Hum Genet (1991) 88:64-70
9 Springer-Verlag1991
A test of the production line hypothesis of mammalian oogenesis Paul E. Polani and John A . Crolla*
Division of Molecular and Medical Genetics,Paediatric Research Unit, United Medical and Dental Schools of Guy's and St. Thomas's Hospitals, 7th Floor, Guy's Hospital Tower, London Bridge, London SE1 9RT, UK Received June 20, 1989 / Revised April 22, 1991
Summary. Germ cells in female mammals become committed to meiosis and enter its prophase sequentially in fetal life and, according to the Production Line Hypothesis, the oocytes thus generated are released after puberty as mature ova in the same sequence as that of meiotic entry in fetu. This hypothesis in its original and complete form has a subordinate proposition that concerns chiasma (Xma) frequency; it postulates that Xma number would decrease with fetal age. Consequently, univalents would increase, leading to errors of chromosome disjunction at the first meiotic division (MI), and thus to maternal age-dependent numerical chromosome anomalies. By using an in vitro/in vivo approach, we radioactively labelled the DNA of germ cells at premeiotic synthesis as they sequentially entered meiosis, while the fetal ovaries were in culture. At the end of this in vitro phase, pachytene/diplotene (P/D) stages were studied to determine their labelled fraction. The ovaries were then transplanted to spayed females and, after the in vivo phase, mature ova were harvested and the proportion of labelled first and second meiotic metaphases (MI/MII) determined. By marking the germ cells with label while in vitro during periods equivalent to early and late gestation, and by comparing the observed proportions of labelled MI/MII with those of oocytes labelled at P/D, we concluded that, in the mouse, ova do not mature at random for release, but are formed according to a production line system in which the time of release after puberty is related to the time of entry into meiosis in fetu.
Introduction
In an attempt to explain the maternal age-dependence to trisomy 21 in man, while taking account of the fact that in female mammals the genetically important events of gametogenesis (crossing-over, chiasma (Xma) formation, * Present address: Wessex RegionalGenetics Laboratory, General Hospital, Salisbury,WiltshireSP2 7SX, UK Offprint requests to: P. E. Polani, "Little Meadow",West Clandon, Surrey GU4 7TL, UK
and recombination) happen before their birth, Henderson and Edwards (1968) proposed that the order of the entry of germ cells into meiosis, i.e. of oocyte generation, dictated the order of their release as mature ova from puberty to the end of fertile life. Furthermore, they suggested that oocytes formed late in fetal life had fewer Xma holding their bivalents together then early ones. Homologous chromosomes that are paired to form bivalents, but that are devoid of Xma, would, because of the absence of the mechanical function of the latter, behave as univalents and so would not disjoin regularly from each other, frequently producing unbalanced ova and zygotes. The frequency of the latter would increase progressively with advancing age of the female because the old ova would be derived from oocytes generated late prenatally. This, the "Complete Production Line Hypothesis" (Complete PLH), is clearly relevant not only to the origin of trisomy 21, but also to that of all numerical chromosome anomalies of maternal derivation traceable to errors of MI and correlated with maternal age, viz. the majority of such anomalies; this should be true not only for man, on whom data of such anomalies are substantial, but also for experimental animals. The Complete PLH originated from Henderson and Edwards' observations on adult mice; they found that oocytes at MI, harvested from older females, had lower Xma counts and more univalents than those from younger ones. These results were interpreted in the light of an essential aspect of Xrna, namely that "there is no loss of chiasmata once formed" (Henderson and Edwards 1968) in the early prophase of meiosis, right up to the time of diakinesis and MI (Henderson 1969, 1970; but see Slizynski 1960, and John 1990): this is the basis for the hypothesis. The hypothesis was subsequently subjected to a variety of tests. Using cytological techniques, Luthardt et al. (1973) confirmed the decrease in Xma frequency and the increase in univalents with age, as did Polani and Jagiello (1976), who also studied MI segregation of chromosomes. They concluded that most MI univalents, although incremental with age, were artefactual. Suyawara and Mikamo (1986; and 1983 in the Chinese hamster) supported this view; Speed (1977), who had noted the drop in Xma, could not confirm the increase of univalents. Hummler
65 et al. (1987) found evidence in the Djungarian h a m s t e r arguing against a production line system. O t h e r workers based their tests on the use of structural c h r o m o s o m e anomalies. Thus, T e a s e and Fisher (1986) investigated meiotic pairing in inversion heterozygosity; their results were not inconsistent with an oogenesis production line system in the mouse. H o w e v e r , in a further study (1989), which included analysis of M I at different ages, they aligned themselves against such a system. This too was the stance of B e e r m a n n et al. (1987) on the grounds of an increase with age and not a decreased calculated cross-over frequency within a translocation; de B o e r and van der H o e v e n (1980) exploited the long product of a translocation and found no decrease of X m a with age; they concluded that non-disjunction was mostly the result of postnatal oocyte ageing. Finally, using a novel and demanding approach as a test of a f u n d a m e n tal stipulation of the C o m p l e t e P L H , Jagiello and Fang (1979) showed that late c o m p a r e d with early fetal oocytes at diplonema had fewer X m a and m o r e univalents. H o w ever, these findings could not be duplicated b y Speed and Chandley (1983) on a different Swiss strain of mice because, in their hands, X m a identification p r o v e d difficult. G e n e recombination frequencies, as tests of the C o m plete P L H , could be used to c o r r o b o r a t e the cytological work, and older studies showing age-related changes in recombination seemed supportive ( B o d m e r 1961; Reid and Parsons 1963). H o w e v e r , ad hoc tests (Wallace et al. 1976) indicated that the changes at the different tested loci were not all in the direction expected f r o m the hypothesis, and that changes in recombination location might also occur with age. All the above work addresses the C o m p l e t e P L H , but is an indirect test of the core of the hypothesis. It is centred either on testing for a decrease with age (usually postn a t a l , exceptionally fetal) of X m a frequency with a corresponding increase of univalents, accepting implicitly that numbers of X m a , once formed, are i m m u t a b l e , or else on tests based on age-related changes of r e c o m b i n a tion frequencies. On balance, the bulk of the w o r k does not support the hypothesis. H o w e v e r , n o n e of the studies have set out to probe directly the "first in first out", last in last out" stipulation, the key tenet of the hypothesis to which the X m a reduction with univalent formation (and therefore the non-disjunction effect) are special subordinate propositions. In order to test the P L H in this n a r r o w e r but fundamental sense, the following experiments were conducted, based on tracking and analysing mature ova derived from oocytes g e n e r a t e d at different times during fetal life and marked before entry into fetal meiosis. A previous a t t e m p t to track fetal oocytes by marking them at premeiotic D N A synthesis, by means of injected radioactive DNA precursors into the m o t h e r during pregnancy, had not been successful (G. Jagiello and F. Giannelli, personal communication). In o r d e r to bypass blocking by maternal and fetal barriers, we m a d e use of an in vitro/in vivo technique that we had developed for this purpose (Polani et al. 1979; 1981). It consists of explanting and culturing i m m a t u r e fetal ovaries, during which time the oocytes, generated as if in fetu, are
m a r k e d radioactively at premeiotic D N A synthesis. After the in vitro phase, heterotopic ovarian transplantation into spayed females permits maturation of oocytes in vivo under hormonal control, and their subsequent harvesting and study.
Materials and methods
Mouse strains and breeding Two strains of laboratory mouse were used, a random bred A Strong and a congenic C3H/He. Timed rantings were carried out, and the day of vaginal plug was called day 1 of gestation.
Culture In all experiments and for all labelling protocols, pregnant females were culled on day 14 by cervical dislocation and the fetal ovaries were dissected free of their adherent mesonephric tissue prior to being set up in an organ culture system, the details of which are described by Polani et al. (1979 and 1981; here, we adopted a multiwell system in preference to the raft method). In experiments I-III, ovaries from several fetuses (derived from the same pregnant female) were pooled prior to setting up the culture. In experiments IV and V, the ovaries from individual fetuses were cultured in pairs until specific manipulations were carried out. While in vitro, the ovaries were exposed to 3HTdR (0.1 ~tCi/ml; specific activity 0.5 Ci/mmol) by transferring them to multiwells on agar and by incubating them in medium preconditioned with 3HTdR. The developing ovaries were exposed to the label for 10-18 h on day 14, or for the whole of day 14, or day 15, or days 16/17 gestation equivalents. At the end of the labelling incubation, the ovaries were removed from the culture wells, washed in three changes of complete medium without label, and prepared for either air-dry spreading, or for transplantation as indicated by the protocol of the experiment.
Transplantation After 8 days in culture, the fetaI ovaries were removed from the culture medium, washed in phosphate buffered saline and transplanted under the kidney capsule of a previously spayed young adult female of matching strain.
Air-drying and spreading of meiotic chromosome We studied both prophase and metaphase cells. For the former, at the pachyteneldiplotene stages, the fetal ovaries were placed into 0.7% sodium citrate for 30rain at room temperature, transferred onto microscope slides in a drop of hypotonic citrate, and finely mashed with iris knives; the resulting cell suspension was spread over the centre of the slide. A small drop of 3:1 methanol: acetic acid was immediately dropped onto the cells and the preparation allowed to air-dry; the prophase cells were stained in lacto-aeetic orcein prior to autoradiography. Successfully transplanted ovaries, removed from the renal site, were mechanically dissociated and their freed mature germinal-vesicles-containing oocytes (GV) were selected under low power phase and kept in vitro for 4 h at 37~ in Ham's F10 supplemented with 10% donor calf serum (Flow). Chromosome spreads (MI/MII) were prepared according to Tarkowski (1966) and stained in lacto-orcein, prior to autoradiography.
Histology and autoradiography In experiment v , transplanted ovaries were washed in two changes of phosphate buffered saline, fixed in Bouin's for 4-4,5 h, and
66 paraffin-embedded for serial-section histology. Sections were cut at a thickness of 10 Isrn, de-waxed in xylene, dehydrated through a series of alcohols, floated onto poly-L-lysine pretreated microscope slides, and coated with autoradiographic stripping film (Kodak AR10). The preparation of P/D and MI/MII for autoradiography was similar. Slides were exposed in the dark at 4~ for 2 weeks (prophase/metaphase) or 6 weeks (histology). All slides were developed in Kodak D19 and fixed with Kodak Rapid Fixer. The sections were stained through the stripping film with toluidine blue in borax (the techniques followed are in principle described by Gude [1968]). The serial sections were systematically analysed and drawn
for follicle reconstruction. Autoradiographic g a i n s were visualised and counted under direct light or phase contrast illumination with a Zeiss photomicroscope III. The autoradiographie analysis and histological scoring of all experimental and control material (see Results) were performed blind. Slides were coded by an independent person and decoded only at the end of the whole experiment. The laminarity of follicles and the compactness, or otherwise (antral and Graafian follicles) of the follicular cells were recorded, and the autoradiographic grains were assessed and counted over the GV, the follicular and stromal cells and the spaces within follicles.
Table 1. Summary of experimental and control data (with percentages of labelled prophases, P/D, and metaphases, MI/MII) Experiment
System and day (d) 3HTdR labelling
I
Ovaries for P/D a
12
Direct ageing
Ovaries transplanted
77
Ovaries harvested b
Experimental
31 702 892
(86%) (48%)
MI/MII yield
d14 d16/17
106 100
(93%) (2%)
6.6 (206/31)
Ovaries for P/D
Early harvest
Ovaries transplanted
28
(57% take)
223 8.0 (233/28)
29
Ovaries harvested
40
133 114
(86% take)
P/D yield part
d14 d14 d15 d16/17
4486 2310 1420 963
(55%) (86%) (78%) (37%)
MI/MII yield part
d14 d14 d15 d16/17
177 343 180 98
(92%) (98%) (41%) (2%)
7.0 (798/114)
MI/MII per ovary IV
Ovaries for P/D
16
Littermate
Ovaries transplanted
16
Ovaries harvested
Histoautoradiography
(40% take)
d14 d16/17
II and III
V
49
P/D yielda
MI/MII per ovary
Control
14
39
(98% take)
404
10.4 (404/39)
16 (88% take)
P/D yield
d14 d16/17
608 849
(90%) (23%)
MI/MII yield
d14 d16/17
50 70
(94%) (0%)
14
(88% take)
B
82
MI/MII per ovary
8.6 (120/14)
5.9 (82/14)
Ovaries for P/D
8
m
Ovaries transplanted
8
3c
Ovaries for histology
8
3c
P/D yield
d14 d16/17
1600 960
(90%) (43%)
GV yield
d14 d16/17
202 184
(83%) (35%)
GV per ovary
48.25 (386/8)
a In all experiments, data on prophases are control data for the expectations of metaphase labelling b Productive takes only c Littermates to animals in experiment IV
158 52.7 (158/3)
67 Finally, with respect to all our experiments, the critical point of the in vitro/in vivo method is that, whereas it ensures the required control, it nevertheless permits oogenesis to proceed normally. The in vitro phase allows fetal germ cells to become committed to meiosis in a regular fashion; the in vivo heterotopically transplanted ovaries support normal maturation of the resident oocytes. During the two phases, the numbers of oocytes that are generated and that mature are close to normal. A further measure of normality is that, in orthotopic transplantation experiments using genetically marked fetal ovaries, the mated recipient females produced and reared normal young; thus, both the maturation and the fertilisability of the transplanted oocytes are normal (Polani et al. 1981).
Results
Following exposure of the fetal ovaries to 3HTdR in vitro, we set out to establish the proportions of radioactively labelled m a t u r e oocytes available after transplantation, and to c o m p a r e them with the proportions of
Table 2. Labelled mature ova (MI/MII) derived from fetal germ cells labelled in vitro. The expected labelling values of MI/MII, calculated from the proportion of meiotic prophase celis found to be labelled at the end of the in vitro phase, assume that the PLH is not valid. The expected values are all statistically significantly different from the observed labelling data. The days on which the in vitro cultured fetal ovaries were supplied with the radioactive DNA precursor correspond to gestation ages. The metaphases were harvested early after transplantation
Ovaries Labelled explanted on day 14, 3HTdR in vitro on d14
Unlabelled
Total
Observed
484 (97.0%)
15
499
Expected
431
68
499
495
3620
14
177
80
177
2019
4486
107
180
40
180
317
1420
264
268
171
268
1722
2 704
MI/MII (86.3%) P/D 10-18 h of d14
Observed Observed
3125 163 (92.0%)
MI/MII Expected
97
(55.0%) P/D
Observed Observed
dt5
2467 73 (41.0%)
MI/MII Expected
140 (77.7%)
P/D
d16/17
Observed Observed
1103 4 (1.5%)
MI/MII Expected
97 (36.3%)
Total
P/D
Observed
MI/MII P/D
Observed Observed
982
1124 12 230
aHTdR-o-o-o-to ovaries Littermotes in vitro on day 1L I 15 116 ~ 17 linvivo for 19doys Then .o-o-o-
2t 3t
(ARG}~ .~
,
o.o .....
Fig. 1. Schema
ff the littermate/shared-recipient experiment (IV), comparing metaphases autoradiographically (MI/MII [ARG]) in experimental animals with their littermate controls, obtaining background labelling on P/D (Pachy) from sister ovaries of experimental animals, and using the two kidneys of the same recipient for engraftment of ovaries labelled in vitro on day 14 or on days 16/ 17 (cranially) together with unlabelled control ovaries (caudally) from a littermate
labelled prophases at the end of the in vitro phase, as a labelling baseline; the comparison provides the test of the P L H . W e carried out five sets of e x p e r i m e n t s (I-V), and the numerical data are set out in T a b l e 1. In experim e n t I, following labelling in vitro on either day 14, or o n gestation equivalent days 16/17, we set out to obtain M I / M I I on day 19 after transplantation and at weeks 5, 7, 9, 11, 15 and 20. H o w e v e r , no m e t a p h a s e s could be harvested f r o m ovaries 9 weeks or longer after transplantation, although ovaries survived well and o v a could be harvested with similar success rates as f r o m control ovaries up to, and including, 7 weeks post-transplantation. Hence, these results f r o m the early harvests have b e e n pooled and included in the final d a t a (Table 2). T h e subsequent experiments t o o k these results into account and were all early harvest (day 19) experiments. T h e first experiments also showed that m o s t prophases were labelled (see Discussion). T h e r e f o r e , we tested the P L H m o r e critically by labelling a m o r e discrete population of gonia, in experiment I I I and we labelled the cultures for only 10-18 h on day 14. In e x p e r i m e n t II, label was supplied to cultured ovaries on day 14 and days 16/ 17, and also on day 15. O v a were harvested early. In experiment IV, which was designed to minimise variation, fetuses of the same litter were allocated to e x p e r i m e n t a l or control groups (Fig. 1). A m o n g the f o r m e r , pairs of ovaries (from one fetus) were labelled on day 14, or, alternatively, on days 16/17 ( f r o m a littermate). O f each pair of these ovaries, one was used to establish the proportion of prophases m a r k e d , whereas the other was transplanted for the in vivo phase. Control ovaries (from a third littermate) were co-transplanted with the experimental ovaries: the day-14-1abeUed ovary and its control under the capsule of the right kidney, the days-16t17labelled ovary, with its control, to the left side of the same recipient. Finally, 8 fetuses were used in experiment V. We could select 3 littermates to the fetuses in experiment IV. O f each pair of these three ovaries, one was labelled (day 14 or days 16/17), and the o t h e r was an unlabelled control. A f t e r 19 days, the test and control gonads were serially sectioned for histo-autoradiography.
68 Table 3. Distribution of GV labelling in histological sections of transplanted fetal ovaries exposed to 3HTdR in vitro. Fetal ovaries were exposed to the radioactive DNA precursor at two gestational ages. The oocytes at the GV stage were analysed autoradiographitally in reconstructed serial histological sections of the ovaries matured in vivo (experiment V). The proportions of labelled GV correspond well with the proportions of labelled control P/D determined at the end of the in vitro phase 3HTdR in vitro to fetal ovaries on
No. of ovaries
Prophases (P/D) and germinal vesicles (GV)
Total
Labelled
UnlabeUed
d14
5
GV 176 (87%) P/D 1440 (90%)
26 (13%) 160 (10%)
202 1600
d16/17
3
GV 64 (35%) P/D 414 (39%)
120 (65%) 546 (61%)
184 960
In a further 5 instances, non-littermates were used, one ovary for histology, the other (with one ovary from each of three additional animals) for prophase label analysis. The results of the basically similar experiments I, II and IV were concordant and so the data for day 14 and days 16/17 of in vitro labelling have been pooled and are presented in Table 2, which gives separately the results for day 15 (experiment II) and for the partial day 14 (experiment III). The expected values have been computed from the P/D labelling proportions on the assumption that the mature ova derived from them were randomly released, viz. they were not derived from a production line system. The results of the fifth experiment are set out in Tables 3 and 4. Whereas G V labelling in the larger follicles could readily be assessed, such an assessment proved very difficult in the many, very small follicles/ oocytes, and a quantitative analysis and numerical breakdown were not attempted. Discussion We will consider first the early post-transplantation results from the first set of experiments. These yielded a very high proportion of labelled MUMII from ovaries previously exposed in vitro to 3HTdR on day 14 of gestation. This is about the time when gonia start being committed to meiosis and enter prophase. In contrast, MI/ MII harvested equally early after transplantation, but
previously exposed to label on day 16/17 of gestation, were almost completely unlabelled. Day 16/17 of gestation is about the time when a considerable proportion of gonia has completed premeiotic D N A synthesis and is at pachy- diplo-nema (Polani et al. 1981). These two contrasting results would be expected if a production hne system operated such that gonia completing premeiotic D N A synthesis, incorporating the D N A label and entering meiosis early in fetu, are preferentially released soon after puberty; those entering later in fetu (and labelled then) would not be mature and ready for release and so would not be harvestable as MUMII soon after puberty. T h e oocytes that are mature at this time would be those that had been committed to meiosis much earlier and that were already at pachy- diplo-nema when exposed to the D N A label. Clearly, given that this production hne system operates, late post-transplantation harvests should show complementary labelhng patterns: early (day 14) in vitro labelled ovaries should yield largely unlabelled MI/ MII, whereas late-labelled ovaries (days 16/17) should yield mostly labelled MI/MII. Unfortunately, these predictions have not been verified in our experimental system, as the first set of experiments showed that a direct ageing approach was not feasible, a result, probably, of a combination of radiation sensitivitiy of oocytes and heterotopic graft maintenance problems. Nevertheless, the dual results from early post-transplantation harvests indicate that an indirect approach (as adopted in experiments II and IV) can provide a valid test of the hypothesis. However, in the hght of results from experiment I, two conditions have to be met. First, the day-14 in vitro labelling results should demonstrate a marked discrepancy between the proportion of day-14 in vitro labelled P/D (as an index of premeiotic labelling of gonia) and post-transplantation MI/MII, the excess being in fav o u r of the latter. The observed difference in labelling proportions between P/D and MI/MII is in the right direction, but did not seem to us to be sufficiently marked to be conclusive, because of the high proportion of P/D labelling, and this, we assumed, was the result of several generations of gonia dividing mitotically, so that not only those at the last premeiotic D N A synthesis incorporated the label. Hence, we performed experiment III, labelling for only part of day 14, in which the expected percentage of labelled MI/MII (calculated from P/D labelling on the assumption that a production line system did not operate) was around 55%, whereas the observed value was 92% (Table 2). This sizeable discrepancy con-
Table 4. Distribution of GV labelling by type of follicles in histological sections after in vivo maturation of fetal ovaries exposed to 3HTdR in vitro. For details of prophase and GV labelling, see Table 3. Paucilaminar follicles were mostly below stage 4, multilaminar at about stage 5 and the rest at stages 6-8 of Pedersen and Peters (1968 in Peters and McNatty 1980) 3HTdR in vitro to fetal ovaries on
Paucilaminar follicles
Multilaminar follicles
Antral and Graafian follicles
Labelled
Unlabelled
Labelled
UnlabeUed
Labelled
Unlabelled
d14
115 88%
16 12%
52 87%
8 13%
9
2
202
50 43%
67 57%
14 23%
46 77%
0
7
184
d16/17
Total
69 firms that the oocytes maturing first after puberty had originated from gonia that had completed premeiotic DNA synthesis and entered the meiotic prophase earliest in fetu. The second condition that has to be met concerns ovaries exposed to 3I-ITdR in vitro on the equivalent of days 16/17 of gestation. Here it is essential to ascertain whether or not there are ample numbers and a substantial proportion of labelled oocytes within the ovary, in spite of the fact that the harvested MI/MII are almost completely unlabelled (Table 2), although labelled mature oocytes are as easily released as unlabelled ones, and could yield labelled MI/MII. The histo-autoradiographic experiment V was carded out to explore this point. The results (Table 3) show that, for the later labelling time, over one third of the GV was labelled. Table 4 examines follicle types and shows, for days 16/17, that there is a statistically significant differential decrease in the proportions of labelled GV in oocytes, from the less mature in paucilaminar follicles, to those in the multilaminar ones, to the most mature in antral and Graafian follicles. Therefore, for days 16/17, Tables 3 and 4 indicate that failure to obtain mature oocytes as labelled MI/MII depends on the underlying fact that different classes of follicles are labelled, and that those follicles whose oocytes had incorporated the DNA label in vitro at a late fetal stage were still immature at the time of the early post-transplantation histological study. Conversely, the GV of mature antral and Graafian follicles in situ were completely unlabelled, and thus the mechanically released mature oocytes yielded almost only unlabelled MI/MII (Table 2). These unlabelled mature oocytes can therefore be taken to represent those germ ceils that had entered meiosis in the fetal ovary earlier than days 16/17, and that are ready to be released soon after puberty, unlike the labelled immature oocytes that, having incorporated the DNA label late in vitro, would mature and be released later. In contrast with days 16/17, Table 4 shows that, after labelling in vitro on day 14, most GV in antral and Graafian follicles bear the label (the difference in Table 4, penultimate column, between day 14 and days 16/17 is significant: P=0.0023 from Fisher's exact 2 • 2 test); the pattern of day-14 labelling in situ and that of mature oocyte release (MI/MII) are congruent with a system in which oocytes that enter early into meiosis, mature early after puberty and are released early, as already discussed. The few unlabelled GV of mature antral follicles can be assumed to be representatives of those gonia that had become committed to meiosis very early and had completed their last premeiotic DNA synthesis before the explanted gonads had been exposed to, and penetrated by, the radioactive DNA label. The results from the numerous smaller oocytes have not tabulated because their precise scoring proved impossible. Table 2 also shows separately the results from in vitro labelling on day 15. These are intermediate with respect to MI/MII labelling between those corresponding to day 14 and those of days 16/17 in vitro exposures, and support the general conclusions of the experiments discussed. Therefore, in summary, the results of our experiments validate the PLH, viz. that the process of murine
oogenesis rests on a "first in first our, last in last out" sequence. It would be surprising if such a system of oogenesis, did not operate in other mammals, including man. However, for this we must turn to the subordinate proposition of the PLH, not tested here, which concerns Xma and univalents, viz. to the Complete PLH. We note that the experimental results of most workers in this field do not seem to support it (see Introduction); nevertheless, on balance, it seems plausible that some maternal agerelated non-disjunctional events might be associated with the production line system of oogenesis, as originally proposed by Henderson and Edwards (1968). These may occur in a strain-dependent manner, as the original experiments seem to indicate, and so be subject to genetic variation. If so, then the origin of a proportion of human non-disjunctional maternal-age-dependent chromosome anomalies, (trisomy 21, other trisomies and related errors) might reside in mechanisms that operate prenatally so that oocytes generated in late fetal life would produce ova that would tend to be chromosomally deviant with increasing maternal age. Clearly, for this part of the hypothesis, non-disjunctional chromosomes, for example trisomic ones, should be non-recombinant, at least for that proportion of recombination events that depend on Xma formation and crossing-over. Thus, direct tests of these chromosomes should provide evidence for the collateral stipulations of the PLH; such tests would be easier to perform in man than in any experimental mammal because chromosome anomalies are common, a proportion is compatible with prolonged survival after birth and they are easily recognised. Furthermore, the testing of trisomic chromosomes for recombination becomes practical now that recombinant DNA methods and polymorphisms are available, as P. A. Jacobs and T. J. Hassold (personal communication) have shown. Indeed their preliminary results for sex chromosome trisomies suggest that the relevant chromosomes are not only often non-recombinant but also that maternal age at the birth of trisomics with non-recombinant chromosomes tends to be higher than that of subjects with recombinant chromosomes. With respect to maternal age, the results for trisomy 21 are generally not in line with the above and suggest that failure to pair-exchange is relatively unimportant for this specific non-disjunction (see also Takaesu et al. 1990). However, a map of the long arm of trisomic chromosomes 21 of maternal derivation appears to be significantly shorter than that of normal female chromosomes 21, suggesting that pairing or recombination problems may underlie the non-disjunctional origin of these autosomal trisomies (Sherman et al. 1990). Furthermore, an age-related decrease in recombination in specific chromosome 21 regions has been noted (Warren et al. 1987; Perroni et al. 1990; Tanzi et al. 1990; cf. with the findings of Wallace et al. 1976, in the mouse) and attributed to failure of Xma formation. Clearly, then, at least some of these observations support the Complete PLH. It should be added however that, even so, probably only a proportion of maternalage dependent errors at the first meiotic division might
70 b e s e e d e d p r e n a t a l l y in t h e m a n n e r e n v i s a g e d by H e n d e r s o n a n d E d w a r d s (1968). P r e s u m a b l y , o t h e r e r r o r s at t h e first a n d s e c o n d d i v i s i o n s in t h e f e m a l e , a n d p a t e r n a l e r r o r s h a v e d i f f e r e n t o r i g i n s ( s e e r e v i e w b y P o l a n i 1981, a n d m o n o g r a p h o f B o n d a n d C h a n d l e y 1983). A p a r t f r o m g e n e t i c f a c t o r s , e n v i r o n m e n t a l c o n d i t i o n s in t h e f e m a l e a r e l i k e l y t o act p o s t n a t a U y o n a n o v a r y t h a t for y e a r s m a y b e a t a r g e t f o r t h e i r c u m u l a t i v e influences. O n t h e o t h e r h a n d , s o m e f a c t o r s m a y act only at the e x t r e m e s o f r e p r o d u c t i v e life, f o r e x a m p l e in o l d age w h e n h o r m o n e s m a y e x e r c i s e a n u n w a n t e d activity o n o o c y t e - d e p l e t e d o v a r i e s . A l t h o u g h p o i n t e r s s u p p o r t t h e s e ideas, m u c h is b a s e d o n s p e c u l a t i o n . T h e triggers o f p a t e r n a l l y derived numerical chromosome anomalies are even more hypothetical.
Acknowledgements. We are grateful to Professor P. Armitage for his valuable comments on the statistical aspects of the results. We thank Professor M. Bobrow for his interest, Action Research for the Crippled Child, the Mr. and Mrs. Archie Sherman Foundation, the Spastics Society and the Generation Trust for their financial support, Mrs. B. Merchant for her help with the bibliography, and Miss A. Knight for work on the manuscript.
References Beermann F, Bartels I, Franke U, Hansmann I (1987) Chromosome segregation at meiosis I in female T (2;4)lGo/+mice: no evidence for a decreased crossover frequency with maternal age. Chromosoma 95:1-7 Bodmer W F (1961) Effects of maternal age on the incidence of congenital abnormalities in mouse and man. Nature 190:11341135 Boer P de, Hoeven F A van der (1980) The use of translocation-derived "marker-bivalents" for studying the origin of meiotic instability in female mice. Cytogenet Cell Genet 26: 49-58 Bond D J, Chandley AC (1983) Aneuploidy. (Oxford monographs on medical genetics 11) Oxford University Press, Oxford, pp 55-76 Gude WD (1968) Autoradiographic techniques: localization of radioisotopes in biological material. Prentice-Hall, Englewood Cliffs, NJ Henderson SA (1969) Chromosome pairing, chiasmata and crossing-over. In: Lima-de-Faria A (ed) Handbook of molecular cytology. North-Holland, Amsterdam, pp 326-357 Henderson SA (1970) The time and place of meiotic crossing-over. Annu Rev Genet 4: 295-324 Henderson SA, Edwards R G (1968) Chiasma frequency and maternal age in mammals. Nature 218:22-28 Hummler E, Theuring F, Hansmarm I (1987) Meiotic nondisjunction in oocytes from aged Djungarian hamsters correlates with an alteration in meiosis rate but not in univalent formation. Hum Genet 76: 357-364 Jagiello G, Fang JS (1979) Analyses of diplotene chiasma frequencies in mouse oocytes and spermatocytes in relation to ageing and sexual dimorphism. Cytogenet Cell Genet 23:53-60 John B (1990) Meiosis. Cambridge University Press, Cambridge, pp 70-73 Luthardt FW, Palmer CG, Yu P-L (1973) Chiasma and univalent frequencies in aging female mice. Cytogenet Cell Genet 12: 68-79
Peters H, McNatty KP (1980) The ovary. A correlation of structure and function in mammals. Granada, London, p 15 Polani PE (1981) Chiasmata, Down syndrome and nondisjunction: an overview. In: Cruz F de la, Gerald PS (eds) Trisomy 21 (Down syndrome) research perspectives. University Park Press, Balitmore, pp 111-130 Polani PE, Jagiello GM (1976) Chiasmata, meiotic univalents and age in relation to aneuploid imbalance in mice. Cytogenet Cell Genet 16: 505-529 Polani PE, Crolla JA, Seller MJ, Moir F (1979) Meiotic crossing over exchange in the female mouse visualised by BUdR substitution. Nature 278: 348-349 Polani PE, Crolla JA, Seller MJ (1981) An experimental approach to female mammalian meiosis: differential chromosome labelling and an analysis of chiasmata in the female mouse. In: Jagielio G, Vogel HJ (eds) Bioregulators of reproduction. (P & S biomedical sciences symposia series) Academic Press, New York, pp 59-97 Perroni L, Bricarelli D, Grasso M, Piefluigi M, Baldi M, Pedemonte C, Strigini P (1990) Crossing over and chromosome 21 nondisjunction: a study of 60 families. Am J Med Genet 7[Suppl] : 141-147 Reid DH, Parsons PA (1963) Sex of parent and variation of recombination with age in the mouse. Heredity 18:107-108 Sherman SL, Takaesu N, Freeman S, Phillips C, Blackston RD, Kanta B J, Jacobs PA, Cockwell AE, Kurnitt D, Uchida I, Hassold TJ (1990) Tfisomy 21: association between reduced recombination and non-disjunction. Am J Hum Genet 47: A97 Slizynski BM (1960) Sexual dimorphism in mouse gametogenesis. Genet Res 1 : 477-486 Speed RM (1977) The effects of ageing on the meiotic chromosomes of male and female mice. Chromosoma 64: 241-254 Speed RM, Chandley AC (1983) Meiosis in the fetal mouse ovary. II. Oocyte development and age related aneuploidy. Does a production line exist? Chromosoma 88:184-189 Sugawara S, Mikamo K (1983) Absence of correlation between univalent formation and meiotic nondisjunction in aged female Chinese hamsters. Cytogenet Cell Genet 35: 34-40 Sugawara S, Mikamo K (1986) Maternal ageing and nondisjuncfion: a comparative study of two chromosomal techniques on the formation of univalents in first meiotic metaphase oocytes of the mouse. Chromosoma 93 : 321-325 Takaesu N, Jacobs PA, Cockwell A, Blackston RD, Freeman S, Nuccio J, Kurnit DM, Uchida I, Freeman V, Hassold T (1990) Nondisjunction of chromosome 21. Am J Med Genet 7[Suppl] : 175-181 Tanzi RE, Haines JL, Gusella JF (1990) Detailed genetic linkage map of human chromosome 21: patterns of recombination according to age and sex. Prog Clin Biol Res 360: pp 15-26 Tarkowski AK (1966) An airdrying method for chromosome preparations for mouse eggs. Cytogenetics 5: 394-400 Tease C, Fisher G (1986) Further examination of the productionline hypothesis in mouse foetal oocytes. I. Inversion heterozygotes. Chromosoma 93: 447-452 Tease C, Fisher G (1989) Further examination of the productionline hypothesis in mouse foetal oocytes. II. T(14;15)6Ca heterozygotes. Chromosoma 97: 315-320 Wallace ME, MacSwiney FJ, Edwards RG (1976) Parental age and recombination frequency in the house mouse. Genet Res 28: 241-251 Warren AC, Chakravarti A, Wong C, Slaugenhaupt SA, Halloran SL, Watkins PC, Metaxotou C, Antonarakis SE (1987) Evidence for reduced recombination on the nondisjoined chromosomes 21 in Down syndrome. Science 237: 652-654
9 Springer-Verlag1991
A test of the production line hypothesis of mammalian oogenesis Paul E. Polani and John A . Crolla*
Division of Molecular and Medical Genetics,Paediatric Research Unit, United Medical and Dental Schools of Guy's and St. Thomas's Hospitals, 7th Floor, Guy's Hospital Tower, London Bridge, London SE1 9RT, UK Received June 20, 1989 / Revised April 22, 1991
Summary. Germ cells in female mammals become committed to meiosis and enter its prophase sequentially in fetal life and, according to the Production Line Hypothesis, the oocytes thus generated are released after puberty as mature ova in the same sequence as that of meiotic entry in fetu. This hypothesis in its original and complete form has a subordinate proposition that concerns chiasma (Xma) frequency; it postulates that Xma number would decrease with fetal age. Consequently, univalents would increase, leading to errors of chromosome disjunction at the first meiotic division (MI), and thus to maternal age-dependent numerical chromosome anomalies. By using an in vitro/in vivo approach, we radioactively labelled the DNA of germ cells at premeiotic synthesis as they sequentially entered meiosis, while the fetal ovaries were in culture. At the end of this in vitro phase, pachytene/diplotene (P/D) stages were studied to determine their labelled fraction. The ovaries were then transplanted to spayed females and, after the in vivo phase, mature ova were harvested and the proportion of labelled first and second meiotic metaphases (MI/MII) determined. By marking the germ cells with label while in vitro during periods equivalent to early and late gestation, and by comparing the observed proportions of labelled MI/MII with those of oocytes labelled at P/D, we concluded that, in the mouse, ova do not mature at random for release, but are formed according to a production line system in which the time of release after puberty is related to the time of entry into meiosis in fetu.
Introduction
In an attempt to explain the maternal age-dependence to trisomy 21 in man, while taking account of the fact that in female mammals the genetically important events of gametogenesis (crossing-over, chiasma (Xma) formation, * Present address: Wessex RegionalGenetics Laboratory, General Hospital, Salisbury,WiltshireSP2 7SX, UK Offprint requests to: P. E. Polani, "Little Meadow",West Clandon, Surrey GU4 7TL, UK
and recombination) happen before their birth, Henderson and Edwards (1968) proposed that the order of the entry of germ cells into meiosis, i.e. of oocyte generation, dictated the order of their release as mature ova from puberty to the end of fertile life. Furthermore, they suggested that oocytes formed late in fetal life had fewer Xma holding their bivalents together then early ones. Homologous chromosomes that are paired to form bivalents, but that are devoid of Xma, would, because of the absence of the mechanical function of the latter, behave as univalents and so would not disjoin regularly from each other, frequently producing unbalanced ova and zygotes. The frequency of the latter would increase progressively with advancing age of the female because the old ova would be derived from oocytes generated late prenatally. This, the "Complete Production Line Hypothesis" (Complete PLH), is clearly relevant not only to the origin of trisomy 21, but also to that of all numerical chromosome anomalies of maternal derivation traceable to errors of MI and correlated with maternal age, viz. the majority of such anomalies; this should be true not only for man, on whom data of such anomalies are substantial, but also for experimental animals. The Complete PLH originated from Henderson and Edwards' observations on adult mice; they found that oocytes at MI, harvested from older females, had lower Xma counts and more univalents than those from younger ones. These results were interpreted in the light of an essential aspect of Xrna, namely that "there is no loss of chiasmata once formed" (Henderson and Edwards 1968) in the early prophase of meiosis, right up to the time of diakinesis and MI (Henderson 1969, 1970; but see Slizynski 1960, and John 1990): this is the basis for the hypothesis. The hypothesis was subsequently subjected to a variety of tests. Using cytological techniques, Luthardt et al. (1973) confirmed the decrease in Xma frequency and the increase in univalents with age, as did Polani and Jagiello (1976), who also studied MI segregation of chromosomes. They concluded that most MI univalents, although incremental with age, were artefactual. Suyawara and Mikamo (1986; and 1983 in the Chinese hamster) supported this view; Speed (1977), who had noted the drop in Xma, could not confirm the increase of univalents. Hummler
65 et al. (1987) found evidence in the Djungarian h a m s t e r arguing against a production line system. O t h e r workers based their tests on the use of structural c h r o m o s o m e anomalies. Thus, T e a s e and Fisher (1986) investigated meiotic pairing in inversion heterozygosity; their results were not inconsistent with an oogenesis production line system in the mouse. H o w e v e r , in a further study (1989), which included analysis of M I at different ages, they aligned themselves against such a system. This too was the stance of B e e r m a n n et al. (1987) on the grounds of an increase with age and not a decreased calculated cross-over frequency within a translocation; de B o e r and van der H o e v e n (1980) exploited the long product of a translocation and found no decrease of X m a with age; they concluded that non-disjunction was mostly the result of postnatal oocyte ageing. Finally, using a novel and demanding approach as a test of a f u n d a m e n tal stipulation of the C o m p l e t e P L H , Jagiello and Fang (1979) showed that late c o m p a r e d with early fetal oocytes at diplonema had fewer X m a and m o r e univalents. H o w ever, these findings could not be duplicated b y Speed and Chandley (1983) on a different Swiss strain of mice because, in their hands, X m a identification p r o v e d difficult. G e n e recombination frequencies, as tests of the C o m plete P L H , could be used to c o r r o b o r a t e the cytological work, and older studies showing age-related changes in recombination seemed supportive ( B o d m e r 1961; Reid and Parsons 1963). H o w e v e r , ad hoc tests (Wallace et al. 1976) indicated that the changes at the different tested loci were not all in the direction expected f r o m the hypothesis, and that changes in recombination location might also occur with age. All the above work addresses the C o m p l e t e P L H , but is an indirect test of the core of the hypothesis. It is centred either on testing for a decrease with age (usually postn a t a l , exceptionally fetal) of X m a frequency with a corresponding increase of univalents, accepting implicitly that numbers of X m a , once formed, are i m m u t a b l e , or else on tests based on age-related changes of r e c o m b i n a tion frequencies. On balance, the bulk of the w o r k does not support the hypothesis. H o w e v e r , n o n e of the studies have set out to probe directly the "first in first out", last in last out" stipulation, the key tenet of the hypothesis to which the X m a reduction with univalent formation (and therefore the non-disjunction effect) are special subordinate propositions. In order to test the P L H in this n a r r o w e r but fundamental sense, the following experiments were conducted, based on tracking and analysing mature ova derived from oocytes g e n e r a t e d at different times during fetal life and marked before entry into fetal meiosis. A previous a t t e m p t to track fetal oocytes by marking them at premeiotic D N A synthesis, by means of injected radioactive DNA precursors into the m o t h e r during pregnancy, had not been successful (G. Jagiello and F. Giannelli, personal communication). In o r d e r to bypass blocking by maternal and fetal barriers, we m a d e use of an in vitro/in vivo technique that we had developed for this purpose (Polani et al. 1979; 1981). It consists of explanting and culturing i m m a t u r e fetal ovaries, during which time the oocytes, generated as if in fetu, are
m a r k e d radioactively at premeiotic D N A synthesis. After the in vitro phase, heterotopic ovarian transplantation into spayed females permits maturation of oocytes in vivo under hormonal control, and their subsequent harvesting and study.
Materials and methods
Mouse strains and breeding Two strains of laboratory mouse were used, a random bred A Strong and a congenic C3H/He. Timed rantings were carried out, and the day of vaginal plug was called day 1 of gestation.
Culture In all experiments and for all labelling protocols, pregnant females were culled on day 14 by cervical dislocation and the fetal ovaries were dissected free of their adherent mesonephric tissue prior to being set up in an organ culture system, the details of which are described by Polani et al. (1979 and 1981; here, we adopted a multiwell system in preference to the raft method). In experiments I-III, ovaries from several fetuses (derived from the same pregnant female) were pooled prior to setting up the culture. In experiments IV and V, the ovaries from individual fetuses were cultured in pairs until specific manipulations were carried out. While in vitro, the ovaries were exposed to 3HTdR (0.1 ~tCi/ml; specific activity 0.5 Ci/mmol) by transferring them to multiwells on agar and by incubating them in medium preconditioned with 3HTdR. The developing ovaries were exposed to the label for 10-18 h on day 14, or for the whole of day 14, or day 15, or days 16/17 gestation equivalents. At the end of the labelling incubation, the ovaries were removed from the culture wells, washed in three changes of complete medium without label, and prepared for either air-dry spreading, or for transplantation as indicated by the protocol of the experiment.
Transplantation After 8 days in culture, the fetaI ovaries were removed from the culture medium, washed in phosphate buffered saline and transplanted under the kidney capsule of a previously spayed young adult female of matching strain.
Air-drying and spreading of meiotic chromosome We studied both prophase and metaphase cells. For the former, at the pachyteneldiplotene stages, the fetal ovaries were placed into 0.7% sodium citrate for 30rain at room temperature, transferred onto microscope slides in a drop of hypotonic citrate, and finely mashed with iris knives; the resulting cell suspension was spread over the centre of the slide. A small drop of 3:1 methanol: acetic acid was immediately dropped onto the cells and the preparation allowed to air-dry; the prophase cells were stained in lacto-aeetic orcein prior to autoradiography. Successfully transplanted ovaries, removed from the renal site, were mechanically dissociated and their freed mature germinal-vesicles-containing oocytes (GV) were selected under low power phase and kept in vitro for 4 h at 37~ in Ham's F10 supplemented with 10% donor calf serum (Flow). Chromosome spreads (MI/MII) were prepared according to Tarkowski (1966) and stained in lacto-orcein, prior to autoradiography.
Histology and autoradiography In experiment v , transplanted ovaries were washed in two changes of phosphate buffered saline, fixed in Bouin's for 4-4,5 h, and
66 paraffin-embedded for serial-section histology. Sections were cut at a thickness of 10 Isrn, de-waxed in xylene, dehydrated through a series of alcohols, floated onto poly-L-lysine pretreated microscope slides, and coated with autoradiographic stripping film (Kodak AR10). The preparation of P/D and MI/MII for autoradiography was similar. Slides were exposed in the dark at 4~ for 2 weeks (prophase/metaphase) or 6 weeks (histology). All slides were developed in Kodak D19 and fixed with Kodak Rapid Fixer. The sections were stained through the stripping film with toluidine blue in borax (the techniques followed are in principle described by Gude [1968]). The serial sections were systematically analysed and drawn
for follicle reconstruction. Autoradiographic g a i n s were visualised and counted under direct light or phase contrast illumination with a Zeiss photomicroscope III. The autoradiographie analysis and histological scoring of all experimental and control material (see Results) were performed blind. Slides were coded by an independent person and decoded only at the end of the whole experiment. The laminarity of follicles and the compactness, or otherwise (antral and Graafian follicles) of the follicular cells were recorded, and the autoradiographic grains were assessed and counted over the GV, the follicular and stromal cells and the spaces within follicles.
Table 1. Summary of experimental and control data (with percentages of labelled prophases, P/D, and metaphases, MI/MII) Experiment
System and day (d) 3HTdR labelling
I
Ovaries for P/D a
12
Direct ageing
Ovaries transplanted
77
Ovaries harvested b
Experimental
31 702 892
(86%) (48%)
MI/MII yield
d14 d16/17
106 100
(93%) (2%)
6.6 (206/31)
Ovaries for P/D
Early harvest
Ovaries transplanted
28
(57% take)
223 8.0 (233/28)
29
Ovaries harvested
40
133 114
(86% take)
P/D yield part
d14 d14 d15 d16/17
4486 2310 1420 963
(55%) (86%) (78%) (37%)
MI/MII yield part
d14 d14 d15 d16/17
177 343 180 98
(92%) (98%) (41%) (2%)
7.0 (798/114)
MI/MII per ovary IV
Ovaries for P/D
16
Littermate
Ovaries transplanted
16
Ovaries harvested
Histoautoradiography
(40% take)
d14 d16/17
II and III
V
49
P/D yielda
MI/MII per ovary
Control
14
39
(98% take)
404
10.4 (404/39)
16 (88% take)
P/D yield
d14 d16/17
608 849
(90%) (23%)
MI/MII yield
d14 d16/17
50 70
(94%) (0%)
14
(88% take)
B
82
MI/MII per ovary
8.6 (120/14)
5.9 (82/14)
Ovaries for P/D
8
m
Ovaries transplanted
8
3c
Ovaries for histology
8
3c
P/D yield
d14 d16/17
1600 960
(90%) (43%)
GV yield
d14 d16/17
202 184
(83%) (35%)
GV per ovary
48.25 (386/8)
a In all experiments, data on prophases are control data for the expectations of metaphase labelling b Productive takes only c Littermates to animals in experiment IV
158 52.7 (158/3)
67 Finally, with respect to all our experiments, the critical point of the in vitro/in vivo method is that, whereas it ensures the required control, it nevertheless permits oogenesis to proceed normally. The in vitro phase allows fetal germ cells to become committed to meiosis in a regular fashion; the in vivo heterotopically transplanted ovaries support normal maturation of the resident oocytes. During the two phases, the numbers of oocytes that are generated and that mature are close to normal. A further measure of normality is that, in orthotopic transplantation experiments using genetically marked fetal ovaries, the mated recipient females produced and reared normal young; thus, both the maturation and the fertilisability of the transplanted oocytes are normal (Polani et al. 1981).
Results
Following exposure of the fetal ovaries to 3HTdR in vitro, we set out to establish the proportions of radioactively labelled m a t u r e oocytes available after transplantation, and to c o m p a r e them with the proportions of
Table 2. Labelled mature ova (MI/MII) derived from fetal germ cells labelled in vitro. The expected labelling values of MI/MII, calculated from the proportion of meiotic prophase celis found to be labelled at the end of the in vitro phase, assume that the PLH is not valid. The expected values are all statistically significantly different from the observed labelling data. The days on which the in vitro cultured fetal ovaries were supplied with the radioactive DNA precursor correspond to gestation ages. The metaphases were harvested early after transplantation
Ovaries Labelled explanted on day 14, 3HTdR in vitro on d14
Unlabelled
Total
Observed
484 (97.0%)
15
499
Expected
431
68
499
495
3620
14
177
80
177
2019
4486
107
180
40
180
317
1420
264
268
171
268
1722
2 704
MI/MII (86.3%) P/D 10-18 h of d14
Observed Observed
3125 163 (92.0%)
MI/MII Expected
97
(55.0%) P/D
Observed Observed
dt5
2467 73 (41.0%)
MI/MII Expected
140 (77.7%)
P/D
d16/17
Observed Observed
1103 4 (1.5%)
MI/MII Expected
97 (36.3%)
Total
P/D
Observed
MI/MII P/D
Observed Observed
982
1124 12 230
aHTdR-o-o-o-to ovaries Littermotes in vitro on day 1L I 15 116 ~ 17 linvivo for 19doys Then .o-o-o-
2t 3t
(ARG}~ .~
,
o.o .....
Fig. 1. Schema
ff the littermate/shared-recipient experiment (IV), comparing metaphases autoradiographically (MI/MII [ARG]) in experimental animals with their littermate controls, obtaining background labelling on P/D (Pachy) from sister ovaries of experimental animals, and using the two kidneys of the same recipient for engraftment of ovaries labelled in vitro on day 14 or on days 16/ 17 (cranially) together with unlabelled control ovaries (caudally) from a littermate
labelled prophases at the end of the in vitro phase, as a labelling baseline; the comparison provides the test of the P L H . W e carried out five sets of e x p e r i m e n t s (I-V), and the numerical data are set out in T a b l e 1. In experim e n t I, following labelling in vitro on either day 14, or o n gestation equivalent days 16/17, we set out to obtain M I / M I I on day 19 after transplantation and at weeks 5, 7, 9, 11, 15 and 20. H o w e v e r , no m e t a p h a s e s could be harvested f r o m ovaries 9 weeks or longer after transplantation, although ovaries survived well and o v a could be harvested with similar success rates as f r o m control ovaries up to, and including, 7 weeks post-transplantation. Hence, these results f r o m the early harvests have b e e n pooled and included in the final d a t a (Table 2). T h e subsequent experiments t o o k these results into account and were all early harvest (day 19) experiments. T h e first experiments also showed that m o s t prophases were labelled (see Discussion). T h e r e f o r e , we tested the P L H m o r e critically by labelling a m o r e discrete population of gonia, in experiment I I I and we labelled the cultures for only 10-18 h on day 14. In e x p e r i m e n t II, label was supplied to cultured ovaries on day 14 and days 16/ 17, and also on day 15. O v a were harvested early. In experiment IV, which was designed to minimise variation, fetuses of the same litter were allocated to e x p e r i m e n t a l or control groups (Fig. 1). A m o n g the f o r m e r , pairs of ovaries (from one fetus) were labelled on day 14, or, alternatively, on days 16/17 ( f r o m a littermate). O f each pair of these ovaries, one was used to establish the proportion of prophases m a r k e d , whereas the other was transplanted for the in vivo phase. Control ovaries (from a third littermate) were co-transplanted with the experimental ovaries: the day-14-1abeUed ovary and its control under the capsule of the right kidney, the days-16t17labelled ovary, with its control, to the left side of the same recipient. Finally, 8 fetuses were used in experiment V. We could select 3 littermates to the fetuses in experiment IV. O f each pair of these three ovaries, one was labelled (day 14 or days 16/17), and the o t h e r was an unlabelled control. A f t e r 19 days, the test and control gonads were serially sectioned for histo-autoradiography.
68 Table 3. Distribution of GV labelling in histological sections of transplanted fetal ovaries exposed to 3HTdR in vitro. Fetal ovaries were exposed to the radioactive DNA precursor at two gestational ages. The oocytes at the GV stage were analysed autoradiographitally in reconstructed serial histological sections of the ovaries matured in vivo (experiment V). The proportions of labelled GV correspond well with the proportions of labelled control P/D determined at the end of the in vitro phase 3HTdR in vitro to fetal ovaries on
No. of ovaries
Prophases (P/D) and germinal vesicles (GV)
Total
Labelled
UnlabeUed
d14
5
GV 176 (87%) P/D 1440 (90%)
26 (13%) 160 (10%)
202 1600
d16/17
3
GV 64 (35%) P/D 414 (39%)
120 (65%) 546 (61%)
184 960
In a further 5 instances, non-littermates were used, one ovary for histology, the other (with one ovary from each of three additional animals) for prophase label analysis. The results of the basically similar experiments I, II and IV were concordant and so the data for day 14 and days 16/17 of in vitro labelling have been pooled and are presented in Table 2, which gives separately the results for day 15 (experiment II) and for the partial day 14 (experiment III). The expected values have been computed from the P/D labelling proportions on the assumption that the mature ova derived from them were randomly released, viz. they were not derived from a production line system. The results of the fifth experiment are set out in Tables 3 and 4. Whereas G V labelling in the larger follicles could readily be assessed, such an assessment proved very difficult in the many, very small follicles/ oocytes, and a quantitative analysis and numerical breakdown were not attempted. Discussion We will consider first the early post-transplantation results from the first set of experiments. These yielded a very high proportion of labelled MUMII from ovaries previously exposed in vitro to 3HTdR on day 14 of gestation. This is about the time when gonia start being committed to meiosis and enter prophase. In contrast, MI/ MII harvested equally early after transplantation, but
previously exposed to label on day 16/17 of gestation, were almost completely unlabelled. Day 16/17 of gestation is about the time when a considerable proportion of gonia has completed premeiotic D N A synthesis and is at pachy- diplo-nema (Polani et al. 1981). These two contrasting results would be expected if a production hne system operated such that gonia completing premeiotic D N A synthesis, incorporating the D N A label and entering meiosis early in fetu, are preferentially released soon after puberty; those entering later in fetu (and labelled then) would not be mature and ready for release and so would not be harvestable as MUMII soon after puberty. T h e oocytes that are mature at this time would be those that had been committed to meiosis much earlier and that were already at pachy- diplo-nema when exposed to the D N A label. Clearly, given that this production hne system operates, late post-transplantation harvests should show complementary labelhng patterns: early (day 14) in vitro labelled ovaries should yield largely unlabelled MI/ MII, whereas late-labelled ovaries (days 16/17) should yield mostly labelled MI/MII. Unfortunately, these predictions have not been verified in our experimental system, as the first set of experiments showed that a direct ageing approach was not feasible, a result, probably, of a combination of radiation sensitivitiy of oocytes and heterotopic graft maintenance problems. Nevertheless, the dual results from early post-transplantation harvests indicate that an indirect approach (as adopted in experiments II and IV) can provide a valid test of the hypothesis. However, in the hght of results from experiment I, two conditions have to be met. First, the day-14 in vitro labelling results should demonstrate a marked discrepancy between the proportion of day-14 in vitro labelled P/D (as an index of premeiotic labelling of gonia) and post-transplantation MI/MII, the excess being in fav o u r of the latter. The observed difference in labelling proportions between P/D and MI/MII is in the right direction, but did not seem to us to be sufficiently marked to be conclusive, because of the high proportion of P/D labelling, and this, we assumed, was the result of several generations of gonia dividing mitotically, so that not only those at the last premeiotic D N A synthesis incorporated the label. Hence, we performed experiment III, labelling for only part of day 14, in which the expected percentage of labelled MI/MII (calculated from P/D labelling on the assumption that a production line system did not operate) was around 55%, whereas the observed value was 92% (Table 2). This sizeable discrepancy con-
Table 4. Distribution of GV labelling by type of follicles in histological sections after in vivo maturation of fetal ovaries exposed to 3HTdR in vitro. For details of prophase and GV labelling, see Table 3. Paucilaminar follicles were mostly below stage 4, multilaminar at about stage 5 and the rest at stages 6-8 of Pedersen and Peters (1968 in Peters and McNatty 1980) 3HTdR in vitro to fetal ovaries on
Paucilaminar follicles
Multilaminar follicles
Antral and Graafian follicles
Labelled
Unlabelled
Labelled
UnlabeUed
Labelled
Unlabelled
d14
115 88%
16 12%
52 87%
8 13%
9
2
202
50 43%
67 57%
14 23%
46 77%
0
7
184
d16/17
Total
69 firms that the oocytes maturing first after puberty had originated from gonia that had completed premeiotic DNA synthesis and entered the meiotic prophase earliest in fetu. The second condition that has to be met concerns ovaries exposed to 3I-ITdR in vitro on the equivalent of days 16/17 of gestation. Here it is essential to ascertain whether or not there are ample numbers and a substantial proportion of labelled oocytes within the ovary, in spite of the fact that the harvested MI/MII are almost completely unlabelled (Table 2), although labelled mature oocytes are as easily released as unlabelled ones, and could yield labelled MI/MII. The histo-autoradiographic experiment V was carded out to explore this point. The results (Table 3) show that, for the later labelling time, over one third of the GV was labelled. Table 4 examines follicle types and shows, for days 16/17, that there is a statistically significant differential decrease in the proportions of labelled GV in oocytes, from the less mature in paucilaminar follicles, to those in the multilaminar ones, to the most mature in antral and Graafian follicles. Therefore, for days 16/17, Tables 3 and 4 indicate that failure to obtain mature oocytes as labelled MI/MII depends on the underlying fact that different classes of follicles are labelled, and that those follicles whose oocytes had incorporated the DNA label in vitro at a late fetal stage were still immature at the time of the early post-transplantation histological study. Conversely, the GV of mature antral and Graafian follicles in situ were completely unlabelled, and thus the mechanically released mature oocytes yielded almost only unlabelled MI/MII (Table 2). These unlabelled mature oocytes can therefore be taken to represent those germ ceils that had entered meiosis in the fetal ovary earlier than days 16/17, and that are ready to be released soon after puberty, unlike the labelled immature oocytes that, having incorporated the DNA label late in vitro, would mature and be released later. In contrast with days 16/17, Table 4 shows that, after labelling in vitro on day 14, most GV in antral and Graafian follicles bear the label (the difference in Table 4, penultimate column, between day 14 and days 16/17 is significant: P=0.0023 from Fisher's exact 2 • 2 test); the pattern of day-14 labelling in situ and that of mature oocyte release (MI/MII) are congruent with a system in which oocytes that enter early into meiosis, mature early after puberty and are released early, as already discussed. The few unlabelled GV of mature antral follicles can be assumed to be representatives of those gonia that had become committed to meiosis very early and had completed their last premeiotic DNA synthesis before the explanted gonads had been exposed to, and penetrated by, the radioactive DNA label. The results from the numerous smaller oocytes have not tabulated because their precise scoring proved impossible. Table 2 also shows separately the results from in vitro labelling on day 15. These are intermediate with respect to MI/MII labelling between those corresponding to day 14 and those of days 16/17 in vitro exposures, and support the general conclusions of the experiments discussed. Therefore, in summary, the results of our experiments validate the PLH, viz. that the process of murine
oogenesis rests on a "first in first our, last in last out" sequence. It would be surprising if such a system of oogenesis, did not operate in other mammals, including man. However, for this we must turn to the subordinate proposition of the PLH, not tested here, which concerns Xma and univalents, viz. to the Complete PLH. We note that the experimental results of most workers in this field do not seem to support it (see Introduction); nevertheless, on balance, it seems plausible that some maternal agerelated non-disjunctional events might be associated with the production line system of oogenesis, as originally proposed by Henderson and Edwards (1968). These may occur in a strain-dependent manner, as the original experiments seem to indicate, and so be subject to genetic variation. If so, then the origin of a proportion of human non-disjunctional maternal-age-dependent chromosome anomalies, (trisomy 21, other trisomies and related errors) might reside in mechanisms that operate prenatally so that oocytes generated in late fetal life would produce ova that would tend to be chromosomally deviant with increasing maternal age. Clearly, for this part of the hypothesis, non-disjunctional chromosomes, for example trisomic ones, should be non-recombinant, at least for that proportion of recombination events that depend on Xma formation and crossing-over. Thus, direct tests of these chromosomes should provide evidence for the collateral stipulations of the PLH; such tests would be easier to perform in man than in any experimental mammal because chromosome anomalies are common, a proportion is compatible with prolonged survival after birth and they are easily recognised. Furthermore, the testing of trisomic chromosomes for recombination becomes practical now that recombinant DNA methods and polymorphisms are available, as P. A. Jacobs and T. J. Hassold (personal communication) have shown. Indeed their preliminary results for sex chromosome trisomies suggest that the relevant chromosomes are not only often non-recombinant but also that maternal age at the birth of trisomics with non-recombinant chromosomes tends to be higher than that of subjects with recombinant chromosomes. With respect to maternal age, the results for trisomy 21 are generally not in line with the above and suggest that failure to pair-exchange is relatively unimportant for this specific non-disjunction (see also Takaesu et al. 1990). However, a map of the long arm of trisomic chromosomes 21 of maternal derivation appears to be significantly shorter than that of normal female chromosomes 21, suggesting that pairing or recombination problems may underlie the non-disjunctional origin of these autosomal trisomies (Sherman et al. 1990). Furthermore, an age-related decrease in recombination in specific chromosome 21 regions has been noted (Warren et al. 1987; Perroni et al. 1990; Tanzi et al. 1990; cf. with the findings of Wallace et al. 1976, in the mouse) and attributed to failure of Xma formation. Clearly, then, at least some of these observations support the Complete PLH. It should be added however that, even so, probably only a proportion of maternalage dependent errors at the first meiotic division might
70 b e s e e d e d p r e n a t a l l y in t h e m a n n e r e n v i s a g e d by H e n d e r s o n a n d E d w a r d s (1968). P r e s u m a b l y , o t h e r e r r o r s at t h e first a n d s e c o n d d i v i s i o n s in t h e f e m a l e , a n d p a t e r n a l e r r o r s h a v e d i f f e r e n t o r i g i n s ( s e e r e v i e w b y P o l a n i 1981, a n d m o n o g r a p h o f B o n d a n d C h a n d l e y 1983). A p a r t f r o m g e n e t i c f a c t o r s , e n v i r o n m e n t a l c o n d i t i o n s in t h e f e m a l e a r e l i k e l y t o act p o s t n a t a U y o n a n o v a r y t h a t for y e a r s m a y b e a t a r g e t f o r t h e i r c u m u l a t i v e influences. O n t h e o t h e r h a n d , s o m e f a c t o r s m a y act only at the e x t r e m e s o f r e p r o d u c t i v e life, f o r e x a m p l e in o l d age w h e n h o r m o n e s m a y e x e r c i s e a n u n w a n t e d activity o n o o c y t e - d e p l e t e d o v a r i e s . A l t h o u g h p o i n t e r s s u p p o r t t h e s e ideas, m u c h is b a s e d o n s p e c u l a t i o n . T h e triggers o f p a t e r n a l l y derived numerical chromosome anomalies are even more hypothetical.
Acknowledgements. We are grateful to Professor P. Armitage for his valuable comments on the statistical aspects of the results. We thank Professor M. Bobrow for his interest, Action Research for the Crippled Child, the Mr. and Mrs. Archie Sherman Foundation, the Spastics Society and the Generation Trust for their financial support, Mrs. B. Merchant for her help with the bibliography, and Miss A. Knight for work on the manuscript.
References Beermann F, Bartels I, Franke U, Hansmann I (1987) Chromosome segregation at meiosis I in female T (2;4)lGo/+mice: no evidence for a decreased crossover frequency with maternal age. Chromosoma 95:1-7 Bodmer W F (1961) Effects of maternal age on the incidence of congenital abnormalities in mouse and man. Nature 190:11341135 Boer P de, Hoeven F A van der (1980) The use of translocation-derived "marker-bivalents" for studying the origin of meiotic instability in female mice. Cytogenet Cell Genet 26: 49-58 Bond D J, Chandley AC (1983) Aneuploidy. (Oxford monographs on medical genetics 11) Oxford University Press, Oxford, pp 55-76 Gude WD (1968) Autoradiographic techniques: localization of radioisotopes in biological material. Prentice-Hall, Englewood Cliffs, NJ Henderson SA (1969) Chromosome pairing, chiasmata and crossing-over. In: Lima-de-Faria A (ed) Handbook of molecular cytology. North-Holland, Amsterdam, pp 326-357 Henderson SA (1970) The time and place of meiotic crossing-over. Annu Rev Genet 4: 295-324 Henderson SA, Edwards R G (1968) Chiasma frequency and maternal age in mammals. Nature 218:22-28 Hummler E, Theuring F, Hansmarm I (1987) Meiotic nondisjunction in oocytes from aged Djungarian hamsters correlates with an alteration in meiosis rate but not in univalent formation. Hum Genet 76: 357-364 Jagiello G, Fang JS (1979) Analyses of diplotene chiasma frequencies in mouse oocytes and spermatocytes in relation to ageing and sexual dimorphism. Cytogenet Cell Genet 23:53-60 John B (1990) Meiosis. Cambridge University Press, Cambridge, pp 70-73 Luthardt FW, Palmer CG, Yu P-L (1973) Chiasma and univalent frequencies in aging female mice. Cytogenet Cell Genet 12: 68-79
Peters H, McNatty KP (1980) The ovary. A correlation of structure and function in mammals. Granada, London, p 15 Polani PE (1981) Chiasmata, Down syndrome and nondisjunction: an overview. In: Cruz F de la, Gerald PS (eds) Trisomy 21 (Down syndrome) research perspectives. University Park Press, Balitmore, pp 111-130 Polani PE, Jagiello GM (1976) Chiasmata, meiotic univalents and age in relation to aneuploid imbalance in mice. Cytogenet Cell Genet 16: 505-529 Polani PE, Crolla JA, Seller MJ, Moir F (1979) Meiotic crossing over exchange in the female mouse visualised by BUdR substitution. Nature 278: 348-349 Polani PE, Crolla JA, Seller MJ (1981) An experimental approach to female mammalian meiosis: differential chromosome labelling and an analysis of chiasmata in the female mouse. In: Jagielio G, Vogel HJ (eds) Bioregulators of reproduction. (P & S biomedical sciences symposia series) Academic Press, New York, pp 59-97 Perroni L, Bricarelli D, Grasso M, Piefluigi M, Baldi M, Pedemonte C, Strigini P (1990) Crossing over and chromosome 21 nondisjunction: a study of 60 families. Am J Med Genet 7[Suppl] : 141-147 Reid DH, Parsons PA (1963) Sex of parent and variation of recombination with age in the mouse. Heredity 18:107-108 Sherman SL, Takaesu N, Freeman S, Phillips C, Blackston RD, Kanta B J, Jacobs PA, Cockwell AE, Kurnitt D, Uchida I, Hassold TJ (1990) Tfisomy 21: association between reduced recombination and non-disjunction. Am J Hum Genet 47: A97 Slizynski BM (1960) Sexual dimorphism in mouse gametogenesis. Genet Res 1 : 477-486 Speed RM (1977) The effects of ageing on the meiotic chromosomes of male and female mice. Chromosoma 64: 241-254 Speed RM, Chandley AC (1983) Meiosis in the fetal mouse ovary. II. Oocyte development and age related aneuploidy. Does a production line exist? Chromosoma 88:184-189 Sugawara S, Mikamo K (1983) Absence of correlation between univalent formation and meiotic nondisjunction in aged female Chinese hamsters. Cytogenet Cell Genet 35: 34-40 Sugawara S, Mikamo K (1986) Maternal ageing and nondisjuncfion: a comparative study of two chromosomal techniques on the formation of univalents in first meiotic metaphase oocytes of the mouse. Chromosoma 93 : 321-325 Takaesu N, Jacobs PA, Cockwell A, Blackston RD, Freeman S, Nuccio J, Kurnit DM, Uchida I, Freeman V, Hassold T (1990) Nondisjunction of chromosome 21. Am J Med Genet 7[Suppl] : 175-181 Tanzi RE, Haines JL, Gusella JF (1990) Detailed genetic linkage map of human chromosome 21: patterns of recombination according to age and sex. Prog Clin Biol Res 360: pp 15-26 Tarkowski AK (1966) An airdrying method for chromosome preparations for mouse eggs. Cytogenetics 5: 394-400 Tease C, Fisher G (1986) Further examination of the productionline hypothesis in mouse foetal oocytes. I. Inversion heterozygotes. Chromosoma 93: 447-452 Tease C, Fisher G (1989) Further examination of the productionline hypothesis in mouse foetal oocytes. II. T(14;15)6Ca heterozygotes. Chromosoma 97: 315-320 Wallace ME, MacSwiney FJ, Edwards RG (1976) Parental age and recombination frequency in the house mouse. Genet Res 28: 241-251 Warren AC, Chakravarti A, Wong C, Slaugenhaupt SA, Halloran SL, Watkins PC, Metaxotou C, Antonarakis SE (1987) Evidence for reduced recombination on the nondisjoined chromosomes 21 in Down syndrome. Science 237: 652-654
