Parental origin effects in mice

Nuclear transplantation experiments in mice, reviewed elsewhere in this Symposium, have clearly demonstrated that the maternal and paternal genomes from which the embryo is formed are not functionally equivalent. The paternal genome appears to be essential for the normal development of extraembryonic tissues and the maternal genome for some stage of embryonic development. These findings provide some explanation for the observations that in mammals diploid parthenotes possessing two maternal genomes fail to survive (Markert, 1982) and that, in man, embryos with two paternal chromosome sets are inviable, forming hydatidiform moles (Kajii & Ohama, 1977). It has been proposed that a specific ‘imprinting’ of the paternal genomes occurs during gametogenesis so that the presence of both a female and male pronculeus is essential in an egg for full-term development (Barton, Surani & Norris, 1984; McGrath & Solter, 1984a; Surani, Barton & Norris, 1984).

The term ‘imprinting’ was coined by Crouse (1960) to describe the modification of chromosomes in the germlines of Sciara that causes the maternal and paternal autosomes and X-chromosomes to behave differently in the early cleavage stages of the embryo and at meiosis in the adult. Selective chromosome elimination was the phenomenon observed. The term has also been used in connection with the heterochromatic behaviour and genetic inactivation of the paternal chromosome set in male mealy bugs (Brown & Nelson-Rees, 1961) and, more recently, it has been applied to a process that causes the paternal X to be preferentially inactivated and heterochromatic in mouse extraembryonic membranes (Takagi, 1983). Imprinting can therefore be regarded as a chromosome phenomenon. Indeed, genetic studies in the mouse using various chromosome rearrangements have shown that it cannot be the whole genome that is imprinted because parental origin seems unimportant for many chromosomes. With others, however, a zygotic lethality, known as noncomplementation lethality, results when two members of a homologous pair of chromosomes or chromosome regions derive from only one parent. With yet others, anomalous opposite phenotypes result and this suggests a differential functioning of the chromosomes or regions involved according to parental origin. This paper summarizes the genetic data on these parental origin effects with special reference to those causing anomalous phenotypes.

Chromosomally balanced mice carrying a pair of homologous chromosomes derived from only one parent, i.e. maternal or paternal disomy (with corresponding nullisomy) can readily be produced with the use of Robertsonian translocations that cause high frequencies of nondisjunction when heterozygous. Thus, intercrosses between Robertsonian heterozygotes allow a chromosomal gain from one parent to be complemented by a corresponding chromosome loss from the other, and, when the parents differ for alleles at a marker gene locus, those young that are monoparental for the chromosome concerned can readily be distinguished by marker gene phenotype (Tease & Cattanach, 1986). Further detail is given in Fig. 1. The frequency of maternal and paternal disomic mice produced in this way is dependent upon the nondisjunction levels associated with the various Robertsonian translocations but, typically, is about 2%. Higher frequencies can be obtained by intercrossing animals heterozygous for two different Robertsonian translocations having one chromosome arm in common (monobrachial homology) since this combination forces yet higher levels of nondisjunction (White, Tjio, van der Water & Crandall, 1972; Gropp & Winking, 1981).

Intercrosses between heterozygotes for reciprocal translocations can similarly generate chromosomally balanced young which have two copies of a region of chromosome either distal or proximal to the translocation breakpoint from one parent, i.e. maternal or paternal duplication (with corresponding deficiency) (Snell, 1946; Searle & Beechey, 1985). Further detail is given in Fig. 2. The frequency of distal duplication/deficiency is high, about 17 %; that for proximal duplication/deficiency is much lower, less than 5 % (Searle & Beechey, 1978).

Studies with Robertsonian translocations have demonstrated that mice that are monoparental for either maternal or paternal chromosomes 1,3,4,5,9,13,14 and 15 (denoted maternal and paternal disomy, respectively) are viable and normal. Paternal disomy 6 is also normal. Maternal or paternal disomy for chromosomes 2, 7, 10, 12, 16, 17 and 18 have not been investigated. Crosses to generate disomy 19 have been carried out (Lyon, Ward & Simpson, 1976) but because the Robertsonian translocation employed gives only very low frequencies of nondisjunction, the failure to detect any disomy 19 progeny may not be significant. Data consistent with the viability of some monoparentals have been obtained from work with reciprocal translocations. In addition, such studies have shown normal complementation occurs for certain regions of chromosomes 7,10 and 18 (Searle, 1985).

The evidence on noncomplementation lethality has recently been reviewed (Searle & Beechey, 1985). Only a brief summary of the facts will therefore be given here. The phenomenon has been reported for four different chromosomes, 2,6,7 and 8, and a possibly related finding has been described for chromosome 17.

Searle & Beechey (1985) have provided data from studies with two reciprocal translocations, T(2;9)11H and T(2;8)26H, that indicate that maternal duplication/paternal deficiency for a distal region of chromosome 2 (distal 2) causes perinatal death. With T11H the reciprocal type was indicated to be viable, but with T26H, which involves a smaller distal region, perinatal death was found. The latter was attributed to noncomplementation lethality associated with the chromosome 8 constitution (maternal duplication/paternal deficiency for distal 8). But more recent data, to be described shortly, have shown that perinatal lethality and anomalous phenotypes are typically associated with distal 2 genotypes of monoparental origin, these deriving from a number of different reciprocal translocations involving chromosome 2.

In these crosses and in studies using another translocation, T(2;4)lSn, maternal deficiency/paternal duplication for a proximal region of chromosome 2 (proximal 2) was detected with a frequency (0·7 %) consistent with expectation for adjacent-2 disjunction. No offspring of the reciprocal type were found among a relatively small number of young classified (88) which suggests, but does not establish, a lethality for this class (P = 0·53). More recent work (Cattanach & Kirk, 1985), to be described shortly, has clearly indicated noncomplementation lethality involving the most proximal region of the chromosome.

Studies with the Robertsonian translocation, Rb(4.6)2Bnr (Tease & Cattanach, 1986), have shown that whereas paternal disomy 6 is viable and normal (10 marked young in 1113 progeny scored), maternal disomy 6 is lethal (0 marked young in 948 progeny; P = 0·0002). This has been confirmed in crosses between mice carrying two Robertsonian translocations having monobrachial homology for chromosome 6 (Beechey & Searle, 1985); among 130 offspring there were 8 cases of paternal disomy 6 but none of maternal disomy 6 (P = 0·0003).

Evidence of noncomplementation lethality involving this chromosome was first suggested in the early work of Snell (1946) on a translocation now described as T(7;?)2Sn and has recently been verified and established in more detail by Searle (1985) using T(7;19)9H and T(7;18)50H. Two different noncomplementation lethalities have been identified. One causes a lethality for maternal duplication/paternal deficiency for a region between the T2Sn and T50H breakpoints (0 marked progeny among 31 classified; P = 0·004); the other causes a lethality of maternal deficiency/paternal duplication for a region distal to T50H (by deduction from other data).

The T(2;8)26H data described for chromosome 2 have given some indication that maternal duplication/paternal deficiency for distal 8 may be lethal and this is supported by limited data with T(8;17)17H (Searle & Beechey, 1985); no marked young representing this class were found among 30 progeny classified. This departs significantly from the 17 % expectation (P = 0·004) but in view of the occurrence of marked young in the T26H crosses, albeit as perinatal lethals, and with a chromosome 2 explanation for this lethality, further data would seem necessary to establish that noncomplementation lethality occurs with chromosome 8.

Evidence of incomplete complementation involving this chromosome has been provided by the experiments of Lyon & Glenister (1977) with the translocation T(9;17)138Ca. Maternal deficiency/paternal duplication for a region of the chromosome proximal to the translocation breakpoint was found in a number of crosses to give a lower frequency of successful complementation than its reciprocal. It was suspected that this phenomenon was related to one described by Johnston (1974, 1975) involving the hairpin-tail mutation, T^hp, which is actually a small deficiency located proximally in chromosome 17. Johnston found that T^hp was lethal when inherited from the mother but viable when inherited from the father; and both genetic studies (Winking, 1981) and nuclear transplantation experiments (McGrath & Solter, 1984b) have since proven this to be a nuclear rather than a cytoplasmic or maternal effect. Deficiency mapping has allowed the specific chromosome region responsible, denoted T maternal effect (Tme), to be more precisely identified (Winking & Silver, 1984).

Anomalous phenotypes have recently been described in chromosomally balanced mice that are monoparental for certain chromosome regions. The chromosomes involved are 2 and 11; the chromosome 11 effect will be described first.

Intercrosses between heterozygotes for the Robertsonian translocation, Rb(11.13)4Bnr (Fig. 1) regularly produce maternal and paternal disomy 11 and 13 young. The latter have been found to be normal, irrespective of parental origin, but both maternal and paternal disomy 11 animals invariably have been found to differ in size from their normal sibs; significantly, maternal disomics were consistently smaller than their litter mates and paternal disomics were consistently larger (Fig. 3). The anomalous size effects were thus opposites or reciprocals of each other.

The size differences were first recognized at birth with the use of the chromosome 11 marker gene, vestigial tail, but were subsequently found with another marker (wavy coat) and also without any marker simply by size alone (Table 1).

The genotypes were confirmed in all cases by test matings. Postnatal growth rates of the small maternal and large paternal disomics appeared similar to that of their normal sibs with the consequence that the size differences remained evident through to adulthood (Fig. 4). The two types of disomics were produced in near-equal frequencies (Table 1) and among an overall total of 87, males and females were about equally represented (small, 19 ♀24 ♂large 26 ♀ 18 ♂).

Both disomics were proportionately normal externally and organ weights were in proportion to body weights. Plasma bioassay able somatomedin activities have recently been checked and found to be similar to those of their normal sibs (Dr K. Ashton, personal communication). Neither type of disomy has exhibited viability problems and both sexes have proven fertile. The size differences did not persist into the next generation.

Intercrosses between heterozygotes for the reciprocal translocation T(2;11)30H have shown that only the proximal region of chromosome 11 is associated with the size phenomenon (Fig. 2). Both products of adjacent-1 disjunction were of normal size at birth, but maternal duplication/paternal deficiency for the region of chromosome 11 proximal to the translocation breakpoint (deriving from adjacent-2 disjunction) resulted in mice which, like those with maternal disomy 11, were small at birth and at later ages. Eight such animals have been detected among 714 progeny (1·12%) either with the use of the proximal 11 marker, waved-2, or by size alone and their genotypes verified by test mating (Fig. 2). By contrast, no large marked or unmarked young representing maternal deficiency/paternal duplication for proximal 11 were found in these matings (P = 0·0003). Their absence may be attributed to their chromosome 2 genotype (maternal duplication/paternal deficiency for proximal 2) and, as such, provides the evidence for the proximal 2 noncomplementation lethality mentioned earlier. Limited data from intercrosses involving T30H and the proximal chromosome 2 marker gene pallid, which affects eye colour, have suggested that the lethality may occur early in embryonic development (0 marked young among 204 classified prenatally as compared with 8 of the reciprocal type detected among 714 young classified postnatally; P = 0·100). Further evidence is clearly necessary to establish this point, however.

The T(2;ll)30H intercrosses used to investigate the chromosome 11 size phenomenon incidentally provided the second example of anomalous phenotypes associated with chromosome parental origin. Although both maternal duplication/paternal deficency for distal 11 (i.e. maternal deficiency/paternal duplication for distal 2) and its reciprocal were normal in size they proved to be postnatal lethals. This, in part, accords with the noncomplementation lethality previously reported for this region of chromosome 2 (Searle & Beechey, 1985), but the novel discovery was the occurrence of characteristic abnormal phenotypes (Fig. 5). Maternal duplication/paternal deficiency for distal 2 resulted in flat-.sided, arch-backed, hypokinetic newborn that, failing to suckle effectively, usually died within 24 h. And, remarkably, the reciprocal genetic type showed a clear opposite phenotype; these mice had short, square bodies and broad, flat backs and were notably hyperkinetic. These often survived for several days but failed to thrive, their behaviour becoming increasingly extreme with tremor and balance defects developing. Viability was found to be dependent upon genetic background (Table 2).

Almost identical though sometimes less obvious phenotypes have recently been generated by intercrossing other translocations involving chromosome 2 (Table 2). With T(2;4)1Sn the hyperkinetic maternal deficiency/paternal duplication type appeared more viable than those deriving from the other translocations; thus the behaviour was less evident and some animals survived for up to 12 days. The variability in survival times of this class no doubt accounts for its reported viability in the earlier noncomplementation studies with T(2;8)26H; in fact, reanalysis of the T26H breeding records has shown that the maternal deficiency/paternal duplication young were sometimes described as fat at birth and liable to loss. Survivors were often described as runted (Searle, personal communication).

Intercrosses between heterozygotes for T(2;16)28H whose chromosome 2 breakpoint lies distal to that of T30H failed to produce mice of the two anomalous phenotypes (Table 2). In view of the proven viability of at least one of the two complementation types produced in T(8;16)17H intercrosses (see section on chromosome 8) it seems unlikely that this is a consequence of noncomplementation lethality involving chromosome 16. It is more probable that the chromosome 2 ‘anomaly factor’ is located between the T30H and T28H breakpoints on this chromosome (Fig. 6). Likewise, the appearance of both types of anomalous young from intercrosses of the three translocations with more proximally located breakpoints (Table 2) suggests that the proximal ‘noncomplementation factor’ must lie close to the centromere (Fig. 6).

Each of the parental origin effects described suggests that maternal and paternal copies of certain chromosome regions in the mouse may function differently during embryonic development.

The T^hp (or Tme) phenomenon (Johnston, 1974,1975; Winking & Silver, 1974) indicates that a maternal copy of a region of chromosome 17 must be present for normal foetal development, and this finds a degree of support from the genetic studies with T138Ca (Lyon & Glenister, 1977).

The noncomplementation lethality for one region of chromosome 7 indicates much the same thing; a maternal copy may be required for embryonic survival but, alternatively, it could be concluded that two paternal copies cause the lethality.

For another region of chromosome 7 and also for regions of chromosomes 6 and possibly 8 the reverse appears to be true; either a paternal copy is required or two maternal copies cause the lethality. Whether the cause of the lethality is the absence of one parental copy or the duplication of the other it seems likely that parental origin effects can involve either maternal or paternal chromosomes with the consequence that they are functionally inequivalent.

A differential functioning of maternal and paternal chromosome regions is strongly indicated by the anomalous phenotypes discovered in the genetic studies with chromosomes 2 and 11. With both chromosomes the presence of two maternal copies of the regions involved caused opposite or reciprocal phenotypes to those caused by two paternal copies (hypokinetic and hyperkinetic, small and large) and this suggests a shortage versus excess of genetic activity. This could mean either single or earlier activity of regions of paternal chromosomes 2 and 11, and inactivity or later activity of the corresponding maternal chromosome regions. Clearly, the same phenomenon could explain the T^hp and noncomplementation lethality effects; excess or earlier activity of one parental component may allow normal embryonic development, but subnormal or late activity of the other might well lead to embryonic death. Precedence for activity of a single chromosome or chromosome region is well established through the behaviour of the paternal X in mouse extraembryonic membranes (Tagaki, 1983); precedence for earlier/later activity is available at the gene locus level in diverse species and interfamily hybrids (Ohno, 1969; Klose & Wolf, 1970; Whitt, Cho & Childers, 1972; Whitt, Childers & Cho, 1973; Schmidtke, Kuhl & Engel, 1976; Krietsch et al. 1982).

At present it is not clear whether the parental origin effects really involve whole chromosome regions, and thus many genes, or perhaps only single loci within the regions crudely defined by the genetic studies. Multiple gene action cannot be excluded but deficiency mapping has located the factor responsible for the T^hp phenomenon to within a very small region of chromosome 17 (Winking & Silver, 1984) and the size phenomenon associated with chromosome 11 most probably requires the action of only a single locus. With regard to the latter it may be no coincidence that a growth hormone locus (dwarf) is located within the region of chromosome specifically involved (Cattanach & Kirk, 1985) and, although the dwarf mutant is of normal size at birth despite a severe deficiency of pituitary hormones (Bartke, 1965), the homologous chromosome region of man (Buckle et al. 1984) carries a cluster of growth hormone loci, at least one of which acts in the placenta. A hormonal control of foetal growth mediated through the placenta has therefore been suggested; and it is of interest in this regard that there is overdevelopment of the trophoblast and extraembryonic membranes in embryos with two paternal genomes and underdevelopment in those with two maternal genomes (Surani et al. 1984). Experiments to investigate growth hormone production by the placenta in normal, maternal and paternal disomic mice using a direct assay and with a probe for prolactin are being initiated. In the case of the anomalous mice associated with chromosome 2, the behavioural problems suggest neurological effects and hence the adenosine deaminase deficiency of the wasted mutant whose locus lies in the relevant region of the chromosome has been suspected (Cattanach & Kirk, 1985). However, blood and liver ADA levels have proved to be very similar in normal and both types of abnormal mice. Some other locus/loci must be responsible for the effects observed.

Single or preferential action of regulator loci, or of receptor loci, to give the latter different affinities for regulatory substances, may underlie the parental origin effects described. Whatever the mechanism, the phenomena provide explanations for the lethality manifested by parthenogenic and androgenic embryos, for not only is there the possible connection between the size effects and trophoblast and extraembryonic membrane development discussed above, but the noncomplementation lethalities attributable to proximal 2, central and distal 7, chromosome 6 and possibly distal 8 could be largely or entirely responsible. It remains to be established that these lethalities are expressed in the early stages of embryogenesis.

The parental origin effects may clearly be attributed to the imprinting of the chromosomal regions involved during gametogenesis in the parents. However, as imprinting processes, perhaps only concerning single loci, they differ considerably from other examples of imprinting that have involved whole chromosomes and that have been variously expressed as chromosome loss, heterochromatic behaviour, late DNA replication, as well as genetic inactivation (Lyon & Rastan, 1984; Tagaki, 1983).

Chromosomal imprinting may give rise to at least two other parental origin phenomena. For chromosome regions in which parental duplication (as opposed to deficiency) causes abnormality, as perhaps with chromosome 2 and 11, trisomies of appropriate parental origin should show these abnormalities in addition to others caused by chromosome imbalance. Moreover, as suggested by Searle & Beechey (1985), human trisomies in which the duplicated chromosomes have homologies with mouse chromosomes showing these parental origin effects should show equivalent abnormalities. Likewise, triploids might be expected to show abnormalities attributable to parental duplications, though again this would be dependent upon the parental origin of the extra chromosome set. In this regard it is noteworthy that all 54 paternally derived triploids studied by Jacobs et al. (1982) in which pathological diagnosis could be made were partial hydatidiform moles whereas only 3 out of 15 maternally derived triploids were molar. Significantly, all 3 of the maternally derived molar triploids resulted from a failure of the first meiotic division. This might suggest that maternal imprinting takes place at the second division.

A second phenomenon that might result from imprinting is a differential expression of semidominant genes dependent upon parental origin. One such example in the mouse might be fused (Fu). It has been observed that the offspring of Fu+ or Fu/Fu mothers tend to be less affected than those of + 4-mothers and it might be significant that the gene lies very close to the Tme region responsible for the T^hp phenomenon. Curiously, the direction of the effect is not the same. In the case of Tme, a normal maternal region is required but, with Fu, the paternal allele appears to be predominantly expressed. Huntington’s chorea might represent an example of the phenomenon in man. Thus the juvenile form of the disorder seems most often to have been inherited from the father rather than from the mother; it will be interesting to see which mouse chromosome carries the gene.

The parental origin effects described indicate that maternally and paternally derived regions of certain chromosomes function differently during embryogenesis and this is clearly of fundamental biological significance. Further studies in the mouse are needed to specify which other chromosomes show these effects and to define the regions involved. And, given that the resulting abnormalities can be recognized prenatally, the effects in trisomies of maternal and paternal origin could readily be sought. The differential expression of semidominant genes according to parental origin may or may not represent a further form of the phenomenon. However, the fundamental problem to be resolved is the biochemistry of the imprinting process.