Using the cryo-fixation/freeze-substitution method, we studied the ultrastructural changes and behavior of vacuoles and related organelles (rER and Golgi bodies) during microspore and pollen development, and pollen maturation of Arabidopsis thaliana. In young microspores forming exine (pollen outer cell wall), vacuoles looked like those of somatic cells. In microspores during the formation of intine (inner cell wall), a large vacuole appeared which was made by fusion of pre-existing vacuoles and probably absorption of solutions. In the young pollen grain after the first mitosis, a large vacuole was divided into small vacuoles. The manner of division was not by binary fission and centripetally, but by the invagination of tonoplasts from one side to the opposite side of a vacuole. After the second mitosis, somatic type vacuoles disappeared. In mature pollen grains just before germination, membrane-bound structures containing fine fibrillar substances (MBFs) appeared. The MBFs were considered to be storage vacuoles. In pollen grains from flowers in bloom, MBFs changed to lysosomal structures with acid phosphatases (lytic vacuole). They gradually increased in number and volume, and decomposed the cytoplasm. The autolysis of pollen grains is the first finding in this study, which may contribute to the loss of ability of pollen germination after anthesis.
(Received May 6, 2003; Accepted August 29, 2003)
Pollen development takes place within the anther. Pollen mother cells (meiocytes) produce a tetrad of haploid microspores after meiosis. The microspores become free from meiocyte cell wall after dissolution of the cell wall. During the tetrad stage and microspore formation, the exines (outer pollen walls) and intines (inner pollen walls) are synthesized around the cells. Following the appearance of a large vacuole and the migration of the nucleus toward the cell periphery, the asymmetric division (microspore mitosis) occurs, producing a vegetative cell containing a generative cell. Then pollen grains mature to accumulate substances which will be used for pollen germination and early pollen tube elongation (for details, see reviews, Bedinger 1992, McCormick 1993, Hesse 1995). These drastic phenomena occurring during pollen development have attracted the interest of botanists and ultrastructural studies on this process have been carried out in various plant species. The majority of these studies was carried out with conventional chemical fixation for electron microscopy. In contrast, it was shown that cryo-fixation/freeze substitution technique was superior to conventional chemical fixation in its ability to preserve fine structures in pollen grains of Tradescantia (Noguchi and Ueda 1990) and Brassica (Iwano et al. 1999), and pollen tubes of Tradescantia (Noguchi 1990). This technique also has progressed the research of the cytoskeleton system in pollen tubes of Nicotiana (Lancelle et al. 1987) and Pyrus (Tiwari and Polito 1988).
Arabidopsis thaliana whose complete DNA sequence is now known, is used as an experimental material in various fields of plant biology. Several ultrastructural studies of microspore and pollen development have been reported: exine formation (Paxson-Sowders et al. 1997); final maturation and rehydration (Van Aelst et al. 1993) and the development stages (Owen and Makaroff 1995, Kuang and Musgrave 1996, Zhang et al. 2002). All of these were carried out using chemical fixation methods. Therefore, we re-examined the ultrastructures of pollen grains during developmental stages by cryo-fixation/freeze substitution technique and have got many new findings. In this paper, we focus our attention on vacuoles and related organelles (the endoplasmic reticulum and Golgi bodies), and report their behavior and ultrastructural changes during pollen development and maturation.
In the young microspores during exine formation, a nucleus was located at the cell center, and plastids with small amount of starch grains and thylakoids that did not stack to grana occupied more than half of the cytoplasm (Fig. 1a, b). Vacuoles were distributed throughout the cytoplasm and some of them were dumbbell shaped. The electron density of contents in vacuoles was similar to that of the cytoplasm. Golgi bodies seemed to not actively produce vesicles. The rough endoplasmic reticulum (rER) did not develop.
In microspores during the intine-formation, where nucleus was located at the cell periphery and plastids markedly decreased in number and size, a large vacuole appeared in addition to small vacuoles (Fig. 1c). The electron density of the contents in the vacuoles was much less than those of cytoplasm (Fig. 1c, d). The rER and Golgi bodies were distributed throughout the cytoplasm. Golgi bodies increased in number and their cisternae, and Golgi vesicles seemed to migrate not to the vacuole but to the cell surface (Fig. 1d).
After generative cell formation, a large vacuole started to split into small vacuoles (Fig. 2a). The partition did not occur by binary fission and centripetally, but by the invagination of tonoplasts that occurred at several points and developed from one side to the opposite side of a vacuole (Fig. 2a, b). Electron-transparent droplets without a clear membrane, which began to appear before the first mitosis, increased in number and surrounded only the generative cells (Fig. 2a). As these droplets were stained with p-phenylene diamine (Boshier et al. 1984) (Fig. 2c), they were identified not as vacuoles but as lipid bodies. Small partite vacuoles accumulated globular materials, <160 nm in diameter, with higher electron densities than those of other inclusions. Sometimes, fusion of these small vacuoles with the plasma membrane was observed (Fig. 2b, large arrow). Golgi bodies were mainly located at the cell periphery, especially near the caving plasma membrane that separated the cell from the intine (Fig. 2a). In the following stage (Fig. 2d), dividing vacuoles and small vacuoles dispersed throughout the cytoplasm, and the plasma membrane became intensely convoluted. Between the convoluted plasma membrane and the intine there were many membrane-bound structures (profile in section). They seemed to be secreted globular materials and the cross-sections of the convoluted plasma membrane.
The pollen of Arabidopsis is of the tri-cellular type, which means the generative cell divides again to form two sperm cells before pollen germination. After the second mitotic division, vacuoles with the similar appearance to those in the previous stage were no longer observed. Numerous electron-dense structures about 250 nm wide appeared (Fig. 3a). As the connection between electron-dense structures and rER was observed (Fig. 3b), they were considered to be produced by ER cisternae. The rER became more conspicuous, and its cisternae were dilated and attached with many ribosomes on their surface.
In mature pollen grains in dehiscing anthers of flowers before anthesis, whose stamens and pistil were the same height, the electron-dense structures in the previous stage had disappeared (Fig. 4). Many membrane-bound structures with fibrillar substances (MBFs) appeared. The MBFs occupied the central part, more than half of the cytoplasm in many central sections of the pollen grains (Fig. 4a). Highly dilated rER cisternae surrounded spherical materials with thinner outlines than a unit-membrane (Fig. 4b, c). As these spherical materials reacted with p-phenylene diamine (data not shown), they were identified not as vacuoles but lipid bodies. Golgi bodies increased in number and seemed to actively produce vesicles, because many Golgi vesicles 120–150 nm in diameter were observed throughout the cytoplasm, especially near the plasma membrane. These vesicles did not fuse with the MBFs and were considered to be P-particles (polysaccharide particles) appeared in mature pollen grains and pollen tubes of various plants (Weber 1988). When pollen grains were treated with 0.07% neutral red, a dye that is taken up selectively by vacuoles (Mahlberg 1972, Ehara et al. 1996), they became red (Fig. 5a). The redness was not homogeneous within the cells, and the neutral-red-positive area coincided with the location of MBFs (compare Fig. 5a with Fig. 4a). Just after pollen germination, the cytoplasm at the bases of the pollen grains was stained more readily than the proximal cytoplasm near the pollen tube (Fig. 5b); this also coincided with the distribution of MBFs in the germinated pollen but not pollen tubes (data not shown).
In the pollen grains harvested from flowers in bloom whose stamen became higher than the pistil, the cytoplasm showed various profiles (Fig. 6). Some were similar to those of mature pollen grains from flowers before anthesis (data not shown). The first structural change occurred within the MBFs. They began to contain osmiophilic globoids <90 nm in diameter and decrease in volume (Fig. 6a). In some pollen grains, both MBFs with osmiophilic globoids and electron-transparent membrane-bound structures with osmiophilic globoids were observed. In other pollen grains, only the latter were observed, and their globoids increased in diameter, to <150 nm (Fig. 6b). The electron-transparent membrane-bound structures fused to each other (Fig. 6b large arrow), became larger, and contained several osmiophilic globoids and tubular and vesicular structures (Fig. 6c, d). Their image liked as lysosomes observed in animal cells, therefore we performed a cytochemical detection test for acid phosphatase (a marker enzyme of lysosomes) to determine the chemical nature of the structure. Because products of reactions involving acid phosphatases were detected in the structures (Fig. 6e), we considered that the structures were lysosomes. With the appearance of lysosomal structures, the lipid droplets surrounded by rER disappeared (Fig. 6c, d). The ratio of the total lysosome volume to the cell volume in a pollen grain varied. In pollen grains in which more than half of the cytoplasm had decomposed, MBFs and the lipid droplets surrounded by rER had completely disappeared, and the rER, Golgi body cisternae and mitochondria were intensively swollen (Fig. 6d).
The transformation of vacuoles during the development of microspores and pollen, as observed in the present investigation, is summarized in Fig. 7.
The appearance of a large vacuole before generative cell formation (first mitosis) and its degradation into many small vacuoles are generally observed in pollen of various plant species. Owen and Makaroff (1995) reported that the enlargement of vacuoles depends on the fusion of pre-existing small vacuoles in A. thaliana. One of our results supports it, because dumbbell-shaped vacuoles observed before the appearance of a large vacuole (Fig. 1a) are considered as the fusion process rather than the division process of vacuole. In addition, our result that the electron density of contents in the large vacuole decreased than those of pre-existing small vacuoles (compare Fig. 1c, d to Fig. 1a, b) indicates the absorption of some solutions into vacuoles. The microspores also increased in volume (about 1.2 times) during the period in spite of inactive appearance of rER and Golgi bodies. Probably, the enlargement of vacuole occurs by both the fusion of pre-existing small vacuoles and the absorption of some solutions.
Until now, the information about the manner of vacuole degradation was limited not only in A. thaliana but also in other plant species. This was probably because the tonoplasts of vacuoles were not smooth but wavy in pollen prepared by chemical fixation, which did not give enough structural resolution for determining the manner of division. Our data clearly showed that the vacuoles divide not by binary fission and not centripetally like plastids and mitochondria (Kuroiwa et al. 1998) but by the invagination of tonoplast that occurred at several points and developed from one side to the opposite side of a vacuole (Fig. 2a, b). We observed the fusion of divided vacuoles with the plasma membrane (Fig. 2b). Vacuoles (both lytic and storage) usually do not exocytose in somatic cells, but exocytosis seems to occur in pollen, because the vacuoles disappeared completely from the pollen grains at the next stage. If the large vacuole (the origin of small vacuoles) developed by the absorption of solutions, then the partite small vacuoles would be expected to dilute the lytic abilities and never disturb the surface environment of the cells, even if they secreted their inclusion by exocytosis.
The behavior of vacuoles during and just after sperm cell formation in pollen grains of A. thaliana, varied in different reports. Regan and Moffatt (1990) reported that, according to the neutral red staining, vacuoles disappeared during the first or second mitotic divisions. In contrast, the existence of vacuoles during these periods was reported in studies using electron microscopy (Owen and Makaroff 1995, Kuang and Musgrave 1996). In our study, the typical vacuoles that are similar in appearance to those observed in young pollen grains (Fig. 1, 2) were not observed in the pollen grains just after sperm formation (Fig. 3). During these periods, electron-dense structures were actively produced from the rER cisterna. Although, we have not identified whether it is a kind of vacuole, it is clear that the continuity between the electron-dense structure and the vacuoles in young pollen grains does not exist, because the former is newly produced from the rER cisterna in these periods.
In mature pollen grains of A. thaliana prepared by chemical fixation, many vacuoles with an average diameter of 300 nm have been reported (Van Aelst et al. 1993, Kuang and Musgrave 1996). However, their identification as vacuoles probably depended on their appearance: circular membrane-bound structures in electron microscopic sections. In our study, MBFs appeared only in the mature pollen grains from flowers just before anthesis (Fig. 4). The profiles of MBFs are very different from those of vacuoles observed by chemical fixation; the former contained fine fibrils but the inclusions in the latter are transparent. Therefore, we identified the MBFs as a kind of vacuoles for the following reasons. Regan and Moffatt (1990) reported that neutral red positively stained mature pollen grains of A. thaliana, and we ourselves confirmed the stainability of mature pollen grains by neutral red (Fig. 5). In both experiments, the neutral-red-positive area coincided with the location of MBFs. Further, the redness was not homogeneous within the cells. Therefore, we consider that the neutral red had entered sub-cellular compartments. However, we could not find vacuole-like structures other than MBFs in mature pollen grains. The specific appearance of MBFs in only mature pollen grains suggests that they are storage vacuoles. It has already been reported that neutral red stains not only vegetative vacuoles but also protein-storage vacuoles isolated from maturing and germinating pumpkin seeds (Hara-Nishimura et al. 1987). In addition, we already have supporting data: expression of the gene for the β-vacuolar processing enzyme (β-VPE) that is involved in the maturation of seed proteins in protein storage vacuoles (Kinoshita et al. 1999) was detected in mature pollen grains harvested from flowers just before anthesis of Arabidopsis plants transformed with pβVPE::gus genes (unpublished data). The origin and detailed behavior of the MBFs will be reported elsewhere.
It is thought that mature pollen grains usually become desiccate and enter a phase of stasis. After pollination, the pollen grain undergoes a period of rehydration and its metabolic machinery is reactivated (Heslop-Harrison 1987, Van Aelst et al. 1993). In contrast, we observed the autolysis of mature pollen by lysosomal structures with acid phosphatases. Our results suggest that the appearance of osmiophilic globoids in MBFs is the start of pollen autolysis. However, similar electron-dense globoids in mature pollen grains in Chlorophytum elatum were considered to be a calcium pool that might be utilized in the course of pollen germination (Butowt et al. 1997). It is not clear whether the osmiophilic globoids in A. thaliana include calcium or not. We prefer to consider that they come from the inclusions of MBFs, because the osmiophilic globoids increased in size with the decrease of the fibrillar substances and the size of MBFs. If the MBFs are storage vacuoles containing materials for pollen germination, as we assumed, then it would be reasonable to hypothesize that the MBFs are decomposed when the mature pollen grains are not able to germinate. Taken together, our results suggest that the process of mature pollen autolysis is as follows. MBFs decompose their contents and transform into the electron-transparent membrane-bound structures with osmiophilic globoids. Then they fuse with each other, increase in volume, and acquire a lysosomal nature through the addition of acid phosphatases. As the lysosomal structures increase, the highly dilated rER surrounding lipid bodies (one of the prominent structures in mature pollen grains) disappears, and then the rER, Golgi bodies, and mitochondria are decomposed after swelling. The autolysis of mature pollen grain is the first finding not only in A. thaliana but also other plant species. Therefore, we examined many pollen grains by electron microscopy samples made at different times and concluded that they are not artifacts but a real phenomenon in A. thaliana. The following facts may support our conclusion. (1) In A. thaliana, the height of stamen is same as that of pistil in the flowers just before anthesis; in contrast, the pistil became higher than pistils after anthesis (Fig. 7). (2) The germination ratio of pollen in vitro was highest in pollen grains from flowers just before anthesis and declined after anthesis of the flowers (unpublished data). (3) The cytoplasm of pollen just after germination looks more like those of pollen grains harvested from flowers just before anthesis than that of pollen grains from flowers in bloom. Pickert (1988) reported that pollen of A. thaliana has a longevity of less than 3 d, which is much shorter than that of Nicotiana tabacum pollen. N. tabacum pollen keeps the germination ability more than 3 d in room temperature (Noguchi and Morré 1991) and more than 1 year in the refrigerator. In A. thaliana, the pollen grains are gradually decomposed by lysosomal structures after anthesis and may lose their ability to germinate in a short time. The autolysis of mature pollen grains may contribute to the self-pollination in A. thaliana.
Material and Methods
Plant materials and growth conditions
Seeds of A. thaliana L. (Heynh) ecotype Wassilewskija (WS) were sown on vermiculite mixed with pumice grains (1 : 1 by volume) and grown at 22°C under continuous fluorescent light at a photon flux density of 45–55 µmol m–2 s–1.
Cryo-fixation and freeze-substitution
Microspores and immature pollen grains in non-dehiscing anthers were extruded onto aluminum foil 4 mm square and rapidly frozen in liquid propane at –190°C. Most of the mature pollen grains in dehiscing anthers were attached to formvar films mounted on wire loops 8 mm in diameter and then frozen. Some pollen grains were frozen by a high-pressure freezing apparatus (BAL-TEC HPM010). The frozen cells were transferred to cooled acetone at –85°C containing 4% osmium tetroxide and 0.2% uranyl acetate. After substitution for 48 h at –85°C, the cells were washed with acetone and embedded in Spurr resin. Sections were stained with lead citrate and examined under an electron microscope (Hitachi H-7000).
Cytochemical detection of acid phosphatase activity
Pollen grains were fixed with 3% glutaradehyde buffered with 0.05 M Na-cacodylic acid (pH 7.2) for 10 min at 4°C. The cells were washed for 30 min with distilled water and then incubated for 2 h at 37°C in reaction medium containing 0.15 g of Na2-β-glycerophosphate as substrate, 0.012 g of lead nitrate, 0.11 g of potassium sodium tartrate, 0.004 g of manganese chloride and 10 ml of 0.05 M acetate buffer (pH 5.2), with a transfer to fresh medium every 30 min. The cells were then washed successively with distilled water, 1.5% acetic acid and distilled water again, and were immersed in 1% ammonium sulfide solution for 2 min. The cells were washed with distilled water, and then post-fixed with a 1% osmium tetroxide solution buffered with 0.05 M Na-cacodylic acid (pH 7.2) for 2 h, washed with water, dehydrated with acetone, and embedded in Spurr resin (for details, see Noguchi 1976).
Pollen grains were pre-fixed with 4% glutaraldehyde and post-fixed in 1% osmium tetroxide. During dehydration with serial dilutions of ethanol, 0.1% p-phenylene diamine was added to 70% ethanol to stain the lipids (Boshier et al. 1984).
Neutral red treatment
Pollen grains were soaked in 0.075% neutral red for 60 min for differentiation of vacuoles.
Corresponding author: E-mail, firstname.lastname@example.org; Fax, +81-742-20-3416.