Chromosomes in trisomy 21

Jérôme Lejeune

Annals of the new york academy of sciences Volume 171, Article 2, Pages 381-390. September 24, 1970.


So much work has been devoted to the chromosomal study of trisomy 21 that it seems unnecessary to add to it. Nevertheless, the organizers of this conference have rightly believed that research was probably not complete, since we do not yet know why, how and when the chromosomal aberrations appear. Before focusing our attention on this most important point for the understanding of the causes of the disease, we can briefly summarize the general data now established in regard to many thousands of trisomics.


General statements

To date, all cases of clear-cut clinical 21 trisomy syndrome have been carriers of extra chromosomal material. In all the cases in which identification was possible, either by cytological or genealogical analysis, the extra fragment was chromosome 21. It is then entirely correct to assume that the morphological and biochemical syndrome results from a pure gene dosage effect.

The second general statement we can accept is that the risk of the disease increases exponentially with the aging of the mother, as first recognized by Shuttelworth [1] and carefully analyzed by Penrose [2]. This aging factor is related exclusively to the mother; parity and father's age play no detectable role.

If trisomies 21 are classified by their karyotypes, it appears that at least 95 % are due to de novo free trisomies; the rest are related to translocation trisomies. Among these, around half are newly arisen mutations and half due to abnormal segregation of a preexisting familial balanced translocation. Interestingly enough, the risk for the progeny of a translocation carrier is much higher if the carrier is the mother; however, no maternal age effect has been detected in this category.

Our purpose in this paper is to investigate whether these apparently disparate data can be fitted together into a coherent picture so that a tentative answer can be given to our three basic questions, why, how and when ?


Origin of the extra 21 in free trisomies


General Morphology

If we could easily recognize each chromosome 21 individually, we would immediately detect the origin of the extra one and determine more exactly whether the malsegregation occurs in the male or the female gamete. Unfortunately, the situation is not so easy. First, chromosomes 21 and 22 are very similar. In exceptionally good preparations, the five acrocentrics can be split into two obvious groups; such a classification, however, is often not possible. Still more troublesome is the fact that if the three 21 chromosomes are clearly detected, it is due to their intraclass similarity; thus there is no possibility of differentiating the one coming from one parent from the two others. Many speculations have been made in regard to the satellites, for it sometimes happens that one of the parents has large satellites on only one of these 21 chromosomes. Unfortunately, no case has yet been reported in which the trisomic received two large-satellited 21 chromosomes.


The Meiotic Analysis

First Division Error

If for the sake of simplicity the occurrence of crossing-over is neglected, malsegregation occurs if rupture of the synapses does not take place early enough. The two duplicated chromosomes migrate at the same pole and, as a consequence, the two chromatids included unduly in the egg are homologous; i.e., one is received from the father and the other from the mother of the individual in which meiosis takes place.

Fig. 1. Normal meiosis: 1- first division; synapses rupture, but centromeres do not split; 2- second division; cleavage of centromeres.

Fig. 2. First division error. Synapses do not split; both 21s are homologous.

Second Division Error

Again, neglecting crossing-over, the accident occurs when the centromere does not split before anaphase. The two chromatids migrating together to the same pole are thus sister chromatids.

In the first accident, the homologous chromatids are independent insofar as their allelic content is concerned. In the second type, on the other hand, the allelic content of sister chromatides is by necessity identical. It follows, as has been shown independently by Penrose [3] and Gremy [4], that quantitative characters would be differently affected by the two types of accidents and that parentchild or sib-sib correlations would be different.

Another conclusion can be drawn. To avoid a full mathematical demonstration which is quite simple but rather cumbersome, we can focus on the problem of the mean and the variance. Let us suppose that a gene has two alleles in the population, a and b, the effect of these alleles being exactly additive, so that as people have a (2a) value; ab people have a (a+ b) value; and bb people have a (2b) value.

The three phenotypes have the frequencies p2, 2pq, and q2, if p is the frequency of allele a and q is the frequency of allele b; i.e., p2 (aa) + 2pq (ab) + q2 (bb). It follows that the expected mean is M=2(pa+qb), with a variance V=2pq(a- b)2.

In a trisomic if the extra chromatid is a homologue of the other (first division accident), the three chromosomes 21 are independent and the distribution of the aaa aab abb and bbb types would be given by the expansion of (p+q)3; i.e., p3 (aaa) + 3p2q (aab) + 3pq2 (abb) + q3 (bbb). The expected mean is thus : M=3(pa + qb) and the variance V=3pq(a-b)2.

If, however, the extra chromatid is a sister of the other, the trisomic contains only two independent sets of alleles, one of the chromosomes being a copy of another. Hence the four genotypes and phenotypes will be given by the expansion of (p + q)2 and have the frequencies p2 (aaa) + pq (aab) + pq(abb) + q2 (bbb) . The mean is obviously 3(pa + qb), as shown by elementary calculus, however, the variance is now V = 5pq(a-b)2; thus it is much greater than in the preceding case.

From these formulas, we can deduce that the ratio V/M should be the same in normals and in trisomics by first meiotic error: Vn/Mn(normals) = 2pq(a - b)2 / 2(pa + qb) = V1/M1(trisomics) = 3pq(a - b)2 / 3(pa + qb).

On the other hand, the ratio for the second meiotic error, V2/M2 = 5pq(a - b)2 / 3 (pa +qb), is expected to be 5/3 higher than the V/M ratio for either normal or first meiotic error.

By analysis of available data on alkaline phosphatase, galactose uridyl transferase and kynurenine excretion, it can be shown that the V/M ratio is higher in trisomics than in normals, thus pointing toward a second division error.

It can be seen intuitively and demonstrated algebraically, that if crossing-over occurs the V2/M2 parameter of the second division will tend to resemble the V1/M1 parameter of the first division error. The phenotypic distribution would still be different but the discrepancy would be less striking; it follows that the increased V/M ratio in trisomics does not prove that all cases are second division errors, only, that a sizable number of them must be. This is in agreement with the occurrence of mosaics, which are mitotic events, with the same phenotypic result as second division error (if crossing-over is overlooked).

From this statistical aspect we can conclude that possibly a very high number of trisomics are carriers of sister chromatids and, since free trisomics alone seem to be sensitive to the aging of the mother, we can infer that aging must act somehow on the segregation mechanism, impairing either the first or the second division-predominantly the second.

Fig. 3 . Second division error. Centromeres do not split; both 21s are sister chromatids.


The Marker Chromosome

If these deductions were true, we should find cases in free 21 trisomics in which the extra chromosome is demonstrably a copy of the other.

Recently, Dr. Jean de Grouchy came across such an instance, and he was kind enough to allow me to quote these unpublished data. A typical trisomic 21 has two Gpo chromosomes in his karyotype. His mother is carrier, in the heterozygous stage, of the Gpo marker. This Gpo must be a 21, for the child who has received two of them is a typical 21 trisomic. Thus, we have here the proof that noncleavage of the centromere at the second division is the cause of the disease, at least in this instance.


Origin of the free 21 in the translocation trisomies


The G ~ D Translocation

Without reviewing all kinds of translocations involving chromosome 21 we can focus on the most common type, the translocation between acrocentrics. If we exclude the very rare 21 ~ 21 type, which can only produce 21 trisomics or monosomic 21 offspring, the effects are comparable no matter whether the translocation is a 21 ~ D, a 21 ~ 22 or a 21 ~ Z [5].

In observing the most frequent type, the 21 ~ D, it is remarkable that among the newly arisen mutants for 21 ~ D translocations, most of them are also carriers of 21 trisomy as demonstrated by Dutrillaux [6]. This phenomenon could be expected if the exchange occurs at the four chromatids stage. Moreover, if the translocation process is akin to a chromatid exchange and has a tendency to stabilize the segregation, most of the newly arisen translocations should produce a disomic 21 gamete, thus explaining why de novo balanced translocations are comparatively rare.


The Origin of the Extra 21

During meiosis in a translocation carrier, the error could occur at either the first or the second meiosis. If it occurs at the first meiosis, the resulting trisomic would be a carrier of the translocation and of the free 21 homologue. If it occurs at the second meiosis, the trisomic would either be a free trisomy or a trisomy for both 21 and D, due to the nondisjunction of the centromere of the translocated chromosome.

Among more than 300 children born from a D ~ G translocated parent, all the trisomics but one were carriers of the translocation. Hence, all of them resulted from a first meiotic error, and the extra 21 was by necessity a homologue of the translocated one (if crossing-over is not taken into consideration). According to our preceding considerations, the ratio of the variance to the mean should thus be greater for free trisomics than for translocated trisomics if a quantitative character is analyzed. From the recent data of Jerome (this monograph), a simple calculation shows a deviation in the expected sense, albeit not a significant one.

Fig. 4. Normal meiosis in a G/D translocation carrier.

Fig. 5. First division error in a G/D carrier. In the trisomic, carrier of the G/D, both 21s are homologous.


The Risk of Malsegregation

To estimate correctly the risk of the occurrence of a 21 trisomic in the progeny of a translocation carrier, a careful statistical analysis of the published data must be worked out. Not only must the bias of ascertainment by a patient be corrected, but also the likelihood of detection of the whole genealogical tree must be accounted for. As a general conclusion, confirming earlier findings, it can be estimated [2] that the risk of trisomy in the progeny of a woman carrier of a D/G is approximately 1/5 or 1/6. This is definitely lower than the 1/3 expected in the case of random segregation with lethality of the monosomic 21 zygotes. In carrier fathers, the risk is much lower-between 1/20 and 1/100.

A simple way to examine this difference is to count all the D ~ G trisomics known to be born from D ~ G carrier mothers and all those born from a D ~ G carrier father. Whey amount, respectively, to 104 vs 11. Without any other calculation, we can thus conclude that the risk is tenfold higher in women than in men.


Discussion of causes of the aberrations

Excluding the eventual effects of viral diseases, x-rays, and other risk exposures, we are finally confronted with apparently disparate findings which can be summarized as follows: (1) Free trisomies are very often second meiotic errors, and their global frequency increases with the aging of the mother. (2) Inherited translocation trisomies are apparently independent of maternal age, but are first meiotic errors and are much more frequent if the carrier is the mother.

How can these two observations be combined into a general hypothesis ? To investigate this possibility it is interesting to consider, as recently proposed by Gillois,[7] that every meiotic (or mitotic) step is enzymatically controlled and automatically regulated by the preceding one. For our purposes we can consider a simplified scheme and suppose that if the whole machinery is prepared before the starting signal is given, all the steps will succeed each other in appropriate sequence.

First division :

- fusorial apparatus

- rupture of synapses (without cleavage of centromeres)

- anaphase

Second division :

- fusorial apparatus

- cleavage of centromeres

- anaphase

Let us consider what would be the most appropriate way of using a system with built-in self-regulation. Obviously, best results would be achieved if the system was constantly running at its own speed, but a disadvantage would be to increase the number of point mutations if their frequency was related to the number of copies of a gene.

To circumvent this difficulty, an immediate answer would be to stop meiosis and to have it functioning only at the very moment that a new gamete is necessary. To achieve this trick we would need another regulatory mechanism controlling the meiotic process, turning it on or off on demand. This solution would result in a drastic reduction of cell division between the egg and the gamete, thus, in our hypothesis, reducing the rate of point mutations. The disadvantage would be to superimpose another regulatory mechanism on the already built-in regulation of meiosis. Any "give" in the coupling would increase the risk of error.

Both solutions have advantages that are greater than their inconveniences; they are, however, mutually exclusive. Thus, an optimization of the process could be to use both of there with the aid of two separate systems: one using continuous meiotic flow and the other using meiosis on request. The fact that this solution was chosen by nature is not proof positive that man and woman have been invented for that sole purpose. Nevertheless, this broad view has some practical implications which could be investigated.


The Translocation Trisomies

As noted earlier, any "give" between ovarian regulation superimposed upon meiotic regulation would add to abnormal migration. The way this coupling is realized is poorly understood. Nevertheless, as demonstrated by the experiments of Edwards, [8] the mechanical rupture of the follicle seems to trigger the whole meiotic process.

Let us suppose that the D/G translocation slows down the synaptic process between the free and the translocated 21. No effect is expected in male meiosis progressing at its own rate; in the ovary, however, if ovulation triggers the system before synapses are established, the; free 21 will migrate at random. Every random migration will produce either a trisomic or unviable monosomic zygote after fertilization.


The Free Trisomies and Aging

In female meiosis, a second trigger is the entry of the sperm that occurs before the expulsion of the second polar body. It is even possible that sperm entry is normally the triggering mechanism for the second anaphase. If it occurs earlier than the preparation of the necessary enzymatic machinery, the cleavage of centromeres could fail, leading to a second meiotic accident. This new type of accident occurs in addition to the first type due to early triggering by premature ovulation.

On the other hand, if fecundation is delayed, the meiotic mechanism, not retriggered in due time, could fail and become dormant. Thus, no fusorial apparatus would be produced at all, and triploidy by digyny could occur. Even aging of the membrane could let more than one sperm in, Hence triploidy by dispermy. At the most, a retarded second anaphase could result in the lagging of one element, hence a free trisomy. The difficulty here again stems from inappropriate timing between meiotic regulation and the circumstantial triggering mechanisms.

Various conjectures can be made about the possible correlation between the follicle chronology and the risk of aberration. For translocation carriers, early rupture of the follicle could be the plain danger, and that seemed to be the case in at least one family that was observed. For normal mothers precocious or late triggering could be dangerous, therefore no straightforward prediction can be ventured. It can only be surmised that aside from the possible affect of the aging of the egg, [9] premature rupture of the follicle could play a very dangerous role. If this were the case, fertilization would occur too early in the cycle rather than too late.

These reflections do not pretend to provide a solution to the whole problem, but may possibly lead to some heuristic implication. Geneticists probably unduly overlook physiological conditions. If we could understand them and possibly correct or prevent their distortion, many children could be protected from a severe disease which is considered a chance accident because its actual causes have not yet been sufficiently studied. In contrast to the case of amniocentesis, much discussed these days, a physiological understanding of trisomy could save many lives without destroying any.



1. SHUTTLEWORTH, G. E. 1909. Mongolian imbecility. Brit. Med. J. II: 661.

2. PENROSE, L. S. 1934. A method of separating the relative etiological effects of birth order and maternal age. Ann. Eugen. 6: 108.

3. PENROSE, L. S. 1963. Measurements of likeness in relative of trisomics. Ann. Hum. Genet. 27: 182-187.

4. GREMY, F., D. SALMON & J. LEJEUNE. 1967. Les modèles de genèse prézygotiques des trisomies. Ann. Génét. 10: 167-178.

5. DUTRILLAUX, B. 1968, Etude des translocations du chromosome 21. Thèse de Médecine. Paris.

6. DUTRILLAUX, B. & J. LEJEUNE. 1969. Etude de la descendance des porteurs d'une translocation t (21q Dq) Ann. Génét. 12: 77-82.

7. GILLOIS, M. 1969. La pseudo sexualité des cellules somatiques en culture: modèle de l'induction des pseudo-meioses. Ann. Génét. 21: 5-14.

8. EDWARD, R. G. 1965. Maturation in vitro of human ovarian ovocytes. Lancet ii: 926-929.

9. GERMAN, J. 1968. Mongolism, delayed fertilization and human sexual behaviour. Nature 217: 516-518.

10. DE GROUCHY, J. 1969. Personal communication.