Male mice, heterozygous for the T26H reciprocal translocation, were mated to normal females. A systematic histological study at day 6 and day 8 of pregnancy, covering altogether 303 implantation chambers, yielded the following results: the average number of decidual reactions per female was 8 ·91, the average number of morphologically normal embryos at day 6 was 4 ·61 and at day 8, 4 ·25. No significant embryonic loss occurred between day 8 and the end of pregnancy. Most (89 ·0 %) of the embryonic loss due to genetically unbalanced (deficiencies combined with duplications) progeny occurs between the moment of the induction of a decidual reaction and day 6, i.e. in ±1 ·5 days. When the genetically unbalanced embryos were grouped in three classes according to the size of their deficiencies, the numbers and types of the aberrant histological pictures found could be reconciled with the expectations of the three classes, based on cytological analysis. According to this, embryos with the smallest deficiencies take just over 2 days before all have died. Embryos with mediumsized deficiencies take ±1 ·5 days, and the group with the largest deficiencies take just over 1 day. At day 6, 8 ·8 % of the implantation chambers had an intact epithelium whereas at day 8, this had dropped to 1 ·3 %. This is consistent with the view that erosion of the uterine epithelium at the site of implantation is an autonomous process, but helped by the phagocytic action of the trophoblast cells.
Mammalian radiation genetics has mainly used the mouse as its experimental animal. Radiation-induced reciprocal translocations are characterized by excess embryonic lethality when a carrier, heterozygous for the translocation (T/+), is mated to a normal individual. The occurrence of dead implants in such a situation has even been used as an endpoint for measuring genetic damage (for instance Searle et al. 1974). No detailed study into the nature of the T/ + dependent dead implants, observed during the second half of pregnancy, has hitherto been carried out.
The development of human cytogenetics has yielded numerous cases of reciprocal translocations (Ford & Clegg, 1969), arousing interest in the timing and duration of embryonic death accompanying these translocations.
The meiotic segregation of the chromosomes involved in a reciprocal translocation yields a definite number of gametic types, all but two characterized by the presence of a deficiency and a duplication, either separately or combined (Searle, Ford & Beechey, 1971). These unbalanced gametes are capable of taking part in fertilization (Ford, 1972) and the resultant unbalanced zygotes will develop and then die at a particular stage. For the histological study of the embryological effects of reciprocal translocations in the mouse, we have taken the translocation T(2; 8)26H which, in heterozygous condition, is known to produce mainly two types of deficiency/duplication gametes (de Boer, 1976).
The aim of this study was (a) to establish when the genetically unbalanced embryos die and (b) to see if there is any correlation between the size of the deficiencies (and duplications) and the time of death.
MATERIALS AND METHODS
Female T(2; 8)26H translocation carriers were of Harwell origin and obtained through the courtesy of Dr A. G. Searle. The animals were back-crossed to a Swiss random-bred stock (Cpb:(SE)S) for five generations; T/+ males were then mated to pure Swiss random-bred females, which were killed on days 6 and 8, yielding embryos of approximately 5 and 7 days old. (The day the vaginal plug is found is day 1.) Altogether 18 females were killed on day 6 and 16 on day 8. Care was taken to distribute them as equally as possible over the seven T/+ males used. In control matings, Swiss random-bred males were mated to females of the same stock. Autopsy was performed between 9-10 a.m. This should yield embryos of just over 5 and 7 days old on average, when the dark period lasts from 6 p.m. to 6 a.m. (Braden, 1957). At autopsy, the uterus and oviducts were fixed in Bouin’s fixative for 3 h. The ovaries were kept for corpora lútea counts. The uterus was bleached in 70 % ethanol and processed for histo-logical sections at 7 μm in the usual way. The sections were stained with Delafield’s haematoxylin-eosin. Each deciduoma was sectioned in toto and four out of every eight sections were prepared for scoring. In our view, the chance of overlooking an embryo or its remnants is small with this procedure. With the aid of Snell & Stevens (1966), Rugh (1968) and Theiler (1972), criteria have been developed for the recognition of normal development with a developmental stage of 4 ·5, 5 ·0, 5 ·5, 6 ·0, 6 ·5, 7 ·0 and 7 ·5 days post-fertilization (see Table 1). Most sections were sagittal to the embryo. Where the embryo was sectioned transversely, the age of the embryo has been estimated as closely as possible, with the aid of Snell & Stevens (1966). Implantation chambers were considered to contain embryonic lethals if embryonic structures were absent or aberrant, or if the developmental stage was not in concordance with the day of autopsy.
The T26H reciprocal translocation is between chromosomes 2 and 8. The translocation chromosome 28 is longer than the original chromosome 2 and the translocation chromosome 82 is shorter than the original chromosome 8 (see Fig. 1). During prophase of the first meiotic division, homologous chromosome segments pair (see Fig. 1) and the segregational behaviour of the potential quadrivalent determines the number of gametic types found. The cytological analysis of secondary spermatocytes of T26H/ + males gives no definite assessment of the fraction of primary spermatocytes going through translocation-caused numerical non-disjunction, because not all four chromosomes involved in the translocation can be recognized. The data of de Boer (1976), together with the pattern of embryonic and foetal death obtained here, suggest that it is low. The predominant type of segregation is alternate/adjacent-1 (de Boer, 1976), whereby non-homologous centromeres go to the same pole. This yields two types of balanced gametes (with either two normal or two translocation chromosomes) and two types of unbalanced gametes. Less frequently, homologous centromeres go to the same pole, i.e. adjacent-2 segregation occurs. This yields six types of unbalanced gametes. Table 2 summarizes the gametic types thus produced, the location of the deficiency and duplication and their size according to measurements in G-banded chromosome preparations (de Boer & van Gijsen, 1974). The gametic frequencies given in Table 2 have been based on cytological observations on secondary spermatocytes and on relative litter-size data (de Boer, 1976). The assumption has been made that numerical non-disjunction caused by translocation is negligible.
The term ‘small mole’ is used for the little blood-filled deciduomata, observed during the second half of pregnancy (day 14). ‘Large mole’ refers to a dead implant, observed at the same day of pregnancy, in which at least the amnion can be seen macroscopically.
(1) The amount and timing of the embryonic lethality
Table 3 gives mean numbers of corpora lútea, decidual reactions, histologically normal embryos at days 6 and 8 and young born. The difference between the number of decidual reactions from the two types of mating is not significant (t73 = 0 ·822). The unbalanced embryos are thus capable of inducing a decidual reaction, as found by Searle et al. (1971). The difference between the size of litters sired by T26H/+ males at day 8 and at term is also not significant (t64 = 0 ·343).
Very little further embryonic death, due to translocation-dependent genetic imbalance or other reasons, was seen after day 8.
A number of females were killed at day 14 to test males for translocation heterozygosity. In the control, 7 out of 1423 implants had an amnion in which at superficial inspection the embryo was absent or dead (i.e. a large mole). For implants from T26H/ + origin, the figures are 6 out of 959. There is no difference between the two genotypes as to the number of large moles they produce, confirming that genetic imbalance does not cause embryonic death during the second half of pregnancy.
Table 4 divides decidual reactions at 6 and 8 days into those yielding a normal embryo and those containing a dead or retarded embryo , indicating that a proportion of the total embryonic death occurs between day 6 and day 8. If the total embryonic death amounts to 8 ·91 –4 ·08 = 4 ·83 decidual reactions (see Table 3) which at the end do not contain a live embryo, 8 ·91 –4 ·61/4 ·83 × 100 = 89 ·0% occurs between the induction of the decidual reaction and day 6, an additional 7 ·5 % occurs between day 6 and day 8 and the remaining 3 ·5 % between day 8 and the end of pregnancy. Knowing that ‘spontaneous’ embryonic lethality (caused by +/+ males and +/ + females and observed as a dead implant) is low in the mouse (1 ·3 % on the basis of the number of corpora lútea, see Table 3), the ‘true’ distribution of translocation caused death among the age classes will not deviate much from the distribution thus given.
Among normal embryos (Table 5), morphological variation with respect to developmental stage covered three age classes (as defined in the Materials and Methods). This means a variation in development of between 1 ·0 and 1 ·5 days at each day of gestation.
(2) The histology of abnormal implants
The variation in histological appearance at each stage of pregnancy is listed in Table 6 and illustrated in Fig. 2. At day 6 of gestation, 13 implantation chambers showed intact uterine epithelium and no embryonic remnants. An additional eight implantation sites were characterized by partially degenerated epithelial cells, with no embryonic remnants. Thirty-four implantation sites contained embryonic cells, usually singly or in clumps. Six implants contained embryonic structures with some degree of internal organization which was clearly aberrant. Another site with an unimplanted blastocyst was judged to be a case of delayed implantation. At day 8, only two implantation sites had intact epithelium. A large group (64) was characterized by blood in the implantation chamber, only occasionally encountered at day 6. Only in three of those were there any embryonic cells left. In seven sites embryonic remnants were mixed with blood. At an additional 13 sites the embryos either developed abnormally (see Fig. 2 F where the amniotic cavity is composed of two parts, one containing mesoderm), or were retarded in their growth. No specific syndromes representative of any of the gametic types produced by the T26H/ + male parent could be traced. The aberrant histological pictures occurred within the same uteri, as would be expected if (a) the fertilizing capacity of the male gametes is unrelated to their genetic contents (Ford, 1972) and (b) the embryos with a larger deficiency die earlier than embryos with a smaller deficiency.
(3) The relation between the spectrum of deviant implantation sites found and the spectrum of genetically unbalanced zygotes expected
Returning to Table 2, all unbalanced gametes and thus zygotes are characterized by the simultaneous occurrence of deficiencies and duplications, usually of chromosome segments, but in two cases, originating from adjacent-2 disjunction, of whole chromosomes. From the study of the embryological and foetal effects of primary monosomies and trisomies (Gropp, Giers & Kolbus, 1974), it is known that the primary monosomies die considerably earlier than the primary trisomies. Keeping this in mind, we have classified the gametic types into three classes, postulating that the embryos with the largest deficiencies die first. Class 1 contains the types with a deficiency greater than 4 ·6%, i.e. (2, 2); (8, 8) and (82, 82). Class 2 contains the types (2, 82) and (28, 28), the latter with a deficiency of only 2 ·6% but with a large duplication of 8 ·4%. Class 3 contains two types with a small deficiency (2, 28), 1 ·4% and (28, 8), 1 ·2%.
Table 7 sets out this classification and gives the expected frequencies with 8 ·5 % adjacent-2 segregation. This figure has been chosen on the basis of cytological observations in secondary spermatocytes (de Boer, 1976).
In section 1 of the Results, it has been shown that there is excess embryonic mortality between days 6 and 8, amounting to 56 ·1 –43 ·2 = 12 ·9% (Table 6). In 12 ·9 % of the day 8 implants (8.4 of Table 6), embryonic structures were seen. In 8 · 8 % of the implants at day 6 there were embryonic cell clumps or abnormal embryonic structures. Together, this accounts for 21 ·7 % and makes up most of class 3 of Table 7. We propose that the embryos of this class start to die before day 6 and are all abnormal by day 8. Their degeneration would thus take well over two days. Embryos which give rise to a decidual reaction but do not contribute to the invasion of the uterine epithelium must die after a shorter time interval. The blastocyst, but not the morula, is able to evoke a decidual reaction (McLaren, 1969). Autolysis of the uterine epithelium takes place between 113 and 129 h post coitum (El-Shershaby & Hinchliffe, 1975). If we assume the blastocyst stage to be reached days post coitum, these embryos must die in a period of just over 1 day. At day 6, we found 8 ·8 % of the implantation chambers to be empty with an intact uterine epithelium. We propose that this category is occupied by the class 1 embryos with perhaps some of the class 2 embryos. Table 7 suggests that all the embryos of class 2 have died by day 6. The bulk of this class falls in 6 ·2 of Table 6 (i.e. about 23 ·6% of all implants), with few or no embryonic cells present. Because we also found some more complete embryonic structures at day 6, the embryos of 6.2 must have been dead for a couple of hours at the time of autopsy. The data given in Table 6 are consistent with the assumption that the embryos of class 2 die over a period of 1 ·5 days.
The total embryonic lethality was 56 ·1 % at day 8 with probably not much lethality occurring thereafter. If there is no embryonic lethality between the blastocyst stage and day 8, other than that associated with translocation heterozygosity, the expected survival is (Searle et al. 1971), where p is the fraction of adjacent-2 segregation plus translocation-caused non-disjunction. With p ≈ 0 ·08 the expected survival is 46% and death 54%. This is close to the 56 ·1 % found in this study. Dead implantations occur among control Swiss matings (see Table 3); although the numbers will be small, it is therefore not correct to estimate the adjacent-2 frequency from the embryosurvival data in this study.
In section 1 it was reported that the frequency of large moles at day 14 did not differ between T/ + × 4-/4- and control matings. If numerical non-disjunction for the translocation-involved chromosomes occurs, a class of zygotes must be formed mainly trisomic for chromosome 8 or 82 (de Boer, 1976). The latter is especially likely to be picked up as a large mole (Gropp, Kolbus & Giers, 1975). The fact that we did not find a surplus of large moles argues against numerical non-disjunction due to translocation and thus against the production of trisomic types in T26H/+ males.
This study is the first in the mouse to locate the period in which embryonic lethality in translocation heterozygotes occurs, although earlier workers agreed that it must be shortly after implantation, as judged by the small blood-filled implantation chambers of rather uniform size, scored at day 10 or later (Snell, 1933; Hertwig, 1940; Carter, Lyon & Phillips, 1955).
The interpretation of the histological sections in terms of genetically unbalanced zygotes, due to heterozygosity for the T(2; 8)26H translocation in the male parent, has been based on the supposition that embryos with a larger deficiency die first. Although this seems reasonable to us, there is as yet no experimental support for this in the mouse or any other mammal, because the primary monosomies have not yet been studied embryologically. Moreover, the different primary trisomies show effects during the second half of pregnancy that are not strictly correlated with their size (Gropp et al. 1974, 1975). Deficiencies and duplications of only two chromosomes, 2 and 8, are involved in the present study. From the primary effects of these chromosomes on development it is known that monosomie embryos of chromosome 8 must evoke a decidual reaction and that trisomies for chromosome 8 die relatively early among the trisomies so far studied. Death occurs about day 11 or 12 (Gropp et al. 1975). The fit between the spectrum of morphological descriptions of the implantation chambers with abnormal or no contents and the spectrum of gametic classes seems to be good, especially if one realizes that with the number of embryos employed the standard deviations of the gametic classes and embryonic classes (Table 6) are rather wide.
Results obtained by Oshimura & Takagi (1975) on embryos karyotyped at 6 ·5 –7 ·5 days gestation and from matings between T6Ca/ + and normal mice show that the pattern of death of the unbalanced alternate/adjacent-1 and numerical non-disjunction types follows the size of their deficiency, large deficiencies dying first, which is further support for the explanation of our data.
Genetic lethals are more variable at a later stage of development, both as individuals and as a group. This variation in developmental potential seems to increase when the genetic imbalance is less severe (Hamerton, 1971 ; de Boer & Groen, 1974). On the contrary, when death is brought about by mutant genes in homozygous condition during the first week of development, the actual disintegration takes a short period of time (McLaren, 1974).
A human reciprocal translocation with the same embryological consequence as the T26H mouse reciprocal translocation studied here would lead to embryonic death between days 4 and 17 after fertilization. Abortions in the third week of pregnancy can be clinically recognized (Boué & Boué, 1974). It is known that in couples where one parent is the carrier of a reciprocal translocation, there is only a slight elevation (from ± 15 –21 ·7 %) in the level of clinically recognizable abortions (Ford & Clegg, 1969). This sample of reciprocal translocations was biased, however, in that most were ascertained through a defective child, indicating a non-representative distribution of the translocation breakpoints.
In another, much smaller sample of human reciprocal translocations, not ascertained through a defective proband, the same tendency was noted, with an increase in the number of recognizable abortions only at the 5 % probability level (Jacobs et al. 1970). Thus, the findings in T26H/+ are compatible with those for human reciprocal translocations, indicating that lethality due to gross chromosomal imbalance occurs at comparable embryonic stages.
In the present study we found a difference between the fraction of implantation sites with an intact epithelium at day 6 (8 ·8%) and at day 8 (1 ·3 %). According to El-Shershaby & Hinchliffe (1975) the invasion of the uterine epithelium at the implantation site is an autonomous process, although the trophoblast digests the dead epithelial cells by phagocytosis. This point is illustrated in our material. The uterine epithelium disappears even when there is no trophoblast, but the process takes less time when trophoblast is present.
We thank Mr F. A. van der Hoeven for skilled histological assistance. Mr C. V. Beechey and Dr A. G. Searle commented on the manuscript. Thanks to them as well.