compel-ling [Speroff et al., 1994]. Aetiological theories havebeen proposed based on estrogen’s interaction with do-pamine [Cookson, 1982; Weik et al., 1991] and seroto-nin systems [Fink & Sumner, 1996].

Estrogen and Bipolar Disorder

A number of lines of evidence suggest the possibleinvolvement of estrogen pathways in the pathophysiol-ogy of bipolar disorder: (i) the close relationship be-tween bipolar disorder and episodes of PP; (ii) the spe-cial vulnerability of women to mood and anxietydisorders, which may result from the cyclical nature ofestrogen secretion from puberty to menopause and itssubsequent almost total withdrawal [Seeman, 1997];(iii) the marked effect of estrogen on mood state and itsantidepressant action [Henderson et al., 1991]; and (iv)the action of estrogen on the expression of several neu-ronally relevant genes including those encoding dopa-mine receptors [Weik et al., 1991; Guivarch et al.,1995], the 5HT2A receptor [Fink & Sumner, 1996],

MAOA [Ma et al., 1995], and tyrosine hydroxylase[Raab et al., 1995].

Steroid hormones such as estrogen regulate gene ex-pression in target cells through their interaction withspecific receptors. The ESR 1 (a) and ESR 2 (b) sub-types of the estrogen receptor are members of the su-perfamily of nuclear receptors, which can transduceextracellular signals into transcriptional responses.ESR 1 (6q25.1) is more that 140 kb long and containseight exons (Ponglikitmongkol et al., 1988]. Two knownpolymorphisms identified by the PvuII and XbaI re-striction enzymes are located in intron 1 of the geneand are not known to be functional.

Recruitment of a large sample of PP probands is on-going in our centre. In the current study we have ex-amined the hypothesis that variation at these two poly-morphisms influences susceptibility to bipolardisorder. Information on puerperal episodes was ob-tained for female probands, which made an analysis ofa subsample with PP possible. Therefore, we have ex-amined two hypotheses regarding the relationship be-tween ESR 1 and affective disorder: (i) that variation attwo polymorphisms in this gene influences susceptibil-ity to bipolar disorder, and (ii) that these polymor-phisms play a role in increasing susceptibility to thepuerperal trigger.

MATERIALS AND METHODSSamples

Bipolar Probands. Two hundred nineteen unre-lated British Caucasians (92 male [42%]; mean age of46 years, standard deviation 13 years), meeting DSM-IV (Diagnostic and Statistical Manual of Mental Dis-orders, 4th ed.; APA, 1994) criteria for bipolar I disor-der, were recruited from psychiatric outpatient clinicsin South Wales and the Midlands of England. Probandswere interviewed by a trained investigator using eitherSADS-L [Schedule for Affective Disorders and Schizo-phrenia, Lifetime Version; Endicott & Spitzer, 1978] orSCAN [Schedules for Assessment in Neuropsychiatry;Wing et al., 1990] and case note information was ob-tained. Best estimate lifetime diagnoses were made onthe basis of all available clinical information. Parousfemale probands were specifically asked about episodesof illness related to childbirth and classified into threegroups: (1) PP probands—DSM IV mania or other psy-chotic episode within 12 weeks of childbirth; (2) Non-PP bipolar women with no episodes of illness within 6months of childbirth; and (3) women with perinatal epi-sodes other than the narrow definition of PP given inthe preceding paragraphs, or where the relationshipbetween illness and childbirth could not be established.

Controls. Two hundred nineteen unrelated BritishCaucasian controls, not screened to exclude a history ofpsychiatric illness, were recruited from the blood trans-fusion service in South Wales. Controls were groupmatched for age and sex to the bipolar group (94 male[43%]; mean age 44, standard deviation 10 years).

Genotyping

High molecular weight genomic DNA was extractedfrom whole blood according to routine procedures. A

Bipolar Disorder and Puerperal Psychosis at Two Polymorphisms in ESR 1 851

1.3kb DNA fragment from intron 1 of ESR 1 was am-plified by the polymerase chain reaction (PCR) usingpreviously published primers [Yaich et al., 1992]. Thefollowing PCR conditions were used: initial denatur-ation at 94°C for 5 min followed by 30 cycles of dena-turation at 94°C for 60 sec, annealing at 63°C for 60sec, and primer extension at 72°C for 90 sec, plus afinal extension at 72°C for 5 min. Each reaction con-tained 50 ng genomic DNA, 25 pmol of each primer(Alta Bioscience, Birmingham, U.K.), 1.5 mM MgCl2,200 mM dNTPs, 1.0 unit Taq polymerase (Bioline, Lon-don, U.K.), and 1 × Taq buffer (as supplied by manu-facturer). A double digest was performed with PvuIIand XbaI and examination of the resulting fragmentsallowed determination of the haplotype for these twopolymorphisms. Separation was on 1% agarose gel withvisualisation by ethidium bromide. The four haplo-types resulting from digestion with XbaI and PvuIIwere xp (850, 390, 60 bp) Xp (850, 450 bp), XP (1.3 kb),and xP ( 910, 390 bp) (upper case letters—X, P—denoteloss of restriction site) in combination giving 10 pos-sible genotypes.

Statistical Analysis

Departure from Hardy-Weinberg equilibrium wastested using a x2 goodness of fit test. Statistical signifi-cance of differences between proband and control dis-tributions for the four possible haplotypes and geno-types based on the combination of these haplotypeswas assessed using x2 tests. For sparse contingencytables P-values were calculated via Monte Carlo simu-lations using the CLUMP program [Sham & Curtis,1995].

RESULTS

ESR 1 genotyping was successfully performed on 210bipolar probands and 208 controls. The control geno-type distribution was close to that expected underHardy-Weinberg equilibrium for both polymorphisms(P >0.83). The distributions of the four haplotypes inprobands and controls are shown in Table I. There wasno significant difference between haplotype (Clump x2

4 3.92, P 4 0.26) or genotype distributions (Clump x2

4 5.41, P 4 0.64) between bipolars and controls.Analysis of the subset of female probands and controlsalso revealed no significant differences between haplo-type (Clump x2 4 2.02, P 4 0.67) or genotype (Clumpx2 4 2.11, P 4 0.99) distributions.

PP Probands

Comparisons between the PP group and the controlsshowed no significant difference between haplotype(Clump x2 4 7.87, P 4 0.14) or genotype (Clump x2 49.76, P 4 0.16) distributions. A comparison of PP pro-bands with parous bipolar probands without episodesof puerperal illness again revealed no significant dif-ference between haplotype (Clump x2 4 1.67, P 40.70) or genotype (Clump x2 4 6.99, P 4 0.44) distri-butions.

DISCUSSION

Our methodology allowed us to examine the haplo-type distribution of two polymorphisms in ESR 1 inbipolar and control samples and to test two hypotheses.

Hypothesis One. We found no support for the hy-pothesis that variation at these polymorphisms influ-ences susceptibility to bipolar disorder. Our samplesize gives a power of greater than 99% to detect aneffect with an odds ratio of 2 and a power in excess of80% to detect an effect with an odds ratio of 1.5. We cantherefore reject this hypothesis with some confidence.

Hypothesis Two. We also have examined the pos-sibility that although not conferring susceptibility tobipolar disorder itself (the diathesis), variations atthese polymorphisms increase the risk of suffering pu-erperal episodes of illness (the trigger). We found nosupport for this hypothesis although the small numberof women with PP episodes (n 4 26) results in modestpower to detect all but very large effect sizes (power of80% to detect an effect with an odds ratio of 3.5). There-fore, we are unable to rule out these polymorphismsinfluencing susceptibility to the puerperal trigger withmore modest effect sizes.

In addition it must be stressed that we cannot ex-clude other polymorphisms in either this gene or in theESR 2 subtype as playing a role in susceptibility tobipolar disorder or the puerperal trigger. Indeed, in ratmodels there is work to suggest that the b subtype ofthe gene is more active in the brain than the a form[Kuiper et al., 1998].

Focusing on a homogenous subtype (PP) of a complexdisorder (bipolar) allows us to examine a logical set ofhypotheses regarding the involvement of estrogen andother neurosteroid pathways in pathophysiology. Thelarge sample of PP probands currently under collectionwill provide greatly increased power to test the involve-ment of polymorphisms in a hierarchy of candidatesincluding estrogen receptor genes, genes involved inestrogen action or metabolism, and neuronally relevantgenes containing estrogen response elements. Elucida-tion of the pathophysiological basis of the puerperaltrigger will lead to benefits in treatment and preven-tion of PP and is likely to inform research on a range ofother affective disorders.

In summary, our data do not support the hypothesis

TABLE I. Distributions (Frequency with Proportion inParentheses) of Haplotypes of Two Polymorphisms at the

Estrogen Receptor Gene

Haplotypesxp Xp XP xP

Total controls 236 (0.57) 0 (0.00) 147 (0.35) 33 (0.08)(n 4 208)

Total bipolar 218 (0.52) 2 (0.01) 160 (0.38) 40 (0.09)probands(n 4 210)

Female controls 130 (0.54) 0 (0.00) 88 (0.37) 22 (0.09)(n 4 120)

Female bipolar 129 (0.53) 2 (0.01) 88 (0.37) 23 (0.10)probands(n 4 121)

PP probands 29 (0.56) 1 (0.01) 18 (0.35) 4 (0.08)(n 4 26)

Non-PP bipolars 40 (0.46) 1 (0.01) 34 (0.40) 11 (0.13)(n 4 43)

852 Jones et al.

that variation at two polymorphisms in the ESR 1 geneinfluences susceptibility to either the bipolar disorderdiathesis or the puerperal trigger. However, genes en-coding proteins involved in estrogen pathways remainattractive candidates for affective disorders in generaland for bipolar affective PP in particular.

ACKNOWLEDGMENT

We would like to thank all those who participated inthe study.

REFERENCESAmerican Psychiatric Association (APA). 1994. Diagnostic and statistical

manual of mental disorders, 4th ed. Washington, DC: APA.

Bennett P, Mulchahy T, Owen MJ, Craddock N, Gill M. 1998. The Well-come Trust UK-Irish Bipolar sib pair study: chromosome 21. Am J MedGenet 81:541.

Bratfos O, Haug JO. 1966. Puerperal mental disorders in manic depressivefemales. Acta Psychiatrica Scand 42:285–294.

Brockington IF. 1996. Puerperal psychosis. In Motherhood and mentalhealth, p 200–284. Oxford: Oxford University Press.

Cookson JC. 1982. Poststpartum mania, dopamine, and estrogens. Lancetii:672.

Craddock N, Brockington I, Mant R, Parfitt E, McGuffin P, Owen M. 1994.Bipolar affective psychosis associated with consanguinity. Brit J Psy-chiatry 164:359–364.

Craddock N, Khodel V, Van Erdewegh P, Reich T. 1995. Mathematicallimits of multilocus models: the genetic transmission of bipolar disor-der. Am J Hum Genet 57:690–702.

Craddock N, Owen MJ. 1996a. Modern molecular genetic approaches topsychiatric disease. Brit Med Bull 52:434–452.

Craddock N, Owen MJ.1996b. Candidate gene association studies in psy-chiatric genetics: a SERTain future? Mol Psychiatry 1:434–436.

Craddock N, Jones I. 1999. Genetics of bipolar disorder. J Med Genet36:585–594.

Davidson J, Robertson E. 1985. A follow-up study of postpartum illness,1946-1978. Acta Psychiatrica Scand 71:451–457.

Dean C, Kendell RE. 1981. The symptomatology of puerperal illness. BritJ Psychiatry 139:128–133.

Dean C, Williams RJ, Brockington IF. 1989. Is puerperal psychosis thesame as bipolar manic-depressive disorder? A family study. Psychologi-cal Med 19:637–647.

Endicott, J, Spitzer, RL. 1978. A diagnostic interview. The Schedule forAffective Disorders and Schizophrenia. Arch Gen Psychiatry 35:837–844.

Fink G, Sumner BEH. 1996. Estrogen and mental state. Nature 383:306.

Guivarch D, Vernier P, Vincent JD. 1995. Sex steroid hormones change thedifferential distribution of the isoforms of the D2 dopamine receptormessenger RNA in the rat brain. Neuroscience 1:159–166.

Henderson AF, Gregoire AJP, Kumar R, Studd JWW. 1991. Treatment ofsevere postnatal depression with oestradiol skin patches. Lancet 338:816–817.

Kadramas A, Winokur G, Crow R. Postpartum mania. 1979. Brit J Psy-chiatry 135:551–554.

Kallman FJ. 1938. The genetics of schizophrenia. New York: J.J. Augustin.

Kane FJ Jr. 1968. Postpartum psychosis in identical twins. Psychosomatics9:278–281.

Kendell RE, Chalmer JC, Platz C. 1987. Epidemiology of puerperal psy-choses. Brit J Psychiatry 150:662–673.

Kuiper GGJM, Shughrue PJ, Merchenthaler I, Gustafsson JA. 1998. Theestrogen receptor b subtypr:a novel mediator of estrogen action in neu-roendocrine systems. Frontiers Neuroendocrinol 19:253–286.

Lander ES, Schork NJ. 1994. Genetic dissection of complex traits. Science265:2037–2048.

Ma ZQ, Violani E, Villa F, Picotti G, Maggi A. 1995. Estrogenic control ofmonoamine oxidase A activity in human neuroblastoma cells express-ing physiological concentrations of estrogen receptor. Euro J Pharma-col 284:171–176.

Marks MN, Wieck A, Checkley SA, Kumar R. 1991. Life stress and post-partum psychosis: a preliminary report. Brit J Psychiatry 158:45–49.

McNeil TF. 1986. A prospective study of postpartum psychoses in a highrisk group. I. Clinical characteristics of the current postpartum epi-sodes. Acta Psychiatrica Scand 74:205–216.

McNeil TF. 1988. A prospective study of postpartum psychoses in a highrisk group. 3. Relationship to mental health characteristics duringpregnancy. Acta Psychiatrica Scand 77:604–610.

Meltzer ES, Kumar R. 1985. Puerperal mental illness: clinical features andclassification: a study of 142 mother-and-baby admissions. Brit J Psy-chiatry 147:647–654.

Platz C, Kendell RE. 1988. A matched-control follow-up and family study of“puerperal psychosis.” Brit J Psychiatry 153:90–94.

Ponglikitmongkol M, Green S, Chambon P. 1988. Genomic organization ofthe human estrogen receptor gene. Embo J 7:3385–3388.

Protheroe C. 1969. Puerperal psychoses: a long term study 1927-1961. BritJ Psychiatry 115:9–30.

Raab H, Pilgrim C, Reisert I. 1995. Effects of sex and estrogen on tyrosinehydroxylase mRNA in cultured embryonic rat mesencephalon. MolBrain Res 33:157–164.

Reich T, Winokur G. 1970. Postpartum psychoses in patients with manicdepressive disease. J Nervous Mental Dis 151:60–68.

Seeman MV. 1997. Psychopathology in women and men: focus on femalehormones. Am J Psychiatry 154:1641–1647.

Schopf J, Bryois C, Jonquiere M, et al. 1985. A family hereditary study ofpost-partum “psychoses.” Euro Arch Psychiatry Neurol Sci 235:164–170.

Sham PC, Curtis D. 1995. Monte Carlo tests for associations between dis-ease and alleles at highly polymorphic loci. Ann Hum Genet 59:97–105.

Speroff L, Glass RH, Kase NG. 1994. Clinical gynecologic endocrinologyand infertility. Baltimore: Williams and Wilkins.

Thuwe I. 1974. Genetic factors in puerperal psychosis. Brit J Psychiatry125:378–385.

Weik A, Kumar R, Hirst AD, Marks MN, Campbell IC, Checkley SA. 1991.Increased sensitivity of dopamine receptors and recurrence of affectivepsychosis after childbirth. Brit Med J 303:613–616.

Whalley LJ, Roberts DF, Wentzel J, et al. 1982. Genetic factors in puer-peral affective psychoses. Acta Psychiatrica Scand 65:180–193.

Wing JK, Barbor T, Brugha T, Burke J, Cooper JE, Giel R, Jablenski A,Regier D, Sartorius N. 1990. SCAN. Schedules for clinical assessmentin neuropsychiatry. Arch Gen Psychiatry 47:489–593.

Yaich L, Dupont WD, Cavener DR, Parl FF. 1992. Analysis of the PvuIIrestriction-fragment-length-polymorphism and exon structure of theestrogen-receptor gene in breast-cancer and peripheral-blood. CancerRes 52:177–183.

Bipolar Disorder and Puerperal Psychosis at Two Polymorphisms in ESR 1 853