A Sense of Self

Polyploid organisms contain a number of chromosomes that is some multiple of x (basic chromosome number) greater than the 2x content of related diploid species. Winkler (1916)introduced the term in a study of vegetative grafts in Solanum, when he discovered tetraploid plants regenerated from callus tissue on cut stems of diploid S. nigrum. Winge (1917)observed that interspecific hybrids give rise to hybrid plants that are sterile as a result of the failure of correct chromosome pairing at meiosis; he proposed that spontaneous chromosome doubling, resulting from accidental somatic doubling in mitosis or nonreduction in meiosis, could convert such hybrids into a fertile type. As illustrated by these two early reports, polyploidy can arise from a doubling of a single genome (autopolyploidy) or from the merger of two distinct genomes after hybridization (allopolyploidy). Since then, polyploidy has been found to be very common in plants. Polyploidization is currently viewed as a highly dynamic process and a major force in the evolution of higher plants (Soltis and Soltis, 1995). Wendel (2000)stated that certainly 50% and perhaps more than 70% of angiosperms have experienced chromosome doubling at some point in their evolutionary histories, and Grant (1981)reported an estimated frequency of as high as 95% among pteridophytes. Genomic analysis of a growing number of organisms may produce further upward revisions of these numbers as numerous previously recognized diploids, includ-ing Arabidopsis (Grant et al., 2000), maize (Gaut and Doebley, 1997), soybean (Shoemaker et al., 1996), cotton (Muravenko et al., 1998), and sorghum (Gómez et al., 1998), are discovered to have undergone ancient rounds of chromosome doubling followed by gene loss. Wendel (2000)wrote that it “is difficult to overstate the importance of doubling in the evolutionary history of flowering plants.” Polyploidy also occurs in vertebrates, some well-known examples being among frogs, fish, and chickens. Spring (1997)provided a convincing argument that humans and other vertebrates may be ancient polyploids; a survey of more than 50 gene families from aldolases to zinc finger transcription factors showed that usually a single invertebrate gene corresponds to up to four equally related human genes on different chromosomes. The ancient polyploid nature of numerous diploid organisms has been revealed only by genomic approaches, such as genome sequencing and comparative mapping, because polyploids appear to undergo a process of “diploidization” along with extensive gene loss or inactivation and chromosomal rearrangements that make ancient rounds of chromosome doubling difficult to detect. As Winge observed in 1917, two sets of chromosomes in a diploid hybrid may be sufficiently different from one another that they fail to pair correctly at meiosis. Chromosome doubling and the creation of a polyploid thus solves the immediate problem of meiotic pairing of homologous chromosomes in a hybrid organism; each chromosome can pair with its own duplicate. However, the polyploid organism faces another immediate problem during meiosis: the two sets of homeologous chromosomes (i.e., from different parental stock) may be sufficiently similar to one another that pairing of homeologous chromosomes may occur, disrupting the correct pairing of the new truly homologous (i.e., duplicated) chromosomes. The process of dip-loidization may serve to accentuate differences between homeologous chro-mosomes in a polyploid organism, facilitating correct pairing of homologous chromosomes during meiosis (Feldman et al., 1997). If this is the case, we might expect to see evidence of substantial diploidization occurring within the first generation after polyploidization. Evolution of Hexaploid Bread Wheat T. aestivum and Other Wheat Cultivars. The S genome of Ae. speltoides or of another closely related Aegilops species is the closest known relative of the B genome. Thus, B was either contributed by an unknown and now extinct genome, or it diverged from the S genome progenitor after polyploidization. Adapted from Feldman (2001). Evolution of Hexaploid Bread Wheat T. aestivum and Other Wheat Cultivars. The S genome of Ae. speltoides or of another closely related Aegilops species is the closest known relative of the B genome. Thus, B was either contributed by an unknown and now extinct genome, or it diverged from the S genome progenitor after polyploidization. Adapted from Feldman (2001). Allopolyploidy in Wheat. The wheat spike shown in the center is from a newly synthesized allopolyploid with a genomic combination (SlSlAmAm) similar to that of durum wheat (genome BBAA). It is flanked by its two-diploid parent: Ae. sharonensis (genome SlSl) on the left and T. monococcum ssp aegilopoides (genome AmAm) on the right. The articles by Ozkan et al. and Shaked et al. describe the genetic and epigenetic rearrangements that occur in F1 interspecific and intergeneric hybrids between various species of the wheat (Aegilops and Triticum) group and in the subsequent generations (S1, S2, and S3) of the derived amphiploids. DNA elimination and methylation were found to be rapid and major responses of the genome to wide hybridization and to allopolyploidy in wheat. (Image kindly provided by Avi Levy.) Allopolyploidy in Wheat. The wheat spike shown in the center is from a newly synthesized allopolyploid with a genomic combination (SlSlAmAm) similar to that of durum wheat (genome BBAA). It is flanked by its two-diploid parent: Ae. sharonensis (genome SlSl) on the left and T. monococcum ssp aegilopoides (genome AmAm) on the right. The articles by Ozkan et al. and Shaked et al. describe the genetic and epigenetic rearrangements that occur in F1 interspecific and intergeneric hybrids between various species of the wheat (Aegilops and Triticum) group and in the subsequent generations (S1, S2, and S3) of the derived amphiploids. DNA elimination and methylation were found to be rapid and major responses of the genome to wide hybridization and to allopolyploidy in wheat. (Image kindly provided by Avi Levy.) Previous work (Feldman et al., 1997; Liu et al., 1998b) showed that sequence elimination occurs very early (within the first few generations) after allopolyploidization in the wheat group. In these earlier studies, 3- to 6-year-old synthetic allopolyploids were compared with their diploid parental accessions rather than with the exact parental plants. Ozkan et al. analyzed diploid parental plants, F1 progeny, and the first three generations (S1, S2, and S3) of synthetic allopolyploids obtained from hybridization of several species of Aegilops and Triticum, allowing a more precise determination of the timing and rate of genomic change. Differential sequence elimination between the parental genomes in an allopolyploid within the first generation would have the immediate effect of increasing divergence between homeologous chromosomes. The hypothesis of Feldman et al. (1997)that nonrandom sequence elimination leads to diploidization and a more well-behaved meiotic system is attractive and appears to fit the data from wheat (Ozkan et al., 2001; Shaked et al., 2001) and Brassica species (Song et al., 1995). Feldman et al. (1997)postulated that this mechanism of diploidization operates independently of a second well-known mechanism by which the Ph1 locus controls homeologous chromosome pairing in allopolyploid wheat. The activity of the Ph1 locus suppresses homeologous pairing in polyploid wheat, and ph1 mutant plants show high levels of homeologous recombination (Luo et al., 1996). Ozkan et al. examined allopolyploids produced by crossing wild-type (Ph1Ph1) and mutant (ph1ph1) lines of hexaploid wheat with diploid Ae. longissima. Sequence elimination in these allooctoploids was reduced compared with that in synthetic tetraploids or hexaploids, but there was no difference in the pattern of sequence elimination between allooctoploids carrying or lacking Ph1. The polyploids examined in these studies were produced by colchicine treatment of the diploid F1 hybrids. Colchicine is an antimitotic agent that arrests mitosis in metaphase by preventing the formation of spindle fibers; it has long been recognized as an inducer of polyploidy in plants. Thus, the question is raised whether the characteristics of synthetic polyploids are shared by naturally occurring polyploids. The study by Ozkan et al. included several polyploids that arose spontaneously (without colchicine treatment) after hybridization, and no significant differences were observed in the pattern of sequence elimination in these plants versus the synthetic polyploids. Also, the pattern of elimination in synthetic allohexaploids was similar to that of the natural hexaploid wheat. Interestingly, the patterns of sequence elimination were different in synthetic polyploids that did not have a genomic makeup analogous to any naturally occurring allopolyploid. Sequence elimination was found to start earlier in the synthetic allopolyploids that resembled naturally occurring allopolyploids (Ozkan et al., 2001). Although intriguing, these results raise many more questions than they answer, and the mechanisms and bases of sequence elimination remain elusive. It has been observed that synthetic allopolyploid plants tend to be phenotypically and genetically unstable, in contrast to wild and cultivated allopolyploids (Comai, 2000). Rapid sequence elimination and consequent diploidization may be part of a mechanism for stabilizing allopolyploid genomes. In a previous study of allopolyploidy, Song et al. (1995)observed extensive and rapid genome changes after the production of synthetic polyploids derived from various combinations of three Brassica species, and concluded that homeologous recombination was a likely cause of the changes. The results of Song et al. (1995)suggested that the frequency of genomic change associated with polyploidization is correlated positively with the degree of divergence between parental diploid genomes (see also Soltis and Soltis, 1995). Ozkan et al. and Shaked et al. failed to observe this characteristic in the allopolyploid wheat progeny. It should be noted that Song et al. (1995)examined polyploids derived from only two sets of crosses: B. rapa (A genome) used as the male or the female parent with B. nigra (B genome) to produce AB and BA polyploids, and B. rapa as the male or female parent with B. oleracea (C genome) to produce AC and CA polyploids. Of the 80 to 90 sequence fragments examined, the AB and BA polyploids showed about twice the number of changes as did the AC and CA polyploids, and the A and B genomes were more highly divergent than the A and C genomes. Although these data are consistent with the idea that a higher degree of sequence divergence gives rise to a greater frequency of genomic change in the new polyploid, we should remember that this represents a rather small data set and that generalizations may be premature. Ozkan et al. examined eight DNA sequences in a total of 35 interspecific and intergeneric F1 hybrids and 22 derived polyploids from numerous Aegilops and Triticum species, whereas Shaked et al. analyzed a large number of loci (more than 3600 AFLP fragments) in diploid and tetraploid progeny of a small number of crosses. These studies did not find a correlation between the frequency of genomic change in polyploid wheat and the degree of divergence between the parental lines. Song et al. (1995)also found that the cytoplasmic donor may play an important role in the formation of the polyploid, such that DNA sequence was preferentially eliminated from the genome corresponding to the male parent in some cases. In contrast, Ozkan et al. found that the nature of the cytoplasmic donor did not affect the pattern of sequence elimination in polyploids arising from crosses between two wheat species, T. aestivum ssp aestivum and Ae. speltoides, that were tested for this effect. In another cross between Ae. sharonensis and Ae. umbellulata, 14% of loci from the cytoplasmic (female) parental genome of Ae. sharonensis were eliminated compared with only 0.5% for the male parental genome of Ae. umbellulata. The reasons for this nonrandom pattern of sequence elimination are unknown. It might be of interest to compare these genomic changes with those produced from a reciprocal cross using Ae. umbellulata as the cytoplasmic donor and Ae. sharonensis as the pollen donor. In contrast to the results of Song et al. (1995), Axelsson et al. (2000)could find no evidence of rapid genome change in synthetic and natural B. juncea polyploids (AABB genome) using restriction fragment length polymorphism linkage analysis. While they could not rule out the possibility that such changes occurred, their results strongly suggested that homeologous recombination was minimal in the Brassica polyploids. In a recent study of cotton (Gossypium), Liu et al. (2001)analyzed six sets of synthetic allopolyploids using AFLP and MSAP and found no evidence of rapid genomic change among some 22,000 loci. In that study, only two generations of allopolyploids were examined, but the nearly complete absence of sequence elimination and of changes in methylation were in striking contrast to the widespread changes observed within the first two generations of polyploidy in wheat by Ozkan et al. and Shaked et al. Clearly, there is much more to be learned about the mechanisms of rapid genomic change in wheat and possibly Brassica allopolyploids on the one hand and the mechanisms for the maintenance of a relatively quiescent genome in cotton allopolyploids on the other hand. An alteration in cytosine methylation patterns was a second major response to allopolyploidization observed within the first few allopolyploid wheat generations (Shaked et al., 2001). Alterations in cytosine methylation occurred in ∼13% of the loci examined with MSAP analysis, and methylation was found to affect both repetitive and low-copy DNA sequences and apparently was independent of sequence elimination. Liu et al. (1998a)found that methylation was a contributing factor to changes that occurred in low-copy coding sequences in newly synthesized wheat allopolyploids. Methylation has been shown to be associated with some types of gene silencing (Holliday and Ho, 1998). Typically, hypermethylation is associated with gene inactivation, although gene activation could result from hypermethylation of a negative regulator of transcription (Comai, 2000). The silencing of one parental set of ribosomal RNA (rRNA) genes in an interspecifc hybrid, termed “nucleolar dominance,” is a widespread phenomenon in plants. Chen and Pikaard (1997)showed that silenced rRNA genes in Brassica allotetraploids are maintained by DNA methylation and histone deacetlyation. Gene silencing has been found to be a rapid response of Brassicaceae genomes to allopolyploid formation. Using AFLP analysis, Comai et al. (2000)estimated that at least 0.4% of the genes in Arabidopsis suecica, an allotetraploid of A. thaliana and Cardaminopsis arenosa, were silenced within one or two generations of polyploidization. Although this may seem like a small number, it should be remembered that changes in the expression of a small number of genes, particularly regulatory genes, have the potential to cause enormous changes in the development of an organism. Lee and Chen (2001)extended this work and identified a variety of genes, including rRNA and transcription factor genes, that are silenced in polyploid A. suecica in a methylation-dependent manner. Gene silencing associated with methylation is believed to have evolved as a defense response mechanism against invasive transposons (Matzke and Matzke, 1998). Comai (2000)suggested that gene silencing in allopolyploids may occur in genes that contain transposon-like sequences as a result of the activation of this plant defense mechanism. Wendel (2000)discussed three primary evolutionary fates of duplicated genes in polyploid organisms: gene silencing, functional diversification, and the retention of original or similar function. Gene silencing would encompass both genetic (e.g., mutation, sequence elimination) and epigenetic (e.g., methylation, chromatin structure) phenomena, and functional diversification would include changes in noncoding regulatory regions that affect transcription. Divergence of function among duplicated genes is widely considered to be a reason for the evolutionary success of polyploids. However, Wendel (2000)noted that there is very little concrete evidence of functional divergence of genes occurring after polyploidization. This area is sure to garner more attention in the near future. Recent work in comparative genomics and the study of polyploidization, such as that of Ozkan et al. and Shaked et al., has engendered a radical change in our views of genome size and organization. For example, Bennetzen and Kellogg (1997)concluded that plants may have a “one-way ticket” to larger genome sizes via amplification of retrotransposons, based on the phylogeny of some diploid grasses of known genome size. These authors noted that the grasses examined were all diploids, according to the best evidence available, so that increases in genome size did not appear to be associated with polyploidy. Furthermore, it was difficult to envision a process whereby repetitive elements could be removed from genomes at a rate that would balance or reverse the trend of retroelement amplification. Because interspersed repetitive DNA sequences make up the majority of the repetitive DNA in complex plant genomes, a removal process must be able to excise these sequences without removing adjacent plant genes (Bennetzen and Kellogg, 1997). However, we must now consider the possibility that many of these “diploid” species are derived from ancient polyploids. And the work of Ozkan et al. (2001), like that of Song et al. (1995), shows that allopolyploidy-induced sequence elimination in noncoding regions (albeit low-copy sequences that were examined in these reports) can occur rapidly in a sizable fraction of various plant genomes. The idea of the “dynamic genome” has been a paradigm in genetics since the confirmation and widespread acceptance, in the late 1970s and early 1980s, of McClintock's discovery—some 30 years earlier—of transposable elements, but we are perhaps only beginning to appreciate the extreme dynamic range, the rates of change that are possible, and the various mechanisms for promoting change or maintaining homeostasis within genomes.