“What went wrong? And will it happen again?” These are the common questions from “chromosomal families” that bring people to the genetic clinic. We can recast these questions as: “Did I, or one of us, produce an abnormal gamete? If so, why? What gamete might be produced next time? Or, if the chromosomes were normal at conception, what went wrong thereafter?” To deal intelligently with these questions, the counselor needs a broad knowledge of how gametes form, how chromosomes behave, and how the early conceptus grows. In this chapter we consider the distinction between abnormality due to structural defect (full or segmental aneuploidy) and that due to functional defect (aberrant activation or imprinting status). Most of the chromosome abnormalities in individuals that counselors see in the clinic will have arisen from errors during formation of the germ cells, and we focus here particularly on meiosis, the specialized cell division of gametogenesis. Chromosome defects can arise postzygotically, and abnormalities of mitotic cell division in the cleavage stage embryo and in the embryo proper can produce chromosome mosaicism; we will review the possible consequences of this. We refer in passing to the concept of dynamic mutation, but leave its fuller discussion for the fragile X chapter.
First, we will look at etiology. We discuss three chromosomal settings within which genetic abnormality may arise, namely meiosis, mitosis, and genomic imprinting (and mention also the special category of trinucleotide-repeat transmission). Within each, we consider what types of abnormality may happen. In meiosis and mitosis, irregular segregations can produce aneuploidy for a whole chromosome, while asymmetric segregation of a structural rearrangement produces an incorrect amount of part of a chromosome (partial or segmental aneuploidy). In genomic imprinting, the defect is qualitative, with abnormal expression of what can be a normal amount of chromosome. Sometimes there is overlap—for example, a meiotic error can subsequently lead to an abnormality of imprinting. Sometimes we cannot be sure which is the correct category: a supposed meiotic error, for example, could actually have arisen in a premeiotic mitosis. Nevertheless, this format is not too arbitrary, and provides a useful framework within which the generality of chromosomal abnormality can be appreciated. Second, we shall consider pathogenesis, the process by which the underlying genetic defect then leads to phenotypic abnormality. Third, we will make some general comments about which categories of abnormality are likely to recur, and those for which sporadic occurrence is the rule.
Meiosis in Chromosomally Normal Persons
The purpose of meiosis is to achieve the reduction from the diploid state of the gonadal stem cell (2n = 46) to the haploid complement of the normal gamete (n = 23), and to ensure genetic variation in the gametes. The latter requirement is met by enabling the independent assortment of homologs (the physical basis of Mendel's second law) and by providing a setting for recombination between homologs. While we do not dwell on recombination per se, this is a raison d'être of the chromosome to the classical geneticist: “from the long perspective of evolution, a chromosome is a bird of passage, a temporary association of particular alleles” (Lewin, 1994).
The mature gamete is produced after the two meiotic cell divisions: meiosis I and meiosis II (Fig. 2-1). In meiosis I, the primary gametocyte (oocyte or spermatocyte) gives rise to two secondary gametocytes, each with 23 chromosomes. These chromosomes have not divided at the centromere, and remain in the double-chromatid state. In meiosis II, the chromosomes of the secondary gametocyte separate into their component chromatids. In the male, the daughter cells produced are the four spermatids, which mature into spermatozoa. In the female, the daughter cells are the mature ovum and its polar bodies. (In fact, it is not until sperm penetration that meiosis II in the ovum is completed.) Each gamete contains a haploid set of chromosomes. The diploid complement is restored at conception with the union of two haploid gametes. The moment of conception, as the embryologist sees it, is not at sperm penetration, but only when the two pronuclei have fused to form a single nucleus (“syngamy”).
Figure 2-1. Outline of chromosomal behavior and distribution during gametogenesis. Each primary spermatocyte divides symmetrically at the sequential meioses into four spermatids. Division of the oocyte is asymmetric, as it buds off the first polar body (PB1) at meiosis I, and the second polar body (PB2) at meiosis II. (The first polar body may or may not divide at meiosis II; as shown here, it has not).
Note that spermatogenesis divides the cytoplasm evenly, so that after meiosis II there are four gametes of equal size. The sperm head that penetrates the ovum comprises virtually entirely nuclear material; the tail is cast off. In oogenesis, cytoplasmic division is uneven, producing a secondary oocyte and first polar body after meiosis I, and the mature ovum and second polar body at meiosis II. The chromosomes of the first polar body may or may not undergo a second meiotic division (which would in any event be a pointless achievement). The ovum and its polar bodies each have a haploid chromosome set, but the ovum retains almost all of the cytoplasm.1 Another major sex difference concerns the timing of gamete maturation. In the female, meiosis is partway through, in the late prophase of meiosis I, by the eighth month of intrauterine life (the actual process of recombination takes place during weeks 16–19 of fetal life). At birth, there are around 3 million oocytes. Most of these disappear, but those destined to mature stay in a “frame-freeze” until they enter ovulation, some 10 to 50 years thereafter, and meiosis recommences. Testicular stem cells, by contrast, do not begin to enter meiosis until the onset of puberty. Thereafter, millions of mature sperm are continuously produced.
Figure 2-2. Closer detail of chromosomal behavior during meiosis I. One crossover has occurred between the long arms of one chromatid of each homolog. In oogenesis, one of the two cells at (d) would be the first polar body.
We now examine more closely the details of meiosis. During their final mitotic division in the gametic stem cell, the homologous pairs of chromosomes have (as with any mitosis) replicated their DNA to change from the single-chromatid to the double-chromatid stage. They now enter into the meiotic cell cycle (Fig. 2-2a). As meiosis I proceeds to prophase, chromosomes conduct a “homology search” and come together and pair, with matching loci alongside each other (Fig. 2-2b). This process, termed synapsis, continues with a more intimate pairing of the homologs, starting at the tips of the chromosomes and proceeding centrally (Barlow and Hultén, 1996), and the synaptonemal complex is formed. The paired chromosomes themselves are called bivalents2. Synapsis sets the stage for an exchange of matching chromosome segments; this is the process of recombination, or crossing-over (Fig. 2-2c). Next, desynapsis occurs (the diplotene stage), with dissociation of the synaptonemal complex and the formation of chiasmata. Now, the two homologous chromosomes disjoin and go to opposite poles of the cell. This is the anaphase stage; the orderly movement of chromosomes during this sequence is facilitated if synapsis, recombination, and chiasmata formation have proceeded normally. Finally, the cell divides into the two daughter cells (Fig. 2-2d). How the chromosomes are distributed—i.e., which chromosome goes to which pole, is called segregation. Normally, each daughter cell gets one of each of the pair of chromosomes, and this is referred to as 1-to-1 (1:1) segregation. Uniquely in the meiosis I cell division, daughter cells are produced with double-chromatid chromosomes.
These cells then enter meiosis II (with the possible exception of the first polar body, as noted above). In this cycle, the chromosomes do not replicate, because they are already in the double-chromatid state. The chromosomes separate at the centromere, and the resulting single-chromatid chromosomes disjoin, one going to each pole, resembling a mitotic division (Fig. 2-2e). The course of meiosis is discussed in much fuller detail by Miller and Therman (2001).
Chromosomal pathology arises when these processes of disjunction and segregation go wrong, as nondisjunction and malsegregation.
Meiosis in Chromosomally Abnormal Persons
Two main categories fall under this heading. The first, and most important, is the phenotypically normal person heterozygous for a balanced structural rearrangement (translocation, inversion, and insertion being the major forms). Second, there is the rare instance of persons who are themselves chromosomally unbalanced with either a full or a partial aneuploidy, and thus phenotypically abnormal, and who present with questions of their reproductive potential. We will deal with each situation in detail in separate chapters, but consider the broad principles here.
In heterozygotes for some balanced rearrangements involving only small segments, the chromosomes may “ignore” the nonhomologous material they contain, and pair (“heterosynapsis”) and segregate much as would happen at a normal meiosis. In other balanced rearrangements, the inherent tendency to pairing dictates that homologous segments of rearranged chromosomes will align, as well as they are able to achieve this (“homosynapsis”). This may require the chromosome to be something of a contortionist, forming complex configurations such as multivalents and reversed loops. According to either scenario, the stage is set for the possibility of unbalanced segregation. The gametes produced—and therefore the conceptuses that arise—are frequently unbalanced. In this context a segmental aneuploidy is usually involved, that is, a part of a chromosome is present in the trisomic or monosomic state; or, rather frequently, a combination of trisomy for one segment and monosomy for another. Partial trisomy and partial monosomy are also referred to as duplication and deletion, respectively.
In some rearrangements, recombination presents a further hazard. Inversions and insertions may produce a new recombinant (rec) chromosome that has a different genetic composition from that of the original rearrangement. A conceptus forming from it would inevitably be genetically unbalanced.
In the individual who has a full aneuploidy, and in whom gametogenesis is able to proceed, in theory a trivalent may form, or a bivalent and an “independent” univalent. Either could lead, effectively, to a 2:1 segregation. This appears actually to be the case in trisomy 21; whereas in gonosomal states (XXX, XXY, and XYY) the “third” chromosome is, as it were, disposed of, and the great majority of gametes are normal (see pp. 211 and 197). In the person with a partial aneuploidy due to an unbalanced rearranged chromosome, whether 46,(abn) or 47,+(abn), the abnormal chromosome may have an even (or near even) chance to be transmitted in the gamete. However, the opportunity to observe such outcomes rather infrequently arises. The picture is less straightforward in mosaicism.
MITOSIS AND MOSAICISM
The purpose of a mitotic cell division is to pass on faithfully an intact and complete copy of the genome to the progeny cells. The mitotic cycle consists of the following sequence: gap-1 period (G1) → synthesis period (S) → gap-2 (G2) → cell division. The G1 → S → G2 components together comprise the interphase period of the cell cycle. During the S period, the chromosomes replicate their DNA, thus converting from the single-chromatid to the double-chromatid state. Genetically active segments of chromosomes replicate earlier during the S period, while inactive segments, which include almost the entire inactivated X chromosome in the female, are late-replicating. The cell division period is further subdivided into prometaphase → metaphase → anaphase → telophase. The chromosomes condense to enter prometaphase, and condensation continues into metaphase. Metaphase chromosomes align on the equatorial plate, and the spindle apparatus becomes attached to the (bipartite) centromere of each chromosome, consisting of its two kinetochores. Pulled at the kinetochores, the chromatids of each chromosome then separate (disjoin) and are drawn in opposite directions (anaphase) and arrive at the opposite poles of the cell (telophase). Then, the nuclear membranes reconstitute, the cytoplasm constricts and divides, and two daughter cells now exist.
A mitotic error can cause phenotypic abnormality by generating an abnormal cell line at some point during embryogenesis. If we focus on the end result, the feature distinguishing mitotic from meiotic errors is that the former typically produce a mosaic conceptus, while meiotic errors produce a non-mosaic abnormality. We define chromosomal mosaicism as the coexistence, within the one conceptus, of two or more distinct cell lines that are genetically identical except for the chromosomal difference between them, these cell lines having been established by the time that embryonic development is complete (the point at which the embryo becomes a fetus). Thus, the different cell lines are fixed in the individual, and are a part of his or her chromosomal constitution. The earlier in embryogenesis that a mitotic error happens the greater the likelihood for a substantial fraction of the soma to be aneuploid, leading to increasing departure from normality of the phenotype. It is probable that many mitotically arising abnormalities lead to cell death, leaving no trace.
Chimerism, which is to be distinguished from mosaicism, is the coexistence of more than one cell line in an individual due to the union of two originally separate conceptions. It could be imagined that twin blastocysts happen to make contact and then fuse. Alternatively, there might have been a double (dispermic) fertilization of an ovum and a polar body. A 46,XX plus 46,XX chimera would most likely present as a normal female, whereas 46,XX plus 46,XY could manifest an abnormality of sexual differentiation. An extraordinary example is recorded in Wiley et al. (2002) of a malformed stillborn with 47,XY,+21 plus 47,XX,+12 chimerism. Less spectacularly, a normal woman with 46,XY in 99% of peripheral lymphocytes but 46,XX on other tissues, including ovary, had probably been the recipient of a twin-to-twin transfusion with marrow colonization from a male co-twin (Sudik et al., 2001). The chimera of classical mythology was herself “in the forepart a lion, in the hinder a serpent, and in the midst a goat.”3
Considering the enormous numbers of mitoses that proceed successfully, it is clear that the ordering of chromosomal disposition during cell division must be a marvelously robust mechanism. A complex system of interacting components underlies the mechanism, including the cohesin multiprotein complex, among which are the similar RAD21 and REC8 proteins having crucial roles in mitosis and meiosis, respectively, and the synaptonemal complex proteins (Buonomo et al., 2000; Gregson et al., 2001; Yuan et al., 2002). Rare instances of marked mitotic instability indicate the existence of errors in the system, some at least of which may be genetically determined (Mikkelsen, 1966), and the mosaic variegated aneuploidy syndrome (p. 308) is the classic example of a presumed “mitosis control mutation.” For example, Miller et al. (1990) karyotyped a child because of major physical and neurodevelopmental defects, and he had cells trisomic and monosomic for almost every chromosome; only about a quarter were 46,XY. As a possible milder manifestation of this phenomenon, Fitzgerald et al. (1986) described a mother, herself physically normal, who had had three trisomic 21 conceptions. Her own tissues (blood, skin) were mostly 46,XX, but some cells had a variety of aneuploidies (47,+21, 47,+18, 47,XXX), indicating a proneness to chromosome maldistribution apparently operating in both meiosis and mitosis. A very early example of familial mosaicism, back in the days of solid-stain cytogenetics, concerns a family in which a mother and son were mosaic for a Robertsonian translocation, and the son had two mosaic children, a girl with Down syndrome who had four separate cell lines, and a boy with 45,X/46,XY mosaicism (Zellweger and Abbo, 1965).
The quite common finding of loss of an X chromosome in an occasional cell in the normal female population (Horsman et al., 1987; Catalán et al., 1995) may reflect normal age-related anaphase lag, as may the similar loss of the Y chromosome in males (Kirk et al., 1994). Aviv and Aviv (1998) refer to “age-dependant hidden mosaicism” and propose a role for the progressive shortening of telomeres in leading to somatic aneuploidies in older persons. Possibly, this chromosomal change might be an agent, rather than just a passive consequence, in the ageing process.
If we do a chromosome test on any normal person—a routine analysis from a sample of peripheral blood—we would probably get a normal result (46,N). We would conclude from an analysis of a dozen or so cells from one specialized tissue that the rest of the soma is also 46,N. In most of the person's tissue, this will be truly the case. But the body comprises a vast number of cells—a trillion (1012) or so—which required a vast number of mitoses for their generation. The dozen cells checked in the laboratory are only a billionth of 1% of all the per-son's cells, and we routinely (and, for practical purposes, not unreasonably) regard this minute fraction as a valid representative of the remaining 99.999999999%. Notwithstanding, we can surely suppose that one or more errors will have happened during one or some of the many mitoses, and that these will have produced a chromosomally abnormal cell line and the person is really a chromosomal mosaic. It seems plausible to imagine that unrecognized islands of mosaicism involving a tiny number of cells— only a few thousand or a few dozen, perhaps— could well be a fairly frequent state. Perhaps everyone is a mosaic.
Mosaicism from the First Divisions of the Zygote
The first few mitotic divisions from the one-cell zygote are particularly vulnerable to error, and we need to consider this brief period of development separately. Insight into this vulnerability has come from experience in the IVF (in vitro fertilization) laboratory, with the application of preimplantation diagnosis. It may be that the early cleavage pre-embryo has to rely on an inadequate supply of maternal cell-cycle control factors, conveyed in the egg, before being able to bring about its own autonomous production. Surprisingly large fractions of pre-embryos are chromosomally mosaic. In one series, for example, in which 216 apparently normal IVF zygotes were followed through to the 2–4 cell and 5–8 cell stages, almost half were mosaic (Bielanska et al., 2002a). Going from the 2–4 cell to the 5–8 cell stage, the fraction rose from 15% to 50%. Often, the mosaicism was “chaotic,” that is, different cells had different aneuploidies. Pre-embryos that failed to advance had much higher levels of mosaicism than those whose development appeared to proceed smoothly, as naturally might have been expected. Munné et al. (2002b) studied over a thousand IVF embryos, and deduced that a substantial fraction reflected mitotic nondisjunction, these embryos typically consisting of cells with normal, trisomic, and monosomic chromosomal constitutions. Interestingly, this category of embryo mosaicism was associated with increasing maternal age, possibly reflecting the decay of stored factors in the oocyte just mentioned. One must bear in mind that all these observations are made in the unnatural setting of in vitro development, and that the picture may be less abnormal in vivo. These matters are dealt with in some detail in Chapter 24.
Insight can also be gained from inference in the study of mosaic individuals. Jacobs et al. (1997), in a study of Turner syndrome, observed that patients with Xq isochromosome mosaicism hardly ever have a 46,XX cell line: most are 45,X/46,X,i(Xq). This is what would be expected if the error happened at the very first mitosis of the initially 46,XX zygote. If it happened at the next two or three divisions, a 46,XX cell line would have been present, and presumably predominant. Similar inferences from observations in “corrected” trisomy, and in trisomy mosaicism (see p. 314), support the conclusion that the first division of the zygote is especially prone to error.
Dizygous twinning is more frequent in mothers in their late thirties, and so it is not remarkable that occasional instances are seen of co-twins, one with normal chromosomes, and the other with a maternal age-related aneuploidy. Monozygous twinning could happen in an abnormal conception just as in a normal one, and the occasional instance of twins concordant for an abnormal karyotype is to be expected (Schlessel et al., 1990). Rather more remarkable is the case of monozygous twins discordant for karyotype—clearly the adjective “identical” is inappropriate here! Rogers et al. (1982) studied monochorionic twin brothers, one 46,XY and the other 47,XY,21, in whom marker analysis supported a diagnosis of monozygosity. The skin fibroblast karyotypes were nonmosaic, but both infants showed blood mosaicism, presumably from twin–twin transfusion in utero. In this type of twinning, the assumption is that either an initially 47,XY,21 conceptus underwent splitting, with loss of a no. 21 then occurring in one of the newly created embryos; or, vice versa, a mitotic nondisjunction occurred in one monozygous embryo from an initially normal conception. Similarly, Nieuwint et al. (1999) describe two sets of monozygous twins, one of each pair being chromosomally normal and the other abnormal, monosomy X in one and trisomy 21 in the other. A scenario whereby monozygous twins could be of opposite sex is noted below (under Nondisjunction in Mitosis).
Somatic Recombination in Homologs
Genetic exchange can take place, as a (presumably uncommon) normal event, during a mitotic cycle, involving either the pair of homologous chromosomes or the sister chromatids of one chromosome. The cytogenetic demonstration of sister chromatid exchange (SCE) is rather dramatic (Fig. 19-2). Should the SCE be unequal, tandem duplication and deletion lines may be generated. If the deletion line is lost, a normal/duplication mosaicism results (Rauen et al., 2001). According to the somatic extent of the abnormal cell line, the phenotype may or may not be affected; and according to its involvement in the gonad, a reproductive risk may or may not apply.
Nondisjunction in Meiosis
Nondisjunction is remarkably frequent, and in consequence many human conceptions, perhaps one-third, are trisomic or monosomic. Nondisjunction is defined as the failure of homologous chromosomes to segregate symmetrically at cell division. The classical description of the mechanism of meiotic nondisjunction is as follows. In a chromosomally normal person, if the pair of homologs comprising a bivalent at meiosis I fail to separate (fail to disjoin4), one daughter cell will have two of the chromosomes and the other will have none. This is 2:0 segregation (Fig. 2-3a and Fig. 2-4, upper). In other words, one cell is disomic for that homolog, and the other is nullisomic. Nondisjunction may occur in meiosis II, meiosis I having proceeded normally. In meiosis II, it is the chromatids that fail to separate (Fig. 2-3b). Then, at fertilization, the conceptus ends up trisomic or monosomic, assuming the other gamete to be normal (Fig. 2-5a, b). Trisomy or monosomy in the offspring of normal parents is called primary trisomy or primary monosomy.
Figure 2-3. The classical view of the mechanics of nondisjunction. The asterisked cell reflects the complement of the oocyte in Figure 2-4 (upper). In oogenesis, one of the two cells following meiosis I would be the first polar body, which might or might not proceed to meiosis II.
Making inferences from the rates of recombination, the assumption is that most nondisjunction, at least in oogenesis, occurs at the first meiotic division.5 This is the conclusion reached with the most extensively studied chromosomes, nos. 8, 13, 15, 16, 18, and 21 (Nicolaidis and Petersen, 1998). Indeed, in the case of trisomy 16, every case is proposed to be due to a maternal meiosis I error. For the X chromosome, about 90% of nondisjunctions leading to the 47,XXX state are of maternal origin. At least half of the maternal X nondisjunctions that cause 47,XXX and 47,XXY arise from meiosis I errors, and about a third occur in meiosis II (Thomas et al., 2001). Meiosis II is the site of some autosomal nondisjunctions, and, in fact, most trisomy 18 may result from this stage. In spermatogenesis, the 22 autosomes have about an equal likelihood of undergoing nondisjunction, although some (nos. 16, 21, and 22) appear to be more vulnerable (Shi and Martin, 2000c). Only in trisomy 2 among the autosomes is there a substantial paternal contribution, with close to half reflecting a meiotic error in spermatogenesis (Hassold, 1998; Robinson et al., 1999b). The presence or absence of recombination is associated, in oogenesis, with normal disjunction or nondisjunction for at least chromosomes 15, 16, 18, and 21, the detail of which differs for different chromosomes (Thomas et al., 2001).
Figure 2-4. Oocytes at metaphase of meiosis II, showing nondisjunction of a G-group chromosome having occurred at the preceding first meiotic division. Upper, oocyte with classical nondisjunctional disomy, showing an additional G-group double-chromatid chromo some. The arrowed pair may be no. 21s, and the karyotype 24,X,+21. Lower, Oocyte with “predivisional” disomy, showing an additional G-group single chromatid. The arrowed pair may be no. 21s, and the karyotype 24,X,+21cht. (From Kamiguchi et al., 1993, courtesy Y. Kamiguchi.)
While autosomal nondisjunction overwhelmingly has its origin in oogenesis (and meiosis I at that), male gametogenesis has a major part to play in the malsegregation of sex chromosomes. Obviously enough, the 47,XYY state, when it is due to a meiotic error, must be due to paternal nondisjunction (at meiosis II, logically), with the production of a 24,YY sperm. A few may represent a postzygotic error (Robinson and Jacobs, 1999). As much as half of 47,XXY Klinefelter syndrome is due to nondisjunction having occurred in spermatogenesis, giving a 24,XY sperm. X-Y nondisjunction is predisposed following an absence of recombination in the primary pseudoautosomal regions (PAR1) of the X and Y at meiosis I; paternal age has little or no influence on this process, the degree of recombination showing no difference between younger and older men (Thomas et al., 2000; Shi et al., 2001b, 2002). Monosomy X is mostly due to absence of a paternally contributed sex chromosome, although the mechanism for this error is unlikely to be a simple obverse of the X-Y nondisjunction that causes 47,XXY (Hassold, 1998).
Figure 2-5. Aneuploid gametes producing an aneuploid conceptus (a and b), and aneuploid gametes producing uniparental disomy (c).
Sequential nondisjunctions at both meiotic divisions could lead to tetrasomy, and this is the basis of some X chromosomal polysomy (Hassold et al., 1990; Deng et al., 1991). Simultaneous nondisjunctions of two chromosome pairs can lead to double aneuploidy. The reader with a sense of history will want to review the 48,XXY,+21 case described in Ford et al. (1959); some 40 years later Chen et al. (2000a) listed the various combinations that have been reported. Their own case of 48,XXX,18 (diagnosed prenatally) resulted from nondisjunctions of X and 18 in maternal meiosis II. Simultaneous parental nondisjunctions, with both gametes being disomic, is rare but not unknown, and is another route to double aneuploidy; for example, Robinson et al. (2001) describe 48,+14[pat],+21[mat] in a spontaneous abortion. If one gamete is disomic and the other nullisomic for the same chromosome, one parent has contributed both members of the homologous pair, and the other none (Fig. 2-5c). This is uniparental disomy (UPD) due to “gametic complementation,” an event of extreme rarity. Complete nondisjunction is an expression that could be applied in the case of triploidy when this is due to the retention of the polar body within the ovum (Martin et al., 1991).
The Chromatid “Predivision” Hypothesis of Angell
An alternative mechanism for nondisjunction, put forth by Angell (1997), is based on the premise that precocious separation of chromatids (“predivision”) during meiosis I is the crucial factor. This hypothesis is based on direct oocyte observations. Three sequential events comprise the gist of this theory (Fig. 2-6). First, homologs fail to pair6 during meiosis I; if they do pair, they separate again before meiosis I is complete. In other words, instead of homologous pairs existing as a bivalent in meiosis I, they exist as two separate univalents. Second, these univalents are prone to “predivide”—that is, the separation of the two chromatids that should normally happen at meiosis II instead takes place while they are still in the first meiotic cycle. This could happen to both univalents or just the one. Third, at anaphase of meiosis I, these double- or single-chromatid chromosomes segregate to the oocyte and polar body independently. The oocyte in Figure 2-4 (lower) may be an example of asymmetric segregation due to this process, having received a double-chromatid and a single-chromatid chromosome. Sandalinas et al. (2002) provide some corroborative support for An-gell's hypothesis in their findings on direct analysis of fresh oocytes, with both predivision and nondisjunction being more frequent in the gametes of women over age 35 years than those under 35, and more often observed in the smaller chromosomes. Similarly, using oocytes obtained during IVF procedures, Pellestor et al. (2002, 2003) showed that predivision is a more frequent cause of aneuploidy than classical nondisjunction, and furthermore is related to increasing maternal age.
Figure 2-6. Nondisjunction following “predivision” of one homolog into its component chromatids in meiosis I (Angell's hypothesis). The asterisked cell reflects the complement of the oocyte in Figure 2-4 (lower). In oogenesis, one of the two cells following meiosis I would be the first polar body, which might or might not proceed to meiosis II.
Causes of Nondisjunction
Meiosis is a mechanical process, in which component parts need to separate, and to move. Proteins are produced whose role it is to control and to enable the process, and checkpoints are set up along the way. The great majority of aneuploidy due to nondisjunction arises in oogenesis. A particular vulnerability of maternal meiosis may lie in the degradation, over time, of factors that underpin the adhesion of the homologous chromatids of the bivalent. This failure of snug apposition, then, leads the chromosomes to adopt unstable positions when meiosis resumes. The particular unstable position will depend, where there is just one chiasma, on whether its site is towards the middle or towards one end of the chromosome. This can then allow the pairs of chromatids, only loosely attached to each other at this single chiasma, to act as independent univalents at the first meiotic division (Wolstenholme and Angel, 2000; Pellestor et al., 2002; Yuan et al., 2002). Another theory has it that the motor proteins associated with the centromere may be a point of vulnerability. Chromosomes move along the spindle to their appropriate destinations by the active intervention of these motor proteins, and if they are not working properly, chromosomes may end up being not where they ought to be (Hodges et al., 2002). Quality checking, which is stringently applied in the male, is poorly effective in the female, and so the maturing of an aneuploid oocyte is not prevented; as Hunt and Hassold (2002) suggest, Nature seems to have erred in putting less protective investment into the more scarce gamete.
While these meiosis-control factors may be the proximate cause of failed disjunction, what background attributes might lead to a loss in their integrity? Of course, advanced childbearing age is an obvious answer. A very telling insight comes from the work of Battaglia et al. (1996). These investigators sampled oocytes at meiosis II metaphase from younger (20–25 years) and older (40–45) volunteers who were having normal menstrual cycles. They did not look at individual chromosomes but rather at the disposition of the spindle and the metaphase chromosomes as a whole. They made the most striking findings according to the ages of the women: a symmetrical and neatly arrayed complex was seen in the younger women, while in the older women the spindle was all askew and the chromosomes a-jumble, as shown in Color Figure 2-7. (see separate color insert) It is not difficult to accept that this structural disorganization (or failure to have become organized, which Volarcik et al., 1998, call “congression failure”) would undermine the ability of the chromosomes of the oocyte to then undergo regular segregation. Thus, support may be drawn for the view that a gradual deterioration in the integrity of the meiotic apparatus with increasing age of the mother is the major factor in predisposing to meiotic nondisjunction in female gametogenesis. Individual variation may mean that some women are more prone to these effects than others.
Another factor contributing to maternal meiotic error may relate to the store of oocytes remaining in the gonad, with respect to the rate of attrition with increasing age (Kline et al., 2000). This is the concept of “ovarian reserve”: oocyte numbers peak at about 7,000,000 during the fifth month of intrauterine life, falling to 300,000 at menarche, with the loss rate accelerating in the 10 years prior to menopause, by which time the numbers are down to 1000 (van der Spuy and Alberts, 2002). A general marker of a reduced oocyte store is skewing of X chromosome inactivation, and Beever et al. (2003) have shown X-skewing in the context of recurrent miscarriage due to trisomy, and with trisomy detected at prenatal diagnosis—bear in mind that there is an association between trisomic abortion and premature menopause, and with removal or congenital absence of an ovary, each of these states conferring a diminished ovarian reserve. Yet a further proposition is that the accumulation of mitochondrial DNA mutations might compromise the functioning of the support structures underpinning the meiotic mechanisms (Schon et al., 2000), and Seifer et al. (2002) demonstrated an increasing frequency with age of a particular mtDNA deletion, which is otherwise known to be age-related, in the cells that surround the oocyte within the follicle (the granulosa cells). Thus, in summary, it may be a decay in cohesion and motor-control proteins and their supporting infrastructure with age, in combination with local factors (such as the remaining complement of oocytes in the ovary) and extra-ovarian factors (e.g., pituitary hormone levels) that alter with age, and with the final insult of poor quality control, that together affect the integrity of the ovarian meiotic process, and that make it more likely that the oocytes produced by older women will be aneuploid.
Nondisjunction in Mitosis
Mitotic (somatic, postzygotic) nondisjunction is the major mechanism in the causation of mosaicism. Nondisjunction can occur in an initially normal (46,N) zygote, with the generation of mosaicism for a trisomic and a concomitant monosomic line, as well as the normal line (Fig. 2-8a). In autosomal nondisjunction, growth of the monosomic cell line is severely disadvantaged, and it may well die out, leaving just the normal and the trisomic cell lines comprising the individual. Mosaic Down syndrome, with the karyotype 46,N/47,+21, is the classic example. In one autosomal trisomy, trisomy 8 mosaicism, somatic nondisjunction accounts for the great majority of cases, perhaps all of them (Robinson et al., 1995).
Figure 2-8. Generation of mosaicism. (a) Postzygotic nondisjunction in an initially normal conceptus. In this example, one cell line (monosomic 21) is subsequently lost, with the final karyotype 46,N/47,+21. (b) Postzygotic nondisjunction in an initially 46,XX conceptus, resulting in 45,X/46,XX/47,XXX mosaicism. (c) Postzygotic anaphase lag in an initially 47,+21 conceptus.
Actually, about 5% of standard apparently nonmosaic 47,+21 is also due to a mitotic defect (Antonarakis et al., 1993), with the “third” chromosome 21 equally likely to be maternal or paternal. In 3% of apparently nonmosaic 47,XXY and 9% of 47,XXX, the error was postzygotic, presumably prior to the formation of the inner cell mass (MacDonald et al., 1994). If there is a nondisjunction of an X chromosome later in embryonic life, both abnormal cell lines may remain—such as, for example, 45,X/46,XX/47,XXX mosaicism, which is presumed to come from an initially 46,XX zygote (Fig. 2-8b). More than one mitotic error can happen, separate in time and place; for example, DeBrasi et al. (1995) identified concomitant 45,X and 47,XX,+8 (and 46,X,+8) in a woman with clinical features of both trisomy 8 and Turner syndrome, in whom the molecular study supported the hypothesis of an originally 46,XX conception.
We referred above to the vulnerability of the first few mitoses from the zygote stage. A very early mitotic nondisjunction in a 46,N conceptus, with both chromatids of one homolog passing to the same pole, could produce trisomy in a cell destined to give rise to the pre-embryo. Subsequent loss of one homolog could restore a disomic karyotype; depending on which one, the disomy would be biparental or uniparental. Paternal UPD 15 (producing Angelman syndrome) is characteristically due to this sequence of events (Robinson et al., 1993a). A fraction of Prader-Willi syndrome, about 15%, is similarly due to postzygotic nondisjunction of the maternal chromosome 15, with uniparental disomy then resulting following the loss of the paternal chromosome.
Two sequential abnormal events can lead to the intriguing situation of nonidentical monozygous twins of opposite sex. The mitotic loss of a Y chromosome in a 46,XY conceptus could produce X/XY mosaicism; then a splitting of the conceptus could produce monozygous twins in which the distribution of the X and XY lineages might differ. Costa et al. (1998) studied two sets of X/XY mosaic twins, each set comprising a girl and a boy. In one set, the girl was diagnosed clinically as having Turner syndrome, and had a 45,X[23%]/46,XY[77%] karyotype on blood analysis, while her male co-twin was 45,X[16%]/46,XY[84%]. In the other set, the girl had mixed gonadal dysgenesis, and her karyotype was 45,X[24%]/46,XY[76%]. Her brother's karyotype, on blood, was 45,X/46,XY with the same ratios as his sister; but on fibroblast culture he was nonmosaic 46,XY. He had absence of one testis. Fetal blood mixing, due to vascular connections in the shared placenta, may have contributed to his 45,X cell line. If so, this would be a sort of pseudochimerism superimposed upon the mosaicism.
Nondisjunction can occur in a postzygotic mitosis in a conceptus that is initially trisomic for an autosome (say, 47,21). Thus, one copy of the homolog in question is lost. The same result may be due to the mechanism of anaphase lag. (In this latter case, the chromosome fails to connect to the spindle apparatus, or is tardily drawn to its pole, and fails to be included in the reforming nuclear membrane. On its own in the cytoplasm, it will form a micronucleus and soon be lost.) This converts the trisomy in this cell to 46,N. Its descendant cells are 46,N, and the karyotype of the conceptus is, say, 46,N/47,+21 (Fig. 2-8c). Most mosaic trisomy/disomy 13, 18, 21, and X arises in this way, e.g., 47,XXY → 46,XY/47,XXY (Robinson et al., 1995).
A conceptus with an “interchange tertiary trisomy”—that is, a 47-chromosome count, with the two translocation chromosomes and an additional copy of one of the derivative chro-mosomes—might generate a cell line with the balanced state, if one of the derivatives is lost postzygotically. Thus, a zygote with, for example, a 47,t(1;2),+der(1) karyotype might generate a cell line with 46,t(1;2). If this cell line included blood-forming tissue, but if much of the soma otherwise consisted of cells with the unbalanced state, a phenotypically abnormal child could have, on blood sampling, a balanced translocation karyotype. Such a case is presented in Dufke et al. (2001), and it is possible that this scenario could contribute to an explanation for the apparent excess of abnormal children among the balanced carrier offspring of translocation carrier parents (see p. 94).
Postzygotic “Correction” of Aneuploidy and Uniparental Disomy
If the conversion of trisomy to disomy, sometimes referred to as correction or rescue, occurs prior to the formation of the pre-embryo, and if the 46,N line then gives rise to the pre-embryo, the embryo will be nonmosaic 46,N.7 According to which one of the three chromosomes was lost, normal biparental disomy in the embryo could be restored, or UPD could result (Fig. 2-9). This is a much more common mechanism of UPD than the scenario of coincidental nondisjunctions in the genesis of both gametes as depicted in Figure 2-5c. It is at prenatal diagnosis, typically, that the fact of this rescue mechanism comes to be discovered, with trisomy seen at chorionic villus sampling (CVS), and disomy at a subsequent amniocentesis (Sirchia et al., 1998). Chromosome 15 is of particular concern, and Purvis-Smith et al. (1992) and Cassidy et al. (1992) provide historic illustrations in pregnancies showing 47,15 at CVS, with conversion to 46,N at amniocentesis; the infants had upd(15) mat, which produces the phenotype of Prader-Willi syndrome. Walczak et al. (2000) showed the same thing retrospectively, in demonstrating trisomy 15 by FISH on archived placental tissue.
Postzygotic correction can also happen in the other direction, as it were: conversion of a monosomic zygote into a disomic one. Quan et al. (1997) report a girl, 46,XX, with Duchenne muscular dystrophy due to a homozygous deletion of exon 50 of the dystrophin gene. She had homozyosity of the X chromosome for all of the tested marker loci—apparently, a complete maternal uniparental isodisomy X. Even a meiosis II nondisjunction would probably have had some heterozygosity, due to recombination at meiosis I, so Quan et al. propose a mitotic mechanism. A 45,XO conception, from a 22,O sperm + 23,X egg at syngamy, underwent duplication, or possibly nondisjunction, of the single X chromosome. Unfortunately, this X chromosome carried a de novo dystrophinmutation. It remains an open question as to how many monosomy X conceptions, the X chromosome being normal, are successfully corrected to produce normal 46,XX,upd(X) females (Schinzel et al., 1993).
The following structural rearrangements may be listed: translocations, insertions, inversions, isochromosomes, duplications, deletions, and complex rearrangements. All arose de novo at one point, whether with the index case in whom the abnormality was discovered, or in a parent or more distant ancestor, with a balanced (or, rarely, unbalanced) form transmitted thereafter in the family. Jacobs (1981) has derived the following mutation rates for the generation of de novo rearrangements: 1.6 × 10-4 for the balanced reciprocal translocation, and 2.9 × 104 for unbalanced rearrangements. The illegitimate breakage and reunion that produces these rearrangements is typically due to the apposition of chromosomal segments containing DNA sequences with a high degree of homology, or “paralogous sequences,” also known as “duplicons” (Ji et al., 2000; Peoples et al., 2000). Among these are the olfactory receptor gene clusters, which are widely spread throughout the genome (Giglio et al., 2001). Palindromic AT-rich sequences comprise the basis of “hot-spots” at 11q23, 17q11, and 22q11, leading to translocations t(11;22) and t(17;22) arising with breakpoints at these sites (Kurahashi et al., 2003). Most breakpoints are in nontranscribed DNA and thus, for the most part, contribute no untoward effect upon the phenotype. If the breakpoint in one chromosome disrupts a gene, phenotypic defect is the likely consequence. We discuss possible mechanisms of formation in more detail in the appropriate chapters.
Figure 2-9. Uniparental disomy from “correction” of a trisomic conceptus by loss of a homolog. Nondisjunction* at meiosis I, followed by postzygotic loss** of one homolog, causes uniparental heterodisomy. (If, for example, this were chromosome 15, and the meiotic nondisjunction occurred in the mother, the child would have Prader-Willi syndrome.) Nondisjunction at meiosis II would cause uniparental isodisomy.
Setting in Which Rearrangement Occurs
Mutations causing chromosomal rearrangement can occur during both meiosis and mitosis. Classically, meiosis has been considered the mutational setting par excellence; chromosomes are particularly active during meiotic synapsis and recombination, and it is plausible to suppose that error could readily creep in. Spermatogenesis, during which there are great changes in cell size as meiosis proceeds, may be a particularly hazardous environment; in fact, most de novo rearrangements arise in male gametogenesis (Shaffer and Lupski, 2000). Some of these mutations may have arisen after the second meiotic division is complete, but before the mature sperm has been produced. In Turner syndrome variant due to mosaicism with a structurally abnormal X, more often the abnormality arose during paternal meiosis (Uematsu et al., 2002). Oogenesis is not, however, immune (Giglio et al., 2001). If a conceptus is initially 46,rea, the abnormal chromosome may be discarded in fetal life, and the remaining homolog replicated (“compensatory isodisomy”), to give a 46,N karyotype at least in some tissues. But if the 46,rea cell line was operative during embryogenesis, the developmental damage it caused is irreversible (Bartsch et al., 1994).
Mitosis offers ample opportunity for mutation in terms of the number of individual cell divisions, and this would lead to the mosaic state, that is, 46,rea/46,N or 47,+rea/46,N. In fact, mosaicism for a structural rearrangement is only infrequently recognized.
Balanced and Unbalanced Rearrangements
Structural rearrangements can be balanced or unbalanced, with respect to the amount of genetic material per cell. Arguing somewhat circularly, in the phenotypically normal person it is inferred that, although such an individual's genetic material is in a different chromosomal arrangement, it is present in the proper (balanced) amount and functioning properly. It is irrelevant to the person's health, other than his or her reproductive health. It may be helpful in explaining this to think of the person's genome as a recipe book—a series of instructions for everything that is genetically determined. If an error occurs in the pagination (a translocation) and, say, pages 17 to 24 are inserted between pages 36 and 37, the recipes are all still there; they are still perfectly capable of being read. If a sequence of pages is inserted upside down (an inversion), one need only turn the book around to read them. At the cytogenetic level, one can only use the term apparently balanced; even the highest-resolution banding does not reveal a genetic imbalance at the level of a few kilobases of DNA. A problem that often cannot be resolved is the instance of a phenotypically abnormal individual who has an apparently balanced rearrangement. Might it be, at a submicroscopic level, that genetic imbalance really exists? If the (normal) parent has the apparently identical chromosome, the problem becomes more complex. Either some subtlety of the rearrangement can lead to phenotypic normality in the parent and abnormality in the child (e.g., Fig. 11-7), or the phenotypic abnormality is coincidental, and not the result of the karyotypic anomaly.
Duplications and deletions that once were at or beyond the limits of light microscopy now yield to dissection by molecular cytogenetics. With FISH methodology, comparative genomic hybridization, microarray analysis, or variation on these themes, previously unseen translocations have come to light (Biesecker, 2002). Most of the imbalances discovered by this approach have been subtelomeric deletions, these regions being, generally speaking, gene-rich. Clinical acumen remains of central importance: DNA technologies (e.g., Fig. 1-7) are not (yet!) suitable for screening of every case of “dysmorphism/developmental delay,” and skilled examination is necessary to enable a focused molecular-cytogenetic search.
EPIGENETICS AND GENOMIC IMPRINTING
A formal definition of an epigenetic effect includes these points: the DNA sequence of a particular gene remains unaltered, but the ability of this gene to be expressed is alterable following its transmission through the germline, or somatically, and this alteration is reversible. The expression “genomic imprinting” is applied in the setting of epigenetic effects that operate in germline transmission. Some parts of some chromosomes are subject to genomic imprinting, and this imprinting is parent-specific: that is, genes in the chromosomal segment are expressed, or not expressed, according to whether the chromosome had been transmitted in the sperm or in the ovum. An imprinted segment takes up an “epigenetic mark,” and the gene or genes in this segment are not expressed, leaving it to the corresponding locus or loci on the homologous chromosome to be the only source of expression. Imprinting therefore can lead to a differential activation status of the two alleles of the locus or loci concerned: one of the pair is functional, and the other is “silent.” When the phenomenon was first appreciated in humans, it was naturally suspected that many forms of congenital abnormality might be due to aberrant imprinting. The practical application of genomic imprinting in cytogenetics however appears to be confined to a rather small number of conditions. Nevertheless, the theoretical interest is considerable.
Most of the autosomal genome is not subject to imprinting and is functionally disomic. That is, with each locus having a pair of alleles, each of the pair is functionally active, contributing more or less equally to the genetic output from that locus. This is biallelic gene expression. A minority of the genome is subject to imprinting and requires only one of the pair of alleles to be active, while the other one becomes inactivated. In other words, the locus is functionally monosomic, with a genetic output from only one allele. This is monoallelic expression. If the allele of maternal origin is inactivated, only the allele of paternal origin is functionally active, and vice versa. Following conception, the imprint remains through cycles of postconceptional somatic mitoses: the chromosome “remembers” the sex of the parent who contributed it (put differently, it retains its epigenetic mark). The imprinting pattern may be specific to a certain tissue or to a certain developmental stage. Thus, some tissues may express monoallelically, while in other tissues biallelic expression is retained; or a tissue may express monoallelically at one stage in embryogenesis, and biallelically thereafter (Giddings et al., 1994; Ekstrom et al., 1995). X chromosome inactivation is a special case.
Imprinting is a normal mechanism of gene regulation. It is mediated through a process that takes place during gametogenesis, the physical basis of which includes methylation of cytosine bases within the gene(s), or within controlling sequences upstream of it. This process is reversible, and in the life of an autosomal allele or chromosomal segment, as it passes from individual to individual down the generations and across the centuries, imprinting— the epigenetic mark—will be acquired, maintained, lost (“erased”), reacquired (“reset”), and lost again according to the sexes of the individuals through whom it is transmitted. Throughout, it retains the same DNA sequence.
Mechanisms Whereby Functional Genetic Abnormality Can Arise
In the context of imprinting, we may consider three categories of functional genetic defect: uniparental disomy with overexpression and/or nonexpression; deletion with nonexpression; and relaxation of imprinting with overexpression. (1) Uniparental disomy will lead to either biallelic expression or no expression at the locus or loci within the imprintable segment. (2) If a deletion removes a chromosomal segment that would otherwise have been “silenced,” all that is lost is a nonfunctioning genetic segment, and there is no untoward consequence. If the deletion removes the segment on the active chromosome, however, the corresponding part of the other homolog is inactive, and so neither chromosome will be genetically functioning in this segment; in a sense, the silent allele is unmasked. (3) If, in a chromosomally normal conceptus, a segment that should have been imprinted loses its imprint (“relaxation” of imprinting), the locus or loci contained in it will be operating at double normal capacity.
Every counselor is familiar with the concept of the triplet expansion disorders. The first to have been recognized, in 1991, was Kennedy spinobulbar muscular atrophy, followed almost immediately by the fragile X syndrome. In the fragile X syndrome, the number of trinucleotide repeats can change as the allele is transmitted from parent to child. Having reached a critical number, the gene's function is compromised, and the phenotypic abnormality declares itself. The trinucleotide of cytogenetic importance is cytosine-cytosine-guanine (CCG). The rare folate-sensitive fragile sites (p. 239) are most likely due to expansions of naturally occurring polymorphic CCG sequences.
CONSEQUENCES OF GENETIC ABNORMALITY
Chromosome imbalances are harmful because of the fundamental reason that some (not all) genes are dosage-sensitive. In duplications, there is 150% of the normal amount of this chromosomal segment; and in the deletion, 50% of the normal amount. The imbalance involves a whole chromosome (full aneuploidy) or a part of a chromosome (partial aneuploidy, segmental aneusomy). An incorrect amount of dosage-sensitive genetic material in every cell of the conceptus distorts its development to a greater or lesser extent. Large losses or gains almost invariably set early development so awry that abortion occurs. Lesser imbalances may be compatible with continued intrauterine survival, but with the eventual production of a phenotypically abnormal child. Very minor imbalances may cause defects that are not readily detectable in early infancy, and some chromosomal “defects” may be without effect. However, as a first principle, anything but 100% of the normal amount of (at least autosomal) genetic material produces a less than 100% normal phenotype. Mental defect is the almost universal consequence of autosomal imbalance, and vice versa, much mental defect is due to a chromosome abnormality (Raynham et al., 1996).
It is generally too simplistic to think of duplication producing more of an effect, and deletion less of an effect, in terms of an opposite quality of eventual phenotypes resulting. But in some instances the concept of “type and countertype,” originally proposed by Lejeune (1966), may be invoked. Deletion of 7p15 may cause the cranial bones to fuse prematurely (craniosynostosis), due to abnormal behavior of osteoblasts at their periphery, whereas duplication leads to underdevelopment of the skull, with a large and confluent fontanelle (Stankiewicz et al., 2001c). Deletion of 15q26.1–qter (which removes the growth factor locus IGFR1) is associated with intrauterine growth retardation, whereas dup(15)(q26.1–qter) may cause a syndrome of postnatal overgrowth (Nagai et al., 2002; Faivre et al., 2002).
It is an obvious point, but worth restating: the defect in these aneuploid states involves too much or too little of what is normal chromosome material. The “third” chromosome in standard trisomy 21 is a perfectly normal no. 21 chromosome, with a perfectly normal complement of chromosome 21 genes. How, therefore, could it be that an additional amount of normal genetic message leads to an abnormal reading of that message? This is one of the great remaining unanswered questions of biology, which we touch upon in Chapter 16.
How do we determine what is a large or a small degree of genetic imbalance? First, we can take a quantitative approach—how much material is involved. The fine detail, in terms of megabases of DNA, is known for each chromosome band (Table 2-1). For practical purposes, the blunter measurement of haploid autosomal length (HAL) is useful (see Appendix A). The largest chromosome, no. 1, comprises 8.4% of the HAL while no. 21, the smallest, is 1.9%. As a very general rule, if the imbalance consists of less than 1% of HAL, the conceptus is often viable in utero, and live birth frequently results. If the excess is greater than 2%, abortion is likely. Imbalance involving autosomal deficiency (partial monosomy) is generally much less survivable than is duplication (partial trisomy).
Table 2.1. Actual Lengths of DNA Accommodated in Each Band, Measured in Bases, Using Chromosome 19 As an Example
Quantitative assessment is rather crude. In general, dark G-bands are low in gene content, and light G-bands are gene-rich; the telomeric regions have the highest gene density. More precisely, Saccone et al. (1999) have determined, on the basis of probing for regions rich in GC sequences, those parts of the chromosomes that carry a greater quantum of genes (Color Fig. 2-10; see separate color insert). Gene density is not the only factor: it is also a matter of whether the functioning of the genes in the duplicated/deleted segment is susceptible to their being present in an abnormal dose. Assessment is made, however, not on this theory, but on the empiric observations of phenotypes. Some segments (e.g., 9p, all of 21) appear to have a substantial pre- and postnatal survivability in the trisomic state, while a lesser number of segments (e.g., distal 4p) are sometimes viable when monosomic. Chromosome 13 provides the most impressive examples of viability for a large autosomal imbalance. Trisomy for the whole of chromosome 13—fully 3.7% of the HAL—frequently goes through to live birth, and in the 13q deletion syndrome, monosomy occurs for up to 2.5% of HAL. This presumably reflects a low gene density on this chromosome (Fig. 2-10) and a relative paucity of genes that are sensitive to dosage imbalance. The same likely applies to chromosomes 18 and 21. Occasionally, imbalance is so small that theeffect on the child's physical phenotype is only very minor, and intellectual function can remain within the normal range, albeit toward the lower end of that range. There are some segments which, when duplicated or deleted, appear to cause no abnormality at all. For example, trisomy for the segment 9p12–p21.3 (comprising as much as ~0.6% of HAL) was found in a physically and mentally normal man, who had “learned a technical profession” (Stumm et al., 2002). The concept of heritable “euchromatic deletions and duplications without phenotypic effect” is discussed in Chapter 15.
Other segments impart a serious trisomic/monosomic effect even for a tiny amount of chromatin. Deletion of the single band 17p13.3, for example, causes the severe phenotype of the Miller-Dieker syndrome. The least number of loci that can be removed in a deletion is one: in Rubinstein-Taybi syndrome, for example, the loss of a single copy of the gene CBP within 16p13.3 suffices to produce the phenotype (see p. 284). Differing lengths of deleted segments enable a dissection of the specific segmental contributions to components of an abnormal phenotype, and the example of 5p deletions provides an illustration (see Fig. 17-6). These various observations of karyotype–phenotype correlations permit judgment concerning whether a particular duplication or deficiency may be of minor, major, or lethal effect.
Certain chromosome regions, in the unbalanced state, are associated with particular types of malformation. A “malformation map” can be produced from documenting the association of certain congenital defects with certain segmental aneusomies (Brewer et al., 1998, 1999; Fig. 2-11). One of the most commonly seen defects is a heart malformation, even more so in fetal than in postnatal aneuploidy (Yates, 1999). The complex twisting and folding of the cardiac tube is, apparently, particularly vulnerable to an incorrect dosage of certain genes. These genes may reside in those chromosomal regions associated with heart defects, and van Karnebeek and Hennekam (1999) have documented these associations. A “susceptible cardiological karyotype” does not, however, necessarily determine that a heart defect will happen, as attested, for example, by the observation of discordant monozygous twins with a dup(4)(q28.3qter) (Celle et al., 2000). We have undertaken similar phenotype map studies with respect to kidney defects (Amor et al., 2003) and to epilepsy (Singh et al., 2002a), in hopes that the chromosome regions thus illuminated may serve as candidate regions for the discovery of renal and epilepsy genes.
Figure 2-11. A duplication–malformation correlation map. Some chromosomal regions, in the duplicated state, are particularly associated with certain types of malformation. Presumably, these regions harbor genes that have roles in the formation of these particular organs. Other regions (including all of chromosome 19) are unrepresented, and some of these may contain “triplo-lethal genes.” ACC, agenesis of the corpus callosum; ASD, atrial septal defect; AVSD, atrioventricular septal defect; PDA, patent ductus arteriosus; VSD, ventricular septal defect. (From C. Brewer et al., Am. J. Hum. Genet. 64, 1702–1708, 1999, courtesy D. R. FitzPatrick, © The American Society of Human Genetics, and with the permission of the University of Chicago Press.) A similar map has been drawn for deletions (Brewer et al., 1998).
Database resources. Two important hardcopy sources provide information on, firstly, the clinical features of specific duplications and deficiencies and, second, viability of a particular segment. Schinzel has compiled his Catalogue of Unbalanced Chromosome Aberrations (2001), an invaluable resource documenting the clinical phenotype in about 2000 different aneuploid states that are compatible with live birth, and many of which are associated with survival through early childhood or beyond. Stengel-Rutkowski et al. (1988) have gathered data from 1120 translocation pedigrees, determining for numerous segments in the partially trisomic and partially monosomic states the likelihood that a pregnancy would proceed through to live birth (referred to in detail in Chapter 4).
The newer methodologies noted above enable yet finer cytogenetic assessments to be made, and yet smaller deletions and duplications to be detected. Subtelomeric FISH probes are revealing previously unseen defects, and this cytogenetic group may comprise, in total, almost as much of the chromosomal basis of a retarded population as does trisomy 21; it may be appropriate that, in suitably chosen patients, subtelomeric FISH should be the next test to be done after a normal result from a standard G-band analysis (Knight and Flint, 2000; Dawson et al., 2002). There has been a steady stream of reports in recent times, as laboratories develop the necessary methodology, and clinicians the necessary diagnostic skills (Anderlid et al., 2002a; Baker et al., 2002a; Clarkson et al., 2002; Dawson et al., 2002; Van Karnebeek et al., 2002a). The importance of subtelomeric rearrangements may reflect the fact of these being in gene-dense regions, with at least some of these genes being dose-sensitive.8 It appears that somewhere in the region of 2%–10% of patients in the category of unclassified multiple congenital anomalies/mental retardation (MCA/MR) will prove positive. The detection rate has varied according to the ascertainment of cases. Those that have been more recently studied have a lower yield (since some “nearcryptic” deletions or duplications will already have been picked up by advanced G-banding). A positive family history increases the rate (since inherited rearrangements are enriched in this population). Other indicators may include growth retardation of prenatal onset, microcephaly, a degree of facial dysmorphism, and genital abnormality in the male (de Vries et al., 2001b), although not all authors are convinced of the usefulness of these clinical signs (Baker et al., 2002a). Persistence may be required: for example, Holinski-Feder et al. (2000) only found a subtelomeric t(3;16)(q29;p13.3) segregating with mental retardation in a large kindred after they had done a linkage exercise, which pointed to 16p. A number of originally “recessive” syndromes have since fallen to the power of this new cytogenetic methodology (Riegel et al., 2001). For example, the Pitt-Rogers-Danks and Lambotte syndromes are really due to a subtle 4p deletion and to a 4p/2q deletion/duplication, respectively (see p. 278).
The Sex Chromosomes
Sex chromosome imbalances need to be considered separately. Any X chromosomes in excess of one are genetically inactivated. Thus, indicating the inactivated X in lower case, normal females are 46,Xx, normal males are 46,XY, Turner females are 45,X, Klinefelter males are 47,XxY, and other X aneuploidies are 47,Xxx, 48,Xxxx, 48,XxYY, 49,XxxxY, and 49,Xxxxx. As for the Y chromosome, its active genetic material is confined to only a small segment, these genes being mostly related to sex determination and testicular function. Thus, in spite of the presence of one or more whole X or Y chromosomes in excess in the 47-, 48- and 49-chro-mosome states, in utero survival remains possible. Indeed, for 47,XXX, 47,XXY, and 47,XYY, survival is apparently quite uncompromised. Gonadal development in X aneuploid males is particularly affected, and intellectual function is jeopardized to a mild or moderate extent in the n ≥ 47 states in both sexes. 45,X has a high in utero lethality, although the small fraction surviving to term as females with Turner syndrome show in comparison a remarkably mild phenotypic effect.
Phenocopies. Similar phenotypes may flow from different genotypes. Pseudotrisomy 13 may be an autosomal recessive condition (Amor and Woods, 2000). The expression Di-George syndrome refers to an ensemble of signs that characterize the 22q11 deletion. Somewhat similar clinical pictures can be seen in deletions of 10p13 and of 4q34.2 (p. 282 and Tsai et al., 1999). The CHARGE (coloboma of the eye, heart anomaly, choanal atresia, retardation, and genital and ear anomalies) association is typically associated with a normal karyotype, but chromosomal imbalance is infrequently identified (Sanlaville et al., 2002). Some apparent genetic heterogeneity might fade if more stringent diagnostic criteria were to be applied.
THE MOSAIC STATE
Whether mosaicism matters depends upon which tissue, and how much of that tissue, is abnormal. If a majority of the soma is chromosomally abnormal, the phenotype is likely to be abnormal. If only a tiny fraction of some tissue were involved, in which the aneuploidy would have essentially no effect—if, say, some of the bony tissue of the distal phalanx of the left little toe were trisomic 21, and the rest of the person 46,N—it would never be known. Indeed, as mentioned above, it could be that many people carry a tiny and completely unimportant abnormal cell line somewhere in their soma. A very minor degree of mosaicism could still be important if a crucial tissue carried the imbalance. An abnormal chromosome confined to tissues of, for example, a localized area or cell type in one part of the brain, could theoretically cause neurological dysfunction. (Perhaps twenty-first century technology will devise a “quantitative cytogenetic MRI scan” that could map out brain regions with a disomic 21 and a trisomic 21 genetic complement!) Abnormality involving a gonad or part of a gonad (“gonadal mosaicism”) could lead to a child being conceived with that aneuploidy. Mosaicism confined to extraembryonic tissue may be without phenotypic effect, although it certainly causes anxiety if it produces an abnormal test result at prenatal diagnosis. Confined placental mosaicism (CPM) may exist unbeknown in pregnancies producing normal infants, as Lestou et al. (2000) showed in a study of 100 placentas, with 5 revealing CPM for trisomies 2, 4, 12, 13, and 18. Mosaicism may frequently be observed at the IVF laboratory in the early cleavage embryo, and of spectacular degree, with different cells having different aneuploidies—a state of affairs that becomes very relevant in preimplantation genetic diagnosis (Chapter 24).
Mosaicism for a Full Aneuploidy
As a general principle, an individual with an aneuploid line in only some tissues is likely to have a less severe but qualitatively similar phenotype to that of someone with the nonmosaic aneuploidy. The ascertainment of these individuals is biased: those with a more obvious phenotypic defect are, naturally, more likely to be detected. Mosaic Down syndrome— 47,+21/46,N—can be less obvious than standard trisomy 21, and with a lesser compromise of intellectual function. The existence of 46,N cells in some of the brain tissue presumably has a moderating effect. Some aneuploidies can only, or almost only, exist in the mosaic state, the nonmosaic form being lethal in utero. Examples of this are 47,+8/46,N and 47,9/46,N. If the distribution of the aneuploid cell line is asymmetric, body shape may be asymmetric, generally with the hypoplasia in regions of aneuploidy. De Ravel et al. (2001) described hemifacial microsomia (one side of the face being underdeveloped) and other body asymmetry in two children with autosomal mosaicism, one for trisomy 9 and the other for trisomy 22. The child with 47,XY,+22/46,XY had 9/10 cells 22 on skin fibroblasts from the arm on the right (underdeveloped) side, compared with 5/11 on the left arm (blood karyotype was 46,XY). Molecular analysis supported there having been a postzygotic anaphase lag that had produced the 46,XY line from an initially 47,XY,+22 conception. A surprising case is that of Greally et al. (1996): a child with mosaic trisomy 16 had a cardiac malformation and was otherwise (barring a unilateral simian crease) not dysmorphic, and her neurodevelopmental progress has been quite normal. One might suppose (but could not prove) that the trisomic cell line was somewhat confined in distribution, and excluded the brain. Mosaicism excluding the bone marrow will give a normal blood karyotype and, vice versa, mosaicism confined to marrow would be seen on routine peripheral blood analysis, but not on other samplings; mosaic trisomy 8 may provide examples in both directions. Examples of presumed very low–level trisomy mosaicism have come to light through prenatal diagnosis, such as “trisomy 13 mosaicism” in an apparently normal child with 1 cell out of 400 on cord blood (Delatycki et al., 1998). In sex chromosome mosaicism, fertility can exist when otherwise infertility is the rule—for example, in formes frustes of Turner syndrome with 45,X/46,XX and of Klinefelter syndrome with 47,XXY/46,XY.
Mosaicism for a Structural Rearrangement
We reviewed a three-decade experience in New Zealand, and only 12 cases of mosaicism for a structural rearrangement had ever been recognized, of which at least three were regarded as balanced and eight presumably unbalanced. This equated, crudely, to an incidence of around 1 case per year per 1,000,000 population (Gardner et al., 1994). This is likely to be an underestimate. Leegte et al. (1998) reviewed the literature and recorded 29 cases of mosaicism for a balanced reciprocal translocation. One would expect that this state would be without any phenotypic effect. The only practical implication would be if the mosaicism extended into the gonad, an example of which is provided by Shapira et al. (1997c). A man, himself quite normal, was mosaic for a pericentric inversion, 46,XY,inv(9)(p24q34.1)/46,XY, and he had a child with a recombinant chromosome that had caused a deletion 9p/duplication 9q syndrome. Some cases of “balanced” translocation mosaicism have been ascertained through the phenotypic abnormality of a proband, and it is arguable whether the rearrangement caused the abnormality or was coincidental (Aughton et al., 1993).
With an unbalanced karyotype, the broad rule applies that the mosaic form is likely to be less severe than the nonmosaic form. Pigmentary skin anomaly is a notable and clinically useful phenotypic trait that can characterize this type of unbalanced mosaicism, the important categories being hypomelanosis of Ito, linear and whorled nevoid hypermelanosis, and “phylloid” (leaf-like) pigmentary disturbance (Verghese et al., 1999; Noce et al., 2001). The distribution of the abnormal cells in hypomelanosis of Ito, and thus of dyspigmentation, follows the lines of Blaschko, and Magenis et al. (1999) use the expression “Blaschkolinear malformation complex.” Asymmetry is a further clinical pointer (Woods et al., 1994). “Functional mosaicism” for a structural rearrangement is exemplified by the X-autosome translocation in which different regions of the body have differing ratios of inactivation of the translocation and the normal X, and this also can lead to hypomelanosis of Ito (Hatchwell et al., 1996).
Tissue Sampling in the Detection of Mosaicism. Mosaicism can in theory be very widespread, and the distribution of the different cell lines can vary considerably. Kingston et al. (1993) described a fetal study in which several tissues taken post-termination had various fractions of mosaicism for an additional abnormal chromosome, including 88% of brain cells, while only 3% of amniotic fluid cells and no cells from a sample of fetal blood had the abnormal chromosome. One clinically useful and easy investigation that allows multiple site sampling, albeit in a restricted field, is the FISH version of the buccal smear. Reddy et al. (1999) studied a retarded woman with mosaicism for an “add(3),” whose true constitution proved on spectral karyotyping (of peripheral blood) to be 46,XX,der(3)t(3;14)(q29;q31)/46,XX. Using a 14q subtelomeric probe, they were able to show that 86% of buccal cells showed three signals, and so contained the der(3), while the 14% of cells with two signals were normal. In fact, this ratio was very similar to that of the peripheral blood, which was 83:17.
Mosaicism from Somatic Recombination Between Homologs
Probably, in the great majority of instances, and providing the exchange is balanced and not involving imprintable chromatin, somatic recombination would not adversely influence development. If the exchange between two homologs were uneven, mosaicism could be produced with duplication and deletion cell lines (see Fig. 17-2). The amount, mix, and survivability of the different cells throughout the soma will influence the extent to which normal morphogenesis is compromised.
Gonadal (and Somatic-Gonadal) Mosaicism
The classical view is that gametes with a chromosomal abnormality are typically produced by 46,N parents, whose gonads are chromosomally normal. The abnormality is presumed to have arisen at meiosis, and involves only the gamete(s) arising from that single meiosis. Indeed, Chandley (1991) has suggested that many structural rearrangements transmitted by the male may have arisen during a postmeiotic phase of spermatogenesis; a mutation at this site would involve just the one spermatozoon. If, however, an abnormality arises during formation of a germ cell prior to the onset of meiosis, an abnormal cell line can become established, and occupy a part of the gonad or gonads. This is gonadal mosaicism, a condition that cannot usually be recognized until after two siblings are born with the same de novo abnormality. The most direct demonstration would be by karyotyping gametes, and sperm karyotyping is a theoretical (some propose it should be a practical) means to identify paternal gonadal mosaicism ahead of having a second affected child. However, Brandriff et al. (1988) analyzed sperm from a 46,N man who had fathered two children with del(13)(q22q32), and found none with a del(13).
Cells destined to give rise to gametocytes originate from the yolk sac in early embryogenesis and migrate to the gonadal ridge on the dorsal wall of the abdominal cavity where, along with the supporting cells, they come to comprise the tissue of the gonad. In doing so, gametocytes must replicate many times, going through about 30 cycles of division. Thirty cycles produces 230 (about 1,000,000,000, a billion) progeny cells, and the potential for error exists at each cell division contributing to this population. These errors could be nondisjunctions or the production of structural rearrangements. Edwards (1989) has offered a startling insight into the actuality of gonadal mosaicism, pointing out that, in the male, the total length of seminiferous tubule is about 1 kilometer. If a mutation were to occur in a spermatogonium in, say, the 20th cycle of division, its progeny would then go through 10 more cycles, and comprise 210 (about 1000) cells. This would be only a millionth (1000/1,000,000,000) of the 1 km of tubule— a mere 1 mm. So a man mosaic in such a way would have a risk of only 1 in a million to father a conception with this particular abnormality. From similar reasoning, a defect arising at the 10th cycle could affect 1 meter of tubule, and thus carry a risk of 1 in 1000. Oogonia need to go through a lesser number of cycles but the same principles broadly apply.
Figure 2-12. A family exemplifying somatic-gonadal mosaicism. (a) Pedigree. The father had the mosaic karyotype 46,XY,del(1)(q25q31.2) /46,XY on lymphocyte study. Two children have the del(1) (q25q31.2) in nonmosaic state. The family was ascertained following routine prenatal diagnosis. The older del(1) sibling's development was judged, at age 5 years, to be in the low–average range; height, weight, and head circumference were in the range 20th–25th centiles. The father worked as an electrician. (b) Partial karyotype showing the father's two cell lines: two normal no. 1s, and one normal and one deleted no. 1. The segment 1q25–q31.2 is shown crosshatched. (Case of G. Dawson.)
If the abnormality arises in embryogenesis prior to the differentiation of the germ cell line, the soma may also be involved: this is somatic-gonadal mosaicism. Sachs et al. (1990) demonstrated ovarian mosaicism by direct gonadal samplings in a woman who had had one Down syndrome child and three other trisomic 21 pregnancies, whose blood karyotype was 47,21[3%]/46,N[97%]. Tissue cultured from ovarian biopsies showed almost half the cells in each ovary to be 47,XX,21. Figure 2-12 shows an example of somatic-gonadal mosaicism for a structural rearrangement. The index case was identified with a small intrachromosomal del(1) at routine prenatal diagnosis. The father was mosaic for this deletion, in 20% of lymphocytes. Of his two other children, one had normal chromosomes, and the other had the same deletion. The father is phenotypically normal, and the older child with the deletion has an IQ in the low–normal range. A similar circumstance is recorded in Fan et al. (2001): a father with the blood karyotype 46,XY,dup(8)(p21.3p23.1)/46,XY, who was a university-educated man working as a financial planner, had two children with 46,XX,dup(8)(p21.3p23.1). These girls had poor language development, clumsy motor abilities, and minor facial dysmorphism. If the proportion of abnormal cells in the mosaic parent were higher or differently distributed, that parent might manifest some signs of the partial aneuploid state. The father reported in Kennedy et al. (2001) had a dup(8)(p23.1p23.1) in mosaic state, in the ratio normal/duplication of 17:8 on blood sampling, and he himself had aheart defect, as did his nonmosaic dup(8) daughter. Her defect was, however, more severe than her father's: she had a fairly complex malformation, including a right-sided aortic arch, while he had only a right-sided arch. Notably, the daughter was described as “achieving top grades in school,” a very unusual phenotypic commentary on a child with a non-mosaic chromosome duplication.
About 1%–2% of placentas can have a different chromosomal constitution from that of the embryo, with usually the embryo being normal and the placenta trisomic. This is confined placental mosaicism. Thus, in 1%–2% of chorionic villus sampling (placental biopsy) there will be a potentially misleading result. Fortunately, these uncommon instances can, as a rule, be recognized as such, although not without causing some anxiety at the time. In a few confined placental aneuploidies, the function of the placenta may be compromised, and fetal wellbeing may be affected (see pp. 402, 409).
Amniotic Fluid Cell Mosaicism
Infrequently, true mosaicism is recognized at amniocentesis. Occasional cells with a chromosomal abnormality, if they are solitary or involving a single clone, are generally regarded as having arisen in vitro (artifactual mosaicism). At least most of the time, this is probably the correct interpretation. We consider placental and amniotic fluid cell mosaicism in detail in Chapter 25.
The idea that abnormality could be due to unequal parental contributions of an overall correct amount of chromosome material seemed most remarkable in 1980 when Engel first made the suggestion and coined the expression “uniparental disomy.” It came to be accepted fact. The two disorders that, par excellence, exemplify the concept of qualitative imbalance are Prader-Willi syndrome (PWS) and Angel-man syndrome (AS). The concept of genomic imprinting, discussed above, is central to an understanding of the etiology. Each syndrome is due to failure of expression of different (but closely linked) segments within the proximal long arm of chromosome 15. A PWS critical region is normally expressed from only one chromosome (functional monoallelism), in this case, the paternally originating chromosome. The maternal region is normally inactive and alleles in this region are not transcribed. If the paternal PWS region is absent, the maternal one cannot “fill the gap”; and this functional nullisomy is the root cause of PWS. An AS critical region, lying just a little distal from the PW region, likewise needs only monoallelic expression for normal phenotypic function. In this case, the maternal region is active, and the paternal region, having been imprinted, is inactive. If the maternal region is absent, there can be no genetic activity, and this causes the AS phenotype.
Absence of the paternal PWS region or maternal AS region can arise from two major mechanisms. First, in UPD, one parent fails to contribute a no. 15 chromosome, and the “correcting” presence of two copies from the other parent cannot restore a proper balance. This can be heterodisomy (the two homologs being different), or isodisomy (they are identical). Second, there can be a deletion within proximal 15q that removes a segment of chromatin containing the two regions. These issues are dealt with in some detail in Chapter 20.
As the imprinting story has evolved, it has emerged that several, indeed most, chromosomes appear not to be subject to imprinting. For these chromosomes, and with both homologs equally genetically active, regardless of the parent of origin, UPD will have no untoward effect. Or, if the UPD-contributing parent should happen to be heterozygous for a recessive gene, and if this is the isodisomy category of UPD, the child will display the condition concerned, due to homozygosity (“isohomozygosity”) for that recessive gene. Rare instances of this scenario are known.
Similar considerations may apply in the trisomies. Naturally, one parent must have contributed more than one homolog. Consider the example of Down syndrome: does the parent from whom the disomic gamete came contribute two different no. 21 chromosomes? In other words, does the child inherit a no. 21 chromosome from three of the grand-parents—a “heterotrisomy”? Or, does the parent contribute two identical (isodisomic) no. 21 chromosomes? Whether phenotypic differences stem from these different possibilities is uncertain, although Baptista et al. (2000) suggest that heterotrisomy 21 may of itself convey a greater risk for a specific heart malformation, ventricular septal defect, perhaps because of a damaging interaction of three subtly different protein products from a 21q “heart locus.”
Uniparental disomy for the entire chromosome set, or uniparental diploidy, has a devastating effect on development. If a conceptus has lost its maternal complement and the paternal complement is doubled, no embryo at all forms, and the chorionic villi are grossly abnormal. This is a hydatidiform mole (p. 353). If a 46,XX ovum at meiosis I attempts a parthenogenetic development, a grossly disorganized mass of embryonic tissue results: an ovarian teratoma (p. 313). If a triple set of chromosomes (triploidy) is present at conception, there is either a diploid maternal set plus a haploid paternal set, or vice versa. These different parental origins determine different very abnormal fetal and placental phenotypes (p. 259).
Segmental Uniparental Disomy
A mitotic mechanism that can lead to functional imbalance, if the segments exchanged are in a region subject to imprinting, is somatic recombination. The first shown example of this causing a dysmorphic syndrome is the segmental paternal (pat) UPD for 11p that underlies some instances of Beckwith-Wiede-mann syndrome (BWS) (p. 316), 11p being a segment that is normally maternally imprinted. In the partially UPD(pat) cell line, this segment is now expressed biallelically at distal 11p. The asymmetry of body growth in this syndrome reflects the distribution of two cell lineages: the normal biparental disomic line and the functionally imbalanced UPD(pat) line.
SPORADIC AND RECURRENT ABNORMALITIES
Chromosomally normal parents can produce abnormal gametes by nondisjunction, or by one of the other mutational mechanisms discussed above. The combination of factors that causes these defects in an individual case is unknown. No convincing case has ever been made for the agency of diet, illness, chemical exposure, or lifestyle factors in maternal no. 21 meiotic nondisjunction (Chapter 26), nor is there much support from epidemiological studies. Noting the similarity of Down syndrome prevalence rates worldwide, Carothers et al. (1999) commented that “the totality of published data could well be consistent with no real variation at all, and [this] might explain why a search for environmental factors associated with Down syndrome has been so unproductive.” The maternal age effect is of course important, indeed central, and any search for causes of chromosomal aneuploidy must take this into account. A plausible view is that there is a natural degeneration of the oocyte, as discussed above (and see Fig. 2-7). Simply put, eggs get older, and they show their age.
Chromosomes are plastic, dynamic entities, and cell division is a complex mechanical process; these qualities alone may suffice to endow the vulnerability that causes human aneuploidy and rearrangement. Given the assumption that all persons with intact gametogenesis are capable of producing an abnormal gamete, one view is that a certain background abnormality rate is intrinsic to the human species, and it is a chance matter whether this or that couple will have the misfortune to conceive the abnormality which, inevitably, someone has to bear.
Parental Predisposition to Nondisjunction? An alternative view is that some 46,XX and 46,XY people are more prone than others to produce chromosomally unbalanced gametes, at least with respect to full aneuploidies. An intrinsic fault in one parent in the mechanism of chromosome distribution at cell division could be the basis of the rare examples of multiple recurring defects, and given the complexity of the apparatus and process of cell division, it is logical that error-causing mutants would exist. Whether there might be milder alleles at postulated cell division loci, which could more widely be the cause of occasional nondisjunction, remains entirely a matter for speculation. Nevertheless, a geneticist could scarcely ignore the possibility that there might exist subtle genetic variation potentially setting the stage for nondisjunction. The recent work of D. Warburton (pers. comm., 2002) showing increased recurrence risks after a trisomic pregnancy (Table 16-2) is germane to this argument.
With regard to the ageing of the oocyte, might some women be more prone to have a trisomic pregnancy (for whatever chromosome) due to an accelerated ovarian ageing? The timing of menopause depends on the attrition rate of eggs: when the oocyte pool reaches a certain stage of depletion, the ovary can no longer maintain itself, in either its gamete-producing or its endocrine functions. Kline et al. (2000) studied the age of menopause in women who either had or had not had a trisomic pregnancy. The former reached menopause a year earlier than the latter. The investigators drew the conclusion that a woman's inherent rate of oocyte loss is related to her risk of producing a disomic egg, and thus a trisomic pregnancy. This might be due to a herd effect, with, for example, an oocyte's maturation being compromised in the setting of a falling oocyte complement in the whole ovary. Or, if there are fewer oocytes available as candidates for maturation, a faulty oocyte might simply have a greater chance of becoming mature. Does the case of a couple having sequential +21, +18, and +13 pregnancies, the mother being in her forties (FitzPatrick and Boyd, 1989), illustrate some form of meiotic “weakness”? Or was this simply pure bad luck three times running?
In the male, Rubes et al. (2002) propose that some men may have a consistent tendency to both meiotic and mitotic aneuploidy, although the increased levels are very small. They group sperm aneuploidy into three categories: those arising as rare sporadic events, those due to extrinsic factors (such as chemotherapy), and those from the consistent aneuploid producers due to intrinsic predisposition. A slight increase in sperm disomy 13, 21, 22, and XY reported in fathers of 45,X daughters might reflect this latter group (Soares et al., 2001b). These observations on the sexes offer hints, and little more, that gametogenesis in some persons may be more susceptible than in others.
Actual biological components that might underlie such a susceptibility have been pursued, including variation with respect to the following four factors: (1) mismatch repair genes, (2) a meiosis-control protein, (3) the centromeric α-satellite, and (4) folate metabolism. The mismatch repair genes are well known as the basis of the Mendelian disorder hereditary nonpolyposis colorectal cancer (HNPCC). Given observations of infertility in homozygous mice, Martin et al. (2000a) analyzed sperm from a number of men from a large Newfoundland HNPCC kindred who were heterozygous for an MSH2 mutation. They had slightly higher rates, from 1.4 to 1.8 times, of sperm with disomies 13 and 21, XX, and diploidy, compared with a control group. None had a history of reproductive abnormality.
With regard to other meiosis-control proteins, Yuan et al. (2002) bred mice lacking the gene for synaptonemal complex protein 3 (SCP3), this gene normally being expressed during prophase of meiosis I. The rate of oocyte and zygote aneuploidy rose considerably, associated with a greater number of achiasmate meiotic chromosomes (again bringing to mind Angell's hypothesis of pre-division), and the effect was more marked in older mice.
A possible role for the differences in the structure of the centromere is suggested by Maratou et al. (2000). The length of the α-satellite (alphoid) arrays might plausibly have an influence on the integrity of the disjunctional process, these sequences comprising the major molecular component of the centromere. Of course, the centromere is crucial in directing the movement of chromosomes at cell division. Maratou et al. assessed the length of the α21-I array in parents of children with trisomy 21 and that in controls, and observed that meiosis I non-disjoining mothers had the smaller of the no. 21 chromosomes with a notably small α21-I length (mean size 835 kb, compared to 1293 kb in the smaller chromosome of normally disjoining fathers). It is speculated that small alphoid size on one chromosome 21 might result in a premature loss of sister chromatid cohesion (Pellestor et al., 2002).
Several groups have assessed maternal polymorphism for certain folate metabolism enzymes, with somewhat conflicting results, and a link with aneuploidy appears inconclusive (Hobbs et al., 2000; Hassold et al., 2001; O'Leary et al., 2002; Stuppia et al., 2002; Yanamandra et al., 2003). There was no change in prevalence of trisomy 21 in Canada comparing before and after the institution of folic acid fortification of flour in the late 1990s (Ray et al., 2003).
Figure 2-13. Chromosomes at synapsis exist as double-chromatid structures (e.g., the reciprocal translocation quadrivalent at right). But, for simplicity, we generally represent them with just the one chromatid (left).
A NOTE ON THE DIAGRAMS. Following the progress of rearranged chromosomes during meiosis is not easy, so we have taken some liberties in simplifying the accompanying diagrams. Most of these diagrams depict the synapsing chromosomes at meiosis with just one chromatid; of course, the chromosome has actually replicated at this point and exists as a double chromatid entity (Fig. 2-13).
1. Cytoplasm contains the mitochondria, and transmission of mitochondrial DNA is (almost entirely) maternal. The mitochondrial genome has been described, somewhat whimsically, as chromosome 25, or the M chromosome. In not otherwise referring to this “chromosome,” we are not seeking to deny its importance or interest!
2. Since, at the level of the chromatid, there are four el-ements, the word tetrad can also be used in this setting. In one sense, the cell at this stage of the cycle has 23 × 4 = 92 chromosomes. At the molecular level, the number of single DNA strands is 8.
3. Somewhat stretching the analogy, Bianchi (2000) makes the intriguing suggestion that, because of the retention and persistence of fetal blood cells following delivery, every mother is, in a sense, a chimera. Bazopoulou-Kyrkanidou (2001) offers an elegant essay, “Chimeric Creatures in Greek Mythology and Reflections in Science,” for the reader of broader view.
4. Note that disjunction is a normal process, and nondisjunction is not; there is no such word as “dysjunction.”
5. Wolstenholme and Angell (2000) point out that these assumptions about the meiotic stage may be mistaken. Verlinsky et al. (2001a) have drawn a yet more controversial inference from their studies of polar bodies from oocytes obtained for IVF. Meiosis II errors were interpreted to occur with a frequency similar to that of meiosis I errors, at least for chromosomes 13, 16, 18, 21, and 22. However, while direct in vitro observations on oocytes and polar bodies certainly have an attractiveness, artifact remains a possible confounder (Dailey et al., 1996), and we should be cautious about discarding the constructions assembled from indirect (but in vivo) analysis.
6. If the homologs had never joined together, then theycould not disjoin. In that sense, nonconjugation might be a more accurate word than nondisjunction, a point Sturtevant and Beadle (1962) made many years ago. Following Angell, predivision may be a proper word. Nonetheless, nondisjunction is well entrenched in the genetic lexicon, and its general meaning of “the inclusion of both daughter chromosomes in the same nucleus, by whatever mechanism” (Miller and Therman, 2001) is well accepted.
7. Some aneuploidies with a degree of viability, such as trisomy 18, may allow retention of the abnormal cell line in the inner cell mass (Magli et al., 2000).
8. Chromosome 16p is an exception to this rule, at least considering the distalmost part of the subtelomeric region (Horsley et al., 2001). The genes here may be, as it happens, dosage-insensitive. In distal 6q, the subtelomeric region may be gene-sparse, and the phenotypic consequence due to deletion rather mild (Kraus et al., 2003b).