The molecular basis of genetic dominance

Jf Med Genet 1994;31:89-98

Review article

The molecular basis of genetic dominance

Andrew 0 M Wilkie

AbstractStudies of mutagenesis in many organ-isms indicate that the majority (over90%) of mutations are recessive to wildtype. If recessiveness represents the'default' state, what are the distinguish-ing features that make a minority ofmutations give rise to dominant or semi-dominant characters? This review drawson the rapid expansion in knowledge ofmolecular and cellular biology to classifythe molecular mechanisms of dominantmutation. The categories discussed in-clude (1) reduced gene dosage, expres-sion, or protein activity (haploinsuffi-ciency); (2) increased gene dosage; (3)ectopic or temporally altered mRNAexpression; (4) increased or constitutiveprotein activity; (5) dominant negativeeffects; (6) altered structural proteins; (7)toxic protein alterations; and (8) new

protein functions. This provides a frame-work for understanding the basis ofdom-inant genetic phenomena in humans andother organisms.

(J Med Genet 1994;31:89-98)

Institute of MedicalGenetics, UniversityHospital of Wales,Heath Park, CardiffCF4 4XW, UKA 0 M Wilkie

Correspondence toDr Wilkie, Institute ofMolecular Medicine, JohnRadcliffe Hospital,Headington, Oxford OX39DU, UK.

The concepts of dominance and recessiveness(or recessivity), originally formulated by Men-del,' are so fundamental to genetics that theyare often taken for granted. But why are some

diseases dominant and others recessive? Thisquestion is frequently ignored in textbooks ofgenetics, and it is surprisingly difficult to findmuch written on the subject. With the rapidaccumulation of molecular data on diploidorganisms as diverse as yeasts and humans,unifying themes are beginning to emerge.2This review attempts to classify and elaboratethese ideas, and collates some of the more

useful references. I will first outline some

definitions and concepts, and then giveillustrative examples of different molecularmechanisms of dominance. Although I havefocused on human disorders where possible,many additional lessons can be learned fromthe study of non-human systems.

Dominance, semidominance, andrecessivenessIt should first be remembered that dominanceis not an intrinsic property of a gene or mutantallele, but describes the relationship between

the phenotypes of three genotypes: an allelemay behave as a dominant, semidominant, orrecessive depending both on its partner allele,and the character under consideration.3 Con-sider alleles A and B, with genotypes AA, AB,and BB. If a particular phenotypic character isobserved in the AA and AB genotypes, butdiffers from BB, then allele A is dominant toallele B. When the AB phenotype is inter-mediate between or combines characters fromboth the AA and BB phenotypes, alleles A andB are semi- or codominant. Most wild typealleles are dominant over other alleles, as thewild type and heterozygote phenotypes areusually indistinguishable; thus most geneticdiseases are recessive (fig 1).A potential source of confusion when consi-

dering dominance phenomena in human gen-etic disease, is that only the wild type andheterozygous mutant phenotypes are generallyencountered. Examples of homozygous mu-tants both for relatively common disorders(thalassaemia, familial hypercholesterolaemia)and rarer conditions (achondroplasia, pie-baldism) indicate that the phenotype of thehomozygote usually tends to be more severethan the heterozygote, hence the wild type andmutant alleles are, strictly speaking, semidom-inant.6 The Huntington's disease mutationprovides an unusual instance of a mutant allelethat is truly dominant to wild type in thathomozygotes appear no more severely affectedthan heterozygotes7-9 (fig 1). Although it isinteresting to speculate on the differences inmechanism giving rise to semidominance andcomplete dominance, there are insufficientmolecular data to attempt a synthesis. Themore simple, but perhaps more fundamental,question addressed in this review may be sum-marised as follows: what aspects of a mutantallele'sfunction cause it to affect the phenotype inthe presence of a wild type allele? For simplicityI will use the term 'dominant mutation' todescribe a mutant allele in this context.

Dominant mutations are much rarerthan recessive onesAlthough dominant disorders outnumber re-cessives by a ratio of nearly 4:1 in McKusick's1992 compilation,'0 ascertainment in thehuman is undoubtedly biased in favour of milddominantly inherited phenotypes. By contrast,it has long been known from systematic muta-genesis of a variety of diploid organisms thatthe majority of mutations are recessive to wild

89

Wilkie

GenotypeAA AB BB

Descri ptiorn Examples

A is dominant to B Recessive 1isetnesC,r

° ___ A a rl di B a se r-incm o rn n a ri I 3-tih a a ssa na aC' :farmlial hvyerchoestwo ann-

B 's d orniT riai A: Rare: HUntngtongA s iseasebe an exampie

_ 1 1 ? 0BB pher otue: irkiowr Most sn callec: o;'ar-cdiseases

Figure 1 Relationship between genotype and clinical phenotype. A is a wild typeallele and B a mutant allele; different phenotypes are represented by different shadedblocks.

type. " For example, insertional inactivation byrandom integration of retroviral DNA into themouse genome produces recessive and domin-ant phenotypes with a ratio of about 10-20:1.12 13The search for an explanation of the reces-

sive behaviour of most mutations generated alively debate in the 1930s between SewallWright, who believed that it arose intrinsicallyfrom the physiology of gene action, and RAFisher, who proposed that the accumulation ofmodifier alleles at other loci was responsible.Fisher's theory has now generally lost favour,and Orr'4 showed that in the alga Chlamydomo-nas, which is usually haploid (so that Fisherianselection cannot apply), most mutations never-theless showed recessive behaviour when ex-amined in a transiently diploid background,supporting Wright's theory. Indeed, diploidymay have evolved because it protects againstrecessive mutations. '5'7 Thus it is dominance,rather than recessiveness, that demands specialexplanation; but why should the 'default' stateof mutations be recessive?The usual explanation is as follows. The

most likely effects of a random gene mutationare that it will either be neutral (normal pheno-type) or inactivating. If the latter, the questionis whether the inactivation would be clinicallymanifest in the heterozygote (dominance orsemidominance, specifically, haploinsuffi-

Muller's classification

ciency), or only in the homozygote (recessive-ness). In 1981 Kacser and Burns'8 proposed atheory of metabolic fluxes to explain why mostinborn errors of metabolism are recessive.Assuming that a metabolic pathway has manynon-rate limiting steps, control of flux at anyparticular point in a pathway will be small.Hence, many pathways show a saturable rela-tionship between enzyme level and metabolicflux, with fluxes fully saturated at wild typeenzyme level; a 50% reduction in enzymeactivity would therefore cause little reductionin flux below its saturation level.Although this theory fits metabolic path-

ways well, it is not applicable to critical ratelimiting steps of such pathways, nor to muta-tions causing qualitatively altered function,especially when structural or controlling/sig-nalling proteins are involved. It is perhaps notsurprising that most dominant mutants belongto one of these latter categories, and frequentlyinvolve developmental malformations.An additional explanation for the rarity of

dominant mutations is suggested by work onthe nematode Caenorhabditis elegans. Reces-sive mutations at a series of loci termed smgmay alter the behaviour of mutations at otherloci from recessive to dominant (cryptic dom-inance). It seems that the wild type smg lociencode proteins that can recognise and select-ively degrade many mutant mRNA species,forming part of a mutant surveillance sys-tem.'920 The relevance of this finding tohumans is not yet clear.

Finally, note that although the number ofknown recessive conditions in the human mayconsiderably underestimate the total, the truefigure is unlikely to approach the total numberof genes. There is a growing list of murinegenes for which targeted disruption is notassociated with any phenotypic abnormality intransgenic mice.21 A similar situation applies tothe mutational spectrum in C elegans, and it isnoteworthy that dominant "gain of function"mutants exist at several loci for which thehomozygous null phenotype is entirely nor-mal.22

Molecular classification

Haploinsufficiency

+ Gene dosage

+/ Ectopic mRNA expressi

+/Constitutive protein acti

i Dominant negative

dStructural function

i Toxic protein

i New protein

Figure 2 Relationship between genetic and molecular mechanisms of dominance.genetic classification is that formulated by Muller23; the relationship between wild(A) and mutant (B) alleles is indicated diagrammatically. Thick lines join categ,that commonly show equivalence; dashed lines connect less frequent groupings.

Types of dominant mutationI In 1932, Muller23 suggested a classification of

dominant mutations that is still widely quoted.I He coined the terms amorph, hypomorph, and

hypermorph to reflect quantitative changes to aZ pre-existing wild type character; antimorph to

on I describe mutual antagonistic interaction with-j wild type; and neomorph for a new phenotype,

not fully antagonised by wild type. His propo-I sal, made when the molecular nature of muta-

tion was still uncertain (and predating theidentification of DNA as the genetic materialby 12 years),24 was remarkable for its pres-cience. Unfortunately, later authors have

I sometimes tended to assume a one to onerelationship between this classification, based

I on classical genetics, and underlying molecularmechanism. While clear parallels exist, these

The are inexact (fig 2). As this review focuses ontype molecular mechanisms of dominance, I haveaormes

avoided using Muller's terms to highlight the

90


distinction between the genetic and molecularlevels of analysis. The following classificationseems to accommodate most situations, al-though some ambiguities and overlaps are

inevitable.(1) Reduced gene dosage, expression, or pro-tein activity: haploinsufficiency.(2) Increased gene dosage.(3) Ectopic or temporally altered mRNAexpression.(4) Increased or constitutive protein activity.(5) Dominant negative effects.(6) Altered structural proteins.(7) Toxic protein alterations, not covered inother categories.(8) New protein functions.Some examples of these mechanisms are

shown in the table and are further discussedbelow. A general distinction can be madebetween category (1), which involves loss offunction, and categories (2) to (8), which rep-resent gain of function. Note that the latter,frequently used term encompasses a widerange of mechanisms, and is thus only appli-cable in a broad context. The table includestwo other mechanisms that may give rise to a

"dominant" pattern of inheritance, but inwhich the inherited mutation is not dominantat a cellular level. These involve recessive anti-oncogenes and genomic imprinting, and are

discussed briefly in a later section.

REDUCED GENE DOSAGE, EXPRESSION, ORPROTEIN ACTIVITY: HAPLOINSUFFICIENCYFor the minority of cases in which the abnor-mal phenotype results from inactivation of oneof a pair of alleles, the term "haploinsuffi-ciency" is used ("haplolethality" if earlyembryonic death occurs). Haploinsufficientloci are relatively unusual: a careful survey ofthe Drosophila genome showed only 56 lociassociated with an altered phenotype whenpresent as a single copy, of which four were

lethal.25 However, such loci are more import-

ant than their rarity might suggest, for tworeasons. First, mutation may arise from anymechanism producing loss of function: dele-tion, chromosome translocation, truncationcaused by nonsense and frameshift mutation,and some promoter and splice site mutationsand amino acid substitutions may all be re-

sponsible. Such variety will tend to increasethe frequency with which the disease isobserved. Second, dosage sensitive genes seem

to be an intrinsically interesting group.26Genes showing haploinsufficiency fall into

two broad categories. A few code for tissuespecific proteins synthesised in large quantit-ies, for instance, type 1 collagen27 (but see alsothe section on structural mutations), globins,low density lipoprotein receptor,28 haem syn-thesis (porphyrias),2 and Cl esterase inhibitor(hereditary angio-oedema).29 In the first twocases, the abnormal heterozygous phenotypemay be because of the resulting imbalance witha matched component protein; in the latterthree, because of interference with a rate limit-ing step of a metabolic pathway. Of particularnote, levels of Cl esterase inhibitor associatedwith heterozygous deficiency are only 15 to20% of normal, even during remission. This isbecause the normal inhibitor is "mopped up"relatively rapidly by complexing with plasmaenzymes, and the rate at which this occurs islargely independent of inhibitor concentration(zero order kinetics).29 The quantitative defi-ciency is hence greater than the expected valueof 50%.A second category comprises regulatory

genes working close to a threshold level fordifferent actions. Examples in humans includePAX3 (Waardenburg syndrome),303' PAX6(aniridia),32 GLI3 (Greig cephalopolysyndactyly,GCPS),33 34 WT1 (Wilms's tumour/genito-urinary abnormalities),35 36 RD S/peripherin(retinitis pigmentosa),37 and KIT (pie-baldism).38 Such threshold dosage effects maybe clinically manifest in only a subset of thetissues in which the gene is expressed (aniridia

Major categories and mechanisms of genetic dominance, with the types of mutation commonly responsible. See textfor further examples and references. D = large deletion, T truncation (nonsense or frameshift mutation),M= missense mutation or small in frame deletion, S= splice site mutation, P= promoter mutation, Tr = translocationor other rearrangement, Dup = duplication, A = amplification, () = inconsistent association.

Category of mutation Mechanism Types of mutation Examples

Loss offunctionHaploinsufficiency Subunit imbalance D, T, S, (M) a and ( globins

Metabolic rate determining step D, T, S, (M) LDL receptorDevelopmental regulator D. T, S, (M), (Tr) PAX3, PAX6

Gain offunctiontGene dosage Duplication Dup PMP-22

Amplification A MDM2

T/Ectopic mRNA expression Altered temporal pattern P,Tr,(D) y globin, MYCAltered tissue distribution P, Tr Ubx, Antp, MYCTmRNA stability D lin-14

T/Constitutive protein TStability (PEST deletion) T CLN3, glp-lactivity Constitutive activation M RAS, Gso, SCN4ADominant negative Disruption of dimer M, (T) KIT, p53

Competition for substrate M, (T) RASStructural protein Disruption of structure M, S, (T) Collagen, fibrillinToxic protein Disruptive interaction M Rhodopsin, amyloidosesNew protein Altered substrate specificity M a, antitrypsin

Exon shuffling Tr BCR/ABLOther mechanisms

Recessive antioncogene - RetinoblastomaGenomic imprinting - Beckwith-Wiedemann syndrome

91

Wilkie

and GCPS are examples) and the phenotypemay be sensitive to the genetic background.To understand the mechanisms of dosage

sensitivity in this "regulatory" group requires adetailed knowledge of the molecular interac-tions involved, something not yet achieved forany human gene. However, simpler organismsprovide some excellent model systems. Forinstance, sex determination in Drosophila re-quires the ability to distinguish between X:au-tosomal ratios of 1 in females and 0 5 in males.This may be achieved by titration of "numer-ator" X chromosome genes against "denomina-tor" autosomal ones, possibly by competitionof the cognate proteins for binding to a regula-tory DNA sequence.'9 Further insight may begained by studies of morphogenic proteins, forexample the Drosophila transcription factordorsal (dl), which is distributed in a nuclearconcentration gradient along the dorsoventralaxis of the early embryo. Dosage dependentactivation of different sets of downstreamgenes by dl correlates with the strength of thedl binding sites in their promoters: genes withhigh affinity dl binding are activated orrepressed by dl at lower threshold levels.4'Correspondingly, female flies heterozygous fordl null mutations produce abnormal embryosthat fail to develop mesoderm, which requiresthe highest level of dl activity.42

INCREASED GENE DOSAGEApplication of Kacser and Burns' principles'8predicts that an increase in gene dosage tothree copies should affect the phenotype evenless often than a reduction to one copy. Experi-mental analysis supports this: for example, thesurvey of aneuploidy in Drosophila previouslymentioned25 identified only one triplo-lethaland one triplo-abnormal locus. Nevertheless,cytogenetically visible trisomy in humans(which will usually encompass at least 40 to 50genes) is usually associated with phenotypicabnormality, indicating that a significantminority of loci must be sensitive to 3 versus 2dosage. It may be relevant that the increase indosage at the mRNA and protein level canexceed the expected factor of 1-5; considerablygreater rises are observed for some geneson chromosome 21 in Down's syndrome.4'Although the distinctive phenotypes associ-ated with certain trisomies may therefore beattributable to a small number of critical genes,few of these have been specifically identified.An exception is PMP-22, duplication of whichis likely to be the principal cause of type ICharcot-Marie-Tooth disease.44 The PMP-22region is also haploinsufficient, giving thedifferent phenotype of dominant pressurepalsies45; however, the cellular mechanismsof these contrasting dosage effects are notunderstood.Gene amplification in somatic cells to much

higher copy numbers frequently occurs in cer-tain neoplasias.46 A particularly clear exampleof how this causes a dominant phenotype isprovided by the amplification of the MDM2gene in sarcomas. MDM2 protein binds to andinactivates the tumour suppressor gene P53

(discussed further below), leading to escapefrom normal p53 regulated cellular growthcontrol.47

ECTOPIC OR TEMPORALLY ALTERED mRNAEXPRESSIONThis group is characterised by disturbance ofthe exquisite controls of mRNA expressionthat dictate the normal cellular distribution,temporal restriction, and absolute levels ofmRNA. In principle, altered gene expressioncan arise in any gene or message that contains aregulatory domain, and the molecular patho-logy of such mutants is correspondingly di-verse.48A fairly specific illustration of loss of tem-

poral regulation is provided by hereditary per-sistence of fetal haemoglobin (HPFH). Knowncauses include point mutation of the y globinpromoter, which alters binding of the eryth-roid transcription factor GATA-1,49 certain 3'deletions encompassing the 6 and P globingenes,50 and alterations of unidentified transacting factors. The effect of all these mutationsis to abrogate the normal switch from expres-sion of y to 6 and 1 globin, which occursaround the time of birth. The resulting HPFHdominantly ameliorates the effects of 1 thalas-saemia mutations.An example of ectopic expression is pro-

vided by the contrabithorax (Cbx) mutationsof Drosophila, which involve the ultrabithorax(Ubx) gene, normally expressed in the poster-ior part of the embryo with an anterior bound-ary in the third thoracic segment (T3). In Cbxmutants, which comprise insertions, inver-sions, and other chromosomal rearrange-ments,5' Ubx is also expressed in T2 and thisresults in the homeotic transformation of T2into a T3 like structure. Similarly, dominanthomeotic mutations of the Antennapedia geneoccur because of ectopic expression: in onecase studied in detail (Antp73b), a chromosomalinversion results in the entire Antp codingregion being placed under a new promoter.52More commonly, the disease phenotype may

reflect a combination of alterations in the tem-poral specificity, tissue distribution, and abso-lute level of mRNA expression. The primaryabnormality usually lies at the level of tran-scription, but sometimes mRNA processingmay be affected. Examples of transcriptionalalterations include the following. Chromoso-mal translocations resulting from errors inrecombinase mediated gene rearrangement inlymphoid cells activate expression of tran-scription factors like MYC, causing B and Tcell neoplasms.5'54 Promoter mutations in theCaenorhabditis sex determining gene her-1(the only member of this pathway subject totranscriptional control) increase expressionlevels and result in partial transformation ofXX worms into phenotypic males.55 Increased,ectopic expression of a chimeric mRNAencoding a normal protein accounts for thelethal yellow mutant at the mouse agoutilocus.5657

Control of expression at the level ofmRNA

92


processing is illustrated by the heterochronic(defining developmental time) C elegans genelin-14. Dominant mutants, which cause the re-expression of early cell lineages at later de-velopmental stages, delete the 3' untranslatedregion (UTR) of the mRNA and lead to raisedprotein levels. This 3'UTR may regulateexport of the transcript from the nucleus,transcript stability, or translation.48 Splice sitemutations of mRNA subject to differentialsplicing will alter the pattern of mature mRNAisoforms: this is observed at the WT1locus.35 58 59

INCREASED OR CONSTITUTIVE PROTEIN ACTIVITYAt the protein level, increased activity may becaused by increased half life or by loss ofnormal inhibitory regulation (constitutiveactivity). One class of mutations conferringincreased half life are those occurring in PESTsequences (rich in proline, glutamic acid, ser-ine, and threonine),6' which act as recognitionsignals for proteolytic degradation: loss ofthese sequences by C-terminal truncation sta-bilises the protein. Examples of PEST dele-tions include mutations of the CLN3 gene ofSchizosaccharomyces pombe (WHI-l/DAF-1cell cycle mutants)6' and the glp-1 gene of Celegans.62 glp-1is required for induction ofgermline proliferation and embryogenesis, andthe glp-l(q35) point mutation is particularlyinstructive, as it causes both semidominant(multivulva) and recessive (sterility/embryonic

A let-60 ras

*)

RecessiveConstitutive activationDominant negative

GTP/GDP binding

B Toll

f f + 4

TransmembraneC KIT

EI aeuar *)(*)Extracellular

(Oligomerisation)

D P53

Transmembrane Tyrosinekinase

+ +# +* ++ +* * *

Mutational hotspots Oligomerisati

Recessive'Class lI' dominantConstitutive activation

HaploinsufficientDominant negative

RecessiveDominant negative

on

Figure 3 Diagrammatic representation offour gene products, illustrating the complexrelationship of point mutation to phenotype. The N-terminus is on the left. *indicates a

missense alteration, I a frameshift or nonsense mutation leading to truncation. (A)let-60 ras.6465 Additional mutations documented in human tumours63 are denoted (*).(B) Toll.66 (C) KIT.38 Truncations of the tyrosine kinase domain may have mixedhaploinsufficient and dominant negative effects. Murine W mutations3 are shown inbrackets. (D) P53.67-9 Clusters of missense mutations are represented by black boxes.Four specific missense mutations tested in vitro for presence or absence of dominantnegative effects69 are shown individually.

lethality) phenotypes. The former is attribu-table to stabilisation of the truncated proteinowing to the PEST deletion, while the lattermay result from counteracting destabilisationof the mutant mRNA.62A paradigmatic example of constitutive pro-

tein activation is provided by the RAS genes:oncogenic point mutations prevent GTP hy-drolysis, thus maintaining the protein in anactivated state6'65 (fig 3A). Similarly, activat-ing missense mutations at the 201Arg residueof the Gsot protein (which stimulates adenylylcyclase) have been documented (as somaticmosaics) in five cases of McCune-Albrightsyndrome.707' Different point mutations in theadult skeletal muscle sodium channel a subu-nit gene SCN4A cause hyperkalaemic periodicparalysis7273 and paramyotonia congenita,74 byinterfering with normal voltage sensitive inac-tivation of the sodium current. In view of thediffering effects of single and triple dosage ofthe PMP-22 gene described above, it is inter-esting that the phenotype associated with aheterozygous missense mutation (1 6Leu -.Pro) resembles that of triple dosage, that is,Charcot-Marie-Tooth disease.75 This suggeststhat the missense mutation may increase PMP-22 activity, but this has not yet been shown.A particularly complex spectrum of muta-

tions is encountered at the Drosophila locusNotch, which encodes a transmembranereceptor protein that transduces a variety ofcellular signals, and includes an extracellulardomain rich in epidermal growth factor (EGF)like repeats. Increased intrinsic activity is as-sociated with some of the so called Abruptexmutations: these are missense, clustered in theEGF domain, and are thought to perturb thenormal balance of homo- and heterodimericprotein interactions.7677 Heterozygous null(loss of function) mutations of Notch give adifferent phenotype, and yet other mutantsexist that have recessive or dominant negativeeffects76 77 (see below). Another interestingDrosophila locus is Toll. This encodes a trans-membrane protein that provides an unusualexample of two distinct activation mechanisms(fig 3B).66 "Class I" mutants are missense andact constitutively, possibly owing to directstructural modulation of the protein. "ClassII" mutants are truncations that retain theextracellular component: this activates wildtype Toll by an undefined mechanism. Mu-tants in the class II group differ geneticallyfrom other categories of active protein mutantsin that they are non-functional when heterozy-gous to a null.66 Such truncation mutants morecommonly cause dominant negative effects, asdescribed below.

DOMINANT NEGATIVE MUTATIONSIn the heterozygous state these mutants anta-gonise the activity of the remaining wild typeallele, giving a phenotype approaching a null;when homozygous, or heterozygous to a nullmutation, they are non-functional. Hersko-witz78 drew attention to the value of thesemutations in experimental studies and pro-posed a classification. The major group com-

93

Wilkie

prises multimeric proteins dependent on oligo-merisation for activity: the presence in a mul-timer of a mutant subunit with intact bindingbut altered catalytic domains can abrogate thefunction of the entire multimer. For example,if the protein normally dimerises, admixture ofequal numbers of normal and mutant subunitswill result in only 25% normal dimers, poten-tially causing a 75% reduction in activity. Formonomeric proteins, dominant negative muta-tions could occur if substrate was limiting: amutant able to bind the substrate, but notmetabolise it, would have this effect. Muta-tions of polymeric structural proteins, some-times classed as dominant negative, are dis-cussed separately (next section).Dominant negative effects have been de-

scribed in many types of protein with signal-ling or transcriptional functions. A specificexample is provided by the DNA bindingactivity of Drosophila dorsal, mentioned pre-viously, which depends on dimerisation: mostmutations are true recessives, but one particu-lar mutant exerts a dominant negative effect.This is an Arg-*Cys substitution that maps tothe DNA binding domain but does not affectoligomerisation: it appears to act by abolishingthe DNA binding of normal/mutant hetero-dimers.40 Similarly the more severe phenotypeassociated with WT1 mutations in Denys-Drash syndrome, as compared with Wilms'stumour/genitourinary abnormalities, may beexplained by the dominant negative behaviourof specific zinc finger mutations in the formercondition: it is not yet certain whether this ismediated by WT1 dimers.3536 Specific Abrup-tex missense mutations at the DrosophilaNotch locus are dominant negative, as men-tioned above.767A wider variety of mutations may cause

dominant negative effects in the KIT proto-oncogene, manifested as white spotting (W) inthe mouse and piebaldism in the human. KITencodes a receptor tyrosine kinase and dimeri-sation, which occurs in response to ligandbinding, is essential for activity. Whereas thepiebald phenotype associated with completeKIT deletion is relatively mild in the hetero-zygote (haploinsufficiency), point mutationsinvolving the intracellular tyrosine kinase do-main cause severe disease338 (fig 3C). Trunca-tions in the same domain tend to have a vari-able, intermediate phenotype: although partlythe result of haploinsufficiency, a dominantnegative effect is probably also contributing, asseen in an analogous truncation of the fibro-blast growth factor receptor (FGF-R).79 Thereovirus cell attachment protein provides afurther example.80Dominant negative effects may be very im-

portant in neoplasia, a paradigm being thetumour suppressor P53: a wide variety ofacquired mutations has been described, themany missense mutants being concentrated infour clusters6768 (fig 3D). p53 oligomerises invitro and can adopt two conformations, oneactive and the other inactive; wild type proteinis normally in the active state. Cotranslationwith certain missense mutants results in mixedoligomers that adopt the inactive conforma-

tion.6" Thus, although P53 is conventionallyviewed as a "recessive" tumour suppressorgene, some mutants can deregulate p53 func-tion in a dominant negative fashion. In con-trast, no alteration in wild type activity isinduced by a missense mutant associated withthe Li-Fraumeni syndrome, suggesting thatLi-Fraumeni p53 mutants may be relatively"weak" ones.9 Note that the p53 oligomerisa-tion domain lies at the extreme C-terminus (fig3D); prematurely truncated forms cannot bindwild type and therefore do not act in a domin-ant negative fashion.A possible example of Herskowitz's second

class of dominant negative effect, involving amonomeric protein, is provided by certainpoint mutations in the RAS gene (fig 3A).6365 Amutant protein able to bind the guanine nuc-leotide exchange factor, but not be activated byit, will deplete the pool of this limiting factoravailable for activation of normal RAS.

Intriguingly, the dominant negative prin-ciple seems to have been exploited by certainnaturally occurring regulatory systems, andrepresentatives of both Herskowitz's classesare known. Belonging to the first class is thenegative regulation, by formation of inactiveheterodimers, of the transcription factorsMyoD and c-Jun by the proteins Id and JunBrespectively. Id is a truncated helix loop helixprotein that forms dimers with MyoD, butlacks the adjacent basic region required forDNA binding.8' Similarly, critical amino acidsubstitutions in JunB abolish its homodimeri-sation and DNA binding, but favour forma-tion of inactive JunB/c-Jun heterodimers.82 InHerskowitz's second class is the interferonactivator IRF1 and its antagonist IRF2; IRF2has enhanced DNA binding and displacesIRF1 from the interferon promoter, but isonly weakly activating.83

ALTERED STRUCTURAL PROTEINSAt a simplistic level it is easy to understandwhy mutations of structural proteins are fre-quently dominant: admixture of normal andabnormal structural components will disruptthe integrity of the overall structure on a"weak links in a chain" principle. Careful cellbiochemical analysis shows a more complexpicture: additional modulators of the abnormalphenotype will include mRNA stability, andthe degree of abnormality in cellular process-ing, secretion, and extracellular incorporationinto mature fibrils, of the nascent protein.Mutations of type I collagen in osteogenesisimperfecta (OI)8485 and of fibrillin in Marfansyndrome" 87 provide the best studied ex-amples; in OI, there is a reasonable correlationbetween predicted disruption of structurecaused by point mutations and exon skips andseverity of phenotype.8485 By contrast, loss offunction mutations have milder effects (seesection on haploinsufficiency).27 Other struc-tural proteins showing dominant mutationsinclude myosin heavy chains (unc-54(d) muta-tions in C elegans88 and hypertrophic cardio-myopathy in humans89), and keratins 5 and 14(Dowling-Meara epidermolysis bullosa).90

94


TOXIC PROTEIN ALTERATIONSThe common thread to these mutations, whichare usually missense, is that they cause struc-tural alterations in mono- or oligomeric pro-teins. These disrupt normal function and leadto toxic precursors or waste products from sidereactions that poison the cell. Dominant nega-tive effects are excluded. There are clear paral-lels with the mutational mechanisms in truestructural proteins, but the phenotypic effectsof toxic mutants are more unpredictable.One commonplace example is the sickle

mutation (haemoglobin S, P6Glu--Val).(Although this shows largely recessive be-haviour, coinheritance of a second mutationin cis (haemoglobin S Antilles, P23Val-+Ile)causes sickling to manifest in the heterozy-gote9'). Other examples include missense mu-tations of C elegans degenerins, mec-4 anddeg-1, which cause specific neuronal cells toswell up, vacuolate, and lyse92; a point muta-tion in mouse tyrosinase-related protein-I(Light mutant) that disrupts melanosomestructure93; and various point mutations inrhodopsin associated with slow degenerationof rod photoreceptor outer segments.37 A par-

ticularly striking group comprises the domi-nantly inherited hereditary amyloidoses, a di-verse collection of diseases associated withalterations in the structure of soluble proteinsthat increase stability of the protein andpredispose to multimerisation. Proteins impli-cated include transthyretin, ,B amyloid precur-sor protein, gelsolin, cystatin C, prion protein,apolipoprotein AI, lysozyme, and fibrino-gen. 94-96

NEW PROTEIN FUNCTIONSThe creation of new, advantageous proteinfunctions by mutation is the life blood ofevolution, but occurs over a protracted timescale. Proteins with truly new functions are

only rarely encountered in natural human mu-tation and are usually pathological. Two cat-egories may be recognised; missense mutationswith specific functional effects, and assortativeshuffling of exons. In protein engineering,which seeks to accelerate the evolutionary pro-cess and develop proteins with new functions,the same principles apply in the design of newmutants.97The serine protease inhibitors (serpins),

popular targets for protein engineering, pro-

vide perhaps the best natural example involv-ing a missense mutation. A (358Met-+Arg)substitution in a l antitrypsin converts its ac-

tivity to antithrombin, by altering the specifi-city of the active site.98 Another example, iden-tified only in vitro, is a missense mutantprotein (mev) that facilitates cellular uptake ofmevalonate99; the wild type protein lacks thisactivity, but its normal function is unknown.The juxtaposition of domains from different

proteins to generate potentially new functionsis best illustrated by the chimeric fusion pro-

teins produced by some oncogenic chromo-some translocations.5354 The c-ABL/BCRfusion products in the 9;22 Philadelphia trans-location provide the most well characterised

example, distinct chimeric fusion proteinsbeing associated with chronic myeloid andacute lymphatic leukaemia.100 These proteinshave a higher tyrosine kinase activity thannormal c-ABL, and may also differ in sub-strate specificity. The PAX3 gene providesanother example. Haploinsufficiency causesWaardenburg syndrome (see above), buttranslocation to a specific region of chromo-some 13 is associated with alveolar rhabdo-myosarcoma.'0'

OTHER MECHANISMS OF DOMINANCEIn this section are summarised briefly a varietyof other more obscure, but nevertheless inter-esting mechanisms of dominance acting at acellular level.

Position effect variegation in Drosophila is thevariable reduction in expression of a genejuxtaposed to heterochromatin by chromo-some rearrangement. Variegating mutationsare generally recessive in that they reduceexpression only from the rearranged (cis) chro-mosome. The brown locus is unusual in thatexpression is also reduced from the normal(trans) allele. This dominant effect seems todepend on somatic pairing between the homo-logous chromosomes, but the mechanisms ofthis and other 'trans sensing' effects are stilluncertain. 102 103The phenomenon of nucleolar dominance in

wheat reflects the relative expression of tan-dem ribosomal DNA from allelic loci. Expres-sion at an individual locus correlates with thenumber of upstream regulatory sequences.These appear to compete for binding to limit-ing amounts of an activating protein, so thatthe more repeats present, the greater the likeli-hood of activation.'04

Segregation distortion loci subvert the nor-mal pattern of 1:1 gametic segregation, leadingto meiotic drive. This may occur either atmeiosis, when some property of the generalstructure or size of a chromosome gives it areplication advantage on the spindle (chromo-somal drive), or postmeiotically, when directcompetition between the gametes occurs(genic drive).'05 This may allow disadvanta-geous mutations to spread through the popula-tion, by virtue of close linkage to the drivelocus. A well known example is the t complexof mouse.

Unlinked non-complementation occurs whenheterozygous mutations occur at two genescoding for interacting proteins. Whereas theheterozygous state for either locus on its own issilent, concurrent mutations at both loci causethe phenotypic threshold to be exceeded, andthe disease becomes manifest. Examples in-clude the interaction of a and 3 tubulin muta-tions in Drosophila'06 and, more speculatively,the enhanced severity of dystrophin mutationsin trans to an abnormal allele for autosomalrecessive Fukuyama congenital muscular dys-trophy. 107An allied phenomenon, called negative com-

plementation or metabolic interference, occurswhen two alleles at the same locus interact togive a more severe phenotype in the compound

95

Wilkie

heterozygote than in either homozygote. Forexample, Abruptex (Abx) mutations of theDrosophila Notch gene fall into two genetictypes, "enhancers" and "suppressors" ofNotch. Homozygotes for either type are viable(characterised by gapping of the wingveins), yet compound enhancer/suppressor Abxheterozygotes are lethal.7677 Metabolic interfer-ence may theoretically result in various patternsof phenotypic segregation'08 and has beeninvoked to explain the atypical inheritance ofseveral human genetic diseases; none has yetbeen corroborated at the molecular level.

DOMINANT INHERITANCE, WITHOUTDOMINANCE AT A CELLULAR LEVELAlthough the vertical transmission of an ab-normal character is usually assumed to implydominance of the mutation at the cellular level,this is not always the case. In humans, twoexceptions are sufficiently important to havebeen included in the table: recessive antionco-genes and imprinted loci.

Retinoblastoma provides the paradigmaticexample of a phenotype that segregates in adominant pattern, yet is the result of a muta-tion (in the RB1 gene) that is recessive at acellular level. Cells carrying a heterozygousRB 1 mutation are entirely normal, but a"second hit" somatic mutation of the normalallele in at least one retinal cell (a relativelylikely event) causes retinoblastoma.'09110 Ana-logous putative "antioncogenes" or "tumoursuppressors" have been cloned in several otherdominantly inherited cancer syndromes,including Li-Fraumeni syndrome (P53), neuro-fibromatosis types 1 and 2, familial adenoma-tous polyposis (APC), and Von Hippel-Lin-dau disease. At the cellular level, evidence for apurely recessive mechanism of gene action is,however, less certain than with RB 1, and vary-ing contributions from haploinsufficient anddominant negative effects are possible, as dis-cussed for P53 and APC.Genomic imprinting may give rise to a com-

plex pattern of dominant inheritance. If a geneis transcribed only from the chromosome ori-ginating from one of the two parents, the locusis effectively hemizygous. Mutation of theallele on the 'active' chromosome will com-pletely inactivate the locus, whereas mutationof the allele on the other chromosome will haveno phenotypic effect. Apparent dominanttransmission of the disorder can occur, but thiswill show dependence on the sex of the trans-mitting parent. Representative pedigrees areprovided by transgenic mutation of the mouseinsulin-like growth factor-II gene,"1' and inthe human diseases Beckwith-Wiedemannsyndrome"2 and hereditary paraganglioma."3

Perspectives on human genetic diseaseAlthough this classification may initially ap-pear to be an academic exercise, appreciationof these various mechanisms is helpful forthinking about disease processes. For example,perusal of the table and fig 3 indicates that a

different mutational spectrum may be antici-pated in different diseases, according to theircellular mechanism. A wide variety of muta-tions cause loss of function: disease genes witha high mutation rate will often be haploinsuffi-cient and be involved in regulatory pathwaysor act as tumour suppressors or both. A searchfor constitutional chromosomal abnormalities(deletions, translocations), which provide suchan invaluable resource for disease location andpositional cloning,' '4 is much more likely to besuccessful in this group than in the "gain offunction" categories. By contrast, acquiredchromosomal abnormalities in neoplasia mayoften pinpoint specific oncogenes involved in"gain of function" transformation. The pheno-type associated with missense mutations willusually be critically dependent on their exactposition and nature, except in structural pro-teins; hence multiple, independent point mu-tations as a cause of dominant disease are mostcommonly encountered in such proteins.

In understanding mechanisms of cancer, thedominant negative effects illustrated for p53may occur in other tumour suppressor genes.For instance, germline mutations of the APCgene cause familial adenomatous polyposis/Gardner's syndrome, and somatic mutationsoccur in sporadic colon cancer. The amino acidsequence of APC predicts that it will formcoiled coils, structural elements that permitoligomerisation. 15 116 The majority of APCmutants, both germline and somatic, are mis-sense17 118 and some could disrupt normal oli-gomers to give dominant negative effects.Analysis of the particular mutations presentmay therefore guide prognosis.The mechanisms of dominance in con-

ditions associated with unstable triplet repeats(for example, fragile X syndrome, myotonicdystrophy, and Huntington's disease) are notyet clear, and probably heterogeneous, witheffects owing to alterations in both mRNAexpression and protein function. Although the(CGG)n expansion in the fragile X syndrome isassociated with DNA methylation and absenceofFMR- 1 gene expression,' 9 in myotonic dys-trophy, DMK alleles containing (CTG)nexpansions may actually be overexpressed'20(although this is disputed'2' 122). Other poten-tial variables are whether the expanded tripletlies in the coding or non-coding region of theprotein, and the sequence of the repeat itself.'23Complete elucidation of the mechanisms ofdominance associated with triplet repeatexpansion may well yield some surprises.

Finally, an understanding of the molecularmechanism of a disease is a prerequisite forattempting gene therapy. Nearly all diseasescurrently targeted for gene therapy are reces-sive,'24 in which the goal is simply to replacethe missing product. It should be evident thatmost categories of dominant disease pose aformidable challenge to gene therapy, butalready the "molecular engineers" are contem-plating strategies to overcome these problems.Examples include antisense RNA therapy toantagonise selectively the action of dominantnegative mutants; or conversely, the introduc-tion of such mutants to counteract the effects

96


of increased mRNA expression or protein ac-tivity.

The idea for this review originated from a meeting of theGenetical Society ("Dominance and recessiveness revisited",Edinburgh, 25 September 1992). I am very grateful to theorganiser, Veronica van Heyningen, and all contributingspeakers for putting together a stimulating meeting. DouglasHiggs, Peter Harper, William Reardon, Sarah Slaney, and ananonymous referee made helpful comments on the manuscript.

1 Bennett JH, ed. Experiments in plant hybridisation by GMendel. London: Oliver & Boyd, 1965.

2 Vogel F, Motulsky AG. Human genetics: problems andapproaches. 2nd ed. Berlin: Springer Verlag, 1986:228-333.

3 Jackson IJ. Mouse coat colour mutations: a moleculargenetic resource which spans the centuries. Bioessays1991;13:439-46.

4 Pauli RM, Conroy MM, Langer LO, et al. Homozygousachondroplasia with survival beyond infancy. Am Jf MedGenet 1983;16:459-73.

5 Hulten MA, Honeyman MM, Mayne AJ, Tarlow MJ.Homozygosity in piebald trait. Med Genet1987;24:568-71.

6 Pauli RM. Dominance and homozygosity in man. AmMed Genet 1983;16:455-8.

7 Wexler NS, Young AB, Tanzi RE, et al. Homozygotes forHuntington's disease. Nature 1987;326:194-7.

8 Myers RH, Leavitt J, Farrer LA, et al. Homozygote forHuntington disease. Am Jf Hum Genet 1989;45:615-8.

9 The Huntington's Disease Collaborative Research Group.A novel gene containing a trinucleotide repeat that isexpanded and unstable on Huntington's disease chromo-somes. Cell 1993;72:971-83.

10 McKusick VA. Mendelian inheritance in man: catalogs ofautosomal dominant, autosomal recessive, and X linkedphenotypes. 10th ed. Baltimore: Johns Hopkins Univer-sity Press, 1992:xxi.

11 Vogel F, Motulsky AG. Human genetics: problems andapproaches. 2nd ed. Berlin: Springer Verlag, 1986:128-9.

12 Jaenisch R. Transgenic animals. Science 1988;240:1468-74.13 Friedrich G, Soriano P. Promoter traps in embryonic stem

cells: a genetic screen to identify and mutate develop-mental genes in mice. Genes Dev 1991;5:1513-23.

14 Orr HA. A test of Fisher's theory of dominance. Proc NatlAcad Sci USA 1991;88:11413-5.

15 Charlesworth B. When to be diploid. Nature 1991;351:273-4.

16 Kondrashov AS, Crow JF. Haploidy or diploidy: which isbetter? Nature 1991;351:314-5.

17 Perrot V, Richerd S, Valero M. Transition from haploidy todiploidy. Nature 1991;351:315-7.

18 Kacser H, Burns JA. The molecular basis of dominance.Genetics 1981;97:639-66.

19 Hodgkin J, Papp A, Pulak R, Ambros V, Anderson P. Anew kind of informational suppression in the nematodeCaenorhabditis elegans. Genetics 1989,123:301-13.

20 Hodgkin J. Fluxes, doses and poisons: molecular perspec-tives on dominance. Trends Genet 1993;9:1-2.

21 Li E, Sucov HM, Lee KF, Evans RM, Jaenisch R. Normaldevelopment and growth of mice carrying a targeteddisruption of the oe1 retinoic acid receptor gene. ProcNatl Acad Sci USA 1993;90:1590-4.

22 Park EC, Horvitz HR. Mutations with dominant effects onthe behavior and morphology of the nematode Caeno-rhabditis elegans. Genetics 1986;113:821-52.

23 Muller HJ. Further studies on the nature and causes of genemutations. In: Jones DF, ed. Proceedings of the sixthinternational congress of genetics. Brooklyn Botanic Gar-dens, Wisconsin, 1932:213-55.

24 Carlson EA. Defining the gene: an evolving concept. AmHum Genet 1991;49:475-87.

25 Lindsley DL, Sandler L, Baker BS, et al. Segmental aneu-

ploidy and the genetic gross structure of the Drosophilagenome. Genetics 1972;71:157-84.

26 Ingham P, Smith J. Crossing the threshold. Curr Biol1992;2:465-7.

27 Willing MC, Pruchno CJ, Atkinson M, Byers PH. Osteo-genesis imperfecta type I is commonly due to a COLlAlnull allele of type I collagen. Am Hum Genet1992;51:508-15.

28 Goldstein JL, Sobhani MK, Faust JR, Brown MS. Hetero-zygous familial hypercholesterolemia: failure of normalallele to compensate for mutant allele at a regulatedgenetic locus. Cell 1976;9:195-203.

29 Lachmann PJ, Rosen FS. The catabolism of Cl-inhibitorand the pathogenesis of hereditary angio-edema. ActaPathol Microbiol Immunol Scand 1984;92 (Sect C, suppl284):35-9.

30 Tassabehji M, Read AP, Newton VE, et al. Mutations inthe PAX3 gene causing Waardenburg syndrome type 1

and type 2. Nature Genet 1993;3:26-30.31 Gruss P, Walther C. Pax in development. Cell 1992;69:719-

22.

32 Glaser T, Walton DS, Maas RL. Genomic structure, evolu-tionary conservation and aniridia mutations in the humanPAX6 gene. Nature Genet 1992;2:232-8.

33 Vortkamp A, Gessler M, Grzeschik KH. GLI3 zinc-fingergene interrupted by translocations in Greig syndromefamilies. Nature 1991;352:539-40.

34 Schimmang T, Lemaistre M, Vortkamp A, Ruther U.

Expression of the zinc finger gene Gli3 is affected in themorphogenetic mouse mutant extra-toes (Xt,. Develop-ment 1992;116:799-804.

35 Hastie ND. Dominant negative mutations in the Wilmstumour (WTl) gene cause Denys-Drash syndrome -

proof that a tumour-suppressor gene plays a crucial rolein normal genitourinary development. Hum Mol Genet1992;1 :293-5.

36 Little MH, Williamson KA, Mannens M, et al. Evidencethat WTI mutations in Denys-Drash syndrome patientsmay act in a dominant negative fashion. Humi Mol Genet1 993;2 :259-64.

37 McInnes RR, Bascom RA. Retinal genetics: a nullifyingeffect for rhodopsin. Nature Genet 1992;1:155-7.

38 Spritz RA, Holmes SA, Ramesar R, Greenberg J, Curtis D,Beighton P. Mutations of the KIT (mast stem cellgrowth factor receptor) proto-oncogene account for acontinuous range of phenotypes in human piebaldism.Am J Hum Genet 1992;51:1058-65.

39 Younger-Shepherd S, Vaessin H, Bier E, Jan LY, Jan YN.deadpan, an essential pan-neural gene encoding an HLHprotein, acts as a denominator in Drosophila sex determi-nation. Cell 1992;70:911-22.

40 Isoda K, Roth S, Nusslein-Volhard C. The functionaldomains of the Drosophila morphogen dorsal: evidencefrom the analysis of mutants. Genes Dev 1992;6:619-30.

41 Jiang J, Levine M. Binding affinities and cooperative inter-actions with bHLH activators delimit threshold re-sponses to the dorsal gradient morphogen. Cell1993;72:741-52.

42 St Johnston D, Nusslein-Volhard C. The origin of patternand polarity in the Drosophila embryo. Cell1992;68:201-19.

43 Holtzman DM, Epstein CJ. The molecular genetics ofDown syndrome. Mol Genet Med 1992;2:105-20.

44 Patel PI, Roa BB, Welcher AA, et al. The gene for theperipheral myelin protein PMP-22 is a candidate forCharcot-Marie-Tooth disease type lA. Nature Geniet1992;1:159-65.

45 Chance PF, Alderson MK, Leppig KA, et al. DNA deletionassociated with hereditary neuropathy with liability topressure palsies. Cell 1993;72:143-51.

46 Bishop JM. Molecular themes in oncogenesis. Cell1991;64:235-48.

47 Oliner JD, Kinzler KW, Meltzer PS, George DL, Vogel-stein B. Amplification of a gene encoding a p53-associ-ated protein in human sarcomas. Nature 1992;358:80-3.

48 Ruvkun G, Wightman B, Burglin T, Arasu P. Dominantgain-of-function mutations that lead to misregulation ofthe C elegans heterochronic gene lin-14, and the evolu-tionary implications of dominant mutations in pattern-formation genes. Development 1991;S1:47-54.

49 Martin DIK, Tsai SF, Orkin SH. Increased y-globinexpression in a nondeletion HPFH mediated by anerythroid-specific DNA-binding factor. Nature1 989;338:435-8.

50 Feingold EA, Forget BG. The breakpoint of a large deletioncausing hereditary persistence of fetal hemoglobin occurswithin an erythroid DNA domain remote from the 3-globin gene cluster. Blood 1989;74:2178-86.

51 White RAH, Akam ME. Contrabithorax mutations causeinappropriate expression of Ultrabithorax products inDrosophila. Nature 1985;318:567-9.

52 Schneuwly S, Kuroiwa A, Gehring WJ. Molecular analysisof the dominant homeotic Antennapedia phenotype.EMBO J 1987;6:201-6.

53 Cleary ML. Oncogenic conversion of transcription factorsby chromosomal translocations. Cell 1991;66:619-22.

54 Rabbitts TH. Translocations, master genes, and differencesbetween the origins of acute and chronic leukemias. Cell1991;67:641-4.

55 Trent C, Wood WB, Horvitz HR. A novel dominanttransformer allele of the sex-determining gene her-i ofCaenorhabditis elegans. Genetics 1988;120:145-57.

56 Miller MW, Duhl DMJ, Vrieling H, et al. Cloning of themouse agouti gene predicts a secreted protein ubiqui-tously expressed in mice carrying the lethal yellow muta-tion. Genes Dev 1993;7:454-67.

57 Michaud EJ, Bultman SJ, Stubbs LJ, Woychik RP. Theembryonic lethality of homozygous lethal yellow mice(A! A-V) is associated with the disruption of a novel RNA-binding protein. Genes Dev 1993;7:1203-13.

58 Bruening W, Bardeesy N, Silverman BL, et al. Germlineintronic and exonic mutations in the Wilms' tumour gene(WTI) affecting urogenital development. Nature Genet1992;1: 144-8.

59 Bickmore WA, Oghene K, Little MH, Seawright A, vanHeyningen V, Hastie ND. Modulation of DNA bindingspecificity by alternative splicing of the Wilms tumor wtlgene transcript. Science 1992;257:235-7.

60 Rogers S, Wells R, Rechsteiner M. Amino acid sequencescommon to rapidly degraded proteins: the PESThypothesis. Science 1986;234:364-8.

61 Reed SI. GI-specific cyclins: in search of an S-phase-promoting factor. Trends Genet 1991;7:95-9.

62 Mango SE, Maine EM, Kimble J. Carboxy-terminal trun-cation activates glp-l protein to specify vulval fates inCaenorhabditis elegans. Nature 1991 ;352:81 1-5.

63 Bourne HR, Sanders DA, McCormick F. The GTPasesuperfamily: conserved structure and molecular mechan-ism. Nature 1991;349:117-27.

64 Beitel G, Clark S, Horvitz HR. The Caenorhabditis elegansras gene let-60 acts as a switch in the pathway of vulvalinduction. Nature 1990;348:503-9.

65 Han M, Sternberg PW. Analysis of dominant-negative

97

Wilkie

mutations of the Caenorhabditis elegans let-60 ras gene.Genes Dev 1991;5:2188-98.

66 Schneider DS, Hudson KL, Lin TY, Anderson KV. Dom-inant and recessive mutations define functional domainsof Toll, a transmembrane protein required for dorsal-ventral polarity in the Drosophila embryo. Genes Dev1991;5:797-807.

67 Vogelstein B. A deadly inheritance. Nature 1990;348:681-2.68 Vogelstein B, Kinzler KW. p53 function and dysfunction.

Cell 1992;70:523-6.69 Milner J, Medcalf EA. Cotranslation of activated mutant

p53 with wild type drives the wild-type p53 protein intothe mutant conformation. Cell 1991;65:765-74.

70 Weinstein LS, Shenker A, Gejman PV, Merino MJ, Fried-man E, Spiegel AM. Activating mutations of the stimula-tory G protein in the McCune-Albright syndrome. NEnglJ7 Med 1991;325:1688-95.

71 Schwindinger WF, Francomano CA, Levine MA. Identifi-cation of a mutation in the gene encoding the a subunit ofthe stimulatory G protein of adenylyl cyclase inMcCune-Albright syndrome. Proc Natl Acad Sci USA1992;89:5152-6.

72 Ptacek LJ, George AL, Griggs RC, et al. Identification of amutation in the gene causing hyperkalemic periodicparalysis. Cell 1991;67:1021-7.

73 Rojas CV, Wang J, Schwartz LS, Hoffman EP, Powell BR,Brown RH Jr. A Met-to-Val mutation in the skeletalmuscle Na + channel a-subunit in hyperkalaemic periodicparalysis. Nature 1991;354:387-9.

74 McClatchey AI, Van den Bergh P, Pericak-Vance MA, etal. Temperature-sensitive mutations in the III-IVcytoplasmic loop region of the skeletal muscle sodiumchannel gene in paramyotonia congenita. Cell1992;68:769-74.

75 Valentijn LJ, Baas F, Wolterman RA, et al. Identical pointmutations ofPMP-22 in Trembler-J mouse and Charcot-Marie-Tooth disease type IA. Nature Genet 1992;2:288-91.

76 Kelley MR, Kidd S, Deutsch WA, Young MW. Mutationsaltering the structure of epidermal growth factor-likecoding sequences at the Drosophila Notch locus. Cell1987;51 :539-48.

77 Xu T, Rebay I, Fleming RJ, Scottgale TN, Artavanis-Tsakonas S. The Notch locus and the genetic circuitryinvolved in early Drosophila neurogenesis. Genes Dev1990;4:464-75.

78 Herskowitz I. Functional inactivation of genes by dominantnegative mutations. Nature 1987;329:219-22.

79 Amaya E, Musci TJ, Kirschner MW. Expression of adominant negative mutant of the FGF receptor disruptsmesoderm formation in Xenopus embryos. Cell1991;66:257-70.

80 Leone G, Maybaum L, Lee PWK. The reovirus cellattachment protein possesses two independently activetrimerization domains: basis of dominant negativeeffects. Cell 1992;71:479-88.

81 Benezra R, Davis RL, Lockshon D, Turner DL, WeintraubH. The protein Id: a negative regulator of helix-loop-helix DNA binding proteins. Cell 1990;61:49-59.

82 Deng T, Karin M. JunB differs from c-Jun in its DNA-binding and dimerization domains, and represses c-Junby formation of inactive heterodimers. Genes Dev1993;7:479-90.

83 Harada H, Fujita T, Miyamoto M, et al. Structurallysimilar but functionally distinct factors, IRF-1 and IRF-2, bind to the same regulatory elements of IFN and IFN-inducible genes. Cell 1989;58:729-39.

84 Sykes B. Bone disease cracks genetics. Nature 1990;348:18-20.

85 Byers PH, Wallis GA, Willing MC. Osteogenesis imper-fecta: translation of mutation to phenotype. J Med Genet1991;28:433-42.

86 Dietz HC, Cutting GR, Pyeritz RE, et al. Marfan syndromecaused by a recurrent de novo missense mutation in thefibrillin gene. Nature 1991;352:337-9.

87 Milewicz DMcG, Pyeritz RE, Crawford ES, Byers PH.Marfan syndrome: defective synthesis, secretion, andextracellular matrix formation of fibrillin by cultureddermal fibroblasts. J Clin Invest 1992;89:79-86.

88 Bejsovec A, Anderson P. Functions of the myosin ATP andactin binding sites are required for C elegans thickfilament assembly. Cell 1990;60:133-40.

89 Watkins H, Rosenzweig A, Hwang DS, et al. Characteris-tics and prognostic implications of myosin missensemutations in familial hypertrophic cardiomyopathy. NEnglJ Med 1992;326:1108-14.

90 Fuchs E, Coulombe PA. Of mice and men: genetic skindiseases of keratin. Cell 1992;69:899-902.

91 Monplaisir N, Merault G, Poyart C, et al. Hemoglobin SAntilles: a variant with lower solubility than hemoglobinS and producing sickle cell disease in heterozygotes. ProcNatl Acad Sci USA 1986;83:9363-7.

92 Driscoll M, Chalfie M. The mec-4 gene is a member of afamily of Caenorhabditis elegans genes that can mutate toinduce neuronal degeneration. Nature 1991;349:588-93.

93 Johnson R, Jackson IJ. Light is a dominant mouse mutationresulting in premature cell death. Nature Genet1992;1:226-9.

94 Citron M, Oltersdorf T, Haass C, et al. Mutation of the 3-

amyloid precursor protein in familial Alzheimer's diseaseincreases f-protein production. Nature 1992;360:672-4.

95 Benson MD, Liepnieks J, Uemichi T, Wheeler G, CorreaR. Hereditary renal amyloidosis associated with a mutantfibrinogen a-chain. Nature Genet 1993;3:252-5.

96 Pepys MB, Hawkins PN, Booth DR, et al. Human lysozymegene mutations cause hereditary systemic amyloidosis.Nature 1993;362:553-7.

97 Fersht A, Winter G. Protein engineering. Trends BiochemSci 1992;17:292-4.

98 Owen MC, Brennan SO, Lewis JH, Carrell RW. Mutationof antitrypsin to antithrombin. al -antitrypsin Pittsburgh(358 Met--Arg), a fatal bleeding disorder. N Engl_3 Med1983;309:694-8.

99 Kim CM, Goldstein JL, Brown MS. cDNA cloning ofMEV, a mutant protein that facilitates cellular uptake ofmevalonate, and identification of the point mutationresponsible for its gain of function. J Biol Chem1992;267:231 13-21.

100 Kurzrock R, Gutterman JU, Talpaz M. The moleculargenetics of Philadelphia chromosome-positive leukemias.N Engl Jf Med 1988;319:990-8.

101 Barr FG, Galili N, Holick J, Biegel JA, Rovera G, Ema-nuel BS. Rearrangement of the PAX3 paired box gene inthe paediatric solid tumour alveolar rhabdomyosarcoma.Nature Genet 1993;3:113-7.

102 Dreesen TD, Henikoff S, Loughney K. A pairirrg-sensi-tive element that mediates trans-inactivation is associatedwith the Drosophila brown gene. Genes Dev 1991;5:331-40.

103 Tartof KD, Henikoff S. Trans-sensing effects from Dro-sophila to humans. Cell 1991;65:201-3.

104 Flavell RB. Variation in structure and expression of ribo-somal DNA loci in wheat. Genome 1989;31:963-8.

105 Lyttle TW. Cheaters sometimes prosper: distortion ofmendelian segregation by meiotic drive. Trends Genet1993;9:205-10.

106 Hays TS, Deuring R, Robertson B, Prout M, Fuller MT.Interacting proteins identified by genetic interactions: amissense mutation in a-tubulin fails to complementalleles of the testis-specific 3-tubulin gene of Drosophilamelanogaster. Mol Cell Biol 1989;9:875-884.

107 Beggs AH, Neumann PE, Arahata K, et al. Possibleinfluences on the expression of X chromosome-linkeddystrophin abnormalities by heterozygosity for autoso-mal recessive Fukuyama congenital muscular dystrophy.Proc Natl Acad Sci USA 1992;89:623-7.

108 Johnson WG. Metabolic interference and the + - hetero-zygote. A hypothetical form of simple inheritance whichis neither dominant nor recessive. Am J Hum Genet1980;32:374-86.

109 Knudson AG. Mutation and cancer: statistical study ofretinoblastoma. Proc Natl Acad Aci USA 1971;68:820-3.

110 Cavenee WK, Dryja TP, Phillips RA, et al. Expression ofrecessive alleles by chromosomal mechanisms in retino-blastoma. Nature 1983;305:779-84.

111 DeChiara TM, Robertson EJ, Efstratiadis A. Parentalimprinting of the mouse insulin-like growth factor IIgene. Cell 1991;64:849-59.

112 Koufos A, Grundy P, Morgan K, et al. Familial Wiede-mann-Beckwith syndrome and a second Wilms tumorlocus both map to l1p15.5. Am J Hum Genet1989;44:71 1-19.

113 Heutink P, van der Mey AGL, Sandkuijl LA, et al. A genesubject to genomic imprinting and responsible for here-ditary paragangliomas maps to chromosome 1 lq23-qter.Hum Mol Genet 1992;1:7-10.

114 Tommerup N. Mendelian cytogenetics. Chromosome re-arrangements associated with mendelian disorders. JMed Genet 1993;30:713-27.

115 Bourne HR. Consider the coiled coil Nature1991;351:188-90.

116 Bourne HR. Suppression with a difference. Nature1991;353:696-8.

1 17 Miyoshi Y, Ando H, Nagase H, et al. Germ-line mutationsof the APC gene in 53 familial adenomatous polyposispatients. Proc Natl Acad Sci USA 1992;89:4452-6.

118 Powell SM, Zilz N, Beazer-Barclay Y, et al. APC muta-tions occur early during colorectal tumorigenesis. Nature1992;359:235-7.

119 Pieretti M, Zhang F, Fu YH, et al. Absence of expressionof the FMR-1 gene in fragile X syndrome. Cell1991;66:817-22.

120 Sabouri LA, Mahadevan MS, Narang M, Lee DSC, SurhLC, Korneluk RG. Effect of the myotonic dystrophy(DM) mutation on mRNA levels of the DM gene. NatureGenet 1993;4:233-8.

121 Fu YH, Friedman DL, Richards S, et al. Decreasedexpression of myotonin-protein kinase messenger RNAand protein in adult form of myotonic dystrophy. Science1993;260:235-8.

122 Hofmann-Radvanyi H, Lavedan C, Rabes JP, et al. Myo-tonic dystrophy: absence of CTG enlarged transcript incongenital forms, and low expression of thle normal allele.Hum Mol Genet 1993;2:1263-6.

123 Mandel JL. Questions of expansion. Nature Genet1993;4:8-9.

124 Miller AD. Human gene therapy comes of age. Nature1992;357:455-60.

98

Documents

The molecular basis of genetic dominance