ABSTRACT
Genomic imprinting, where the genes from one parent have different expression properties to those of the other parent, occurs in plants. It has potentially significant consequences because of the importance of hybrids in plant evolution and plant breeding, and provides a mechanism that can hide genetic variation for many generations. The study of nuclear organization shows that chromosome and genome position relates to imprinting in F1 hybrids, with peripheral genomes tending to be expressed preferentially. In some inbred, polyploid hybrids, such as Triticale (a wheat x rye hybrid), treatment with the demethylation agent azacytidine releases hidden variation, which was perhaps lost because of imprinting phenomena.
Occurrence and importance
Genetic phenomena that do not follow Mendelian segregation ratios have been noted - and occasionally published - in hybrid plants over the last 50 years. They have been explained by phrases including ‘block transferance of characters,’ ‘genetic affinity,’ ‘suppression,’ ‘selectivity’ of expression and ‘cryptic structural differentiation.’ Other unusual segregation ratios have been explained using ‘skewed backcross ratios,’ ‘polygenes,’ ‘expression modifiers/modification,’ ‘linkage’ and ‘homoeostasis.’ In some cases, elimination of chromosome sets, leaving the chromosomes of only one parent, may be involved (e.g. Davies, 1958, 1974; Lange, 1971). However, some of the non-Mendelian events can be explained by genomic imprinting - where genes from one parent have different expression properties to those of genes from the other parent, solely as a consequence of their parental origin. In the present work, ‘parents’ will be used to describe both the direct parents of an individual plant, and, for inbred and hybrid plants, the original lines used to make the cross that gave rise to the stock. Even within inbred lines, parental differences may give imprinting, but, in plants, the evidence for this remains limited and largely unpublished.
In crop plants
When breeding crop plants, phenomena leading to genomic imprinting can prevent the production of new plant hybrids that combine desirable characteristics of the parents. Another consequence of imprinting can be the unexpected and undesirable release of variation after several generations of inbreeding, which would lead to non-uniformity in crop stands.
Although the mechanism of imprinting is not certain, it is possible that imprinting might be manipulated. For example, particular sets of genes could be incorporated in a crop, which could them be activated in a subsequent generation. Such a property would be useful both for the induction of new characteristics at particular times (e.g. the period when cereal grains axsre filling) or years (e.g. drought). Alternatively, genomic imprinting could be used to ensure the expression of certain desirable groups of genes in all the progeny of a cross, without the possibility of non-expression because of dominance relationships or supression.
In wide hybrids
In the plant kingdom, sexual interspecific and even intergeneric hybrids can be made relatively easily (see, e.g. Stephens, 1949; Finch and Bennett, 1980). In evolutionary terms, such wide hybridization may lead to genetic introgression in plants, where genes are transferred between different evolutionary lines. The recombination of characters between alien species is also important for enabling the introduction of new genetic variation into the gene pools of crops which may have been restricted by many centuries of inbreeding or intensive selection for homozygosity. Many such hybrids do not resemble intermediates between the parental species, but exhibit a form of genomic imprinting or parental dominance, which is not necessarily gamete specific. The present paper aims to show that such forms of species-specific imprinting occur in plants in both F1 hybrids and inbred lines.
Parental dominance in wide hybrids
In F1 hybrids between cereals
F1 hybrids are of both practical and research interest because they are the source of new combinations of genes. Within a species, two varieties may be intercrossed to produce a new variety combining the best characteristics of both parents. Wide hybrids may be made to transfer alien genes into a crop species or as a preliminary to producing a polyploid plant. F| hybrids also exhibit characteristics referred to as ‘hybrid vigour,’ which is important for increasing yield and yield stability.
Finch and Bennett (1980) made intergeneric hybrids between various barley (Hordeum) and rye (Secale) species. The F1 hybrid H. vulgare×S. africanum does not resemble an intermediate between the two parents but resembles the Secale parent for many different characters (Bennett, 1984; Heslop-Harrison and Bennett, 1984; Fujigaki and Tozu, 1987). Fig. 1 shows photographs of a leaf base of the parental species and a new hybrid between H. vulgare cv. Tuleen 346 (Finch and Bennett, 1982) and 5. africanum, made following the techniques of Finch and Bennett (1980). The ligule and auricle characteristics of the hybrid resemble the 5. africanum parent more than the H. vulgare parent. Fig. 2 shows leaf surface peels from representative individuals of the three plants, again with the hybrid resembling the S. africanum parent more than the H. vulgare in hairiness and cell form. Other characteristics include those of the perenniality (the hybrid and .S’. africanum are perennial, while the barley flowers and dies in one growing season) and cold hardiness (the hybrid and S. africanum are extremely sensitive to cold temperatures). Thus, the resemblance is not only for morphological but some physiological characteristics.
Morphological characteristics of the bases of leaves of (A) Hordeum vulgare (barley). (B) Secale africanum, a wild rye species and (C) the intergeneric F, sexual hybrid H. vulgare×S. africanum. The hybrid resembles the S. africanum parent for characteristics such as the ligule (I) length, auricle claw (au) size and hairiness.
Morphological characteristics of the bases of leaves of (A) Hordeum vulgare (barley). (B) Secale africanum, a wild rye species and (C) the intergeneric F, sexual hybrid H. vulgare×S. africanum. The hybrid resembles the S. africanum parent for characteristics such as the ligule (I) length, auricle claw (au) size and hairiness.
Leaf surface peels from the lower surface of leaves of (A) H. vulgare, (B) the hybrid H. vulgare×S. africanum and (C) S. africanum. As in Fig. 1, the hybrid resembles the 5. africanum parent more than the H. vulgare parent.
In this hybrid, the male parent was S. africanum, and we have not succeeded in making the cross in the reverse direction. The difficulty is probably related to the breeding methods of the plants, since the rye species is outbreeding and produces large amounts of vigourous pollen, in contrast to the largely self-pollinating barley. Nevertheless, although direct evidence is as yet limited, it might be expected that the phenotypic dominance observed is due to genotypic effects rather than a consequence of the parental origin of the gametes.
Correlations of phenotype with chromosome position
The physical positioning of the chromosomes in the hybrid H. vulgare×S. africanum has been investigated by reconstructing metaphases from sets of electron micrographs of serial sections. These results have shown that the chromosomes originating from the two parental genomes are not randomly positioned within the hybrid, but tend to be spatially separated with those from one species lying in a peripheral domain, while those from the other are more central. In the H. vulgare×S. africanum hybrid, the chromosomes of H. vulgare origin tend to be central in the nucleus, while those of S. africanum origin are peripheral (Bennett, 1982, 1984; Finch and Bennett, 1981; Bennett, 1988; Heslop-Harrison and Bennett, 1984). However, at metaphase, the chromosomes are generally inactive in gene expression and DNA replication, so it is necessary to know if the genomes are also spatially separated at interphase, when the chromosomes are active in gene expression and DNA replication.
In hybrids where the DNA from the two parents is sufficiently different, the chromosomes of the two genomes can be distinguished at all stages of the cell cycle by a method using in situ hybridization of total genomic DNA. The basis of the technique is illustrated in Fig. 3, which shows Southern transfers of restriction enzyme digests of DNA from two parents and an intergeneric hybrid. These have been hybridized with labelled genomic DNA from one of the parents, H. chilense, alone (Fig. 3A), or in the presence of a large excess of unlabelled genomic DNA from the other parent, S. africanum, to block sequences that are common between the two genomes so the labelled DNA cannot hybridize (Fig. 3B; see Anamthawat-Jónsson et al. 1990, for further details). The tracks with DNA from the H. chilense parent show stronger hybridization than the S. africanum tracks, and the differentiation is emphasized in the blocked blot where the two parents and hybrid can be distinguised clearly.
Luminographs showing sites of hybridization of labelled total genomic Secale africanum DNA to Southern transfers of size-fractionated restriction enzyme digests of DNA from S. africanum (Sa), H. chilense (Hc) and the F1 hybrid between the two species (×). (A) Hybridized with labelled H. chilense DNA alone; (B) hybridized with labelled genomic H. chilense DNA in the presence of an excess of unlabelled genomic DNA from S. africanum to block non-specific hybridization of labelled probe (see Anamthawat-Jónsson et al. 1990). The probe enables discrimination between the two species and the hybrid, and the specificity of probing is increased by the addition of the blocking DNA. (\ lambda HindIII digest as size marker, top to bottom 23.1, 9.4, 6.6, 4.4, 2.3 and 2.1 kb).
Luminographs showing sites of hybridization of labelled total genomic Secale africanum DNA to Southern transfers of size-fractionated restriction enzyme digests of DNA from S. africanum (Sa), H. chilense (Hc) and the F1 hybrid between the two species (×). (A) Hybridized with labelled H. chilense DNA alone; (B) hybridized with labelled genomic H. chilense DNA in the presence of an excess of unlabelled genomic DNA from S. africanum to block non-specific hybridization of labelled probe (see Anamthawat-Jónsson et al. 1990). The probe enables discrimination between the two species and the hybrid, and the specificity of probing is increased by the addition of the blocking DNA. (\ lambda HindIII digest as size marker, top to bottom 23.1, 9.4, 6.6, 4.4, 2.3 and 2.1 kb).
Fig. 4 shows the results following in situ hybridization using genomic probe and blocking (Heslop-Harrison et al. 1990; Schwarzacher et al. 1989) in the F1 hybrid H. vulgare×S. africanum. The micrographs show that the two genomes, originating from different parents, tend to be spatially separated both at prophase and during interphase, as well as at metaphase. However, these data are from spread material, where it is possible that differential penetration of probe or detection reagents could lead to the observed results. Fig. 5 shows a single section through nuclei of the hybrid, which detnonstrates that the genome separation is also seen in sectioned material where the three-dimensional structure of the cell has been preserved (Leitch et al. 1990). Reconstructions now being made from such interphase nuclei tend to confirm the impression of the spatial separation of parental chromosome sets at interphase in the wide hybrids (Leitch, Schwarzacher and Heslop-Harrison. unpublished data).
In situ hybridization to chromosome spreads of the hybrid H. vulgare×S. afncanum using the genomic probing with blocking method (Schwarzacher et al. 1989; Heslop-Harrison et al. 1990). (A.B) A telophase nucleus where the chromosomes are not fully condensed and the major axes can be followed. (C.D) An interphase nucleus. (A,C) After staining with the DNA-specific dye DAP1 which fluoresces blue when excited with ultra-violet light. It shows all the chromatin in the nucleus. (B,D) Following in situ hybridization with labelled S. africanutn DNA in the presence of unlabelled H. vulgare DNA. The sites of hybridization of the labelled DNA were detected by blue light excited yellow fluorescence, while unlabelled chromatin fluoresces orange with the propidium iodide counterstain. The micrographs show that the two genomes, originating from different parents, tend not to be intermixed both at telophase and interphase, and the DNA originating from S. afncanum tends to be peripheral.
In situ hybridization to chromosome spreads of the hybrid H. vulgare×S. afncanum using the genomic probing with blocking method (Schwarzacher et al. 1989; Heslop-Harrison et al. 1990). (A.B) A telophase nucleus where the chromosomes are not fully condensed and the major axes can be followed. (C.D) An interphase nucleus. (A,C) After staining with the DNA-specific dye DAP1 which fluoresces blue when excited with ultra-violet light. It shows all the chromatin in the nucleus. (B,D) Following in situ hybridization with labelled S. africanutn DNA in the presence of unlabelled H. vulgare DNA. The sites of hybridization of the labelled DNA were detected by blue light excited yellow fluorescence, while unlabelled chromatin fluoresces orange with the propidium iodide counterstain. The micrographs show that the two genomes, originating from different parents, tend not to be intermixed both at telophase and interphase, and the DNA originating from S. afncanum tends to be peripheral.
A single 0.25 μm thick section through a metaphase (M). prophase (P) and interphase (1) nuclei of the hybrid H. vulgare×S. africanutn which demonstrates that chromatin of different parental origin is not intermixed in sectioned material where the three-dimensional structure of the cell has been completely preserved (Leitch et al. 1990). (A)The section stained with the DNA-specific dye DAP1, showing that all chromatin fluoresces relatively uniformly. (B)The same section after probing with labelled genomic S. africanutn DNA and detection of the sites of hybridization with Texas Red (fluorescing red under green light excitation). The probe hybridizes strongly to the DNA of S. afncanum origin, which occurs in domains that are not intermixed with the DNA of H. vulgare origin (compare A and B).
A single 0.25 μm thick section through a metaphase (M). prophase (P) and interphase (1) nuclei of the hybrid H. vulgare×S. africanutn which demonstrates that chromatin of different parental origin is not intermixed in sectioned material where the three-dimensional structure of the cell has been completely preserved (Leitch et al. 1990). (A)The section stained with the DNA-specific dye DAP1, showing that all chromatin fluoresces relatively uniformly. (B)The same section after probing with labelled genomic S. africanutn DNA and detection of the sites of hybridization with Texas Red (fluorescing red under green light excitation). The probe hybridizes strongly to the DNA of S. afncanum origin, which occurs in domains that are not intermixed with the DNA of H. vulgare origin (compare A and B).
The two hybrids H. vulgare×S. africanum and H. chilense×S. africanum have been studied in greater detail than others. However, further examples of hybrids where the phenotype resembled the parent that contributed the peripheral chromosomes in the metaphase have also been reported (Bennett, 1984).
Parental dominance in segregating populations after hybridization
Another class of parental dominance occurs in the hybrid between two species of cotton, Gossypium hirsutum and G. barbadense. The characteristics and behaviour of this hybrid were first described many years ago (see Stephens, 1950 and Wallace, 1960, for references). In summary, after interspecific crossing and self-pollination of subsequent generations, the phenotype of the progeny often reverted to resembling one or other of the parental species showing that groups of parental characters were preserved (Schwendiman, 1974). In more formal terms, the progenies of interspecific hybrids do not give clear Mendelian ratios for the segregation of genes that are known to be allelomorphic on the basis of intraspecific tests. In some cases, which are well documented and comprehensively discussed by Stephens (1950), interspecific hybrids followed by backcrossing to the female parent can give a ‘marked deficiency’ in the expression of the gene from the male parent. He goes on to report that ‘the deficiency cannot be due to the effect of the introduced gene itself because with successive backcrosses the normal 1:1 segregation ratio tends to be more closely approached.’ Cytoplasmic effects cannot play a major role since there will be little introgression of the male cytoplasm into the hybrids and backcrosses. Although some results can be explained by the segregation of expression modifiers, and by cryptic structural differentiation of chromosomes which leads to selection against chromosome segments introduced from one parent, it seems probable that some of the effects are the result of the imprinting of genes, and certainly the area would be worthy of further investigation.
Parental dominance within inbred cereal hybrids
The phenotype of first generation hybrids between different cereals is discussed above. Intergeneric hybrids between cereals can be made and the chromosome number doubled to give polyploid plants, which are fertile and can be grown as a crop. One such crop is Triticale (2n=6x=42; × Triticosecale Witt.), a hybrid between a tetrapioid wheat (2n=4x=28; Triticum durum) and rye, 5. cereale (2n=2x=14). This crop is widely grown, particularly in eastern Europe and Canada, for use in animal feed and occasionally for milling for flour. However, like many of the other hybrids discussed above, it does not resemble a true intermediate between the parental species, but has many features of rye rather than wheat (see, e.g. Percival, 1923).
Methylation and gene expression in Triticale
Other papers in this volume discuss the importance of methylation, particularly of the cytidine nucleoside, in the control of gene expression and perhaps genomic imprinting (see also Michalowsky and Jones, 1989; Holliday et al. 1990). The experiment to be described here was initiated because of the potential importance or correlation between DNA methylation and gene expression - including the species-specific imprinting - in the wide hybrids and hybrid crops discussed above. The results obtained are preliminary, but worthy of publication because of their surprising nature. If methylation is important to imprinting in such hybrids, then demethylation might allow expression of otherwise supressed genes from one of the parents.
Demethylation has been correlated with transcriptional activation of genes in animals (Bird, 1986) and plants (Flavell et al. 1986). During germination of wheat seedlings, the amount of methylcytosine has been reported to diminish by about 15%, and later during development, new sequences are methylated (see Vanyushin, 1984); although, in common with most plants, a high proportion of the genome remains methylated. The drug 5-azacytidine can be used to demethylate DNA; it is incorporated into DNA during replication instead of cytidine, but cannot be methylated because of a nitrogen atom at the 5′ position (Jones and Taylor. 1980). Almost complete demethylation of genomic DNA occurs when a small percentage of substitution with azacytidine has occurred because of the effect of the compound as a false substrate and inhibitor of the methylases (see Parrow. Alestrom and Gautvik, 1989).
Demethylation effects on Triticale
Seeds of the Triticale variety Lasko were treated with various concentrations of azacytidine for up to two days. The plants arising from the treated seeds were referred to as the M1 (first mutation) generation, and the plants from these seeds were the M2 generation. Young leaves were removed from plants of both generations when they were about two months old for DNA extraction and subsequent analysis of restriction enzyme digests using modifications of standard methods (Sharp et al. 1988. 1989). Total genomic DNA from young leaves was digested to completion using a range of methylation-sensitive and -insensitive restriction enzymes including HpaII. MspI and ApaI. The nonradioactive chemiluminescence method. ECL (Amersham) was used for probe labelling, hybridization and the detection of sites of probe hybridization, following the manufacturer’s instructions and methods described by Anamthawat-Jónsson et al. (1990).
The luminographs in Fig. 6 show the lengths of restriction fragments which are homologous to the wheat ribosomal DNA clone pTa71 (Gerlach and Bedbrook, 1979; kindly provided by R. B. Flavell and M. O’Dell, IPSR Norwich, after recloning in pUC19). Both the M| and M2 generations show similar patterns of probe hybridization. Some remethylation of DNA is apparent between the two generations, but additional, short, fragments are found consistently from some of the restriction digests of DNA from the azacytidinetreated plants. The digests with MspI (recognition sequence CCGG, but methylation sensitive and cleaving only when the outer C is unmethylated) and ApaI (recognition sequence GGGCCC, and cleaving only if the internal C is unmethylated) show that there are differences in methylation of particular cytidine residues between treated and untreated plants within the ribosomal DNA repeat unit.
Luminographs showing restriction enzyme digests of DNA from azacytidine (AZC)-treated and control (Cont) Triticale plants of the M2 generation. All five enzymes are methylation sensitive. The Southern transfers were probed with the wheat ribosomal DNA clone pTa71 (see text). Additional lower molecular weight fragments are visible in the Mspl and Apal digests.
Luminographs showing restriction enzyme digests of DNA from azacytidine (AZC)-treated and control (Cont) Triticale plants of the M2 generation. All five enzymes are methylation sensitive. The Southern transfers were probed with the wheat ribosomal DNA clone pTa71 (see text). Additional lower molecular weight fragments are visible in the Mspl and Apal digests.
In summary, the luminographs show that the azacytidine treatment is effective in the demethylation of some cytidine residues in the Triticale DNA, and that these differences are inherited through at least two generations. Inheritance of methylation patterns through generations has been observed in the human (e.g. Silva and White, 1988); in the Triticale experiment, future work will aim to find other methylation differences, and to examine further plant generations. The substantial overall level of methylation observed in the Southern blots is expected, since remethylation of the DNA will occur over the first few cell cycles after removal or degradation of the azacytidine (Gruenbaum et al. 1981); clearly not all sites in the azacytidinetreated DNA are remethylated to the level observed in the controls in either the M1 or M2 generation.
Morphological analysis of plants
When the earliest plants were beginning to ripen, various morphological characteristics shown in Table 1 were scored (without knowledge of the treatment of the plant), following the Draft Guide-lines for the Conduct of Tests for Distinctness, Uniformity and Stability for Triticale (1UPOV, 1988). Visual assessment of other field characteristics given in the table VII of 1UPOV (1988) showed that there was little or no variation between the plants. For example, both treated and control plants had similar semi-erect growth habits, very weak or no anthocyanin coloration in awns or anthers, and similar ear lengths.
Results from the scoring of some of the characteristics of the M| plants are shown in Fig. 7. When compared with control plants, the azacytidine-treated plants showed some significant differences in ripeness, height and number of ears or tillers. Different times and chemical concentrations made relatively little difference to the plant performance (Fig. 7). Fig. 8 shows photographs of two plants of the glasshouse-grown M2 generation plants, which illustrate some of the characteristics scored in Fig. 7 in the M1 generation.
Results from the scoring of some of the characteristics of the M1 plants. All characters were scored on coded plants using a relative scale system. Shaded bar height represents the mean; and lines above and below, the standard deviation of the character. (A) Plant height from 1 (short) to 4 (tall). (B) Maturity from 1 (early to ripen) to 6 (late). (C) Number of tillers from 1 (few) to 6 (many). There are large differences between the treated and control plants, but few significant differences between the four treatments.
Results from the scoring of some of the characteristics of the M1 plants. All characters were scored on coded plants using a relative scale system. Shaded bar height represents the mean; and lines above and below, the standard deviation of the character. (A) Plant height from 1 (short) to 4 (tall). (B) Maturity from 1 (early to ripen) to 6 (late). (C) Number of tillers from 1 (few) to 6 (many). There are large differences between the treated and control plants, but few significant differences between the four treatments.
Photographs of two pots, each containing three glasshouse-grown Triticale plants, which illustrate some of the characteristics scored in Fig. 7 in the M1 generation. (A) M2 plant; the seed of the parent of these plants was treated with azacytidine. (B) Control; as (A) but treated with water only.
Photographs of two pots, each containing three glasshouse-grown Triticale plants, which illustrate some of the characteristics scored in Fig. 7 in the M1 generation. (A) M2 plant; the seed of the parent of these plants was treated with azacytidine. (B) Control; as (A) but treated with water only.
The experiments were carried out on a variety of Triticale, Lasko, which has passed standard tests for stability and uniformity of the crop, and hence would be expected to show negligible plant-to-plant variation. The azacytidine treatment was able to induce a range of new characteristics, presumably by enabling or inducing the expression of genes that were repressed in the hybrid. We do not know whether genes or control regions belonging to the parental genome which is less expressed are methylated more than those of the expressed genome; nor do we know the parental origin of the genes that are responsible for the altered characteristics. However, if the supression of blocks of genes originating from one or other parent were involved, then the system would be a good example of genomic imprinting.
Further verification and experimental work
The number of plants grown in the glasshouse was small, and their performance may not be a good guide to field performance, so detailed conclusions must await the scoring of larger numbers of plants grown outdoors. It will be important to study the heritability of the new characteristics (when the treated and control plants are used as both male and female parents), the methylation patterns in sequences other than the ribosomal RNA genes, the cytology of the plants and nucleolar expression patterns. While mutagenesis induced by the azacytidine cannot be ruled out, it is unlikely that such consistent DNA restriction fragment differences would be observed (Fig. 6 and additional data not shown), and unlikely that the morphological characteristics of many different plants would be consistent (Figs 7 and 8). It is conceivable that the range of variability induced by azacytidine is similar to that generated when plants (particularly of the Solanaceae) are regenerated from somatic cells. If so, then demethylation and activation of repressed genes may be involved in both systems.
Prospects
It will be important to examine the experimental wide hybrids discussed in the first section of the present work to see if there are any differences in the methylation of the two genomes. Perhaps gene expression, and hence the morphology of the hybrids, can be altered by demethylation; such changes may have effects on chromosome and genome disposition. Conversely, it will be important to examine aspects of chromosome and interphase gene disposition within the methylated and demethylated Triticales to see if there are any changes that correlate with the different expression patterns. Finally, investigation of the heritability patterns of the methylation polymorphisms in the two Triticale lines may give particularly significant results. The phenotype and methylation patterns of the progeny will point to any imprinting phenomena that correlate with parental methylation pattern, while excluding genetic differences between the parents.
ACKNOWLEDGEMENTS
I thank BP Venture Research Unit for support of this work, Mrs Julia Coates for much help with the Triticale azacytidine experiment, and Mr Phil Webb for information on breeding of Triticale and the supply of seed. I also thank my colleagues in the Karyobiology Group (Dr Andrew Leitch, Dr Trude Schwarzacher and Mrs Kesara Anamthawat-Jónsson) and Professor MD Bennett for assistance with the nuclear organization work.