Although duplication of chromosomes is the basic system of propagation
of genetic information, very little is known about this remarkable
Not attempting to make a review of the actual knowledge on chromosome
structure, nor on its function, simply two phenomena observed in human cell
will be discussed: selective endoreduplication and behaviour of ring structure.
1. Selective endoreduplication
Almost all the chromosomal aberrations already known in man can be
attributed to three fundamental mechanisms :
Malsegregation of normal element at diakinesis :
Maldistribution of elements involved in a previous rearrangement,
leading to duplication or deficiencies in the progeny of translocation
heterozygotes for example;
- And aneusomie de recombinaison, resulting of chromatid exchange
between rearranged chromosomes leading also to duplication deficiency but
without numerical shift and, eventually without perceptible changes of the
Selective endoreduplication is another mechanism, previously
overlooked, which is related to the mechanism of the duplication of chromosomes
themselves and not to their movement or to their meiotic exchanges.
It must be stressed that selective endoreduplication differs from the
well known endoreduplication in the number of chromosomes involved in the
process. Endoreduplication is produced by two successive duplications of the
whole complement, so that a 2n cell, splits finally, in two 4n daughter
Selective endoreduplication involves only one element or only a part
of it, the rest of the chromosomes having perfectly normal behaviour. Hence due
to this process a 2n cell splits finally into two daughter cells each of them
being 2n +1 or 2n+ part of a chromosome.
First quoted in 1963, and established on a few sporadic instances in
1966, this phenomenon was found to correspond to a particular lesion of a
chromosome in a family about which detailed information was available.
A feeble minded woman, who had attempted suicide, was found to have a
peculiar gap on the long arm of ones of her chromosomes no. 2. This gap was
obvious in half of the cells in blood culture and could be suspected in many of
the rest. It was very different from the ordinary (and normal) secondary
constrictions. In four cells out of 640, the distal fragment was found
endoreduplicated, all the rest of the karyotype being normal.
In two cells one exemplar of this piece was found in excess, and a
cell containing two exemplars and another containing four were also observed.
This person had two daughters-one normal having perfectly normal chromosomes.
She had received the normal 2. The other girl was feeble minded and suffered
from a cleft palate which was conveniently surgically corrected. She also had
an abnormal chromosome no. 2, exhibiting exactly the same gap as the chromosome
of the mother.
Identically, selective endoreduplication of the distal segment was
observed in one cell out of 287; loss of this piece in two cells, gain of one
in one cell and gain of two in two cells were recorded.
It is difficult to assess whether the mental condition of the two
carriers of the abnormal chromosome was related either to the gap or to the
duplication deficiencies in mosaic for the distal fragment, although this is a
very real possibility.
But it is quite certain that selective endoreduplication of the distal
fragment qh (2) was related to the pre-existing change of structure, to the
"gap" of no. 2 transmitted from the mother to the abnormal daughter.
No other case of familial selective endoreduplication is yet
published, but the consistency of the phenomenon leads one to consider that it
could have a general significance.
Only three weeks ago, another instance was found in this laboratory of
a retarded infant, exhibiting a very similar gap (but on a chromosome no. 8
this time) and having one cell in 63 in which the distal part of the fragment
is selectively endoreduplicated, exactly as it was in the first family.
As previously quoted, selective endoreduplication occurs sporadically
and, in this laboratory, the frequency with the culture techniques used is of
the order of 10-4, Personal inquiries to other laboratories gave the
conviction that sporadic selective endoreduplication is effectively seen from
time to time in one particular cell and impression is that 10-4 is
an approximate order of magnitude.
Hence here is a duplication accident extremely rare no doubt, but the
frequency of it can be multiplied at least by a factor of hundred (and possibly
much more if the identity of the target is taken into accounts when a
structural gap exists on a chromosome.
Considering the mechanics of duplication it seems difficult to escape
the conclusion that there exists in each chromosome a built-in regulation
system which not only determines the synchronisation of the whole set (normal
mitosis) but also controls the duplication of each chromatid, sending its order
from the centromere towards the end.
The argument is that in all instances observed either sporadic or
predetermined by an existing structural anomaly, the selective
endoreduplication involves only the distal part of the element. When this
accident is partial, it never "crosses" the centromere.
Significance of selective endoreduplication is still difficult to
assess. If it involves the whole chromosome (this has been observed in one cell
in author's laboratory and concerned the Y) this process leads to two daughter
cells with an extra element. In other words, the results differ from the
malsegregation accident because no reciprocal clones are produced (one
monosomic, the other trisomic) but only trisomic clones are produced. This
could explain the occurrence of sexual mosaicism like XX/XXX/XXXX, etc. without
haplo-X clone being detected. Also, in aneuploid cancerous cells it is very
often observed that chro-mosomes appear in excess, without the reciprocal cells
having lost these chromosomes. In both cases a selection system could supply an
explanation, but the very existence of selective endoreduplication furnishes a
new understanding of this point.
The second error of replication I would discuss concern the ring
chromosome and duplication of rings.
As extensively studied by McClintock the duplication of ring
chromosome is not always conformed but can lead to various accidents. Many
instances are now known in man of constitutional ring chromosomes, and it is
remarkable that the resulting phenotypes are always a mixture of symptoms of
trisomy and of monosomy. A theoretical analysis of the behaviour of ring can
explain this apparent paradox.
Constitution of a ring "Telomeric hypothesis" postulating that each
end of a rod chromosome is terminated by a special "cork", the telomere, which
prevents the sticking of chromosomes end to end in an endless chain, implies
that no ring can be formed without loss of genetic material, say X of the short
arm and Y of the long one.
It is theoretically possible that these deletions can be extremely
localized, for example, segment X being reduced to the sole telomere. An
example of this fact has been fortuitously observed in the laboratory three
months ago. This picture is an excellent demonstration that the telomeric
hypothesis holds true for the human chromosomes in vitro although it does riot
preclude the possible existence of so called interstitial telomeres as
envisaged by Hsu.
Duplication of ring can happen normally but exchange of chromatid can
lead as demonstrated by McClintock to dicentric chromosomes.
Two accidents could be envisaged, the first being if one break
between sister chromatids is healed this leads to a dicentric with centromeres
evenly disposed on the resulting double sized ring. In the second instance, a
chromatid exchange produces also a dicentric, but the two centromeres are
That this second mechanism is likely to be at work, in accordance
with the ideas of McClintock, is inferred from the observation that all the
dicentric rings observed to date in human material had diametrically opposed
centromeres. In this system, the chromatid exchange transforms the ring in a
moebius surface, which indeed cannot be cleaved in two rings but only in a
double sized single ring.
An interesting topological consequence is that centromere must not
play the role of a swivel, because, if it could, the moebius surface would be
transformed into a normal ring and no dicentrics could be formed.
From experimental data of Lima-de-Faria it appears that early
duplication of the centromere and its fine structure are in obvious
contradiction with its eventual swivel role. Hence we moebius effect.
An interesting question arises when the dicentric ring, pulled in
opposite direction by its centromeres, breaks and produces two daughter rings
by secondary healing of the broken end. If the rupture points of the two
chromatids are diametrically opposed, the two daughter rings are identical and
no change is achieved. If the paints of rupture are not diametrically opposed,
if they are "de guinagois", the two daughter rings differ, one having an extra
piece included (duplication) the other having lost it.
The question can thus be asked what is the probability that a given
gene, will be included doubly or excluded during the process ? This can be
easily pre dicted on the basis of simple assumptions. Let us assume the
position of a gene being defined by its rank, starting with gene 1 close to
centromere, then 2, 3. . . up to n (close to the other side of the centromere)
if the ring has "n" genes.
Let it be assumed that we are interested in the fate of a gene having
the rank r between 1 and n.
Of all the combinations excluding or including one gene, one will
interest the gene situated at rank r. Among those interesting two genes, two
will interest it. Thus, for the combination interesting 1-2 . . . and r genes,
the gene will be involved (i, integer the same
is obviously true for combination interesting n, n-1 . . . (n - r + 1) genes,
and they will involve the gene r (i integer)
Finally the combination, involving a number of genes comprised between
(r + 1) and (n - r) will affect the rth gene r times a total of: (rn - 2r).
Together the rth gene is involved 2 +
r(n - 2r) ive : r (n - r + 1) and the total number of possible combinations
involving or not involving the rth gene is r
(n - r + 1), for r can vary from 1 to n, the possibility that the rth gene is
involved is Pr = r ( n - r + 1)/[n (n + 1)(n + 2)/6]
The numerator is defining the different terms of the diagonal of
Pythagorean table and the denominator the sum of these terms. The statistical
distribution is different from the Gaussian one and shows that a gene in the
middle of a ring is at much greater risk than a gene close to the centromere. A
curious remark is thus that the law of probability governing a ring is in fact
included in the oddest squared table, that of Pythagoras !
A generalisation of the chromatid exchange hypothesis does not need to
go further than a few successive exchanges. If two successive exchanges occur
the result is interlocked rings. If one breaks, reditus ad integrum is
possible. The same is true with more exchanges (of the same paths with the
reservation that interlocked rings too complexly interwoven can possibly not
duplicate, with a pulverisation, of the ring resulting from the process.
Generally if successive exchanges occur in the same ring n, nl = n
turning to left and nr = n turning right, their topological result is equal to
their algebraic sum nt = nr - ne. These considerations, based upon the fact
that centromere is not a swivel, raise many difficulties.
One results from the fact that spiralization of ring chromosomes does
occur during the contraction from prophase to metaphase. It is impossible to
spiralize regularly a ring as anybody can convince himself with a ring of
electric wire, except by turning for an equal amount of time to the left and an
equal amount of time to the right. Whether this occurs in ring chromosomes in
unknown but it is interesting to quote that changes of the sense of
spiralisation have been directly observed in rod chromosomes by Holm and Bajer
To extend this reasoning from chromosomes to ring molecule of DNA is
tempting but possibly hazardous. It can be remarked nevertheless, that Cairn
was obliged to postulate a hypothetical swivel to explain duplication of ring
DNA. The structure of this swivel which should rotate at the speed of an
ultracentrifuge is unknown.
If we suppose that by same topological tricks, not requiring any
swivel, ring DNA follows the model previously discussed, one can forsee that
exchanges between the strands (molecular crossing oven would lead to three
kinds of events:
- rings of various size (equivalents of reconstituted monocentric
rings after rupture of the dicentric),
- double size rings (equivalents of the dicentrics, i.e. one
- interlocked rings or catenated (two paths of the same signs).
These three phenomena have been observed lay Clayton and Vinograd and
Hudson and Vinograd in the ring DNA molecule found in mitochondria of human
If the model holds true, the result should be duplication-deficiency
accumulating progressively, explaining eventually, by a background noise
steadily increasing, the loss of information being endured by the cytoplasm,
ultimately resulting into the ageing of somatic cells. To accept actually the
mode of reasonning I have just delineated is a matter of personal choice
because of the purely theoretical characteristics of the model.
By the way, this simplification would lead to the conclusion that DNA
molecule can sometimes be dextrogyre and sometime sinistrogyre, the topological
trick being previously asked for. This problem is relevant to specialist in
molecular structure and is far beyond the grasp of a mere cytogeneticist. These
points are nevertheless worth mentioning because rings (either chromosome or
molecules could thus obey the same simple laws. Alternatively, if the model is
rejected, the fine description of the swivel in molecular term, would be of a
In conclusion, it must be stressed that this discussion about breaks
and healings is especially dangerous when rings are involved.
The closer the discussion comes to the very nature of the ring, the
greater is the risk of the argument itself becoming circular.
I tried my best to break this vicious circle, but the healing process
frequently evoked in this paper does not preclude interlocked reasoning.
A. R. Banerjee (Calcutta) : Has viscosity of the cell fluid anything
to do with the formation of ring chromosome ?
J. Lejeune (Paris): More experience is required before an answer could
G. K. Manna (Kalyani) : Could you please give an idea of the
distribution of lethal genes from the chromosomal malformation study ?
J. Lejeune (Paris): To this broad question, present stage of
cytogenetics is too crude to offer any acceptable answer.
K. Patau ( Wisconsin) : In one of Bajer's mitosis movies, two
interlocked ring chromatids are shown to move through each other at anaphase. I
would suggest that this scene also suggests the mechanism by which prompt
healing of the break is assured (sketch). Do you have any comment on this ?
J. Lejeune (Paris): Indeed breakage of one ring is the only way two
interlocked rings can separate-quasi immediate healing, as you suggest, is a
J. H. Hunziker (Buenos Aires): Don't you think it would be better to
call it "partial" rather "selective" endo-reduplication ? The phenomenon seems
to be haphazard and not connected at all with natural selection.
J. Lejeune (Paris): "Partial" endoreduplication could as well have
been chosen. Nevertheless it would not define the fact that the same part is
repeatedly affected, or to be clear, "selected." Terminology is always
difficult and selection is taken here in the sense of "chosen," "elected" and
with no connotation of competition or advantage. But I would not defend too
much the term I use fearing to become partial.
J. Nizam (Hyderabad): l; Do the structural changes on the somatic
chromo somes tend to be altered during the course of subsequent cell division
or they remain constant throughout? 2. Are there attempts to set right the
structural changes by artificial means?
J. Lejeune (Paris) : Most of changes seem to be stable but for the
ring accidents. I am not aware of any attempts to repair them artificially.
S. D. Sharma (Chandigarh): Do the ring chromosomes always persist
through same D-mitotic cycle and whatever percentage of the cells may carry
these rings ?
J. Lejeune (Paris) : If in the first cell, the zygote contained a
ring, all the cells contain it; with the exception of the 45 chromosomed cells
which have lost the ring.
C. Pavan (Sao Paulo : In one of your slides you have shown a
"pulverized" chromosome and have concluded that this was due to multiple
interlocked rings. How may exchanges would you think occurred in the ring
chromosome to cause that pulverization ? Do you exclude the possibility that
one break of two in special places of the chromosome may cause an error in
chromosome spiralization or duplication?
J. Lejeune (Paris): I agree entirely that some sensitive points, if
broken, could interfere with the replication and lead to pulverisation. As to
the number of exchanges necessary, I would only venture more than two-but this
is an uninformed guess.
J. Moutschen (Liege) : I would consider :Dr. Lejeune's selective
endoreduplication as a marvellous explanation for our multiple satellitic
chromosome. I showed in Vicia-there were till 4 appended to the same N. O.
However, I wonder if this explanation is always true since Levan in Allium and
ourselves observed translocations of a satellite into centromere. In such
cases, could it be that near centromere regions endo-reduplicate on
heterologous satellitic region?
J. Lejeune (Paris) : I am not familiar with these very interesting
data mentioned by Prof. Moutschen. Indeed it would be most useful to
investigate thoroughly this matter. As a fact the particular point at which
selective endoreduplication begins is in some case very close to the centromere
and in others, rather far apart.
R. B. Singh (Hissar): Is it so that chromosome number 13 undergoes
ring formation more often that other chromosomes? If so, why?
J. Lejeune (Paris) : I am not sure 13 is more at risk than other
elements. A possible explanation would rely an the fact that disequilibrium and
loss of other pairs, like the number one for example would be so deleterious
that it is incompatible with embryonic development.