The Cytological Mechanism of Apospory in Paspalum notatum Analyzed by Differential Interference-Contrast Microscopy

Abstract

Bahia grass (Paspalum notatum Flugge) is an important tropical forage grass and sets seed by apospory. I) To clarify the mechanisms of aposporous embryo sac initial cell (AIC) appearance and apomictic embryo sac formation, and II) to make it clear the mechanism of multiple embryo seed set a development in polyembryonic ovules, several apomictic and sexual varieties of bahia grass were studied cytologically and quantitatively by Nomarski differential interference-contrast microscopy. The results were I) there was no difference between sexual and apomicts to megasporogenesis; and then, the megaspore degenerated in apomicts; at the same time, AIC originated from nucellar tissue appeared and its numbers increased as the ovary grew before anthesis; II) at anthesis, the sac derived from AIC located in the micropylar end (first sac) were 92.5 to 100%, and those in the chalazal ends (other sacs) were 40.4 to 86.0% among the apomicts; the first sac divided dominantly and were 56 to 87% comparable to 0 to 1% of the other sacs at 4 days after anthesis; however, 4 to 17% of the other sacs also showed embryo formations but endosperm. In final, the first sac occupied the whole space of the ovule, in which the embryos in the other sacs coexisted.

1. Introduction

Apomixis provides a method for cloning plants through seeds, so that it is of value for agriculture used to fix hybrid vigor and other hopeful breeding materials that are positioned in the middle breeding process [1, 2]. Apomixis is usually classified into three major mechanisms, apospory, diplospory and adventitious embryogeny [1]. Among the three mechanisms, apospory is considered as the most important one in agriculture because it does not undergo meiosis to propagate through the seed. Bahia grass (Paspalum notatum Flugge) is an apoamphimictic perennial that sets seed by apospory, a form of gametophytic apomixis [3, 4, 5, 6]. Studies on chromosome, cytology, cross-compatibility, and colchicine treatment have been done in bahia grass [7, 8, 9]. However, studies in the field of molecular level have not been developed yet and still need more research.

Recently, some challenges using differential interference-contrast microscopy (DIC) technology have been conducted in all organs of plants. For example, in the meiotic chromosome [10], in petal development and ethylene biosynthesis [11], in nucleolus morphological changes [12], in pharmacology and cell biology [13], in xylem differentiation [14], in single microtubules [15]. In particular, near field DIC provides the ability to illuminate two neighboring points on the sample simultaneously, which shows that by modulating the two wavelengths employed in exciting such a probe, phase difference information can be retrieved through measuring the near field photoinduced force at the difference of the two modulation frequencies [16]. And more, two phases (cell structure and fluorescence) that appeared concurrently in the same sample and could be observed in ASG-1 transgenic rice [17], and Arabidopsis (Chen et al. in contribution) while using the general DIC system. From the above description, it is understood that DIC shows a bright future for the clarification of not only the structures but also the mechanisms, and not only in plants but also in animals, as well as the microbes.

We choose bahia grass as a monocotyledons species as that would be particularly amenable to a molecular study of apomixis. It is shorter a plant and easily to cultivate among the important forage grasses. Recently, the somatic embryogenesis and plant regeneration system of bahia grass has been established in developing gene introduction techniques [18]. To clarify the molecular process controlling apomixis in P. notatum, it is important to determine the developmental timing and location of apomictic events in suitable laboratory strains that are currently available. It is a fact that the timing of apomictic gene expression, the key to cloning the apomixis genes has not been identified in bahia grass. Recently, it is reported in guinea grass (Panicum maximum) analyzed ultrastructurally and cytologically by DIC and Transmission electron microscopy (TEM) that, aposporous embryo sac initial cell (AIC) appearance is related with the increasing of ovary length [19, 20, 21]. And based on the ovary length as an index, AIC-specific clones have been obtained named as Apomixis-specific gene-1 (ASG-1) [22, 23, 24]. In the other species, molecular approaches to apomixis research have also been reported, i.e., Citrus aurantium [25], Arabidopsis thaliana [26], Brachiaria brizantha [27], Tripsacum [28], Pennisetum [29]. That bahia grass has or has not the same mechanisms should be understood for the molecular studies. In P. notatum, Quarin [8] reported the method to observe the effect of pollen source and pollen ploidy on endosperm formation and seed set in pseudogamous apomict. However, an efficient embryo sac analysis method cannot be found in bahia grass. Therefore, it is essential for the analysis of genetic and breeding in apomixis that the mechanism of embryo sac formation in apomict gets clear using an efficient analysis method [30].

In this study, the major objectives were, using the microscopy method of DIC I) to make it clear the cytological and quantitative observations of AIC appearance and its development in bahia grass, and to estimate the period of AIC appearance using ovary length as an index; and II) to clarify the process of polyembryonic seed set in facultatively apomictic ovules, and to provide information for estimation of the degree of apomixis or sexual of P. notatum. And the multiple embryo formation and the balance of maternal and parental to endosperm formation were also discussed.

2. The mechanism of AIC appearance and its development

2.1 The process of AIC appearance

Four apomictic bahia grass varieties and two obligate sexual varieties were chosen for this study. These materials were kindly provided from Osumi breeding branch, Kagoshima Prefecture Agricultural Experiment station, Japan. For the 2 obligate sexual materials, Nangoku and C 1, the former is a diploid (2n = 20) variety [31], and the latter is a diploid (2n = 20) trace. About four apomictic varieties, they are tetraploid [32].

One hundred to over 300 buds or flowers before and at anthesis were collected for each variety for embryo sac analysis. The buds and flowers were fixed in FPA50 (formalin propionic acid: 50% ethanol = 5: 5: 90) for 5–7 days at 4°C [20, 21, 33, 34]. Ovaries were picked out carefully from fixed buds or flowers under microscope by using needle and tweezers, placed in 70% ethanol followed dehydration series (70, 80, 90, 100% ethanol), and cleared in Herr's benzyl-benzoate-four-and-a-half fluid [35] for over 2 h at 0–4°C. The observations were conducted using DIC.

The frequencies of apospory and sexual were calculated at anthesis according to the schematic of Figure 1 [36]. Here, apospory was classified into two types, Panicum type (PN) and Paspalum type (PS). PN type represents 4-nucleate embryo sac with one polar nucleus, one egg cell and two synergids [20, 37, 38, 39]. PS type represents 4-(or 5-) nucleate embryo sac with two polar nuclei, one egg cell and one (or two) synergid(s). Sexual means Polygonum type (S) 8-nucleate embryo sac with one egg cell, two polar nuclei, two synergids and three antipodals [40]. Therefore, the frequency of apospory in this study was estimated according to the total percentage of PN and PS types.

Figure 1.

Schematic representation of embryo sac development during sexual gametogenesis and the apospory forms in P. maximum and P. notatum.

Until megasporogenesis there showed similar behaviors in both sexual and apospory varieties. After megasporogenesis, however, different events from sexual varieties were observed in apospory varieties. While the formed megaspore became almost unfunctional and degenerated with the membrane disappeared, AIC-derived from enlarged unreduced nucellar cells appeared from a different direction and usually entered the space around the degenerated or surviving megaspore (Figure 2(1), [36]). In Figure 2(2) and (3), there showed coexistence of AIC and degenerated or surviving megaspore.

Figure 2.

Aposporous embryo sac initial cell (AIC) appearance and AIC-derived embryo sac maturity in apomictic bahia grass (P. notatum). (1) AIC appearance (big arrow) and degenerating megaspore remained (small arrow). Big arrow indicates AIC derived from nucellar cells, with a pointed (sharp) cell wall inserting into the space where only one megaspore without nuclear membrane remaining. 2 and 3 are 2 focal planes of same specimen. (2) Megaspore degenerated (small arrow). (3) Functional AIC formation (big arrow) with a circular cell wall in chalazal end in one ovule. (4) Four-nucleate aposporous embryo sac formation after AIC underwent two times of division. The nucleates were gathered up from each other in the micropylar end. (5) Differentiation of four-nucleate nuclei, an egg cell firstly formed (e, one synergid cell (s) and two polar nuclei (p). (6) Aposporous embryo sac maturity with one egg cell (e, two nuclei (p) and one synergid cell (usually invisible) in micropylar end. (7) Polyembryonic ovule containing three aposporous embryo sacs marked with single, double and third arrows, respectively. e: egg cell; p: polar nucleus; s: synergid cell; mi: micropylar end. Bar = 30 μm in Fig. 2(1)–(3), Bar = 45 μm in Fig. 2(4) and (5), Bar = 25 μm in Fig. 2(6) and (7).

2.2 The process of AIC-derived aposporous embryo sac formation

In general, the AIC forms the embryo sac through a special process. The AIC undergoes mitosis two times and forms 2-, 4-nucleate. And no antipodal was found (Figure 2(4)). The cell division only occurred in the half-space of the embryo sac in the micropylar end. It is different from the sexual one which usually occupied the whole space of the sac. And then, the 4 nucleates developed to complete their parts in order (Figure 2(5)). In common, a mature embryo sac formed with one egg cell, one synergid and two polar nuclei (Figure 2(6)). In some rare cases, Panicum type 4-nucleate embryo sac with one polar nucleus was observed. In most ovules multiple apomictic embryo sacs were observed from one to six (Figure 2(7)). The distinctive features of mature embryo sacs were observed. 1) There is one egg cell, the nucleolus, ca 8 μm in diameter, being visible and surrounded with bright starch grains. With which sometimes the nucleus and its membrane could be distinguished, and the cytoplasm was denser than the other cells. 2) In most of the embryo sacs, there is one synergid cell with the nucleolus, ca 4 μm in diameter, usually observed. And the cytoplasm was very few and the cell was occupied with bigger vacuoles. Moreover, near the synergid filiform apparatus usually were observable. 3) Two polar nuclei were almost observed with the nucleolus, ca 12 μm in diameter, surrounded by a nuclear membrane, and the cytoplasm was few; 4) No antipodal. The appearance of filiform apparatus is also evidence of embryo sac maturity in both of sexual and apomictic plants. The nucleoli are stained deeply with the clearing fluid, so that we can easily distinguish the stage of embryo sac formation by counting the numbers of nucleoli. Apomictic embryo sac in the micropylar end, matured about one day before anthesis, but in sexual varieties embryo sac did at the day of anthesis.

In most ovules of apomictic plants, AICs appeared continuously as the ovary length increased. The numbers of AICs were increased between the formations of functional megaspore and mature 4- or 8-nucleates according to the ovary length. For example, the ovary length in "competitor " was ca 520 μm when the first AIC appeared, and ca 496 μm when the third one did. However, 5 AICs appeared between 520 μm to 624 μm (688 μm) in ovary length. So were the other varieties. These values of ovary length indicated that AICs in the same ovule did not seem to differentiate synchronously.

From the length of ovaries, the stages of the ovary containing degenerated embryo sac were from 4-nucleate to their embryo sac maturity in apomictic plants, and from 4- and 8-nucleate to embryo sac maturity in sexual plants, respectively. On the other hand, the stage of functional megaspore showed a range of ovary length so wide it became very close to the value of degenerated ovaries.

2.3 The types of embryo sacs appeared in apomictic plants

Ten types of embryo sacs were observed and most of the embryo sacs belonged to PS type (Table 1, [36]). In apomicts, most ovules showed the number of embryo sacs more than one. And the ovules with different embryo sacs (5PS to 2PS, PS +PNn to 4PS +PNn, S, PNn) were observed. The percentage of S type (Polygonum type) was 13.6% (8/59), 9.1% (3/33), 12.9% (4/31) and 5.4% (2/37) in competitor, Nanou, Tifton and Common, respectively. PNn type (Panicum type) was only observed in one ovule among the tested varieties.

Varieties	No. ovules observed	No. sterile ovules	Types of embryo sacs1
			5PS	4PS	3PS	2PS	S	PS+PNn	2PS+PNn	3PS+PNn	4PS+PNn	PNn
Competitor	60	1	4	7	14	14	8	4	4	1	2	1
Nanou	35	2	1	2	7	11	3	3	2	1	3
Tifron	44	14	2	1	4	13	4	2	2	1	1
Common	43	6	3	5	10	13	2			3	1

Table 1.

Number of ovules with different types of embryo sacs at anthesis in apomictic varieties of P. notatum.

¹

PS: Paspalum type 4-(or 5-)nucleate with 2 polar nuclei, PN: Panicum type 4-nucleate with one polar nucleus, PNn: number of PN embryo sac (one or more PN), S: Polygonum 8-nucleate embryo sac.

Among the 4 varieties, the frequencies of PN type were from 5.0% to 30.3%, and that of PS type were from 60.7% to 90.0%. Total frequencies of apospory were 86.4%, 91.1%, 87.1% and 95.0% in competitor, Nanou, Tifton and Common, respectively.

2.4 The general discussions concerning the appearance of AIC and AIC-derived embryo sac formation

There are no differences observed between obligate sexual and apomictic plants until megasporogenesis in bahia grass. After megasporogenesis, however, different events are followed in embryo sac formation. Sexual ovules proceeded in a manner typical of the Gramineae family, i.e., functional megaspores divided and formed a mature 8- nucleate embryo sac as reported in P. notatum [6] and in P. maximum [20, 40]. In contrast, megaspore does not divide in apomictic ovules and becomes degenerated (Figure 2(2) and (3)). Consequently, AIC (2n) derived from nucellus tissue, different from megaspore (n) appeared, divided, and directly formed mature 4-nucleate embryo sac. Most of the embryo sacs contain one egg cell, one synergid cell and two polar nuclei. It is different from that reported in P. maximum, i.e., one egg cell, two synergids and one polar nucleus [20, 38]. Here, it was called as Paspalum type of 4-nucleate embryo sac. Koltunow [2] indicated that in Hieracium MMC (mother megaspore cell) and AIC appeared together. It is different from Paspalum reported here and Panicum [20]. That means apospory has different reproductive process in different species, genus, or families. In this study, however, when megaspore developed normally, AIC appearance was not observed. By the way, after AIC appeared megaspore coexisted with AIC for a movement, and finally, degenerated. In any case, once AIC appeared, megaspore did not develop. This also differs from Panicum type reported by Chen and Kozono [20]. AICs appearance stage is distinctly different from sexual ones, and AIC occurs only in apomictic varieties. So, the stage could be considered as a stage related to apomixis gene expression. Here, we can set up a hypothesis that, AIC gene exited and usually waited for a chance to express, only when the megaspore gave out a signal not fulfilling its mission to form embryo sac. AIC appearance and the embryo sac formation also have an important evolution meaning to protect from any unforeseen happenings.

The earliest AIC that appeared in ovule always located in micropylar end, as the ovary grows, the later appeared are located along with the first AIC and being apart from it. To understand the mechanism of AIC appearance, we selected ovary length as an index and measured the ovaries when they were observed in different AIC appearance. From the range of ovary length, AICs do not appear together in same time. Instead, they seemed following a continuous course and appeared one by one during the period from megasporogenesis even to the first embryo sac maturity. According to the ovary length compared with the morphology of spikes, AICs appeared in the period of spike emerging to open at anthesis. With regard to the ovary length measured we could collect every stage of embryo sac to apply apomixis gene cloning. Sterile ovules with degenerated embryo sac appeared in both sexual and apomictic varieties based on the observation and quantitative analysis of ovary length. The ovary length of the ovary in which the first AIC appeared was longer than that of ovary the functional megaspore appeared in all varieties, indicating that the aposporous phenomenon of AIC appearance is initiated after megasporogenesis. Further, the ovary length of the ovary staged in functional megaspore was wide and close to the ovary lengths of the ovaries showing degeneration of megaspores in different developed stages. These results indicated indirectly that the development of sexual embryo sac derived from megaspore is often terminated accompanied by AIC appearance in many aposporous apomicts around the stage of megasporogenesis [41, 42]. Which one of megaspore or AIC firstly showed the signal to terminate or to appear will be interesting to further researches of apomixis. In the present study, 10 types of embryo sac formation were observed in Paspalum notatum (Table 1). Here we must issue that "S" types (sexual embryo sacs) observed were almost 4- or 5- nucleate embryo sacs. Except S type, the ovules with over 2S types contain only the same, 8-nucleate ones were not observed. For the case of ovules containing one, two or more 4-, 5- (8)- nucleate embryo sacs in one ovule, two pathways could be considered as follows. 1) The sexual embryo sac formation results from the direct division of one, two or over two megaspore(s) though the AIC(s) appeared (or not) in the same ovules. 2) They are derived from AIC(s). In particular, as the ovules with two megaspores in chalazal end were not observed in this study while AIC(s) appeared in the micropylar end, the former pathway could be hardly considered as a putative one. So, the later pathway seems reasonable based on that AICs develop into not only 4-nucleate [39] but also, at a low frequency, 5- nucleate embryo sacs in Panicum [38], or 8-nucleate ones in Hieracium [42]. In Panicum, the 4- nucleate embryo sac formation with an egg cell, one synergid and two nuclei, was reported by Bashaw and Hanna [37]. And 5- nucleate one with an egg cell, two synergies and two nuclei reported by Nakajima and Mochizuki [38]. For the mechanism of 5-nucleate embryo sac formation in Panicum, Chen and Kozono [20] set up a hypothesis. That is, after megaspore or AIC, whether which is located in micropylar end or not, divided firstly into two nuclei, only the micropylar nucleus continued to divide twice secondly and to form 4-nuclei, and in final, 5 nuclei formed totally in an embryo sac. For the distribution of the 5 nuclei, the chalazal nucleus derived from the first division of megaspore or AIC, and one of four micropylar nuclei derived from the second division of megaspore or AIC, pair with each other to form two polar nuclei, and for the remaining three nuclei, one becomes one egg cell and two being synergies. From the above reports, we could conclude that facultatively apomictic bahia grass prefers to produce Paspalum type 4-(or 5-) nucleate (one egg cell, one (or two) synergid(s), and two polar nuclei) rather than to produce Panicum type 4-(or 5-) nucleate (one egg cell, two synergids, and one (or two) polar nucleus). Especially, there may be no Polygonum type (8-nucleate) embryo sac in polyembryonic ovules. Why did the two different apospory reproductive processes in the same 4-nucleate types of Paspalum and Panicum occur? This question means what should be clarified in the next future experiments.

3. The mechanism of polyembryonic seed set in facultatively apomictic ovules

3.1 The dominantly developmental process of the embryo sac formed in micropylar end by parthenogenesis

At 0 DAA, most of the embryo sac located in micropylar end became typical ones. In the accessions tested, the percentage of the typical embryo sacs in micropylar end was 90.9–100% higher than that in the other end (40.4–86.0%) (Table 2, [43]). If the typical one is S type, it consists of 8-nucleates of egg cell, 2 synergids and 3 antipodal. If the typical one is PN or PS type, it consists of 4-nucleates of egg cell, synergid and 2 polar nuclei.

Varieties	No. ovules observed (A)	No. embryo sacs (B)	Mean no. (B/A)	No. embryo sacs
				Micropylar	Others
Competidor	59	171	2.9	59 (100%)	112 (78.6%)
Nanou	33	85	2.6	30 (90.9%)	52 (40.4%)
Tifton	31	51	1.6	31 (96.8%)	20 (80.0%)
Common	40	90	2.3	40 (92.5%)	50 (86.0%)

Table 2.

Distribution of matured embryo sacs at anthesis in facultatively apomictic varieties of P. notatum.

At 6 h after anthesis, polar nuclei firstly started to divide to mean 9.0 cells in 55 of 59 ovules of Competitor. At 1 DAA, first division of egg cell was observed while the endosperm had reached mean 38.8 cells. This phenomenon is same to P. maximum that after the fertilization between the sperm and polar to form endosperm, with stimulation of the fertilization, the egg cell automatically divides into an embryo by parthenogenesis [20, 21]. Here, the endosperm always appeared as free-nuclear endosperm during 0–2 DAA. From 1 to 4 DAA, the mean numbers of egg embryo and endosperm cells were increased ca. 4 times per day, respectively. The number of ovules containing developed egg embryo and endosperm were increased as the days after anthesis increased. However, the ovules unfertilized remained in a certain number showed in Table 2. At 2–4 DAA, formation of endosperm cell wall started from the position farthest and nearest to the embryo, respectively. At 4 DAA, egg cell has undergone 6 cycles of division and became an embryo containing mean >64 cells. The endosperm was well-developed and almost occupied the whole space of the sac where it is in. After 4 DAA, the endosperm and young embryo developed so fast that the numbers of both cells could not be counted under the microscope. On the other hand, during 1–4 DAA, nucellar cells were changed to be vacuolated, as endosperm cells hanged from free-nuclear to cell-wall-formed with well-developed cytoplasm. No morphological differences were observed between the developed sacs either derived from sexual sac or apomictic sac.

In polyembryonic ovules, the embryo sac in micropylar end developed dominantly when compared with that in the other ends. Figure 3 showed the developments of both embryo and endosperm of the sac in micropylar end, and the other sacs were crowded out to chalazal end, with dividing egg embryo and polar nuclei [36].

Figure 3.

Multiple embryo formation in the same embryo sac in P. notatum flowers at 7–10 d after anthesis (DAA). A1. Two embryos with well-developed endosperm appeared in the chalazal end. A2. A globular-stage embryo around by endosperm in micropylar end at 7 DAA. (A1 and A2 are the same specimen). B) One embryo around by endosperm in micropylar end and one embryo in the side of ovule tissue were observed at 8 DAA. C) Two embryos side by side are located in the micropylar end. e = egg cell, p = polar cells, s = synergid, em = embryo, en = endosperm, mi = micropylar end. Bar = 45 μm.

The ovules containing dominantly developed embryo sacs were investigated at 4 DAA in 4 varieties. The results showed in Table 3 indicated that, 56–87% of the ovules contained developed sacs in micropylar end in all the accessions, and in contrast, on other ends, 0% in 3 varieties and 3% in one variety.

Varieties	No. ovules observed A	No. embryo sacs B	ES in miciopylar end			ES in the other end
			embryo and endosperm C(C/A)	embryo only	endosperm only	embryo and endosperm D(D/A)	embryo only E(E/B)	endosperm only
Competitor	30	41	26 (87%)	0	4	0 (0%)	7 (17%)	0
Nanou	27	28	15 (56%)	1	0	0 (0%)	1 (4%)	0
Tifton	30	35	21 (66%)	0	2	0 (0%)	3 (9%)	0
Common	30	38	23 (77%)	0	0	1 (3%)	5 (13%)	0

Table 3.

Development of embryo sac (ES) in the flowers 4 d after anthesis in facultatively apomictic varieties of P. notatum.

3.2 The developmental process of other types of embryo sacs formed in chalazal end

Some other cases appeared different from the above. 1) A single embryo sac in micropylar end contains a well-developed embryo and unfertilized 2 nuclei. 2) A single embryo sac contains 2 embryos and well-developed endosperm. 3) The embryo sac degenerated in micropylar end, and in chalazal end, the sac with developed embryo and unfertilized 2 nuclei. And the data of other types could be known from Table 3. Among the 4 varieties, the number of the embryo sac in micropylar end containing only embryo was 1 of 27 ovules in Nanou, and the numbers containing only endosperm were 4 of 30 ovules in Competitor, 2 of 30 ovules in Tifton, respectively. In contrast, in the other end, the numbers of the sacs containing only embryo were 4–17% in all the varieties, and the number containing only endosperm was 0% in the varieties.

3.3 The mechanism of seed forming embryo development in polyembryonic embryo sacs

At 6 DAA, the developed embryo and endosperm of the sac in micropylar end occupied the embryo sac it located in, and at the same time, the space of the whole ovules was almost occupied by the developed sac (Figure 3a). The embryo usually exists with near globular shape, and it is surrounded by well-developed endosperm. For the embryo sacs in the other ends, they usually were squeezed out to outside of the developed sac. However, those sacs showed continuous development. Some egg cells divided well, and formed embryos usually located in opposite side to micropylar end (Figure 3a), or the neighbor (Figure 3b and c). At 10 DAA, the formed 2 embryos showed the same morphology (Figure 3c), and it is difficult to distinguish their origins between them. From Table 3 [36], we can find that about 4–17% of ovules observed contain two or more embryos in the same ovule.

In emasculated ovules observed at 4 DAA, no developed embryo sac distinguished in 60 ovules of 4 varieties. The parthenogenesis rate was 0% in all the 4 varieties. At 15 DAA, the inflorescences that were emasculated and then isolated from any pollen source failed to produce seed. So, pseudogamous is essential for seed set in P. notatum.

3.4 The general discussion concerning the dominant development of the sac in micropylar end and seed formation by parthenogenesis

In facultative apomictic bahia grass, AICs appeared one by one, and then, they became multiple embryo sacs in same ovule, as the ovary length increased [36]. And as the first AIC usually appeared and located in the micropylar end, 92.5 to 100% of embryo sacs closest to micropylar end of ovule matured at anthesis observed in this study (Table 2). On the other hand, the embryo sacs located in other end showed 40.4 to 86.0% mature rates at anthesis. The AIC appearance age (order) maybe influence the mature of apomictic embryo sacs themselves. It could be considered that the first appeared AIC in micropylar end has the temporal dominant in formation and maturity of the embryo sac when compared with the other sacs. And for the fertilization chance, the sac has also the positional dominant, as it was closest to synergid cell through which pollen tube penetrates and finishes fertilization. Therefore, the sac derived from first AIC located in micropylar end has the advantage of fertilization. On the observations of ovules at 4 DAA, the rates of developed embryo sacs with embryo and endosperm were from 56 to 87% in the sacs of micropylar end (Table 3). On the other hand, the other sacs were 0% in 3 varieties, and one variety was 3%. Therefore, the sac in micropylar end has the advantage of seed set. This result also supported the hypothesis that the embryo of developed sac in micropylar end, in final, became a seed-forming embryo [20]. Using this method, we can estimate the degree of apomixis or sexual of any facultative apomictic materials used, based on the analysis of embryo sacs in micropylar end at anthesis.

Different events were observed on the seed set between guinea grass and bahia grass. In guinea grass, the other embryo sacs were crowed out to the chalazal end by the developed micropylar sac, and in final, they were completely degenerated after 10 DAA [21]. In contrast, the rates of embryos formed in the other embryo sacs were 4–17% in 4 accessions of bahia grass used in this study. These are higher than that (0%) in 5 accessions and that (2%) in one accession of guinea grass [21]. This evident was also observed from embryo sac analysis (Figure 3). As the sac derived from AIC contains 2n level reproductive cells, egg cell does not need fertilization. However, for the endosperm formation, fertilization between central cell and sperm cell is needed. And egg cell usually starts division followed the endosperm cell formation. The other sacs also follow the same manner. As the embryo sac developed advantageously in micropylar end, it could be considered that the egg cells in the other sacs were developed vigorously in different places of ovules. In that case, only embryo formed but no endosperm. In the polyembryonic ovules, the embryos located in the other sacs usually presented close to the well-developed endosperm of the micropylar sac. The 2 kinds of embryos in the same ovule seemed sharing the endosperm of the micropylar sac. Maybe that is why the embryos in different embryo sacs could coexist in the same ovule. At the germination experiment, twines, or multiple seedlings (>5%) were observed (data not shown). This means the different embryos have the same germination capacity. For the endosperm balance number, some reporters have discussed the requirement for balance between maternal and paternal contributions to the endosperm formation [44, 45, 46, 47, 48]. When they used different ploidies and sexual materials, endosperm balance in terms of maternal to paternal ratio, 2:1 was considered balanced and should produce normal endosperm. And they indicated that unreduced embryo sacs with one central-cell nucleus (4 factors), when fertilized by a sperm (2 factors), would result in the proper endosperm ratio of 2m:1p, and the closed percentages were obtained between embryo sacs with a single central-cell nucleus and ovaries with endosperm developing 6 and 8 d after pollination. The result of aposporous guinea grass reported by Chen and Kozono [21] also supported the above explanation. In grasses, the most common type of unreduced embryo sac is the 4-nucleates Panicum type (PN) developing into one egg, two synergids, and one polar nucleus or, more rarely, one egg, one synergid, and two polar nuclei [2, 42]. In Paspalum notatum, Chen et al. [36] reported that the percentages of unreduced embryo sacs with 2 central-cell nuclei (PS) were 60.7–90.0%, and that with one central cell nucleus (PN) were 5.0–30.3% in 4 varieties observed, respectively. However, the percentages of ovaries with endosperm developing 4 DAA were 56–87% in the same 4 varieties in this study. If we follow the report of Morgan et al. [47], only 5.0–30.3% of developed endosperm should be obtained. Recently, Quarin [8] reported the related information about the endosperm balance number that, apomictic 4x P. notatum is a pseudogamous species with effective fertilization of the 2 unreduced (2n = 4x) polar nuclei by a reduced (n = 1x) sperm. In that case, endosperm development and seed set occurred independently in the species. In that case, endosperm development and seed set occurred independently of the species or the ploidy level of the pollen donor. In his explanation, as sexual Paspalum plants fit the endosperm balance number (EBN), the EBN insensitivity is observed in apomictic apomixis. The EBN insensitivity could have arisen as an imprinting consequence of a high maternal contribution.

Recently, the mechanisms of reproductive characterization [49], molecular and genetic regulation [50, 51, 52], and its utilization [53, 54] have been discovered consequently in P. notatum. Together with the clarification of the mechanisms of aposporous embryo sac initial cell appearance and the cell-derived aposporous embryo sac formation descripted, here, in P. notatum and previous and similar report in P. maximum [19], this study will provide the essential and important information for successful cloning of apospory genes in P. notatum. And therefore, the P. notatum as one of the main players will be chosen in apomixis research and give another extension for the breeding program and productive utilization in agriculture and forage grasses.

4. Conclusion

In this study, using the plant materials of several apomictic and obligately sexual varieties of bahia grass (Paspalum notatum) and the method of differential interference-contrast microscopy (DIC), we have clarified the mechanisms of aposporous embryo sac initial cell (AIC) appearance, the numbers of AICs increased as the ovary length grew before anthesis and AIC-derived apomictic embryo sac formation at anthesis, and after the anthesis, the process of polyembryonic seed formation by parthenogenesis. With the DIC technique giving the 3–D image recognition, AIC appearance, a different event from obligate sexual one, was firstly observed and recognized in P. notatum, so it could be expected as a related candidate with apomixis gene expression, as the ASG-1, an apomixis-specific gene-1 as reported in P. maximum based on the ovary length as an index to sample the AIC stage ovaries according to DIC observations [22, 24]. Therefore, the results of this study will be useful to provide the information on isolation of apomixis gene from apomictic varieties and the production of gene transgenic plants [17].

Acknowledgments

This study was partly supported by the Grant-in-Aid for Scientific Research (C) of the Ministry of Education, Culture, Science and Sports of Japan, No. 10660010.

Conflict of interest

Authors have declared that no competing interests exist.