1.6. Objetivos de la investigación
2.2.3. Dialogía: pluralidad de voces y discursos
The results from Experiment 1 show that X^O fetuses are equivalent in size to ‘XX’ sibs (or perhaps larger) and almost indistinguishable from ‘XY’ sibs at 10.5 àpc. A previous study of the weights of XO fetuses demonstrated that XO fetuses with a paternal X-chromosome were significantly smaller than XX sibs. Considering these two results, it appears that the developmental progression of XO fetuses with a maternally inherited X-chromosome is not equivalent to X^O fetuses.
To rule out genetic background effects it is necessary to produce both X^O and X^O individuals within the same litters for direct comparison (Experiment 2). It is clear from this experiment that a single X-chromosome causes developmental retardation if it is paternal in origin (confirming the X^O effect described by Burgoyne et al., 19836) but not if it is maternal in origin. The difference in fetal weights between X^O and X^O sibs at a timepoint midway through gestation (10.5 àpc), is emphatic in both crosses 2a and 26; the X^O fetuses being nearly twice the size of the X^O fetuses.
At what time does the imprint affect development?
In the light of the present results, the developmental retardation originally described in the X^O offspring borne of XX mothers (Burgoyne et al, 19836) is undoubtedly due to an imprint on the X-chromosome rather than a pseudoautosomal gene dosage effect. In that study, the reduced size of X^O fetuses was already evident by 7.25 àpc. Thus, a paternally imprinted X has a retarding effect on development at some point prior to 7.25 dpc. Preliminary data suggest that cell number in X^O blastocysts is not significantly lower than that in XX blastocysts (P.S.Burgoyne, pers. comm ), but a more rigorous test for an imprinting effect at 3.5 àpc would be to compare cell numbers in and X^O blastocysts. However, the lack of an imprinting effect at this time can be indirectly inferred from other data. The Y™ chromosome is unusual in that it confers no developmental advantage in the preimplantation period - XX and XY™ blastocysts having equivalent cell numbers (Burgoyne, 1993). This absence of a Y-effect is reflected in the equivalent size of X^O and XY™ fetuses at 10.5 dpc (Burgoyne et al, 1995), and indirectly implies that X^O and XX embryos (like XX and X^O) would show no cell number difference at 3.5dpc.
The X-chromosomal imprint, which results in a developmental advantage for fetuses with a single maternal X-chromosome at 10.5 6pc, must therefore take effect at some time between the blastocyst (3.5 dpc) and the egg cylinder stage (7.25 6pc) - the earliest stage at which X^0-XX differences have been measured (Burgoyne et al., 19836).
Mechanism by which the X-chromosomal imprint operates
A paternally derived X-chromosome may be less efficient than a maternally derived X- chromosome in supporting early development, and this could most simply be explained by the paternal imprint reducing X-linked gene expression. In fact, reduced levels of enzyme encoded by the paternal allele of X-linked Hprt have been observed in preimplantation mouse embryos (Moore and Whittingham, 1992). Variation in the strength of the imprint could perhaps explain the variability of the X^O weight data in the present study.
A possible explanation for reduced X^ expression comes from studies of Xist, the gene involved in X-inactivation. Paternal Xist expression has been observed at preimplantation stages, as early as the 4-cell embryo, regardless of the presence or absence of another X- chromosome (Kay et al, 1994). Presumably, an X^O embryo must have an active X- chromosome to ensure survival, since the X-chromosome is known to contain many important housekeeping genes and furthermore, YO embryos do not develop beyond the 8- cell stage (P.S.Burgoyne, pers. comm ). Nevertheless, it is possible that in X^O embryos, as a result of early paternal Xist expression, some X-linked genes could either be inactivated for a short period of time or, alternatively, be delayed in their activation, at a time when the embryo has used up the majority of maternally-derived resources and embryonic genome activation begins. The net effect of this reduced X-chromosome activity might be an overall developmental delay, rather than lethality.
If this hypothesis is correct, one might expect a preimplantation X^0-XX difference. While this is not supported by preliminary data collected at 3.5 dpc (P.S.Burgoyne, pers. comm ), there could be some time lag before the reduced X-activity is manifest as a developmental delay. Furthermore, the maternal X-chromosome originates in the oocyte, the DNA of which
difiference could cause the maternal X to become transcriptionally active earlier or perhaps transcribe at a higher level than the paternal X.
Clearly, in cells containing a single X-chromosome, complete inactivation of that chromosome would be lethal. However, certain tissues may be more tolerant than others of the effects of X-dosage deficiency. The paternal X-chromosome is known to be preferentially inactivated in the trophectoderm and primitive endoderm just prior to implantation at around 4.5 dpc (Takagi and Sasaki, 1975; West et al, 1977; Harper et al, 1982). While it has been shown that the single paternal X is active in extraembryonic membranes, even in those tissues where X^ is normally almost exclusively expressed (Frels and Chapman, 1980; Papaioannou and West, 1981), it is possible that ‘weaker’ X-chromosome activity is present in specific extraembryonic tissues with concomitant developmental retardation of the embryo proper.
Intiiguingly, no X^O-XX difference was observed when offspring were obtained fi"om XO mothers on a Jcl/ICR genetic background (Omoe and Endo, 1993). A number of different developmental parameters (not including fetal weights) were measured at 10 dpc, but a difference was only seen between XX and XY fetuses and this was not statistically significant. In another study, Tada et a/.(1993) found only 12 out of 35 X^O and X^Y embryos which showed growth retardation and concluded that a paternal X-imprint is either weak or absent in mice. These data underline the importance of genetic background as a context for differential gene expression and may provide a means to investigate the imprinting effect described above in more detail.
Are other developmental consequences o f X-monosomy in mice affected by the parental origin o f the X-chromosome?
males
Handel and Hunt (1991) assessed the effects of inheriting a single X-chromosome fi'om an inappropriate parental source by producing X^Y^ mice. These males were fertile with no obvious phenotypic abnormalities. However, prenatal development progression was not monitored and the breeding scheme used did not allow the accurate prediction of X^Y^ genotype frequency (hence prenatal viability) amongst the offspring. Furthermore, one would
predict that the preimplantation advantage conferred by the Y-chromosome (Burgoyne, 1993) would counteract, in part, any developmental retardation resulting from the single paternal X-chromosome.
Selection against fetuses?
It is well established that XO mice produce fewer XO ofrspring than expected, and there has been debate as to whether this is solely due to a deficiency of ‘O’ eggs or whether there is also preferential XO loss during pregnancy (Kaufinan, 1972; Luthardt, 1976; Brook, 1983). Hunt (1991) studied the breeding of XY*^ females (on a C57BL/6 inbred background) which produced markedly fewer XY*^ offspring than expected due to preferential loss during pregnancy. In both examples cited above, the fetuses carry a single paternally derived X- chromosome and are borne by X-monosomic mothers. Hunt (1991) therefore suggested that a fetal X^ effect, perhaps interacting with some consequence of the maternal X-monosomy, may be the underlying cause of XY*^ and XO fetal loss. The finding that X^O, but not X^O fetuses, are retarded in early development, provides some support for this suggestion. However, it is clear that a single paternal X at fetal stages is not a sufficient cause in itself.
Previously collected data concerning X^O frequency throughout pregnancy and at birth, for outbred mothers heterozygous for the X-inversion - In(X)lH (Evans and Phillips, 1975), showed that there was no detectable selection against the X^O fetuses (preimplantation XO XX ratio = 1:3.15, XO XX ratio at birth = 1:3.10 - P.S.Burgoyne and E.P.Evans, pers. comm ). Furthermore, XY*^ females have been produced on a random bred background, and these females show no marked deficiency of XY*^ offspring (B.Peitz and P.S.Burgoyne, pers. comm ). However, as part of a study comparing X^O with XX fetuses at 10.5 6pc, Burgoyne et a/.(19830) described a sub-population of severely retarded X^O fetuses, termed runts. One could speculate that in a less favourable maternal environment, these XO runts might be eliminated during pregnancy.
In summary, in the mouse it seems likely that X^ monosomy is a factor which puts fetuses ‘at risk’ (Mahadevaiah et al, 1993) by causing developmental retardation, but only if the mother
Oocyte loss and a single paternal X
XO mice are fertile but have a reduced reproductive lifespan (Lyon and Hawker, 1973; Deckers et al, 1981) which is due to a deficiency of oocytes; XO mice having only approximately 40 per cent the normal numbers of oocytes. This deficiency is a consequence of excess perinatal oocyte loss (Burgoyne and Baker, 1981a) due to selective elimination of cells with an unpaired X-chromosome at meiosis - a form o f‘meiotic quality control’ (Miklos,
1974; Burgoyne and Mahadevaiah, 1993). It is a formal, but unlikely, possibility that the parental origin of the X-chromosome could influence such a quality control mechanism, but this has not been tested.
The XO mouse - a modelfor Turners syndrome?
In many ways, the XO mouse is an inadequate model for Turner syndrome. Clearly the effects of X-monosomy in humans are much more severe than in mice. XO mice are viable, fertile and have no obvious somatic defects while, of the very few human XO conceptions (approximately 0.3% - Jacobs, 1990) that develop to term, most are affected by a wide range of somatic malformations (Hook and Warburton, 1983) and are generally sterile. It is currently thought that Turner syndrome is a consequence of haploinsufBciency for specific genes shared on the X and Y-chromosomes, rather than monosomy per se. While haploinsufBciency, for some or all of these genes, may explain the main features of the complex Turner phenotype (observed in both X^O and X^O individuals) more subtle effects of X-chromosomal imprinting cannot be excluded.
Evidence for imprinting effects in human X-monosomy
If an X-chromosomal imprint is operating during early development in humans, in a similar fashion to that described above in the mouse, one might expect X^O individuals to be more severely developmentally compromised than X^O individuals and thus be over-represented among 45,X abortuses. However, a comparison of the X^O X^O ratio for live births and abortuses suggests that this is not the case. The ratio of ^ 0 :X ^ 0 live births is approximately 70:30 as determined by Xg blood groups (Sanger et al, 1977) and, more recently, RFLP (Restriction Fragment Length Polymorphism) analysis (Ross et a l, 1990; Mathur et al.,
1991; Lorda-Sanchez et al, 1992). RFLPs have been particularly usefiil in determining the parental origin of the X-chromosome in spontaneously aborted 45,X fetuses (most of which
occur in the first trimester). The X^O X^O ratio for these abortuses approximates to 75:25 (Hassold et al, 1988; Jacobs et al, 1990), which is very similar to the livebom ratio. Similarly, the X^O X^O ratio for the small proportion of X-monosomics lost later in pregnancy (in the second and third trimester) is comparable to the livebom ratio (Cockwell et al, 1990). The presence of the maternal X-chromosome apparently does nothing to rescue the XO fetus fi'om early fetal death.
It is possible that the presence of a paternal imprint on the X-chromosome in Turner syndrome might contribute to specific somatic defects rather than to overall mortality, and this might contribute to the variable phenotype commonly seen. Furthermore, a single gene or subset of genes within an imprinted region of the X-chromosome could give rise to particular features of the Turner phenotype, depending on the parental origin of the X- chromosome inherited.
A possible indication of an X-imprinting effect in the human, which results in a differential phenotype for X^O and X^O fetuses comes firom hospital pathology reports in which X^O abortuses were found to be associated with small gestational sacs - a feature not recorded for any of the X^O abortuses examined (Hassold et al, 1988) - although the significance of these differences is not clear. In addition, data pooled fi'om a number of studies investigating livebom 45,X individuals suggests a correlation between the inheritance of a single matemal X-chromosome and cardiovascular abnormalities, and to a lesser extent neck webbing and pretreatment height (Ross et al, 1990; Lorda-Sanchez et al, 1992; Mathur et a l, 1991; Chu et al, 1994). This possible imprinting effect in the human is at odds with the expectation fi'om the mouse data that X^O individuals should be more severely affected developmentally thanX^Os.
In summary, the evidence supporting a role for imprinting in Tumer syndrome is scarce. More detailed clinical investigations combined with re-karyotyping of patients, molecular analysis of parental origin of the X-chromosome and sensitive screening for mosaicism in both livebom, and particularly fetal 45,X individuals are required to resolve this issue.
)^0-X Y, )^0-XXandXX-XYdifferences
There is now a substantial body of data showing that XY embryos are developmentally more advanced than XX embryos in the preimplantation period (Tsunoda et al, 1985; Burgoyne, 1993; Valdivia et al, 1993) due to an accelerating effect of the Y-chromosome (Burgoyne, 1993). In view of this, it was anticipated that XY embryos should also have a developmental advantage over X^O embryos which may be manifest as a size advantage at 10.5 6pc.
In Experiment 1, ‘XY’ (pooled XY* and XY*^) embryos were, on average, slightly larger than X^O embryos but this difference did not approach significance. This could be partially explained by the fact that approximately half the ‘XY’ fetuses carried no Y-specific DNA. Moreover, both XY* and XY*^ genotypes show some weight deficit postnatally, which may be indicative of a partial trisomy effect associated with the presence of the ‘foreign’ Y* centromere (P.S.Burgoyne, pers. comm ). If this ‘trisomy’ effect is already present at 10.5 àpc, this would mask the X^O - XY’ difference.
In Experiment 2a (Table 2.2a) the X^O-XY difference borders on significance using a one tailed test (0.1>/^0.05). It was therefore felt that the possibility of a ‘carry-over’ of the preimplantation Y-effect should be further investigated. Chapter 5 addresses the question as to whether the Y^^" present in the Pq/* stock confers a preimplantation advantage. Chapter 3 uses the cross X^‘^X^‘^x X^‘^Y to generate litters containing the three genotypes XX, X^O and XY to provide further 10.5 àpc data. Since XX mothers produce larger litters than XO mothers and this cross produces three rather than four genotypes, the pairwise comparison of genotypes within litters should be much more sensitive.
One surprising result was the finding that X^O fetuses are significantly larger than XX fetuses (Table 2.2c), suggesting that there is an effect of the difference in X-chromosome constitution on fetal size at 10.5 àpc. This is also fiirther addressed in Chapter 3.