Common wheat, Triticum vulgare, is a hexaploid species with 42 chromosomes in which the genomes of diploid wheat and of the diploid species Aegilops speltoides and Aegilops squarrosa are combined together1-5. There is considerable similarity between the chromosomes of these three diploids. Indeed, in hybrids between diploid wheat and Aeg. squarrosa and between diploid wheat and Aeg. speltoides, complete pairing at meiosis occasionally results in the formation of seven bivalents, although the mean bivalent frequency is somewhat less, being between three and four per cell5-6. Moreover, the attraction which results in bivalent formation in the hybrids is of such strength as to cause multivalent formation in tetraploids induced from them. However, each chromosome normally conjugates with its complete homologoue at meiosis in T. vulgare, and only bivalents are formed. Thus, the pairing attractions between the equivalent, homoeologous, chromosomes of different genomes either no longer exists or is no longer expressed. Polyploid wheat has therefore evolved to behave cytologically as a diploid. Further, the shift has been of such efficiency that there is very little chromosome pairing even in 21-chromosome haploids of T. vulgare, in which intergenome pairing is not restricted by the affinity of complete homologues. The cytological distinctiveness of the equivalent chromosomes of different genomes is all the more striking in view of the close genetic relationships which have enabled Sears7, by nullisomic-tetrasomic compensation, to recognize the seven homoeologous groups into which the complement of wheat can be arranged.
Recent results with various aneuploids have revealed something of the genetic control of the diploid behaviour of T. vulgare. The evidence centres on the derivatives of a 41-chromosome monosomic line, designated HH, which arose at Cambridge in the variety Holdfast. The formation of twenty bivalents and one univalent (Fig. 6) in HH monosomics is like the behaviour of all other monosomics of common wheat8. Meiosis is normal in the 42-chromosome segregants in the monosomic line, and in the 42-chromosome hybrids from crosses of HH monosomics to euploid plants of Holdfast. One arm of the chromosome monosomic in the HH line is rather more than twice the length of the other, but the chromosome has not yet been positively identified relative to the numbering based on the variety Chinese Spring.
Monosomics of HH have been used in crosses with 44-chromosome plants which had the full complement of wheat chromosomes plus a single pair of rye chromosomes. From these crosses five 20-chromosome, nulli-haploid, plants have been obtained, which, by the nature of their origin, must have had the haploid complement of Holdfast minus the chromosome which is monosomic in the HH line. The nulli-haploids arose in (a) the F1 of the cross HH monosomic x rye chromosome III diomic addition, and in (b) the F2 of crosses of HH monosomic x rye II, or rye III, disomic additions9. All the F1's which produced nulli-haploids in F2 had twenty bivalents composed of wheat chromosomes and the HH chromosome and a rye chromosome as univalents. In each situation, therefore, the 20-chromosomes in the embryo sacs which functioned parthenogenetically must have been those which had segregated from bivalentsthat is, all the wheat chromosomes except HH.
The meiotic pairing of the HH nulli-haploids has been compared with the pairing in 21-chromosome euhaploids of Holdfast. The meiotic behaviour of the euhaploid plant listed in Table 1 was similar to that of twelve other 21-chromosome plants of Holdfast examined over a period of four years. The number of bivalents in these plants never exceeded four, with a mean frequency of between 1.3 and 1.7 per cell, and trivalents were rare (Fig. 1). There was uaually only one chiasma per bivalent.
By contrast, in the HH nulli-haploid enumerated in Table 1, which is typical of five such plants examined in two successive years, the mean bivalent frequency exceeded the maximum observed in euhaploids. Trivalents were quite frequent with as many as five in some cells (Fig. 3), although there was never more than one per cell in the euhaploids. Very many more chromosomes per cell, a mean of 11.0 compared with a mean of 2.9, were involved in chiasma-associations and a much higher frequency of closed bivalent and of 'pan-handle' trivalents resulted from more intimate pairing. As many as nineteen of the twenty chromosomes of the nulli-haploid have been observed in various associations simultaneously, although never more than nine were involved in associations in the same cell in euhaploids.
The most reasonable hypothesis to account for these observations was that the deficient chromosome carried a gene, or genes, which restricted intergenomic, homoeologous, pairing; the alternative being that association between randomly distributed duplicate segments was normally inhibited. On either view, there being no other independent mechanism, in the absence of the pairing restriction normally undetected affinities were expressed. Crucial evidence that the pairing was homoeologoous rather than random was afforded by the high frequency of trivalents and infrequency of quadrivalents.
If homoeologous pairing could take place in HH nulli-haploids, it was argued that 40-chromosome nullisomics deficient for the HH pair should deviate from the strictly bivalent-forming regime of T. vulgare. Consequently, three HH nullisomics have been examined. All had large and frequent multivalents, and associated univalents, at meiosis. No method has yet been found of consistently preparing well-spread pollen mother cells of this material, but the complexity of the behaviour is readily apparent (Fig. 4). More than half the cells have at least one multivalent, usually a quadrivalent, and many have several. Associations of three, four, five and six are common, but no higher multivalents have been observed. However, the greater the number and size of multivalents the less is the chance of disentangling the snarl, so there is danger of observational bias. No doubt, also, there are homoeologous, or allosyndetic, bivalents which would be undetected unless markedly heteromorphic. The magnitude of the meiotic disturbance is therefore hard to assess; nevertheless, it is clear that, in the absence of the HH chromosome, T. vulgare ceases to be a stable bivalent former, a classical example of the autosyndetic pairing allohexaploid, and behaves as an intermediate, auto-allopolyploid.
The diploidizing mechanism is effective in the hemizygous state in monosomics and in euhaploids. Furthermore, in a haploid plant which had twenty normal wheat chromosomes plus the iso-chromosome formed from the long arm of the HH chromosome (Fig. 5), meiotic pairing was the same as in euhaploids. Moreover, there was no multivalent formation in a 41-chromosome monosomic which had twenty normal pairs of Holdfast chromosomes and in which the HH chromosome was represented only by a single telocentric of the long arm. The long arm alone is thus effective in prohibiting homoeologous association; but no evidence is available on the effects of the short arm alone. It may be that both arms have genes with equally effective control over the pairing behaviour, but this seems unlikely. The control of bivalent formation is probably restricted to one arm, and may indeed be effected by quite a localized region.
The implications of these results are both theoretical and practical. First, there has been considerable discussion on the methods by which many polyploids have attained their cytologically diploid character, with its obvious advantages in fertility and genetic stability. Several authors have favoured the theory that this situation has been achieved by the selective accumulation of many small changes of chromosome structure, which lead to the divergence of homoeologues. However, the selection of the localized product of a single mutational step, such as seems likely to have happened in wheat, is very much simpler to envisage, and would clearly be a more rapid and efficient process. To compare the frequencies of the genetical, or cytological, determination of diploid meiotic behaviour, other allopolyploid species should be examined for evidence of a mechanism similar to that in wheat.
Secondly, knowledge of the pairing restricting mechanism may be useful in wheat improvement. The utilization of related diploid species, for example, in the genus Aegilops, in wheat breeding is hampered because the chromosomes of the diploids do not pair with those of wheat, in intergeneric hybrid. To overcome this obstacle, Sears10 was forced to use X-rays to induce the translocation of a disease-resistance gene of Aeg. umbellaulata on to a wheat chromosome, and others9,11,12 have been compelled to explore alien chromosome addition and substitution lines. However, this intergeneric allosyndetic pairing may well take place in the absence of the HH chromosome, just as it does between chromosomes of the different genomes; aliens genes could then be introduced into wheat chromosomes by normal recombination. Critical hybrids of this constitution can be extracted from crosses of HH monosomics by the appropriate diploid.
Finally, nullisomic plants of this constitution may be used either within single varieties or in intervarietal hybrids in 'intergenome exchange' breeding. Thus a useful gene which showed a dose effect could be obtained duplicated or triplicated, represented on each chromosome of the homoeologous group. The chromosome structure within a variety could be re-patterned by breeding from HH nullisomics. Further, if an HH nullisomic were also an intervarietal hybrid, the release of variation, and the range of segregation in later generations, would be very much greater than from a euhaploid hybrid. Thus, the HH nullisomic may afford access to otherwise unavailable genetic variation.
However, the exploitation of HH nullisomics depends upon their fertility. There is no natural self-fertility, but a reasonable seed set has been obtained by pollinating HH nullisomics with Holdfast euploids, although the reciprocal cross hs been unsuccessful. There were several large multivalents at meiosis in the one 41-chromosome derivative, so far examined, of an HH nullisomic pollinated by a euploid. The HH chromosome regained from the euploid pollen was always a univalent, and homoeologous pairing must have been prohibited by its presence. The multivalents were thus indicative of translocation heterozygosity, the outcome of homoeologous pairing in the nullisomic parent. Homozygotes for new structural conditions can be derived from such plants, lthough further back-crossing may first be necessary to reduce the extent of structural alteration.
Knowledge of the situation described already begins to make plain some of the problems of wheat cytogenetics and perhaps of polyploidy in general. Further investigation cannot fail to extend this advantage and may well contribute to the cytogenetic manipulation of wheat for practical breeding purposes.
The interest of Dr. G.D.H. Bell in the development of this work is gratefully acknowledged.
International Congress of Genetics X (1958)
Chromosome Pairing and Haploids in Wheat
Plant Breeding Institute, Cambridge, England
Four kinds of haploid have been obtained in Triticum vulgare var. Holdfast (6x=42) from lines involved in making additions and substitutions of single pairs of rye chromosomes. (1) Euhaploids with 21 chromosomes had the complete haploid chromosome set of wheat. (2) Two types of 22-chromosome addition haploid had the haploid complement of wheat plus one rye chromosome. (3) Three types of 20-chromosome nulli-haploid had the haploid set of wheat chromosomes minus one. (4) One 21-chromosome substitution haploid had 20 chromosomes of the haploid set of wheat plus one rye chromosome. At first metaphase of meiosis in euhaploids, in the addition haploids, and in one of the nulli-haploids, there were mean bivalent frequencies of 1.0 to 1.7 per cell, and there was little difference between the three classes. Nulli-haploids of the second type, and the substitution haploid with the same 20 wheat chromosomes, had fewer bivalents than the euhaploids (mean 0.7 to 0.9 per cell), suggesting that the missing chromosome often participated in bivalent formation in euhaploids. However, in the third nulli-haploid there was a mean of 4.2 bivalents and 0.8 trivalents per cell. Cells with three trivalents were not uncommon and 29% of cells had five to seven bivalents, in contrast to the other haploids examined in which trivalents were very rare and not more than four bivalents per cell were formed. Further, in this nulli-haploid 20% of the total bivalents were rings, whereas no more than 4% were rings in the other haploids. It seems, therefore, that the chromosome deficient in this nulli-haploid carries a gene which reduces pairing, so restricting association between homologous chromosomes. Such a gene would have selective value in hexaploid wheat for the effect which cytologically diploid behaviour would have on fertility and genetic stability.
The evolutionary significance of this situation and its potential value in wheat breeding, especially in the introduction of alien genes, will be considered, reference being made to the behaviour of nullisomics and alien chromosome substitutions.