psychiatric genetics: the case of single gene disorders

European Child & Adolescent Psychiatry11:201–209 (2002) DOI 10.1007/s00787-002-0284-0 REVIEW

ECA

P 284

■ Abstract Quantitative and mo-lecular genetics have made impor-tant developments in the last threedecades. There is increasing evi-dence of the role of heredity in thefield of neuropsychiatric disordersin children. So far, only a few path-ways between genes and behaviourhave been unravelled.

Accepted: 28 August 2002

J. Steyaert (�) · J.-P. FrynsCentre for Human GeneticsUniversity of LeuvenHerestraat 493000 Leuven, BelgiumE-Mail: [email protected]. SteyaertCentre for Clinical GeneticsUniversity of Maastricht, The Netherlands

Quantitative genetics puts poly-genic inheritance models forward.Molecular genetic research basedon these models seems promising,but until now has provided only alimited explanation for the vari-ance in the studied neuropsychi-atric disorders. In these models thecomplexity of the expression of asingle gene grows exponentiallywith the number of genes involved.Consequently, research on thegene-phenotype relationships andphenotypical variability in suchmodels is extremely complex.

The candidate gene approach, inwhich the gene-phenotype pathwayof a single gene is studied, is moremanageable, and in our opinion es-sential in understanding multiple

gene models. We discuss recentfindings in the field and their rele-vance for neuropsychiatric pheno-types. Single gene defects will onlyexplain a part of the range of neu-ropsychiatric disorders in children,but the evidence that this approachcan generate may help to clarifyneuropsychiatric phenotypes. Thediscovery of single gene disordersin subgroups of subjects with aneuropsychiatric phenotype mayresult in new perspectives for theirtreatment.

■ Key words behaviouralphenotype – child psychiatry –genetics – autism – candidategenes

Jean SteyaertJean-Pierre Fryns

Psychiatric genetics: the case of single gene disorders

Introduction

Our cultural ancestors had clear ideas about the hered-ity of behaviour and personality characteristics. Homerattributed to Telemachos the wisdom and godlikeness ofhis famous father Odysseus [22]. Telemachos, however,had never met his father during the first twenty years ofhis life. Nature was stronger than nurture at that time.Much later Sir Francis Galton [39] approached this phe-nomenon in a more scientific way by studying thehereditability of cognitive abilities. This was the basis ofquantitative genetics. Ever since, and in particular in thepast three decades, the heredity of behavioural traitsand psychiatric conditions has been widely studied [17,39, 45]. This has not only led to interesting findings onthe heredity of personality characteristics, cognitive

abilities and psychiatric conditions, but also to more ac-curate descriptions of shared and non-shared environ-mental effects on development [33, 45]. In adult psychi-atry, the heredity of schizophrenia and bipolar disorderhas been studied extensively [35]. In developmental psy-chiatry, autism has been a particular focus of research[44]. This has resulted in valuable information on theheredity and the phenotype of autism, e. g. the conceptof a broader phenotype in some non-autistic familymembers of autistic probands [31, 34]. The developmentof statistical techniques [45] has allowed the study of arange of questions like hypotheses about possible mech-anisms involved in sex differences, or the number ofgenes that might play a role in the susceptibility for achild psychiatric disorder.

Several methods can be applied in the search forgenes affecting behaviour and/or mental disorder (see

202 European Child & Adolescent Psychiatry, Vol. 11, No. 5 (2002)© Steinkopff Verlag 2002

Table 1 for an overview).We refer to other articles on thissubject for a more comprehensive review of how thesemethods can be applied in neuropsychiatric research[45].Family and genetic epidemiology studies have beenaccompanied by molecular genetic studies, in search forgenes affecting behaviour and psychiatric disorders. Ini-tially, linkage analysis has been used to look for singlegenes that could cause the heredity of neuropsychiatricdisorders. Up to now, successes like the discovery of thegene for Huntington’s disease [19], in which only onegene (defect) is necessary and sufficient for a particulardisease phenotype, have not been reproduced in psychi-atric conditions. This has different reasons.

Firstly, research suggests that neuropsychiatric disor-ders are often caused by several interacting genetic loci,or quantitative trait loci (QTL’s), rather than by a singlemajor locus. This, for example, has been demonstratedin schizophrenia [42] and in autism. In autism the con-cordance in monozygotic twins is more than twice thatin dizygotic twins, suggesting that more than one geneis involved in the heredity of the disorder in these cases[45]. Linkage analysis searching for such QTL’s in autismhas shown that probably more than 15 loci are involved[43], while not a single gene at these loci has been iden-tified yet. Though there are robust arguments that thephenomenon of QTL’s operates in a substantial numberof cases, it is a particularly complex field of research, inpart because we do not exactly know what to look for.This obstacle can be alleviated by looking in the other

direction: instead of going backward and look for QTL’sstarting from the phenotype, one can focus on the asso-ciation of candidate genes with a particular neuropsy-chiatric disorder [11]. This has been performed, e. g. inAttention Deficit Hyperactivity Disorder (ADHD) [8].However, a problem that remains in a forward geneticapproach is that it is not clear how different genes com-bine to cause a particular neuropsychiatric phenotype.This is not necessarily through several defective genes,but also through particular combinations of polymor-phic genes. In the population, different alleles of onegene co-exist. These polymorphisms lead to small dif-ferences in the functioning of the gene product. Whenseveral such genes are involved,particular combinationsmay lead to dysfunction, as proposed in the study on theadditive effects of polymorphisms of three genes in-volved in dopamine function in Tourette syndrome [9].Secondly, different and independent genetic errors cancause the same psychiatric disorder. Different chromo-somal abnormalities are independently associated withautism [36]. This phenomenon demonstrates the ge-netic heterogeneity of neuropsychiatric disorders. It is alimitation in linkage analysis based on a QTL model.Such research requires large groups of subjects or fami-lies, and these groups are likely to be genetically hetero-geneous. In different families, completely independentgenes or combinations of genes may cause a similar neu-ropsychiatric phenotype, blurring linkage analysis.Thirdly, a psychiatric phenotype and the underlying ge-

Table 1 Brief overview of main strategies in molecular research for genes affecting behaviour or mental disorders

Searching for → Single genes Multiple genes

Backward approach: + Relative routine for linkage analysis. If single locus + This is most likely closer to the reality of the geneticsexists, there is a good chance of finding it. of neuropsychiatric disorders.

from behaviour to genes: + Good indication when there is a clear disease + QTL models can be taken into accountlinkage analysis phenotype with monogenic inheritance – Long way from finding multiple loci to finding genes.

in a large pedigree. – Even if genes are found, there is a good chance– Long way from finding locus to finding gene. that their pathogenetic mechanism is unknown.– No results if no single major gene effect, or if – At molecular level, understanding the simultaneous

penetrance of major gene is low. effect of multiple genes is extremely complex.– Finding a single major gene will often only

reflect part of reality in complex disorders.

Forward approach: + Effect of one gene on the phenotype can be + Not only individual effect, but also additive effectstudied in depth. and interactions of several (candidate) genes

from genes to behaviour: + Mechanisms of expression and variability can be on a phenotype can be studied.association/candidate gene studies studied for that gene, as well as function of the + QTL models can be studied.

gene product. – What happens at the molecular level is difficult+ The procedure may point to other related to study.

candidate genes. – Small or non-significant effect of candidate genes– Inherently, this approach does not allow describing in a complex model may erroneously lead to

combined and quantitative effects of sets of genes. removing these genes from the model.– Association studies with genes with an unknown – Inherent multiple analysis may lead to erroneously

function may lead to false positive results if there significant effects that do not reflect an effect atis another reason for association between the gene the molecular level.and the phenotype (e. g. ethnic group)

“+”-signs indicate advantages of the approach, and “–”-signs limitations. The list of advantages and limitations is not exhaustive

J. Steyaert et al. 203Single gene disorders

netic abnormalities may not always coincide [45]. Thesame genetic liability seems to underlie conduct disor-der and oppositional defiant disorder, which are gener-ally considered as different diagnostic categories. A ge-netically influenced behaviour trait may be a risk factorfor a psychiatric disorder, independent of the genetic li-ability for the disorder itself. This was demonstrated forthe genetically influenced personality trait neuroticismas a risk factor for depressive disorder [28].A fourth rea-son is that, ideally, the definition of a particular diseasephenotype should precede the search for causativegenes. On the other hand, it often only becomes appar-ent which phenotype is associated with a particular ge-netic defect or variant once the latter has been identi-fied. Only after it became possible to detect the fragile-Xsite Xq27.3, did it become clear that not only males, butalso carrier females often have clinical characteristics ofthe fragile-X syndrome [14, 21]. This paradox compli-cates research on gene-behaviour relationships whenstarting from the psychiatric phenotype.

Nevertheless, in some neuropsychiatric disorders, ithas been possible to define a phenotype that could belinked to a causative gene. In a single family aggressivebehaviour and borderline intellectual functioning werefound to be associated with a mutation in the MAO-Agene [5]. In the KE-family, a specific language impair-ment is associated with the FOXP2 gene [24, 30]. How-ever, even in a seemingly one-gene-one-behaviour situ-ation like Brunner’s syndrome, it became clear that therelationship between the gene defect and the behaviouris not so straightforward. The aggressive behaviour ofthe subjects with a MAO-A gene mutation may be due todistant effects of the gene on the metabolism of otherneurotransmitters [4].

In conclusion, linkage analysis looking for singlecausative genes in neuropsychiatric disorders has onlyled to very modest results. Linkage analysis looking formultiple genes is somehow like looking for unknownobjects in a haystack. Sometimes this may be a success-ful endeavour, but it is generally more rewarding toknow more about the kind of object one is searching.This consideration is in favour of a forward approachbased on multiple (candidate) genes. However, this doesnot decrease the complexity of the problem. Consider-ing the given arguments and questions, we advocate aforward single gene approach as preliminary step to-wards more complex models whith multiple candidategenes. Understanding the variability and mechanismsinfluencing the expression of single genes is in our viewelementary in dealing with multiple gene models, justlike a mason should be acquainted with bricks and mor-tar before building houses. Moreover, in families wherea single gene has a major effect on the occurrence of aneuropsychiatric disorder, it may help to understandwhy there remains such variability in the phenotype.

Aim of this review

Variability in phenotypic expression is rather the rulethan the exception in single gene disorders, e. g. fragileX syndrome or neurofibromatosis type 1.A similar vari-ability may be expected for genes that have an effect inthe pathogenesis of a particular neuropsychiatric disor-der. We reviewed a number of the mechanisms thatcause variability, to gain a better understanding of thedifficulties occurring in our research with single candi-date genes in neuropsychiatric disorders, in particularmental retardation and autism. These difficulties are atleast multiplied in similar research with multiple candi-date genes. We publish this review, as it may help to un-derstand why for many years to come child psychiatristswill have to deal with a discrepancy between generalknowledge about the hereditability of neuropsychiatricdisorders and what can actually be demonstrated in anaffected individual or family.

Methods

Literature on gene/behaviour research was selectedfrom more than 6000 Medline references on the topic.One group of articles covers the past four decades, andconsists mainly of review articles by influential authors,and articles that have been seminal in the field ofgene/behaviour research. The second group of selectedarticles are specific research findings on molecularmechanisms in single gene disorders. For obvious rea-sons, we favoured papers on single gene disorders withan associated behavioural phenotype, though this wasnot a strict limitation.Another bias was that we favouredpapers on psychiatric and genetic conditions that aresubject of research in our centre, e. g. autism, mental re-tardation, dynamic mutations, Velo-Cardio-Facial Syn-drome (VCFS).

■ Gene/behaviour relationships in single-genedisorders with a behavioural phenotype

Behaviour researchers often assume that single genedisorders have a dichotomous effect on development,with a clear-cut phenotype when the gene defect is pres-ent versus normality when it is not [39]. While this maybe true in Huntington’s chorea, the dichotomy betweennormality and impairment is not always so clear in otherconditions. In neurofibromatosis type 1 (NF1), evenwhen pathognomonic physical signs are present, the be-haviour of the subject may not be affected, while othersubjects manifest marked learning problems and/or be-havioural impairment like poor social abilities and poorimpulse control. In a family, some affected membersmay only show minor physical signs, while others, with


exactly the same NF1 gene mutation have marked be-havioural impairment [27]. The same phenomenon oc-curs in females with the fragile X full mutation, of whichat least 25 % have no cognitive or behavioural impair-ment. It is also seen in VCFS caused by a microdeletionof a small group of genes on chromosome 22q11 (22q11-deletion). Some of these subjects have marked cognitiveimpairment, a higher than expected number of themhave psychotic disorders, but others have no marked be-havioural impairment. This phenomenon is indepen-dent of cardiac malformations that also occur with greatvariability in VCFS [49]. In tuberous sclerosis some sub-jects have severe autism and/or epilepsy, while othersare normal [23].

These examples demonstrate that disorders of singlegenes that are expressed in the central nervous systemdo not necessarily have a dichotomous effect on behav-iour, but rather lead to a spectrum of both quantitativeand qualitative impairment. In some single gene disor-ders, the impairment will be severe and look like a di-chotomy. Only a small minority of fragile-X males havea cognitive level in the normal range [46], while the ma-jority have moderate mental retardation. In other singlegene disorders, e. g. NF1, this dichotomy is not presentand the phenotype is a continuum ranging from im-pairment to normality. These genes have a reduced pen-etrance in a number of individuals, and these carrierstransmit the disease through their pedigree while theyare clinically unaffected. Thus, in a number of familieswith several members with a neuropsychiatric disorder,a single gene anomaly can be associated with a distinctphenotype, while in other families the same phenotypeis caused by an additive effect of multiple genes. For ex-ample, in a sample of subjects with schizophrenia, asmall but higher than expected number have 22q11-deletion [2], while in other subjects, other genetic ornon-genetic factors underlie the disorder. We do notknow how often single gene defects or variants occur ina large population of subjects with a particular psychi-atric phenotype: a large number of single gene disordersis still unknown, and so is their prevalence. These con-ditions are not necessarily very rare: 22q11-deletion,and the NF1-gene mutations are present in 1 in 4000 in-dividuals.Moreover,single gene disorders are not alwaysassociated with a recognisable physical phenotype: in X-linked mental retardation (XLMR) at least 33 genes onthe X chromosome are associated with moderate ormild mental retardation in males, without being associ-ated with a particular physical phenotype [6]. It is pos-sible that these so-called non-specific XLMR-genes areactually associated with specific behavioural pheno-types [13].

To understand the possible effects of single genes onbehaviour, it is important to understand the mecha-nisms that can explain the large variability in pheno-type. History has shown that discovering such a mecha-

nism in one gene can help to discover the gene-pheno-type pathways in other genes in which a similar mecha-nism operates. When intergenerational amplification ofa trinucleotide repeat was discovered as the underlyingmolecular mechanism in fragile-X syndrome, thishelped elucidate the phenomenon of anticipation1 inseveral neurological diseases [38]. The different impactof a single gene on the phenotype can be explained bysome of the mechanisms discussed below, though thisoverview is certainly not exhaustive.

Phenotypical differences and genotypical heterogeneity

One reason why the phenotype of a clinical disorder candiffer in severity between subjects is the fact that the dis-order may have different, though often related, geneticcauses. In early-onset Alzheimer’s disease (EOAD), atleast three different genes coding for enzymes in the bio-chemical pathway of amyloid β peptide lead to a similarneuropsychiatric disorder, though the average age of on-set of EOAD may differ depending on which of the genesis mutated [10].

Another form of genetic heterogeneity is when dif-ferent mutations of a disease gene coexist in the popula-tion, each of them with a different impact on the pheno-type. In the autosomal recessive metabolic disorderphenylketonuria (PKU), a deficit of the enzyme phenyl-alanine hydroxylase results in elevated levels of phenyl-alanine and a varying degree of mental retardationwhen untreated. Different mutations in this gene resultin varying degrees of elevated phenylalanine and men-tal retardation [20]. In cystic fibrosis (CF), an autosomalrecessive disorder with severe pulmonary and pancre-atic involvement, at least 300 different mutations in theCFTR-gene have been described [7].These different mu-tations have various impacts on the phenotype: the typ-ical severe CF-phenotype is associated with a “severe”mutation on each of both alleles, while “milder” muta-tions on both alleles or a severe mutation on one allelecombined with a normal allele are associated with bilat-eral absence of the vas deferens in males, but not withCF.

Dynamic repeats, anticipation, variable expression

In the genome many forms of particular sequences oftwo, three or four base pairs occur. These di-, tri- ortetranucleotide repeats are generally stable. The term“dynamic mutations” is used for these polynucleotiderepeats that may become unstable and expand over gen-erations [41]. A CGG-trinucleotide repeat was first de-

1 Anticipation means that a particular hereditary disease occurs at ayounger age and has a more severe course in offspring than in par-ents.


scribed in the fragile-X mental retardation gene (FMR1gene) [58], and later found to be the cause of the fragile-X phenotype [53]. Healthy subjects have less than 50CGG repeats, and this number is stable from parents tooffspring. Normal individuals can, however, also be car-riers of a so-called permutation, with 50 to 200 CGG re-peats. This number of CGG repeats becomes unstablewhen transmitted by a female. Once the expansion ex-ceeds 200 CGG repeats the full mutation is present.Translation of the FMR1 gene is impaired, resulting inabsence of the FMR1 protein [12], as in the fragile-Xphenotype. This mechanism may take several genera-tions before the syndrome occurs. In fragile-X syn-drome, the phenotype varies in a rather dichotomousway: males with the permutation do not have the fra-gile-X phenotype,while males with the mutation do.Dy-namic mutations are the cause of at least seven disorderphenotypes (see Table 2) [37, 41]. In some of these dis-orders, e. g. Friedreich’s Ataxia and spinocerebellarataxia (SCA), there is additional genetic and phenotypi-cal heterogeneity. It is striking that all the dynamic mu-tations that have been described until now primarily af-fect the central nervous system. In trinucleotide repeatdisorders the severity of the phenotype may depend onthe expansion length of the triplet repeat, and there areoften threshold effects. In Huntington’s disease the dis-ease occurs above a threshold of 37 CAG repeats in thegene, and above that threshold longer repeats are asso-ciated with an earlier age of onset and a more malignantdisease process [3]. It also has been suggested that inmultiplex families with major psychoses an expandingtrinucleotide repeat might explain the finding that thedisease process can get worse in consecutive generations(anticipation) [38]. This hypothesis has not been con-firmed so far.

Another aspect of trinucleotide expansions is that wedo not know much about the phenotype of expansions

that exceed the normal range but which are not associ-ated with the classical phenotype of that disorder. Somefindings suggest that in some genes, intermediate repeatlengths could be associated with different phenotypes.The fragile-X permutation, an expansion of 55 to 200CGG repeats near the FMR1 gene, is not associated withthe fragile-X phenotype. However, there is strong evi-dence that female carriers of the permutation are at riskfor premature ovarian failure [1], and there are sugges-tions that a number of them have a different cognitivedevelopment [47, 52]. In the SCA8 gene, an expansionbetween 107 and 127 CTG-repeats is associated withspinocerebellar ataxia,while much longer repeats are as-sociated with major psychosis in some individuals, andnot with spinocerebellar ataxia [54]. It is clear that theimportance of trinucleotide repeats in neuropsychiatricdisorders should be studied further.

Sex effects

An obvious reason for phenotypical variation is sex ef-fect in X-linked disorders. Females can benefit from thecompensatory effect of the homologue gene on theirsecond X-chromosome. In some conditions, e. g. colourblindness, females will be clinically unaffected carriersof the gene. In other conditions, e. g. fragile-X, femalesmanifest a broad phenotypical spectrum, ranging fromcomplete normality to moderate mental retardation.However, more indirect sex effects also exist in the ex-pression of genes associated with particular pheno-types. Genetic factors play an important role in theautism phenotype.Autism is strikingly more frequent inmales than in females, but no data have been found infavour of the location of “autism” genes on the X chro-mosome. Therefore, it is assumed that pre-natal hor-monal effects may play a role in the expression of theautism phenotype [32].

Disorder Mutation

Fragile X syndrome (CCG)n repeat expansion in promoter region of FMR1 gene;locus Xq27.3

Myotonic Dystrophy (CTG)n repeat expansion in untranslated region of DMPK gene;locus 19q13.3

Huntington’s chorea (CAG)n repeat expansion in huntington gene; locus 4p16.3

Spinocerebellar ataxias type 1–17 At least 17 different genes. In at least 7 of these, the mutation(SCA1–SCA17). is a dynamic (CAG)n repeat expansion.

Friedreich’s ataxia (FRDA) (GAA)n repeat expansion in frataxin gene; locus 9q13. Other gene(frataxin-2) at locus 9p23.

Dentatorubral pallidoluysian atrophy (CAG)n repeat expansion in atrophin-1 gene; locus 12p13.31

Spino-bulbar muscular atrophy (CAG)n repeat expansion in androgen receptor gene*;(Kennedy disease) locus Xq11–q12

The data are based on “Online Mendelian Inheritance in Man”database (37). This overview does not include dis-orders that have only been found in a very small number of subjects*Smaller than average repeat lengths at this site possibly result in a higher risk for prostate cancer (25)

Table 2 Overview of central nervous system disor-ders associated with dynamic mutations


Effects of other genes

Other genes can affect the outcome in a single gene dis-order. The effect of a defective gene can depend on thefunctioning of another gene. In a number of metabolicpathways, poor or absent expression of two genes maybe necessary before any phenotypical effect is present.This is of course the case in autosomal recessive disor-ders, where the presence of one normal copy of the geneis sufficient for normal functioning, while absence ofboth alleles causes disease. In addition, more subtlemechanisms exist: in some monogenic mental retarda-tion disorders, the mutated gene has such an impact oncognitive function that the level of mental retardation isindependent of other factors. In other disorders, thoughthe affected subjects have a marked cognitive impair-ment, the intelligence quotient correlates with the cog-nitive level of the subject’s parents. This was demon-strated in VCFS (22q11-deletion) [48]. It is assumed thatin VCFS “background” genes still influence intelligence,while they have no effect in fragile-X syndrome.

It becomes even more complicated when mutationsin several different genes are necessary for the disease tooccur, as suggested in the theory of polygenic inheri-tance of disorders. Few examples in molecular geneticssupport this. One example is that of Bardet-Biedl syn-drome (BBS), an autosomal recessive disorder includingpigmentary retinal dystrophy, polydactyly, obesity, de-velopmental delay and renal defects. Two different genesfor BBS have been found, BBS2 and BBS6, and it was ex-pected that subjects with two mutated BBS2 or BBS6genes would manifest the BBS phenotype [26]. However,in the studied pedigrees subjects were found with twomutated BBS2 genes and a normal phenotype, while anadditional mutation in one of the BBS6 genes is neces-sary to manifest the BBS phenotype.

Distant effects of gene defects

Gene defects can have molecular consequences that ex-tend far beyond the obvious pathway of gene,gene prod-uct and the metabolic function of the gene product.Firstly, through feed-back mechanisms, poor or absentexpression of a gene may increase translation to mes-senger RNA (mRNA). Eventually this specific mRNAwill accumulate. There is evidence that in MyotonicDystrophy, the myopathy is caused by the mutantmRNA, while cardiac conduction defects and cataractsare caused by a deficiency in the DMPK protein, the geneproduct of one of the two known genes responsible forMyotonic Dystrophy [50]. In carriers of the fragile-Xpremutation the mRNA levels of the FMR1 gene aremuch higher than in normal controls [51], and it isthought that this may play a role in the phenotypicalanomalies found in some fragile-X premutation carriers(see above).

Secondly, the absence of a particular gene productmay have positive or negative feed back effect on thetranslation of other genes. In mice without FMR1 gene(knock-out fragile-X mice), the expression of at least 200genes has changed quantitatively [15, 16].

Thirdly, in neurotransmitter metabolism changes ina coding gene may alter the efficiency of an enzyme, re-sulting in other metabolic equilibriums and conse-quently different concentration of the neurotransmitter.In Brunner’s syndrome [5] the MAO-A deficiency hascomplex effects on the function of several neurotrans-mitters and it is not clear which of these changes affectbehaviour [4].

Chance

Randomness is a fundamental characteristic of proba-bilistic events and consequently these are difficult to de-scribe. However, stochastic effects (the cumulative effectof probabilistic events) play a role in the development ofa complex organism. This is one of the reasons whymonozygotic twins (MZT) are not completely similarand may have different anatomical defects although theyshare the genetic susceptibility for this defect. Individu-als with trisomy 21 or 22q11-deletion often have atrio-ventricular septum defects (AVSD). However, MZT pairswith these genetic syndromes may be discordant forcardiac defects [55]. Even in one individual, a gene de-fect may affect one side of the body and not the other, asseen in the Bardet-Biedl syndrome (see above), wherethe polydactyly may only affect one hand and not theother. This phenomenon of asymmetry due to proba-bilistic events can also be observed in normal develop-ment, e. g. selective breeding shows that the averagelength of the white leg markings in horses is hereditary,but the differences of these white markings between leftand right legs are probabilistic and not determined bygenes [57].

To demonstrate the stochastic basis of closure of theatrioventricular septum, a computer model simulatingthe growth of the septum between the two endocardialcushions during embryonic development was made[29]. The septum closes through cell division, migrationand adhesion. The computer model allowed for skewingof (genetically determined) parameters for cell division,migration and adhesion between cells,as seems to be thecase in trisomy 21. Consecutive runs of the computerprogram with an identical skewing resulted in non-clo-sure of the atrioventricular septum in a number of runs,but not in others. Thus, changing these genetically de-termined parameters resulted in an increased chance todevelop an AVSD, but did not predict normal or abnor-mal development in a dichotomous way. This is exactlywhat happens in trisomy 21, where some children withDown syndrome are born with AVSD and others with anormal heart, even in monozygotic twins. This shows


that stochastic effects can yield significant phenotypicvariability among individuals with identical genotypesand environments.

Discussion

Quantitative behaviour genetics investigate to what ex-tent a particular behavioural trait or disorder can be ge-netically determined and which mechanisms may be in-volved in its heredity. Molecular behaviour geneticsinvestigate which genes can be involved in behaviourand psychiatric disorders and by which mechanismsthese genes operate. In the past few decades,quantitativeand molecular behaviour genetics have yielded impor-tant insights in developmental psychiatry and valuablenew techniques in both fields. Though psychiatric dis-orders are often strongly genetically determined theycannot be seen as one-gene-one-disorder diseases [40].Their heredity generally does not follow classicMendelian rules, and more complex forms of heredityhave been suggested. Quantitative genetics has providedstrong arguments in favour of polygenic causality. Ac-cording to this model, a large number of genes, or quan-titative trait loci, in interaction with environment fac-tors, have an additive effect contributing to thepsychiatric phenotype. These studies necessitate largesamples of patients and families, which are inevitablygenetically heterogeneous. The examples given in thisreview compel us to assume that a large sample with aparticular phenotype, e. g. schizophrenia, consists of atleast a subgroup in which the additive effect of a partic-ular set of genes is responsible for the phenotype, and asubgroup in which a number of isolated genes, or veryfew genes cause the phenotype. One argument in favourof the existence of this mono- or paucigenic group isthat some single gene disorders and isolated chromoso-mal defects are associated with a psychiatric phenotype.A second argument is that the few molecular studies onpolygenic mechanisms have showed so far that, evenwhen the additive effect of a large number of genes isanalysed, this only explains a modest part of the vari-ance, even in disorders with a high hereditability [8]. Tocomplicate matters, the polygenic group may in its turnconsist of several subgroups in each of which a differentset of genes contributes to the phenotype. However,there are no simple clinical means to separate the poly-genic and the mono- or paucigenic groups, and we donot know their relative importance and contribution.

Differentiation of the phenotype may be a rewardingtechnique to obtain different and genetically more ho-mogeneous subgroups in these samples, which in turnwould facilitate the search for the involved genes. By dif-ferentiating the phenotype “early onset Alzheimer’s Dis-ease”, this method led to the discovery of genes that playa role in Alzheimer’s disease. However, we have only dim

cues on how to achieve such a redefinition of pheno-types. Ideally, it should be based on genetic characteris-tics, and that is exactly what we are looking for and donot have yet. Eventually, research based on other pheno-typic categories might lead to the conclusion that the re-lation between genes and phenotype is often too com-plex and indirect,or should be studied in a different way.

Considering these methodological difficulties in thestudy of gene–behaviour pathways using a backward ap-proach starting from the phenotype side, it is prudentalso to study what we can learn when starting from theside of molecular biology, or forward approach. Disor-ders involving only one or very few genes can have aphenotypic expression that is much more variable thaninitially expected, as we have illustrated above. In thepast 10 years, in-depth study of the molecular biology ofsome of these disorders has revealed formerly unknownmechanisms, which operate in more genes and disor-ders. Simultaneous effect of other genes, distant effect ofgene defects, dynamic mutations, probabilistic mecha-nisms, and yet undiscovered mechanisms explain thebroad phenotypic variance of single gene disorders andtheir seemingly non-Mendelian patterns of inheritance.They illustrate how complex the gene-behaviour path-way already is when the genetic causes are compara-tively simple, and give us a faint idea of the complexityof the problem when a large number of genes are in-volved. The rule of parsimony in research suggests tostudy first how a single gene may affect neuropsychiatricphenotypes. We do not know the prevalence of singlegene disorders in a population with a particular psychi-atric phenotype, but can learn much from the study ofphenotypes associated with single gene disorders.Firstly, it gives insight in the pathway between that par-ticular gene and behaviour. Secondly, through showingwhich behaviour is influenced by a particular gene, itmay help to redefine the phenotype of neuropsychiatricdisorders, and thus open new alleys in gene-behaviourresearch of these disorders. Thirdly, it may lead to previ-ously unknown mechanisms in molecular biology. Oneway to do this is to study individual patients or familiesin whom a neuropsychiatric disorder is associated witha demonstrable gene defect. This is the case in chromo-somal rearrangements, microdeletions, or microdupli-cations. In chromosomal rearrangements, the genes dis-rupted at the breakpoints can be identified. Inmicrodeletions the effect of deleted genes can beanalysed, as is the case in the association of schizophre-nia with 22q-deletion. In microduplications we canstudy the effect of redundant genes, as in the recentlydiscovered association between anxiety disorders and amicroduplication on chromosome 15 [18].Fourthly,bet-ter knowledge of the building blocks may improve ourunderstanding of how multiple genes operate together.Finally, focussing on single genes in an array of QTL-genes underlying a neuropsychiatric disorder may open


new alleys to treatment, as was illustrated by the findingthat children with ADHD – a polygenic disorder – havedifferent therapeutic responses to methylphenidate de-pending on which alleles of the dopamine receptor D2genes they have [56].

Clarifying pathways and hopefully phenotypes usingthis molecular genetic method will probably lead to newapproaches in quantitative genetics. This may allow new

insights into the genetics of polygenic neuropsychiatricphenotypes, and will certainly bring both fields of re-search even closer to each other.

■ Acknowledgements We are most grateful for the critical and lin-guistic comments from dr. Th. de Ravel and the constructive sugges-tions from the anonymous referees.

References

1. Allingham-Hawkins DJ, Babul-Hirji R,Chitayat D, Holden JJ, Yang KT, Lee C,Hudson R, Gorwill H, Nolin SL, Glicks-man A,Jenkins EC,Brown WT,Howard-Peebles PN, Becchi C, Cummings E, Fal-lon L, Seitz S, Black SH, Vianna-Morgante AM, Costa SS, Otto PA, Min-groni-Netto RC, Murray A,Webb J, Mac-Swinney F, Dennis N, Jacobs PA, SyrrouM, Georgiou I, Patsalis PC, GiovannucciUzielli ML, Guarducci S, Lapi E, CecconiA, Ricci U, Ricotti G, Biondi C, ScarselliB, Vieri F (1999) Fragile X permutationis a significant risk factor for prematureovarian failure: the International Col-laborative POF in Fragile X study – pre-liminary data. American Journal ofMedical Genetics 83:322–325

2. Bassett AS, Hodgkinson K, Chow EW,Correia S, Scutt LE, Weksberg R (1998)22q11 deletion syndrome in adults withschizophrenia. American Journal ofMedical Genetics 81:328–337

3. Brandt J, Bylsma FW, Gross R, Stine OC,Ranen N, Ross CA (1996) Trinucleotiderepeat length and clinical progressionin Huntington’s disease. Neurology 46:527–531

4. Brunner HG (1996) MAOA deficiencyand abnormal behaviour: perspectiveson an association. Ciba FoundationSymposium 194:155–164

5. Brunner HG, Nelen M, Breakefield XO,Ropers HH, van Oost BA (1993) Abnor-mal behavior associated with a pointmutation in the structural gene formonoamine oxidase A. Science 262:578–580

6. Chiurazzi P, Hamel BC, Neri G (2001)XLMR genes: update 2000. EuropeanJournal of Human Genetics 9:71–81

7. Claustres M, Guittard C, Bozon D,Chevalier F, Verlingue C, Ferec C,Girodon E, Cazeneuve C, Bienvenu T,Lalau G, Dumur V, Feldmann D, Bieth E,Blayau M, Clavel C, Creveaux I, MalingeMC, Monnier N, Malzac P, Mittre H,Chomel JC, Bonnefont JP, Iron A, CheryM, Georges MD (2000) Spectrum ofCFTR mutations in cystic fibrosis andin congenital absence of the vas defer-ens in France. Human Mutation 16:143–156

8. Comings DE,Gade-Andavolu R,Gonza-lez N,Wu S,Muhleman D,Blake H,DietzG, Saucier G, MacMurray JP (2000)Comparison of the role of dopamine,serotonin, and noradrenaline genes inADHD, ODD and conduct disorder:multivariate regression analysis of 20genes. Clinical Genetics 57:178–196

9. Comings DE, Wu S, Chiu C, Ring RH,Gade R, Ahn C, MacMurray JP, Dietz G,Muhleman D (1996) Polygenic inheri-tance of Tourette syndrome, stuttering,attention deficit hyperactivity, conduct,and oppositional defiant disorder: theadditive and subtractive effect of thethree dopaminergic genes – DRD2, Dbeta H, and DAT1. American Journal ofMedical Genetics 67:264–288

10. Cruts M, Van Broeckhoven C (1998)Molecular genetics of Alzheimer’s dis-ease. Annals of Medicine 30:560–565

11. Evans KL, Muir WJ, Blackwood DH,Porteous DJ (2001) Nuts and bolts ofpsychiatric genetics: building on theHuman Genome Project. Trends in Ge-netics 17:35–40

12. Feng Y, Zhang F, Lokey LK, Chastain JL,Lakkis L, Eberhart D, Warren ST (1995)Translational suppression by trinu-cleotide repeat expansion at FMR1. Sci-ence 268:731–734

13. Fisch GS (2000) Psychological assess-ment in XLMR: a proposal for settinginternational standards. Genetic Coun-seling 11:85–101

14. Fryns JP (1984) The fragile X syn-drome. A study of 83 families. ClinicalGenetics 26:497–528

15. Gantois I, Kooy RF (2002) Targetingfragile X. Genome Biol 3(5):1014.1–1014.5

16. Gantois I, Reyniers E, Kooy F (2002)Proceedings of the 10th InternationalWorkshop on Fragile X and XLMR:Identification of genes differentially ex-pressed in the fragile X knockoutmouse. Genetic Counseling 13:207–261

17. Gelernter J (1999) Clinical moleculargenetics. In: Charney DS, Nestler EJ,Bunney BS (eds) Neurobiology of Men-tal Illness.Oxford University Press,NewYork, pp 108–120

18. Gratacos M, Nadal M, Martin-Santos R,Pujana MA,Gago J,Peral B,Armengol L,Ponsa I, Miro R, Bulbena A, Estivill X(2001) A polymorphic genomic dupli-cation on human chromosome 15 is asusceptibility factor for panic and pho-bic disorders. Cell 106:367–379

19. Gusella JF (1984) Genetic linkage of theHuntington’s disease gene to a DNAmarker. Canadian Journal of Neurolog-ical Sciences 11:421–425

20. Guttler F, Guldberg P (2000) Mutationanalysis anticipates dietary require-ments in phenylketonuria. EuropeanJournal of Pediatrics 159 Suppl 2:S150–153

21. Hagerman RJ, Chudley AE, Knoll JH,Jackson AWd, Kemper M, Ahmad R(1986) Autism in fragile X females.American Journal of Medical Genetics23:375–380

22. Homer (1995). Odyssey, books 1–12.Loeb Classical Library. Vol. 1. HarvardCollege

23. Hunt A, Shepherd C (1993) A preva-lence study of autism in tuberous scle-rosis. Journal of Autism and Develop-mental Disorders 23:323–339

24. Hurst JA,Baraitser M,Auger M,GrahamF, Norell S (1990) An extended familywith a dominantly inherited speech dis-order. Developmental Medicine andChild Neurology 32:347–355

25. Irvine RA,Yu MC, Ross RK, Coetzee GA(1995) The CAG and GGC microsatel-lites of the androgen receptor gene arein linkage disequilibrium in men withprostate cancer. Cancer Research 55:1937–1940

26. Katsanis N, Lupski JR, Beales PL (2001)Exploring the molecular basis ofBardet-Biedl syndrome. Human Molec-ular Genetics 10:2293–2299

27. Kayl AE, Moore BD, 3rd (2000) Behav-ioral phenotype of neurofibromatosis,type 1. Mental Retardation and Devel-opmental Disabilities Research Reviews6:117–124

28. Kendler KS (1996) Models for the jointeffect of genotype and environment onliability to psychiatric illness. AmericalJournal of Psychiatry 143:279–289


29. Korenberg JR, Kurnit DM (1995) Mo-lecular and stochastic basis of congeni-tal heart disease in Down syndrome. In:Clark EB, Markwald RR, Takao A (eds)Developmental Mechanisms of HeartDisease. Futura Publishing Co, Inc, NewYork, pp 581–596

30. Lai SL, Fisher S, Hurst JA, Vargha-Khadems F, Monaco AP (2001) A fork-head-domain gene is mutated in a se-vere speech and language disorder.Nature 413:519–523

31. Le Couteur A, Bailey A, Goode S, PicklesA, Robertson S, Gottesman I, Rutter M(1996) A broader phenotype of autism:the clinical spectrum in twins. Journalof Child Psychology and Psychiatry37:785–801

32. Manning JT, Baron-Cohen S, Wheel-wright S, Sanders G (2001) The 2nd to4th digit ratio and autism. Develop-mental Medicine and Child Neurology43:160–164

33. McGue M, Bouchard TJ, Jr. (1998) Ge-netic and environmental influences onhuman behavioral differences. AnnualReview of Neuroscience 21:1–24

34. Murphy M, Bolton PF, Pickles A, Fom-bonne E, Piven J, Rutter M (2000) Per-sonality traits of the relatives of autisticprobands. Psychological Medicine 30:1411–1424

35. Nurnberger JI, Jr, Goldin LR, GershonES (1994) Genetics of psychiatric disor-ders. In: Winokur G, Clayton P (eds)The Medical Basis of Psychiatry. WB.Saunders, Philadelphia, pp 459–492

36. O’Brien G, Yule W (1995) BehaviouralPhenotypes. Clinics in DevelopmentalMedicine. Cambridge University Press,Cambridge

37. Online Mendelian Inheritance in ManOMIM (TM) (2002) McKusick-NathansInstitute for Genetic Medicine, JohnsHopkins University, Baltimore MD andNational Center for Biotechnology In-formation. National Library of Medi-cine, Bethesda, MD. World Wide WebURL:http://www.ncbi.nlm.nih.gov/omim/

38. Petronis A, Kennedy JL (1995) Unstablegenes – unstable mind? American Jour-nal of Psychiatry 152:164–172

39. Plomin R, DeFries JC, McClearn GE,Rutter M (2000). Behavioral Genetics.4th edition ed. WH. Freeman and Com-pany, New York

40. Plomin R, Owen MJ, McGuffin P (1994)The genetic basis of complex humanbehaviors. Science 264:1733–1739

41. Richards RI (2001) Dynamic muta-tions: a decade of unstable expandedrepeats in human genetic disease. Hu-man Molecular Genetics 10:2187–2194

42. Risch N (1990) Linkage strategies forgenetically complex traits. I. Multilocusmodels. American Journal of HumanGenetics 46:222–228

43. Risch N, Spiker D, Lotspeich L, Nouri N,Hinds D, Hallmayer J, Kalaydjieva L,McCague P, Dimiceli S, Pitts T, NguyenL, Yang J, Harper C, Thorpe D, VermeerS, Young H, Hebert J, Lin A, Ferguson J,Chiotti C, Wiese-Slater S, Rogers T,Salmon B, Nicholas P, Petersen PB, Pin-gree C, McMahon W, Wong DL, Cavalli-Sforza LL, Kraemer HC, Myers RM(1999) A genomic screen of autism: ev-idence for a multilocus etiology. Amer-ican Journal of Human Genetics 65:493–507

44. Rutter M (2000) Genetic studies ofautism: from the 1970s into the millen-nium. Journal of Abnormal Child Psy-chology 28:3–14

45. Rutter M, Silberg J, O’Connor T, Si-monoff E (1999) Genetics and childpsychiatry: I Advances in quantitativeand molecular genetics. Journal ofChild Psychology and Psychiatry 40:3–18

46. Steyaert J, Borghgraef M, Legius E,Fryns JP (1996) Molecular–intelligencecorrelations in young fragile X maleswith a mild CGG repeat expansion inthe FMR1 gene. American Journal ofMedical Genetics 64:274–277

47. Steyaert J, Legius E, Borghgraef M,Fryns JP (2001) A distinct neurocogni-tive phenotype in female fragile X per-mutation carriers assessed with visualattention tasks. American Journal ofMedical Genetics (In press)

48. Swillen A, Devriendt K, Legius E,Eyskens B, Dumoulin M, Gewillig M,Fryns JP (1997) Intelligence and psy-chosocial adjustment in velocardiofa-cial syndrome: a study of 37 childrenand adolescents with VCFS. Journal ofMedical Genetics 34:453–458

49. Swillen A, Devriendt K, Legius E,Prinzie P, Vogels A, Ghesquiere P, FrynsJP (1999) The behavioural phenotypein velo-cardio-facial syndrome (VCFS):from infancy to adolescence. GeneticCounseling 10:79–88

50. Tapscott SJ, Thornton CA (2001) Bio-medicine. Reconstructing myotonicdystrophy. Science 293:816–817

51. Tassone F, Hagerman RJ, Taylor AK,Gane LW, Godfrey TE, Hagerman PJ(2000) Elevated levels of FMR1 mRNAin carrier males: a new mechanism ofinvolvement in the fragile-X syndrome.American Journal of Human Genetics66:6–15

52. Tassone F, Hagerman RJ, Taylor AK,Mills JB, Harris SW, Gane LW, Hager-man PJ (2000) Clinical involvement andprotein expression in individuals withthe FMR1 permutation.American Jour-nal of Medical Genetics 91:144–152

53. Verkerk AJ, Pieretti M, Sutcliffe JS, FuYH, Kuhl DP, Pizzuti A, Reiner O,Richards S, Victoria MF, Zhang FP,Eussen BE, van Ommen GJ, BlondenLA, Riggins GJ, Chastain JL, Kunst CB,Galjaard H, Caskey CT, Nelson DL,Oostra BA, Warren ST (1991) Identifi-cation of a gene (FMR-1) containing aCGG repeat coincident with a break-point cluster region exhibiting lengthvariation in fragile X syndrome. Cell65:905–914

54. Vincent JB,Yuan QP,Schalling M,Adolf-sson R, Azevedo MH, Macedo A, BauerA,DallaTorre C,Medeiros HM,Pato MT,Pato CN, Bowen T, Guy CA, Owen MJ,O’Donovan MC, Paterson AD, PetronisA, Kennedy JL (2000) Long repeat tractsat SCA8 in major psychosis. AmericanJournal of Medical Genetics 96:873–876

55. Vincent MC, Heitz F, Tricoire J, Bour-rouillou G,Kuhlein E,Rolland M,CalvasP (1999) 22q11 deletion in DGS/VCFSmonozygotic twins with discordantphenotypes. Genetic Counseling 10:43–49

56. Winsberg BG, Comings DE (1999) As-sociation of the dopamine transportergene (DAT1) with poor methyl-phenidate response. Journal of theAmerican Academy of Child and Ado-lescent Psychiatry 38:1474–1477

57. Woolf CM (1995) Influence of stochas-tic events on the phenotypic variationof common white leg markings in theArabian horse: implications for variousgenetic disorders in humans. Journal ofHeredity 86:129–135

58. Yu S, Pritchard M, Kremer E, Lynch M,Nancarrow J, Baker E, Holman K, Mul-ley JC, Warren ST, Schlessinger D,Sutherland GR, Richards RI (1991)Fragile X genotype characterized by anunstable region of DNA. Science 252:1179–1181

psychiatric genetics: the case of single gene disorders

Documents