A high maternal genome excess causes severe seed abortion leading to ovary abscission in Nicotiana interploidy-interspecific crosses.

Seed abortion and ovary abscission, two types of postzygotic reproductive barriers, are often observed in interspecific and/or interploidy crosses in plants. However, the mechanisms underlying these reproductive barriers remain unclear. Here, we show that the distinct types of seed developmental abnormalities (type I and type II seed abortion) occur in a phased manner as maternal to paternal genome dosage increases and that type II seed abortion is followed by ovary abscission. We revealed that these two types of seed developmental abnormalities are observed during seed development in the interploidy-interspecific crosses of Nicotiana suaveolens and N. tabacum. Moreover, in the cross showing type II seed abortion, several events, such as changes in abscission-related gene expression and lignin deposition, occurred in the ovary abscission zone, eventually leading to ovary abscission. Notably, successive increases in maternal ploidy using ploidy manipulated lines resulted in successive type I and type II seed abortions, and the latter was accompanied by ovary abscission. Conversely, both types of seed abortion and ovary abscission could be overcome with a ploidy manipulation technique that balances parental ploidy levels. We thus concluded that a high maternal genome excess cross may cause severe seed developmental defects and ovary abscission. Based on our findings, we propose a model explaining the abortion phenomena, where an interaction between the promotive and inhibitive effects of the parental genomes determines the developmental destiny of seeds. SIGNIFICANCE STATEMENT: We demonstrate that a stepwise increase in maternal ploidy results in a stepwise increase in seed abortion severity, leading to ovary abscission in plants. We propose a model explaining the abortion phenomena, where an interaction between the promotive and inhibitive effects of the parental genomes determines the developmental destiny of seeds.

INTRODUCTION

In angiosperms, the process of seed development begins with double fertilization. One of the sperm cells fertilizes the egg cell to form the zygote, giving rise to the embryo, while another sperm cell fuses with the two polar nuclei to form the triploid endosperm. The endosperm is a nourishing tissue that ensures adequate nutrient transfer between the mother tissues and the embryo (Hehenberger et al., 2012). Meanwhile, the seed coat differentiates from maternally derived integuments after fertilization. Coordinated embryo, endosperm, and maternal seed coat development is necessary for normal seed formation. However, seed abortion, a typical postzygotic reproductive barrier preventing seed development or seed germination, is widely observed in interspecific and interploidy crosses (Baek et al., 2016; Coughlan et al., 2020; Oneal et al., 2016; Rebernig et al., 2015; Roth et al., 2018; Scott et al., 1998; Sekine et al., 2013; Tonosaki et al., 2018;). Studies on Arabidopsis, Solanum, Mimulus, and other species have strongly suggested that endosperm developmental failure is the primary cause of seed abortion, as this failure leads to embryo arrest and seed abortion (Hehenberger et al., 2012; Rebernig et al., 2015; Scott et al., 1998; Sekine et al., 2013; Tonosaki et al., 2018; Oneal et al., 2016; Coughlan et al., 2020; Roth et al., 2018).

The endosperm formed by sexual reproduction between diploid parents is typically triploid. According to the endosperm balance number (EBN) hypothesis, normal development of endosperm usually requires a relative maternal (m): paternal (p) EBN ratio of 2:1 in the endosperm (EBN is an arbitrary number allocated to each species) (Carputo et al., 1999; Ehlenfeldt and Hanneman, 1992; Ehlenfeldt and Ortiz, 1995; Johnston et al., 1980; Leblanc et al., 2002; Lin, 1984). However, in interploidy crosses, deviation from the 2:1 genome or EBN ratio results in endosperm developmental failure. Maternal genome or EBN excess generally results in precocious developmental transition in the endosperm, while paternal genome or EBN excess generally results in delayed or failed endosperm development (Coughlan et al., 2020; Scott et al., 1998; Sekine et al., 2013). Similar abnormal endosperm development has also been reported in interspecific crosses between species with the same ploidy level. In these cases, it has been suggested that the influence of maternal and paternal genomes on endosperm development may not be balanced or that parental species have different EBNs or “effective ploidy,” leading to endosperm developmental failure (Baek et al., 2016; Bushell et al., 2003; Coughlan et al., 2020; Ishikawa et al., 2011; Johnston et al., 1980;). The abnormal endosperm development in interspecific crosses between species with the same ploidy level can be bypassed by changing the ploidy of one parental species (Bushell et al., 2003; Johnston and Hanneman, 1982; Lafon‐Placette et al., 2017; Tonosaki et al., 2018). Moreover, genome‐wide studies on endosperm gene expression have revealed that genomic imprinting is extensively perturbed in failing hybrid endosperm in Solanum, Capsella and several other species (Lafon‐Placette et al., 2018; Roth et al., 2019; Tonosaki et al., 2018).

The genus Nicotiana comprises 76 species that are predominantly distributed thoughout the Americas and Australia. The wild tobacco, Nicotiana suaveolens, which belongs to Nicotiana section Suaveolentes, is a self‐compatible (SC) species found in Australia. In our previous study, we found that when SC octoploid N. suaveolens accession PI 555565 (8x) is used as the seed parent for a cross with SC tetraploid species N. tabacum (cultivated tobacco; Nicotiana section Nicotiana; 4x), enlarged ovaries (immature fruits) are dropped at 12–17 days after pollination (DAP) and hybrid seeds are never obtained (He et al., 2019). Meanwhile, seed abortion but not enlarged ovary abscission was observed in another interploidy‐interspecific cross between N. suaveolens PI 555561 (SC; 8x) as the seed parent and N. tabacum (4x) as the pollen parent (He et al., 2019). The detailed causes underlying these postzygotic reproductive barriers have not been clarified. Therefore, we hypothesized that a high excess of maternal genome results in enlarged ovary abscission, which is a postzygotic reproductive barrier. Using two interploidy‐interspecific crosses, we conducted comparative analyses of ovary abscission zone (AZ) and seed development. Our cross experiment using ploidy manipulated lines added strong support to this hypothesis, where successive increases in maternal ploidy resulted in successive abnormal seed development.

MATERIALS AND METHODS

Plant materials and growth conditions

The following three accessions of wild species N. suaveolens were used in this study: PI 555561 (2n = 8x = 64), PI 555565 (2n = 8x = 64), and PI 555568 (2n = 4x = 32). The seeds were obtained from the United States National Plant Germplasm System (https://www.ars‐grin.gov/npgs/ index.html). The cultivated species N. tabacum ‘Red Russian’ (2n = 4x = 48), whose seeds were obtained from the Leaf Tobacco Research Center (Japan Tobacco Inc.), was also used. All plants were grown under fluorescent lamps (FL40S·BRN; Toshiba) in a cultivation room (16‐h light/8‐h dark; ~70 μmol m−2 s−1; 25°C) and used for cross experiments. Selfed seeds of PI 555561, PI 555568, and N. tabacum were used for chromosome doubling experiments. The resulting doubled plants were grown under LED light (WPRW01; Tomy) in a cultivation room (16‐h light/8‐h dark; ~70 μmol m−2 s−1; 25°C) and used for cross experiments.

Ploidy level manipulation

Seeds of PI 555561 (8x), PI 555568 (4x), and N. tabacum (4x) were separately soaked in 0.5% gibberellic acid (GA3) solution for 30 min and sown on two layers of filter paper saturated with 15 mL distilled water (DW) in 90‐mm Petri dishes (80–100 seeds per Petri dish). The dishes were placed at 28°C under continuous illumination (~150 μmol m−2 s−1). When seedlings developed fully expanded cotyledons (~1 week after sowing), they were immersed in 100 mL of 0.1% colchicine solution with 2% dimethyl sulfoxide for 12 h at 28°C under continuous illumination. After washing with water twice, the seedlings were transferred to soil‐containing pots. Doubled plants were screened by flow cytometry and chromosome analysis as previously described (He et al., 2019). Eventually, ~200 seeds per accession/cultivar were processed until doubling individuals were obtained, and one PI 555561 (16x) individual, three PI 555568 (8x) individuals, and two N. tabacum (8x) individuals were obtained.

Cross experiments

Five plants of each N. suaveolens accession and N. tabacum (both without ploidy level manipulation) were used for self‐ and interspecific crosses. For chromosome doubled plants, one individual of each accession/cultivar was selected at random and used for self‐ and interspecific crosses. Hand pollination was conducted as follows, for emasculation, corollas of flower buds on female parents were cut longitudinally using a scalpel and anthers were removed using a pair of tweezers 1 day before anthesis. Then, the stigmas were pollinated with pollen grains from the male parents. After pollination, corollas were sealed to avoid pollen contamination. During fruit development, the number of seeds in self‐ and interspecific crosses using PI 555561 (8x) and PI 555565 (8x) was counted using six immature fruits (six replicates) at 6 DAP, the day ovule culture was carried out. Seeds were then collected from the dehiscent fruits (capsules).

Seed weight (n = 30) was determined using an analytical balance (AB54; Mettler Toledo, Greifensee, Switzerland) with each of the three capsules (three replicates) for each cross and expressed as a single seed weight (weight of 30 seeds/30). Seed germinability was evaluated by in vitro seed culture. Seeds were soaked in 0.5% GA3 solution for 30 min and sterilized with 5% sodium hypochlorite for 15 min. The sterilized seeds were then sown on Petri dishes (90 mm diameter, 17 mm deep) containing about 25 mL half‐strength MS medium (Murashige and Skoog, 1962) supplemented with 1% sucrose and 0.8% agar (pH 5.8), followed by culturing for 30 days at 28°C under continuous illumination (~150 μmol m−2 s−1).

Ovule culture was carried out by collecting the flowers of female parents at 6 DAP, after which their corolla, sepals, and styles were removed. The ovaries were surface‐sterilized with 70% ethanol for 30 s and 5% sodium hypochlorite for 5 min. Ovary walls were then peeled using a scalpel to expose the placentas with intact ovules. Fertilized and enlarged ovules were excised using a scalpel, placed in Petri dishes containing 25 mL MS medium supplemented with 7% sucrose and 0.8% agar (pH 5.8), and cultured for 40 days at 28°C under continuous illumination (~150 μmol m−2 s−1).

Measurement of ovary/fruit break strength

The break strength of ovary/fruit AZ was measured using a digital force gauge (FGP‐0.5; Nidec‐Shimpo) every other day starting from 2 to 18 DAP. The break strength was quantified as the force (in gram equivalents) required to detach the ovary/fruit from the branch. Six ovaries/fruits were tested per measurement.

Observation and measurement of pollen tubes

Pollen tube elongation in the styles was examined at 6, 14, 24, 48, and 64 h after pollination. For pollen tube staining, we followed a previously described protocol (Lu et al., 2011). Pollen tubes were observed using a fluorescence microscope (BX50; Olympus, Tokyo, Japan) equipped with a U‐MNIBA2 filter (490–520 nm; Olympus). The tip of the longest pollen tube was marked with ink on the cover glass, and the distance from the stigma to the ink mark was measured as the equivalent pollen tube length. Pollen tube length was measured over three independent experiments.

Histology

Histological analyses of ovary/fruit AZ and developing seeds after pollination were conducted as follows: collected samples were fixed in FAA (formalin: acetic acid: 50% ethanol in a ratio of 5:5:90 v/v/v), after which air in the tissues was evacuated using a vacuum pump and the samples stored at room temperature until further use. The fixed samples were dehydrated in an ethanol and t‐butyl alcohol series (ethanol: t‐butyl alcohol: water = 4:1:5, 5:2:3, 10:7:3, 9:11:0, 1:3:0, 0:10:0). t‐Butyl alcohol was gradually replaced with paraffin at 63°C over a one week period in an open bottle to evaporate traces of t‐butyl alcohol, followed by embedding in paraffin. The embedded samples were then cut into 10–12‐µm‐thick sections using a microtome (PR‐50; Yamato Kohki). Ribbons were placed on glass slides with DW and the slides dried at 50°C on a warming plate overnight. The slides were deparaffinized in xylene for 30 min (twice), then hydrated in a graded ethanol series (100, 95, 85, 70 and 50% in DW). All sections were treated with 3% iron alum first, and then stained with 1% fast green (90% ethanol) for 1 min or 4% safranin (50% ethanol) for 24 h at room temperature. Finally, the sections were imaged using a microscope (BX50; Olympus) under conventional bright field illumination.

Reverse transcription real‐time PCR

Approximately 3 mm of tissue between the peduncle and the branch (AZ represents the intersection) was collected at 2, 6, 10, and 14 DAP. Five tissues were bulked for RNA extraction to obtain sufficient sample amounts. All samples were frozen in liquid nitrogen and stored at −80°C until RNA extraction. Total RNA was extracted using the RNeasy Plant Mini Kit (Qiagen) and then treated with RNase‐free DNase (Promega, Madison, WI) according to manufacturers’ instructions. cDNA was synthesized from 1 μg RNA using oligo (dT)20 primer and ReverTra Ace reverse transcriptase (Toyobo) according to manufacturer’s instructions. Real‐time PCR was performed on a QuantstudioTM 3 Real‐Time PCR system (Thermo Fisher Scientific) with 1X SYBR Green PCR Master Mix (Takara Bio, Shiga, Japan), 0.4 µM of each gene‐specific primer, and 1 µL cDNA template for a final reaction volume of 20 µL. All primer sequences are described by Wu et al., (2012). Cycling conditions were as follows: denaturation at 95°C for 3 min, 40 cycles of 95°C for 15 s, and annealing at 58°C–65°C for 1 min. Relative expression levels were normalized against the NtHistone3 gene. All analyses were performed with three technical and three biological replicates.

Statistical analysis

Data were analyzed using SPSS Statistics software (version 22; IBM). The number of seeds in an ovary, seed weight, ovary/fruit break strength, and gene expression levels were compared using Tukey’s multiple comparisons test.

RESULTS

Different types of reproductive barriers are observed in two interploidy‐interspecific crosses

We compared the reproductive barriers observed in two interploidy‐interspecific crosses, PI 555565 (8x) × N. tabacum (4x) and PI 555561 (8x) × N. tabacum (4x). Seeds were easily obtained in the type I cross PI 555561 (8x) × N. tabacum (4x) and self‐crosses of PI 555561 (8x), PI 555565 (8x), and N. tabacum (4x), whereas 0% of flowers produced capsules and hybrid seeds could not be obtained in the type II cross PI 555565 (8x) × N. tabacum (4x) (Figure S1a and Table S1). In the type II cross PI 555565 (8x) × N. tabacum (4x), ovaries were enlarged after pollination but naturally dropped or were easily removed by an external force (light touch with a hand) during 12 to 17 DAP. Ovary abscission occurred at the intersection of the peduncle and the branch, which is called the AZ (Figure 1a). The morphological appearances of seeds derived from self‐crosses of maternal parents and the type I cross PI 555561 (8x) × N. tabacum (4x) were normal, and the white or ivory‐white seed color at 10 DAP turned light or dark brown at 14 DAP. However, seeds derived from the type II cross PI 555565 (8x) × N. tabacum (4x) were severely shriveled and showed a light brown color at 10 DAP that turned brown at 14 DAP (Figure 1b). The number of seeds per ovary investigated at 6 DAP was significantly low in the type II cross PI 555565 (8x) × N. tabacum (4x) compared with self‐cross of PI 555565 (8x) (Figure 1c). Seeds from the type I cross PI 555561 (8x) × N. tabacum (4x) were also significantly lighter compared with those from both parents (Figure S1b) and failed to germinate (Figure S1c and Table S1). To rescue these seeds, we conducted ovule culture at 6 DAP, which was successful in the type I cross PI 555561 (8x) × N. tabacum (4x) but unsuccessful in the type II cross PI 555565 (8x) × N. tabacum (4x), highlighting the difference in severity of seed abortion between the two interploidy‐interspecific crosses (Figure 2 and Table S2).

FIGURE 1

Ovary abscission and phenotypes of seeds from interploidy‐interspecific crosses. (a) The PI 555565 (8x) ovary after pollination with N. tabacum (4x). Arrowheads indicate the abscission zone (AZ) at the intersection between the peduncle and branch. Scale bars = 10 mm. (b) Appearance of the developing seeds from two self‐crosses and two interploidy‐interspecific crosses. The ovary walls were peeled off and the seeds were photographed at 10 and 14 DAP. Scale bars = 200 µm. (c) Number of seeds per ovary for the indicated crosses. Error bars represent the mean ± standard error of six replicates. Different lower case letters indicate significant differences (<0.05; Tukey’s test). DAP, days after pollination

FIGURE 2

Histological and gene expression analysis of ovary/fruit AZ in self‐crosses and interploidy‐interspecific crosses. (a) Safranin‐stained longitudinal thin sections of AZ. Arrowheads indicate AZ and the arrow indicates lignin deposition. Scale bars = 200 µm. (b) Measurements of ovary/fruit break strength. (c) Relative expression of abscission‐related genes in AZ by real‐time PCR. Expression levels of NtCEL5, NtCEL8, NtEXP1, and NtEXP4 at 2, 6, 10, and 14 DAP were determined and normalized to NtHistone3. Error bars represent the mean ± standard error of six replicates (b) and three biological replicates (c). Different lower case letters indicate significant differences (<0.05; Tukey’s test). DAP, days after pollination

Distinct events occur in the AZ before ovary abscission

To better characterize ovary abscission in terms of AZ development, we histologically monitored the AZ including the adjacent peduncle and branch of two interploidy‐interspecific crosses and two self‐crosses of PI 555565 (8x) and PI 555561 (8x) using longitudinal sections at 10, 12, and 14 DAP. Cells of the AZ are smaller and more densely filled with cytoplasm than are cells in adjacent regions (Addicot, 1982). In all four crosses, smaller, denser cells than the surrounding cells were observed, indicating that the AZ had developed (Figure 2a). Notably, only the AZ of the type II cross PI 555565 (8x) × N. tabacum (4x) exhibited lignin deposition at 12 DAP, while the areas of the connective region between the peduncle and branch were reduced at 14 DAP. To quantitatively measure the adhesion strength of the AZ, we determined the force required to detach the ovary from the branch (break strength; Figure 2b). Break strength in self‐crosses of PI 555565 (8x) and PI 555561 (8x) increased until 12 and 16 DAP, and then decreased to ~150 and 200 g at 18 DAP, respectively. In the type I cross PI 555561 (8x) × N. tabacum (4x), break strength increased until 8 DAP and then gradually decreased; however, the break strength was ~100 g even at 18 DAP, and no ovary dropped even if the ovary was touched by hand. In these three crosses, mature seeds were obtained during 18 to 22 DAP. Conversely, in the type II cross PI 555565 (8x) × N. tabacum (4x), break strength increased until 10 DAP but rapidly decreased after 12 DAP. Indeed, ovaries at 14 DAP dropped when touched by hand. Thus, break strength reduction in the type II cross PI 555565 (8x) × N. tabacum (4x) was consistent with the lignin deposition and reduction of the connective region revealed by histology.

Several abscission‐related genes encoding cell wall hydrolyzing enzymes, such as cellulase, expansin and polygalacturonase, have been identified and their expression patterns characterized in the AZ during tomato flower abscission, Arabidopsis floral organ abscission, and tobacco corolla abscission (Kim et al., 2015; Meir et al., 2010; Wu et al., 2012). In the present study, we performed real‐time PCR to investigate five cellulase genes, six expansin genes, one pectate lyase gene, and one polygalacturonase gene (Table S3), which were previously analyzed in tobacco corolla AZ (Wu et al., 2012). Among them, four genes, NtCEL5, NtCEL8, NtEXP1, and NtEXP4, exhibited a distinct expression pattern (Figure 2c). Regarding the two cellulase genes, NtCEL5 showed significantly higher expression at 10 and 14 DAP, while NtCEL8 showed significantly higher expression at 10 DAP in the type II cross PI 555565 (8x) × N. tabacum (4x) compared with the two self‐crosses and type I cross PI 555561 (8x) × N. tabacum (4x). The two expansin genes NtEXP1 and NtEXP4 exhibited overall low expression levels and little change in expression patterns in the type II cross PI 555565 (8x) × N. tabacum (4x). However, NtEXP1 expression was significantly increased at 10 and/or 14 DAP in PI 555565 (8x) and PI 555561 (8x). This gene was expressed to some degree in the type I cross PI 555561 (8x) × N. tabacum (4x) during the investigation period; however, unlike for PI 555565 (8x), NtEXP1 expression did not increase beyond 10 DAP in the type II cross PI 555565 (8x) × N. tabacum (4x). Similarly, NtEXP4 expression was increased at 10 DAP in PI 555565 (8x) but not in the type II cross PI 555565 (8x) × N. tabacum (4x). This gene also showed a constantly high expression in PI 555561 (8x) and was increased at 10 DAP in the type I cross PI 555561 (8x) × N. tabacum (4x).

Specific abnormal development is observed in developing seeds before ovary abscission

We assumed that the ovary abscission process, including lignin deposition and specific expression patterns of cellulase and expansin genes, may result from fertilization failure. Thus, we investigated whether pollen tube elongation after pollination was normal. In all crosses analyzed, pollen tubes elongated in a straight manner and penetrated the ovaries. Although the timing of when pollen tubes reached the ovaries was slower in the type II cross PI 555565 (8x) × N. tabacum (4x) compared with the self‐cross PI 555565 (8x) (Figure S3), this retardation was not considered crucial for ovary abscission because ovary abscission was not observed in the type I cross PI 555561 (8x) × N. tabacum (4x), where similar retardation of pollen tube elongation occurred.

We next hypothesized that seed developmental defects, as evidenced by appearance and ovule culture, are a key feature of ovary abscission. To test this hypothesis, developing seeds were examined via histology. In two self‐crosses, embryo and endosperm development appeared normal and we observed successive stages of embryogenesis; globular, heart‐shaped, and torpedo‐shaped embryos; and normal developmental transition of the endosperm until 8 DAP (Figures 3 and 4). In contrast, developing seeds in the type I cross PI 555561 (8x) × N. tabacum (4x) showed abnormal development (Figure 3), an earlier developmental transition of the endosperm (at least before 6 DAP) compared with that of the self‐cross PI 555561, and arrested endosperm development. The embryos showed abnormal hypertrophy in the globular state and void space was observed between seed coat and endosperm after 12 DAP. This type of abnormal seed development was termed type I seed abortion. Interestingly, developing seeds from the type II cross PI 555565 (8x) × N. tabacum (4x) also showed endosperm precocious developmental transition before 6 DAP, after which the endosperm region narrowed as if pressed by surrounding cells, while embryos remained in the early globular stage and further development was not observed (Figure 3). This type of abnormal seed development was termed type II seed abortion. These results suggest that type I and type II seed abortion are associated with parental ploidy levels.

FIGURE 3

Abnormal embryo and endosperm development in the type I interploidy‐interspecific cross PI 555561 (8x) × N. tabacum (4x). More severe abnormalities, similar to those in the type II cross PI 555565 (8x) × N. tabacum (4x), were observed by increasing ploidy level in the female PI 555561 (16x) parent, whereas the abnormalities were restored by increasing ploidy level in the male N. tabacum (8x) parent. Longitudinal thin sections of developing seeds at the indicated DAP were stained with fast green. Scale bars = 200 um. VS, void space

FIGURE 4

Abnormal embryo and endosperm development in the type II interploidy‐interspecific cross PI 555565 (8x) × N. tabacum (4x). Abnormalities were restored by increasing ploidy level in the male N. tabacum (8x) parent, as revealed by histological analysis. Longitudinal thin sections of developing seeds at the indicated DAP were stained with fast green. Scale bars = 200 um

Seed abortion and ovary abscission are suppressed by parental ploidy balanced crosses

Notably, the parental species used in the present study have different ploidy levels, and thus two interspecific crosses were 8x (♀) × 4x (♂). It is known that one cause of abnormal endosperm development is unbalanced ploidy levels or effective ploidy (EBN) in both parents (Baek et al., 2016; Coughlan et al., 2020; Johnston et al., 1980; Scott et al., 1998; Sekine et al., 2013; Tonosaki et al., 2018). Hence, we aimed to investigate the effect of varying paternal or maternal ploidy levels on endosperm and embryo development and ovary abscission. To perform crosses between the same ploidy levels, we first increased paternal N. tabacum ploidy levels, which was expected to restore endosperm and embryo development in the type II cross PI 555565 (8x) × N. tabacum (4x) and type I cross PI 555561 (8x) × N. tabacum (4x). We generated octoploid N. tabacum (8x) by colchicine treatment of N. tabacum plantlets (4x) (Figure S4). This increase in chromosome number was critical for restoring both type I and II seed abortion. In both PI 555565 (8x) × N. tabacum (8x) and PI 555561 (8x) × N. tabacum (8x), more than 52.4% of flowers produced capsules with seeds after pollination (Figure S1), while enlarged ovary abscission did not occur. The hybrid seeds were viable, as judged by germination rate (≥96.2%), and their weights were nearly the same as those from self‐crosses of the maternal parents (Figure S1). Moreover, histological observation demonstrated that the endosperm and embryo developed normally as seen in the self‐crosses of PI 555565 (8x) and PI 555561 (8x), although some delay in development was recognized in the cross PI 555565 (8x) × N. tabacum (8x); this delay may be negligible considering the possible differences in fertilization timing between interspecific crosses and self‐crosses (Figure S3).

Next, we conducted other parental ploidy balanced crosses by reducing maternal ploidy levels. We attempted to produce tetraploids of PI 555565 and PI 555561 via anther culture. Although we successfully produced haploid N. tabacum (2x), we could not obtain PI 555565 (4x) and PI 555561 (4x). Thus, we instead used the existing SC tetraploid accession of N. suaveolens, PI 555568, which is closely related to PI 555565 (8x) and PI 555561 (8x) (He et al., 2019), for the cross with N. tabacum (4x). After pollination with N. tabacum (4x) pollen, 100% of PI 555568 (4x) flowers produced capsules with seeds, whose weights were not significantly different from those of the PI 555568 (4x) self‐cross. Moreover, the hybrid seeds were viable with a germination rate of 96.4% (Figure S1) and seed development was normal and similar to that of the PI 555568 (4x) self‐cross, as evidenced by histology (Figure S5).

Type II seed abortion and ovary abscission are triggered by high maternal excess

Although PI 555568 (4x) produced normal seeds after crossing with N. tabacum (4x), it may be possible that genetic differences between PI 555568 (4x) and PI 555565 (8x) or PI 555561 (8x) influenced the crossing results. Furthermore, the question remains of whether type I and type II seed abortion are related or that specific N. suaveolens accessions show type I or type II seed abortion and ovary abscission (i.e., some genetic factors specifically related to each seed abortion type and ovary abscission exist in each N. suaveolens accession). Moreover, whether ovary abscission is directly caused by type II seed abortion remains unanswered. To resolve these questions, two N. suaveolens polyploids, PI 555568 (8x) and PI 555561 (16x), were artificially produced by colchicine treatment (Figure S4). When PI 555568 (8x) was crossed with N. tabacum (4x), enlarged ovary abscission was not observed and 36.2% flowers produced capsules with seeds. Seed weights were significantly lower than those of the maternal parent PI 555568 (8x) and the cross PI 555568 (4x) × N. tabacum (4x), while the germination rate was 33.3%, which was lower than that of the self‐crosses of PI 555568 (8x) and N. tabacum (4x) and the cross PI 555568 (4x) × N. tabacum (4x) (Figure S1). Furthermore, abnormal endosperm and embryo development in the cross PI 555568 (8x) × N. tabacum (4x) was similar to that observed in the type I cross PI 555561 (8x) × N. tabacum (4x) (Figure S5). Thus, it appears that an increase in the ploidy level of PI 555568 (4x) to 8x caused type I seed abortion in the cross with N. tabacum (4x).

Interestingly, enlarged ovaries were dropped at 8–14 DAP and no hybrid seeds were obtained for the cross PI 555561 (16x) × N. tabacum (4x) (Figure S1a). Precocious developmental transition of the endosperm was already observed at 6 DAP, as evidenced by histology (Figure 3). Further endosperm development was not observed and the endosperm region narrowed, similar to that of the type II seed abortion seen in the cross PI 555565 (8x) × N. tabacum (4x). Globular embryos were observed at 6–8 DAP and abnormal hypertrophy of the embryos was observed after 10 DAP. These results demonstrate that an increase in the ploidy level of PI 555561 (8x) to 16x resulted in type II seed abortion and ovary abscission in the cross with N. tabacum (4x), although some differences in embryo sizes were observed between the cross PI 555561 (16x) × N. tabacum (4x) and the type II cross PI 555565 (8x) × N. tabacum (4x). Thus, the PI 555561 (16x) × N. tabacum (4x) cross provided direct evidence that type I and II seed abortion are related phenomena and that the degree of ploidy difference between both parents determines the type of seed abortion, which leads to ovary abscission.

DISCUSSION

In the present study, we showed that two types of seed abortion, type I and II, are observed in the maternal genome excess crosses of PI 555561 (8x) × N. tabacum (4x) and PI 555565 (8x) × N. tabacum (4x), respectively. Early endosperm developmental transition was commonly observed in both types. This is consistent with the EBN or effective ploidy theory where maternal genome or effective ploidy (or EBN) excess generally results in precocious endosperm development, leading to seed abortion. In several plant species, it has been reported that abnormal endosperm development in interspecific crosses can be bypassed by altering the ploidy of one parental species (Bushell et al., 2003; Johnston and Hanneman, 1982; Lafon‐Placette et al., 2017; Tonosaki et al., 2018). In agreement, the relative increase in paternal to maternal genome dosage in the present study (e.g., increase of N. tabacum ploidy from 4x to 8x) restored endosperm and embryo development in both types of seed abortion.

Surprisingly, when we increased N. tabacum ploidy, type II seed abortion was prevented, leading to suppression of another postzygotic reproductive barrier—ovary abscission. This is likely due to the direct relationship between type II seed abortion and ovary abscission. Successive crossing experiments using N. suaveolens accessions with or without chromosome manipulation demonstrated that type I seed abortion and type II seed abortion with ovary abscission occur in a phased manner as maternal to paternal genome dosage increases. Considering that abnormal endosperm development appears to precede the morphological changes and aberrant gene expression patterns in the AZ of ovaries in the type II cross PI 555565 (8x) × N. tabacum (4x), ovary abscission is thus a consequence of type II seed abortion.

Based on our findings, the effective ploidy hypothesis can be extended to include a graded severity of abnormal endosperm development and ovary abscission in maternal excess crosses. When maternal and paternal species have a similar effective ploidy, as in the cross PI 555568 (4x) × N. tabacum (4x), the hybrid endosperm achieves normal development resulting in normal seed development. An increase in the effective ploidy of maternal species in the endosperm gives rise to type I seed abortion or type II seed abortion leading to ovary abscission. Conversely, an increase of the effective ploidy of paternal species in the endosperm restores normal seed development. Recent studies on the genera Solanum, Capsella and Mimulus suggests that naturally evolved lineages may exhibit fractional amounts in the increase/decrease of effective ploidy, leading to various degrees of abnormal endosperm development in inter‐lineage crosses (Coughlan et al., 2020; Lafon‐Placette et al., 2018; Roth et al., 2019;). In the present study, we can infer that the relative effective ploidies between N. suaveolens accessions are PI 555565 (8x) ≈ PI 555561 (16x)> PI 555561 (8x) ≈ PI 555568 (8x)> PI 555568 (4x).

According to the parental conflict hypothesis, maternal and paternal genomes have opposite effects on offspring development (Haig and Westoby, 1989; Haig and Westoby, 1991). In plants, the parental conflict hypothesis states that the maternal genome or maternally expressed imprinted genes reduce seed size and potentially reduce seed set under unfavorable conditions. Conversely, the paternal genome or paternally expressed imprinted genes increase seed size and promote seed set (Bai and Settles, 2015). Therefore, the parental conflict and effective ploidy hypotheses can explain our findings. Based on these hypotheses, we propose a model in which the interaction between the promotive and inhibitive effects of the parental genome determine the developmental destiny of seeds (Figure 5). The promotive effect derived from the paternal genome and the inhibitive effect derived from the maternal genome offset each other in a cross between parents with similar effective ploidy (Figure 5a). In the present study, the maternal genome excess crosses generally had negative effects on endosperm growth. Therefore, we can consider that the inhibitive effect exceeds the promotive effect in the crosses where maternal effective ploidy>paternal effective ploidy, resulting in endosperm and embryo developmental failure (type I seed abortion; Figure 5b). Further increases in the inhibitive effect, such as in crosses where maternal effective ploidy>> paternal effective ploidy, result in more severe endosperm and embryo developmental failure (type II seed abortion; Figure 5c) and lead to premature ovary abscission. By increasing paternal effective ploidy, the promotive effect becomes equal to the inhibitive effect, leading to restoration of endosperm and embryo development (Figure 5d).

FIGURE 5

Model of the interaction between promotive and inhibitive effects that affect endosperm and embryo development in interploidy‐interspecific crosses. (a) In crosses between parents with similar effective ploidy, the promotive effect is equal to the inhibitive effect, and thus normal endosperm and embryo development are achieved. This is applicable to the cross PI 555568 (4x) × N. tabacum (4x). (b) In crosses where maternal effective ploidy is higher than that of paternal effective ploidy, the inhibitive effect is increased and exceeds the promotive effect, causing type I seed abortion. This is applicable to the type I crosses PI 555561 (8x) × N. tabacum (4x) and PI 555568 (8x) × N. tabacum (4x). (c) In crosses where maternal effective ploidy is excessively high compared with that of paternal effective ploidy, a further increase in the inhibitive effect causes type II seed abortion, which leads to ovary abscission. This is applicable to the type II crosses PI 555561 (16x) × N. tabacum (4x) and PI 555565 (8x) × N. tabacum (4x). (d) When paternal effective ploidy is increased experimentally, the promotive effect becomes equal to that of the inhibitive effect, resulting in normal endosperm and embryo development. This is applicable to the crosses PI 555561 (8x) × N. tabacum (8x) and PI 555565 (8x) × N. tabacum (8x). em, embryo; en, endosperm; sc, seed coat

Among the abscission‐related genes that encode cell wall hydrolyzing enzymes, two cellulase genes (NtCEL5 and NtCEL8) and two expansin genes (NtEXP1 and NtEXP4) showed aberrant expression patterns in the AZ of PI 555565 (8x) after pollination with N. tabacum (4x) pollen. These results indicate that NtCEL5, NtCEL8, NtEXP1, and NtEXP4 play key roles in ovary abscission after type II interspecific‐interploidy crosses. When the same abscission‐related genes were investigated during corolla abscission in tobacco (N. tabacum), NtCEL5 and NtEXP5 expression was found increased in the AZ (Wu et al., 2012). It was also reported that CEL5 expression is increased in the AZ during unfertilized flower abscission in tomato (Nakano et al., 2014). Furthermore, lignin deposition—which was observed in the present study—is observed in the AZ during ripened fruit abscission, but not during unfertilized flower abscission in the tomato (Tsuchiya et al., 2015). Therefore, lignin deposition may be a commonly observed phenomenon during ovary/fruit abscission regardless of developmental stage and self‐cross or interspecific‐interploidy cross. Although cellulase and expansin genes are also predicted to be involved, the specific genes are likely to differ based on organ abscission under different situations.

In conclusion, different doses of maternal and paternal genomes (or effective ploidy) disturb endosperm development in interploidy‐interspecific crosses. We provided evidence that stepwise maternal genome (or effective ploidy) excess shows stepwise deterioration in endosperm growth, eventually leading to premature ovary abscission with a similar, but somewhat different, mechanism to that of other organ abscissions. It is also remarkable that ploidy manipulation can bypass seed abortion and ovary abscission. Our findings provide insights for understanding the mechanisms of reproductive barriers in plants.

CONFLICT OF INTERESTS

The authors declare no competing interests.

AUTHOR CONTRIBUTIONS

H.H., S.Y., and T.T. designed the research; H.H. performed the research; H.H., S.Y., and T.T. analyzed the data; and H.H. and T.T. wrote the paper.

Supplementary Material

Click here for additional data file.