Special Behavior of Homo(eo)logous Chromosomes at Meiosis and the Mechanism of Partial Female Fertility of Allotriploid Lilium ‘triumphator’ (LLO)

Background: The abnormal meiosis of triploid is paid little attention; however, triploid not only is an important trend of ornamental breeding but also is regarded as a bridge of plant polyploidy speciation. In the present research, an allotriploid lily ‘Triumphator’ (LLO, 2n = 3x = 36 ) was investigated. Results: The results showed that the homoeologous chromosomes of two L-genomes formed bivalents and then disjoined and their half-bivalents moved to opposite poles at anaphase I; The chromosomes of O-genome formed univalents, among which, some attached to spindle �bers on two sides, their sister chromtids moved to the opposite poles slower than half-bivalents; some attached to spindle �bers on one side, they move to one pole at similar speed to half-bivalents; some of which deattached to spindle �bers at kinechores on either side, they became micronuclei at anaphase I. Conclusions: Interploid hybridizaitons showed that the LLO lily, regarless of male sterile, could be female parents to crossed with appropriate males to produce aneuploids, supporting the hypothesis -- �ve same genomes is essential for its endosperm development in interploid hybridizations of Lilium. The discussion well not only explained why triploid lilies with Fritillaria embryo sacs, regardless of male sterile, are usually partial female fertile while other plants with polygonum embryo sacs are seedless; but also explained that why 3x × 2x/4x lilies and tulips produce predominantly aneuploids while 3x × 2x/4x and 2x/4x × 3x of other plants, including 2x/4x × 3x of lilies and tulips, produce euploids and near so.


Background
Triploid, containing three sets of chromosomes, is usually sterile and seedless.It is often regarded as bottle neck for introgression breeding due to its sterility, especially for eld crops which are propagated usually by sowing seeds; however, triploid has been used to breed seedless watermelon [1], banana [2], grape [3], Citrus [4], and many ornamental crops, like tulip and lily [5,6], because they can be propagated through cutting, grafting or tissue culture.In addition, tripoid is regarded as a bridge to polyploidy speciation naturally [7,8].Since 2n gametes occur popularly in plants with very low rate [9], the possibility of unilateral sexual polyploidization is much higher that bilateral sexual polyploidization, i.e., triploid is rst formed much easier than tetraploid [7].Actually, not all triploids are completely sterile.Once partially fertile triploids have been formed, they could produce not only auepoloid, but also euploids like tetraploids, pentaploids, and hexaploids through 3x × 2x, 2x × 3x, 3x × 3x; and then, may produce 4x, 5x, etc., by 3x × 4x; and then other ploidy level popluations through more interploidy hybidization [7,10].Thus, triploids play important roles in plant polyploidy speciation.
Triploid lilies have special characteristics on their fertility.Different from that other triploid plants are usually highly sterile both male and female as well, for example, seedless watermelon, grape, banana, mandarin, and others, triploids lilies are partial female fertile regardless of completely male sterile, i.e., they can be used female parents to cross with appropriate diploid/tetraploid to predominantly produce aneuploid progeny [24][25][26].The phenomena are well explained by analysis the characteristics of different megasporogenesis and their signi cance on the success/failure of 3x × 2x/4x crosses between Fritillariatype and Polygonum-type plants (See detail in discussion).
Meiosis is the key step in the sexual life cycle, not only for species to continue from generation to generation and but also for species to evolve others, because it is the basis of heritage, mutation, assortment, and recombination [27].However, there is very little attention paid on abnormal meiosis of triploids.The reasons for that, probably, are 1) it is hard to obtain triploid; 2) triploid can't perpetuate sexually because they are usually sterile and seedless, and thus not so valulable for eld crops.Since triploid is important for plant polyploidy speication and triploid lilies are very popular and good materials for cytogenetic study, we investigated the microsporogenesis of an allotriploid lily, analyzed its male and female fertility, and the genomic composition of its progeny, and then, discussed its signi cance on lily breeding and plant speciation.

The microsporogenesis
As lily breeder claimed, GISH showed that 'Triumphator' has two sets of L-genomes and one set of Ogenome, i.e., an allotriploid (LLO, 2n = 3x = 36) (Fig. 1a).The allotriploid lily displayed the characteristics.1) At metaphase I (Fig. 1 b and c), the homologous chromosomes of two L-genomes usually associated to form bivalents (white "II"), while their homoeologous chromosomes of one O-genome remained univavent (red "I"); In a few cases, bivalents were formed by one O-genome (Fig. 1b: red "II" ) or by L-and O-genomes (Fig. 1b: red"I" + white "I"), multivalents formed by either L-genome (Fig. 1b and c: III and IV) or L-and O-genomes (Fig. 1b: white "II" + red "I") were also observed; sometimes, multivalents might not be true, just because two bivalents, or one univalent and one bivalent, adjoined by chance.2) At anaphase I (Fig. 1d, e and f), the bivalents were disjoined and moved to the two opposite poles (Fig. 1d, e and f: ) while some univalents stayed on the equatorial plates ( ), possibly because spindle bers exerted equal opposite forces on the kinetochores on the bivalents and the univalents, but bivalents need less force to separate than univalents; However, some L-chromosomes ( ) moved seemingly to the poles faster than L-chromsomes ( ), possibly because the O-chromosomes never attached to spindle bers and they just stayed there occasionally; and there are a few O-chromosomes ( ) which moves the same as half-bivalent ( ), possibly because one side of them were attached with spindle ber.Besides, in most PMCs, some O-chromosomes were apparently missing.It was not certain that they were really degraded or a few chromosomes might be associated together.The mechanism needs further and elaborate work.
3) At telophase I (Fig. 1 g, h and i), chromosomal bridges (Fig. 1 g: ) and micronuclei (Fig. 1 h and i: ) were commonly seen at this stage.The former might be caused by multiple crossovers, the latter might be due to detachment to spindle microtubule polymer due to the dysfunctional kinechores.At this stage, the PMCs with such abnormal phenomenon were take account 43.3%.4) At anaphase II and telophase II (Fig. 1 j, k and l), chromosomal bridges could be seen in 12.2% of PMCs; Most PMCs produced tetrad, while 3.9% produce triad.Based on observation of microsporogenesis, it was concluded that the LLO lily had abnormal meiosis.
Here we summarize the abnormal behaviour of chromosomes of LLO lily with an ideogram (Fig. 2).
Homoeologous chromosomes of two L-genomes formed bivalents and their two sides attached to spindle bers (microtubule polymers) at anaphase I, then disjoined and half-bivalents moved to opposite poles at anaphase I (Fig. 2: ); One O-genome chromosomes form univalents, some of which attached to spindle bers at kinechores on two sides, and sister chromotids move to the opposite poles slower than half-bivalents (Fig. 2: ); some of which attached to spindle bers at kinechores on one side, they move to one pole at similar speed to half-bivalents (Fig. 2: ); some of which deattached to spindle bers at kinechores on either side, they became micronuclei at anaphase I (Fig. 2: ) .

Male and female fertility of LLO
In vitro pollen germination test showed that no pollen grains of triploid LLO germinated and the other diploid lilies were 33.5-67.50%(Table 1).Though the LLO was completely male sterile; however, when it was hybridized with LL, TT and OO using normal pollination respectively.Their fruits of all combinations developed to different extents (Table 2, Fig. 3).Obviously, the fruits of LLO × LL /TT developed better than those of LLO × OO.Seemingly, the fruits of LLO × LL were similar to those of LLO × TT; however, the ovules in the fruits of LLO × LL developed much better than those of LLO × TT and LLO × OO, because some embryos or endosperm could be isolated from LLO × LL. 14 seedlings were obtained from LLO × LL, and only callus were from LLO × TT /OO (Fig. 2).It was concluded that LLO could be female parents to hybridize with appropriate diploid males.

Genomic composition of the progeny
Nine seedlings of LLO × LL were analyzed using GISH.The results showed that they were aneuploid, with chromosome number ranging from 27-32 (Fig. 4 and Table 3).All of them contained 24 L-chromosomes, indicating 12 of them were from LLO because LL contributed 12 L-chromosomes, and had Ochromosomes from three to eight.In addition, no recombinant chromosomes were found in the progeny.
Because the homologous chromosomes of two L-geomes usually form bivalents (II) and their homoeologous chromosomes of one O-genome remained univavent (I) in most PMCs of LLO, it was expected that their gametes should contain 12 L-chromosomes and variable O-chromosomes from such abnormal meiosis.The genomic compositions of the progeny were agreeable to the chromosomal behaviors during meiosis.

Discussion
There are very a few reports on meiosis of allotriploid lily.For example, 'Cocossa' (OOT) is partial male fertile and micronuclei are obviously observed at telophase of its abnormal meiosis [28].The same phenomenon occurs in the present research (Fig. 1h and i).Micronuclei are common in human cancerous cells and they possibly arise from hypomethylation of repeat sequences in pericentromeric DNA, irregularities in kinetochores, dysfunctional spindle apparatus, or awed anaphase checkpoint genes [29].
When allotriploid LLO meiosis is compared with distant diploid LA F1 hybrids [21], we notice both LLO and LA have form bivalents and univalent at metaphase I; however they show different chromosome behaviour at anaphase I: in LA hybrids, both bivalents and univalents are separated and pulled to the opposite poles [21]; while in LLO, bivalents are disjoined and move to the opposite pole, some univalents are scattered irregularly.Since the chromosomal movement at anaphase in normal meiosis is caused by microtubulin polymerdegrade [30,31], we speculate that the bivalents in LLO attach the spindle microtubule polymer the same as in normal meiosis; however, some univalents attach the spindle microtubules on either or both of their two sides; some univalents do not detach to any spindle microtubule (Fig. 2).
Although triploid lilies have abnormal meiosis and are usually male sterile, however, they are partially female fertile.They can be used as female parents to cross with appropriate diploid/tetraploid to predominantly produce aneuploid progeny (Table 4, Fig. 5a).Lilium have Fritillaria embryo sac [32].In Fritillaria embryo sacs, the ploidy levels of the secondary nucleus do not vary with the egg cell [19,24,26] and constantly twice as somatic cell.For an example: the allotriploid lily 'Triumphytor' (LLO) in the present research produce aneuploid egg cells due to abnormal meiosis but its secondary nuclei are constantly hexploid (6x = 4L + 2O).So, In lily 3x × 2x/4x, the embryos are usually aneuploid while the endosperm of 3x × 2x is 7x and that of 3x × 4x is 8x (Fig. 6: b1, b2).Since euploid endosperm could develop due to balanced chromosomes, some aneuploid embryos in the lily 3x × 2x/4x crosses can survive [26].Based on Table 4, it is observed that the endosperm with ≥ 5 same genomes developed well and the crosses are highly successful, while the endosperm with < 5 same genomes usually aborted and the crosses were usually unsuccessful.Considering that endosperm of normal 2x × 2x crosses contain 5 same genomes in Lilium, the hypothesis --Five Same Genomes of Endosperm is Essential for its Development in interploid hybridizations of Lilium, has been proposed to explain the partial female fertility of triploid lilies [24].Because LLO × LL, whose endosperm genome composition (EGC) is 5L + 2O, is successful while LLO × OO/TT, whose EGC is 4L + 3O/4L+2O+T, are hard successful, the present results support the hypothesis.So far, however, there are three exceptions for the hypothesis, LLO × TTTT, LLO × OTOT [33] and LLO × AA [34].Possibly, genomic imprinting, i.e., an excess dosage of paternal genomes promotes endosperm development [35,36], is the reason why LAA (AAA) × AAAA is more successful than LAA (AAA) × AA in Lilium [24].Similarly, a paternal T-genome in the endosperm of LLO × TTTT more than of LLO × TT is bene cial to its success, even a paternal O-genome in the endosperm of LLO × OTOT more than of LLO × TT is also good for the endosperm development.LLO × AA is very special exception; however, it does not mean that its endomsperm develop as well as the endosperm of LLO × LL, because a very few seedlings are obtained from a large amount of cultured ovules rather than cultured embryo sacs or embryos.The best way to solve the problem would be to use LLO as female to hybridize with LL, OO, TT, AA, LTLT, OTOT, OAOA as males to compare their success or failure in one season and on same site.Anyhow, it is concluded that triploid lilies can be used female parents to cross with appropriate diploid or tetraploid males to produce aneuploids regardless of male sterility.The hypothesis not only explain well the suceess or failure of 3x × 2x/4x crosses, but also can guide breeders to combine to different genomes into one cultivar step by step.For example, we may combine L-, A-, Tand O with LALA, LOLO and TATA through the following two steps: Tulip and Fritillaria also important Fritillaria-type plants.Triploid tulips also can be female parents to cross with diploid or tetraploid males [37][38][39].Firitillaria is an important medicinal herb.It is expected 3x × 2x/4x can produce aueuploid Fritillaria, the variations caused by aneuploid would be good chance to select new varieties containing higher amount of effective chemicals.
Unlike lilies, Tulips, and Fritillaria with Frititillaria-type embryo sacs, most plants, like watermelon and banana, produce Polygonum embryo sacs.Their triploids are usually male and and female sterile.Once they have partially male and female fertility, the 3x × 2x/4x or 2x/4x × 3x crosses produced more euploids or near euploids than other aneuploids as illustrated in Fig. 5.This phenomenon could be explained with their monosporic embryo sacs.Based on normal megasporogenesis, it is deduced that the ploidy levels of the secondary nucleus in Polygonum embryo sacs are invariably twice as that of the egg cell.So, whether in the 3x × 2x/4x or 2x/4x × 3x crosses, both the embryos and endosperm are usually aneuploid (Fig. 6. b3, b4, c3 and c4 ); since aneuploids are little viable due to unbalanced genes, these triploids are usually seedless; only when some triploids have partially female or male fertility, 3x × 2x/4x or 2x/4x × 3x crosses can produce some euploids and or near so due to inter-embryo competition [10].
Triploid is not only the source to produce aneuploid varieties but also is the bridge in plant polyploidy speciation.In angiosperms, 40-70% of them have been identi ed as polyploids [40].Though polyploidy mechanisms include bi-sexual polyploidisation, uni-sexual polyploidisation and chromosome doubling, it is increasingly accepted that the 2n gamete is the main reason for polyploid speciation [9,41].The occurrence of bi-sexual polyploidisation and chromosome doubling is much rarer than uni-sexual polyploidisation, indicating that a triploid resulting from uni-sexual polyploidisation may act as a bridge role in polyploid speciation [7,41,42].As analyzed in Fig. 6, once triploids are formed in Polygonum-type plants, 3x × 2x or 2x × 3x possibly produce diploid, triploid, and even tetraploid or pentaploid though the possibility is rare.3x × 3x also possibly produce different ploidy level populations, but it is more rarer because there is very little chance for very few functional pollen grains to meet very few functional eggs in this combination.Since aneupolids are poor viable and they could not perpetuate sexually, then, during the process of polypoloidy speciation with triploid as a bridge, their basic chromosome numbers usually remain unchanged in different ploidy level taxa.Different from Polygonum-type plants, once triploids occur in Fritillaria-type plants, 3x × 2x can produce lots of aneploids with quite variable chromosome numbers.If they can be propagated asexually, many new species with different basic chromosome numbers might be formed after adaptation and evolution.Limonium (2n = 12, 14, 16, 24, and 36) of Plumbaginaceae, also an tetrasporic-type plant, is good example [43] for this point.However, this is the rst time to propose the relationship between the embryo sac type and the basic chromosome number during angiosperm speciation.It needs further elaborate and extensive work unveil the complex mechanism.A--Asiatic; L--Longi orum; O --Oriental; T --Trumpet; H --Lilium henryi; M --L.martagon 1 indicating the male produce 2n pollen; 2 indicating three successful 3x × 2x/4x hybridizations without 5 sames genomes in their endosperm.

Conclusions
The present study and above discussion prove that all triploid lilies are partially female fertile regardless of their complete male sterility because they are tetrasporic-type plants, and then can be used to cross with appropriate diploid or tetraploid males to breed aneuploid cultivars; The hypothesis, Five Same Genomes of Endosperm is Essential for its Development in 3x × 2x/4x crosses in Lilium, can guide breeders to select parents to successful combine different genomes in lily or even other tetrasporic-type plant breeding; Possibly, the embryo sac plays some roles in the change of basic chromosome numbers during plant speciation.In addition, the abnormal chromosome behavior at anaphase I would give a new way to unveil the molecular mechanism of the microcuble disassembly and chromosomal movements.

Microsporogenesis
Microsporogenesis was according to Zhou et al. [21].When ower buds were about 27-38 mm, their anthers were taken out and put in a vial containing a Carnoy's solution (ethanol: acetic acid = 3:1, V/V) and stored for at least 30 min.Then, some pollen mother cells (PMCs) were put on slides and mixed with a drop of 2% Carbol Fuchsin (Beijing Solarbio Science & Technology Co. Ltd. ), and the slides were immediately covered with square covers and checked under a microscope (ZEISS Scope.A1).

Table 1
The genome composition (GC), chromosome number (Ch No.), and pollen germination (PG) of

Table 2
The crosses and their pollinated owers, rescued fruits, developed ovules, isolated embryos, and survival seedlings

Table 3
Genome compositions of 9 seedlings of LLO × LL based on GISH analysis, indicating their chromosome numbers (Ch No.) including L-chromosome numbers (L-No.)and O-chromosome numbers (O-No.), and the chromosome numbers contributed by pollen (No. by Pollen) and those by egg (No. by

Table 4 .
The summary of reported 3x × 2x/4x of Lilium, indicating the interploid crosses are usually successful when their endosperm contain ≥ 5 same genomes