12.1 Mode, Time, and Place of Origin of the Allopolyploids

Allopolyploidization is a biological process that has played a central role in plant speciation and evolution (Manton 1950; Stebbins 1950, 1971; Grant 1971; Soltis and Soltis 1993, 1995; Soltis et al. 2009; Masterson 1994, de Bodt et al. 2005; Tang et al. 2008), and has driven and shaped the evolution of vascular plants, perhaps more than any other evolutionary process (Feldman and Levy 2005, 2009). It constitutes a radical and rapid mode of speciation that produces a new species by means of inter-specific or inter-generic hybridization of two diverging diploid species, followed by chromosome doubling of the F1 hybrids. Allopolyploidization produces a new species in a single step, a novel taxon that is immediately isolated genetically from its two parental species.

The pioneering discoveries of the accurate chromosome numbers of the different Triticum and Aegilops species (Sakamura 1918; Kihara 1919, 1924, 1937; Sax 1921a, b, 1922, 1927; Sax and Sax 1924; Percival 1921, 1923; Schiemann 1929; Sorokina 1937; Lilienfeld 1951, and reference therein) showed that these two genera comprise a polyploid series, containing diploids, tetraploids and hexaploid species. Subsequent studies of chromosomal pairing in hybrids between the allopolyploid and the diploid species of the group revealed the allopolyploid nature of the polyploids, namely, each polyploid species contains two or three different subgenomes that derived from diverging diploid-level genomes (Lilienfeld 1951, and reference therein). The conclusion that the polyploids of the wheat group are allopolyploids was also supported by the fact that only bivalents were formed at first meiotic metaphase of the polyploids (except that of the auto-allohexaploid T. zhukovskyi, which has the genomic constitution GGAAAmAm and therefore, produces few quadrivalents at meiosis). The diploid-like meiotic pairing pattern, i.e., regular bivalent formation due to exclusive homologous pairing, is characteristic of allopolyploids (Stebbins 1950, 1971). Hence, the two genera Aegilops and Triticum contains 18 allopolyploid species, 12 allotetraploids (10 of Aegilops and 2 of Triticum), and 6 allohexaploids, (4 of Aegilops and 2 of Triticum) (Tables 2.8, 9.3, and 10.5). These species have been subjected to extensive taxonomic, cytogenetic, genetic, biochemical, molecular, phylogenetic, and evolutionary studies by numerous scientists (see reviews of Kihara 1954; Mac Key 1966; Morris and Sears 1967; Kimber and Sears 1987; Feldman et al. 1995; Feldman 2001; Gupta et al. 2005; Dvorak 2009), which resulted in the identification of the diploids that were involved in the formation of most allopolyploids.

Whereas the diploid species of the wheat group evolved in the Pliocene [5.3–1.8 million years ago (MYA)] and early Pleistocene (1.8–0.01 MYA), the allotetraploids were produced afterwards, in the mid- or late Pleistocene (Tables 10.7 and 11.4). Ae. speltoides forms a monophyletic clade with the allopolyploid Triticum species, the emmer lineage (allotetraploid T. turgidum and allohexaploid T. aestivum) and the timopheevii lineage (allotetraploid T. timopheevii and auto-allohexaploid T. zhukovskyi) (Gornicki et al. 2014). The geographical distribution of chloroplast haplotypes of the wild allotetraploid wheats and Ae. speltoides illustrates the possible geographic origin of the emmer lineage in the southern Levant, the present-day chloroplast diversity center of wild emmer (Gornicki et al. 2014). This is in accord with the finding of Jan Dvorak (personal communication) suggesting that wild emmer was formed in the vicinity of Mt. Hermon. The origin of the timopheevii lineage was in northern Iraq, around Arbil, where many accessions of major haplotypes of wild T. timopheevii and Ae. speltoides were found (Gornicki et al. 2014). Two accessions of wild T. timopheevii, collected near Arbil, carry the dominant haplotype of Ae. speltoides, hinting that they are the ancestral state of the species. The high level of karyotype diversity in wild T. timopheevii collected in northern Iraq (Badaeva et al. 1994), is consistent with the claims that the species originated there. Two chloroplast haplotypes, H08 and H10, were present in both domesticated T. timopheevii ssp. timopheevii and in T. zhukovskyi, either due to recurrent allohexaploidization events or outcrossing between these two species (Gornicki et al. 2014).

Most of the allopolyploid species of Aegilops were presumably produced in the east Mediterranean region and west Asia, the distribution center of their diploid parents. The Aegilops allotetraploids then spread further westwards along the Mediterranean basin, as well as in more northern and eastern directions (Kihara 1954). The distribution of the Aegilops allotetraploids has been halted by natural boundaries and lack of suitable environments, such as the Saharan and Arabian deserts, the central Asian steppes, the Tian Shan and Himalayan Mountains or the coldness of the continental climate affecting their spread to the north and east. In contrast to the Aegilops allotetraploids, wild Triticum allotetraploids remained in or near the site of their origin (Harlan and Zohary 1966). Wild emmer, originating in the southern Levant, only spread to the northern Levant, where several new chloroplast haplotypes were established (Gornicki et al. 2014).

Middleton et al. (2014), based on sequencing of the chloroplast genome of 12 Triticeae species, estimated that the B subgenome donor to allopolyploid wheat diverged from Ae. speltoides approximately 980,000 years ago, but Li et al. (2022), based on whole genome sequencing of the Sitopsis species, suggest that this divergence occurred much earlier, 4.49 MYA (Table 11.3). The divergence of the S genome of Ae. speltoides and the B genome of the assumed parent of allopolyploid wheat, might have been followed by a speciation. In accord with this possibility, none of the chloroplast haplotypes carried by the 391 accessions of the emmer lineage was found among the 450 Ae. speltoides accessions (Gornicki et al. 2014). Hence, the female donor of the cytoplasm and the B subgenome to the emmer lineage is either a yet undiscovered relative of Ae. speltoides, or perhaps even extinct. Alternatively, the time span from the formation of wild emmer allowed the speltoides accession(s) donor to undergo evolutionary changes, and consequently, the current speltoides accessions differ from it.

The beginning divergence between T. urartu and the A-subgenome donor to allopolyploid wheat is estimated to have occurred 1.28 MYA and between Ae. tauschii and the D-subgenome donor to T. aestivum to have occurred somewhat more than 880,000 years ago (Li et al. 2022; Table 11.3). Gornicki et al. (2014) based on chloroplast DNA sequencing, suggested, that wild emmer, T. turgidum ssp. dicoccoides, was formed 700,000 years ago, whereas Marcussen et al. (2014), based on sequencing of a several hundred nuclear genes, suggested that this event occurred 800,000 years ago, and Li et al. (2022), based on whole genome sequencing, considered it to be somewhat earlier.

Tsunewaki et al. (1991) and Wang et al. (1997), based on intraspecific levels of restriction fragment length polymorphism and single-strand configuration polymorphism in chloroplast DNA, concluded that formation of wild emmer is more ancient than that of T. timopheevii. Indeed, Mori et al. (1995), based on restriction fragment length polymorphism (RFLP) analysis of nuclear DNA, proposed that the wild form of T. timopheevii, ssp. armeniacum, formed 50,000–300,000 years ago. Yet, the data of Gornicki et al. (2014) of sequence-based divergence time of chloroplast DNA, showed that the timopheevii lineage diverged from Ae. speltoides 400,000 years ago. Thus, the allotetraploidization events of wild emmer and of wild T. timopheevii occurred within the last 800,000 and 400,000 years ago, respectively. The Timopheevii lineage and Ae. speltoides chloroplast genomes are very closely related, consistent with Ae. speltoides being the maternal donor of the cytoplasm and G subgenome to T. timopheevii (Shands and Kimber 1973; Kimber 1974; Dvorak and Zhang 1992a, 1992b; Wang et al. 1997). Yet, Li et al. (2022) reported that the G-subgenome donor to T. timopheevii is not Ae. speltoides but rather, a species that diverged from Ae. speltoides 2.85 MYA (Table 11.2).

The allopolyploid species of the wheat group were formed 1.0 MYA or later (Table 11.3). Tsunewaki (2009), based on plasmon comparison between related allotetraploid species and their closest diploid relatives, concluded that wild emmer wheat, T. turgidum ssp. dicoccoides, Ae. crassa and Ae. geniculata (formerly Ae. ovata), are the oldest allotetraploids in the wheat group, followed by other allotetraploids, some of which originated relatively recently. Middleton et al. (2014) reached similar conclusions, in their report that Ae. geniculata is one of the oldest allotetraploid Aegilops species and formed about 1.62 MYA. On the other hand, Ae. cylindrica is a younger allotetraploid whose D subgenome diverged from D genome of Ae. tauschii, approximately 180,000 years ago (Middleton et al. 2014).

Many allopolyploids with combinations similar to natural ones, were synthetically produced, indicating that these genomic combinations can be created in the lab (e.g., Ozkan et al. 2001). Yet, some genomic combinations that were formed in the lab, are absent in nature. The reasons for this can be partly explained by the current eco-geographical isolation of the corresponding parental species and possibly also by the low viability of some combinations that might have hybrid weakness and consequently, could not compete with their parental species and establish themselves in nature.

12.2 Cytological Diploidization

12.2.1 Genetic and Epigenetic Changes Due to Allopolyploidization

The newly formed allopolyploid species, are hybrid species containing two or three different subgenomes enveloped within one nucleus. This situation exerts significant genetic stress on the nascent allopolyploids that must overcome several immediate challenges in order to be able to successfully establish themselves and survive in nature (Levy and Feldman 2002, 2004; Feldman and Levy 2005, 2009, 2011, 2012). Overcoming these challenges is achieved through immediate triggering of a variety of cardinal genetic and epigenetic changes that affect genome structure and gene expression.

Genomic changes of newly formed allopolyploids of Triticum and Aegilops comprise chromosome rearrangements, elimination of coding and noncoding DNA sequences, transposable element (TEs) and tandem repeat elimination or amplification, and gene expression modifications (Feldman et al. 1997; Ozkan et al. 2001; Shaked et al. 2001; Ma and Gustafson 2005, 2006; Cheng et al. 2019). In addition, allopolyploidization triggers the activity of a variety of TEs, which may affect gene expression and induce many structural rearrangements, including deletions or duplications of chromosomal segments (Kashkush et al. 2002, 2003). Yaakov and Kashkush (2011a) and Bariah et al. (2020) reviewed the accumulated data on genetic and epigenetic dynamics of TEs, particularly in newly formed allopolyploid wheats, and discussed the underlying mechanisms and the potential biological significance of TE dynamics following allopolyploidization. Similarly, Parisod and Senerchia (2012) presented evidence that TEs play a central role in driving genome reorganization subsequent to allopolyploidization, predominantly involving deletion of DNA sequences, as opposed to transposition. Genome reorganization generally occurs in the first generations following allopolyploidization and involves extensive epigenetic changes in the vicinity of TEs. Since massive transpositional activation of TEs could be highly deleterious to the nascent allopolyploid, Parisod and Senerchia (2012) speculated that only allopolyploids with transposition controlled through substantial repatterning of epigenetic marks and/or having lost TE fragments, could be viable.

Genetic and epigenetic alterations brought about by allopolyploidization may be induced by small RNAs, which affect key cellular processes, including TE activity, chromatin acetylation, cytosine methylation, and gene expression. Kenan-Eichler et al. (2011), performing high-throughput sequencing of small RNAs of parental, intergeneric hybrids, and of synthetic allopolyploid plants of the wheat group, found that the percentage of small RNAs corresponding to miRNAs increased whereas the percentage of siRNAs corresponding to TEs decreased soon after allopolyploidization. The reduction in siRNAs, together with decreased CpG methylation of the Veju TE element, shown by the same group, represent hallmarks of TE activation. TE-siRNA downregulation in the newly-formed allopolyploids may contribute to their genome destabilization at the initial stages of speciation.

One of the major challenges of the nascent allopolyploids of the wheat group stems from the fact that the homoeologous chromosomes of the different subgenomes are still genetically very closely related, as shown by the ability of four doses of one chromosome to compensate for the deficiency of either of its two homoeologues in allohexaploid wheat (Sears 1952a, 1966). Moreover, molecular studies revealed a high level of gene synteny and collinearity in the homoeologous chromosomes of allopolyploid wheat (Gale et al. 1995; Gale and Devos 1998). This genetic relatedness should enable the homoeologues to pair and recombine during meiosis. However, since homoeologous pairing leads to reduced adaptiveness, namely, partial sterility and multisomic inheritance, mechanisms restricting meiotic chromosome pairing to fully homologous chromosomes, i.e., a rapid process leading to cytological diploidization, have been acquired in allopolyploids of this group.

Since the different subgenomes of the wheat allopolyploids are genetically and structurally closely related, allopolyploids of this group are, in fact, segmental allopolyploids, rather than genomic allopolyploids. Yet, cytologically, they behave as genomic allopolyploids, namely, there is only intra-genomic pairing in the form of bivalents between fully homologous chromosomes (diploid-like meiotic behavior). By restricting pairing to homologous chromosomes, and preventing inter-subgenomic pairing, i.e., between homoeologous chromosomes (chromosomes of different subgenomes that are partly homologous), the cytologically diploidizing systems ensure bivalent pairing at meiosis and, consequently, regular segregation of genetic material, complete fertility, genetic stability, and disomic, rather than polysomic, inheritance, sustaining one of the advantages of allopolyploidy, i.e., heterosis between subgenomes. The establishment of allopolyploids of the wheat group as successful competitive taxa in nature, required the development of cytological and genetic systems that prevent inter-subgenomic pairing, while allowing intra-genomic pairing. Meeting this challenge is presumably a critical prerequisite for the success of the newly formed allopolyploids, that ensures their increased fitness and successful establishment in nature as competitive entities.

One means of bringing about cytological diploidization in newly formed allopolyploids of the wheat group is via rapid differential elimination of a number of DNA sequences, either low-copy or high-copy, from one subgenome in allotetraploids and from the additional subgenome in allohexaploids, leaving the concerned sequences in only one homologous pair, and thus, rendering them homologous-specific sequences (Feldman et al. 1997; Liu et al. 1998a, b; Ozkan et al. 2001; Han et al. 2003, 2005; Salina et al. 2004; Baum and Feldman 2010; Guo and Han 2014). Such differential elimination, i.e., some sequences are eliminated from one subgenome, whereas others are eliminated from the second subgenome, brings to a rapid, further divergence of the homoeologous chromosomes. The subsequent homologous-specific sequences may determine chromosome homology and strengthen homology search and attraction at the commencement of meiosis. Such DNA sequences exist in all the diploid progenitors of the allopolyploid species of the wheat group, whereas in the natural allopolyploids, they exist in only one homologous pair of one subgenome, implying that they were eliminated during or soon after the allopolyploidization event. (Feldman et al. 1997; Ozkan et al. 2001). This rapid elimination, occurring during or soon after the formation of the allopolyploids, was designated as a revolutionary change (Feldman et al. 1997; Ozkan et al. 2001; Table 12.1). No noteworthy further elimination of sequences occurs in the subsequent generations of the allopolyploids. The elimination of sequences is reproducible, as shown by elimination of the same sequences in synthetic and natural allopolyploids bearing the same genomic combinations (Ozkan et al. 2001; Han et al. 2005). It has been concluded that instantaneous elimination of DNA sequences in the first generation(s) of the newly formed allopolyploids of Triticum and Aegilops was one of the major and immediate responses to allopolyploidization.

Table 12.1 Allopolyploid species of the wheat group
Full size table

Rapid elimination of DNA sequences in newly formed allopolyploids was also observed in Triticale, a synthetic allopolyploid derived from wheat and rye (Boyko et al. 1984, 1988; Ma and Gustafson 2005, 2006; Bento et al. 2011). Elimination of DNA sequences were also reported in newly formed allopolyploids of Brassica (Song et al. 1995), Nicotiana (Skalická et al. 2005), and in Arabidopsis (Madlung et al. 2005). It seems that sequence elimination in newly formed allopolyploids is a widespread phenomenon.

DNA elimination seems not to be random at the intra-chromosomal level. For example, Liu et al. (1997a) found that the chromosome-specific sequences on chromosome arm 5BL in allohexaploid wheat are not distributed along the chromosome arm but, rather, cluster in terminal (sub-telomeric), subterminal and interstitial regions of this arm, rendering these regions extremely homologous-specific. Hence, it is tempting to suggest that these chromosome-specific regions, the only regions that determine homology, are equivalents to the classical “pairing-initiation sites” that play a critical role in homology search and initiation of chromosomal pairing at the beginning of meiosis (Feldman et al. 1997).

The extent of DNA elimination was estimated by determining the amount of nuclear DNA in natural allopolyploids and in their diploid progenitors, as well as in newly synthesized allopolyploids and in their parental plants (Ozkan et al. 2003; Eilam et al. 2008, 2010). Allopolyploid species of the wheat group contain 2–10% less DNA than the additive sum of their diploid parents, and synthetic allopolyploids exhibit a similar loss, indicating that DNA elimination occurs soon after allopolyploidization (Nishikawa and Furuta 1969; Furuta et al. 1974; Eilam et al. 2008, 2010; Table 12.1). In addition, the narrow intra-specific variation in DNA content of the natural allopolyploids indicates that the loss of DNA occurs immediately after allopolyploid formation, and that there is almost no subsequent change in DNA content during the evolution of the allopolyploid species (Eilam et al. 2008). Boyko et al. (1984, 1988) and Ma and Gustafson (2005) found that there was a major reduction in DNA content in the course of Triticale formation, amounting to about 9% for the octoploid and 28–30% for the hexaploid Triticale. In this synthetic allopolyploid, the various subgenomes were not equally affected; the wheat genomic sequences were relatively conserved, whereas the rye genomic sequences underwent a high level of variation and elimination (Ma et al. 2004; Ma and Gustafson 2005; Bento et al. 2011). Bento et al. (2011) reanalyzed data concerning genomic analysis of octoploid and hexaploid Triticale and found that restructuring depended on parental genomes, ploidy level, and sequence type (repetitive, low copy non-coding and/or coding). Similarly, in hexaploid wheat, subgenome D underwent a considerable reduction in DNA, while the A and B subgenomes were not reduced in size (Eilam et al. 2010). The fact that DNA elimination in Triticale and in allohexaploid wheat hardly affects the allopolyploid 4x and 6x wheat parents may be explained by the assumption that the subgenomes of these parents already underwent elimination during their formation and the additional subgenome of the diploid parent had to be adjusted.

12.2.2 Suppression of Homoeologous Pairing by the Ph1 Gene

12.2.2.1 Discovery and Induction of Mutations in Ph1

Superimposed on the divergence of the homoeologous chromosomes due to differential sequence elimination, is a genetic system that contributes to the maintenance and reinforcement of the exclusive bivalent pairing in the allopolyploid Triticum species, ensuring that chromosome pairing is wholly restricted to homologous chromosomes (reviewed by Sears 1976b). The most potent suppressor of homoeologous pairing in allopolyploid wheats themselves and in their interspecific or intergeneric hybrids, is a gene located on the long arm of chromosome 5B (5BL). The first evidence that chromosome arm 5BL of cv. Chinese Spring (CS) of bread wheat, T. aestivum ssp. aestivum, carries a gene that suppresses chromosome pairing in hybrids came from the observation of Okamoto (1957), who showed that F1 hybrids between CS, deficient for 5BL, and the amphiploid T. monococcum ssp. aegilopoides-Ae. tauschii (genome AADD), exhibited much higher pairing than the hybrids with 5BL. Soon after, it was independently deduced, by Sears and Okamoto (1958) in the USA and Riley and Chapman (1958) in the UK, that a gene on chromosome arm 5BL (the former designation of chromosome 5B is V), designated Ph (pairing homoeologues; Wall et al. 1971a), and Ph1 by Sears (1982) and Jampates and Dvorak (1986), suppresses homoeologous chromosomes pairing, without imparting any effect on homologues chromosome pairing. Consequently, only bivalents are formed at meiosis and allohexaploid wheat, in spite of being segmental allopolyploid, behaves like a typical genomic allopolyploid. In allohexaploid wheat plants lacking chromosome 5B or chromosome arm 5BL, a low level of pairing occurs between the homoeologues of the A, B, and D subgenomes (Riley 1960; Sears 1976a). On the other hand, in haploids of allopolyploid wheats (Riley and Chapman 1958; Jauhar et al. 1999) or in hybrids between wheat lacking 5BL and related species, when homologous chromosomes do not exist, the otherwise normally suppressed homoeologous pairing occurs at a relatively high frequency (Riley 1960, 1966; Sears 1976b). Supporting evidence that in the absence of 5BL, meiotic pairing indeed involves homoeologous chromosomes, has come from the finding of Okamoto and Sears (1962), who showed that the pairing in haploids of hexaploid wheat is largely between chromosomes of different subgenomes belonging to the same homoeologous group. Additionally, the work of Riley and Kempanna (1963) showed that the absence of chromosome arm 5BL in bread wheat induced pairing of homeologues, in addition to that of homologues.

Chromosome 5B of the allotetraploid T. turgidum compensates for the absence of chromosome 5B in T. aestivum, indicating the presence of a Ph gene, similar to that on chromosome 5B of hexaploid wheat, at the tetraploid level (e.g., Riley 1960; Dhaliwal 1977; Dvorak et al. 1984). Also, it was shown (Feldman 1966a) that chromosome 5G of T. timopheevii compensates for the absence of 5B of T. aestivum, implying that this allotetraploid species contains a gene system similar to that in chromosome 5B of T. aestivum. Since T. timopheevii contributed the G and A subgenomes to hexaploid T. zhukovskyi, it is assumed that the latter also contains a Ph-like gene. Thus, all the four allopolyploid species of Triticum contain a Ph-like gene system suppressing homoeologous pairing, which drives cytological diploidization.

The discovery that homoeologous pairing is prevented by chromosome arm 5BL explained why the chromosomes of the three subgenomes of allohexaploid wheat are unable to pair with each other, neither in the allopolyploid Triticum species nor in their inter-generic hybrids (Riley 1960, 1966; Sears 1976b), even though they are genetically so closely related. The discovery that a deletion for 5BL provides the means of inducing the chromosomes of wheat to pair and recombine with those of related species and genera, has been used widely in the transfer of valuable genes from wild relatives to wheat (e.g., Sears 1976b, Feldman 1988). No wonder, therefore, that from the time of its discovery, Ph1 has had a great impact on wheat cytogenetics, breeding and beyond, and its isolation, structure, and mode of action have been the subjects of intensive cytogenetic and molecular research.

The suppressive effect of Ph1 on homoeologous pairing in inter-generic Triticum hybrids is absolute. In contrast, its effect on homoeologous pairing in hexaploid wheat itself might be dispensable, as plants deficient for this gene exhibit relatively little homoeologous pairing [less than one multivalent per cell, which results from inter-subgenomic pairing (Sears 1976b)]. Interestingly, a Ph-like effect has not been found in any of the allopolyploid species of the closely related Aegilops genus (Riley and Law 1965; Sears 1976b). None of the allopolyploid Aegilops genotypes studied by McGuire and Dvorak (1982) compensated for the absence of Ph1 in hybrids with T. aestivum lacking this gene. Likewise, Cünado (1992) reported that all the polyploid species of Aegilops display a strict bivalent pairing behavior, with the exception of allotetraploid Ae. biuncialis and allohexaploid Ae. juvenalis, which occasionally form a few multivalents at first meiotic metaphase. These findings substantiated the belief that the allopolyploid Aegilops species do not possess a gene that has a Ph-like activity. Nevertheless, these species also exhibit exclusive bivalent pairing of fully homologous chromosomes, presumably due to the structural changes that were created by the sequence-elimination system.

The use of bread wheat plants lacking chromosome 5B or chromosome arm 5BL to induce homoeologous pairing in hybrids between allopolyploid wheats and their wild relatives, encounters some difficulties. The deletion of chromosome 5B or chromosome arm 5BL in ssp. aestivum resulted in reduced vigor and fertility and was associated with difficulty maintaining them as laboratory lines. To offset these deleterious effects, an extra dosage of chromosome 5D must be provided to compensate for the absence of 5B. This complicated the use of such a line (nullisomic 5B-tetrasomic 5D) in hybrids with wild relatives that would contain a pair of 5D. What was really needed is a simple mutation or a deletion of Ph1.

The first attempt to produce a mutation in Ph1 was made by Wall et al. (1971a), who induced a mutation, designated 10/13, by ethylmethanesulfonate (EMS) treatment of ditelosomic 5BL plants of cv. CS of ssp. aestivum. The homozygous mutant had no effect on homoeologous pairing, but in F1 hybrids with rye, one dose this mutation increased the level of homoeologous meiotic pairing. Wall et al. (1971b) considered 10/13 to be a recessive allele to Ph1 and designated it ph1a. Yet, the 10/13 mutant did not entirely meet the requirements of being an allele of Ph1, as the level of pairing it induced in hybrids was not as high as that in nullisomic 5B (Wall et al. 1971a). Ph1 deficiency is actually expected to cause higher pairing than 5B nullisomy, because the short arm of 5B, 5BS, carrying a promoter of pairing (Feldman 1966b), would still be present in a mutant line. Sears (1982, 1984), based on genetic studies (test for allelism), found that the 10/13 mutation is actually a mutant of another homoeologous-pairing suppressor, Ph2, located on chromosome arm 3DS, rather than of Ph1. Accordingly, the correct designation of the 10/13 mutation is ph2a, and not ph1a (Sears 1984).

A deletion in chromosome arm 5BL that includes the Ph1 gene, was induced by Sears (1977) via X-irradiation of normal pollen and using it in pollination of on plants monosomic for a 5B chromosome carrying a morphological marker. The 438 monosomic progeny of the cross with an irradiated chromosome 5B, were crossed with Ae. peregrina and the obtained F1 hybrids were analyzed for increased pairing. Only one mutation was obtained, which appeared to be a mutation of Ph1. The homozygote mutant differed appreciably from the wild type in morphology, and also exhibited somewhat reduced vigor and fertility. In addition, male transmission from the heterozygote was less than 40%, leading Sears (1977) to conclude that the mutation is a deficiency. The author supposed that on the long arm of 5B, distal to Ph1, there is at least one gene for male fertility, and therefore, the one fertile mutant recovered would then be an interstitial deficiency that includes Ph1 but not the fertility gene. This lies in accord with the fact that Ph1 was mapped near the middle of the 5BL arm (Jampates and Dvorak 1986), about 1.0 centimorgan (cM) from the centromere (Sears 1984). This mutation, designated ph1b (Sears 1982), exhibits some homoeologous pairing under homozygous conditions (about 0.66 multivalents per pollen mother cell (PMC), (Sears 1977) and therefore, its progeny should be selected against inter-subgenomic translocations that result from homoeologous pairing. On the other hand, in euhaploids of allopolyploid wheats (Jauhar et al. 1999), as well as in interspecific and intergeneric hybrids (Sears 1976b), ph1b allows for somewhat higher homoeologous pairing than obtained in hybrids deficient for 5BL, because of the presence of 5BS.

One year after the induction of ph1b in allohexaploid wheat, a mutation in the ph1 gene of allotetraploid wheat was obtained via X irradiation of cv. Cappelli of Triticum turgidum ssp. durum (Giorgi 1978, 1983; Giorgi and Cuozzo 1980; Giorgi and Barbera 1981a, b). The mutant was originally identified by the altered morphology of chromosome 5B, which, in some plants, appeared to be shorter than in Cappelli, whereas, in other plants in the same progeny, the arm was longer than in Cappelli (Giorgi and Barbera 1981a). It was found that the plants bearing the short 5B had a deletion in the region containing Ph1 while those possessing the long 5B, had a duplication of the chromosome segment that includes Ph1 (Giorgi and Cuozzo 1980; Giorgi and Barbera 1981a, b).

This mutation, designated ph1c (Jampates and Dvorak 1986), is also an interstitial deletion for a segment of the long arm of chromosome 5B containing ph1. In addition to the deletion, plants with a tandem duplication of part of 5BL that includes Ph1 were produced, thus, plants homozygous for this duplication have four doses of Ph1. The deletion and the duplication in 5BL of Cappel1i have a common origin, probably occurred in the same cell and comprised the same chromosome segment (Dvorak et al. 1984). Hybrids between ph1c and wild wheat relatives exhibited high homoeologous pairing as almost hybrids with ph1b. On the other hand, hybrids bearing the duplicated segment of 5BL and consequently, with two doses of Ph1, had less pairing than hybrids with wild type Cappelli that carry only one dose of Ph1.

To infer the approximate location of the Ph1 gene in the 5BL arm, Dvorak et al. (1984) compared the C-banding of the 5BL arm of wild type Cappelli, the 5BL arm of the homozygous mutant ph1c, and the 5BL arm of the line with the duplication. Like Giorgi and Cuozzo (1980) and Giorgi and Barbera (1981a, b), they found that 5BL of ph1c was shorter than 5BL of the wild type, owing to a deletion of one of two inter-band regions in the middle of the arm. In the line with the duplication, the 5BL arm was longer than its Cappelli counterpart and the interband region that was absent in ph1c was twice as long in the line with the duplication. Hence, C-band patterns confirmed that the difference between ph1c and wild-type Cappelli was due to a deletion of a chromosome segment from 5BL of the mutant, whereas the same segment was duplicated in the line with the duplication. Jampates and Dvorak (1986) crossed the two mutant lines, ph1c and the line with the duplication, and cv. Cappelli with several allotetraploid species of Aegilops. Hybrids involving ph1c had higher levels of chromosome pairing than those involving cv. Cappelli, whereas those involving the duplication had lower levels of pairing than those involving cv. Cappelli. Jampates and Dvorak (1986) found that ph1c is a deletion of sub-region 5BL12.3 between C-bands 5BL12.2 and 5BL21 (The C-bands and euchromatin on the 5BL arm were designated according to the system proposed by Gill (1987) for cv. Chinese Spring).

Several translocations involving chromosome arm 5BL were studied with the objective of finding the approximate location of the Ph1 locus (Driscoll and Quinn 1968; Makino 1970; Mello-Sampayo 1972). In each case, it was inferred that Ph1 is in the proximal part of the arm. These data may not conflict with the reported location of the Phl locus in the area of the first C-inter band region in the middle of the 5BL arm.

Gill and Gill (1991) presented direct evidence corroborating that the phlb mutation is a submicroscopic deletion. The probe XksuS1-5, from a genomic library of Ae. tauschii, detects a single fragment of each of the long arms of chromosomes 5A, 5B and 5D. The specific chromosome 5B fragment it recognizes, which was present in Chinese Spring, was missing in phlb and ph1c mutants. Therefore, Gill and Gill (1991) suggested that XksuS1-5 lies adjacent to Phl on the same chromosome fragment that is deleted in phlb and phlc. Thus, XksuS1-5 can be used to tag the Ph1 gene and might also be a useful marker in cloning Ph1 by chromosome walking.

Gill and Gill (1996) also developed an PCR-based screening for the detection of Ph1. This assay is based on the 0.6 kb probe pHvksu8, from barley, which maps in the interstitial region where Ph1 is located. The probe was sequenced and 20 bp forward and reverse primers were generated. Using PCR, these primers amplified a fragment of chromosomes 5A, 5B, and 5D. The specific 5B fragment was not amplified in the phlb mutant, and thus, this probe can be used as a diagnostic fragment to screen plants for the presence or absence of the Phl gene.

Likewise, in an effort to tag the specific chromosomal region where Ph1 is located, Segal et al. (1997) micro-dissected bread wheat chromosome arm 5BL and produced a plasmid library by random PCR amplification and cloning. From this library, a 5BL-specific probe, WPG90, was isolated and mapped within the region corresponding to the interstitial deleted chromosome fragments carrying Ph1 in bread and durum wheat. This WPG90-based PCR assay allows for easy identification of homozygous genotypes deficient for Ph1.

Deletions in bread wheat chromosomes, induced by the gametocidal chromosome of Ae. cylindrica, arose from a single break, followed by the loss of the distal chromosome region (Endo 1988; Gill et al. 1993a). The breakpoints of the deletions in chromosome arm 5BL, namely, 5BL-9, 11, 1 and 5, are at fraction lengths (FLs) 0.76, 0.59, 0.55 and 0.54, respectively. (FL values were calculated by dividing the arm ratio of the deletion line with that of the normal). To determine the presence and absence of Ph1 in the various deletion lines of 5BL, Gill et al. (1993b) crossed the deletion lines with Ae. peregrina and studied chromosome pairing in the Fl hybrids. Low pairing indicates the presence of Phl, whereas high pairing indicates its absence. The hybrids derived from the deletion lines 5BL-9 and 5BL-11, had low pairing, i.e., these deletion lines possess Ph1. On the other hand, hybrids derived from deletion lines 5BL-1 and 5BL-5, had high pairing, indicating the absence of the Phl in these deletion lines. Since deletion 5BL-1 was the smallest deletion lacking the Phl gene and 5BL-11 was the largest deletion in which Ph1 is present, it was concluded that Phl is located in the chromosome region between FL 0.55 and 0.59 (the breakpoints of deletion 5BL-1 and 5BL-11), respectively.

The deletions in the mutations phlb and phlc are 1.05 µm and 0.89 µm, and located proximal to C-band 5BL2.1 (Gill and Gill 1991; chromosome-banding nomenclature is according to Gill et al. 1991). Thus, the deletion of ph1b encompasses 73.5 million bp and that of ph1c 62.3 million bp (estimated from genomic DNA content of 16 billion base pairs divided by 250 µm, the total length of the chromosome complements of bread wheat; Gill and Gill 1991). Similarly, Foote et al. (1997) identified the Ph1 locus of T. aestivum in the 70 Mb region between the Xrgc846 and Xpsr150A markers. Gyawali et al. (2019) delimited the ph1b deletion to a genomic region of 60 Mb by chromosome walking.

Gill and Gill (1991) and Gill et al. (1993b), using physical mapping of DNA markers on the 5BL region corresponding to the ph1b deletion, demarcated the Ph1 gene to a submicroscopic chromosome region, i.e., the Ph1 gene region, whose size is ~ 2.4 Mb. Both Griffiths et al. (2006) and Sidhu et al. (2008) delimited the Ph1 region to a 2.5-Mb region within the ph1b deletion. Gill et al. (1993b) identified three probes, XksuS1, Xpsr128 and Xksu75, that mapped in the Ph1 gene region of Chinese Spring and were missing in the deletion of phlb. These authors reported that the Ph1 gene region is bordered by the breakpoints of two deletions (5BL-1 and phlc) and is marked by the DNA probe XksuS1. Two other DNA probes, Xpsr128 and Xksu75, flank the Ph1 region, Xpsrl28 being proximal and Xksu75 being distal. These two probes map in the interstitial deletion of the phlb mutant, whereas only Xksu128 maps in the phlc deletion, while Xksu75 detects a DNA fragment distal to the deletion Gill et al. 1993b).

In summary, Ph1, along with the DNA marker XksuS1, is located between the breakpoints of deletion 5BL-1 and phlc. The breakpoint of phlb is distal to the Phl region but proximal to the C-band 5BL2.1, the region that possesses Xksu75. Xpsr128 is present proximal to the Ph1 gene region. The chromosome region around Ph1 is high in recombination, as the genetic distance of the region spanned by XksuS1 and Xksu75 is at least 9.3 cM. This chromosome region is also prone to breaks, as the breakpoints of the mutants phlb and phlc also map in the region.

12.2.2.2 Theories Concerning the Mode of Action of Ph1

At present, it is not known how Ph1 prevents homoeologues from pairing at meiosis and what are the product(s) and the subcellular target(s) of its activity. Does Ph1 recognize specific regions that differ between homoeologous chromosomes and thus enable to distinguish between homologous and homoeologous chromosomes through homology search or pairing initiation? In this respect, are the homologous-specific sequences, produced by the differential elimination during the allopolyploidization event, recognized by Ph1? Or does Ph1 operate on the divergence of a large number of differing sequences spreading out along the homoeologous chromosomes? Does the fact that Ph1 impacts a number of traits, as described below, imply that it has a pleiotropic effect, or does the mutation that includes Ph1 comprise several genes?

Over the years, many studies attempted to elucidate the mode of action of Ph1. The accumulated evidence falls into two main categories: (i) those showing that this gene operates during meiotic prophase, affecting processes involved in synapsis and crossing over, and (ii) those suggesting that this gene exerts its effect during premeiotic stages, affecting the premeiotic alignment of homologous and homoeologous chromosomes, thereby controlling the regularity and pattern of meiotic pairing.

Among the first theories concerning the mechanism of action of Ph1, was that of Riley (1960), who suggested that Ph1 reduces the long-range pairing forces that bring chromosomes together in meiotic prophase. Since the attraction between homoeologues can be assumed to be less than that between homologues, the reduced pairing forces were assumed to no longer bring homoeologues together, although they are still sufficient to unite homologues. This suggestion left too many questions unanswered, particularly, if chromosomes are distributed non-randomly in the premeiotic nucleus, how could pairing of homoeologues lying by chance close to one another be prevented if their homologues happened to be on the opposite side of the nucleus? Also, since pairing is believed to be initiated at chromosome ends, how can homologues coming together from different parts of the nucleus avoid interlocking with other pairs that were also finding each other from a distance?

A later suggestion by Riley (1968) was based on the possibility that Ph1 shortens the period available for the chromosomes to pair, thus, preventing the lower-affinity homoeologues from synapsing but allowing the high-affinity homologues to fully pair. But, when this idea was tested (Bennett et al. 1974), no effect of Ph1 on the duration of meiosis was found.

Upadhya and Swaminathan (1967) suggested that the absence of Ph1 causes a decrease in the speed and degree of chromosome condensation, and that this, in turn, allows homoeologous pairing to occur. However, the amount of reduction in condensation observed in the absence of Ph1 was evidently marginal. Furthermore, the suggestion that differences in condensation have an appreciable effect on homoeologous pairing is difficult to accept knowing that the substantial increase in condensation observed in the absence of chromosome 6A did not result in a decrease in homoeologous pairing.

Colas et al. (2008) showed that chromatin remodeling of homologues at the onset of meiosis in allohexaploid wheat, enabling intimate association and recombination, can only occur if the homologues are identical or nearly identical. Failure to undergo such remodeling results in reduced pairing between the homologues. In this respect, Knight et al. (2010) showed that Ph1 delays chromosome condensation at premeiotic and early meiotic stages, while treatment of premeiotic interphase of interspecific hybrids involving allohexaploid wheat and wild relatives with okadaic acid, a drug known to induce chromosome condensation, induced homoeologous pairing even in the presence of Ph1. Thus, the timing of chromosome condensation during the onset of meiosis is an important factor in controlling chromosome pairing.

When studying three-dimensional reconstructions at the ultrastructural level, of late zygotene and early pachytene of T. aestivum cv. Chinese Spring, Hobolth (1981) observed multivalent configurations, indicating pairing of homologues and homoeologues, and bivalents of strict homologous pairing at pachytene. He proposed that the regular bivalent formation in allohexaploid wheat is due to a temporal delay of crossover by the Ph1 gene until pairing correction is completed at early pachytene.

Likewise, Gillies (1987) studied synaptonemal complexes at zygotene-pachytene in spread nuclei of T. aestivum x  Ae. peregrina hybrids with and without Ph1 and concluded that this gene does not impact the ability of homoeologous chromosomes to form synaptonemal complexes, but rather, influences the rate of pairing or the time of crossover.

Holm et al. (1988), Holm and Wang (1988) and Wang and Holm (1988) studied the effect of Ph1 on chromosome pairing of homologous and homoeologous chromosomes and on synaptonemal complex formation at zygotene-pachytene stages in spread nuclei of T. aestivum cv. Chinese Spring, of aneuploid lines of this cultivar, and of hybrids with wild relatives, all with different doses of chromosome arm 5BL. They found that plants lacking 5BL had a large increase in the number of pairing partners at these early stages of meiosis, but only plants with zero or six doses of 5BL contained crossovers between homoeologues. This lies in line with the finding that plants either without 5BL or with six doses exhibit homoeologous pairing (Feldman 1966b). Holm et al. (1988), Holm and Wang (1988) and Wang and Holm (1988) concluded that Ph1 affects both synapsis and crossover.

Similarly, Martinez et al. (1996) analyzed the synaptic process at mid-zygotene, late-zygotene and pachytene in spread nuclei of T. timopheevii. Nuclei at pachytene showed a lower frequency of multivalents than did zygotene nuclei. The authors concluded that a pairing-correction mechanism at pachytene transforms quadrivalents into pairs of bivalents, possibly by the suppression of crossover between homoeologues in the synaptonemal complexes.

Martinez et al. (2001b) reported that the mean number of lateral elements involved in synaptonemal complex multivalent associations at mid-zygotene was relatively high in plants with zero, two, and four doses of Ph1. At pachytene, multivalents were transformed to bivalents. Multivalent correction was more efficient in the presence than in the absence of Ph1. These findings suggest that the main action of the Ph1 locus on the diploidization mechanism is related to a process which checks for homology during first meiotic prophase.

In ph1b mutant plants, that are deficient for Ph1, the number of ring bivalents and chiasmata decreased, while the number of univalents, rod bivalents, trivalents and quadrivalents increased (Martín et al. 2014). Consequently, these authors proposed that Ph1 has a dual effect at meiosis, namely, it promotes early synapsis between homologues and prevents sites on homoeologues from becoming crossovers.

Dubcovsky et al. (1996) and Luo et al. (1996) studied chromosome pairing between chromosome 1A of T. aestivum carrying interstitial segments of 1Am of T. monococcum, and ordinary 1A of T. aestivum. While the 1Am and 1A segments recombined very little in the presence of Ph1, in the absence of Ph1, they recombined practically as if they were homologues. Consequently, these researchers concluded that Ph1 recognizes the structural differences between homoeologous chromosomes and ensures homologous pairing in allopolyploid wheat by processing homology along the entire length of the chromosomes.

All the above hypotheses, inferring that Ph1 operates at first meiotic prophase by affecting synapsis and crossover, do not explain the mechanism of action. However, other lines of evidence support the hypothesis that this gene exerts its effect at presynaptic stages, controlling the premeiotic alignment of homologous and homoeologous chromosomes, and thereby controls the regularity and pattern of pairing at meiosis.

In an attempt to study if Ph1 also affects homologous pairing, Feldman (1966b) studied chromosome behavior at meiosis of T. aestivum cv. Chinese Spring plants carrying six doses of Ph1, as in tri-isosomic 5BL plants. The rationale behind studying plants with higher dose of Ph1 stemmed from the assumption that the normal two-dose Ph1 effect is partly counteracted by its homoeoalleles on 5AL and 5DL that promote pairing. Indeed, six doses of chromosome arm 5BL caused partial asynapsis of homologues (Feldman 1966b), i.e., it reduced homologous pairing to about one half of the normal level. Hence, higher doses of Ph1 can also act on homologous pairing. Wang (1990) also showed that Phl is not an exclusive suppressor of homoeologous pairingand can also affect the pairing of homologous chromosomes.

Apart from the partial suppression of homologous pairing, six doses of 5BL allowed some pairing of homoeologous chromosomes and induced a high frequency of interlocking bivalents (Feldman 1966b). The seemingly contradictory effect of an extra dose of Phl, namely, partial suppression of homologous pairing on the one hand and induction of homoeologous pairing and interlocking of bivalents on the other hand, suggests that Phl does not simply suppress synapsis at meiotic prophase. In light of these phenomena, it has been proposed (Feldman 1966b) that Phl affects the presynaptic alignment of both homologous and homoeologous chromosomes.

A phenocopy of the effects of six doses of Ph1 on chromosomal pairing was observed following pre-meiotic treatment of T. aestivum with colchicine (Driscoll et al. 1967; Yacobi et al.1982; Feldman and Avivi 1988). Such treatments induced partial asynapsis of homologues and pairing of homoeologues in allohexaploid wheat (Driscoll et al. 1967). Induction of interlocking of bivalents by premeiotic treatment with colchicine, in addition to partial asynapsis of homologues and homoeologous pairing, was observed in allotetraploid Ae. kotschyi (M. Feldman, unpublished).

A role for premeiotic interphase chromosome associations in homologous recognition was also reported in bread wheat by other researchers (Aragón-Alcaide et al. 1997a, b; Schwarzacher 1997; Mikhailova et al. 1998; Martínez-Pérez et al. 1999) and in Saccharomyces cerevisiae (Loidl 1990; Weiner and Kleckner 1994). It has been argued that such associations lead directly to meiotic homologue pairing during first prophase (Kleckner 1996). However, the different studies disagree with regards to the extent and role of premeiotic chromosome association, where they start and how long they last (e.g., see Schwarzacher 1997; Mikhailova et al. 1998; Martínez-Pérez et al. 1999). Chromosome arrangement in interphase, somatic as well as premeiotic, nuclei, is presumably accomplished through distribution of centromeres and telomeres during each telophase, at opposite sides of the nuclei into a Rabl configuration (Fussell 1987). This organization eases the homolog search and the subsequent alignment (Pernickova et al. 2019).

Assuming premeiotic chromosome alignment in T. aestivum, the effect of different doses of Phl can be clearly explained. A model was thus proposed whereby Ph1 exerts its effect at the end of each cell division, including in the last pre-meiotic mitosis, where it affects the alignment of homologous and homoeologous chromosomes in telophase, and consequently in interphase, and so shapes the pattern of synapsis commencement and, as a result, controls the regularity and pattern of chromosomal pairing (Feldman 1993). Accordingly, in the absence of Ph1, i.e., in nullisomic 5B plants of cv. Chinese Spring of T. aestivum or in the ph1b mutant, the three subgenomes coexist in the nucleus and consequently, homologues as well as homoeologues would be closely associated at premeiotic stages, albeit the latter to a lesser extent. This results in reduced pairing of homologues whose pairing initiation was interrupted by the presence of homoeologues. The reduced pairing of homologues is expressed by few univalents and increased numbers of rod bivalents, alongside low frequency of multivalent homoeologous pairing. Mingling of homologues and homoeologues may lead to interlocking of bivalents, mainly of homoeologous bivalents (Feldman 1966b; Yacobi et al. 1982). In contrast, in euploid T. aestivum, which carries the normal two doses of Ph1, the association of homologous chromosomes is barely affected, while the somatic and premeiotic association of homoeologues is suppressed to the extent that they no longer lie together, and therefore, are not able to pair at meiosis or to pair somewhat less intimately. With six doses of Ph1, or pre-meiotic treatment with colchicine, even homologues no longer associate somatically, resulting in more or less random distribution of all the chromosomes. Then, at meiosis, homologues still pair, provided they do not lie too far apart, but in coming together from a distance and beginning pairing at their ends, they frequently intermingle with other bivalents to form interlocking bivalents. Homoeologues that lie close enough to each other also pair, if neither has a homologue close enough to generate greater attraction. In tri-isosomic 5BL plants, multivalent associations at first meiotic metaphase were rare because of the general reduction of pairing, while heteromorphic rod bivalents resulting from homoeologous pairing were more frequent (Feldman 1966b). Finally, about 20% of the ring bivalents present were interlocked with one, two or more (up to seven) other bivalents (Feldman 1966b; Yacobi et al. 1982). The broad occurrence of interlocking bivalents, in spite of the reduction in the total number of bivalents, showed clearly that some of the pairing occurred between somewhat separated partners.

The pairing behavior of two or three homologous isochromosomes can distinguish between factors affecting premeiotic alignment and those affecting synapsis and crossover. An isochromosome, consisting of two homologous arms can undergo either intra-chromosomal pairing between the two homologous arms of the same isochromosome, to form a ring univalent at first meiotic metaphase, or inter-chromosome pairing in cells having two or three homologous isochromosomes, to form a bivalent or a trivalent. Factors that disrupt homologous alignment would reduce the frequency of inter-chromosome pairing without affecting intra-chromosome pairing because the two homologous arms of an isochromosome are connected by a common centromere and their relative position remains undisrupted. On the other hand, factors that prevent synapsis or crossover would affect both types of pairing.

In T. aestivum, Sears (1952a) and Driscoll and Darvey (1970) observed almost complete intra-chromosome pairing in a univalent isochromosome at a frequency similar to that of pairing between homologous arms of conventional chromosomes. Application of colchicine during the last premeiotic interphase resulted in pairing failure of conventional homologues at the first meiotic metaphase (Driscoll et al. 1967; Dover and Riley 1973), but did not affect pairing between the two arms of an isochromosome (Driscoll and Darvey 1970). Hence, these authors concluded that colchicine inhibits premeiotic association of homologues rather than their synapsis and crossover. Similarly, high-temperature treatment during the last premeiotic interphase, which considerably reduced pairing of conventional homologous chromosomes, did not interfere with synapsis and chiasma formation between the two arms of an isochromosome (Kato and Yamagata 1980). In contrast to the effect of colchicine and high temperature, absence of wheat chromosome 3B, known to possess a recessive gene whose activity is responsible for normal synapsis and chiasma-formation (Li et al. 1945; Sears 1944, 1954; Kempanna and Riley 1962; Kato and Yamagata 1982, 1983), reduced homologous pairing not only between the arms of conventional chromosomes (Sears 1954) but also the intra-chromosomal pairing between the two arms of an isochromosome (Kato and Yamagata 1982). This suggests that the pairing gene located on 3B controls either synaptic or postsynaptic events.

To determine which of the processes involved in chromosome pairing is affected by Ph1, Feldman and Avivi (1988) and Vega and Feldman (1998a) studied the effect of different doses of Ph1 on the pairing of two or three isochromosomes and of an isochromosome with a telocentric-chromosome homologous to the isochromosome. These researchers showed that in tri-isosomic 5BL plants, the six doses of Ph1 suppressed the inter-chromosomal pairing of the three isochromosomes to the same extent as it affected conventional pairing, without reducing the intra-chromosomal pairing. Under these conditions, the extra doses of Ph1 could not modify the premeiotic alignment of the arms that were connected to one another by a common centromere. Hence, the failure of inter-chromosomal pairing as well as that of conventional chromosome pairing reflected disturbances in the premeiotic alignment. Thus, Ph1 does suppress presynaptic homologous alignment. The similar outcome of premeiotic colchicine treatment and Ph1 on the pattern of isochromosome pairing supports the hypothesis claiming the premeiotic association of homologues (Feldman 1966b; Feldman and Avivi 1988; Vega and Feldman 1998a).

Similar to the effect of Ph1, premeiotic colchicine treatment, which drastically decreased pairing of conventional chromosomes, reduced inter-chromosome but not intra-chromosome pairing of isochromosomes (Driscoll and Darvey 1970; Dover and Riley 1973; Feldman and Avivi 1988; Vega and Feldman 1998a). Based on these genetic and chemical effects, Feldman and Avivi (1988) reasoned that the effect of Ph1 is exerted before the onset of meiosis. They reviewed data showing that in somatic cells of many plant species, the chromosomes are not randomly arranged with respect to each other, but rather, found that homologues are already lying side by side and therefore do not have to find each other from a distance at the beginning of meiosis (Avivi and Feldman 1980). In other words, the hexaploid nucleus still maintains some organizational aspects of the individual subgenomes, i.e., each subgenome occupies a separate region in the nucleus (Avivi et al. 1982b; Feldman 1993) (Also see Chap. 10, Sect. 4.2.8 on relationships between chromosomes of the different subgenomes and between homologues and non-homologues of the same subgenome).

Moreover, application of colchicine to spikes of normal hexaploid wheat before the last premeiotic mitosis (8–9 days before meiosis) resulted in a doubling of the chromosome number, and in bivalent pairing of nearly all the 84 chromosomes during the subsequent meiosis, even though each homologue exists in four doses (Driscoll et al. 1967; Dover and Riley 1973). It is assumed that the paired chromosomes were sister chromatids in the last mitosis and because of the colchicine treatment, their centromeres remained attached to each other until sometime in interphase, and thus, they remained very close to each other, while the homologous partners were randomly distributed. Likewise, in F1 hybrids between allohexaploid wheat and related species, where without treatment, only homoeologous pairing occurs, colchicine treatment before the last premeiotic mitosis induced chromosome doubling, leading to regular bivalent formation, with no detectable homoeologous pairing (Dover and Riley 1973). These results indicate that premeiotic colchicine treatments, like extra dosage of Ph1, do not alter the processes of synapsis or crossover, but, rather, the premeiotic alignment of homologous chromosomes, which appears to be a prerequisite for meiotic pairing regularity.

Aragón-Alcaide et al. (1997a), using FISH with centromeric and telomeric sequences as probes, investigated centromeric behavior in PMCs of allohexaploid wheat and wheat x alien hybrids carrying different combinations of pairing genes. Their study revealed that centromeres are associated in pairs in pre-meiotic interphase, irrespective of the presence or absence of the homoeologous-pairing suppressors, Ph1 and Ph2. Moreover, they found a difference in centromeric structure in pre-meiotic interphase, pachytene, first meiotic metaphase and anaphase plants carrying versus missing Ph1 and Ph2. In plants lacking Ph1 or Ph2, the centromeres exhibited diffuse hybridization sites during pre-meiotic interphase, and from pachytene through anaphase, whereas in plants carrying Ph1 and Ph2, the discrete hybridization sites at premeiotic interphase remained as dense sites at these meiotic stages. The authors suggested that after replication of centromeres in pre-meiotic interphase, centromeres in the presence of Ph1 and Ph2 can form a more condensed structure throughout first meiotic division as compared to plants lacking the homoeologous-pairing suppressor genes. A diffuse structure may result in increased exposure of the centromere during the pairing process, which could increase interactions at these sites via proteins or via DNA sequences. The increased accessibility could expose regions of the centromere that are more conserved in structure, thus reducing the stringency of pairing at the centromere. Similar mechanisms may also impact other chromosomal sites involved in pairing.

To examine the occurrence of homologous association in somatic cells, Feldman et al. (1966) took advantage of the availability of telocentric chromosomes in hexaploid wheat, which are identifiable in somatic metaphases. Chinese Spring plants having two doses of Ph1 with two telocentric chromosomes, either homologous or non-homologous, were produced. In root tip cells, the two non-homologous telocentrics were found to be located at random with respect to each other, while the two homologous telocentrics were significantly closer together than expected on a random basis. Telocentrics for the opposite arms of the same chromosome were also found to be associated (Feldman et al. 1966; Mello-Sampayo 1973). Thus, it was concluded that the centromere, the DNA sequences shared by the two different arms of the same chromosome, determines the position of each chromosome in the nucleus.

In nulli-5B plants (lacking Ph1), telocentric chromosomes of the different subgenomes lay as close together as telocentrics of the same subgenome, and thus, homoeologues were close to each other as homologues were (Feldman and Avivi 1984).

In disomic 5B plants, carrying two doses of Ph1, chromosomes of the same subgenome lay closer to each other than chromosomes of the different subgenomes (Feldman and Avivi 1973, 1984; Avivi et al. 1982a, b). Thus, the location of homologues and homoeologues with respect to each other, at different levels of Ph1, conformed with expectations based on Feldman’s (1966b) hypothesis.

The closer somatic association of unrelated chromosomes of the same subgenome as compared to chromosomes of different subgenomes, is interpreted as being the result of a tendency for the chromosomes to have fixed positions within the nucleus and for the chromosomes of each subgenome to be grouped together and occupy different region in the nucleus. Spatial segregation of parental genomes in somatic and meiotic metaphases of various plant hybrids has been reported by several researchers (Finch et al. 1981; Schwarzacher et al. 1989, 1992; Linde-Laursen and Jensen 1991). Others have made such observations in the wild grass allotetraploid Milium montianum (Bennett and Bennett 1992) and in allotetraploid cotton (Han et al. 2015). In allopolyploid plants, there appears to be subgenome separation of interphase chromosomes, with chromosomes of each subgenome tending to cluster together (Avivi et al. 1982b; Feldman and Avivi 1984; Hilliker and Appels 1989). A similar genome separation phenomenon was found in synthesized tetraploid cotton (genome AAGG) (Han et al. 2015). Given the evidence of parental genome separation in other plants, Han et al. (2015) speculated that genome separation might be a normal phenomenon in diploid hybrids and in allopolyploid species. Concia et al. (2020) analyzed the entire genome interaction matrix in allohexaploid wheat and revealed three hierarchical layers of chromosome interactions, presented here from strongest to lowest: (i) within chromosomes, (ii) between chromosomes of the same subgenome, and (iii) between chromosomes of different subgenomes. This organization may indicate a non-random spatial distribution of the three subgenomes that could mirror the presence of functional “genome territories.”

Concia et al. (2020) confirmed the presence of subgenome-specific nuclear territories using genomic in situ hybridization (GISH) in root meristematic cells of hexaploid wheat. To determine whether other polyploid plants share the same large-scale nuclear organization, Concia et al. (2020) analyzed 14-day-old rapeseed seedlings (Brassica napus), and again revealed a three-layer hierarchy of chromosomal interactions identical to those in allopolyploid wheat, suggesting that this organization is a general feature of allopolyploid plants. They also showed, using two genome-wide complementary techniques, GISH and in situ Hi-C, that the chromatin of hexaploid wheat is not uniformly distributed across the nucleus but, rather, occupies subgenome-specific nuclear compartments. This finding is consistent with previous cytological observations (Feldman and Avivi 1973; Avivi et al. 1982a, b), indicating that chromosomes of the same subgenome tend to be physically closer than chromosomes of different subgenomes. Consequently, they proposed that genome territories are the primary level of chromatin spatial organization in allohexaploid wheat. Little is known about the mechanisms facilitating homologous pairing versus homoeologous pairing during the telomere bouquet stage at the start of meiosis. The establishment of genome territories may be a mechanism that favors the pairing of homologues versus homoeologues by creating territorial “boundaries” between the different subgenomes, i.e., either homoeologues or non-homologues (Concia et al. (2020).

Similarly, Jia et al. (2021) probed the three-dimensional chromatin architecture of a Chinese cultivar of T. aestivum and found that the three subgenomes occupy specific territories in the nucleus. This is in accord with previous studies revealing that the three subgenomes of hexaploid wheat tend to localize to specific nuclear territories (Avivi et al. 1982b; Li et al. 2000; Concia et al. 2020). Moreover, the data of Jia et al. (2021) suggested that transposable elements help promote the higher order subgenome affinity in allohexaploid wheat. Bhat et al. (2021) in a recent review, propose that DNA, RNA and proteins are organized within precise 3D compartments in the nucleus, affecting many aspects of gene regulation. Non-coding RNA contribute to such intra-nuclear organization.

Driscoll and Darvey (1970), Darvey and Driscoll (1971, 1972), and Darvey et al. (1973) were unable to confirm that homologues lie closer than homoeologues in root-tip nuclei of common wheat. Both homologous and non-homologous telocentrics showed the same distribution, and homologous nucleoli showed a similar tendency to fuse than did non-homologous nucleoli. The tendency of proximal nucleoli to fuse, was established by demonstrating that the number of nucleoli per nucleus decreases substantially during interphase (Crosby 1957).

Moreover, Driscoll et al. (1979) estimated the probability of premeiotic association of homologous chromosomes and of chiasma formation, from frequencies of different chromosome configurations at first meiotic metaphase of allohexaploid wheat, several aneuploid lines, and wheat hybrids. Based on their estimations, they suggested that 5BL did not appear to affect the premeiotic association of chromosomes, but rather, the number of chiasmata.

Despite the evidence that presynaptic alignment of homologues ensures the regularity of pairing, there is little agreement regarding the timing of the first alignment of homologous chromosomes (reviewed by Loidl 1990). While some assume that homologues are already associated at the last premeiotic interphase (e.g., Smith 1942; Feldman 1966b; Maguire 1967), others hold that homologues do not associate before the beginning of zygotene (e.g., John 1976; Rasmussen and Holm 1978). Although there have been indications in a number of organisms that premeiotic alignment is a characteristic feature of meiosis (Avivi and Feldman 1980), it was difficult to conclusively demonstrate this phenomenon because individual chromosomes could not be clearly distinguished. In several species, this problem has recently been circumvented by fluorescence in situ hybridization with DNA probes that detect a specific pair of homologous chromosomes or chromosome segments. In the budding yeast Saccharomyces cerevisiae, the homologues were found to be associated via multiple interstitial interactions during the last premeiotic interphase (Weiner and Kleckner 1994). Genomic in situ hybridization in a wheat line carrying a pair of homologues originating from barley, showed that the hybridization signals of the two barley homologues fused into a single fluorescent signal during the last premeiotic interphase, indicating their complete association (Aragón-Alcaide et al. 1997b). Moreover, Aragón-Alcaide et al. (1997a) observed that in the absence of Ph1, the barley homologues were not in contact along their length and proposed that the absence of Ph1 disrupts premeiotic homologue association. Premeiotic association was also observed in a pair of homologous rye telocentrics added to common wheat (E. I. Mikhailova, T. Naranjo, K. Shepherd, J. Wennekes, C. Heyting and J. H. de Jong, unpublished results).

Martinez-Perez et al. (2001) observed that Ph1 also acts somatically by reducing non-homologous centromere associations. This effect during premeiotic interphase leads to exclusive homologue association during the telomere-bouquet stage in meiosis. The authors proposed that non-homologously associated centromeres separate at the beginning of meiosis in the presence, but not in the absence, of Ph1 and concluded that Ph1 is not responsible for the induction of centromere association, but, rather, regulates its specificity. Likewise, Moore (2002) proposed that Ph1 determines correct pairing of homologous chromosomes at premeiotic and early meiosis prophase by giving nonhomologous chromosomes an almost ‘Teflon’-like status, that increases chromosome specificity in the pairing process. In the absence of Ph1, the homoeologous chromosomes might be able to pair as a result of the loss of such a coating. All these findings in yeast and wheat demonstrate that homologous chromosomes recognize each other and associate before meiosis, a process that leads to exclusive synapsis of homologues at first meiotic prophase.

Of importance to the somatic-association hypothesis (Feldman 1966b) is how much, if any, association there is during the last premeiotic interphase. Following the report by Stack and Brown, somatic pairing is expected in premeiotic cells, even though it did not occur in root tips. Walters (1970, 1972) found no evidence of somatic pairing in Lilium longiflorum at any premeiotic stage, but it seems doubtful that the loose and intermittent association expected during pre-meiosis and early meiotic prophase would be detectable cytologically in a plant with such large chromosomes. Indeed, Walters’ excellent photographs of leptotene nuclei, in which the chromosomes are assembled into what Brown and Stack (1968) would call a ball of yarn, raise the question of how homologous strands located in different parts of the nucleus could possibly thread their way through the mass of other strands and align themselves precisely alongside each other in zygotene, with no entanglement, and all in a relatively short time. Scherthan et al. (1996) found no associated homologues until early meiotic prophase in mouse and humans, but failed to compare the distance between the hybridization signals of homologues with that between non-homologues, and, therefore, the results from premeiotic stages are inconclusive.

It can be concluded that the controversy concerning the occurrence of homologous association in root-tip cells should be resolved by further work, but what is really needed is a careful analysis of the last premeiotic mitosis and premeiotic interphase. Unfortunately, this division is not as easy to study as are root-tip mitoses. Unless lack of association in the premeiotic mitosis can be clearly shown, there is little choice but to accept the premeiotic association hypothesis, because it so simply explains several otherwise puzzling phenomena.

Avivi et al. (1972) argue that somatic association has a physiological advantage and therefore occurs throughout the life of the plant. They found a different pattern of activity of the triplicate series of alcohol dehydrogenase genes in plants lacking 5BL than in plants carrying this chromosome arm. This observation needs to be extended to other loci, especially since analysis of isozymes of glutamate oxaloacetate transaminase indicated that even in the presence of Ph1, monomers from genes on homoeologous chromosomes combine with each other as readily as do the monomers from homologues (Hart et al. 1976). Yet, DNA organization in precise 3D compartments in the nucleus is one of prerequisite conditions for regular gene action (Bhat et al. 2021).

Considering the phenocopy of the effect of six doses of Ph1 on meiotic pairing by premeiotic treatments with colchicine (Driscoll et al. 1967; Dover and Riley 1973; Yacobi et al. 1982; Feldman and Avivi 1988), and since colchicine binds specifically to tubulin subunits, thereby preventing them from polymerizing into microtubules (Borisy and Taylor 1967), it was concluded that the microtubule system is one of the subcellular targets of Ph1 (Avivi and Feldman 1973a; Feldman 1993; Feldman and Avivi 1988) and that microtubules are involved in the process of intimate homologous association. This may involve attachment of centromeres, telomeres and other chromosome segments to the nuclear envelope at the end of each telophase, thereby stabilizing chromosome position in the nucleus throughout premeiotic interphase. The disruption of microtubules by colchicine would detach the chromosomes from the nuclear membrane, leading to their movement at interphase and disruption of the association between homologues. In agreement with this view, Vega and Feldman (1998b) showed that Ph1 affects centromere-microtubule interactions at meiotic anaphases.

Spindle inhibitors, particularly in combination with different Ph1 genotypes, have been used to help elucidate the mechanisms of Ph1 actions. Colchicine treatment of root-tip cells disrupted association of homologues (Avivi et al. 1969). The sensitivity of root-tip mitosis to colchicine and other microtubules drugs, such as vinblastine and griseofulvin, decreased with increased doses (up to 4) of 5BL (Avivi et al. 1973b; Ceoloni et al. 1984; Gualandi et al. 1984). This led to the conclusion that Ph1 somehow affects the binding of microtubule subunits to colchicine, thereby protecting the spindle against colchicine action. The differential sensitivity of different genotypes of Phl to colchicine (Avivi et al. 1970, 1973b; Ceoloni et al. 1984) is not confined to hexaploid wheat. Carla Ceoloni (unpublished data) found that the mitotic spindle of a phlc mutant induced in the Italian durum cultivar Cappelli (Giorgi 1983), is much more sensitive to colchicine than that of plants with Ph1.

It has been reported that detyrosination or acetylation of a-tubulin reduces microtubule dynamics (Webster and Borisy 1989). Such microtubules are more stable and less sensitive to depolymerization by antimicrotubule drugs (Piperno et al. 1987; Kreis 1987; Khawaja et al. 1988). Most microtubule-associated proteins (MAPs), e.g., tau, MAP2 and MAP1, stimulate the assembly of tubulin into microtubules (see Olmsted 1986 for a review) and, consequently, increase the stability of the polymers and suppress microtubule dynamics (Murphy et al. 1977). Post-translational modifications of MAPs, mainly phosphorylation, may be involved in reduction of microtubule dynamics. Interaction of microtubules with each other or other cytoskeletal elements (mainly intermediate filaments) may also increase microtubule stability (Gelfand and Bershadsky 1991). In addition, microtubule stability is affected by kinetochores, which are highly differentiated structures at the centromeres that serve as microtubule attachment sites, and which cap the plus-ends of microtubules (McIntosh and Hering 1991). All these interactions are presumably mediated by MAPs. Hence, phosphorylation-dephosphorylation of these MAPs may be the main mechanism underlying the regulation of such interactions (Gelfand and Bershadsky 1991).

How is chromosomal arrangement in the telophase and interphase nuclei affected by the reduced microtubule dynamics induced by Ph1? According to the presynaptic hypothesis, Ph1 controls chromosome arrangement in the somatic as well as premeiotic nucleus, by operating on the subcellular elements that are involved in chromosome positioning: microtubules and centromeres. Vega and Feldman (1998b) assume that Ph1 action may target the interaction of centromeres with spindle microtubules—an interaction that is critical for the movement of chromosomes to their specific interphase positions. Consequently, Vega and Feldman (1998b) studied centromere behavior of univalents at meiosis of monosomic lines in the presence and absence of Ph1 and found that the frequency of centromere misdivision (transverse division) of univalent chromosomes is affected by Ph1. In common wheat, the centromere of unpaired chromosomes may undergo precocious division at first anaphase or telophase (Sears 1952b). This division is either longitudinal, leading to the formation of two sister chromosomes, each consisting of one chromatid, or transverse (misdivision), leading to the formation of telocentric chromosomes and isochromosomes. Transverse division of the centromere of one-chromatid chromosomes may also occur at second anaphase. In the presence of Ph1, the frequency of centromere misdivision in both first and second meiotic divisions was much higher than in the absence of the gene (Vega and Feldman 1998b), suggesting a role for Ph1 in the interaction between kinetochores and microtubules at anaphase.

The ph1b deletion may contain a number of genes that might affect centromere-microtubule interaction. However, the fact that premeiotic treatments with colchicine and other antimicrotubule drugs phenocopy the effect of extra doses of Ph1 on chromosome pairing (Feldman and Avivi 1988) indicates that the effect on pairing and on centromere microtubule interaction is caused by Ph1.

Several lines of evidence support the involvement of the centromere in chromosomal arrangement in wheat, in somatic (Feldman et al. 1966; Mello-Sampayo 1973) and meiotic (Yacobi et al. 1985a, b) cells. There is also direct evidence of nonrandom distribution of centromeres in the interphase nucleus. Using anti-kinetochore antibodies in interphase nuclei of rat-kangaroo and Indian muntjac, Hadlaczky et al. (1986) observed that centromeres were arranged in pairs. In a similar experiment, half the expected number of pre-kinetochores were detected at interphase in Vicia faba (Houben et al. 1995). These observations indicate at least transient association of homologous centromeres at interphase. Su et al. (2019) suggested that variation in centromere satellite sequences and copy number, and their structural rearrangements, result in asymmetries in allohexaploid wheat homoeologues, highlighting the role of centromeres in homolog pairing during meiosis. This asymmetry in centromere organization among the three subgenomes of allohexaploid wheat, was suggested to play a role in proper homologous pairing during meiosis.

Taken together, Ph1 seems to affect several different traits in somatic, premeiotic and meiotic cells of allohexaploid wheat. These traits are: spindle sensitivity to anti-microtubules drugs, somatic association of homologues, separation of the subgenomes into different nuclear compartments, concise centromeres at premeiotic and meiotic stages, chromatin condensation, synapsis of homologues and homoeologues at zygotene, correction of pairing to homologues at pachytene, prevention of crossover between homoeologues in pachytene, and strength of centromere attachment to microtubules. Ph1 also affects the number and shape of pollen grain pores in common wheat (Avivi and Feldman 1973b). This gene has a pronounced effect on the condensation of heterochromatin in hybrids of wheat with related species and on the condensation of euchromatin in wheat itself (Martinez-Perez et al. 2001; Prieto et al. 2005). Hence, Ph1 influences a number of phenotypic traits and, as such, may be either a gene with a pleiotropic effect (a transcription factor?) or as a complicated locus comprising a cluster of genes.

12.2.2.3 Isolation of Ph1

It is generally accepted that Ph1 represents a single locus, but the possibility that it consists of a cluster of tandemly arranged genes, has not been completely ruled out (Sears 1976b). The existence of two pairing loci on 5BL was already suggested by Mello-Sampayo (1972), who noted that the F1 hybrids T. turgidum x Ae. sharonensis and T. turgidum x S. cereale, in which the distal part of 5BL had been replaced by an homoeologous portion of 5DL, had an intermediate, rather than high, level of pairing. Consequently, he concluded that Ph1 was still present but that its effectiveness was reduced by the presence of the 5DL segment, that might carry a pairing promoter. The 5BL homoeoallele of this 5DL promoter could be a low-grade promoter, or, alternatively, a low-grade suppressor.

Some evidence against the assumption that Ph1 contains more than one locus is the failure to obtain intermediate mutants on chromosome arm 5BL by irradiation (Sears 1977). On the other hand, the inability to induce mutations in Ph1 exhibiting the ph1b-like phenotype by EMS treatment, may indicate that its phenotype is determined by more than one gene (Griffiths et al. 2006).

The fact that Ph1 activity is unique to chromosome 5B, and not to its homoeologues, led Griffiths et al. (2006) to suggest that this gene arose through a post-allopolyploidization structural change on chromosome 5B. Support for this conclusion is provided by studies showing that Ae. speltoides and Ae. longissima, two diploids that are closely related to the B and G subgenomes of allotetraploid wheats, do not compensate for the absence of Ph1, while the B and G subgenomes of these allotetraploids do (Griffiths et al. 2006). Since Ph1 has a pronounced effect on several traits, Griffiths et al. (2006) concluded that Ph1 is likely to comprise a multigene family, heterochromatin or both. These authors used two approaches to analyze the genetic structure of the Ph1 region. First, they used genes of the orthologous regions in rice and Brachypodium sylvaticum to obtain markers for the saturation of the Ph1 region, and second, they used five deletions produced by Roberts et al. (1999), using fast-neutron irradiation, that overlapped the ph1b deletion, to physically dissect the Ph1 region. These approaches enabled the narrowing down of the Ph1 locus to a 2.5 Mb region on 5BL, which contains a chromosomal segment derived after allotetraploidization from the sub-telomeric region of chromosome arm 3BL of bread wheat, comprising a block of heterochromatin and a single Zip4 gene, inserted within a cluster of defective cyclin-dependent kinase-(cdk) like genes, between Cdk6 and 7 (Griffith et al. 2006; Al-kaff et al. 2008; Martín et al. 2014, 2017). Griffiths et al. (2006) assumed that this region fulfills all criteria of a Ph1 candidate structure. On the basis of their sequence homology to genes of known function, Griffiths et al. (2006) singled out the cdk-like genes of the Ph1 region as the best candidates for Ph1 function. This is the only multigene cluster in the region, and at least one of its members (cdk2-4) is 5BL-specific. Similarly, Al-Kaff et al. (2008), who further characterized the Ph1 region by exploiting new deletions in this region and conducting expression analysis studies, assigned the Ph1 locus to the region containing the cdk-like cluster and the neighboring heterochromatin segment. In fact, there are seven cdk-like genes on 5BL compared with at least five on 5AL and two on 5DL, however, the sub-telomeric heterochromatin segment inserted between the two cdk-like genes, cdk-like B6 and B7, is unique to chromosome arm 5B (Al-Kaff et al. 2008). The cdk-like gene cluster in the Ph1 locus, designated Ta5B2, shows some similarity to mammalian Cdk2 (Yousafzai et al. 2010). Greer et al. (2012) showed that Cdk2-type phosphorylation plays a major role in determining chromosome specificity during meiosis. Deletion of Ph1 leads to increased phosphorylation at cdk2-type sites during meiosis, implying that the presence of Ph1 decreases Cdk2-type phosphorylation. Consistent with this, treatment with okadaic acid, an inhibitor of phosphatase activity, increases Cdk2-type phosphorylation, and phenocopies the deletion of Ph1 by inducing crossovers (COs) (Knight et al. 2010). Deleting Ph1 or treating with okadaic acid, both of which increase Cdk2-type activity, increases the efficiency of MLH1, a gene involved in DNA mismatch repair, whose active sites on paired homoeologues may lead to crossing overs (Martín et al. 2014). These authors proposed that Ph1 has a dual effect in wheat, namely, it brings about cytological diploidization by both promoting homologous synapsis during early meiosis and by preventing MLH1 sites on synapsed homoeologues from becoming COs later in meiosis. The effect on synapsis occurs during the telomere bouquet stage, when Ph1 promotes homologous synapsis, thereby reducing the chance of homoeologous synapsis (Martín et al. 2014, 2017). The effect on crossing over formation occurs later in meiosis, when Ph1 prevents MLH1 activity.

Later, Martín et al. (2017) explored these two effects and demonstrated that regardless of the presence or absence of Ph1, synapsis between homoeologues does not take place during the telomere bouquet stage; only homologous synapsis takes place during this stage. Furthermore, in wheat lacking Ph1, overall synapsis was delayed with respect to the telomere bouquet, with more synapsis occurring after the bouquet stage, when homoeologous synapsis is also possible. Secondly, they showed that in the absence of Ph1, it was possible to increase the number of MLH1 sites progressing to COs by altering environmental growing condition. They also showed that higher nutrient levels in the soil or lower temperatures increased the level of both homologous and homoeologous COs.

It has been proposed that the effect of Ph1 on synapsis is connected to altered histone H1 CDK2-dependent phosphorylation. Altered phosphorylation in the absence of Ph1 was shown to affect chromatin structure and delay premeiotic replication and consequently, homologue synapsis, thus allowing homoeologous synapsis to take place (Greer et al. 2012; Martín et al. 2017). Arabidopsis lines carrying mutations in the CDK2-like homologue also exhibited reduced synapsis under specific conditions, suggesting a role for these genes in efficient synapsis (Zheng et al. 2014). Greer et al. (2012) previously proposed that the effect of CDK2-like genes on chromatin structure not only affects synapsis but might also affect the resolution of double Holliday Junctions (marked by MLH1) as COs. Okadaic acid treatment affects chromatin structure and can induce homoeologous CO in wheat-wild relative hybrids (Knight et al. 2010). However, given that the locus contains multiple copies of the CDK2-like and methyltransferase genes, it would be complex and laborious to identify EMS-induced mutants within these genes.

In contrast to Griffiths et al. (2006) and Martín et al. (2017) proposed that the ZIP4 gene, located distally to the heterochromatin in the inserted segment on 5BL, is a more appropriate candidate for Ph1 effects than the cdk-like genes. This is largely based on the evidence that the Zip1 protein is a major building block of the synaptonemal complex (SC) in Saccharomyces cerevisiae (Mitra and Roeder 2007), and, in its absence, SC fails to form, cells arrest or delay in meiotic prophase, and crossover is reduced.

Further studies on the effect of Zip4 (designated TaZIP4-B2) on synapsis and crossover were performed by Rey et al. (2017). Although, there are ZIP4 homologues on group 3 chromosomes, TaZIP4-A1 in 3A, TaZIP4-B1 in 3B, and TaZIP4-D1 in 3D, the ZIP4 paralogue (TaZIP4-B2) within the Ph1 locus on chromosome arm 5BL is a single copy. The TaZIP4-B2 gene is expressed during meiosis, has a higher level of expression than the ZIP4 homologues present on group 3 chromosomes, and its expression is significantly reduced upon Ph1 deletion (Rey et al. 2017). Moreover, ZIP4 has been shown to have a major effect on homologous COs, but not on synapsis, in both Arabidopsis and rice (Chelysheva et al. 2007; Shen et al. 2012).

Rey et al. (2017) searched for EMS-induced mutations in the TaZIP4-B2 gene to determine whether they show reduced homologous CO with some homoeologous pairing and CO but exhibit homoeologous COs in hybrids with wild-relatives. For the crossings with wild relatives, the group used two mutant lines in the TaZIP4-B2 gene, selected from the mutants obtained by Rakszegi et al. (2010) in T. aestivum cv. Cadenza. Hybrids between the Tazip4-B2 mutant Cadenza lines and Ae. peregrina, exhibited similar chiasma frequency to that observed in the ph1b mutant x Ae. peregrina hybrids, suggesting that TaZIP4-B2 within the Ph1 locus is involved in the suppression of homoeologous COs.

Since no multivalents and no significant increase in the number of univalents were observed at first meiotic metaphase of the Tazip4-B2 mutant lines, it seems that homoeologous synapsis may not be significantly affected by TaZIP4-B2 (Rey et al. 2017). Hence, TaZIP4-B2 affects only one trait out of the array of phenotypic traits controlled by Ph1, indicating that this is only one gene in the Ph1 gene cluster.

The ph1b mutant accumulates inter-subgenomic translocations due to homoeologous recombination, which reduces fertility (Sears 1977; Sánchez-Morán et al. 2001). It would therefore be most useful to use the TaZIP4-B2 mutant lines, with reduced homoeologous synapsis and CO at meiosis, but which do exhibit homoeologous COs in hybrids with wild relatives. The absence of multivalents at meiosis in the Tazip4-B2 Cadenza mutants shows that these mutants do not cause translocations between homoeologues and can be used instead of the ph1b mutant.

Sidhu et al. (2008), performing a detailed analysis of the 91 putative genes present within the 450-kb region on the rice R9 chromosome, which is orthologous to wheat chromosome arm 5BL, identified 26 candidates for the Ph1 gene, including genes involved in chromatin reorganization, microtubule attachment, acetyltransferases, methyltransferases, DNA binding, and meiosis/anther-specific proteins. Four of these genes share domains/motifs with the meiosis-specific genes Cor1, Scp1, Zip1, and RAD50. Cor1 codes for a protein of the axial element of the synaptonemal complex and Scp1 from mammals and Zip1 from yeast code for the transverse filaments that synapse the axial elements into the synaptonemal complex. The RAD50 gene is required for the induction and processing of double-strand breaks and therefore, in a null mutant of this gene, crossovers cannot be formed.

Bhullar et al. (2014) identified a candidate Ph1 gene (designated C-Ph1) in the Ph1 region, whose silencing resulted in a phenotype which was, to a certain degree, characteristic of the ph1b and ph1c mutants, namely, an increased number of univalents, multivalent pairing, and interlocking bivalents. However, it also disrupted chromosome alignment on the first meiotic metaphase plate. Despite a highly conserved DNA sequence, the C-Ph1 gene homoeologues on 5AL and 5DL exhibit a different structure and expression pattern, further supporting the claim that C-Ph1 is indeed the candidate for the Ph1 gene. Yet, the fact that C-Ph1 is mostly expressed during first meiotic metaphase, rather than during premeiotic interphase and first meiotic prophase, sheds some doubt on this claim.

The suggestion that C-Ph1 is the Ph1 gene was mainly based on the observation of bivalent clumping at first meiotic metaphase of virus-induced gene silencing (VIGS) mutants of T. aestivum. Rey et al. (2017) drew attention to previous studies that had already shown that bread wheat contains a gene, termed Raftin1, with a phenotype similar to that of C-Ph1. This gene was characterized as a tapetal cell gene, whose maximal expression occurs around the first meiotic metaphase, when the tapetum is fully formed (Wang et al. 2003). Disruption of this gene results in chromosome clumping at first meiotic metaphase and consequently, high male sterility. Therefore, Rey et al. (2017) suggested that C-Ph1 is not a Ph1 candidate.

Yet, C-Ph1 is not the Raftin1 tapetal cell gene described by Wang et al. (2003), as claimed by Rey et al. (2017). The deletion of Raftin1, causes chromosome clumping in the equatorial plate at first meiotic metaphase, but no evidence exist that it triggers an increased number of univalents, multivalent pairing, and interlocking bivalents, as does the silencing of C-Ph1. Moreover, whereas Raftin1, whose protein product, RAFTIN, is essential for the late phase of pollen development, is expressed only in the anther, but not in root, stem, leaf tissues, or emasculated inflorescence (Wang et al. 2003), C-Ph1 is expressed during meiosis (maximum expression of the gene was observed during MI), and also in vegetative tissues, such as roots and flag leaves (Bhullar et al. 2014). Therefore, C-Ph1 and Raftin1 are not the same gene, but because C-Ph1 exhibits maximum activity at metaphase I and triggers chromosome clumping at this stage, it cannot be Ph1. Yet, if Ph1 phenotype stems from a cluster of linked genes, each controlling an aspect of Ph1 phenotype, then C-Ph1 may represent one of them. In this respect, it is worth mentioning that Rawale et al. (2019) observed that silencing of C-Ph1 resulted in 26% recombinant gametes between 1BS of wheat and 1RS of rye in hybrids between two bread wheat lines, one carrying intact chromosome 1B and the second containing a translocation in which chromosome arm IRS of rye replaced the 1BS of wheat. No recombination between these two arms took place in hybrids carrying Ph1.

Rey et al. (2018) claimed that recombination between chromosomes of wild relatives and those of allohexaploid wheat can be increased in F1 hybrids with a ph1b deletion, by treating the plants with Hoagland solution. A search for the element in the solution that is responsible for this increase revealed that irrigation of plants with a 1 mM Mg2+ solution caused a significant increase in homoeologous CO frequency in all analyzed wheat x wild relatives hybrids. These observations suggest a role for magnesium supplementation in improving the frequency of recombination in wheat-interspecific hybrids.

12.2.3 Other Suppressors of Homoeologous Pairing in Wheat

In addition to Ph1, there are several other suppressors of homoeologous pairing in T. aestivum, one located on chromosome 3D (Upadhya and Swaminathan 1967; Mello-Sampayo 1971a, b; Driscoll 1972; Mello-Sampayo and Canas 1973) and the other on 3A (Driscoll 1972; Mello-Sampayo and Canas 1973). The 3D and 3A genes, designated Ph2 and Ph3, respectively (Sears 1982, 1984), are less potent than Ph1 (Mello-Sampayo 1971a, b; Mello-Sampayo and Canas 1973). Ph2 is more effective than Ph3, but only about half as effective as Ph1. The effect of Ph2 and Ph3 on homoeologous pairing in allohexaploid wheat is negligible, but in wheat inter-generic hybrids, they have a somewhat more pronounced impact. Ph2 is located on the distal region of chromosome arm 3DS of T. aestivum (Driscoll 1972a, 1973) and presumably derived from the pairing suppressor gene on 3DS of Ae. tauschii, the donor of the D subgenome (Attia et al. 1977, 1979). The suppressor of homoeologous pairing on 3A is also located on the short arm (Driscoll 1972, 1973). It is most likely that the suppressors on 3DS and 3AS are homoeoalleles. Currently, there is no evidence showing that 3BL carries a homoeologous-pairing suppressor.

Aside from the minor suppressors on 3DS and 3AS, there is also a suppressor on chromosome 4D (Driscoll 1973) that appears to be almost as effective as the one on 3AS. Evidence was also obtained that the long arm of chromosome 2D (2DL) and possibly also of 2A and 2B, may carry a minor suppressor(s) of pairing (Ceoloni et al. 1986).

The EMS-induced 10/13 mutation (Wall et al.1971a), that was believed to be in the Ph1 gene (Wall et al. 1971b), was shown to be in the Ph2 locus (Sears 1982, 1984). Accordingly, following Sears (1982), the correct designation of the mutation 10/13 is ph2a, and not ph1a. Another deletion, aside from ph1b, was recovered by Sears (1977), after X-irradiation of pollen of T. aestivum cv. Chinese Spring. This deletion was substantially less potent than ph1b and resulted in an intermediate level of pairing in F1 hybrids with Ae. peregrina (designated Ae. kotschyi var. variabilis by Sears), i.e., approximately five bivalents per PMC, compared with one bivalent in the control with Ph1 and Ph2, and approximately 13 bivalents in the same hybrid carrying ph1b (Sears 1977). Consequently, Sears (1982) concluded that this deletion is not a mutation of Ph1 or of any other gene on chromosome 5B. Because Mello-Sampayo (1971a) and Driscoll (1972) found that chromosome arms 3DS and 3AS carry minor suppressors of homoeologous pairing, the mutation was suspected to involve one of these genes. Indeed, further studies showed that the mutation is on the terminal segment of the short arm of chromosome 3D (Sears 1982). Accordingly, this mutation was designated ph2b (Sears 1982). There is little or no homoeologous pairing when the mutation ph2b is homozygous in cv. Chinese Spring of T. aestivum (Sears 1977, 1982). Male transmission of the mutation is approximately normal, and fertility, while somewhat reduced, is sufficient for easy maintenance of the homozygous line (Sears 1982). The effect of ph2b on chromosomal pairing in wheat hybrids is quite similar to that of ph1a (Sears 1982). Using synteny with rice, Sutton et al. (2003) narrowed down the Ph2 locus, to a terminal 80 Mb of the short arm of chromosome 3D. More recently, however, Svačina et al. (2020) showed that the deletion induced by Sears (1977), the ph2b mutant, is actually larger than expected, comprising about 125 Mb of terminal end of the short arm of chromosome 3D.

Hybrids of allohexaploid wheat with rye, lacking both 3DS and 3AS, exhibit a level of pairing almost as high as that obtained in the absence of Ph1 (Mello-Sampayo and Canas 1973). There is evidently an interactive effect of the two deficiencies, as the pairing level when both 3DS and 3AS are missing is about twice as high as that observed in the absence of the more active suppressor, 3DS. However, no interaction or even additive action of Ph1 with Ph2 has been demonstrated; lack of both 5BL and 3DS resulted in no more pairing than lack of 5BL alone. Perhaps the absence of chromosome arm 5BL permits the maximum possible amount of pairing, and deletion of additional suppressors will therefore have no added effect (Mello-Sampayo and Canas 1973). In contrast, Ceoloni and Donini (1993) studied the effect of the combined deficiencies of ph1b and chromosome arm 3DS on homoeologous pairing in Chinese Spring and in hybrids between this line and Ae. pergrina and Secale cereale. They found that in both Chinese Spring itself and its hybrid with Ae. peregrina, the combined deficiencies reinforced the ph1b effect in promoting homoeologous pairing. On the other hand, no such effect was noted in the hybrid involving S. cereale.

It seems that Ph2 operates in a different way than does Ph1 (Benavente et al. 1998; Martinez et al. 2001a; Prieto et al. 2005). Benavente et al. (1998), using genomic in situ hybridization, studied the effect of Ph1 and Ph2 on wheat-wheat and wheat -rye chromosome pairing in first meiotic metaphase and in anaphase in wheat x rye hybrids carrying either Ph1, ph1b, and ph2b. Their observation revealed distinct mechanisms involved in the control of wheat homoeologous pairing by the two Ph genes. In accordance, it was suggested (Martinez et al. 2001a; Prieto et al. 2005) that Ph2 has a different function to that of Ph1, as it is not involved in recognition of homologous chromosomes but instead affects the progression of synapsis. Thus, Martinez et al. (2001b) concluded that the Ph1 and Ph2 loci bring about cytological diploidization of allohexaploid wheat via a different mechanism, whereby Ph2 affects synaptic progression while Ph1 affects the correction process of multivalents to bivalents at the transition from zygotene to pachytene.

Ceoloni and Feldman (1987) tested the two mutant lines of Ph2, ph2a and ph2b, for their mitotic-spindle sensitivity to colchicine. The data showed clearly that plants deficient for Ph2 or carrying ph2 alleles, were less sensitive to colchicine treatment than those carrying Ph2. Interestingly, the two genes that suppress homoeologous pairing in allohexaploid wheat, Ph1 and Ph2, affect the sensitivity of the mitotic spindle to colchicine, but in opposite ways: Ph1 decreases sensitivity, whereas Ph2 increases it. So, it can be assumed that Ph1 and Ph2 also operate by affecting microtubular stability, possibly interfering with the dynamic equilibrium of assembled-disassembled tubulin subunits, but in different manner. The 3DS Ph2 locus, was recently isolated (Serra et al. 2021). It encodes for the wheat homolog of DNA mismatch repair gene MSH7, TaMSH7. This is consistent with an anti-recombination effect when genetic divergence between the two recombination partners is too high as is the case with homoeologs. Indeed, the mismatch repair machinery detects mismatches in heteroduplexes formed at Holliday junctions at sites of crossover and recruits proteins that disengage the recombination partners.

Sears (1982) proposed three possibilities for making use of the ph2b mutation: (1) inducing the transfer of genes to wheat from closely related but not fully homologous alien chromosomes, (2) assessing the degree of relationship between alien chromosomes and their wheat homoeologues, and (3) improving amphiploids that have a somewhat reduced level of chromosome pairing.

12.2.4 Pairing Promoters in Wheat

There are several pairing-promoting genes that are also involved in the control of chromosome pairing in allohexaploid wheat. The long arms of chromosomes 5A and 5D carry such promoters (Feldman 1966b, 1968). Promoters also exist on the short arms of chromosomes of group 5, 5AS, 5BS, and 5DS (Feldman 1966b; Riley et al. 1966; Feldman and Mello-Sampayo 1967; Dvorak 1976), and on the long arms of chromosomes of group 3, 3AL, 3BL, and 3DL (Mello-Sampayo 1971a; Driscoll 1972, 1973; Mello-Sampayo and Canas 1973). Promoters were also found on the short arms of chromosomes of group 2, 2AS, 2BS, and 2DS (Ceoloni et al. 1986). Hence, the control of pairing in allohexaploid wheat stems from a balance between several suppressors and promoters.

The 5BS promoter induces pairing, both in wheat itself and in interspecific hybrids (Feldman 1966b; Feldman and Mello-Sampayo 1967; Riley and Chapman 1967). Its effect is substantially smaller than that of Ph1, as shown by the increased homoeologous pairing when both are missing, as in nulli-5B. The effect of the 5DL promoter is evidently greater than that of 5BS, for ditelo-5BL (lacking 5BS) has nearly normal synapsis, whereas nulli-5D (lacking 5DL and 5DS) tends to be asynaptic (Feldman 1966b), particularly at low (and presumably at high) temperatures (Riley et al. 1966). Since extra doses of chromosome 5A in (tetrasomic) largely suppress the effect of 5D nullisomy (Feldman 1966b; Riley et al. 1966), there must be pairing promoters on each arm of 5A. Yet, the effect of the 5A promoters is smaller than that of the 5D promoters since nulli-5A has normal synapsis, even at low temperatures (Riley et al. 1966).

The promoters on 5DL and 5AL could be homoeoallelic to Ph1 and those on 5DS and 5AS to the 5BS promoter. Rey et al. (2017) claimed that Ph1 has no homoeoalleles on 5AL and 5DL. If the promoters of 5AL and 5DL are not homoeoalleles of Ph1, then they should be located in a different chromosomal region that Ph1. There is some evidence (Mello-Sampayo 1972) that the promoter on 5DL is located distally to Ph1 and is not homoeoallelic to it. 5BL may also carry a promoter, distally to Ph1, that is homoeoallelic to the promoters of 5AL and 5DL.

The existence of promoter on the long arm of chromosome 3D was deduced from the fact that pairing in hybrids with Ae. peregrina was increased more by deficiency of the entire chromosome 3D than by deficiency of only the short arm of 3D carrying the pairing suppressor Ph2 (Driscoll 1972). Hybrid plants carrying only 3DS (deficient for 3DL) exhibited a greater increase in pairing, indicating the presence of a pairing promoter on this arm. The data of Mello-Sampayo (1971a) and Mello-Sampayo and Canas (1973) regarding the same and other hybrids, also support a slight promoting effect of 3DL.

The presence of the pairing gene on chromosome arm 3BL is necessary for normal synapsis (Li et al. 1945; Sears 1944, 1954; Kempanna and Riley 1962; Kato and Yamagata 1982, 1983). Sears (1944, 1954) allocated this gene to the right arm (=long arm; Sears and Sears 1979) of chromosome 3B and reported that absence of this gene led to failed crossover. Yet, this gene apparently does not interact with Ph1, since it controls different meiotic processes (Kempanna and Riley 1962). Consequently, it was concluded (Kempanna and Riley 1962) that chromosome arm 3BL carries a gene that is responsible for crossover. Hence, this gene may be different from the putative promoter on 3BL, which is homoeoallelic to the promoters on 3AL and 3DL.

Chromosome arm 2AS also evidently carries a gene essential for normal pairing (Sears 1954), although its effect is not as great as that of the gene on 3BL. Some investigations (Upadhya and Swaminathan 1967; Kempanna 1963) have failed to confirm its existence altogether. But, while an extra dose of chromosome 2A of Chinese Spring, previously reported to carry a pairing promoter on its short arm, did not increase pairing between homoeologous chromosomes in F1 hybrids between Chinese Spring and Ae. peregrina, two doses of chromosome 2D or 2B caused a significant increase in homoeologous pairing in this hybrid (Ceoloni et al. 1986). Evidently, chromosomes 2D and 2B carry a pairing promoter(s). Studies of F1 hybrids between aneuploids of CS, either lacking chromosome 2D or carrying an extra dose of it, and Ae. peregrina, Ae. longissima, and S. cereale, supported the finding that this chromosome carries a pairing promoter. Using ditelosomic lines, the promoter was found to be located on the short arm of 2D (2DS) (Ceoloni et al. 1986). It was deduced that the promoter on 2B is also located on the homoeologous short arm, i.e., 2BS. Thus, 2AS, 2DS, and 2BS carry a pairing promoter(s), with the promoter(s) on 2AS seemingly weaker than those on 2DS and 2BS.

Luo et al. (1992) reported that a Chinese landrace of allohexaploid wheat possesses a promoter of homoeologous pairing. The landrace has only bivalents at meiosis but hybrids between this race and S. cereale or Ae. peregrina exhibited an increase in homoeologous pairing (Luo et al. 1992; Liu et al. 1997b, 2003; Xiang et al. 2005). Monosomic analysis indicated that the promoter is located on chromosome 6A and, thus, it is not homoeoallelic to any of the other pairing genes (Liu et al. 2003; Hao et al. 2011). Yet, Fan et al. (2019), using two mapping populations, identified a QTL locus on 3AL, which is possibly responsible for the promotion of homoeologous pairing in hybrids with this race, and which is probably homoeoallelic to the promoter on 3AL.

Effects on spindle sensitivity to antimicrotubule agents seems to be a common feature of a number of genes that impact homoeologous pairing in hexaploid wheat and in its hybrids. For example, in addition to Phl and Ph2, chromosome arm 5BS, which carries a promoter of homoeologous pairing (Feldman 1966b, 1968; Riley et al. 1966; Riley and Chapman 1967; Feldman and Mello-Sampayo 1967), was also found to affect spindle sensitivity to colchicine (Avivi et al. 1970; Ceoloni et al. 1984). Similarly, Ae. speltoides and Amblyopyrum muticum carry genes that suppress Ph1 activity, induce a high degree of homoeologous pairing in hybrids with allohexaploid wheat (Riley 1966), and affect spindle characteristics. Ae. speltoides exhibits a higher mitotic sensitivity to colchicine than other Sitopsis species, which do not induce homoeologous pairing (L. Avivi and M. Feldman, unpublished data). In line with these reports, an A. muticum chromosome added to allohexaploid wheat induced homoeologous pairing and also caused alterations in the premeiotic spindle, while genotypes of allohexaploid wheat carrying other chromosomes of A. muticum did not exhibit homoeologous pairing and showed normal spindle functioning (Dover and Riley 1973).

On the other hand, no correlation between asynapsis and mitotic spindle features, as measured by colchicine sensitivity, was found in 3BL-deficient wheat genotypes. Kato and Yamagata (1983) found that absence of chromosome arm 3BL, which carries a pairing gene (Sears 1944, 1954), did not affect the sensitivity of the mitotic spindle to colchicine. This finding is in accord with earlier observations (Kempanna and Riley 1962) from which an effect of the 3BL gene on chromosome pairing, different from that of Phl, was inferred and led to the conclusion that both the timing and action of the 3BL gene are clearly distinct from those of the suppressors and promoters of homoeologous pairing.

The striking correlation between the effect on homoeologous pairing and alteration of spindle characteristics indicates that the suppressors and promoters of this type of pairing exert their effect on some features of the mitotic and premeiotic spindle. These genes impact the sensitivity of the mitotic spindle to colchicine, vinblastine, and griseofulvin (Avivi et al. 1970; Avivi and Feldman 1973b; Ceoloni et al. 1984; Gualandi et al. 1984), all of which are antimicrotubule agents that specifically bind tubulin, the main protein subunit of microtubules, thus shifting the equilibrium towards depolymerization.

12.2.5 Pairing Genes in Wild Relatives of Wheat

In many of the studied hybrids of T. aestivum with their wild relatives, there has been little or no evidence that the alien genome has any effect on pairing. The amount of pairing observed in these hybrids has been essentially the same as in haploids of T. aestivum (Riley and Law 1965). However, genotypes possessing genes that promote pairing have been found in several diploid and allotetraploid species, namely, Ae. speltoides (Riley et al. 1961), A. muticum (Riley et al. 1961; Riley and Law 1965), Ae. longissima (Mello-Sampayo 1971b), and Ae. peregrina (Farooq et al. 1990). These promoters of pairing have no effect in the species themselves, where complete pairing of homologues is the rule. But in intergeneric hybrids with allopolyploid wheat, they promote homoeologous pairing.

The level of pairing induced by the high-pairing lines of Ae. speltoides and A. muticum is almost as high as that observed in the same hybrids in the absence of chromosome 5B. An early work (Riley and Law 1965) showed approximately the same degree of pairing in hybrids of aestivum with high-pairing speltoides and muticum when 5B was absent as when it was present. In contrast, Feldman and Mello-Sampayo (1967) observed significantly more pairing in aestivum x speltoides lacking 5B. Similar super-high pairing was subsequently found by Dover and Riley (1972a) in aestivum x muticum hybrids lacking chromosome 5B.

The pairing genes of A. muticum and Ae. speltoides promote pairing between homoeologous chromosomes in hybrids involving allopolyploid wheat, by counteracting the effect of ph1 of allopolyploid wheat (Riley 1960; Feldman and Mello-Sampayo 1967; Dover and Riley 1972a; Dvorak 1972). Two such genes were identified in each of these species [Dover and Riley (1972a) in A. muticum and Dvorak (1972) in Ae. speltoides]. In Ae. speltoides, they were assigned to chromosomes 3S (Su1-Ph1) and 7S (Su2-Ph1) (Dvorak et al. 2006a). Interestingly, they were not found to be homoeoallelic to the suppressor and promoters of group 5 of allopolyploid wheat, but the promoter on 3S was suggested to be homoeoallele to the suppressors and promoters of group 3. The speltoides genes did not affect the level of pairing in the inter-specific diploid Ae. speltoides x Ae. tauschii and Ae. speltoides x Ae. caudata hybrids, in which Ph1 is not present (Chen and Dvorak 1984). In contrast, studies of meiotic chromosomal paring in hybrids between A. muticum and other diploid species of the wheat group showed relatively high paring, presumably due to the promotion of pairing by the muticum genes (Ohta 1990, 1991). Kihara and Lilienfeld (1935) observed seven bivalents at first meiotic metaphase of the F1 hybrid between Ae. comosa and A. muticum. In F1 of all crosses of the wild and domesticated forms of T. monococcum, Ae. speltoides, Ae. longissima, and Ae. caudata with A. muticum, Riley (1966) found high chromosome pairing in meiosis. Jones and Majisu (1968) reported high pairing in the F1 hybrid between Ae. tauschii and A. muticum and Ohta (1990, 1991) and M. Feldman (unpublished) observed almost complete pairing in the A. muticum x T. monococcum hybrids.

The simplest assumption concerning the mode of action of the A. muticum and Ae. speltoides pairing genes is that they are promoters of pairing, irrespective of the presence or absence of the homoeologous pairing suppressor genes. In this view, the high-pairing alleles are almost as strong as Ph1 and consequently largely neutralize the effect of Ph1 when it is present. In the absence of Ph1, they raise the pairing level, but only slightly, because it is already near the maximum (Feldman and Mello-Sampayo 1967). Alternatively, Riley (1960) proposed that the muticum and speltoides genes act by suppressing Ph1 activity. However, even the high-pairing alleles are unable to suppress Ph1 completely, and, consequently, pairing is never as high when Ph1 is present as when it is absent. The suggestion that the high-pairing alleles act by suppressing Ph1 gained credence with the finding that in the absence of Ph1, there is no difference between low- and high-pairing genotypes of speltoides or muticum in their effect on pairing in the hybrid with allopolyploid wheats (Dover and Riley 1972a; Rubenstein 1976). This strongly suggests that these alleles act only on Ph1 and have no effect when Ph1 is absent.

It is possible, however, that the existence of strong promoters in A. muticum and Ae. speltoides may be favored by the predominance of cross-pollination in these species. Riley and Law (1965) suggested that genes for high pairing may have a selective advantage in out-pollinators, being necessary to maintain full pairing between highly heterozygous homologues. This argument, however, revokes the proposal that the pairing promoters only act by suppressing Ph1.

It is not unlikely that other diploid relatives of wheat would have a similar effect if subjected to a sensitive test. In fact, partial suppression of Ph1 evidently occurs in hybrids with autotetraploid Ae. caudata (R. Riley, unpublished). Likewise, Upadhya (1966) found that Ae. caudata induces homoeologous pairing in hybrids with T. aestivum carrying chromosome 5B. This finding corroborates the reports of Kihara and Lilienfeld (1935) on the high degree of chromosome pairing in the triploid F1 hybrid between Ae. caudata and T. turgidum ssp. durum.

Similarly, chromosome 5U of Ae. umbellulata is reported to have a promoting effect similar to that of 5D (Riley et al. 1973). Mochizuki (1962) also found that the addition of a particular pair of diploid Agropyron elongatum (currently Elymus elongatus) chromosomes to durum wheat resulted in a substantial amount of homoeologous pairing. Dvorak (1987) investigated chromosome pairing in haploids of T. aestivum lines with added or substituted chromosomes of Elytrigia elongata (currently Elymus elongatus, genome EeEe), and found that promoters of homologous or homoeologous pairing on chromosome arms 3EeS, 3EeL, 4EeS, 5EeS, and 6Ee of E. elongata. Genes that suppressed pairing of homoeologous chromosomes were found on chromosome arms 4EeL and 7EeS of E. elongatus. Consequently, he suggested that genes promoting or suppressing pairing of homoeologous chromosomes are ubiquitous among diploid Triticeae species.

A pairing promoter was also detected in the wild subspecies of T. monococcum, ssp. boeoticum (currently ssp. aegilopoides (Chapman et al. 1976). These researchers proposed that the higher trivalent frequencies seen in the hybrids between allohexaploid wheat and ssp. aegilopoides could be due to homoeologous pairing and that the genotype of ssp. aegilopoides can partly suppress the activity of Ph1 of wheat.

The genome of S. cereale, which has no apparent effect on the level of homoeologous pairing in normal hybrids with T. aestivum, increases such pairing if two or three sets of rye chromosomes are added (Riley et al. 1973). Interestingly, the effect of the rye genome appears to depend on a balance between the long and short arms of chromosome 5R, with 5RL having a suppressive effect on pairing and 5RS an promoting and stronger effect. Gupta and Fedak (1986) studied chromosomal pairing in hybrids between T. aestivum and Petkus and prolific lines of rye and suggested the presence of genes both with major and minor effects on pairing in rye, with the genetic system promoting pairing in Petkus rye differing from that in prolific rye. The genetic variation in rye observed by Gupta and Fedak (1986) was considered by be similar to that of Ae. speltoides. In this respect, Dvorak (1977) observed that some of the rye genotypes promoted homoeologous chromosome pairing in hybrids with T. aestivum, and from the absence of distinct segregation classes among the hybrids, concluded that these pairing genes constitute a polygenic system.

Lelley (1976) tested the effect of single rye chromosomes on the pairing of homoeologous wheat chromosomes by crossing the seven wheat-rye addition lines of Imperial rye to Chinese Spring with S. cereale and S.montanum (currently S. strictum). In euhaploid hybrid plants (2n = 4x = 28; genome ABDR), no homoeologous pairing was induced, whereas in the 29-chromosome hybrids, while some wheat genome-driven suppression of chiasma formation between homologous rye chromosomes was noted, unequivocal evidence for homoeologous pairing of wheat chromosomes was found in several F1 plants. Lelley suggested that S. cereale and S. strictum possess a genetic system that suppresses Ph1 and that this system consists of several genes, located on different chromosomes, which may act additively. Such genes seem to be more frequent in the wild species, S. strictum, than in the domesticated S. cereale.

Halloran (1966) and Yu et al. (1998, 2001) observed promotion of homoeologous pairing in hybrids between bread wheat and Dasypyrum villosum by D. villosum gene(s), even in the absence of Ph1 or in the presence of its mutant ph1b. These results indicate that D. villosum contains promoter(s) of homoeologous pairing, irrespective of Ph1. Another case of induced homoeologous pairing in the presence of Ph1 and Ph2 was reported by Liu et al. (2011) and Koo et al. (2017), who observed frequent recombination between 5M and 5D chromosomes in substitution lines of bread wheat containing 5Mo from Ae. geniculata, indicating that the promoter on 5Mo can counteract Ph1. Further cytogenetic analysis showed that that the promoter may be located in proximal regions of chromosome 5Mo. Later, Koo et al. (2017) used two different 5Mo chromosomes from different accessions of Ae. geniculata in T. aestivum background and observed differential levels of pairing between 5Mo and 5D in both lines, with chiasmata between 5Mo and 5D detected in 6.7% versus 21.7% of studied meiocytes. This might have been caused by the presence of different alleles on 5Mo that repress Ph1. Additionally, homoeologous pairing occurred only between the 5Mo and 5D chromosomes, as no multivalents were detected (Koo et al. 2017).

Several accessions of allotetraploid species of Aegilops also carry promoters of homoeologous pairing, i.e., Ae. peregrina (Farooq et al. 1990), Ae. triuncialis, Ae. crassa (Claesson et al. 1990), and Ae. geniculata (Lacadena and Azpiazu 1969). Only a number of genotypes of each of the above diploid and allopolyploids Aegilops species possess the promoter genes, indicating the occurrence of genetic variation in this trait. It is not unlikely that other diploid relatives of wheat would have a similar effect if subjected to a sensitive cytogenetic test.

Of note, all the promoters of chromosome groups 5, 3, and 2 in allohexaploid wheat are active, implying that no genetic diploidization (mutation leading to neofunctionalization, inactivation or elimination) took place in these genes. It is presumably crucial for the regularity of pairing processes in allohexaploid wheat to maintain all these genes in an active state, despite genetic redundancy. All the common wheat promoters and suppressors are part of a well-coordinated gene system that affects some of the critical processes that are involved in homology recognition, pairing, and crossing over in allohexaploid wheat itself and in hybrids with related species.

Promoters of homologous and homoeologous chromosome pairing are present in many Triticineae species (Table 5.2). Yet, there is variation in the presence and intensity of homoeologous pairing promoters and suppressors (Naranjo and Benavente 2015). Ozkan and Feldman (2001) reported variation in the effect of Ph1 of tetraploid wheat, Triticum turgidum subsp. dicoccoides, on the level of homoeologous pairing in F1 hybrids with Aegilops peregrina. Likewise, promoters in different lines of the diploid species Ae. speltoides and A. muticum induce different levels of pairing (high-, intermediate- and low-pairing) in hybrids with common wheat (Kimber and Athwal 1972; Dvorak 1972; Vardi and Dover 1972; Dover and Riley 1972a). Similarly, low-pairing and intermediate-pairing lines of Ae. longissima induce different levels of pairing in hybrids with T. aestivum (Mello-Sampayo 1971b). Pairing promoters were also found in several lines of Agropyron cristatum (Ahmad and Comeau 1991; Chen et al. 1989) but not in others (Limin and Fowler 1990). Evidence of variation in the effect of the promoters on homoeologous pairing among Secale taxa was obtained by Cuadrado and Romero (1984), Naranjo et al. (1979), and Naranjo and Palla (1982). Similar evidence was obtained by Gupta and Fedak (1986), who studied the effect of two S. cereale cultivars on chiasma frequency in common wheat x S. cereale hybrids.

Viegas et al. (1980) reported that chromosome arm 5DL of a mutant line of T. aestivum carries a potent pairing suppressor. Suppressors of homoeologous pairing were also found in the diploid subspecies of Elymus elongatus (Dvorak 1987; Charpentier et al. 1988). Chromosome arms 4EeL, 7EeL, and 6Ee of this subspecies carry genes that suppress homoeologous pairing in hybrids with species of other genera (Dvorak 1987; Charpentier et al. 1988).

The existence of suppressors of homoeologous pairing in diploid species of the wheat group was suggested by several researchers (Okamoto and Inomata, 1974; Waines 1976). Waines interpreted pairing data of F1 Triticeae diploid hybrids and amphiploids published by Sears (1941a, b), to indicate that diploid species of this group have genetic systems (suppressors) controlling homoeologous pairing in the inter-specific hybrids, which result in an immediate cytological diploidization in the derived amphiploids. These systems, which, according to Waines (1976), function at as isolating mechanisms, are already present in diploid species.

In accord with Waines, Shang et al. (1989) reported on the occurrence of pairing-control genes in diploid wheat, T. monococcum. Maan (1977) also concluded from his cytogenetic studies of T. turgidum ssp. durum x Ae. comosa ssp. heldreichii hybrids, that a Ph1-like gene exist in the latter.

Several studies attributed the low pairing in allopolyploid wheats x Emarginata species (Ae. bicornis, Ae. sharonensis, Ae. longissima, and Ae. searsii) to weak homoeologous pairing suppressors that exist in these diploid species, which, together with Ph1 of allopolyploid wheats, suppress the pairing of homoeologous chromosomes in these hybrids (Feldman 1978). The assumption that several lines of Ae. longissima carry a weak suppressor is supported by Avivi’s (1976) finding that an autotetraploid derived from a low-pairing type of Ae. longissima exhibited fewer multivalents and a larger number of bivalents than the autotetraploid derived from an intermediate-pairing type of Ae. longissima.

A correlation between increase in bivalent pairing on the account of quadrivalent pairing and an effect on mitotic spindle sensitivity was also observed in the low-pairing line of Ae. longissima (L. Avivi, unpublished data). She found that the two induced autotetraploid lines of Ae. longissima, differing in the degree of multivalent pairing, also differed in their mitotic spindle sensitivity to colchicine; the bivalent-forming autotetraploid, assumed to possess a suppressor (Avivi 1976), was less sensitive than the multivalent-forming one. Similarly, it was found (A. Charpentier, unpublished data) that the natural autotetraploid Elymus elongatus (=Agropyron elongatum), which forms only bivalents at meiosis, was much less sensitive to colchicine than an induced autotetraploid of this species that forms several multivalents in every meiocyte. Crosses of the two autotetraploid lines indicated that bivalentization in E. elongatus results from the activity of a recessive gene (Charpentier et al. 1986).

All the allopolyploid species of Aegilops are characterized by diploid-like meiotic behavior and exhibit strictly bivalent pairing at first meiotic metaphase (Riley and Law 1965). In the absence of Ph1, no genotype of an allopolyploid Aegilops species reproduced the effect this gene has on homoeologous pairing (Riley and Law 1965). This is in spite of the fact that they may contain weak suppressors with inconspicuous effects at the allopolyploid level, that were presumably contributed by their diploid parents. Based on the above results, Riley and Law (1965) concluded that if there is a system in these polyploids suppressing homoeologous pairing, then it must be functionally distinct from that of T. aestivum.

Indeed, none of the allopolyploid Aegilops genotypes tested by McGuire and Dvorak (1982) fully compensated for the absence of chromosome 5B, but genotypes of Ae. cylindrica, Ae. juvenalis, Ae. triuncialis, Ae. geniculata, Ae. columnaris, Ae neglecta, and Ae. recta, did have some suppressive effects on homoeologous pairing when 5B was absent from the hybrids. This suggests that the diploid-like meiosis in these allopolyploid species is also a consequence of genetic suppression. This conclusion is in accord with the absence of homoeologous pairing in a haploid of Ae. geniculata (Matsumura 1940; Kihara 1937), despite the pairing (2–5 bivalents and 0–2 trivalents) in the diploid hybrid between Ae. umbellulata and Ae. comosa, the assumed two parents of Ae. geniculata (Kihara 1949). It is, therefore, clear that in this allopolyploid species, completely regular meiosis, with pairing restricted to homologous partners, developed at the allotetraploid level, either by causing some structural changes among the homoeologues via elimination of DNA sequences, as was shown by Feldman et al. (1997) and Ozkan et al. (2001), or via development of a genetic system that suppresses pairing of homoeologues.

12.2.6 The Effect of B Chromosomes on Pairing in Wheat Hybrids

Two diploid species of the wheat group, Amblyopyrum muticum and Ae. speltoides, have B (accessory) chromosomes which bear a suppressive effect, similar to that of Ph1, on chromosome pairing in hybrids with allopolyploid wheats. In T. aestivum x A. muticum or x Ae. speltoides hybrids lacking chromosome 5B but possessing B chromosomes, the level of pairing was generally the same as when 5B was present, irrespective of the presence or absence of muticum or speltoides pairing alleles (Vardi and Dover 1972). Up to six B chromosomes had little, if any, additive effect as compared to just one, on pairing in hybrids of Ae. speltoides with T. aestivum (Vardi and Dover 1972). However, six B chromosomes caused some pairing failure in Ae. speltoides itself (Dover 1975). There seems to be little, if any, interaction between the B chromosomes and the chromosome carrying Ph1; the pairing level is about the same when either or both are present. No supplementation of the effect of Ph1 by B chromosomes could be found at any level of pairing (Dover 1973; Vardi and Dover 1972). Similarly, studies of Ohta (1990, 1991) on pairing in hybrids between A. muticum and all the diploid species of the wheat group, with or without B chromosomes, showed that A. muticum B chromosomes suppress pairing in these hybrids. Based on these results, Ohta (1990, 1991) exploited the effect of B chromosomes of A. muticum on meiotic chromosomal pairing in F1 hybrids with diploid species of the wheat group, to evaluate the degree of genomic divergence between these diploid species and A. muticum.

Roothaan and Sybenga (1976) analyzed hybrids between T. aestivum cv. Chinese Spring and rye, with and without chromosome 5B of wheat and B-chromosomes of rye and found that the absence of 5B resulted in an increase in homoeologous pairing, alongside a decrease in chiasmata in the rye B chromosomes themselves. Two rye B chromosomes were entirely ineffective in compensating for the absence of 5B.

Neijzing and Viegas (1979) analyzed T. aestivum x S. cereale hybrids, with and without chromosome 5B or 5D and with and without B chromosomes of rye. In the absence of chromosome 5B, there was no effect of rye B chromosomes on meiotic synchrony and on chiasma frequency. Absence of 5D appeared to decrease synchrony at 20 and 15 °C, but genetic variation between plants played an important role as well.

Romero and Lacadena (1980) and Cuadrado et al. (1991) analyzed the interaction between homoeologous pairing of rye B chromosomes carrying the wheat pairing suppressors Ph3, Ph2, and Ph1, and the promoters on 3BL, 5AL, and 5DL in wheat-rye hybrids. These authors found that when the pairing suppressors of wheat were absent, rye B chromosomes had a suppressive effect on pairing, but behaved as promotors when the pairing promoters were absent.

Several researchers transferred B chromosomes of rye into an allopolyploid wheat background. Lindström (1965) managed to obtain plants of bread wheat with two B chromosomes of rye but, unfortunately, his sudden death, prevented him from studying the effect of these B chromosomes on wheat chromosome behavior. Müntzing et al. (1969) studied meiosis in this line and reported a reduction in pairing of the B chromosomes, with almost no noticeable effect on the pairing of the wheat chromosomes. Likewise, Niwa et al. (1997) transferred B chromosomes from Korean rye into cv. Chinese Spring of T. aestivum and reported that the presence of the Bs did not disturb pairing between wheat chromosomes. Mamun-Hossain et al. (1992) compared meiotic chromosome behavior in allotetraploid wheat, T. turgidum ssp. durum, carrying B chromosomes from rye, with that of hexaploid wheat with the same B chromosomes (from the Lindström strain). The B chromosomes induced some asynapsis of the wheat chromosomes, i.e., increased the number of rod bivalents and univalents per cell, with effect of Bs in tetraploid wheat was more pronounced than in hexaploid wheat.

An attempt to identify B chromosomes segments that affect homoeologous pairing in T. aestivum x Ae. peregrina hybrids, was made by Kousaka and Endo (2012) using the B-9 and B-10 segments of rye B chromosomes. The B-9 and B-10 segments are derived from reciprocal translocations between a wheat and B chromosomes; B-9 has the B chromosome pericentromeric segment and B-10 has the B distal segment. B-9, like the B chromosome, suppressed homoeologous pairing when chromosome 5B was absent. On the other hand, the B-9 and B-10 segments promoted homoeologous pairing when 5B was present. These results suggest that the effect of the B chromosomes on homoeologous pairing is not confined to a specific region and that the intensity of the effect varies, depending on the presence or absence of 5B and on the segment and dose of the B chromosomes.

12.2.7 Origin of the Ph Genes

It was assumed that the Ph1 gene evolved at the allotetraploid level, in parallel to, or soon after, the allopolyploidization process (Riley 1960; Sears 1976b). Dover and Riley (1972b) proposed three obvious possibilities for the origin of the Ph1 gene: (a) it arose as a mutation following the formation of allotetraploid wheat; (b) it was transferred to chromosome 5B of a nascent allotetraploid from a B chromosome of Ae. speltoides or A. muticum; or (c) it was already present in the diploid species that donated the B subgenome to allotetraploid wheat. The pairing suppressor in the diploid progenitor of allotetraploid wheat has a weak suppressive effect that was gradually or rapidly increased in the allotetraploid.

In the absence of any clear-cut evidence that the diploid species that contributed the B subgenome to allotetraploid wheat, possessed a Ph1-like gene, an origin of Ph1 by mutation in the nascent allotetraploid seems a likely possibility. The presence of promoters, homoeoalleles to Ph1, on 5AL and 5DL (and possibly also in all the diploids of the wheat group) support the theory of evolution of Ph1 through a mutation to an antimorphic allele. Riley (1960), Riley and Law (1965), and McGuire and Dvorak (1982) assumed that Ph1 resulted from a mutation in a pairing-promoting gene that existed in the diploid donor of the B subgenome. The mutation may have altered the activity of the mutant gene, subsequent to the formation of allotetraploid wheat. Mutation of a promoter gene to Ph1 would be a type of mutation (antimorphic) that occurs spontaneously in allopolyploids (Sears 1972). Such a mutation from a promoter on 5DL to a suppressor of homoeologous pairing was described by Viegas et al. (1980). Whichever type of mutation occurred, it presumably benefited from immediate, strong, favorable selection pressure, since it likely considerably increased the fertility of the nascent allotetraploid.

Another possibility for the origin of Ph1 in the nascent allotetraploid wheat was recently proposed by Rey et al. (2017), who identified the Ph1 region on 5BL in a segment containing a block of heterochromatin, the Zip4 gene and several other genes that were duplicated in chromosome arm 5BL from chromosome arm 3BL. This segment has no similar segments in the homoeologous chromosome arms 5AL and 5DL. Mutations in the Zip4 gene affect crossing over between homoeologues, similar to the effect of ph1b.

The likelihood that Ph1 was transferred to chromosome 5B from a B chromosome was suggested by Dover and Riley (1972b), Dover (1973), and Riley et al. (1973), based on the similarity between the effects of Ph1 and B chromosomes of Ae. speltoides and A. muticum. Dover (1973) speculated that Ph1 originated from an ancestral interchange between a B chromosome and an A chromosome of a putative B-genome donor. But there is no evidence that the effect of B chromosomes of these two diploid species is due to a single gene. Moreover, there are some differences between the effects of these B chromosomes and Ph1; an extra dose of B with two doses of Ph1 did not induce homoeologous pairing and interlocking of bivalents the way an extra dose of Ph1 does alone.

Dover (1973) concluded that B chromosomes are not solely responsible for the reduction in homoeologous pairing in T. aestivum x A. muticum hybrids carrying B chromosomes. The presence of B chromosomes did not introduce an additional factor regulating the degree of pairing. Moreover, Vardi and Dover (1972) found that B chromosomes cause disturbances in the mitotic and meiotic spindle and Dover (1973) attributed this effect to the presence of satellited DNA in pericentromeric regions as opposed to the effect of a single gene.

The third possibility assumes the preexistence of weak homoeologous pairing suppressor genes in diploid species. Dover and Riley (1972a) proposed that Ph1 may has evolved directly as the result of incorporation of a low-pairing allele of Ae. speltoides or A. muticum into nascent allotetraploid wheat. The discovery of suppressors other than Ph1 in hexaploid wheat, and particularly the finding (Riley et al. 1973) that one chromosome of rye, chromosome 5R, carries a suppressor as well as a promoter of pairing, reinforces the proposal that the diploid relatives of wheat all carry suppressors. Indeed, such preexistence of suppressors was suggested by Okamoto and Inomata (1974) who, based on the formation of several bivalents in haploids of diploid wheat, barley and rye, concluded that these genomes include several duplications that are located in different chromosomes. Yet, despite the existence of these duplications, multivalents do not form in the diploids. Okamoto and Inomata (1974) therefore suggested that a gene or genes which suppress quadrivalent formation at meiosis is present in these diploid species. These genes suppress ectopic exchanges between the duplicated regions observed within the diploid genomes.

This possibility is strongly defended by Waines (1976) who, interpreted pairing data of F1 Triticeae diploid hybrids and amphiploids, published by Sears (1941a, b), to indicate that diploid species of this group have genetic systems controlling homoeologous pairing in inter-specific and inter-generic hybrids, resulting in immediate cytological diploidization in the derived amphiploids. These systems, which, according to Waines (1976), function at the diploid level as isolating mechanisms, are already present in diploid species and did not necessarily arise by mutation de novo after allopolyploid formation, but rather, underwent changes increasing their efficiency at the tetraploid level.

The existence of low-potency suppressors of homoeologous pairing in diploid species of the wheat group was suggested by several researchers. The probable existence of a pairing suppressor gene in lines of Ae. longissima and possibly also in Ae. bicornis, Ae. sharonensis, Ae. searsii, and Ae. speltoides, may explain the low amount of pairing typically observed in hybrids between allopolyploid wheats and accessions of these species. In this respect, it is assumed that the donor of the B subgenome to allotetraploid wheat was a low-pairing type containing a weak suppressor of homoeologous pairing. As was suggested by Waines (1976), such genes may be a constituent of an inter-specific genetic barrier enabling the divergence of developing taxa into species. The F1 hybrids between the donor of the B subgenome and that of the A subgenome exhibited very little pairing, presumably due to presence of a suppressor, and were sterile. Interspecific or intergeneric F1 hybrids having low chromosomal pairing at meiosis tend to undergo chromosome doubling, due to the formation of unreduced gametes, more readily than hybrids with high pairing, and produce stable and fertile allopolyploids.

Alternatively, it is not improbable that a suppressor was transferred from one of these Sitopsis species to the donor of the B subgenome, if such a gene was not already in the B subgenome donor before, and from there to the B subgenome of allotetraploid wheat. Bernhardt et al. (2020) presented evidence indicating introgression from Sitopsis species, especially Ae. longissima and Ae. searsii, into the donor of the B subgenome of allopolyploid wheats or directly with the B subgenome of allotetraploid wheat.

If Ph1 in tetraploid wheat derived from a suppressor of the diploid species, which is a relatively less potent suppressor, it was not sufficiently effective soon after formation of allotetraploid wheat. It is assumed, therefore, that only later on, a mutation(s) improved its efficiency. Similarly, the Ph1 of hexaploid wheat presumably became stronger than that of the tetraploid. The absence of evidence for intergenomic recombination in tetraploid and hexaploid wheat, may indicate that the original weak Ph1 gene was superimposed on another system which prevented pairing of homoeologous chromosomes, as suggested by Feldman et al. (1997) and Ozkan et al. (2001).

Attia et al. (1977, 1979) demonstrated that the Ph2 gene is present in Ae. tauschii, the donor of the D subgenome to allohexaploid wheat. Their results and those of Ekingen et al. (1977) clearly indicated that several diploid biotypes of Ae. tauschii caused a suppression of homoeologous pairing similar to the effect of chromosome arm 3DS of T. aestivum. This shows that the 3DS suppressor of hexaploid wheats derived from Ae tauschii. These results are in accord with the assumption of Azpiazu and Lacadena (1970) that the 3DS suppressor originated in Ae. tauschii. The alternative suggestion (Driscoll 1972) of origin after the completion of hexaploid wheat, e.g., by mutation (Mello-Sampayo and Lorente 1968), seems to be inconclusive.

The evolution of Ph1 at the tetraploid level likely occurred in several steps rather than in one event. This is inferred from the occurrence of variation in Ph1 in allotetraploid wheat, which include high-, intermediate- and low-pairing alleles of this gene. Ozkan and Feldman (2001) reported on the discovery of genotypic variation in the control of homoeologous pairing among tetraploid wheats. When comparing the levels of homoeologous pairing in hybrids between CS and Ae. peregrina, significantly higher levels were obtained in hybrids between Ae. peregrina and CS substitution lines in which chromosome 5B of CS was replaced by either 5B of wild allotetraploid wheat, T. turgidum ssp. dicoccoides line 09 (TTD09), or 5G of Triticum timopheevii ssp. timopheevii line 01 (TIM01). Similarly, a higher level of homoeologous pairing was found in the hybrid between Ae. peregrina and a substitution line of CS in which chromosome arm 5BL of line TTD140 of ssp. dicoccoides substituted 5BL of CS. The observed effect on the level of pairing is seemingly exerted by chromosome arm 5BL of ssp. dicoccoides, most probably by an allele of Ph1. Searching for variation in the control of homoeologous pairing among lines of wild allotetraploid wheats, T. turgidum ssp. dicoccoides or T. timopheevii ssp. armeniacum, showed that hybrids between Ae. peregrina and lines of these two wild allotetraploid exhibited three different levels of homoeologous pairing: low, low-intermediate, and high-intermediate. The low-intermediate and high-intermediate genotypes may possess less-effective alleles of Ph1. The variation in the activity of Ph1 in different accessions of T. turgidum ssp. dicoccoides and in T. timopheevii ssp. armeniacum suggests a gradual, rather than a single-step, evolution of the homoeologous pairing suppressor in wild allotetraploid wheats.

Variation in Ph1 activity was also found in bread wheat, T. aestivum ssp. aestivum. Martinez et al. (2005) analyzed chromosome pairing in haploids of three different cultivars of bread wheat (Thatcher, Chris, and Chinese Spring) that were obtained from crosses with Zea mays and found differences in their meiotic behavior. Thatcher and Chris haploids had significantly higher levels of pairing at first meiotic metaphase than CS haploids. This pairing correlated with higher levels of synapsis in the Thatcher and Chris first prophase nuclei as compared to the Chinese Spring nuclei. The authors concluded that variation exists in the effectiveness of the diploidizing mechanism among cultivars of bread wheat.

Rawale et al. (2019) characterized the structure and expression of C-Ph1, that according to Bhullar et al. (2014) is the candidate Ph1 gene, in the ssp. dicoccoides accessions found by Ozkan and Feldman (2001) to differ in their effect on homoeologous pairing in hybrids with Ae. pergrina. The C-TdPh1-5B of ssp. dicoccoides transcribed three splice variants as observed in the hexaploid wheat (Rawale et al. 2019). Further, single-nucleotide changes differentiating accessions varying in homoeologous-pairing control were identified. Quantitative expression analysis showed that the wild emmer accessions that induced high homoeologous pairing, had ~ 10,000-fold higher transcript abundance of the C-TdPh1-5B during first meiotic prophase compared to accessions that exhibited low homoeologous pairing. Based on these results, Rawale et al. (2019) concluded that the homoeologous-pairing control is mediated by transcriptional regulation of this gene during meiosis. The presence of genetic variation in the genetic control of homoeologous pairing in different accessions of wild emmer questioned the validity of the proposed single-step evolution of the homoeologous-pairing suppressor.

Both wild and domesticated T. timopheevii contain a Ph1-like gene (Feldman 1966a; Ozkan and Feldman 2001). One possibility is that Ph1 independently originated in both wild allotetraploid Triticum taxa, T. turgidum ssp. dicoccoides and T. timopheevii ssp. armeniacum. This could have resulted from an independent mutation in a pairing promoter gene. Alternatively, the line the Ae. speltoides-related species which donated the B subgenome to ssp. dicoccoides, and the more modern speltoides line that contributed the G subgenome to ssp. armeniacum, contained a weak pairing suppressor. On the other hand, it is reasonable to assume that Ph1 arose in only one wild allotetraploid taxon, most probably in the older ssp. dicoccoides, and was transferred to the younger ssp. armeniacum through introgression.

12.2.8 Conclusion

To sum up, cytological diploidization in allopolyploid Triticum species arises through two independent, complementary systems. One is based on the physical divergence of chromosomes, and the second, on the genetic control of pairing. The Ph-gene system superimposes itself on and takes advantage of, and thereby reinforces, the system of the physical differentiation of homoeologous chromosomes. The stringent selection for fertility in the allopolyploid Triticum species might favor the development of these two systems, to ensure prevention of multivalent formation and promotion of bivalent pairing in nature.

The use of aneuploid lines of T. aestivum enabled the identification of structural changes, mainly elimination, of DNA sequences in the chromosomes of synthetic and natural allopolyploids. The existence of the same changes in nascent and natural allopolyploids indicates that they were induced during the course of allopolyploidization. Elimination of DNA sequences from the homoeologues of one subgenome in allotetraploids and from the additional subgenome in allohexaploids, converts the remaining sequences into homologous-specific, which can presumably serve in homology search and in initiation of pairing at meiosis.

Superimposed on this physical system in allopolyploids Triticum species, is a genetic system comprising of a battery of genes that suppress and promote chromosome pairing. Although much information has accumulated on the occurrence and activity of these genes, some issues remain unclear, especially as to the ways in which these genes operate. What is striking is the complexity of the system, which involves not only several suppressors and several promoters in the allopolyploid wheat itself, but also genes in related diploids that largely inhibit the major suppressor, and B chromosomes in the diploids that have an effect similar to that of the suppressor.

Relatively little information is available on pairing genes in the wild relatives of wheat, but it is clear that these species also carry both suppressors and promoters affecting chromosome pairing. As was shown by Sears (1941b), in at least some hybrids between diploids, the amount of pairing depends upon the particular biotypes used to make the crosses. Rarely have enough different biotypes been tested to adequately assess the potential for pairing in the combination concerned. Also, it is not clear which genetic system, if any, exists in allopolyploids of Aegilops.

The rapid process of cytological diploidization in the newly-formed allopolyploid species of the wheat group has been critical for their successful establishment in nature. The restriction of pairing to completely homologous chromosomes ensures regular segregation of genetic material, high fertility, genetic stability, and disomic inheritance, which prevents the independent segregation of chromosomes of the different subgenomes. This mode of inheritance leads to permanent maintenance of favorable inter-subgenomic genetic interactions, and thereby fixes heterotic interaction between subgenomes. On the other hand, disomic inheritance sustains the asymmetry in the control of many traits by the different subgenomes (Feldman et al. 2012). In addition, since cytological diploidization facilitates genetic diploidization, genes existing in double and triple doses can be diverted to new functions through mutations, thereby giving preferentiality to the creation of favorable, new inter-subgenomic combinations.

12.3 Genetic Diploidization

How the two or three divergent subgenomes, present in a single nucleus of a nascent allopolyploid, were led to operate in a harmonious manner? On the one hand, an extra dose of some homoeoalleles can be of positive adaptive value. On the other hand, overexpression of the increased gene dosage may lead to redundancy, waste, or, in some cases, even to a deleterious effect. To prevent such negative consequences, the expression of the increased dose of homoeoalleles is reduced, a phenomenon described as “gene dosage compensation”. An example of gene-dosage compensation was presented by Galili et al. (1986) in genes coding for the subunits of high-molecular-weight glutenins in T. aestivum ssp. aestivum. Furthermore, because of genetic divergence of the wheat subgenomes at the diploid level and because of mutations in the allopolyploid level, some homoeoalleles in allopolyploid wheat may be variable, representing a different form of the gene, and thus, may differ from one another in their expression profile. In this case, activity of all the duplicated genes may be advantageous, producing favorable inter-subgenomic interactions. Inter-subgenomic gene interactions may be, in some cases, expressed in novel traits that do not exist in their parental diploids. Some of these traits may be of great adaptive value in nature. Inter-subgenomic gene interactions are also of direct relevance to wheat cultivation. For example, the baking quality of bread wheat is due to the unique properties of its gluten—a product derived from the combined contribution of the three subgenomes of ssp. aestivum and thus, exists only at the hexaploid level.

Moreover, allopolyploidy enables to generate a new type of genetic variation through recombination between homoeologous chromosomes. Using cytological and whole-genome sequence analyses, Zhang et al. (2020) identified 37 homoeologous exchange (HE) events in the progeny of a nascent synthetic allotetraploid T. urartu—Ae. tauschii (2n = 4x = 28; genome AADD). HEs exhibit typical patterns of homologous recombination hotspots, being biased toward low-copy, sub-telomeric regions of chromosome arms and showing association with known recombination hotspot motifs. But strikingly, while homologous recombination preferentially takes place upstream and downstream of coding regions, HEs are highly enriched within gene bodies, giving rise to novel recombinant transcripts, which in turn are predicted to generate new protein fusion variants. However, the amphidiploid T. Urartu—Ae. tauschii has a genomic combination (AADD) that does not exist in nature and might be an atypical case of inter-genomic recombination. Nevertheless, while HE is less frequent in natural wheat, several inter-genomic translocations were encountered in lines of natural allopolyploids of the wheat group, (Maestra and Naranjo 1999, and reference therein). Moreover, Intragenic recombination and formation of chimeric genes was detected in HEs of allopolyploids Brassica and Arabidopsis suecica, and in autopolyploids rice, banana, and peanut, indicating that homoeologous exchange may be a broad phenomenon that occurs also in natural allopolyploids. HE thus provides a mechanism for evolutionary novelty in transcript and protein sequences in nascent allopolyploids.

Another challenge of interspecific or intergeneric hybrids is to suppress activity in cases of deleterious effect or homoeoallele redundancy. The processes that bring redundant or unbalanced gene systems in allopolyploids toward a diploid-like mode of expression are called genetic or functional diploidization. Genetic diploidization is a regulatory process that brings redundant or unbalanced gene systems in allopolyploids toward a diploid-like mode of expression (Ohno 1970). It results either from elimination, mutation or repression of genes that, in many cases, restrict the activity of sets of genes to only one subgenome (Liu et al. 1998b; Wendel 2000; Shaked et al. 2001; Levy and Feldman 2002, 2004; Feldman and Levy 2005, 2009; Comai 2005; Chen 2007). These genomic changes may increase the fitness and competitiveness of the newly formed allopolyploid, leading to its successful establishment in nature. Hence, successful allopolyploidizations are those that successfully trigger an array of genetic and epigenetic changes that confer evolutionary advantages. If heightened fitness is not achieved rapidly enough, the nascent allopolyploid would be selected against as in the case of hybrid incompatibility.

Studies with newly-formed allopolyploid wheats as well as genome sequencing data of natural wheats have indicated that a broad range of DNA alterations occurred during or soon after allopolyploidization, leading to genetic diploidization (Levy and Feldman 2004; Feldman and Levy 2005). These alterations included deletion or silencing of genes, or alternatively, neofunctionalization, occurrence of an adaptive mutation enabling one of the homoeoalleles to develop a new function that was not present in the ancestral gene. Silencing of one or more of the homoeoalleles can be in the form of pseudogenization, a process through which one of the homoeoalleles loses its function as a result of disruption of its regulatory or coding sequence, or subfunctionalization, a neutral mutation process causing an homoeoallele to maintain only a subset of its original ancestral gene. In addition, transposon activation and deactivation, are extensive and occur relatively rapidly. Levy and Feldman (2002, 2004), Feldman and Levy (2005, 2009, 2011, 2012), and Feldman et al. (2013)) distinguished between revolutionary changes, occurring during or immediately after allopolyploidization and evolutionary changes that take place throughout the life of the allopolyploid (Table 12.2). Revolutionary changes include genetic and epigenetic alterations that lead to cytological and genetic diploidization, thereby improving the harmonious functioning of the divergent subgenomes, stabilizing the nascent allopolyploid and facilitating its establishment as a new competitive species in nature—all of which are species-specific. Evolutionary changes comprise mostly genetic changes that promote genetic diversity, flexibility and adaptability—all of which are biotype- or population-specific. In contrast to genes that code for enzymes, genetic diploidization in allopolyploid wheats mainly involves genes that code for structural or storage proteins, e.g., histones, tubulin subunits, glutenins and gliadins subunits, and ribosomal RNA (and possibly also tRNA). In such genes, expression of all homoeoalleles might be redundant and even deleterious, due to over-production and inefficiency. In this case, traits controlled by genes from only or mostly one subgenome may have a higher adaptive impact. It is therefore expected that such gene loci would have been targets for genetic diploidization. The Hardness (Ha) locus, controlling grain hardness in Triticum and Aegilops species, represents a classical example of a trait whose variation arose from gene loss after allopolyploidization (Chantret et al. 2005). The previously reported loss of Pina and Pinb genes from the Ha locus of allopolyploid wheat species, was caused by a large genomic deletion that likely occurred independently in the A and B subgenomes. Moreover, the Ha locus in the D subgenome of ssp. aestivum is 29 kb smaller than in the D genome of its diploid progenitor Ae. tauschii, principally because of transposable element insertions and two large deletions caused by illegitimate recombination. The data of Chantret et al. (2005) suggest that illegitimate DNA recombination, leading to various genomic rearrangements, constitutes one of the major evolutionary mechanisms in wheat species.

Table 12.2 Types and characteristics of genome changes in allopolyploid wheat
Full size table

Likewise, Shitsukawa et al. (2007) reported that both genetic and epigenetic alterations occurred in the homoeologs of an allohexaploid wheat class E mads box gene. Two class E genes have been identified in wheat, i.e., wheat Sepallata (WSEP) and wheat Leafy Hull Sterile1 (WLHS1). The three wheat homoeologs of WSEP show similar genomic structures and expression profiles, whereas the three homoeologs of WLHS1 show genetic and epigenetic alterations. WLHS1 of the A subgenome (WLHS1-A) has a structural alteration that contains a large novel sequence in place of the K domain sequence. A yeast two-hybrid analysis and a transgenic experiment indicated that the WLHS1-A protein has no apparent function. The WLHS1-B and WLHS1-D of the B and D subgenomes, respectively, have an intact MADS box gene structure, but WLHS1-B is predominantly silenced by cytosine methylation. Consequently, of the three WLHS1 homoeoalleles, only WLHS1-D functions in allohexaploid wheat.

Lloyd et al. (2014) suggested that duplicates of meiotic genes return to a single copy following allopolyploidization, more rapidly than the genome-wide average. Therefore, it has been assumed that stabilization upon allopolyploidization of wheat also involved rapid changes in the content and expression of meiotic homoeoalleles. Such a genetic diploidization process would facilitate the correct pairing and synapsis of homoeologs during meiosis. However, the results of Alabdullah et al. (2019) do not support neither extensive gene loss nor changes in homeolog expression of meiotic genes upon wheat allopolyploidization.

In addition, an analysis of the sequences of homoeologous group 1 chromosomes of allohexaploid wheat showed significant deviations from synteny, with many of the non-syntenic genes representing pseudogenes (Wicker et al. 2011). Likewise, the discovery of premature termination codons in 38% of genes expressed in 3A double ditelosomic lines in the genetic background of bread wheat was consistent with ongoing pseudogenization of the wheat genome (Akhunov et al. 2013). Ramírez-González et al. (2018), analyzing genome-wide gene expression patterns in allohexaploid wheat, found expression asymmetries along wheat chromosomes, with homoeoalleles showing the largest inter-tissue, inter-cultivar, and coding sequence variations, most often located in high-recombination distal ends of chromosomes. These transcriptionally dynamic genes potentially represent the first steps toward neo- or sub-functionalization of wheat homoeologs.

Epigenetic alterations of one of the homoeoalleles can be achieved either through epigenetic silencing of one of the homoeoalleles via cytosine methylation, or activation of silenced genes due to their demethylation. Such changes in gene expression were observed in newly formed wheat allopolyploids (Shaked et al. 2001; Kashkush et al. 2002). Epigenetic changes can also result from chromatin modifications or remodeling as well as from alteration in the activity of small RNA molecules. In fact, changes in microRNAs, such as miR168 which targets the Argonaute1 gene, were shown to occur in newly-synthesized hexaploid wheat (Kenan-Eichler et al. 2011). In addition, a high proportion of microRNAs showed non-additive expression upon allopolyploidization, potentially reflecting differential expression of important target genes (Li et al. 2014). These observations may provide insights into dynamic small RNA–mediated homoeologous regulation mechanisms that possibly contribute to heterosis in nascent hexaploid wheat.

New interactions between regulatory factors of the parents, i.e., between the transfactor from one species and the cis or transfactors of the other parental species, may account for the observed inter-genomic suppression in allohexaploid wheat (Galili and Feldman 1984). Genetic suppression or reduction of gene activity can also be caused by DNA elimination (Liu et al. 2008; Kashkush et al. 2002). Methylation and demethylation of retrotransposons, affecting their state of activity (Sabot et al. 2005), were also observed in wheat allopolyploids (Yaakov and Kashkush 2011a, b). Activation of retrotransposons, that constitute most of the DNA of allopolyploids of the wheat group and which are normally transcriptionally silent, may silence or activate neighboring genes (Kashkush et al. 2003). Following allopolyploidization of wheat, the steady-state level of expression of LTR retrotransposons was massively elevated (Kashkush et al. 2002, 2003). This activation may promote either read-in transcripts of the transposon itself, or read-out transcripts into flanking host sequences (Kashkush et al. 2002, 2003; Kashkush and Khasdan 2007). Indeed, in many cases, read-out transcripts were associated with altered expression of adjacent genes, knocking-down or knocking-out the gene product if the read-out transcript was in the antisense orientation relative to the orientation of the gene transcript (such as the iojap-like gene), or over-expressing the gene if the read-out transcript was in the sense orientation (such as the puroindoline-b gene) (Kashkush et al. 2003). A recent study tracking methylation changes around a LTR retrotransposon in the first four generations of a newly formed wheat allopolyploid, indicated that this read-out activity is restricted to the first generations of the nascent polyploid species (Kraitshtein et al. 2010).

Likewise, Yuan et al. (2020) studied genome-wide DNA methylation landscapes in extracted tetraploid wheat (genome BBAA), natural hexaploid wheat from which the tetraploid was extracted, resynthesized hexaploid wheat, natural tetraploid wheat, and the diploid donor of the D subgenome of hexaploid wheat. In the endosperm, these authors found that levels of DNA methylation, especially in CHG (H = A, T, or C) context, were dramatically decreased in the extracted tetraploid relative to its allohexaploid parent. Interestingly, those demethylated regions in extracted tetraploid were remethylated in the resynthesized hexaploid wheat after the addition of the D subgenome. In the extracted tetraploid, hypo-demethylated regions correlated with gene expression, and TEs, dispersed in genic regions of the subgenomes, were demethylated and activated, and thus, may regulate the expression of TE associated genes. The genes that became expressed and TEs that became active in the extracted tetraploid, turned out to be silenced again in the newly synthesized allohexaploid. These dynamic and reversible changes in chromatin and DNA methylation correlate with altered gene expression and TE activity when the D subgenome was added to tetraploid wheat.

Functional diversification of duplicated genes, i.e., differential or partitioned expression of homoeoalleles in different tissues and in different developmental stages, is also a form of genetic diploidization. Botley et al. (2006) reported that differential expression of homoeoalleles in different plant tissues is a common phenomenon in allohexaploid wheat. The activity of several silenced genes could be restored in aneuploid lines, suggesting that no mutation was involved but, rather, new cis–trans interactions or reversible epigenetic alterations took place. Mochida et al. (2006) also presented evidence for differential expression of homoeoalleles in hexaploid wheat and suggested that inactivation of homoeoalleles is a non-random effect.

Allopolyploid patterns of gene expression might be intermediate (between that of the two parents), dominant (similar to one of the parents) or over-dominant (greater than that of the parents). Overdominance can produce novel traits not found in the parents and can be caused by novel cis–trans interactions between regulatory elements of the different genomes that coincide in the same nucleus, as shown in yeast (Tirosh et al. 2009). Many studies on gene expression compared the expression level in the allopolyploid to those of its parents and or to the average of its parents, expressed as the mid-parental value. In hexaploid wheat, Pumphrey et al. (2009) found that approximately 16% of the 825 analyzed genes displayed non-additive expression in the first generation of synthetic allohexaploid wheat. Chague et al. (2010) analyzed 55,052 transcripts in two lines of synthetic allohexaploid wheat and found that 7% of the genes had non-additive expression, while Akhunova et al. (2010) found that about 19% of the studied synthetic allohexaploid wheat genes showed non-additive expression. Li et al. (2014) reported that non-additively expressed protein-coding genes are rare but relevant to growth vigor, and that a high proportion of protein-coding genes exhibit parental expression-level dominance. Similar studies by He et al. (2003) showed that the expression of a significant fraction of genes (7.7%) was altered in the synthetic allohexaploid wheat, T. turgidum-Ae. tauschii, and that Ae. tauschii genes were affected much more frequently than those of T. turgidum. Interestingly, silencing of the same genes was also found in natural T. aestivum, indicating the reversibility of the effect and that the regulation of gene expression is established immediately after allohexaploidization and maintained over generations (He et al. 2003; Chagué et al. 2010). In accord with these results, increased small interfering RNA density was observed for transposable element–associated D-subgenome homoeologs in the progeny of newly formed allohexaploid wheat, which may account for biased repression of D homoeologs (Li et al. 2014). On the other hand, several genes, that are silent in the parental species, became active in the newly formed allohexaploid (He et al. 2003). Similarly, cDNA-AFLP gels also revealed several cDNAs that were expressed only in the allopolyploids and not in the diploid progenitors (Shaked et al. 2001; Kashkush et al. 2002).

The genetic system of the Triticum and Aegilops allopolyploid species facilitates the accumulation of genetic and epigenetic variation, through mutations or gene silencing and through interspecific or intergeneric hybridizations that can lead to introgressions or to new species formation. Such variation is tolerated more readily in allopolyploid wheats than in diploid species of the wheat group (Mac key 1954; Sears 1972). This genetic potential may contribute to genetic variability and to creation of populations with archipelagoes of genotypes, thereby increasing their adaptability, fitness, competitiveness and capacity to colonize newly-opened ecological niches.

12.4 Subgenomic Asymmetry in the Control of Various Traits

Most duplicated genes in the allotetraploid T. turgidum and triplicated genes in the allohexaploid T. aestivum remain active, contributing either to a favorable effect of an extra dosage or to the buildup of positive inter-subgenomic interactions when genes or regulation factors on homoeologous chromosomes are divergent. However, in a small number of loci (about 10–15%), genes of only one subgenome are active, while the homoeoalleles on the other subgenome(s) are either eliminated or partially or fully suppressed by genetic or epigenetic means. For several traits, the retention of some homoeoalleles is not random, with one subgenome favored over the other(s), as observed in many cases [morphological, physiological, molecular and agronomical traits, rRNA genes, storage protein production, interaction with pathogens (reviewed by Feldman et al. 2012)]. The allopolyploids of Triticum and Aegilops were classified into three cytogenetic clusters, with each cluster containing allopolyploids that share a subgenome and differ in the second subgenome(s) (Zohary and Feldman 1962; Table 10.7). The allopolyploids morphologically resemble the diploid donors of the shared (pivotal) subgenomes, namely, the donors of the U, D, and A subgenomes, and differ in other traits, e.g., eco-geographical and tolerance to biotic and abiotic stresses, controlled by the differential genome(s) (Feldman et al. 2012). The contribution of the A and B subgenomes to various traits in wild allotetraploid wheat T. turgidum ssp. dicoccoides was studied by Peng et al. (2003a, b), who found that the A subgenome controls morphological traits, including inflorescence structure, grain shape, free caryopsis, glumes with keels, plant habitus, and growth habit (Table 12.3). This genome also controls the autogamy of allotetraploid wheat (assuming that the donor of the B subgenome is allogamous, i.e., closely related species to Ae. speltoides) and harbors many domestication genes, such as the genes for non-brittle spike on 3AS (Rong 1999; Nalam et al. 2006; Millet et al. 2014), and free-threshing genes on 5AL (Sears 1954). The B subgenome regulates ecological adaptation and tolerance to biotic and abiotic stresses (Peng et al. 2003a, b) and plays a leading role in population adaptation to environmental conditions (Fahima et al. 2006).

Table 12.3 Genome asymmetry in the control of various traits in the wild allotetraploid wheat, T. turgidum ssp. dicoccoides (genome BBAA)*
Full size table

There is evidence for molecular manifestation of genomic asymmetry in the allopolyploid wheat sub-genomes. The level of genetic diversity differs between the two or three subgenomes of the allotetraploid and allohexaploid Triticum species, respectively. The B subgenome exhibits a higher marker polymorphism than the A subgenome in allohexaploid wheat (Chao et al. 1989; Liu and Tsunewaki 1991; Devos et al. 1992), and the wild and domesticated allotetraploid wheat (Liu and Tsunewaki 1991; Huang et al. 1999; Rong et al. 1999). Such differences were most pronounced for loci revealed by gDNA rather than by cDNA probes (Huang et al. 1999; Rong et al. 1999). Similarly, higher polymorphism in the B as compared to A subgenome microsatellites, was seen (Röder et al. 1998). B-subgenome chromosomes are characterized by more C-banding than chromosomes of A and D subgenomes (Gill 1987), reflecting a higher quantity of constitutive repetitive DNA sequences. Similarly, more retrotransposons and variations within them have been observed in the B subgenome, when compared with the A and D subgenomes (K Kashkush, personal communication).

Shaked et al. (2001) reported that sequence elimination in the newly formed allotetraploids Ae. longissima–Ae. umbellulata and Ae. sharonensis–T. monococcum, mainly affect one of the parental genomes. Likewise, a much higher level of sequence elimination, occurring immediately after formation of Triticale, was observed in the rye subgenome as compared to the wheat subgenome (Ma and Gustafson 2008). Shaked et al. (2001) reported that cytosine methylation was also asymmetric; twice as many sequences were affected in T. monococcum, when compared with those of Ae. sharonensis in the nascent amphiploid Ae. sharonensis–T. monococcum. Wicker et al. (2011) identified five times more non-syntenic genes on chromosome arm 1BS of T. aestivum than syntenic genes on the homoeologous arms. They proposed that this accumulation of genic sequences is driven by TE activity, and that these findings indicate that homoeologous wheat chromosomes can exhibit different evolutionary dynamics.

In T. aestivum, the nucleolar organizers of the B subgenome suppress the nucleolar organizers of the A and D subgenomes (Crosby 1957; Crosby-Longwell and Svihla 1960; Darvey and Driscoll 1972; Flavell and O’Dell 1979). Similarly, the nucleolar organizers of the B subgenome suppress those of the A subgenome in allotetraploid T. turgidum (Lacadena et al. 1984; Frankel et al. 1987) and those of the R genome in allotetraploid and allohexaploid triticale, respectively (Darvey and Driscoll 1972; Lacadena et al. 1984; Cermeño et al. 1984a; Martini and Flavell 1985; Appels et al. 1986). Nucleolar dominance of one subgenome was observed in all allopolyploid species of Aegilops (Cermeño et al. 1984b). In these species, the U subgenome from Ae. umbellulata completely suppresses the NOR activity of the M subgenome of Ae. geniculata, the S subgenome of Ae. peregrina, the D subgenome of Ae. juvenalis, and the C subgenome of Ae. triuncialis and that of one pair of the nucleolar organizer chromosomes of the M subgenome of Ae. columnaris, Ae. biuncialis, Ae. juvenalis, and Ae. recta. The nucleolar activity of the D subgenome is completely suppressed by the U subgenome in Ae. juvenalis, and the C subgenome in Ae. cylindrica.

Nucleolus formation is considered evidence for rRNA gene expression and their lack thereof, indicates the absence of rRNA gene transcription (Flavell et al. 1986). Moreover, the relative size of nucleoli within the same nucleus has been taken as a measure of the differential activity between one NOR and another (Flavell et al. 1986). Nucleolar dominance in the allopolyploid species of the wheat group is achieved either by elimination of rRNA-encoding genes, as is the case of rRNA genes on 5AS, or by suppression of their activity. Gustafson and Flavell (1996) and Houchins et al. (1997) noted a correlation between inactivation of rRNA-encoding genes and increased cytosine methylation at their CCGG sites. Further evidence suggesting that nucleolar suppression is triggered by cytosine methylation came from the fact that the suppression of the NORs of genome R was reversed in wheat x rye hybrids and triticale by treatment with the demethylating agent 5-aza-cytosine (Vieira et al. 1990; Neves et al. 1995; Amado et al. 1997).

Chromosomes 1A and 5D of the standard laboratory cultivar Chinese Spring of T. aestivum contain a very small proportion (10%) of the rRNA-encoding genes, while chromosomes 1B and 6B possess 30% and 60% of these genes, respectively (2700 and 5500 copies, respectively) (Mohan and Flavell 1974; Flavell and O’Dell 1976). As a result, chromosomes 1A and 5D are associated with very small nucleoli or none at all in Chinese Spring (Crosby 1957; Crosby-Longwell and Svihla 1960).

Newly synthesized allopolyploids exhibit genetic and epigenetic changes in their rRNA-encoding genes similar to those occurring in natural allopolyploids, indicating that these changes are generated during allopolyploid formation (Baum and Feldman 2010). Likewise, Shcherban et al. (2008) detected rapid elimination of the Aegilops sharonensis rRNA-encoding genes in the synthetic allopolyploid Ae. sharonensis–Ae. umbellulata, which stands in agreement with the pattern in the natural allotetraploids carrying similar genomic combination, i.e., Ae. peregrina and Ae. kotschyi.

Wheat 5S DNA also undergoes immediate changes in response to allopolyploidization, followed by the differential elimination of unit classes of 5S DNA (Baum and Feldman 2010). This elimination was reproducible, i.e., the same unit classes were eliminated in natural and synthetic allopolyploids carrying the same genomic combinations, indicating that no further elimination occurred in the unit classes of the 5S DNA during the life of the allopolyploids.

The high molecular weight (HMW) glutenin subunits are encoded by the Glu-A1 and Glu-B1 gene clusters in allotetraploid wheat, and by the Glu-A1, Glu-B1, and Glu-D1 gene clusters in allohexaploid wheat, located on the long arm of homoeologous-group-1 chromosomes (Payne et al. 1982; Galili and Feldman 1983a, and reference therein). In allohexaploid wheat, each of these gene clusters is composed of two multi-allelic gene loci: Glu-A1-1 and Glu-A1-2 on chromosome 1A, Glu-B1-1 and Glu-B1-2 on chromosome 1B, and Glu-D1-1 and Glu-D1-2 on chromosome 1D. The products of Glu-A1-1, Glu B1-1, and Glu-D1-1 comprise the slow-migrating subunits (x) while those of Glu-A1-2, Glu-B1-2, and Glu-D1-2 comprise the fast-migrating subunits (y). The genetic control of HMW glutenin subunits is another example for subgenome asymmetry. Galili and Feldman (1983b) analyzed 109 different lines of allohexaploid wheat, representing a wide spectrum of genetic backgrounds, and found that 22 lines (20.2%) had no HMW glutenin subunits controlled by chromosome 1A, 44 lines (40.4%) had only one such band and 43 lines (39.4%) had two bands. Moreover, in all lines bearing one subunit controlled by 1A, only the fast-migrating subunit was absent, i.e., only Glu-1A-1 was active, while Glu-A1-2 coding for the fast-migrating band, which is generally active in diploid wheat, was inactive (Waines and Payne 1987).

Likewise, on studying the HMW glutenin subunits in 456 accessions of wild emmer, originating from 21 different populations in Israel, Levy et al. (1988a, b) found that in 82% of these accessions the fast-migrating subunit of the A subgenome was absent, and in 17% of the accessions, the slow-migrating subunit of this subgenome was also absent. Namely, only the genes of the B subgenome were active. In addition, the fast-migrating subunit of the A subgenome was absent in all of the 11 studied lines of the primitive domesticated allotetraploid wheat, T. turgidum subsp. dicoccon. Glu-A1-1, the gene determining the slow-migrating subunit, was also inactive in 16% of the accessions. Thus, in both allotetraploid and allohexaploid wheat, inactivation of HMW glutenin genes is non-random and occurs in the genes of the A subgenome (Galili and Feldman 1983a, b; Feldman et al.1987; Levy et al. 1988a, b, and reference therein). This tendency has also been found among HMW gliadin genes in hexaploid wheat (Galili and Feldman 1983a, b). The order of inactivation was also non-random, starting with the fast-migrating subunits and continuing with the slow-migrating ones.

Galili and Feldman (1984) showed that inactivation of endosperm protein genes is also brought about by inter-subgenomic suppression. Endosperm protein genes, located in the A or the B subgenomes, were repressed by gene(s) of the D subgenome, immediately following the formation of allohexaploid wheat, about 9000 years ago. When the D subgenome was removed, as in the extracted tetraploids (Kerber 1964), these genes became active, indicating that they retained their potential for activity throughout the 9000 years. Similarly, Kerber and Green (1980) described the inter-genomic suppression of a rust resistance gene, located in the D-subgenome, by gene(s) of the A or B subgenomes.

Subgenome asymmetry also occurs in the control of various agronomic traits and of disease and pest resistance in domesticated allopolyploid wheats (Table 12.4). The B and D subgenomes control the most important genes associated with reduced plant height (Rht) and gibberelic acid insensitivity (Ga), yielding dwarf and semi-dwarf wheat, the main types of modern wheat cultivars. Dwarf and semi-dwarf wheat varieties are characterized by an improved harvest index and, consequently, are high-yielding varieties. These varieties have replaced the traditional tall, low-yield varieties in many parts of the world during the ‘green revolution’ and thus, have increased global wheat production. The B and D subgenomes control grain protein content (Law et al. 1978; Joppa and Cantrell 1990) and grain hardness (Morris et al. 1999; Chantret et al. 2005). These subgenomes also control wax production (Tsunewaki and Ebana 1999), an important trait that affects drought tolerance. The B and D genomes are also responsible for tolerance to abiotic stresses, with the B subgenome carrying genes associated with boron tolerance (Paull et al. 1991), low cadmium uptake (Penner et al. 1995), and tolerance to iron deficiency (Maystrenko 1992), and the D subgenome containing gene(s) conferring aluminum tolerance (Riede and Anderson 1996) and response to salinity (Dubcovsky et al. 1996). Most genes for herbicide resistance are located in the B subgenome (Snape et al. 1987), and those responsive to photoperiod and most of those responsive to vernalization are located on the B and D subgenomes. The A subgenome controls plant and spike morphology and the main traits of the domestication syndrome, e.g., non-brittle rachis (Nalam et al. 2006) and free threshing (Sears 1954).

Table 12.4 Genome asymmetry in the control of agronomic traits in durum (genome BBAA) and bread wheat (genome BBAADD)*
Full size table

The B subgenome harbors double the number of disease-resistance genes and resistance-gene analogue (RGA) loci than the A and D subgenomes (Peng et al. 2003b; Fahima et al. 2006). Screening the GrainGenes website (http://wheat.pw.usda.gov/) found that among 184 mapped wheat disease resistance genes, 88 (48%) are located in the B subgenome. Moreover, most genes conferring resistance to stem rust, stripe rust, and leaf rust, the most common wheat diseases, that cause significant global yield loss each year, are located in the B subgenome (Table 12.4).

In allopolyploid species of Triticum and Aegilops, as discussed above, different gene types show a differential propensity for homoeologous change or retention. Genes encoding functional proteins (enzymes) constitute one category of genes that shows a high degree of retention of homoeoalleles (Mitra and Bhatia 1971; Hart 1983a, b, 1987). Such retention enables inter-subgenomic interactions at both the transcriptional level and between gene products, giving rise to functional ‘hybrid’ multimeric enzymes consisting of subunits encoded by different subgenomes. These new heteromeric proteins may have new and desirable properties. Similarly, protein complexes, such as gluten, may also be ‘hybrid’. Moreover, the retention of genes corresponding to trans-acting factors, such as transcription factors, suppressors, and microRNAs, may enable the generation of novel trans interactions that may lead to new expression patterns absent in the diploid parents, as seen in yeast (Tirosh et al. 2009).

For other categories of genes, lack of retention of parental genes or expression patterns, is frequent and non-random. This includes the genes that encode for ribosomal RNA, structural proteins, such as histones and subunits of tubulins, and storage proteins, such as subunits of glutenins and gliadins. In these cases, expression of all homoeoalleles may be redundant, resulting in over-production and even deleterious dose effects. In addition, activity of all homoeoalleles may produce intermediate phenotypes in several traits that decrease the viability of the plants (e.g., hybrid incompatibility genes). Hence, for some traits, control by genes from only one genome (genome asymmetry) may be of higher adaptive value than additive expression, by preventing a genomic clash or avoiding deleterious dosage effects.

Genetic diploidization is not a random process, distinctly affecting specific gene categories and their corresponding traits and forming a clear-cut division of tasks between the constituent subgenomes of allopolyploid wheats. The A-subgenome preferentially controls morphological traits, while the B-subgenome in allotetraploid wheat and the B and D subgenomes in allohexaploid wheat preferentially control the reaction to biotic and abiotic factors. Genetic diploidization may occur during or immediately after allopolyploidization (revolutionary changes), e.g., in rRNA-encoding genes, or through the life history of the species (evolutionary changes), for example, in HMW glutenin genes.

Genome asymmetry may be brought about by either transcriptional dominance of one of the parental genomes (Wang et al. 2006; Flagel and Wendel 2009, 2010; Rapp et al. 2009) or inter-genomic suppression of gene activity (Galili and Feldman 1984), due to incompatibility of regulatory elements (He et al. 2003; Tirosh et al. 2009), chromatin modification (Wang et al. 2006) or suppression of genes adjacent to transposable elements (Kashkush et al. 2003). Differential elimination or inactivation of coding sequences from one of the subgenomes in allotetraploids and from two of the subgenomes in allohexaploids also contributes to the asymmetrical control of the constituent subgenomes (Tate et al. 2006; Feldman and Levy 2009; Buggs et al. 2009, 2010a, b; Koh et al. 2010). Some major transcriptional suppressors, or small non-coding RNAs, such as microRNAs (Ha et al. 2009; Kenan-Eichler et al. 2011), may also have genome-wide effects on asymmetry through the suppression of several targets that, in turn, can affect a cascade of genes, thus leading to asymmetry.

The ability of one subgenome to suppress the activity of genes of another subgenome and thus, fully control a set of traits in allopolyploids, may prevent conflicting gene expression that could potentially lead to defective organ shapes. This protective mechanism ensures the development of viable plants. Diploid species that lack this adaptive ability might fail to produce viable allopolyploids. There are two diploid wheat species, 10 diploid Aegilops species, and one Amblyopyrum species (Eig 1929a, b; van Slageren 1994), most of which have geographical contact with one another (Kimber and Feldman 1987; van Slageren 1994). Many more allopolyploid species, apart from the currently existing ones, may have been generated over the 1–4 million years of the existence of the diploid species (Middleton et al. 2014; Gornicki et al. 2014; Marcussen et al. 2014). Allopolyploids involving the AD, AC, AM, AN, AU, AT, UD, UT, DS, and DT genomic combinations can be produced under artificial conditions but have not been found in nature. It is speculated that inter-genomic incompatibilities leading to reduced fitness, due to failure to produce appropriate genomic asymmetry, hampers their establishment in nature. Rapid progress of structural and functional Triticeae genomics will provide further insights into the mechanisms and functional importance of genomic asymmetry in wheat allopolyploids and other allopolyploids of this group.

To sum up, cytological and genetic diploidization in allopolyploid Triticum and Aegilops species led to the construction of two contrasting and highly important genetic systems that contribute to their evolutionary success: (i) retention of expression of all homoeoalleles of those duplicated or triplicated gene loci whose extra gene dosage has a positive effect by itself or may facilitate the build-up of positive inter-subgenomic interactions between divergent regulation factors, i.e., build up and maintenance of lasting inter-subgenomic favorable genetic combinations (inter-genomic heterosis), and (ii) elimination or suppression of genes from one subgenome in allotetraploids and from two subgenomes in allohexaploids, whose extra dosage or new inter-genomic interactions are deleterious, thus bringing about genome asymmetry for various traits. The latter process may be tissue-specific (Buggs et al. 2011).

Inter-genomic pairing would have led to disruption of the linkage of the homoeoalleles that contribute to positive inter-subgenomic interactions and would have led to segregation of genes that participate in the control of certain traits by a single subgenome. Inter-subgenomic recombination would therefore result in many intermediate phenotypes that may negatively affect the functionality, adaptability and stability of the allopolyploids.

12.5 Evolution During the Life of the Allopolyploids

Allopolyploid species also undergo structural genomic changes during the lifetime of the taxon (evolutionary changes; Table 12.2), generating a new variation that scarcely exists in the diploid species. Allopolyploids harboring two or more different subgenomes within each nucleus, may facilitate inter-genomic horizontal transfer of chromosomal segments, transposable elements or genes. Inter-subgenomic invasion of chromatin segments from the B subgenome into the A subgenome was demonstrated by FISH in T. turgidum ssp. dicoccoides (Belyayev et al. 2000). Cytogenetic studies have shown that several inter-subgenomic translocations occur in the allopolyploids of the wheat group (Maestra and Naranjo 1999, and reference therein). Moreover, in contrast to most diploids, which are genetically isolated from each other and have undergone divergent evolution, allopolyploids in the wheat group exhibit convergent evolution because they contain genetic material from two or more different diploid genomes and can exchange genes with each other via hybridization and introgression, resulting in the production of new subgenomic combinations (Zohary and Feldman 1962).

The presence of duplicated or triplicated genetic material in allotetraploids and allohexaploids, respectively, has relaxed constraints on the function of the multiple genes, enabling, in the long run, continued genetic diploidization, achieved by silencing of one of the duplicated or triplicated genes or divergence of one homoeologous locus to a new function. Thus, the accumulation of genetic variation through mutations is more readily tolerated in allopolyploid than in diploid species (Mac Key 1954a; Sears 1972; Dubcovsky and Dvorak 2007).

The evolutionary changes might also occur in an accelerated manner, thanks to the buffering of mutations in the polyploid background that leads to rapid neo- or sub-functionalization of genes and to a further process of diploidization and of divergence from the diploid progenitor genomes. Akhunov et al. (2013) uncovered a high level of alternative splicing pattern divergence between the duplicated homoeologous copies of genes in common wheat. Their observations are consistent with the accelerated accumulation of alternative splicing isoforms, nonsynonymous mutations, and gene structure rearrangements in the wheat lineage, likely due to genetic redundancy created by allopolyploidization (Akhunov et al. 2013). While these processes mostly contribute to the degeneration of a duplicated genome and its diploidization, they have the potential to facilitate new functional variations, which, upon selection in the evolutionary lineage, may play an important role in the development of novel traits (Akhunov et al. 2013).

According to Stebbins (1950), newly formed allopolyploids are often characterized by limited genetic variation, a phenomenon he referred to as the “polyploidy diversity bottleneck.” This bottleneck arises because only a few diploid genotypes were involved in the allopolyploid formation events, because the newly formed allopolyploid is immediately reproductively isolated from its two parental species and because time was not sufficient for the accumulation of mutations. However, despite this diversity bottleneck and despite the fact that all Aegilops and Triticum allotetraploids were formed much later than their ancestral diploids, e.g., during 1.3 MYA (Middleton et al. 2014; Gornicki et al. 2014; Marcussen et al. 2014), and allohexaploids were formed even more recently, most of the allopolyploids of the wheat group display greater genetic variation than their diploid progenitors (Zohary and Feldman 1962). It is most likely that the allopolyploid species were recurrently formed from different genotypes of their parental diploids, thus increasing their intra-specific variation. As a matter of fact, at least in Ae. cylindrica, there is evidence of multiple allopolyploidization events since some forms contain the D cytoplasm while others contain the C cytoplasm (Caldwell et al. 2004; Gandhi et al. 2005). Moreover, allopolyploidy enables genome plasticity that in turn allows for accelerated evolution to take place, as observed in allopolyploid wheats (Dubcovsky and Dvorak 2007) and as was recently shown in experimental evolution studies in yeast (Selmecki et al. 2015).

One way in which the gene pool of the allotetraploid species was greatly enlarged and their evolutionary potentiality correspondingly increased, was by hybridization and introgression between related allotetraploids (Zohary and Feldman 1962). On the basis of plant habitus, spike morphology, and cytogenetic data, these authors classified the allopolyploid species of Aegilops and Triticum into three natural clusters (Table 12.1). Genome analysis of the allopolyploids within each cluster showed that they share one unaltered subgenome (the pivotal subgenome) and a subgenome(s) that is/are modified [the differential subgenome(s)]. In laboratory hybridization studies of allopolyploids, it was found that the homologous chromosomes of the shared subgenome paired ensuring some seed fertility following pollination by one of the parents (Feldman 1965b), while, the F1 hybrids having 9–10 bivalents (7 bivalents involving chromosomes of the shared subgenome and 2–3 between chromosomes of the differential subgenomes) at meiosis, show that the chromosomes of the differential subgenomes, brought together from different parents, may pair to some extent and exchange genes (Feldman 1965c). Consequently, the differential subgenomes of these allopolyploids are recombinant subgenomes containing chromosomal segments that originated from two or more diploid genomes. Such genomic constitution reveals different evolutionary rates for each of the two or three subgenomes of every allopolyploid.

Thus, all seven alltetraploids and the one allohexaploid of the U-subgenome cluster share a genome homologous to that of diploid Aegilops umbellulata (Table 12.1; Kihara 1954), a weedy annual in the center of the genus distribution area, which possesses an unusually efficient method of seed dispersal (umbrella-type) in the form of a small spike with large number of awns on its glumes and lemmas. Seven distinct allotetraploid species contain a subgenome derived from Ae. umbellulata, and a second subgenome that derived from ancestral diploids which have been variously modified during evolution at the allotetraploid level (the modified subgenomes). These allotetraploids are aggressive weeds, the most common of which have become widespread in the Mediterranean basin.

Similarly, all the three allotetraploids and three allohexaploids of the D-subgenome cluster share a subgenome homologous to that of diploid Ae. tauschii (Table 12.1; Kihara 1954; Kihara et al. 1959), that grows in the eastern region of the genus distribution area (with the exception of Ae. ventricosa, which grows in western Mediterranean regions), and has a barrel-type dispersal unit. The two allotetraploids and two allohexaploids of the A-subgenome cluster, including all the wild and domesticated forms, sharing a subgenome closely related to that of diploid Triticum urartu (Table 12.1; Dvorak 1976; Chapman et al. 1976; Li et al. 2022), which grows in the center of the genus distribution area.

The allopolyploids of each cluster exhibit the same basic morphology (stature, leaf shape, and spike and spikelet morphology) seed dispersal unit structure of the diploid donor of the shared genome. They differ in features of the differential subgenome(s) that are primarily responsible for the eco-geographical adaptation of the various allopolyploid species in the cluster. Thus, inter-specific hybridization has played a decisive role in the production of a wide range of genetic variation in these allopolyploid species and probably significantly contributed to their evolutionary success.

The wild allotetraploid forms of the A-subgenome cluster. T. turgidum ssp. dicoccoides and T. timopheevii ssp. armeniacum, differ from the other species of the two subgenome clusters, by exhibiting relatively low morphological variation and distribution in a comparatively small area. One possible explanation for these differences is the fact that ssp. dicoccoides, formed 700,000–800,000 years ago (Gornicki et al. 2014; Marcussen et al. 2014), was, for many years, the only allotetraploid taxon in the A-subgenome cluster. The second wild form, ssp. armeniacum, was formed 300,000–400,000 years after ssp. dicoccoides (Table 10.7; Gornicki et al. 2014). Thus, sporadic hybridization and introgression between the two wild taxa of the A-subgenome cluster could only have occurred after the formation of ssp. armeniacum, and obviously, also after the migration of wild emmer to the northern part of the Fertile Crescent. Indeed, evidence for introgression between these two wild allotetraploid wheats, that currently grow in mixed stands in many locations of the northern part of the Fertile Crescent, were presented by several researchers. Genotypes of wild T. turgidum and wild T. timopheevii carrying recombinant B or G subgenomes were found by Sachs (1953), Wagenaar (1961, 1966), Rao and Smith (1968), Rawal and Harlan (1975), and Tanaka and Kawahara (1976). Cytogenetically, the Turkish–Iraqi race of wild T. turgidum shows a range of chromosome pairing capabilities, from complete affinity to T. turgidum to high affinity to T. timopheevii (Sachs 1953; Rao and Smith 1968; Rawal and Harlan 1975; Tanaka and Kawahara 1976). Likewise, the study of Gornicki et al. (2014) provides molecular evidence that evolution of these allopolyploid wheats was punctuated by chloroplast introgression. They reported that some of the chloroplast haplotypes of the Turkish–Iraqi race of T. turgidum are similar to haplotypes of T. timopheevii, suggesting an introgression from the latter (Gornicki et al. 2014). On the other hand, an accession of T. turgidum ssp. dicoccoides, for example, which shows high chromosome pairing with both T. turgidum and T. timopheevii (Rawal and Harlan 1975), carries the T. timopheevii chloroplast haplotype as a result of a cross between wild emmer and wild timopheevii. Conversely, one wild timopheevii accession carries the wild emmer chloroplast haplotype. Consequently, the northern ssp. dicoccoides presumably differs from the southern one by chromosomal segments of the G subgenome, whereas the B subgenome of the southern form may have introgressed with diploid Aegilops species of subsection Emarginata, mainly, Ae. searsii and Ae. longissima, and by the allotetraploid Ae. peregrina, which are common in the southern Levant. These inter-lineage and inter-generic introgressions are consistent with both allopolyploid wheat lineages sharing a weak Ph1 locus (Ozkan and Feldman 2001).

In sharp contrast to the rarity of inter-specific hybridization at the diploid level, hybridization between allotetraploid species, particularly between those sharing a common subgenome, is common (Zohary and Feldman 1962; Feldman 1965a). The shared subgenome both increases the compatibility between the parents and the ease with which viable, fertile derivatives can be obtained by introgression (Feldman 1965b). In addition, the diploid species that contributes the common subgenome may possess some particular combination of adaptive characteristics which it transmits to all of its hybrids. Such allotetraploid species tend to grow in mixed stands, and many F1 hybrids as well as backcrossed progeny have been repeatedly found in many localities in Israel, Turkey and Greece (Zohary and Feldman 1962; Feldman 1965a). Progeny raised from such hybrids segregate widely and recover almost complete fertility within one or two generations. Because of self-pollination, which is predominant in these species, these fertile introgressed genotypes become genetically fixed with relative ease and, if they are of adaptive value, they may enlarge the ecological amplitude and increase the gene pool of the concerned species. Of note, self-fertilization, while predominant in these species, is never complete; therefore, the introgressed genotypes, if inter-fertile with other individuals of the species concerned, can be regarded as part of its gene pool (Stebbins 1971). Yet, because of cytoplasmic divergence that may cause male sterility, introgression goes in the direction of the original female species. The occurrence of populations comprising several allotetraploid species in numerous habitats, increases the chance of likelihood of introgression between these species.

Additional evidence for the existence of introgressed genomes in allotetraploid Aegilops was obtained from C-banding analysis (Badaeva et al. 2004). An introgression of a DNA sequence from allohexaploid bread wheat to the allotetraploid Aegilops species, Ae. peregrina, was recently described (Weissmann et al. 2005). Hence, hybridization between allotetraploids, particularly between those sharing one common subgenome, and, to a lesser extent, between other allotetraploids, facilitates the rapid buildup of genetic variability at the tetraploid level. The differential subgenomes that are genetically and geographically isolated from one another at the diploid level, where emphasis is on divergence and specialization, are brought together and allowed to recombine at the tetraploid level. The ability to exchange genetic material through spontaneous inter-specific hybridization, further promotes the convergent evolution of these species. This reticulate pattern of evolution tends to produce a range of morphological intermediates between allotetraploids and blurs the morphological boundaries between them, rendering it difficult to define the species border. Natural hybrids also occur between wild species and domesticated allopolyploid wheats. Hybrids have been recurrently recorded in T. turgidum, e.g., between subsp. durum and its progenitor, wild emmer (subsp. dicoccoides, in Israel (Percival 1921; Jakubziner 1932; Huang et al. 1999; Feldman 2001; Dvorak et al. 2006b; Luo et al. 2007), between wild emmer and Aegilops species (Cook 1913; Percival 1921), and between common wheat and wild emmer (Zohary and Brick 1962). On the basis of genome sequence analysis, recent studies have confirmed that a high rate of introgression took place from wild emmer wheat into the background of domesticated emmer wheat or of bread wheat, leading to a higher variability in these genomes (He et al. 2019 and Zhou et al. 2020). For example, the variation in domesticated-emmer genome was estimated to include ~ 73% that of wild emmer (Zhou et al. 2020; Sharma et al. 2021; Keilwagen et al. 2022).

Substantial evidence of spontaneous hybridization between allotetraploids and diploid species in mixed natural populations of tetraploids and diploids was also obtained (Vardi and Zohary 1967; Vardi 1973; Zohary and Feldman, unpublished). The occurrence of hybrid derivatives in such populations, particularly as a result of backcrossing to the allotetraploid parents, implies the possibility of gene flow from the diploid to the tetraploid level.

The diploid species underwent divergent evolution and consequently, have diverse genomes that are, more or less, genetically isolated from one another. Most exhibit relatively limited morphological, cytological, and molecular variation, specialization in their spike and dispersal unit structure, occupation of few well-defined ecological habitats, and distribution throughout relatively small geographical areas (Zhukovsky 1928; Eig 1929a; Zohary and Feldman 1962; Kimber and Feldman 1987; Van Slageren 1994).

In contrast, the allopolyploids, comprising two or three diverse subgenomes in one nucleus, underwent genomic convergence. Nevertheless, most exhibit wider morphological, cytological, and molecular variation than their diploid parents, occupy a greater diversity of ecological habitats, and are distributed over larger geographical area than their diploid progenitors (Zhukovsky 1928; Eig 1929a; Zohary and Feldman 1962; Kimber and Feldman 1987; Van Slageren 1994). The distribution areas of most of the tetraploids overlap, completely or partly, and extend beyond those of their two diploid parents. They grow well in a very wide array of edaphic and climatic conditions and so do not show the marked ecological specificity of the diploids. Their weedy nature is reflected in their ability to rapidly and efficiently colonize a variety of newly disturbed and secondary habitats. Undoubtedly, the expansion of agriculture and the opening up of many new habitats played a key role in the massive distribution of these allopolyploid species throughout the range of the group (Zohary and Feldman 1962; Kimber and Feldman 1987).

Therefore, students of wheat evolution have been fascinated by this paradox of “polyploidy diversity bottleneck” on the one hand and great genetic diversity, on the other, and have dedicated their research efforts to unraveling the processes and mechanisms that contributed to the rapid buildup of genetic diversity in the allopolyploids and to their great evolutionary success in term of proliferation and adaptation to new habitats, including under domestication.

In contrast to the wide distribution of the allotetraploid species, the distribution area of the natural allohexaploid species is, in all cases, smaller than that of their tetraploid and diploid parents. Also, their ecological amplitudes are much more restricted than those of the related tetraploids and even of the diploid parents. They grow in a smaller range of habitats. The morphological variation of the allohexaploids is also relatively limited. All these indicate a relatively recent origin of the allohexaploid species. T. aestivum ssp. aestivum, which was formed under cultivation, exhibits tremendous variation, due both to spontaneous hybridization with wild relatives and to modern breeding, and is currently grown in large parts of the world, and is exceptional in this respect.