Molecular Genetic Features of Polyploidization and Aneuploidization Reveal Unique Patterns for Genome Duplication in Diploid Malus

Polyploidization results in genome duplication and is an important step in evolution and speciation. The Malus genome confirmed that this genus was derived through auto-polyploidization, yet the genetic and meiotic mechanisms for polyploidization, particularly for aneuploidization, are unclear in this genus or other woody perennials. In fact the contribution of aneuploidization remains poorly understood throughout Plantae. We add to this knowledge by characterization of eupolyploidization and aneuploidization in 27,542 F1 seedlings from seven diploid Malus populations using cytology and microsatellite markers. We provide the first evidence that aneuploidy exceeds eupolyploidy in the diploid crosses, suggesting aneuploidization is a leading cause of genome duplication. Gametes from diploid Malus had a unique combinational pattern; ova preserved euploidy exclusively, while spermatozoa presented both euploidy and aneuploidy. All non-reduced gametes were genetically heterozygous, indicating first-division restitution was the exclusive mode for Malus eupolyploidization and aneuploidization. Chromosome segregation pattern among aneuploids was non-uniform, however, certain chromosomes were associated for aneuploidization. This study is the first to provide molecular evidence for the contribution of heterozygous non-reduced gametes to fitness in polyploids and aneuploids. Aneuploidization can increase, while eupolyploidization may decrease genetic diversity in their newly established populations. Auto-triploidization is important for speciation in the extant Malus. The features of Malus polyploidization confer genetic stability and diversity, and present heterozygosity, heterosis and adaptability for evolutionary selection. A protocol using co-dominant markers was proposed for accelerating apple triploid breeding program. A path was postulated for evolution of numerically odd basic chromosomes. The model for Malus derivation was considerably revised. Impacts of aneuploidization on speciation and evolution, and potential applications of aneuploids and polyploids in breeding and genetics for other species were evaluated in depth. This study greatly improves our understanding of evolution, speciation, and adaptation of the Malus genus, and provides strategies to exploit polyploidization in other species.

Apple is an economically important crop world-wide [7], [28]- [33].Polyploidy was identified in apples seventy years ago; several commercial cultivars are triploids [28], [32].Root-tip chromosome count is reliable for cytotype detection but not informative for meiotic analysis [26], [32].Microsatellite markers have been successfully applied to genetic and parentage studies [29], [30], [34], but not previously used for cytotypic analysis.Microsatellite markers are co-dominant, highly polymorphic, reproducible, and transferable [29], [30], [34].The Malus 17 linkage groups (LG) correspond to its 17 chromosomes [7], [31].These advantages of microsatellites enable accurate study of the cytotype genetics and characterization of polyploidization in apples.Molecular phylogenetics accompanying the Malus genome confirmed derivation of Malus species by auto-polyploidization [7], however, detailed mechanisms of polyploidization in this genus (or in other perennial woody species) are unknown.Several models have been postulated for Malus polyploidization based on comparison to the mechanisms described for the herbaceous species [28], [32].We provide empirical data to test these models integrating cytology and molecular genetic characterization.This study seeks to characterize the details of polyploidization in Malus with particularly emphasis on aneuploidization.These data not only add to our knowledge of the evolution and domestication history of the genus, but also are critical in developing breeding strategies and molecular tools for apple breeding programs.

Meiotic mechanisms for polyploidization in Malus
Meiotic mechanisms of non-reduced gametes for polyploidization were previously uncertain in the diploid Malus species, for example, whether from first-division restitution (FDR), or from second-division restitution (SDR), or from both [28], [32].But the distributional features of markers in the non-diploid seedlings showed that all diploid and aneuploid gametes with a cytotype of $n+1 were genetically heterozygous (Table S1, S2, S3, S4, S5, S6, S7, S8, S9, S10, S11).Thus, FDR was the exclusive mechanism for non-reduction of eudiploid gametes and aneuploid spermatozoa (with a cytotype $n+1), perhaps by partial non-disjunction [1], [9], [14].An alternative mechanism may be a two-step process of restitution in the first-division followed by non-disjunction in the second-division, but concurrent diploidization and aneuploidiza- The percentage was calculated based on the summary data in the Table 2; Six aneuploid seedlings with cytotype of 2n21 were found in the seven crosses, and among them two seedlings were affected by the LG02 (See Table 2), thus the percentage of LG02 was approximately estimated 33.333% as contributors to 2n21 cytotype.doi:10.1371/journal.pone.0029449.t003 tion during a meiotic course has not been established in any species to date [1], [8], [9], [14], [22].

Auto-triploidization, auto-tetraploidization and autoaneuploidization in Malus
Triploids are considered very important to speciation and evolution as ''a bridge for polyploidization'' [1], [9].Triploids were 3.8-fold more frequent than tetraploid seedlings across the diploid Malus crosses (Table 1).Triploidy prevails in certain Malus species, e.g.'M.hupehensis', 'M.sikkimensis', and 'M.toringoides', but very few tetraploids have been identified in the Malus species [28].The number of aneuploid seedlings exceeded the number of eupolyploids (Table 1), suggesting that auto-aneuploidization was a major route for genomic duplication in the diploid Malus.

Gametic polyploidization patterns in diploid Malus
The Malus polyploidization patterns presented differently between ova and spermatozoa (Figure 1), and these features are interesting from an evolutionary perspective.Ova, through eupolyploidization, may account for maintenance of genetic stability, preventing biological disturbance and imbalance.And spermatozoa, through a greater range of cytotypes, may convey increased genetic diversity, providing a broad base for natural selection, improving adaptability [1], [2], [6], [10], [11], [22].Environmental exposure during micro-and megasporogenesis may account for their respective features in polyploidization.Deep in the ovary, the developing megaspore may be less affected by abiotic stressors such as radiation and fluctuant temperature than the microspore.Therefore, at once, the meiotic features of the male and female gametes of Malus impart genetic stability and variability, assisting evolution of this genus [2], [9]- [11], [22].

An alternative model for derivation of Malus
In many cases, neopolyploids have displayed a diploid-like meiotic behaviour after their establishment, and underwent diploidization to finally form 'diploids' but with genomic duplication [22], [38].It was recently deduced that Malus (x = 17) was derived from auto-polyploidization of two sister taxa (x = 9, 2n = 18), followed by diploidization and then aneuploidization to x = 17 [7].Given the relatively reproductive instability of aneuploids, we propose an alternative but more parsimonious three-step path to derivation of Malus: autoaneuploidization of two sister taxa (x = 9, 2n = 18) to monosomes (x = 9, 2n = 17), followed by whole genomic duplication in both ova and spermatozoa, involving FDR non-reduction (Figure 1), to tetraploids (x = 9, 4n = 34), and through diploidization to the extant diploid state (x = 17, 2n = 34).The frequency of monosomes was lower than other aneuploids in the diploid Malus (Table 1).How were monosomes (x = 9, 2n = 17) more successful than other aneuploids during Malus derivation?The reasons may be that monosomes had a higher evolutionary pressure to produce a higher frequency of tetraploids (x = 9, 4n = 34) than other aneuploids, and these tetraploids presented a stronger adaptability during speciation.

A path to evolution of odd basic chromosome numbers in the genome
Both auto-polyploidization and allo-polyploidization have been important for plant speciation and domestication, for example apple (Malus) [7] and wheat (Triticale) [39].Speciation resulting in an odd basic chromosome number (x) was thought to have derived from allo-polyploidization of two species with even and odd basic chromosomes [40].But recent studies have elevated the importance of auto-polyploidization for natural speciation [21]- [22].Auto-triploidization of species with odd basic chromosomes such as Malus species and auto-aneuploidization of species with odd or even basic chromosomes may be an important contributor to speciation with odd basic chromosome numbers during evolution.

Impacts of sexual aneuploidization on plant evolution
Our finding in Malus conflicts with prevailing views: (i) that eupolyploidy is primary for genomic duplication in diploid crosses, or (ii) that aneuploidy exceeded eupolyploidy in crosses from the parents with different ploidy, e.g.triploid crossed with diploid parents [1], [9], [14], [20], [22].Impacts of aneuploidization on speciation and evolution have long been ignored. .Aneuploidization can result in speciation with both odd and even basic chromosome numbers, while eupolyploidization can only contribute to even basic chromosome numbers (Table 2 and Table 3).Aneuploids derived from the same taxa can be of much genetic discrepancy.For example, species in Maloideae and certain species in Spiraeoideae have the same basic chromosome numbers (x = 17) [7]; they may derive from the same taxa through aneuploidization but with different chromosomal patterns.Regulation of chromosomal loss or gain contributing to aneuploids may be controlled by different genes, thus their effects on survival and adaptation of aneuploids may be very distinct.And this may result in the different evolutionary fates among the aneuploids under the natural selection from diverse environmental cues [1], [14], [22].Thus aneuploidization can result in multi-directional evolution.Genetic diversity in a polyploid individual may be higher than its diploid ancestor, but within a polyploid population it will consequently decrease.For example, on a 'abxcd' locus [28], four diploid genotypes, 'ac', 'ad', 'bc', and 'bd' can be found in the diploid Malus population, but two genotypes of 'abc' and 'abd', and one genotype of 'abcd' can be detected in the triploids and tetraploids, respectively (Table S8, S9, S10, S11).Thus, eupolyploidization can cause a decrease of genetic diversity on a whole.In contrast, aneuploidization will increase genetic diversity because aneuploids contain more genotypes than diploids (Table S1, S2, S3, S4, S5, S6, S7) [41].Eupolyploidization can cause gene silencing [42], [43].Aneuploidization may also result in gene silencing but it may not be so prevalent as in eupolyploidization.In addition, aneuploids may initiate special gene expression, e.g.resistance to a disease or tolerance to coldness or drought because of 'pseudo-dominant expression' by chromosomal loss or 'superdominant expression' by chromosomal gain in these aneuploids.Thus, aneuploidization provides a broader base and more diverse conditions for natural selection.Speciation through both eupolyploidization and aneuploidization, particularly aneuploidization with chromosomal loss, may have a higher probability of success than by eupolyploidization alone because of a greater evolutionary pressure in aneuploids during speciation and evolution.However, aneuploids are often less reproductively stable than eupolyploids [1], [14], [22].Thus an integrated route of 'aneuploidizationeupolyploidization-diploidization' for speciation and evolution of Malus as defined by our model, is more parsimonious with current data than other routes, e.g.only from eupolyploidization or from aneuploidization.Therefore, advantages and disadvantages in both aneuploidization and eupolyploidization should be properly evaluated, and thus their impacts on speciation and evolution can be appropriately determined.

A protocol using co-dominant markers for triploid apple breeding
Triploid apples, generally characterized by their large fruits, have been very attractive to both growers and consumers: old cultivars such as 'Gravenstein', 'Baldwin', 'Rhode Island Greening', 'Blenheim Orange', 'Stayman Winesap', 'Tompkins County King', as well as newer cultivars 'Jonagold' and 'Mutsu' [28].The fertility of triploids has, however, been found to be lower than that of the diploids and tetraploids, but rarely zero and often with differences between cultivars [1], [9], [28].This character appeals to both apple breeders and growers because of less labour needed for blossom and fruit thinning in triploids [28].Despite years of effort, attempts of chromosome doubling using colchicine [36] and endosperm culture [28] have not contributed to release of triploid apple cultivars.More than 90% commercially superior triploid apple cultivars have proved to be produced by sexual polyploidization [28].Thus, diploid crosses remain the principal method to breed apple triploid cultivars.
There are approximately 8,000 diploid apple cultivars in the world, but fewer than 50 are triploids [28].Given importance of triploid cultivars in the apple industry, we present a protocol using co-dominant markers aimed to efficiently identify a small number of triploid seedlings in a large population and thus to accelerate the triploid breeding program (Figure S1).In addition, parental genetic backgrounds influenced the frequency of triploids in the diploid F 1 seedlings (Table 1), thus parents should be carefully chosen for successful triploid breeding.

Applications of the strategies in this research to the studies for other species
To date, methodology has not been established to determine natural polyploids in a diploid population whether derived by sexual 2n gametes or by somatic chromosomal doubling [1], [22].This question can be addressed by using the strategy outlined in the current work.Polyploids identified by root-tip chromosome count would present extremely close microsatellite features as those in the diploid seedlings, suggesting non-reduced gametes contributing to these polyploids were not genetically heterozygous, further indicating these gametes should have derived from SDR [1], [8], [9].The sexual 2n gametes by FDR are heterozygous and co-dominant markers will present segregation.On the contrary, co-dominant makers will have non-segregation in the asexual 2n gametes.Though co-dominant markers in the polyploids derived from somatic polyploidization present similar patterns to those derived from sexual gametes by SDR, crossovers during meiosis resulting in a few heterozygous loci in the sexual 2n gametes allows discrimination between these modes.At least one in several hundred co-dominant markers in these sexual gametes will present a different pattern from their parents because of crossover occurrence between homologous chromosomes in these diploid gametes.However, asexual gametes do present the same patterns as their diploid mothers.Thus, our strategy is much more precise than the current widely-used flow cytometry and cytological methods for sporogenesis study.
Spermatozoa in triploids and aneuploids are mostly sterile, but their ova are usually fertile because they can produce euhaploid or a few eudiploid ova [1], [6], [14], [22], [26].Thus, triploids and aneuploids may be useful in the seed crop breeding program and for commercial seed production as alternative male-sterile materials.Polyploids and aneuploids with extra chromosomes can be used for analysis of genic or chromosomal dosage effects in the study of evolution and genetics; monosomes can be applied to study of function of certain genes or used for rapid genetic mapping in both plants and animals when corresponding dominant genes are affected in these monosomes.Polyploidy and aneuploidy are important for horticultural crops; they usually express valuable traits in the tree architectures or biological styles, for example dwarfing tree-systems or multiple petals, or as malesterility [44], [45].Vegetative propagation will facilitate extension of these polyploids and aneuploids in the horticultural industry.Polyploidy was found in bacterium, yeast, fungi, and virus [46]-[][] [49].Polyploidization is also common in certain animals, or organs of the higher form animals [22], [50]- [51]; polyploidy occurs in humans and causes miscarriages and sterility [2], [9].This study will encourage people to develop the molecular markers to detect polyploids and aneuploids, thus to precisely diagnose miscarriages and sterility in the early stage in humans or other animals caused by polyploidy or aneuploidy.

Conclusions
In conclusion, diploid Malus presented unique patterns for eupolyploidization and aneuploidization contributing to genomic duplication.For example, aneuploidy exceeded eupolyploidy (Table 1), and ova only contributed euploidy while spermatozoa contributed both euploidy and aneuploidy (Figure 1) for apple polyploidization.These unique characters of polyploidization in the diploid Malus confer genetic diversity and multi-directions for evolution and speciation.This is the first demonstration of a strategy using co-dominant markers to successfully analyze meiotic mechanisms and cytotypic derivation of unreduced gametes contributing to polyploids and aneuploids.This study is the first to provide molecular evidence for the contribution of heterozygous non-reduced gametes to genetic heterozygosity and fitness in polyploids and aneuploids, thus aids us understanding of superiority in polyploids and aneuploids.Impacts of aneuploidization on speciation and evolution were previously ignored.Aneuploidy can increase, while eupolyploidy may decrease genetic diversity in their newly established aneuploid or eupolyploid populations.Aneuploidization can result in a greater range of cytotypes for speciation, a higher genetic diversity and a broader base for natural selection.Eupolyploids and aneuploids should be further applied to the studies of genetics, breeding, and evolutionary science.Marker-assisted breeding using co-dominant markers will greatly accelerate breeding of triploid apples.We proposed a path to evolution of numerically odd basic chromosomes and considerably revised the model for derivation of Malus.The findings in this research will aid better understanding of evolution, speciation, and domestication of Malus.This study provides strategies for a further exploration of genetics of polyploidization and an enhanced polyploid breeding program for other living species, and polyploid genetics for humans.

Root-tip Chromosome Count
Root-tip chromosome count for ploidy identification was assessed according to Pratt and co-authors', with minor modifications [52].Root tips of no more than 3.0 cm, and collected during 10:30 am and 12:30 pm from early June to early August yielded maximum numbers of metaphase cells.Seedlings more than one year old were analyzed by root-tip chromosome count.

Framework Genetic Linkage Analysis
Framework linkage groups were constructed using one reciprocal cross with 62 arbitrarily-chosen diploid F 1 individuals and their parents [41], and compared to public references (http://users.unimi.it/hidras) to confirm transferability and consistency of the markers between crosses.
The marker order, genetic distance between markers, the overall coverage of genetic distance in the linkage differed among the seven crosses.Reasons resulting in these differences would be F 1 diploid seedlings and markers in this study may be insufficient for calculation of a 'precise' genetic distance between markers, and chromosomal crossovers may differently occur among the seven crosses thus influencing calculation of genetic distance and marker order [53].Fortunately, we had a few published Malus maps as the references [29]- [30], [34] and data from the seven crosses were compared with each others.At last, the markers which consistently clustered in the same linkage group were chosen for genetic analysis.
When a pair of primers produced no less than two markers and were mapped in the different loci, an extension numbers (e. g. '_1', '_2', '_3', or '_4') attached the original name were given for each marker.Sequences of primers can be obtained at the public website (HIDRAS: http://users.unimi.it/hidras).

Strategies of Cytotypically Genetics Using Micro-satellites
DNA was isolated by a modified CTAB (cetyltrimethylammonium bromide) protocol.DNA of eupolyploids, aneuploids, two arbitrarily-chosen diploid F 1 seedlings (as the control) identified by cytology, and their parents were genetically analyzed using microsatellite markers.Microsatellites exhibit different distributional features in polyploid and aneuploid seedlings compared to diploids.Consider the markers with the segregation pattern of four different alleles, 'abxcd', where 'ab' is the maternal genotype, 'cd' is the paternal genotype.If the unreduced gametes were heterozygous, we would expect the following allelic distributions in a locus [41]: Diploid seedlings (eudiploids) will present one of four bi-allelic genotypes; 'ac', 'ad', 'bc', or 'bd'.Eutriploid seedlings will present one of two tri-allelic genotypes; 'abc' or 'abd', if the unreduced gametes were maternal.Tetraploid seedlings will present one tetra-allelic genotype, 'abcd', where each parent contributed an unreduced gamete.Aneuploid seedlings will present allelic distributions similar to eudiploids, except that those markers in the chromosomes ''affected'' by the aneuploidy will present distributional features with extra chromosomes; 'acd' or 'bcd', or distributional features with 'lacking' chromosomes; 'a' or 'b', if the extra or missing chromosomes came from the paternal gametes.The genetic features can be confirmed by other markers presenting the segregation patterns of 'efxeg', 'nnxnp', or 'lmxll' alleles [41].Therefore, cytotype genetics in polyploid and aneuploid seedlings can be characterized by distributional features of microsatellites in these seedlings.

Statistical Analysis
Generalized linear models were fitted using the R 2.11.1 (http://www.r-project.org).Analysis of Deviance tables were constructed to investigate comparisons and interactions testing the deviance terms against a chi-squared distribution.Coverage intervals were calculated using an equivalent Bayesian model (OpenBUGS software, http://www.openbugs.info) of the generalized linear model fitted, since the standard errors are not symmetric for low percentages [54].

Figure 1 .
Figure 1.Schematic Summary of the Features of Gametic Combinations for Apple Polyploidization in Diploid Malus.Ova have two cytotypes, n and 2n ova; spermatozoa have a range but classified-into three cytotypes, n, 2n, and aneuploid spermatozoa for apple polyploidization.'n21' and 'n+1' refers to two aneuploid spermatozoa for aneuploidization.Diploid Malus exhibited a unique gametic combinational pattern, ova preserving euploidy exclusively, while spermatozoa presenting both euploidy and aneuploidy, for polyploidization.Molecular features showed that non-reduced gametes were genetically heterozygous, indicating first-division restitution was the exclusive mode for apple polyploidization.Figure depicts only three basic chromosomes with different colours in the legend to elucidate the basic chromosome number in the apples is odd.doi:10.1371/journal.pone.0029449.g001

Table 2 .
Genetic summary of cytotype distribution and the contribution of individual linkage group (LG) to aneuploid cytotypes.
a Based on analysis of all aneuploid seedlings in the study, summed over the crosses.b In column ''2n21'', data represents absence of linkage group, for all other aneu-cytotypes, data represents one duplicate linkage group.For example, of the 16 individual ''2n+10'' seedlings, 11 had a duplicate copy of LG02.doi:10.1371/journal.pone.0029449.t002

Table 3 .
Average percentage (%) of individual chromosome contributing to aneuploids in the seven diploid Malus crosses.