American J o u r n a l of Medical Genetics Supplement 7:31-37 (1990)
The Consequeiices of Chromosome Imbalance Charles J. Epstein Departments of Pediatrics and 0l‘Biochemistry and Biophysics, University of California, S a n Francisco
Review of the clinical cyl ogenetic literature provides compelling evidence for a specific relationship between imblilance of particular chromosomes or chromo soma1 regions and the appearance of defined patterns of phenotypic abnormalities. 111 many instances, detailed phenotypic mapping has made it possible to assign portions of ;L phenotype to relatively small chromosome segments, which are sometimes referred to as “critical regions.” However, since these regions are usually defined by a subset of the phcnotypic manifestations of an aneuploidy sy ndrome-generally those anomalies that are regarded as most characteristic or readily o bservable-it is important not to fall into the trap of thinking that it is imbalance of only these regions that has deleterious effects on development and function. Thus, in Down syndrome, the presence of an extra copy of the proximal part of 21q22.3 appears to result i n the typical physical phenotype-as defined principally in terms of the characteristic facial and hand anomalies and congenital heart defecein addition to mental retardation. But, duplication of proximal 21q also affecits mental development, and the regions rer,ponsible for many other aspects of the Down syndrome phenotype, including Alzhei mer disease, have not been defined at all. Therefore, it remains likely that loci present on many parts of the long arm of chromosome 21 play a role in the development of the ove rall phenotype of Down syndrome. The immediate effect at the molecular level of an aneuploidy-caused alteration in gene dose appears to be a non-compensated commensurate change in the production of gene products. Therefore, the mechanisms invoked to explain the genesis of the phenotype must be based on 50% increases and decreases in gene product synthesis in trisomies and
monosomies, respectively. Despite the specificity of patterns of phenotypic abnormalities, there is both a considerable degree of variability in the expression of the individual components that constitute these patterns and a significant degree of overlap among the patterns of different aneuploid states. This variability is presumed to be the result of a combination of stochastic, environmental, and other genetic factors which impinge on but do not obscure the overall pattern of abnormalities. The overlap of phenotypes may be attributable to the involvement of many gene products of different chromosomal origin in any particular developmental pathway, and hence the susceptibility of such pathways to a variety of perturbations. KEY WORDS: aneuploidy, critical regions, phenotypes, phenotypic mapping, chromosome 21, specificity (of phenotypes), variability (of phenotypes)
INTRODUCTION With the rapid advances th a t have been made in the techniques for mapping of the human genome, considerable attention has focussed on the mapping of chromosome 21. The attractiveness of this particular chromosome is the result of two factors-its small size and its involvement in the causation of Down syndrome (DS). As more genes are mapped to chromosome 21 and their physical locations and linkage relationships are delineated, it is becoming possible to seek to determine which and how many of these genes play a n important role in the pathogenesis of DS. In the pre-molecular mapping days, the distal part of chromosome 21, the segment encompassed by the bands in 21q22, was identified as the “Down syndrome region” ofthe chromosome [Rethore, 1981; Summitt, 19811.This conclusion was based on the analysis of the phenotypes of cases of DS that resulted from duplications of part of chromosome 21, usually as the result of reciprocal transFkceived for publication on July 17, 1989, revision received Seplocations. Accurate determination of the breakpoints in tember 6, 1989. Address reprint requests to Char16 s J. Epstein, M.D., Depart- these translocations was, of course, a frequent problem. ment of Pediatrics, Box 0106, Univer ;ity o f California, San Fran- However, studies now being carried out, which combine high resolution cytogenetics with the use of chromocisco, CA 94143.
0 1990 Wiley-Liss, Inc.
32
Epstein
some-specificmolecular probes, are attempting to delineate smaller parts of 21q22 as being of primary importance in the determination of the DS phenotype. Given the great interest in studies of this type, it seems appropriate to consider a t this time the conceptual basis for such attempts and to anticipate what types of results might be expected and how they ought to be interpreted. PHENOTYPE-GENOTYPE CORRELATIONS IN ANEUPLOID STATES Implicit in all attempts to define a “Down syndrome region” is the assumption that the phenotypic manifestations of a n aneuploidy syndrome can indeed be assigned to a specific region or regions of chromosome imbalance. To assess the validity of this assumption I undertook, several years ago, a n analysis of the clinical and cytogenetic descriptions in the clinical literature of a large number and variety of cases of aneuploidy, both duplications and deletions [Epstein, 19861. From this analysis it was possible to propose a set of principles concerning the pathogenesis of aneuploid phenotypes, and two of these principles are particularly germane to this discussion. Phenotypic Mapping The first principle addresses the just-stated assumption directly and holds that individual phenotypic anomalies or features can often be assigned or m a p p e d to specific regions of the genome, and conversely, the component manifestations of individual aneuploid states can often be added together to generate the p h e n o t y p e s of combined aneuploidies. This principle is based on comparisons of phenotypes resulting either from the simultaneous imbalance of two chromosomes or chromosome segments or from progressively larger degrees of imbalance of a single chromosome. The two elements of this principle are complementary to one another. Taken together they indicate that the phenotype of a n aneuploid state can be decomposed into a series of sub-phenotypes associated with sub-segments of the overall region of imbalance and, furthermore, that these sub-phenotypes can then be added together or recombined to give the overall phenotype. The decomposition of phenotypes is done by a subtraction procedure, much the same as is used for the physical mapping of genes. If a trait not found in association with imbalance of a small chromosome segment is found with imbalance of a larger segment that includes the first one, then the trait is provisionally mapped to the region that constitutes the difference between the larger and the smaller one. To prove that imbalance of the region so identified is in fact responsible for the appearance of the trait, it is necessary to show that aneuploidy for this region alone also produces the trait or that this region is the only one in common when the trait appears as the result of imbalance of two segments that overlap only in the region in question. The reconstruction of phenotypes from sub-phenotypes is not without problems, and there is a certain amount of “noise” in the analysis which extends beyond that expected just from variability of expression. When deduced sub-phenotypes are combined naturally, as oc-
curs in various types of double aneuploidy, certain phenotypic manifestations may disappear and new ones not characteristic of either sub-phenotype may appear. That such events should occur is not particularly surprising, since the combination of two unbalanced regions allows for interactions that are not possible when just one region is unbalanced. However, what is impressive is that the obliteration or appearance of new phenotypic traits does not occur more frequently. From this I have inferred that those features that do remain unaltered result from imbalance of loci that have strong and determinative effects on development and that these are the loci that should ultimately be most readily identifiable. However, it must of course be understood that localization of a particular phenotypic defect to a specific chromosome region does not imply that there is a specific locus or loci for the trait in question or for development of the affected tissue or organ. All that can be understood is that imbalance of the identified region leads to a n alteration in the development of this tissue or organ and thereby results in production of the phenotypic abnormality. The manner by which i t occurs does not alter the specificity of the relationship between the responsible locus or loci and the resulting abnormality Critical Regions Having argued that it is possible to decompose the phenotype of a chromosomal syndrome into a set of subphenotypes that can be mapped to sub-segments of the region of imbalance, i t is important to consider whether so-called “critical regions” exist. A critical region is visualized a s being the smallest chromosome segment which, when unbalanced, will give rise to the phenotype generally associated with imbalance of a larger chromosomal region t h a t contains it. Two such critical regions are indicated schematically on the right-hand side of Figure l-a region for traits A and B, and a region for traits C-F. Proposals for the existence of critical regions have been made for aneuploid syndromes involving deletions or duplications of many different chromosomes [see Epstein, 1986, pp 50-521. Related to the notion of critical regions for phenotype determination are instances in which there appears to be a lack of relationship between the length of a n unbalanced chromosomal segment and the clinical manifestations of severity of the resulting syndrome. It is as though the whole phenotype of a n aneuploid state is entirely attributable to imbalance of only a small chromosome region, and imbalance of the surrounding region is irrelevant. The idea that this might actually be the case is not without precedent, since there certainly do exist recessive disorders, which are presumed to be single gene defects, that have all of the earmarks of multiple malformation syndromes indistinguishable clinically from chromosomal disorders. A case in point is the C syndrome (the Opitz trigonocephaly syndrome) which is characterized by, among other things, trigonocephaly, mental retardation, unusual facial appearance, polydactyly, cardiac anomalies, cryptorchidism, abnormally positioned ears, and loose skin neonatally [Antley et al., 19811. Other examples of recessively inherited multiple malformation and mental syndromes
Consequences of Chromosome Imbalance A B
“C r i t i ca I region ’’
9 h i
j
8eCritical
Fig. 1. Schematic view of characteri:.ation of “critical regions” responsible for specific manifestations (A- F) of an aneuploid phenotype by identification of the smallest regions ,f imbalance that produce one or more anomalies. Other, more subtle defects that may result from imbalance of segments outside the “critical regions” are designated g-j .
include the Zellweger syndrom and glutaric aciduria type 11, syndromes caused by k Town metabolic defects [Frerman and Goodman, 1989; Lazarow and Moser, 1989;Wilson et al., 19891. In adclition to these recessive disorders, there are now several examples of small deletion or contiguous gene syndronies, such as the PraderWilli, Miller-Dieker,Angelman, DiGeorge, and LangerGiedion syndromes, in which complex constellations of multiple anomalies are caused by small deletions involving an unknown but presuinably small number of loci [Ledbetter and Cavenee, 1989; Williams et al., 1989bI. It is not inconceivable 1,hat some of the conditions may be the result of loss of just a single gene. Despite the existence of these examples, I do not think that involvement of only a very small number of genes is really the situation with regard ,o aneuploidy involving whole chromosomes or large segments thereof. Rather, what is being pointed to by the r otion of critical regions is the lack of uniformity in the rolationship between the imbalance of different regions of an unbalanced chromosome and the generation of phtmotypic traits-particularly those relatively few that might be considered as quite characteristic or even pikhognomonic of aneuploidy of the region (for example the “cat cry” in the cridu-chat syndrome). However, even though certain chromosomal segments may not have significant visible impacts on the syndrome, this does not mean that more subtle effects, illustrated on tho right side of Figure 1 with the small letters g-j, may not be occurring. Thus, using the cri-du-chat example again, while the typical phenotype of this syndrome miiy reside in deletion of band 5~15.2,there still appealas to be a relationship between the amount of 5p deleted and the severity of the mental retardation [Carlin and Neadle, 19781. Therefore, while the concept of a criticid region or regions may
33
be useful for identifying the part or parts of a chromosome that seem t o have the greatest effect on the obseruable phenotype, it is important not to fall into the trap of concluding that it is imbalance of only this region or regions that has deleterious effects on development and function.
Is There a Down Syndrome Region? These considerations are directly applicable to both past and current attempts to identify a critical region for the DS phenotype. The sense of the most recent investigations is that duplication of the distal part of 21q22.1 and proximal part of 21q22.3,perhaps only of the latter (the status of21q22.2 is unknown), seems to be all that is required to cause the physical phenotype of DS (Fig. 2) [Pellissier et al., 1988; Korenberg et al., 1990a, b; McCormick et al., 1989; Rahmani et al., 19901. This phenotype is defined principally in terms of the characteristic facial and hand changes, congenital heart defect, and. of course. mental retardation. Do these findinas mean that the rest of chromosome 21 is irrelevant to tge phenotype? The is, for two reasons, clearly not. First, Of the proximal part Of chromosome 21q is not without effect [Williams et al., 19903. The most characteristic finding in the reported cases of du-
MENTAL RETARDATION
q22.1 q22’2
q22.3
II H1
“DOWN SYNDROME’
I IJ
Fig. 2. The “Down syndrome”region of chromosome 21, as defined by analysis of individuals with segmental duplications who manifest the major phenotypic abnormalities (facies, hand anomalies, mental retardation, and heart disease) of Down syndrome. However, imbalance of the proximal region of 21q is also known to cause mental retardation, and the region or regions responsible for other phenotypic anomalies of Down syndrome have not been defined.
34
Epstein
plications involving the region 21pter+q22.1 is some degree of mental impairment. This impairment may range from mild to severe, but it is usually associated with a n I& between 50 and 70 [Williams et al., 1975; Schinzel, 1984; Park et al., 19871.Reports of quite severe mental retardation from duplications of this region are somewhat difficult to interpret in view of the presence in some instances of other regions of imbalance (when translocation chromosomes are involved) or of other karyotypic abnormalities [asin Park e t al., 19871.Ascertainment bias may also be present. Nevertheless, if we consider cases with full trisomy 21, which constitute most cases of DS, i t must be concluded that imbalance of both the proximal and distal parts of 21q probably contributes to the overall degree of mental retardation. The second reason that the proximal and very distal parts of 21q cannot be considered irrelevant to the pathogenesis of DS is that there are more aspects of the DS phenotype than are encompassed in the existing clinical descriptions of individuals with duplications of 21q22. These include other major anomalies (particularly duodenal atresia and other gastrointestinal defects), neonatal hematological abnormalities, susceptibility to leukemia (especially megakaryoblastic), immunological deficits, a variety of cellular alterations (enhanced responses to P-adrenergic agents, decreased platelet serotonin, and possibly a n increased sensitivity to radiation, to mutagenic and carcinogenic agents and to virus-induced transformation),and, perhaps of greatest interest, Alzheimer disease (AD) [see Epstein, 1986 for summary]. The situation with regard to the relationship between DS and AD is of particular interest since the two major genes thought possibly to be playing a role in the pathogenesis of this form AD, the genes AAP and AD1 for the amyloid precursor protein and for familial AD, respectively, both seem to be in 21q21. Likewise, SOD1,the gene for CuZn-superoxide dismutase, for which a role in the pathogenesis of AD has also been suggested [Sinet, 19821, is in proximal 21q22.1, outside of the phenotypic DS region. There is presently no firm evidence to implicate any of these genes in the pathogenesis of AD in DS, and it is possible that the relationship between the two conditions is not attributable to the effects of any of them. Nevertheless, i t would be of great interest to know whether individuals with duplications of chromosome 21 centromeric to the distal part of 21q22.1, with and without the region of 21q22.1 that contains SODl,do or do not develop the pathology of AD when they reach the fourth decade of life.
NEED FOR COLLABORATIVE STUDIES At this point I would like to depart for a moment from the formal text to discuss a practical matter concerning the issue that I have just been addressing. As I have already said, considerable work is now being devoted to the molecular analysis of relatively small duplications of chromosome 21. These analyses will now supersede the less precise cytogenetic assessments of the amount and region of chromosome 21 that is involved in each of these duplications. However, just as crucial as the mo-
lecular delineation of the lesion will be the clinical definition of the phenotype of each individual whose DNA is being analyzed. And, by phenotype I mean not just the obvious manifestations that are usually considered, but also all of the other phenotypic changes that have already been alluded to. These analyses need to be done not only for individuals with clear-cut DS, but for all persons with duplications involving any part of chromosome 21. To make this possible, I urge that we establish a standardized and perhaps centralized means for data collection and for the analysis of manifestations such a s immunological status, sensitivity of cells to various agents, and the like. If we are really to succeed in mapping the phenotypic components of DS, it will be necessary for all of the workers in the field to collaborate with one another in obtaining comparable and meaningful data. There are already many precedents for such collaborative efforts in human genetics, and the field of DS research now seems to be ready for one.
THE SPECIFICITY OF ANEUPLOID PHENOTYPES I turn now to the second principle concerning the pathogenesis of aneuploid phenotypes, a principle that gets to the heart of the types ofmechanisms we should be considering to explain their development. This principle states that different aneuploid phenotypes, while often variable in expression and frequently overlapping in phenotype, are each specific and differentiable from one another. This principle asserts that there is a specific and deterministic relationship between imbalance of a particular region of the genome and the syndrome of malformations and other defects that occurs as a result of this imbalance. It argues against the proposition that most of the manifestations of aneuploidy may be interpreted as the results of a nonspecific disturbance of chromosome balance [BlumHoffman et al., 19871. The latter proposition, and others related to it, have been advanced for two reasons. The first is that there is a considerable degree of variability in aneuploid phenotypes, and any two individuals with the same aneuploid state do not necessarily have identical phenotypic abnormalities. Phenotypic variability is certainly the case for DS, but this still does not prevent us from recognizing when a n individual has trisomy 21. However, phenotypic variability should not really surprise us, since developmental processes and chromosomal aberrations do not occur in a vacuum. Except for identical twins, individuals with the same chromosome imbalance are genetically distinct from one another, and differences in the background genome have been shown in experimental animals to influence the ultimate phenotype [for discussion see Epstein, 1986, pp 183-1901. Even identical twins may not have identical phenotypes [Elder et al., 19841,since stochastic factors (for example, the time and place at which individual cells divide, or whether and in which direction they migrate) introduce a small element of randomness into the operation of a developmental pathway. Local or external environmental factors may have a similar perturbing effect. Kurnit and his collaborators [1985, 19871
Consequences of Chromosome Imbalance have argued strongly for a maj 3r role for stochastic factors in determining, once the genetic imbalance is present, whether or not congenital I- eart disease will develop in embryos with trisomy 21, and it is likely that such factors affect many morphoger letic processes.
35
euploid syndromes a s well a s we are able to if this were truly the case.
MECHANISMS OF ANEUPLOID PHENOTYPES If nonspecific effects of aneuploidy are rejected as being major determinants of aneuploid phenotypes, what types of pathogenetic mechanisms should we be considering, and how should we go about searching for them? This is a subject that I have discussed a t great length in a recent monograph [Epstein, 19861, and a t the present time I believe that it is fair to say that no mechanism has yet been shown to apply. Therefore, all types of mechanisms ought to be considered, although I think that certain systems and processes might be particularly vulnerable to the effects of aneuploidy. Among these I include intercellular interactions (such as recognition, adhesion, and communication); receptors, growth factors and morphogens, regulatory systems (including trans-acting factors); and multi-subunit macromolecules. For each of these it is possible to construct theoretical arguments and to use experimental precedents to demonstrate how relatively small changes in gene product concentrations could have quite large effects on function. Effects of aneuploidy on metabolic pathways must also be considered, but I suspect these are less likely to play a n important role, especially in trisomy. Nevertheless, as the work by Groner and his collaborators [1990] and by ourselves [Chan et al., 1988, Epstein et al., 19901 has shown, modest increases in CuZn-superoxide dismutase (SOD-1) activity, as are found in transgenic mice expressing inserted copies of the human SOD-1 gene [Epstein et al., 19871, do have effects on the sensitivity of cells to exogenous agents which perturb oxygen free radical metabolism, and perhaps even on the structure and integrity of the neuromuscular junctions of the tongue [Yarom et al., 19881. In considering how the presence of a n extra copy of any gene or set of genes might lead to some abnormality of development or function, it must be kept in mind that, unless there is evidence to the contrary, any mechanism that is proposed must ultimately be consistent with existence of a strictly proportional dosage effect for this gene or genes. In the case of trisomy 21 and DS, this means a 50% increase in the rate of synthesis of gene products coded for by chromosome 21 and, in general, a 50% increase in their concentration. Although autosoma1 dosage compensation does exist in Drosophila and plants, there is as yet no firm evidence that it exists in man and other mammals. However, a few possible examples have been suggested [Mangin et al., 1985; TothFejel et al., 19871, and i t is still important to determine whether a dosage effect is present whenever a new locus is mapped to chromosome 21. Since virtually all of the data on dosage effects in all species examined is derived from measurements of enzyme activities [for summary see Epstein, 1986, pp 65-84], analysis of the products of loci coding for other types of molecules would be of particular interest.
Phenotypic Overlap The second reason the issue (If nonspecificity arises is because many aneuploidy syndromes, DS included, share similar phenotypic abnormalities. Mental retardation and growth impairment are virtually universal when significant aneuploidy is wesent. In addition, any two chromosome abnormalitie 3 picked at random are likely to share a number of minor anomalies. Nevertheless, despite these overlaps, it is still possible to discriminate different and often quite ( losely related aneuploid phenotypes from one another with a remarkably high degree of accuracy. While such discriminations can readily be made by formal taronomic analysis [Preus and Ayme, 19831, this is rarely necessary, and a knowledgeable and astute clinical E eneticist can generally make the correct diagnosis. The crucial point, a s concluded by Preus and Ayme [19831, is that “although there are few if any physical fc atures which are exclusive to a particular chromoson a1 defect, the pattern of defects is distinct.” How do we explain these ovcdapping phenotypes? I would prefer to think, as statcd by Opitz [1982], that “human organs are evidently capable of responding to a high number of diverse dysmo~phogeneticcauses with the production of only a limited repertoire of malformations.” If we have learned no1,hing else from human biochemical genetics, we have certainly come to realize that many different enzymatic defects scattered throughout the genome can kave quite similar phenotypic effects. However, the other general explanation is that the effects of aneuploidy are nonspecific in that they may be produced by any type of chromosome imbalance. One mechanism that has Deen suggested for these nonspecific effects is “amplific d developmental instability,” in which there is a decreased regulation or “buffering” of morphogenetic processes and, as a result, a greater variability of outcome [Shapiro, 19831. In support of this proposal are data showing a n increased statistical variance of a variety of metric traits i n persons with aneuploidy [Shapiro, 1970, 1975; Langenbeck et al., 1984: Blum-Hoffman et al. 19871. However, not all traits show such a n increase, and it appears that increased variability and variance, when they do occur in DS and other aneuploid conditions, are quite selective with regard to which developmental systems are affected [Epstein, 19881. As I wrote earlier [Epstein, 13881, I have no difficulty with the notion t h a t particular aneuploid states may indeed affect the stability of ceri ain developmental processes and not of others. However, I do not think that the available evidence supports a conclusion that aneuploidy per se, independently cf which chromosome is involved, leads t o a generalizec disturbance of developCONCLUSIONS mental homeostasis which, in tiirn, produces most of the Based on the two general principles that have been phenotypic abnormalities of thc syndrome. It would just not be possible to discriminate among the different an- discussed and on many other theoretical considerations
36
Epstein
and experimental observations, I [Epstein, 1986, p 3721 have concluded that-
I t is legitimate to look for and to expect to find specific mechanisms to explain the relationships between particular phenotypic characteristics and the imbalance of individual loci or sets of loci. Contributions to the phenotype are likely to come from all parts of the unbalanced genome and not solely from j u s t one or a very few loci. While some loci may have a greaterphenotypic effect or representation than others, it is the cumulative effect of imbalance of many loci that determines the overall phenotype. As complicated as the relationships between genetic loci and phenotypic effects may be, and with whatever element of randomness they may be associated, these relationships should nonetheless exist and be discernible. T h e generation of aneuploid phenotypes is not, therefore, a game of chance but is an exercise of the conventional rules of developmental genetics and developmental biology. As I have already written, it has not as yet been possible to forge a mechanistic link between any phenotypic manifestation of a n aneuploid state and imbalance of a specific locus. This is, I feel, only a temporary impasse, since the necessary information and techniques for attacking the problem are now becoming available. Critical to this endeavor is the mapping of the human and mouse genomes, so that it is rapidly becoming possible to think in terms of the specific genes that are unbalanced in any type of aneuploidy, and particularly in trisomy 21. This, coupled with the rapid strides that are being made in developmental biology and neurobiology, now makes i t possible to approach the pathogenesis of the effects of aneuploidy from two directions: from imbalance of known genes to their potential phenotypic consequences, and from observed phenotypic aberrations to their likely genetic causes. When these two approaches ultimately converge, we shall truly begin to understand how genetic imbalance produces aneuploid phenotypes and, more importantly, how trisomy 21 causes Down syndrome.
ACKNOWLEDGMENTS This work was supported by NIH grants HD-17001 and AG 08938. REFERENCES Antley RM, Hwang DS, Theopold W, Gorlin R J , Steeper T, Pitt D, Danks DM, McPherson E, Bartels H , Wiedemann H-R, Opitz J M (1981):Further delineation of the C (trigonocephaly) syndrome. Am J Med Genet 9:147-163. Blum-Hoffman E, Rehder H, Langenbeck U (1987): Skeletal anomalies in trisomy 21 a s an example of amplified developmental instability in chromosome disorders: A histological study of the feet of 21 midtrimester fetuses with trisomy 21. Am J Med Genet 29:155-160. Carlin ME, Neadle MM (1978): Cri-du-chat syndrome-a correlation between anomalies and size of deletion. In Summitt RL, Bergsma D (eds): “Annual Review of Birth Defects, 1977.” New York: Alan R. Liss, Inc. for the National Foundation-March of Dimes, BD:OAS XIV(6C):428-429. Chan PH, Yu ACH, Chen S,Chu L, Epstein CJ (1988): Oxidative stress
exacerbates cellular damage in primary cultures from human SOD-1 transgenic mice. J Cell Biol 107:726a. Elder FFB, Ferguson J W , Lockhart LH (1984): Identical twins with deletion 16q syndrome: Evidence that 16q12.2-ql3 is the critical band region. Hum Genet 67:233-236. Epstein C J (1986): “The Consequences of Chromosome Imbalance.” New York: Cambridge University Press. Epstein C J (1988):Specificity versus nonspecificity in the pathogenesis of aneuploid phenotypes. Am J Med Genet 29:161-165. Epstein C J , Avraham KB, Lovett M, Smith S, Elroy-Stein 0, Rotman G, Bry C, Groner Y (1987): Transgenic mice with increased CuZnsuperoxide dismutase activity: An animal model of dosage effects in Down syndrome. Proc Natl Acad Sci USA 84:8044-8048. Epstein CJ, Berger CN, Carlson E J , Chan PH, Huang TT (1990): Models for Down syndrome: chromosome 21-specific genes in mice. I n Patterson D, Epstein C J (eds): “Molecular Genetics of Chromosome 21 and Down Syndrome.” New York: Wiley-Liss, pp 215-232. Frerman FE, Goodman SI (1989):Glutaric acidemia type I1 and defects of the mitochondria1 respiratory chain. In Scriver CR, Beaudet AL, Sly WS, Valle D (eds): “The Metabolic Basis of Inherited Disease.” 6th ed. New York: McGraw-Hill, pp 915-931. Groner Y, Avraham KB, Schickler M, Yarom R, Knobler H (1990): Clinical symptoms of Down syndrome are manifested in transgenic mice overexpressing the human CuiZn-superoxide dismutase gene. In Patterson D, Epstein CJ (eds): “Molecular Genetics of Chromosome 21 and Down Syndrome.”New York: Wiley-Liss, pp 233-262. Korenberg J R , Kawashima H, Pulst S-M, Ikeuchi T, Ogasawara N, Yamomoto K, Schonberg S, West R, Allen L, Magenis E, Ikawa K, Taniguchi N, Epstein C J (1990a): Molecular definition of a region of chromosome 21 that causes features of the Down syndrome phenotype. Am J Hum Genet 47:236-246. Korenberg JR, Pulst SM, Kawashima H, Epstein CJ, Allen L, Magenis E (1990b): Down syndrome: Toward a molecular definition of the phenotype. Am J Med Genet, this issue. Kurnit DM, Aldridge J F , Matsuoka R, Matthysse S (1985): Increased adhesiveness of trisomy 2 1 cells and atrioventricular malformations in Down syndrome: A stochastic model. Am J Med Genet 20:385-399. Kurnit DM, Layton WM, Matthysse S (1987): Genetics, chance, and morphogenesis. Am J Hum Genet 41:979-995. Langenbeck U, Blum E, Wilkert-Walter C, Hansmann I(1984): Developmental pathogenesis of chromosome disorders: Report on two newly recognized signs of Down syndrome. Am J Med Genet 18:223-230. Lazarow PB, Moser HW (1989): Disorders of peroxisome biogenesis. In Scriver CR, Beaudet AL, Sly WS, Valle D (eds): “The Metabolic Basis of Inherited Disease.” 6th ed. New York: McGraw-Hill, pp 1479-1509. Ledbetter DH, Cavenee WK (1989): Molecular cytogenetics: Interface of cytogenetics and monogenic disorders. In Scriver CR, Beaudet AL, Sly WS, Valle D (eds): “The Metabolic Basis of Inherited Disease.’’ 6th ed. New York: McGraw-Hill, pp 343-371. McCormick MK, Schinzel A, Petersen MB, Stetten G, Driscoll DJ, Cantu ES, Tranebjaerg L, Mikkelsen M, Watkins PC, Antonarakis SE (1989): Molecular cytogenetic approach to the characterization of the “Down syndrome region” of chromosome 21. Genomics 5:325-331. Mangin M, Ares M Jr, Weiner AM (1985): U1 small nuclear RNA genes subject to dosage compensation in mouse cells. Science 229:272-275. Opitz J M (1982): The developmental field concept in clinical genetics. J Pediatr 101:805-809. Park J D , Wurster-Hill DH, Andrews PA, Cooley WC, Graham J M J r (1987):Free proximal trisomy 21 without the Down syndrome. Clin Genet 32:342-348. Pellissier MC, Laffage M, Philip N, Passage E, Mattei M-G, Mattei J - F (1988): Trisomy 21q223 and Down’s phenotype correlation evidenced by in situ hybridization. Hum Genet 80:277-281. Preus M, Aym6 S (1983): Formal analysis of dysmorphism: Objective mechanisms of syndrome definition. Clin Genet 23:l-16. Rahmani Z, Blouin JL, Creau-Goldberg N, Watkins PC, Mattei J F , Poissonier M, Prieur M, Chettouh Z, Nicole A, Aurias A, Sinet PM, Delabar J M (1990): Critical role of the D21S55 region on chromo-
Consequences of Chromosome Imbalance some 21 in the pathogenesis of Dow 1 syndrome. Am J Med Genet, this issue. Rethore M - 0 (1981):Structural variation ofchromosome 21 and symptoms of Down’s syndrome. In Burgio (:R, Fraccaro M, Tiepolo L, Wolf U (edsj: “Trisomy 21. An Intern xtional Symposium.” Berlin: Springer-Verlag, pp 173-182. Schinzel A (19841: “Catalogue of Unb.ilanced Chromosome Aberrations in Man.” Berlin: Walter de Gruyter, pp 690-695. Shapiro BL (19701: Prenatal dental a iomalies in mongolism: comments on the basis and implications c f variability. Ann NY Acad Sci 171(2):562-577. Shapiro BL (1975):Amplified developmc ntal instability in Down’s syndrome. Ann Hum Genet 38:429-4:7. Shapiro BL (1983):Down syndrome-A jisruption of homeostasis. Am J Med Genet 14:241-269. Sinet P-M (1982): Metabolism of oxygt’n derivatives in Down’s syndrome. Ann NY Acad Sci 396:83-54. Summitt RL (1981): Chromosome specific segments that cause the phenotype of Down syndrome. In dc la Cruz FF, Gerald PS (eds): ‘“hisomy 21 (Down Syndrome).Resexch Perspectives.” Baltimore: University Park Press, pp 225-235. Toth-Fejel S, Magenis RE, Kolbe M, G a l i J , Boyd CD (1987):Normal
37
levels of type IV collagen mRNA and normal basement membrane structure in patients with loss of a type IV procollagen gene. Am J Hum Genet 41:A87. Williams CA, Gray BA, Hendrickson JE, Stone JW, Cantu ES (1989): Incidence of 15q deletions in the Angelman syndrome: A survey of twelve affected persons. Am J Med Genet 32:339-345. Williams CA, Cantu ES, Frias J L , McCormick MK, Antonarakis SE (1990): Clinical, cytogenetic and molecular evaluation of a patient with partial trisomy 21 (21qll-q22) lacking the classical Down syndrome phenotype. Am J Med Genet, this issue. Williams J D , Summitt RL, Martens PR, Kimbrell RA (1975):Familial Down syndrome due to t ( 10;21) translocation: Evidence that the Down phenotype is related to trisomy of specific segment of chromosome 21. Am J Hum Genet 27:478-485. Wilson GN, Chadarevian J-P de, Kaplan P, Loehr J P , Frerman FE, Goodman SI (1989): Glutaric aciduria type 11: Review of the phenotype and report of an unusual glomerulopathy. Am J Med Genet 32:395-401. Yarom R, Sapoznikov D, Havivi Y, Avraham KB, Schickler M, Groner Y (1988):Premature aging changes in neuromuscular junctions of transgenic mice with an extra human CuZnSOD gene: A model for tongue pathology in Down’s syndrome. J Neurol Sci 88:41-53.
The Consequeiices of Chromosome Imbalance Charles J. Epstein Departments of Pediatrics and 0l‘Biochemistry and Biophysics, University of California, S a n Francisco
Review of the clinical cyl ogenetic literature provides compelling evidence for a specific relationship between imblilance of particular chromosomes or chromo soma1 regions and the appearance of defined patterns of phenotypic abnormalities. 111 many instances, detailed phenotypic mapping has made it possible to assign portions of ;L phenotype to relatively small chromosome segments, which are sometimes referred to as “critical regions.” However, since these regions are usually defined by a subset of the phcnotypic manifestations of an aneuploidy sy ndrome-generally those anomalies that are regarded as most characteristic or readily o bservable-it is important not to fall into the trap of thinking that it is imbalance of only these regions that has deleterious effects on development and function. Thus, in Down syndrome, the presence of an extra copy of the proximal part of 21q22.3 appears to result i n the typical physical phenotype-as defined principally in terms of the characteristic facial and hand anomalies and congenital heart defecein addition to mental retardation. But, duplication of proximal 21q also affecits mental development, and the regions rer,ponsible for many other aspects of the Down syndrome phenotype, including Alzhei mer disease, have not been defined at all. Therefore, it remains likely that loci present on many parts of the long arm of chromosome 21 play a role in the development of the ove rall phenotype of Down syndrome. The immediate effect at the molecular level of an aneuploidy-caused alteration in gene dose appears to be a non-compensated commensurate change in the production of gene products. Therefore, the mechanisms invoked to explain the genesis of the phenotype must be based on 50% increases and decreases in gene product synthesis in trisomies and
monosomies, respectively. Despite the specificity of patterns of phenotypic abnormalities, there is both a considerable degree of variability in the expression of the individual components that constitute these patterns and a significant degree of overlap among the patterns of different aneuploid states. This variability is presumed to be the result of a combination of stochastic, environmental, and other genetic factors which impinge on but do not obscure the overall pattern of abnormalities. The overlap of phenotypes may be attributable to the involvement of many gene products of different chromosomal origin in any particular developmental pathway, and hence the susceptibility of such pathways to a variety of perturbations. KEY WORDS: aneuploidy, critical regions, phenotypes, phenotypic mapping, chromosome 21, specificity (of phenotypes), variability (of phenotypes)
INTRODUCTION With the rapid advances th a t have been made in the techniques for mapping of the human genome, considerable attention has focussed on the mapping of chromosome 21. The attractiveness of this particular chromosome is the result of two factors-its small size and its involvement in the causation of Down syndrome (DS). As more genes are mapped to chromosome 21 and their physical locations and linkage relationships are delineated, it is becoming possible to seek to determine which and how many of these genes play a n important role in the pathogenesis of DS. In the pre-molecular mapping days, the distal part of chromosome 21, the segment encompassed by the bands in 21q22, was identified as the “Down syndrome region” ofthe chromosome [Rethore, 1981; Summitt, 19811.This conclusion was based on the analysis of the phenotypes of cases of DS that resulted from duplications of part of chromosome 21, usually as the result of reciprocal transFkceived for publication on July 17, 1989, revision received Seplocations. Accurate determination of the breakpoints in tember 6, 1989. Address reprint requests to Char16 s J. Epstein, M.D., Depart- these translocations was, of course, a frequent problem. ment of Pediatrics, Box 0106, Univer ;ity o f California, San Fran- However, studies now being carried out, which combine high resolution cytogenetics with the use of chromocisco, CA 94143.
0 1990 Wiley-Liss, Inc.
32
Epstein
some-specificmolecular probes, are attempting to delineate smaller parts of 21q22 as being of primary importance in the determination of the DS phenotype. Given the great interest in studies of this type, it seems appropriate to consider a t this time the conceptual basis for such attempts and to anticipate what types of results might be expected and how they ought to be interpreted. PHENOTYPE-GENOTYPE CORRELATIONS IN ANEUPLOID STATES Implicit in all attempts to define a “Down syndrome region” is the assumption that the phenotypic manifestations of a n aneuploidy syndrome can indeed be assigned to a specific region or regions of chromosome imbalance. To assess the validity of this assumption I undertook, several years ago, a n analysis of the clinical and cytogenetic descriptions in the clinical literature of a large number and variety of cases of aneuploidy, both duplications and deletions [Epstein, 19861. From this analysis it was possible to propose a set of principles concerning the pathogenesis of aneuploid phenotypes, and two of these principles are particularly germane to this discussion. Phenotypic Mapping The first principle addresses the just-stated assumption directly and holds that individual phenotypic anomalies or features can often be assigned or m a p p e d to specific regions of the genome, and conversely, the component manifestations of individual aneuploid states can often be added together to generate the p h e n o t y p e s of combined aneuploidies. This principle is based on comparisons of phenotypes resulting either from the simultaneous imbalance of two chromosomes or chromosome segments or from progressively larger degrees of imbalance of a single chromosome. The two elements of this principle are complementary to one another. Taken together they indicate that the phenotype of a n aneuploid state can be decomposed into a series of sub-phenotypes associated with sub-segments of the overall region of imbalance and, furthermore, that these sub-phenotypes can then be added together or recombined to give the overall phenotype. The decomposition of phenotypes is done by a subtraction procedure, much the same as is used for the physical mapping of genes. If a trait not found in association with imbalance of a small chromosome segment is found with imbalance of a larger segment that includes the first one, then the trait is provisionally mapped to the region that constitutes the difference between the larger and the smaller one. To prove that imbalance of the region so identified is in fact responsible for the appearance of the trait, it is necessary to show that aneuploidy for this region alone also produces the trait or that this region is the only one in common when the trait appears as the result of imbalance of two segments that overlap only in the region in question. The reconstruction of phenotypes from sub-phenotypes is not without problems, and there is a certain amount of “noise” in the analysis which extends beyond that expected just from variability of expression. When deduced sub-phenotypes are combined naturally, as oc-
curs in various types of double aneuploidy, certain phenotypic manifestations may disappear and new ones not characteristic of either sub-phenotype may appear. That such events should occur is not particularly surprising, since the combination of two unbalanced regions allows for interactions that are not possible when just one region is unbalanced. However, what is impressive is that the obliteration or appearance of new phenotypic traits does not occur more frequently. From this I have inferred that those features that do remain unaltered result from imbalance of loci that have strong and determinative effects on development and that these are the loci that should ultimately be most readily identifiable. However, it must of course be understood that localization of a particular phenotypic defect to a specific chromosome region does not imply that there is a specific locus or loci for the trait in question or for development of the affected tissue or organ. All that can be understood is that imbalance of the identified region leads to a n alteration in the development of this tissue or organ and thereby results in production of the phenotypic abnormality. The manner by which i t occurs does not alter the specificity of the relationship between the responsible locus or loci and the resulting abnormality Critical Regions Having argued that it is possible to decompose the phenotype of a chromosomal syndrome into a set of subphenotypes that can be mapped to sub-segments of the region of imbalance, i t is important to consider whether so-called “critical regions” exist. A critical region is visualized a s being the smallest chromosome segment which, when unbalanced, will give rise to the phenotype generally associated with imbalance of a larger chromosomal region t h a t contains it. Two such critical regions are indicated schematically on the right-hand side of Figure l-a region for traits A and B, and a region for traits C-F. Proposals for the existence of critical regions have been made for aneuploid syndromes involving deletions or duplications of many different chromosomes [see Epstein, 1986, pp 50-521. Related to the notion of critical regions for phenotype determination are instances in which there appears to be a lack of relationship between the length of a n unbalanced chromosomal segment and the clinical manifestations of severity of the resulting syndrome. It is as though the whole phenotype of a n aneuploid state is entirely attributable to imbalance of only a small chromosome region, and imbalance of the surrounding region is irrelevant. The idea that this might actually be the case is not without precedent, since there certainly do exist recessive disorders, which are presumed to be single gene defects, that have all of the earmarks of multiple malformation syndromes indistinguishable clinically from chromosomal disorders. A case in point is the C syndrome (the Opitz trigonocephaly syndrome) which is characterized by, among other things, trigonocephaly, mental retardation, unusual facial appearance, polydactyly, cardiac anomalies, cryptorchidism, abnormally positioned ears, and loose skin neonatally [Antley et al., 19811. Other examples of recessively inherited multiple malformation and mental syndromes
Consequences of Chromosome Imbalance A B
“C r i t i ca I region ’’
9 h i
j
8eCritical
Fig. 1. Schematic view of characteri:.ation of “critical regions” responsible for specific manifestations (A- F) of an aneuploid phenotype by identification of the smallest regions ,f imbalance that produce one or more anomalies. Other, more subtle defects that may result from imbalance of segments outside the “critical regions” are designated g-j .
include the Zellweger syndrom and glutaric aciduria type 11, syndromes caused by k Town metabolic defects [Frerman and Goodman, 1989; Lazarow and Moser, 1989;Wilson et al., 19891. In adclition to these recessive disorders, there are now several examples of small deletion or contiguous gene syndronies, such as the PraderWilli, Miller-Dieker,Angelman, DiGeorge, and LangerGiedion syndromes, in which complex constellations of multiple anomalies are caused by small deletions involving an unknown but presuinably small number of loci [Ledbetter and Cavenee, 1989; Williams et al., 1989bI. It is not inconceivable 1,hat some of the conditions may be the result of loss of just a single gene. Despite the existence of these examples, I do not think that involvement of only a very small number of genes is really the situation with regard ,o aneuploidy involving whole chromosomes or large segments thereof. Rather, what is being pointed to by the r otion of critical regions is the lack of uniformity in the rolationship between the imbalance of different regions of an unbalanced chromosome and the generation of phtmotypic traits-particularly those relatively few that might be considered as quite characteristic or even pikhognomonic of aneuploidy of the region (for example the “cat cry” in the cridu-chat syndrome). However, even though certain chromosomal segments may not have significant visible impacts on the syndrome, this does not mean that more subtle effects, illustrated on tho right side of Figure 1 with the small letters g-j, may not be occurring. Thus, using the cri-du-chat example again, while the typical phenotype of this syndrome miiy reside in deletion of band 5~15.2,there still appealas to be a relationship between the amount of 5p deleted and the severity of the mental retardation [Carlin and Neadle, 19781. Therefore, while the concept of a criticid region or regions may
33
be useful for identifying the part or parts of a chromosome that seem t o have the greatest effect on the obseruable phenotype, it is important not to fall into the trap of concluding that it is imbalance of only this region or regions that has deleterious effects on development and function.
Is There a Down Syndrome Region? These considerations are directly applicable to both past and current attempts to identify a critical region for the DS phenotype. The sense of the most recent investigations is that duplication of the distal part of 21q22.1 and proximal part of 21q22.3,perhaps only of the latter (the status of21q22.2 is unknown), seems to be all that is required to cause the physical phenotype of DS (Fig. 2) [Pellissier et al., 1988; Korenberg et al., 1990a, b; McCormick et al., 1989; Rahmani et al., 19901. This phenotype is defined principally in terms of the characteristic facial and hand changes, congenital heart defect, and. of course. mental retardation. Do these findinas mean that the rest of chromosome 21 is irrelevant to tge phenotype? The is, for two reasons, clearly not. First, Of the proximal part Of chromosome 21q is not without effect [Williams et al., 19903. The most characteristic finding in the reported cases of du-
MENTAL RETARDATION
q22.1 q22’2
q22.3
II H1
“DOWN SYNDROME’
I IJ
Fig. 2. The “Down syndrome”region of chromosome 21, as defined by analysis of individuals with segmental duplications who manifest the major phenotypic abnormalities (facies, hand anomalies, mental retardation, and heart disease) of Down syndrome. However, imbalance of the proximal region of 21q is also known to cause mental retardation, and the region or regions responsible for other phenotypic anomalies of Down syndrome have not been defined.
34
Epstein
plications involving the region 21pter+q22.1 is some degree of mental impairment. This impairment may range from mild to severe, but it is usually associated with a n I& between 50 and 70 [Williams et al., 1975; Schinzel, 1984; Park et al., 19871.Reports of quite severe mental retardation from duplications of this region are somewhat difficult to interpret in view of the presence in some instances of other regions of imbalance (when translocation chromosomes are involved) or of other karyotypic abnormalities [asin Park e t al., 19871.Ascertainment bias may also be present. Nevertheless, if we consider cases with full trisomy 21, which constitute most cases of DS, i t must be concluded that imbalance of both the proximal and distal parts of 21q probably contributes to the overall degree of mental retardation. The second reason that the proximal and very distal parts of 21q cannot be considered irrelevant to the pathogenesis of DS is that there are more aspects of the DS phenotype than are encompassed in the existing clinical descriptions of individuals with duplications of 21q22. These include other major anomalies (particularly duodenal atresia and other gastrointestinal defects), neonatal hematological abnormalities, susceptibility to leukemia (especially megakaryoblastic), immunological deficits, a variety of cellular alterations (enhanced responses to P-adrenergic agents, decreased platelet serotonin, and possibly a n increased sensitivity to radiation, to mutagenic and carcinogenic agents and to virus-induced transformation),and, perhaps of greatest interest, Alzheimer disease (AD) [see Epstein, 1986 for summary]. The situation with regard to the relationship between DS and AD is of particular interest since the two major genes thought possibly to be playing a role in the pathogenesis of this form AD, the genes AAP and AD1 for the amyloid precursor protein and for familial AD, respectively, both seem to be in 21q21. Likewise, SOD1,the gene for CuZn-superoxide dismutase, for which a role in the pathogenesis of AD has also been suggested [Sinet, 19821, is in proximal 21q22.1, outside of the phenotypic DS region. There is presently no firm evidence to implicate any of these genes in the pathogenesis of AD in DS, and it is possible that the relationship between the two conditions is not attributable to the effects of any of them. Nevertheless, i t would be of great interest to know whether individuals with duplications of chromosome 21 centromeric to the distal part of 21q22.1, with and without the region of 21q22.1 that contains SODl,do or do not develop the pathology of AD when they reach the fourth decade of life.
NEED FOR COLLABORATIVE STUDIES At this point I would like to depart for a moment from the formal text to discuss a practical matter concerning the issue that I have just been addressing. As I have already said, considerable work is now being devoted to the molecular analysis of relatively small duplications of chromosome 21. These analyses will now supersede the less precise cytogenetic assessments of the amount and region of chromosome 21 that is involved in each of these duplications. However, just as crucial as the mo-
lecular delineation of the lesion will be the clinical definition of the phenotype of each individual whose DNA is being analyzed. And, by phenotype I mean not just the obvious manifestations that are usually considered, but also all of the other phenotypic changes that have already been alluded to. These analyses need to be done not only for individuals with clear-cut DS, but for all persons with duplications involving any part of chromosome 21. To make this possible, I urge that we establish a standardized and perhaps centralized means for data collection and for the analysis of manifestations such a s immunological status, sensitivity of cells to various agents, and the like. If we are really to succeed in mapping the phenotypic components of DS, it will be necessary for all of the workers in the field to collaborate with one another in obtaining comparable and meaningful data. There are already many precedents for such collaborative efforts in human genetics, and the field of DS research now seems to be ready for one.
THE SPECIFICITY OF ANEUPLOID PHENOTYPES I turn now to the second principle concerning the pathogenesis of aneuploid phenotypes, a principle that gets to the heart of the types ofmechanisms we should be considering to explain their development. This principle states that different aneuploid phenotypes, while often variable in expression and frequently overlapping in phenotype, are each specific and differentiable from one another. This principle asserts that there is a specific and deterministic relationship between imbalance of a particular region of the genome and the syndrome of malformations and other defects that occurs as a result of this imbalance. It argues against the proposition that most of the manifestations of aneuploidy may be interpreted as the results of a nonspecific disturbance of chromosome balance [BlumHoffman et al., 19871. The latter proposition, and others related to it, have been advanced for two reasons. The first is that there is a considerable degree of variability in aneuploid phenotypes, and any two individuals with the same aneuploid state do not necessarily have identical phenotypic abnormalities. Phenotypic variability is certainly the case for DS, but this still does not prevent us from recognizing when a n individual has trisomy 21. However, phenotypic variability should not really surprise us, since developmental processes and chromosomal aberrations do not occur in a vacuum. Except for identical twins, individuals with the same chromosome imbalance are genetically distinct from one another, and differences in the background genome have been shown in experimental animals to influence the ultimate phenotype [for discussion see Epstein, 1986, pp 183-1901. Even identical twins may not have identical phenotypes [Elder et al., 19841,since stochastic factors (for example, the time and place at which individual cells divide, or whether and in which direction they migrate) introduce a small element of randomness into the operation of a developmental pathway. Local or external environmental factors may have a similar perturbing effect. Kurnit and his collaborators [1985, 19871
Consequences of Chromosome Imbalance have argued strongly for a maj 3r role for stochastic factors in determining, once the genetic imbalance is present, whether or not congenital I- eart disease will develop in embryos with trisomy 21, and it is likely that such factors affect many morphoger letic processes.
35
euploid syndromes a s well a s we are able to if this were truly the case.
MECHANISMS OF ANEUPLOID PHENOTYPES If nonspecific effects of aneuploidy are rejected as being major determinants of aneuploid phenotypes, what types of pathogenetic mechanisms should we be considering, and how should we go about searching for them? This is a subject that I have discussed a t great length in a recent monograph [Epstein, 19861, and a t the present time I believe that it is fair to say that no mechanism has yet been shown to apply. Therefore, all types of mechanisms ought to be considered, although I think that certain systems and processes might be particularly vulnerable to the effects of aneuploidy. Among these I include intercellular interactions (such as recognition, adhesion, and communication); receptors, growth factors and morphogens, regulatory systems (including trans-acting factors); and multi-subunit macromolecules. For each of these it is possible to construct theoretical arguments and to use experimental precedents to demonstrate how relatively small changes in gene product concentrations could have quite large effects on function. Effects of aneuploidy on metabolic pathways must also be considered, but I suspect these are less likely to play a n important role, especially in trisomy. Nevertheless, as the work by Groner and his collaborators [1990] and by ourselves [Chan et al., 1988, Epstein et al., 19901 has shown, modest increases in CuZn-superoxide dismutase (SOD-1) activity, as are found in transgenic mice expressing inserted copies of the human SOD-1 gene [Epstein et al., 19871, do have effects on the sensitivity of cells to exogenous agents which perturb oxygen free radical metabolism, and perhaps even on the structure and integrity of the neuromuscular junctions of the tongue [Yarom et al., 19881. In considering how the presence of a n extra copy of any gene or set of genes might lead to some abnormality of development or function, it must be kept in mind that, unless there is evidence to the contrary, any mechanism that is proposed must ultimately be consistent with existence of a strictly proportional dosage effect for this gene or genes. In the case of trisomy 21 and DS, this means a 50% increase in the rate of synthesis of gene products coded for by chromosome 21 and, in general, a 50% increase in their concentration. Although autosoma1 dosage compensation does exist in Drosophila and plants, there is as yet no firm evidence that it exists in man and other mammals. However, a few possible examples have been suggested [Mangin et al., 1985; TothFejel et al., 19871, and i t is still important to determine whether a dosage effect is present whenever a new locus is mapped to chromosome 21. Since virtually all of the data on dosage effects in all species examined is derived from measurements of enzyme activities [for summary see Epstein, 1986, pp 65-84], analysis of the products of loci coding for other types of molecules would be of particular interest.
Phenotypic Overlap The second reason the issue (If nonspecificity arises is because many aneuploidy syndromes, DS included, share similar phenotypic abnormalities. Mental retardation and growth impairment are virtually universal when significant aneuploidy is wesent. In addition, any two chromosome abnormalitie 3 picked at random are likely to share a number of minor anomalies. Nevertheless, despite these overlaps, it is still possible to discriminate different and often quite ( losely related aneuploid phenotypes from one another with a remarkably high degree of accuracy. While such discriminations can readily be made by formal taronomic analysis [Preus and Ayme, 19831, this is rarely necessary, and a knowledgeable and astute clinical E eneticist can generally make the correct diagnosis. The crucial point, a s concluded by Preus and Ayme [19831, is that “although there are few if any physical fc atures which are exclusive to a particular chromoson a1 defect, the pattern of defects is distinct.” How do we explain these ovcdapping phenotypes? I would prefer to think, as statcd by Opitz [1982], that “human organs are evidently capable of responding to a high number of diverse dysmo~phogeneticcauses with the production of only a limited repertoire of malformations.” If we have learned no1,hing else from human biochemical genetics, we have certainly come to realize that many different enzymatic defects scattered throughout the genome can kave quite similar phenotypic effects. However, the other general explanation is that the effects of aneuploidy are nonspecific in that they may be produced by any type of chromosome imbalance. One mechanism that has Deen suggested for these nonspecific effects is “amplific d developmental instability,” in which there is a decreased regulation or “buffering” of morphogenetic processes and, as a result, a greater variability of outcome [Shapiro, 19831. In support of this proposal are data showing a n increased statistical variance of a variety of metric traits i n persons with aneuploidy [Shapiro, 1970, 1975; Langenbeck et al., 1984: Blum-Hoffman et al. 19871. However, not all traits show such a n increase, and it appears that increased variability and variance, when they do occur in DS and other aneuploid conditions, are quite selective with regard to which developmental systems are affected [Epstein, 19881. As I wrote earlier [Epstein, 13881, I have no difficulty with the notion t h a t particular aneuploid states may indeed affect the stability of ceri ain developmental processes and not of others. However, I do not think that the available evidence supports a conclusion that aneuploidy per se, independently cf which chromosome is involved, leads t o a generalizec disturbance of developCONCLUSIONS mental homeostasis which, in tiirn, produces most of the Based on the two general principles that have been phenotypic abnormalities of thc syndrome. It would just not be possible to discriminate among the different an- discussed and on many other theoretical considerations
36
Epstein
and experimental observations, I [Epstein, 1986, p 3721 have concluded that-
I t is legitimate to look for and to expect to find specific mechanisms to explain the relationships between particular phenotypic characteristics and the imbalance of individual loci or sets of loci. Contributions to the phenotype are likely to come from all parts of the unbalanced genome and not solely from j u s t one or a very few loci. While some loci may have a greaterphenotypic effect or representation than others, it is the cumulative effect of imbalance of many loci that determines the overall phenotype. As complicated as the relationships between genetic loci and phenotypic effects may be, and with whatever element of randomness they may be associated, these relationships should nonetheless exist and be discernible. T h e generation of aneuploid phenotypes is not, therefore, a game of chance but is an exercise of the conventional rules of developmental genetics and developmental biology. As I have already written, it has not as yet been possible to forge a mechanistic link between any phenotypic manifestation of a n aneuploid state and imbalance of a specific locus. This is, I feel, only a temporary impasse, since the necessary information and techniques for attacking the problem are now becoming available. Critical to this endeavor is the mapping of the human and mouse genomes, so that it is rapidly becoming possible to think in terms of the specific genes that are unbalanced in any type of aneuploidy, and particularly in trisomy 21. This, coupled with the rapid strides that are being made in developmental biology and neurobiology, now makes i t possible to approach the pathogenesis of the effects of aneuploidy from two directions: from imbalance of known genes to their potential phenotypic consequences, and from observed phenotypic aberrations to their likely genetic causes. When these two approaches ultimately converge, we shall truly begin to understand how genetic imbalance produces aneuploid phenotypes and, more importantly, how trisomy 21 causes Down syndrome.
ACKNOWLEDGMENTS This work was supported by NIH grants HD-17001 and AG 08938. REFERENCES Antley RM, Hwang DS, Theopold W, Gorlin R J , Steeper T, Pitt D, Danks DM, McPherson E, Bartels H , Wiedemann H-R, Opitz J M (1981):Further delineation of the C (trigonocephaly) syndrome. Am J Med Genet 9:147-163. Blum-Hoffman E, Rehder H, Langenbeck U (1987): Skeletal anomalies in trisomy 21 a s an example of amplified developmental instability in chromosome disorders: A histological study of the feet of 21 midtrimester fetuses with trisomy 21. Am J Med Genet 29:155-160. Carlin ME, Neadle MM (1978): Cri-du-chat syndrome-a correlation between anomalies and size of deletion. In Summitt RL, Bergsma D (eds): “Annual Review of Birth Defects, 1977.” New York: Alan R. Liss, Inc. for the National Foundation-March of Dimes, BD:OAS XIV(6C):428-429. Chan PH, Yu ACH, Chen S,Chu L, Epstein CJ (1988): Oxidative stress
exacerbates cellular damage in primary cultures from human SOD-1 transgenic mice. J Cell Biol 107:726a. Elder FFB, Ferguson J W , Lockhart LH (1984): Identical twins with deletion 16q syndrome: Evidence that 16q12.2-ql3 is the critical band region. Hum Genet 67:233-236. Epstein C J (1986): “The Consequences of Chromosome Imbalance.” New York: Cambridge University Press. Epstein C J (1988):Specificity versus nonspecificity in the pathogenesis of aneuploid phenotypes. Am J Med Genet 29:161-165. Epstein C J , Avraham KB, Lovett M, Smith S, Elroy-Stein 0, Rotman G, Bry C, Groner Y (1987): Transgenic mice with increased CuZnsuperoxide dismutase activity: An animal model of dosage effects in Down syndrome. Proc Natl Acad Sci USA 84:8044-8048. Epstein CJ, Berger CN, Carlson E J , Chan PH, Huang TT (1990): Models for Down syndrome: chromosome 21-specific genes in mice. I n Patterson D, Epstein C J (eds): “Molecular Genetics of Chromosome 21 and Down Syndrome.” New York: Wiley-Liss, pp 215-232. Frerman FE, Goodman SI (1989):Glutaric acidemia type I1 and defects of the mitochondria1 respiratory chain. In Scriver CR, Beaudet AL, Sly WS, Valle D (eds): “The Metabolic Basis of Inherited Disease.” 6th ed. New York: McGraw-Hill, pp 915-931. Groner Y, Avraham KB, Schickler M, Yarom R, Knobler H (1990): Clinical symptoms of Down syndrome are manifested in transgenic mice overexpressing the human CuiZn-superoxide dismutase gene. In Patterson D, Epstein CJ (eds): “Molecular Genetics of Chromosome 21 and Down Syndrome.”New York: Wiley-Liss, pp 233-262. Korenberg J R , Kawashima H, Pulst S-M, Ikeuchi T, Ogasawara N, Yamomoto K, Schonberg S, West R, Allen L, Magenis E, Ikawa K, Taniguchi N, Epstein C J (1990a): Molecular definition of a region of chromosome 21 that causes features of the Down syndrome phenotype. Am J Hum Genet 47:236-246. Korenberg JR, Pulst SM, Kawashima H, Epstein CJ, Allen L, Magenis E (1990b): Down syndrome: Toward a molecular definition of the phenotype. Am J Med Genet, this issue. Kurnit DM, Aldridge J F , Matsuoka R, Matthysse S (1985): Increased adhesiveness of trisomy 2 1 cells and atrioventricular malformations in Down syndrome: A stochastic model. Am J Med Genet 20:385-399. Kurnit DM, Layton WM, Matthysse S (1987): Genetics, chance, and morphogenesis. Am J Hum Genet 41:979-995. Langenbeck U, Blum E, Wilkert-Walter C, Hansmann I(1984): Developmental pathogenesis of chromosome disorders: Report on two newly recognized signs of Down syndrome. Am J Med Genet 18:223-230. Lazarow PB, Moser HW (1989): Disorders of peroxisome biogenesis. In Scriver CR, Beaudet AL, Sly WS, Valle D (eds): “The Metabolic Basis of Inherited Disease.” 6th ed. New York: McGraw-Hill, pp 1479-1509. Ledbetter DH, Cavenee WK (1989): Molecular cytogenetics: Interface of cytogenetics and monogenic disorders. In Scriver CR, Beaudet AL, Sly WS, Valle D (eds): “The Metabolic Basis of Inherited Disease.’’ 6th ed. New York: McGraw-Hill, pp 343-371. McCormick MK, Schinzel A, Petersen MB, Stetten G, Driscoll DJ, Cantu ES, Tranebjaerg L, Mikkelsen M, Watkins PC, Antonarakis SE (1989): Molecular cytogenetic approach to the characterization of the “Down syndrome region” of chromosome 21. Genomics 5:325-331. Mangin M, Ares M Jr, Weiner AM (1985): U1 small nuclear RNA genes subject to dosage compensation in mouse cells. Science 229:272-275. Opitz J M (1982): The developmental field concept in clinical genetics. J Pediatr 101:805-809. Park J D , Wurster-Hill DH, Andrews PA, Cooley WC, Graham J M J r (1987):Free proximal trisomy 21 without the Down syndrome. Clin Genet 32:342-348. Pellissier MC, Laffage M, Philip N, Passage E, Mattei M-G, Mattei J - F (1988): Trisomy 21q223 and Down’s phenotype correlation evidenced by in situ hybridization. Hum Genet 80:277-281. Preus M, Aym6 S (1983): Formal analysis of dysmorphism: Objective mechanisms of syndrome definition. Clin Genet 23:l-16. Rahmani Z, Blouin JL, Creau-Goldberg N, Watkins PC, Mattei J F , Poissonier M, Prieur M, Chettouh Z, Nicole A, Aurias A, Sinet PM, Delabar J M (1990): Critical role of the D21S55 region on chromo-
Consequences of Chromosome Imbalance some 21 in the pathogenesis of Dow 1 syndrome. Am J Med Genet, this issue. Rethore M - 0 (1981):Structural variation ofchromosome 21 and symptoms of Down’s syndrome. In Burgio (:R, Fraccaro M, Tiepolo L, Wolf U (edsj: “Trisomy 21. An Intern xtional Symposium.” Berlin: Springer-Verlag, pp 173-182. Schinzel A (19841: “Catalogue of Unb.ilanced Chromosome Aberrations in Man.” Berlin: Walter de Gruyter, pp 690-695. Shapiro BL (19701: Prenatal dental a iomalies in mongolism: comments on the basis and implications c f variability. Ann NY Acad Sci 171(2):562-577. Shapiro BL (1975):Amplified developmc ntal instability in Down’s syndrome. Ann Hum Genet 38:429-4:7. Shapiro BL (1983):Down syndrome-A jisruption of homeostasis. Am J Med Genet 14:241-269. Sinet P-M (1982): Metabolism of oxygt’n derivatives in Down’s syndrome. Ann NY Acad Sci 396:83-54. Summitt RL (1981): Chromosome specific segments that cause the phenotype of Down syndrome. In dc la Cruz FF, Gerald PS (eds): ‘“hisomy 21 (Down Syndrome).Resexch Perspectives.” Baltimore: University Park Press, pp 225-235. Toth-Fejel S, Magenis RE, Kolbe M, G a l i J , Boyd CD (1987):Normal
37
levels of type IV collagen mRNA and normal basement membrane structure in patients with loss of a type IV procollagen gene. Am J Hum Genet 41:A87. Williams CA, Gray BA, Hendrickson JE, Stone JW, Cantu ES (1989): Incidence of 15q deletions in the Angelman syndrome: A survey of twelve affected persons. Am J Med Genet 32:339-345. Williams CA, Cantu ES, Frias J L , McCormick MK, Antonarakis SE (1990): Clinical, cytogenetic and molecular evaluation of a patient with partial trisomy 21 (21qll-q22) lacking the classical Down syndrome phenotype. Am J Med Genet, this issue. Williams J D , Summitt RL, Martens PR, Kimbrell RA (1975):Familial Down syndrome due to t ( 10;21) translocation: Evidence that the Down phenotype is related to trisomy of specific segment of chromosome 21. Am J Hum Genet 27:478-485. Wilson GN, Chadarevian J-P de, Kaplan P, Loehr J P , Frerman FE, Goodman SI (1989): Glutaric aciduria type 11: Review of the phenotype and report of an unusual glomerulopathy. Am J Med Genet 32:395-401. Yarom R, Sapoznikov D, Havivi Y, Avraham KB, Schickler M, Groner Y (1988):Premature aging changes in neuromuscular junctions of transgenic mice with an extra human CuZnSOD gene: A model for tongue pathology in Down’s syndrome. J Neurol Sci 88:41-53.
