Plant Regeneration and Its Functional Analysis within Transgenic Rice of ASG-1, an Apomixis- specific Gene Isolated from Apomictic Guinea Grass

Aims: To analyze the function of the ASG-1, an apomixis-specific gene has been isolated from apomictic guinea grass, the combination of Agrobacterium-mediated transformation system and pSMA35H2-NG binary vector were used for transformation of ASG-1 to ‘Nipponbare’ of rice, and its functional analysis of transgenic rice was carried out. Study Design: 1) plant regeneration of transgenic rice of ASG-1 was achieved, and 2) it was Original Research Article Chen et al.; BJI, 16(3): 1-13, 2016; Article no.BJI.29146 2 followed by experiments with the obtained non-graining ear in the whole plants (NGEP) using comparative observation in both pollens of guinea grass and the T0 NGEP with Nomarski differential interference-contrast microscopy (DIC). Place and Duration of Study: Faculty of Environmental and Horticultural Science, Minami Kyushu University, between September 2012 and October 2015. Methodology: As a preliminary transformation, a β-glucuronidase (GUS) was used to infect the callus. Plant regeneration of the transgenic rice of ASG-1 were achieved. DNAs of the transgenic rice were used for detection of ASG-1 by PCR. For analysis of NGEP, the anthers and ovaries of the transgenic rice and guinea grass were used for comparative observation with DIC. Results: 1) GUS was successfully expressed in calli of both rice and guinea grass. 2) The transgenic rice plants of ASG-1 were then obtained from the culture of seed-derived calli. 3) The natures of T0 callus and T1 plants were confirmed. 4) A subset (~15%) of T0 plants showed morphological sterility with NGEP. 5) The ovaries of the NGEP showed identical characteristics to “Nipponbare” and other normal T1 plants; however, the pollen showed normal rates of <38%. 6) The pollen of guinea grass at anthesis showed rates of normal pollen as low as that of the NGEP rice. Conclusion: ASG-1 may play a role in regulating pollen sterility. Comparing these findings with our previous reports, we identified co-localization for ASG-1 in pollen of both rice and guinea grass, suggesting that the trait of ASG-1 influencing pollen sterility might be inherited from guinea grass to rice through the transformation process.


INTRODUCTION
Apomixis is a unique reproductive mode in plants in which the egg cell does not fuse with the sperm cell and directly develops into the embryo, although the central cell does and forms an endosperm. Through this phenomenon, it is considered that the formation of asexual seeds in plants leads to populations that are genetically uniform maternal clones [1]. As apomixis provides a method for cloning plants through seed and involves the conservation of existing heterozygosis, apomixis is expected to simplify the development of hybrid cultivars and the production of commercial hybrid seed when used in breeding programs [2][3][4]. The transfer of apomixis to crop plants holds great promise within plant breeding for fixation of hybrid vigor because it would allow the propagation of hybrids over successive generations [5]. However, the use of conventional breeding techniques for the transfer of apomixis to important crops has been largely unsuccessful, as the plants employing apomixis are usually polyploids, and show different types of apomixis, making it difficult to understand the underlying genetic mechanism and to apply it in breeding programs [6]. Therefore, the genetic engineering of apomixis may be a more promising approach to reach this material goal [7]. To reach this goal, understanding the molecular mechanism behind apomixis is an essential initial step. To date, some approaches have been assessed for Panicum maximum [6,8], Hieracium [9], and Pennisetum squamulatum [10]. In P. maximum, we initially reported that the fundamental difference between sexual and apomictic accessions of guinea grass is the appearance of the aposporous embryo sac initial cell (AIC) when the megaspore dies or stops its division during the apomict accession, in contrast to sexual accession, which divides into an eightnucleate embryo sac [11]. We also find that the AICs do not simultaneously appear, but rather appear one by one as the ovary length grows continuously [11]. Based on ovary length as an index for sampling different developmental stages of embryo sac formation in sexual and apomict forms, an AIC stage-specific gene was isolated by using a differential screening method, and was termed apomixis-specific gene-1 (ASG-1) [6].
The amino acid sequence of ASG-1 is related to RD22, a seed-specific and drought-induced gene of Arabidopsis thaliana [12]), USP, an unknown seed protein precursor of Vicia faba [13], and other important genes of auxin-down-regulated cDNA clone [14], apospory-region marker [15] and polygalacturonase isoenzame I [16]. In addition, ASG-1 has been identified for its localization in the cells of AIC and AIC-derived embryo sac by in situ hybridization in apomictic Guinea grass [8]). However, the functional analysis of ASG-1 by gene expression in transformants has not been attempted, as successful transformation procedures and vectors have not been available.
To perform the transformation in guinea grass, some approaches have been conducted using plant regeneration from leaflet culture [17], ovary culture [18], and seed culture [19] However, a practical and efficient procedure for gene transformation in guinea grass has not yet been established. In rice (Oryza sativa) of the same family as guinea grass, there have been two reports using Agrobacterium-mediated transformation: one using japonica rice [20] and the other using indica rice [21]. Recently, an efficient Agrobacterium-mediated transformation system has been established by Toki et al. [22] using a new binary vector, pSMA35H2-NG. As the initial purpose of the apomixis project was to isolate the putative genes from apomictic guinea grass and to then introduce it to rice for breeding application [6,11,23] ASG-1 should be assessed for use for transformation of rice.
In the present study, we report that plant regeneration of transformants can be obtained using the combination of an Agrobacteriummediated transformation system for rice (Dr. Toki, NIAS, Japan) and pSMA35H2-NG binary vector (kindly provided by Dr. Ichikawa, NIAS, Japan) specially expressed in the whole plant of rice for transformation of ASG-1, and to clarify the function of non-graining ear in the whole plants (NGEP) occurred in T 0 and T 1 transformants, with comparative observation of pollen in transgenic rice. In addition, as a subsidiary analysis for NGEP, a discussion is presented based on the fact that in situ hybridization of ASG-1 in guinea grass provided signals ranging from strong to weak in the pollen developmental process while staged from before the appearance of AIC to the AIC-derived embryo sac formation and matured embryo sac at anthesis.

The Preliminary Experiment for β-glucuronidase (GUS) Expression in Induced Calli
Seeds of rice (Oryza sativa) cultivars "Nipponbare" and "Akenohoshi" (Japanese varieties) were used for callus formation. Surfaces of seeds were firstly sterilized with 70% alcohol for 30 s and 2.5% sodium hypochlorite for 15 min with gentle stirring. After rinsing with sterile distilled water three times, the seeds were placed on filter paper for removing excess water. The seeds were then placed on callus induction medium (N6D) [24]. Seed germination and callus induction protocols of Toki et al. [22] were followed. Four weeks later, the calli were selected for GUS treatment.
As a control, the leaflet tissues of guinea grass (Panicum maximum) of sexual accessions N68/96-8-o-5, N68/96-8-o-7, and facultative apomictic accession N68/96-8-o-11, were collected from the farm of Minami Kyusyu University (Miyakonojo city, Japan) and cultured for callus formation according to the procedure previously reported [25]. The calli were selected for GUS treatment in the same manner as that for rice 6-8 weeks later.
Binary vectors pIG121Hm ( Fig. 1) and pSMA35H2-NG ( Fig. 2) were kindly provided by Dr. Nakamura (Nagoya U. Japan) and Dr. Ichikawa (NIAS, Japan), respectively. These two vectors containing the hygromycin B selectable marker gene (hpt), and a GUS, were introduced into Agrobacterium tumefaciens strains EH101 [26] and GV3101/pMp90 (kindly provided by Ichikawa), respectively. For GUS gene transformation, calli from rice seeds (Fig. 3A, C) and the leaflets of guinea grass ( Fig. 3E-G) were used in the present study, and the protocols of infection and removing of Agrobacterium, and GUS dyeing were as according to that of rice (Dr. Toki, NIAS, Japan, personal communication).

Construction of ASG-1 Gene to pSMA35H2-NG
As the binary vector, pSMA35H2-NG provided infection rates higher than those of pIG121Hm in calli of both guinea grass and rice, according to the preliminary result. The pSMA35H2-NG vector was used for ASG-1 transformation. The nucleotide sequence of ASG-1 was referred to the GenBank/EMBL/DDBJ nucleotide sequence database with the accession number AB000809. For the plasmid construction, the PCR products added with Xbal and Sacl recognizable sequences, respectively in 5′ and 3′ sites of sense cDNA region of ASG-1 was inserted into the sites of Xbal and Sacl of pGEM5 (Promega) cloning vector. Based on the analysis of insert sequences, the clones without variation were selected. The sequences of ASG-1 cDNA digested with Xbal and Sacl enzymes were then set into pSMA35H2-NG, which was digested with Xbal and Sacl. That is, the cassette of GUS was replaced with ASG-1 cDNA (Fig. 5). The cassette of ASG-1 was introduced into Agrobacterium tumefaciens GV3101/pMp90 for the transformation experiment. For GUS gene transformation, calli from rice seeds (Fig. 3A, C) and the leaflets of guinea grass ( Fig. 3E-G) were used in the present study, and the protocols of infection and removing of Agrobacterium, and GUS dyeing were as according to that of rice (Dr. Toki, NIAS, Japan, personal communication).

Construction of ASG-1 Gene to pSMA35H2-NG
As the binary vector, pSMA35H2-NG provided infection rates higher than those of pIG121Hm in calli of both guinea grass and rice, according to the preliminary result. The pSMA35H2-NG vector was used for ASG-1 transformation. The nucleotide sequence of ASG-1 was referred to the GenBank/EMBL/DDBJ nucleotide sequence database with the accession number AB000809. For the plasmid construction, the PCR products added with Xbal and Sacl recognizable sequences, respectively in 5′ and 3′ sites of sense cDNA region of ASG-1 was inserted into the sites of Xbal and Sacl of pGEM5 (Promega) cloning vector. Based on the analysis of insert sequences, the clones without variation were selected. The sequences of ASG-1 cDNA digested with Xbal and Sacl enzymes were then set into pSMA35H2-NG, which was digested with Xbal and Sacl. That is, the cassette of GUS was replaced with ASG-1 cDNA (Fig. 5). The cassette of ASG-1 was introduced into Agrobacterium tumefaciens GV3101/pMp90 for the transformation experiment. For GUS gene transformation, calli from rice seeds (Fig. 3A, C) and the leaflets of guinea grass ( Fig. 3E-G) were used in the present study, and the protocols of infection and removing of Agrobacterium, and GUS dyeing were as according to that of rice (Dr. Toki, NIAS, Japan, personal communication).

Construction of ASG-1 Gene to pSMA35H2-NG
As the binary vector, pSMA35H2-NG provided infection rates higher than those of pIG121Hm in calli of both guinea grass and rice, according to the preliminary result. The pSMA35H2-NG vector was used for ASG-1 transformation. The nucleotide sequence of ASG-1 was referred to the GenBank/EMBL/DDBJ nucleotide sequence database with the accession number AB000809. For the plasmid construction, the PCR products added with Xbal and Sacl recognizable sequences, respectively in 5′ and 3′ sites of sense cDNA region of ASG-1 was inserted into the sites of Xbal and Sacl of pGEM5 (Promega) cloning vector. Based on the analysis of insert sequences, the clones without variation were selected. The sequences of ASG-1 cDNA digested with Xbal and Sacl enzymes were then set into pSMA35H2-NG, which was digested with Xbal and Sacl. That is, the cassette of GUS was replaced with ASG-1 cDNA (Fig. 5). The cassette of ASG-1 was introduced into Agrobacterium tumefaciens GV3101/pMp90 for the transformation experiment.

Selection and Regeneration of Putative Transgenic Plants
Seeds of "Nipponbare" were used for callus formation according to the protocol conducted in the preliminary experiment ( Fig. 4A-G). After cocultivation for 3 days with GV3101/pMp90, the calli were washed with sterilized water containing 500 mg/L of carbenicillin disodium salt (CDS), and transferred to the selection medium of N6D containing 300 mg/L of CDS and 50 mg/L of hygromycin B twice for the 2 week culture in 30°C and 3,000 lux for 16 h. The selected transgenic calli were transferred to the Regeneration III medium [27] containing 200 mg/L of CDS and 50 mg/L of hygromycin B twice for 3 weeks of culture in the same conditions as those given above. After the shoots and roots were grown to over 1 cm, they were transferred to the same hormone-free medium. The plants grew sufficiently large and were then transported into the pots for grain formation under 28°C and 1 × 10 4 lux for 16 h (Fig. 4H-M). The grains obtained from T 0 plants were sown to obtain T 1 plants.

Molecular Analysis of Transgenic Plants
Routine DNA manipulations of leaves from the T 0 and T 1 putative transgenic rice plants and the apomictic guinea grass of N68/96-8-o-11 as a control were performed according to the DNA mini plate method [28]. For PCR, the primers were designed according to the ASG-1 sequence (Suppl. file) and the hygromycin B sequence (Suppl. file). The PCR conditions were: 95°C, 1 min; 94°C, 30 sec; 54°C, 30 sec; 72°C, 1 min, for 35 cycles; and finally, 72°C, 5 min, then holding at 4°C. The PCR products were electrophoresed in 1.5% agarose gel for determination of specific bands of ASG-1 and hygromycin B.

Cytological Analysis of Transgenic Rice Plants
Fifty buds and flowers before and at anthesis, respectively, were collected for embryo sac analysis from T 0 and T 1 putative plants with grain formation and no grain formation, respectively, and non-transgenic plants as the control. Embryo sacs were observed according to the method of Chen and Kozono [11,23]. The buds and flowers were fixed in FPA 50 (formalin: propionic acid: 50% ethanol = 5:5:90) for 5 to 7 days at 4°C.
Ovaries were then isolated from fixed buds or flowers, cleared in Herr's benzyl-benzoate-fourand-a-half fluid [29] for >2 h at 0-4°C, and observed with Nomarski differential interferencecontrast optics (DIC). The anthers of transgenic and non-transgenic plants of T 1 were also observed with DIC.

Cytological Analysis of Guinea Grass
Fifty buds and flowers of guinea grass were collected before and at anthesis, respectively, from the facultative apomictic accession of N68/96-8-o-11, from which ASG-1 was isolated. Embryo sacs and anthers were observed according to the same methods as those for rice described above.

Preliminary Experiment of GUS Reporter Gene to Guide Transformation
GUS was used as a reporter gene in the transformation efforts to closely monitor the gene transformation progress. Two kinds of vectors ( Fig. 1 and Fig. 2) provided the GUS expression in calli (Fig. 3A-B) and embryogenic calli ( Fig. 3C-D) from rice seed culture, respectively. However, for Guinea grass, only the pSMA35H2-NG/GV3101/PMP90 provided GUS expressions in the initial callus formation, normal callus, and embryogenic callus from leaflet culture (Fig. 3E, F, G), but not in pIG120Hm/EHA101. In addition, the expressed rates of the former vector were 37.5-75.0% and 70.7-73.9% in guinea grass and rice, respectively, which were higher than the 0% and 49.3-65.3% expression rate achieved by the latter vector, respectively ( Table 1). As the pSMA35H2-NG/GV3101/PMP90 transformation system showed universal application in rice and guinea grass, the system was used in the present study for transformation of the ASG-1 gene.

Plant Regeneration of Transgenic rice of ASG-1 from Culture of Mature Seed
Calli were induced to N6D medium from mature rice seeds after 3-4 weeks of culture ( Fig. 4A-D). The calli, of which approximately 20% were white and compact embryogenic calli with numerous embryoids on the surface (Fig. 4E-F), were then transferred onto modified N6D medium for an additional 3 weeks of culture. To further confirm the efficiency of the transformation system, these calli were partly used for GUS transformation, and as a result, they gradually showed blue spots on the surface (Fig. 4G). The remaining calli were then used for ASG-1 transformation (Fig. 5), transferred onto N6D regeneration medium, and the shoots emerged growing into complete plants (Fig. 4H-K). Finally, the plants were naturalized and cultured in a growth chamber (Fig. 4L-M).

Confirmation of the Transgenic Nature of ASG-1 and Hygromycin B in the Regenerated Rice Plants
Initially, when the primers of A1 and S1 (Suppl. file) and DNA were extracted from calli of T 0 putative transgenic rice and N68/96-8-o-11 were used for analysis of polymerase chain reaction (PCR), the specific bands of ASG-1 were detected from the samples, respectively, indicating that they included the transgenic natures (Suppl. file).
PCR analysis was then performed on T 1 putative transgenic plants of rice. The various patterns were obtained in 903 bp between S1 and A1, 581 bp between S1 and A2, and 294 bp between S2 and A2 (Fig. 6). The pattern in hygromycin B was 513 bp between HML and HMR (Fig. 7, Suppl. file). These results suggest that the T 1 plants were from independent transformation events and no escape was observed, indicating the tightness of the selection scheme. As a control, when HML and A3 primers (Suppl. file) were used for PCR analysis, different bands were detected (Fig. 7), suggesting that recombinant plasmids were set in the chromosomes of transgenic rice with different patterns.

Morphology and DIC Analysis of Putative Transgenic Rice Plants
Among the T 0 putative transgenic rice plants, a subset (~15%) showed NGEP, even though cutting and growing to form ears and flowers was repeated three to four times for the plants of NGEP. However, the remaining plants in T 0 and their T 1 generation provided normal graining and seed maturity. To determine the mechanism of NGEP, the buds and flowers of the T 0 and T 1 plants were collected and observed with DIC, as described in the Material and Methods. The female gametophyte, in particular, the embryo sac, showed similar behavior to the normal egg. Synergies, central, and antipodal cells existed in the remaining transgenic plants; however, in the male gametophyte in particular, the pollen showed normal rates <38% ( Fig. 4O; Table 2), which was less the rate of >90% of the T 1 plants and un-transgenic plants (Fig. 4N). The difference between the NGEP and the T 1 plants was in the normal rates of pollen.

Observation of Pollen and Embryo Sac of Apomictic Guinea Grass
To further determine why the sterile transgenic rice plants appeared after ASG-1 of recombinant plasmid was introduced, and whether it is related to the origin from which ASG-1 was isolated, the apomict of guinea grass accession N68/96-8-o-11 was used to observe the pollen sterility. The pollen in the anther and out of the anther in the young bud stage before anthesis showed a full spherical shape and filled contents with rich cytoplasm containing starch grains (Fig. 8A-B), whereas those of the anthesis stage of flowers showed a wrinkled shape and empty contents ( Fig. 8C-D). The normal rates of pollen in the  anthesis stage were counted as <40%, which was the same as that of sterile NGEP of transgenic rice (Fig. 4N). The changes of morphology of pollen in different developmental stages observed in the present study were compared with those in the previous report of in situ hybridization of ASG-1 in apomictic guinea grass, indicating indirect coincidence between both of the pollen types (Table 2).

DISCUSSION
In the present study, we successfully obtained transgenic rice of ASG-1 using an Agrobacterium-mediated transformation system and hygromycin B selection, based on the result that performed well in the GUS transformation protocol for all three accessions of guinea grass and two cultivars of rice tested. This is the first time that ASG-1 was transformed into rice and ASG-1 was realized as a functional gene to act in the transgenic rice plants.
The Agrobacterium-mediated transformation method has mainly been used for dicotyledonous plants. The breakthrough for the application to monocotyledonous plants was achieved by Hiei et al. [20] in japonica rice and Rashid et al. [21] in anthesis stage were counted as <40%, which was the same as that of sterile NGEP of transgenic rice (Fig. 4N). The changes of morphology of pollen in different developmental stages observed in the present study were compared with those in the previous report of in situ hybridization of ASG-1 in apomictic guinea grass, indicating indirect coincidence between both of the pollen types (Table 2).

DISCUSSION
In the present study, we successfully obtained transgenic rice of ASG-1 using an Agrobacterium-mediated transformation system and hygromycin B selection, based on the result that performed well in the GUS transformation protocol for all three accessions of guinea grass and two cultivars of rice tested. This is the first time that ASG-1 was transformed into rice and ASG-1 was realized as a functional gene to act in the transgenic rice plants.
The Agrobacterium-mediated transformation method has mainly been used for dicotyledonous plants. The breakthrough for the application to monocotyledonous plants was achieved by Hiei et al. [20] in japonica rice and Rashid et al. [21] in anthesis stage were counted as <40%, which was the same as that of sterile NGEP of transgenic rice (Fig. 4N). The changes of morphology of pollen in different developmental stages observed in the present study were compared with those in the previous report of in situ hybridization of ASG-1 in apomictic guinea grass, indicating indirect coincidence between both of the pollen types (Table 2).

DISCUSSION
In the present study, we successfully obtained transgenic rice of ASG-1 using an Agrobacterium-mediated transformation system and hygromycin B selection, based on the result that performed well in the GUS transformation protocol for all three accessions of guinea grass and two cultivars of rice tested. This is the first time that ASG-1 was transformed into rice and ASG-1 was realized as a functional gene to act in the transgenic rice plants.
The Agrobacterium-mediated transformation method has mainly been used for dicotyledonous plants. The breakthrough for the application to monocotyledonous plants was achieved by Hiei et al. [20] in japonica rice and Rashid et al. [21] in   Fig. 4N; 4 DIC: Fig. 4O indica rice, respectively. A rapid and efficient Agrobacterium-mediated transformation system in rice was then reported by Toki [30] and Toki et al. [22]. In the present study, we followed Toki's protocol to produce the transgenic plants, which performed well in both guinea grass and japonica rice.
As a preliminary examination of transformation, two binary vectors of pIG120Hm/EHA101 and pSMA35H2-NG/GV3101/PMP90 containing GUS gene were tested for expression rates on calli of both guinea grass and rice. The latter vector provided higher rates than those of the former vector on the two kinds of calli, and the former vector did not work on the three accessions of guinea grass tested (0%), indicating that compatibility existed not only between the binary vector and the donor plant, but also between the binary vectors. The two vectors were usable for the rice transformation system; however, pSMA35H2-NG/GV3101/PMP90 expressed a Near 100% of globular shape with rich cytoplasm Over 50% of irregular shape with poor or none of cytoplasm DIC 3 Transgenic rice (Nipponbare) Graining Near 100% of globular shape with rich cytoplasm Near 100% of globular shape with rich cytoplasm DIC 4 Transgenic rice (Nipponbare) Non-graining Near 100% of globular shape with rich cytoplasm Over 60% of irregular shape with poor or none of cytoplasm 1 In situ hybridization: Chen et al. 2005; 2 DIC (differential interference-contrast microscope): Fig. 8; 3 DIC: Fig. 4N; 4 DIC: Fig. 4O indica rice, respectively. A rapid and efficient Agrobacterium-mediated transformation system in rice was then reported by Toki [30] and Toki et al. [22]. In the present study, we followed Toki's protocol to produce the transgenic plants, which performed well in both guinea grass and japonica rice.
As a preliminary examination of transformation, two binary vectors of pIG120Hm/EHA101 and pSMA35H2-NG/GV3101/PMP90 containing GUS gene were tested for expression rates on calli of both guinea grass and rice. The latter vector provided higher rates than those of the former vector on the two kinds of calli, and the former vector did not work on the three accessions of guinea grass tested (0%), indicating that compatibility existed not only between the binary vector and the donor plant, but also between the binary vectors. The two vectors were usable for the rice transformation system; however, pSMA35H2-NG/GV3101/PMP90 expressed a   Fig. 4N; 4 DIC: Fig. 4O indica rice, respectively. A rapid and efficient Agrobacterium-mediated transformation system in rice was then reported by Toki [30] and Toki et al. [22]. In the present study, we followed Toki's protocol to produce the transgenic plants, which performed well in both guinea grass and japonica rice.
As a preliminary examination of transformation, two binary vectors of pIG120Hm/EHA101 and pSMA35H2-NG/GV3101/PMP90 containing GUS gene were tested for expression rates on calli of both guinea grass and rice. The latter vector provided higher rates than those of the former vector on the two kinds of calli, and the former vector did not work on the three accessions of guinea grass tested (0%), indicating that compatibility existed not only between the binary vector and the donor plant, but also between the binary vectors. The two vectors were usable for the rice transformation system; however, pSMA35H2-NG/GV3101/PMP90 expressed a higher efficiency, not only in rice, but also in guinea grass than did pIG120Hm/EHA101. This provides important information for further transformation experiments in guinea grass, as up to now, an efficient working protocol for guinea grass has not yet been established for the Agrobacterium-mediated system, even though some transformation systems have been working in the other grasses, for example, toll fescue [31].
Similar to the first report, in the present study, the T 0 and T 1 transgenic rice plants of ASG-1 were successfully achieved, and certified by PCR using the primers designed according to the sequences of ASG-1.
However, among T 0 plants, approximately 15% showed NGEP. To date, the low rates of grain formation in transgenic rice were reported in early studies of rice [20,22], mainly due to the antibiotics applied in the recovery process of the culture. In general, the NGEP is considered to be as a consequence of the sterility of female and/or male gametophytes. When we observed the ovary, no abnormal phenomena were evident in embryo sacs at anthesis (data not shown), indicating that the embryo sacs were ready to receive the sperm for fertilization. The female gametophytes are normally formed at anthesis (Fig. 5M) and are not at this point considered to be the factor affecting NGEP in the T 0 plants.
When we continued to observe the male gametophyte of the NGEP, only 38% of the pollen showed a normal nature, which can be described as a sphere shape and containing rich cytoplasm with starch grains. The fact that approximately 62% of pollen was abnormal with irregular and wrinkled shapes and empty cell content without cytoplasm indicates that due to abnormal development, over half of the pollen could not grow sufficiently in size to break out of the anther at anthesis; therefore, the pollen could not exit the anther and reach the stigma for fertilization. The phenomenon of appearance of abnormal pollen is considered to be the reason why the formation of grain was not achieved in the transgenic rice.
To isolate the relationship between the ASG-1 and NGEP in the transgenic rice, the donor plants of guinea grass accession N68/96-8-o-11, from which ASG-1 was isolated, were used for observation of male gametophytes. In general, apomictic or facultative apomictic guinea grass is defined as self-sterility [3]. However, a detailed analysis of the sterility has not yet been conducted. We observed the female and male gametophytes and determined that the normal pollen rates were <40% in the anther at anthesis (Fig. 8), indicating that the <40% of pollen normality may result in its self-sterility in the apomictic guinea grass accession. This result in guinea grass indicates that the same phenomenon appeared in ASG-1 transgenic rice (Fig. 5N) in the <40% pollen normality (Table 2). An additional report [5] also indirectly found that when ASG-1 was used as a probe to perform in situ hybridization in N68/96-8-o-11, its signals were changed from stronger in the two to four nucleate stage to weaker and/or no signal in the mature pollen stage and/or at anthesis ( Table 2). This result may indirectly indicate that ASG-1 acts functionally in the pollen formation process. The results summarized in Table 2 suggest that the self-sterility in guinea grass might be attributed to pollen abnormality. In addition, it could be considered that the pollen abnormality in the transgenic rice could be due to ASG-1 playing a role in pollen formation in guinea grass; therefore, during the transformation process, it also acts in the rice and results in abnormal pollen. As a result, the stigma of the ovary in selfpollinating transgenic rice could not receive the pollen from the same flower, meaning that the egg cell could not fuse with the sperm of the pollen, fertilization between the male and female could not be achieved, and the flower would finally become NGEP.
The ASG-1 gene isolated from facultative apomictic guinea grass (a tetraploid) was transformed into sexual rice (a diploid) showing abnormal pollen, which resulted in NGEP in the present study. In general, whether the gene from a tetraploid, for example ASG-1, works normally in a diploid is doubtable. Recently, a new report in apomixis research has emerged concerning the transformation between apomict and sexual plants [10]. When an apospory-specific genomic region (ASGR) isolated from apomict of Pennisetum squamulatum grass (4x) was transformed into its sexual form (4x), the gene was expressed in egg cells before fertilization and could induce parthenogenesis and the production of haploid offspring in the transgenic sexual form. This result suggested that the same ploidy level may be necessary for the correct expression of the gene; the gene can work normally in the same ploidy level after transformation. This also hints that the same ploidy level of the gene can work well with normal function. Therefore, the next step is using ASG-1 on tetraploid sexual plants to examine its function. Another surprising result obtained recently in our laboratory (data not shown) is that when ASG-1 was transformed into diploid sexual Arabidopsis, the apomixis-like phenomenon with multiple embryo sac formation (MESF) in the same ovule appeared. This result suggests that the ASG-1 gene plays a main role in MESF, and to our knowledge, no study of MESF has as yet been conducted in sexual Arabidopsis. According to the above report [10], it can be expected that the ASG-1 gene will express exactly and provide real morphology when transformed into the same level of tetraploid sexual Arabidopsis.
However, as Arabidopsis belongs to a different family to that of guinea grass, and is dicotyledonous as opposed to the monocotyledonous form of guinea grass, the ASG-1 gene should be transformed into the same species of P. maximum, namely monocotyledonous, sexual, and tetraploid, for the functional analysis of its gene. To realize this aim, we have successfully established a simple and practical plant regeneration system of guinea grass by using matured seed culture [32]. The project is currently in progress.

CONCLUSION
We obtained transgenic rice plants of "Nipponbare" with ASG-1, an apomixis-specific gene isolated from apomictic guinea grass, using a combination of the Agrobacterium-mediated transformation system and pSMA35H2-NG binary vector. A subset (~15%) of the T 0 ASG-1 transgenic rice plants with NGEP were determined using DIC, indicating that the plants had higher abnormalities of pollen, showing shapes different from normal and containing empty cytoplasm and rates of <38%, in contrast to normal, and the anther could not be opened normally to allow pollen out. Furthermore, to identify the relationship between the rice and guinea grass, pollen of facultative apomictic guinea grass of N68/96-8-o-11 was also observed, and the same high abnormality of pollen was confirmed, suggesting that ASG-1 may play a role in pollen development in both rice and guinea grass. As a reference, the previous report showing that in situ hybridization of ASG-1 in N68/96-8-o-11 provided signals gradually changing from strong to weak and no signals in the pollen developmental process also indirectly supports the above result. Comparing these findings with previously reported pollen observations in relation to ASG-1, we identified a coincidence for ASG-1 in pollen of both rice and guinea grass ( Table 2), suggesting that the function of ASG-1 was transferred from guinea grass to rice through the transformation process. This result indicates that ASG-1 may play a role in pollen development and may also find applications in the synthesis of artificial male gametogenesis lines to reduce the time and cost to produce F 1 in breeding programs.

NOTE
The other data can be found in attached supplementary files.