An asexual flower of Silene latifolia and Microbotryum lychnidis-dioicae promotes sex-organ development

Silene latifolia is a dioecious flowering plant with sex chromosomes in the family Caryophyllaceae. Development of a gynoecium and stamens are suppressed in the male and female flowers of S. latifolia, respectively. Microbotryum lychnidis-dioicae promotes stamen development when it infects the female flower. If suppression of the stamen and gynoecium development is regulated by the same mechanism, suppression of gynoecium and stamen development is released simultaneously with the infection by M. lychnidis-dioicae. To assess this hypothesis, an asexual mutant without a gynoecium or stamen was infected with M. lychnidis-dioicae. A filament of the stamen in the infected asexual mutant was elongated at stages 11 and 12 of flower bud development as well as in the male, but the gynoecium did not form. Instead of the gynoecium, a filamentous structure was suppressed as in the male flower. Developmental suppression of the stamen was released by M. lychnidis-dioicae, but that of gynoecium development was not released. M. lychnidis-dioicae would have a function similar to stamen-promoting factor (SPF), since the elongation of the stamen that is not observed in the healthy asexual mutant was observed after stage 8 of flower bud development. An infection experiment also revealed that a deletion on the Y chromosome of the asexual mutant eliminated genes for maturation of tapetal cells because the tapetal cells did not mature in the asexual mutant infected with M. lychnidis-dioicae.


Introduction
The basidiomycetous genus Microbotryum is a species-rich member of smut fungi that infect a wide range of host plants belonging to the dicotyledonous families Caryophyllaceae, Dipsacaceae, Lamiaceae, and Lentibulariaceae [1]. The smut fungus Microbotryum lychnidis-dioicae was isolated from Silene latifolia [2]. M. lychnidis-dioicae has long been used as a model for PLOS  studying the ecology, genetics, and the propagation of sexual infection [3][4][5]. Smut disease is transmitted by smut spores attached to insects [6]. Spore germination, meiosis, and the mating of paired basidiospores of smut fungi occurs on the host cell surface [7]. The basidiospores of smut fungi form a dikaryotic mycelium after mating, which penetrates the host plant. The dikaryotic mycelium of smut fungi is observed in intercellular regions, vascular bundles, and apical meristem after invasion of the host plant. Subsequently, the dikaryotic mycelium of smut fungi forms smut spores in the pollen sacs in the blossoms of host plants [8].
Silene latifolia is a dioecious flowering plant with sex chromosomes in the family Caryophyllaceae; individual plants have male and female flowers. The female flowers blossom with 22 autosomes and two X chromosomes, and the male flowers blossom with 22 autosomes, one X chromosome, and one Y chromosome. S. latifolia is a model for the study of the evolution of plant sex chromosomes and ecology [9]. We require sex-chromosome-linked markers to better understand plant sex chromosomes. Therefore, several Y chromosome-linked markers were made using amplified fragment length polymorphism (AFLP) [10], random amplified polymorphic DNA (RAPD) [11], and a technique that combines laser microdissection and polymerase chain reaction (PCR) [12]. The male flowers of S. latifolia with Y chromosomes were irradiated with γ rays or heavy ion beams to produce hermaphrodites, an asexual mutant, and a pollen-defect mutant with deletions to a part of the Y chromosome [13][14][15]. Using the deletion status of these mutants, a map of the S. latifolia Y chromosome was created [10,[16][17][18][19][20][21]. Phenotypes differ between smut infected-male and female hosts. In the infected male, M. lychnidis-dioicae replaces the pollen in the pollen sac with its own fungal spores. In contrast, stamen formation is promoted in the infected female. Smut spores are then formed instead of the pollen in anthers [22,23].
It is thought that three male factors are encoded on the Y chromosome of S. latifolia: gynoecium-suppression factor (GSF), stamen-promoting factor (SPF), and male-fertility factor (MFF). The asexual mutant was first isolated by Donnison et al. [14]. The asexual mutant has X and Y chromosomes, but the SPF region is deleted on the Y chromosome. The flowers of asexual mutants have a filamentous structure instead of a gynoecium in the center of the flower (as observed in males), as well as developmentally suppressed stamens at the early developmental stages [14,24]. Sporogenous cells in developmentally suppressed anthers of the asexual mutant form in the flower buds at very early developmental stages, but parietal cell layers are absent [25].
What happens when M. lychnidis-dioicae infects asexual mutants with partial deletions in the Y chromosome? In asexual mutants, development of the stamen and gynoecium is suppressed. If suppression of stamen and gynoecium development results from the same mechanism, then suppression of gynoecium development should be released when suppression of stamen development is released by M. lychnidis-dioicae infection. In addition, what differences exist in stamens between the infected asexual mutant, the infected male, and the infected female? We infer that we are able to search for genes related to anther development, which should exist in the deleted region of the Y chromosome, by comparing the pollen sacs of stamens in the infected male, infected female, and infected asexual mutant.
In this study, progeny of the asexual mutant were successively produced by crossing the female-like flowers of the asexual mutant with the male flower of a wild-type male [24]. Five infected asexual mutants were found after inoculation with M. lychnidis-dioicae using PCR screening. We focused on gynoecium development in the infected asexual mutant and compared the successive morphological changes in floral organs caused by all M. lychnidis-dioicae infection among males, females, and asexual mutants using scanning electron microscopy and tissue section analysis.

Plant materials and plant growth conditions
Silene latifolia seeds were obtained from an inbred line (K-line) and stored in our laboratory. The K-line was propagated for 17 generations of inbreeding to obtain a genetically homogeneous population. We also used asexual mutants obtained from crossing the inbred K-line with an asexual mutant (ESS1), which was originally one of the heavy-ion beam irradiationinduced Y-deletion mutants identified by Fujita et al. [20]. Plants were grown from vernalized seeds in pots in a regulated chamber at 23˚C with a 16 h light/8 h dark cycle.

M. lychnidis-dioicae inoculation
Sporidia of the Lamole A1 and Lamole A2 strains of M. lychnidis-dioicae were cultured on potato dextrose agar (BD Difco) at 23˚C for 5 days and suspended at 2 × 10 6 cells/mL in distilled water. Sporidia mixtures of A1 and A2 at equal concentrations were used throughout the inoculations. Inoculation treatments were performed on 10-day-old seedlings of S. latifolia on 0.8% agar plates. The base of each 10-day-old seedling was injected with 2 μL of the mixture. Inoculation was repeated after 3 days. Three weeks after inoculation, we transferred the seedlings to soil in pots and grew them in a regulated chamber at 23˚C with a 16 h light/8 h dark cycle.

Scanning electron microscopy
Flowers were fixed overnight in 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2) at 4˚C. After being washed three times with 0.1 M phosphate buffer (pH 7.2), fixed flowers were dehydrated in an ethanol series (30,50,70,80,90,95, and 100% each step for 30 min at room temperature) and kept in 100% ethanol overnight at 4˚C. The ethanol was replaced with isopentyl acetate and the flowers were dried with a critical-point dryer (HCP-2, Hitachi, Tokyo) and sputter-coated with platinum palladium using an ion sputter (E-1010, Hitachi). The flowers were examined in an S-3000N scanning electron microscope (SEM) (Hitachi) operated at 5 kV in high-vacuum mode. The gynoecium on the scanning microphotographs was colored using Photoshop 7.0 (Adobe Systems, San Jose, CA).

Light microscopy
Flowers were double fixed overnight in 4% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2) at 4˚C and post-fixed for 4 h in 2% osmium tetroxide in distilled water. After being washed with 0.1 M phosphate buffer (pH 7.2), the fixed flowers were dehydrated in an ethanol series (30,50,70,80,90,95, and 100% each step for 15 min at room temperature) and kept in 100% ethanol overnight at 4˚C. The ethanol was replaced with xylene and embedded in paraffin. Embedded flowers in paraffin were cut into 10-μm sections using a microtome (RV-240, Yamato, Japan). The cut sections were deparaffinized in xylene and rehydrated in an ethanol series (100, 95, 90, 80, 70, 50, and 30%; each step for 10 min at room temperature). The rehydrated sections were stained with Schiff's reagent. The sections were observed with a microscope (BX60, Olympus, Tokyo, Japan).

Screening for asexual mutants with a partial Y chromosome deletion
Artificial smut infection has been established only at the cotyledon stage. Therefore, it is necessary to cross a WT male and female-like flowers of the asexual mutant to obtain progeny. The progeny are treated immediately after germination for smut infection and examined for smut spores in anthers at the first flower. To screen for asexual mutants from crossing of a sexual mutant (ESS1) and inbred line (K-line), we performed polymerase chain reaction (PCR) screening using four STS makers (MS4, MK17, ScQ14, and SlAP3) and flower phenotype screening. The STS markers (MS4, MK17, ScQ14, and SlAP3) are described in Kazama et al. [21]. Genomic DNA was extracted from fresh leaves using a DNeasy Plant Mini Kit (Qiagen), according to the manufacturer's instructions. PCR amplification was performed using Blend Taq polymerase (Toyobo, Tokyo, Japan) with a Thermal Cycler Dice TP600 (TaKaRa Bio, Otsu, Japan). The conditions were 5 min at 94˚C, followed by 35 cycles of 1 min at 94˚C, 1 min at 60˚C, and 1 min at 72˚C, with final 5 min extension at 72˚C. Each 50-μL reaction mixture contained 50 ng of template DNA, 5 μL of 2 mM dNTPs, 5 μL of 10× Blend Taq Buffer, 1 μL of Blend Taq, and 1 μL of each primer at 10 μM. The reaction products were electrophoresed on 1.5% agarose gels. After staining with ethidium bromide, fragments were visualized on a UV illuminator (Atto, Tokyo, Japan). Wild-type male and female genomic DNA was used as a control.

Stages of flower bud development in an asexual mutant of S. latifolia
In this study, we newly defined stages of flower bud development of the asexual mutant to correspond with stages of flower bud development in male and female flowers. Stages of flower bud development of the asexual mutant were divided into 11 stages (Fig 1A-1K). Flower bud development in the asexual mutant at stages 1 to 6 was similar to that of males. The flowers of the asexual mutant with developing stamen primordia and gynoecium primordia at stages 1 to 5 of flower bud development showed hermaphroditic morphology (Fig 1A-1F). However, although expansion of the stamen filament occurred at stage 7 in male flowers [23] (S1 Fig), expansion of the stamen filament was not seen in the asexual mutant at stage 7, (Fig 1G). Stamen development was suppressed at stage 8 of flower bud development, and the gynoecium became a filamentous structure (Fig 1H). Only the petals and filamentous gynoecium developed at stages 9 to 11 (Fig 1I-1K). Suppressed anthers became trapezoidal or rectangular, similar to structures observed in the wild-type female (Fig 1).

Morphological changes in the asexual mutant caused by M. lychnidisdioicae
Both the asexual flower and the female-like flower blossom are produced in the asexual mutant ESS1, which has one X chromosome and one Y chromosome with a deletion of the SPF region. The female flowers had five styles, but the female-like flowers had only two styles. It was possible to produce progeny of the asexual mutant by crossing the female-like flowers in the asexual mutant with the male flower in the wild-type male because gynoecia of the female-like flowers in the asexual mutant were fertile. Artificial smut infection has been established only at the cotyledon stage. Therefore, it is necessary to cross a WT male and female-like flowers of the asexual mutant to obtain progeny. The progeny are treated immediately after germination for smut infection and examined for smut spores in anthers at the first flowering. As a result, we obtained 234 seeds, and those seeds were sowed. Of the germinated seeds, 188 S. latifolia sprouts were inoculated with M. lychnidis-dioicae. We genotyped mock or infected plants 3 months after infection, based on genotyping by PCR using four markers: MK17, ScQ14, SlAP3, and MS4 (Fig 2). This produced a female with two X chromosomes, a male with a Y chromosome, and asexual flower mutants with a Y chromosome containing the deleted region and an X chromosome. However, none of the individuals without an X chromosome grew to the flowering stage. Therefore, the seed likely suffered lethality before germination. Table 1 shows the obtained progeny and the M. lychnidis-dioicae infection rate. We obtained 66 female individuals with two X chromosomes, 91 male individuals with the X chromosome and the intact Y chromosome, and 15 asexual mutant individuals with the X chromosome and the Y chromosome with deletion of the SPF region (Table 1). In these plants, the infected female was found as 48 individuals, the infected male was found as 54 individuals, and the infected asexual mutant was found as 5 individuals (Table 1). Flower bud development in the infected asexual mutant was divided into 12 stages (Fig 3A-3L). The morphology of the infected asexual mutant appeared to be similar to that of the male and healthy asexual mutant at stages 1 to 6 ( Fig 3A-3F). However, extension of the stamen filament in the infected asexual mutant was confirmed, as well as in the male, at stages 7 and 8 of flower bud development, and  developmental suppression of the stamens did not occur at stage 8 in the asexual mutant ( Fig  3G and 3H).

Morphological changes of anther locule caused by M. lychnidis-dioicae
In this study, we observed anther locules of the healthy male, the infected male, as well as those of the infected female and infected asexual mutants (Figs 3 and 4A-4Y, and S1 Fig). Each of the infected plants was observed and compared at stages II to VI of anther development. Archesporial cells and epidermal cells formed at stage II of anther development in the healthy male ( Fig 4A). The anther locule at stage V of anther development was composed of a layer of epidermis, a layer of endothecium, a middle layer, a tapetum layer and pollen mother cells as a result of the differentiation of archesporial cells (Fig 4A-4D and 4F-4I). The middle layer and the tapetum layer caused programmed cell death at stage VI of anther development (Fig 4E and 4J).
In the infected males, the primary parietal cell layer was divided into two layers at stage III of anther development (Fig 4L). Furthermore, the secondary parietal cell layer was divided into two layers at stage IV of the anther development (Fig 4M). In the infected males, the endothecium was present in the two layers, unlike in the healthy males. Furthermore, the tapetum and the middle layer in infected males showed different morphology from those in the healthy males at stage V of anther development (Fig 4N). Unlike in the healthy male, in the infected male, the tapetum rapidly disintegrated at stage VI of anther development, where the hyphae of M. lychnidis-dioicae normally grew (Fig 4O).
In the infected female, only archesporial-like-cells and epidermal cells existed, but primary parietal cells and primary sporogenous cells did not exist in stage III of anther development (Fig 4Q). Differentiation of the archesporial-like-cells did not occur at later stages (Fig 4P-4S). Growth of M. lychnidis-dioicae was also observed until stage VI of anther development in the infected female (Fig 4T). In the infected asexual mutant, development of the anther locules in developed stamens caused by M. lychnidis-dioicae was similar to that in the infected male at stages II to IV of anther development (Fig 4U-4W), whereas the morphology of the tapetum at stage V of anther development was different from that in the infected male (Fig 4X). Growth of M. lychnidis-dioicae was also observed at stage VI of anther development (Fig 4Y). Thus, it was found that M. lychnidis-dioicae in the infected asexual mutant and infected female grew at the center of the anther locule, where the tapetum was absent, although M. lychnidis-dioicae in the infected male grew along the tapetum (Figs 4N, 4S, 4X and 5).  Table 1  An asexual flower of Silene latifolia and Microbotryum lychinidis-dioicae was multi-layered, and the tapetum and middle layer showed different morphology from that of the healthy male. However, the endothecium, middle layer, tapetum, and pollen mother cell in the anther locule of the infected male were present as in the healthy male. The stamen of the female of S. latifolia is elongated by infection with M. lychnidis-dioicae. The anther of the female has only archesporial cells and epidermal cells in its locules and does not cause differentiation of the archesporial cells during the development of sporogenous cells at very early developmental stages. Therefore, we suggest that genes related to formation of the endothecium, middle layer, and tapetum are located on the Y chromosome. Farbos et al. [25] observed developmentally suppressed anthers of a healthy asexual mutant at early stages of anther development. As a result, suppressed anthers of the asexual mutant formed archesporial cells, but parietal cells were absent. In this study, normal differentiation of the tapetum was not observed in the infected asexual mutant. Morphology of the anther locule in the infected asexual mutant resembled the Arabidopsis dysfunctional tapetum1 (dyt1) mutant, in which the DYT1 gene (related to differentiation of the tapetum cells) was mutated [26]. Such genes may be lacking in the asexual mutant infected with M. lychnidis-dioicae. Although ESS1 and previously identified asexual mutants [25] were produced in different An asexual flower of Silene latifolia and Microbotryum lychinidis-dioicae ways and the extent of the contribution of the M. lychnidis-dioicae infection to the production of abnormal tapetum cells is largely unknown, it is possible that the genes related to differentiation of the tapetum and middle layer were located around the SPF region of the ESS1 Y chromosome.

Morphological changes induced by M. lychnidis-dioicae
It is thought that elongation of the stamen occurred in the infected female of S. latifolia because M. lychnidis-dioicae plays an alternative role in the Y chromosome of infected females of S. latifolia, which lack the Y chromosome [27]. SLM2 is a homolog of the PISTALA (PI) Bclass gene and is expressed in the stamen and petal primordia [28]. It is thought that smut fungus regulates the expression of SLM2 since this gene is not expressed in the stamen primordia of the female but expressed in the stamen primordia in the infected female [21]. In addition, SLM2 is not expressed in the stamen primordia of asexual mutants, in which the anther development does not occur as in the female [15]. The asexual mutant has a pair of X and Y chromosomes. In the mutant, genes related to gynoecium suppression located in the GSF region are intact, whereas the SPF region is deleted. Therefore, it is thought that development of the stamen does not occur after stage 7 of flower development. This is due to the suppression of gynoecium development caused by GSF function and a lack of promotion of stamen development due to the deletion of the SPF region. Development of the gynoecium is suppressed and replaced by the filamentous structure because the GSF on the Y chromosome is intact (Figs 1  and 2). When M. lychnidis-dioicae infects the asexual mutant, the gynoecium displays a filamentous structure as in the healthy asexual flower mutant. Therefore, it is suggested that the suppression of gynoecium development was not released by the infection of M. lychnidis-dioicae. In addition, it is thought that M. lychnidis-dioicae has a function similar to SPF since the elongation of the stamen that is not observed in the healthy asexual mutant was observed after stage 8 of flower bud development.
Stamens develop only after infection with M. lychnidis-dioicae, but the maturation of tapetal cells is abnormal in the anther locule. This phenomenon suggests that the asexual flower mutants lack genes for stamen development and for tapetal cell development. In addition, infected female flowers reportedly suppress cell death and promote cell-cycle arrest in the stamen [29]. Therefore, M. lychnidis-dioicae plays a role in promoting the development of stamens with a defective Y chromosome even in asexual flower mutants. In contrast, M. lychnidis-dioicae is likely involved in the promotion of stamen development but not in its maturation because the female flower does not develop tapetal cells. The genetic map in Kazama et al. [30] suggested that the asexual flower mutant (ESS1) in this study has the smallest deletion region among the known mutants. Because the tapetal cells are not mature in the asexual flower mutant infected with M. lychnidis-dioicae ( Fig 5) and based on the results of Kazama et al. [30], we speculate that the genes involved in tapetum maturation are in the vicinity of the SPF.

M. lychnidis-dioicae localization and timing of infection influence anther locules
Microbotryum lychnidis-dioicae was observed among vascular bundles, intercellular regions of epidermal cells, root tip cells, and apical cells after infection [8], [31]. As a result, it is thought that M. lychnidis-dioicae penetrated into the anther through the vascular bundles, and that the timing of M. lychnidis-dioicae infection was at stage 7 of flower bud development. This is because M. lychnidis-dioicae was observed in connective cells at stage III of anther development, which corresponds to stage 7 of flower bud development [31], and development of the stamen did not occur at stage 8 of flower bud development (Fig 3). The growth area of M. lychnidis-dioicae in the anther locule is different among the infected male, the infected asexual mutant, and the infected female (Fig 4N, 4S and 4X). In the infected male, tapetal cells and pollen mother cells collapse rapidly, and M. lychnidis-dioicae hyphae grow normally [23], [31] (Fig 4O). M. lychnidis-dioicae hyphae invade the anther after its formation rather than growing rapidly [31]. We suggest that M. lychnidis-dioicae was able to recognize the tapetum, such that the growth of M. lychnidis-dioicae started with tapetum dissolution and was observed at the position where the tapetum cells were present (Fig 4X). Alternatively, M. lychnidis-dioicae is reported to grow around tapetal cells because these are thinner and more fragile than other cell types [31]. In the absence of, or in the presence of immature, tapetal cells, the lack of disintegrating cells made M. lychnidis-dioicae appear to grow in the middle of the anther. Therefore, we suggest that the different growth areas of M. lychnidis-dioicae observed in infected males, infected asexual mutants and infected females was due to the absence of, or the presence of an immature, tapetum.