We describe some previously uncharacterised stages of fertilization in Arabidopsis thaliana and provide for the first time a precise time course of the fertilization process. We hand-pollinated wild type pistils with wild type pollen (Columbia ecotype), fixed them at various times after pollination, and analysed 600 embryo sacs using Confocal Laser Scanning Microscopy. Degeneration of one of the synergid cells starts at 5 Hours After Pollination (HAP). Polarity of the egg changes rapidly after this synergid degeneration. Karyogamy is then detected by the presence of two nucleoli of different diameters in both the egg and central cell nuclei, 7–8 HAP. Within the next hour, first nuclear division takes place in the fertilized central cell and two nucleoli can then be seen transiently in each nucleus produced. In a second set of experiments, we hand-pollinated wild type pistils with pollen from a transgenic promLAT52::EGFP line that expresses EGFP in its pollen vegetative cell. Release of the pollen tube contents into the synergid cell could be detected in living material. We show that the timing of synergid degeneration and pollen tube release correlate well, suggesting that either the synergid cell degenerates at the time of pollen tube discharge or very shortly before it. These observations and protocols constitute an important basis for the further phenotypic analysis of mutants affected in fertilization.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
In flowering plants, double fertilization is a necessary prelude to embryo and endosperm development within the seed. The structures of the male and female gametophytes involved in this reproduction step have been quite extensively studied in a number of species (see reviews: McCormick, 1993; Reiser and Fischer, 1993) including Arabidopsis thaliana (Christensen et al., 1997; Owen and Makaroff, 1995). Mutants that affect embryo sac or pollen development in A. thaliana (see review: Yang and Sundaresan, 2000), as well as pollen germination or pollen tube growth toward its female target (see Grini et al., 1999; Johnson and McCormick, 2001; Mayfield and Preuss, 2000; Procissi et al., 2001; Shimizu and Okada, 2000) have also been isolated. Thus, some of the molecular and genetic mechanisms that control gametophyte development and function are being revealed using genetic approaches. However, in A. thaliana we continue to lack a precise description and time course of the fertilization steps, i.e. the plasmogamy and karyogamy. We also lack mutants affected in these developmental stages to further understand the molecular mechanisms underlying fertilization. In order to identify key steps in fertilization, and later isolate corresponding mutants in A. thaliana, we carried out a precise cytological study. We used Confocal Laser Scanning Microscopy (CLSM) on fixed hand-pollinated pistils (see Christensen et al., 1997), as well as crosses with pollen expressing GFP (Cheung, 2001). In this paper, we show that one synergid cell degenerates shortly before, or simultaneously with, the release of the pollen tube contents into the embryo sac. We present data suggesting that a very early fertilization step then consists of a change in the egg cell polarity and that karyogamy can be visualized by the transient presence of two nucleoli in both the fertilized egg and central cells.
Unfertilized embryo sacs
As a control, we first fixed 5 unpollinated pistils. All the 70 observed embryo sacs from these pistils had two intact synergids, a very highly polarized egg cell with its nucleus toward the chalaza, a central cell with a single nucleus, and degenerated antipodal cells. Thus, they were 4-celled and unfertilized (stage FG7, according to Christensen et al., 1997). We then fixed and observed 35 hand-pollinated pistils, 3–24 HAP, in 6 independent experiments. All of the following observations on fixed material were obtained from these 634 embryo sacs. At 3 HAP, all of the 62 embryo sacs observed were at stage FG7 (Figure 1a, n = 271). We started to observe few embryo sacs with a high fluorescence level in one of the two synergids (Figure 1b; n = 18) at 5 HAP. The first samples at this stage, already described by Christensen et al. (1997) as FG8, were detected in the upper part of the pistil. We can interpret this increase in fluorescence as a landmark of synergid cell degeneration. Both right and left synergids were seen to have the ability to degenerate.
Egg cell with an altered polarity (F1)
In one embryo sac fixed at 5 HAP and in many other samples fixed at later times (n = 52), we could observe the remains of one degenerate synergid but also an important change in the egg cell polarity (Figure 1c). At this stage, the egg nucleus has moved away from the central cell, toward the micropyle. Many small vacuoles occupy the space between the nucleus and the plasma membrane next to the central cell, so that the egg-central cell boundaries are clearly visible. This change in polarity is maintained at later stages (see Figure 1d,e). In the majority of the embryo sacs we also noticed that the central cell nucleus was no longer spherical, but instead elongated along the micropyle-chalaza axis of the female gametophyte (Figure 1c). We termed this stage ‘F1’, for ‘Fertilization 1’ as this may represent the earliest stage described for fertilization. Again, the first samples that reached this stage were detected in the upper part of the fixed pistils.
Two nucleoli in the central cell nucleus only (F2)
One of the most striking stages first occurs 6 HAP, with the highest frequency at 7 HAP. The central cell nucleus is elongated and contains 2 nucleoli (Figure 1d). The two nucleoli are aligned along the micropyle-chalaza axis in 100% of the samples at this stage (n = 44). The nucleolus in a micropylar position is always smaller (2.0 ± 0.7 µm in diameter, mean ± sd; n = 39) than the one in a chalazal position (5.3 ± 0.4 µm; n = 39), this difference in size being highly significant (t-test: P < 10−33). Based on data from maize fertilization (Faure et al., 1994; Mòl et al., 1994) we interpret the presence of two nucleoli as a karyogamy step in the fertilized central cell, and we have named this stage ‘F2’ accordingly.
Two nucleoli in the egg and central cell nuclei (F3)
We observed several other samples with 2 nucleoli, not only in the central cell nucleus, but also in the egg cell nucleus (Figure 1e,f). We termed this new stage ‘F3’ as it was mainly observed at 8 HAP, i.e. after the ‘F2’ stage. In the central cell, nucleoli have the same position and relative size as at the F2 stage. In the egg cell, we could again observe a significant size difference (t-test: P < 10−7) between the two nucleoli, the smaller being 1.0 ± 0.6 µm (mean ± sd) in diameter and the larger being 2.8 ± 0.2 µm (n = 12). Both the smaller and bigger nucleoli in the egg are significantly smaller than their equivalents in the central cell (t-test: P < 10−3 and P < 10−27, respectively). We also noticed that the smaller egg nucleolus faced the chalazal part of the degenerated synergid (67%, n = 8) or the egg-central cell boundaries (33%, n = 4). This situation with two nucleoli in the egg and central cells probably corresponds to karyogamy in both cells. The delay in detecting two nucleoli in the egg versus the central cell suggests that karyogamy starts later in the forming zygote than in the endosperm. In 5 late samples (8–9 HAP) we could see two nucleoli in the egg only, and a single nucleus in the central cell or the first nuclear division of endosperm. This suggests again that karyogamy in the egg is delayed in comparison to that in the central cell, and can occur simultaneously with the first nuclear division in the endosperm. Finally, karyogamy was much less frequently seen in the egg than in the central cell (12 + 5 samples versus 12 + 44), suggesting that karyogamy in the egg persists for a shorter time than that in the central cell.
Early nuclear divisions in the endosperm
The fertilized central cell can undergo a first nuclear division (Figure 1g) as early as 7 HAP. (n = 2), that is 1 h or less after the first observed karyogamy in the central cell. All of the other divisions (n = 14) were seen at 8–9 HAP. In all samples observed, the division axis aligned with the micropyle-chalaza axis. We named this division stage ‘IIa’ in order to subdivide the chronology of endosperm development established by Boisnard-Lorig et al. (2001). The outcome of the division is a syncytium with two adjacent nuclei, one being very closely associated with the tip of the elongating zygote (Figure 1h,i). We named this stage ‘IIb’. Although some samples at stage IIb could be observed as early as 7 HAP, most of them were seen at 8–9 HAP. In some samples at this IIb stage, we could detect two nucleoli within each endosperm nucleus (Figure 1h, white arrows, n = 10). However, this situation with 2 nucleoli seems to be a transient stage as other samples with slightly more distant nuclei had only one nucleolus in each endosperm nucleus (Figure 1i; n = 11). A big vacuole then forms between the two endosperm nuclei while the zygote stays undivided (Figure 1j; n = 50). This stage that we named ‘IIc’ was observed as early as 8 HAP but most of its occurrences were seen at 12 HAP.
A second nuclear division (n = 5, 12–24 HAP, data not shown) leads to a 4-nucleate endosperm syncytium. Early after the division, sister nuclei are close to each other (n = 8) including cases with two nucleoli per nucleus (n = 2). The 4 endosperm nuclei then become evenly distributed along the micropyle-chalaza axis (n = 39, 12–24 HAP) and a third nuclear division leads to an 8-nucleate endosperm (n = 87, 24 HAP). The zygote is still undivided but is very elongated along the micropyle-chalaza axis, with its nucleus next to its chalazal tip.
Figure 1(k) summarises the fertilization and early developmental steps that we observed in A. thaliana, from 634 embryo sacs. The onset of stages FG8 and F1 was observed as early as 5 HAP. The onset of stages F2 and F3 was observed at 6 HAP, and the first IIa at 7 HAP. However, for each time point between 6 and 9 HAP, the stages observed varied between different samples. This variability seems to be due in part to a gradient of development along the pistil axis. During these early steps of fertilization, embryo sacs in the lower part of the pistil were, in general, at developmental stages observed 1-2 h earlier in the upper part. This variability decreased at later time points: at 12 HAP, pistils mostly contained seeds with a bi-nuclear endosperm. At 24 HAP, the vast majority of seeds had an 8-nucleate endosperm and an undivided zygote.
Pollen tube discharge into the synergid cell
We performed a second set of experiments to estimate when the pollen tube penetrates into the synergid cell and releases its contents compared to synergid degeneration and the subsequent F1-3 stages. We hand-pollinated pistils with pollen from a transgenic promLAT52::EGFP line expressing EGFP in the vegetative cell of the pollen grain and tube (see Cheung, 2001). Embryo sacs that have been penetrated by a pollen tube can easily be identified as shown at 8 HAP in Figure 2a. Pollen tubes are also often seen on the funiculus of the ovules and in the vicinity of the micropyle. Observations made at higher magnifications using CLSM (n = 32) show that EGFP can be detected in one lateral half of the embryo sac in the micropylar region, with a very characteristic hook shaped appearance (Figure 2b). The EGFP released by the pollen tubes is located in a volume that has the same shape as that of the degenerated synergid observed in fixed samples (Figure 2c). Some fluorescent material can be observed between the egg and central cells (white arrows on panels b and c), suggesting some material from the pollen tube to be introduced between the two female gametes. We repeated these observations on a total of 68 pistils in 6 independent experiments. We scored 2402 embryo sacs for the release of the pollen tube contents, 4–9.5 HAP, which allowed us to establish a time course of synergid penetration (Figure 3; black bars). We could detect the first fluorescent synergid at 4.5 HAP, in the upper part of the pistil. More synergids then became fluorescent, in a wave-like fashion from pistil top to bottom. We compared these results with those from fixed embryo sacs (Figure 3; white bars). For each of the 5, 6, 7, 8 and 9 HAP time points, the percentage of degenerated synergids is not significantly different from the percentage of embryo sacs with EGFP (t-tests: P > 0.1). Our data therefore show that the timing of synergid degeneration and pollen tube release correlate well. This suggests that either the synergid cell degenerates at the time of pollen tube discharge or very shortly before it.
After pollination, the first observable event of A. thaliana fertilization sensu lato is the degeneration of one of the two synergid cells, as previously reported (Christensen et al., 1997; Mansfield and Briarty, 1991; Mansfield et al., 1991). We provide here a timing for these observations: synergid degeneration could be seen as early as 5 HAP and most of the embryo sacs had undergone this step at 9 HAP. One key question asks whether synergid degeneration occurs long before, shortly before, or at the time of pollen tube contact and penetration. This question is important (i) to understand if synergid degeneration can play a role in pollen tube guidance or can only be a consequence of pollen tube penetration, and (ii) to position pollen tube penetration with the subsequent fertilization steps. In a number of species, synergids degenerate before pollen tubes reach embryo sacs (Jensen and Fisher, 1968b), or even before pollination (Van Went and Cresti, 1988). In some other species synergid degeneration seems not to precede the arrival of the pollen tube (Russell, 1992; Van Went, 1970). In A. thaliana, very few observations have been made. Christensen et al. (1997) observed 3 ovules of A. thaliana with pollen tubes reaching the embryo sac and touching one intact synergid. This could suggest that the synergid cell degenerates just prior to pollen tube penetration or as a consequence of it. However, these samples may also be rare cases with an abnormal development. In the present study we were able to analyse large populations of ovules. Synergid degeneration and the release of the pollen tube content with EGFP into embryo sacs correlate well in time (Figure 3), suggesting that the synergid cell degenerates at the time of pollen tube discharge, or only very shortly before it.
We then observed that the egg cell nucleus moves away from the egg-central cell boundaries while vacuoles become fragmented. A transient contraction of the fertilized egg cell has been shown in vitro within 15 min following sperm-egg plasma membrane fusion in Zea mays (Faure et al., 1994). The rapid change we noticed in A. thaliana may relate to this in vitro observation of egg activation after gamete fusion. This would make it a very early marker of plasmogamy in the egg. To the best of our knowledge, such a rapid change was never previously observed in flowering plants during in planta fertilization. Only changes in the egg cell structure prior to fertilization (Huang et al., 1999; Mòl et al., 2000), or changes in the zygote structure later in its development (see review, Russell, 1993) were already demonstrated.
Following the egg polarity change, we detected a small additional and transient nucleolus in both the central cell and egg cell nuclei. Previous data on fertilization in cotton and grasses have shown the presence of such a small nucleolus just after male and female nuclear envelopes merge (Batygina, 1974; Jensen and Fisher, 1968a; Mogensen, 1982; Mòl et al., 1994). It was interpreted as a late step of karyogamy. Thus our novel observations in A. thaliana are particularly useful as the supernumerary nucleolus will probably constitute a good marker of karyogamy in this plant species. In the future, such a marker will be important for isolating and identifying mutants affected in the fusion of gamete nuclei.
The super-numerary nucleolus has a very characteristic positioning. In the central cell nucleus, it always faces the egg cell. In the egg cell nucleus, it always faces the chalazal part of the degenerated synergid or the central cell. Interestingly, we showed that contents of pollen tubes expressing EGFP are released into a synergid, and diffuse into the intercellular space between the egg cell and central cell plasma membranes (Figure 2). This space may therefore constitute the site of gamete fusion, and the additional nucleolus may be formed by male chromatin. However, there is no experimental data to support this idea. In the future, it may be informative to look at the equivalent steps of autonomous endosperm development without fertilization in the fis mutants (Chaudhury and Berger, 2001; Grossniklaus et al., 2001). The presence or absence of such a supernumerary nucleolus would confirm or rebut the hypothesis of male origin.
We have established a precise timing for karyogamy. The first samples with two nucleoli in the central cell nucleus were seen at 6 HAP, only 1 h after the first synergid degeneration and therefore < 1 h after gamete fusion. This timing agrees with the only precise time courses established: based on the direct observations of naked embryo sacs in Torenia fournieri (Higashiyama et al., 1997), or made using in vitro fusions of sperm with egg or central cells in Zea mays (Faure et al., 1994; Kranz and Lörz, 1993; Kranz et al., 1998). In A. thaliana, a first nuclear division very rapidly follows karyogamy in the central cell. Interestingly, we could again detect transiently two nucleoli in each of the resulting nuclei. Similar observations were made shortly after the second nuclear division in a few samples. The observations of stages with two nucleoli suggest the intriguing possibility that male and female chromatin stay separate throughout karyogamy and early endosperm development. Such a spatial separation of the parental chromatin complements during the first zygote mitosis was shown in Drosophila melanogaster (Callaini and Riparbelli, 1996; Huettner, 1924). In addition, from our data, there is a 20-fold difference in volume between the two nucleoli during karyogamy in both the fertilized central cell nucleus and the zygote nucleus. In Torenia fournieri, a 2-fold difference in volume was shown between the two nucleoli of the zygote nucleus (Higashiyama et al., 1997). This difference in volume may suggest (see review: Carmo Fonseca et al., 2000) that male chromatin has a lower transcriptional activity than female chromatin, at least during the F2 and F3 stages. A broad silencing of the paternal genome during early development as suggested in A. thaliana (Vielle-Calzada et al., 2000) may be related to these structural observations. Investigations are therefore now required to observe chromatin dynamics during fertilization in living material. In the future, histones fused to variants of GFP (Boisnard-Lorig et al., 2001) and expressed under yet-to-be-isolated sperm cell and embryo sac promoters, should allow such observations.
Karyogamy seems to be delayed in the egg cell compared to the central cell. Based on the frequency of the F2 and F3 stages, we also think that karyogamy lasts for a shorter time in the egg than in the central cell. However, we may be overestimating this difference because super-numerary nucleoli are smaller in the egg than in the central cell and therefore may be sometimes missed during cytological examinations. There is indeed an 8-fold difference in the volumes of the smaller nucleolus between these two cell types. Interestingly, our data indicate that there is a similar difference (7-fold) for the bigger nucleolus. This may indicate that there is a lower transcriptional activity in the forming zygote than in the fertilized central cell (see Carmo Fonseca et al., 2000). It may also indicate that there is a general difference in chromatin condensation between the two fertilization products. Again, GFP markers for chromatin are now required for the further study of nuclear dynamics (Boisnard-Lorig et al., 2001). Differences in transcriptional activity and chromatin condensation may account for the differences in the timing of the first mitosis. The first division in the endosperm is indeed initiated much earlier than in the zygote.
The rapid experimental procedures we used together with the precise time course and key steps we identified should constitute a basis of comparison for the future analysis of fertilization mutants. Further investigations will be required to understand in particular (i) if a direct contact between the pollen tube and the embryo sac leads to synergid degeneration; (ii) the mechanisms responsible for a change in polarity in the fertilized egg cell and the possible consequences of it; (iii) the chromatin dynamics and the mechanisms of male nucleus integration; and (iv) how development is initiated differentially in the fertilized egg and central cells.
Plant material and growth conditions
We used seeds of the Arabidopsis thaliana Columbia ecotype obtained from the Nottingham Arabidopsis Stock Centre (Nottingham, UK). The promLAT52::EGFP line (in the Columbia background) was obtained from A. Cheung (University of Massachusetts, Amherst, MA, USA). Seeds were sown in BP 2 substrate with 6% clay (Klasmann-Deilmann GmbH, Geeste-Gross Hesepe, Germany) and vernalized for 3 d at 4°C in the dark. Plants were then grown in a chamber under short days of 10 h, 19°C, 70% of relative humidity, and nights of 14 h, 16°C, 80% of relative humidity. Light was provided at 200 µE m−2 s−1, by fluorescent lamps (Cool daylight 6500, and Warm white; Philips, Ivry-sur-Seine, France). Plants were watered three times a week and were grown under short days for 1.5 months. Flowering was induced under long days conditions (about 14 h of light at 21°C) in a greenhouse (Ecole Normale Supérieure, Lyon, France).
Flowers at the developmental stage 12c (Smyth et al., 1990) of the wild type plants Columbia were emasculated 40–44 h before hand pollination. Pollinations were performed manually in the greenhouse, between 9:00 and 11:00 a.m. We used an excess of pollen from two anthers from two different flowers at anthesis. Control pistils were left unpollinated. Plants were then left in the greenhouse up until pistil collection.
Observations from fixed material
Pistils pollinated with wild type Columbia pollen as well as control unpollinated pistils were collected at precise times, from 0 up until 24 h After Pollination (HAP). All pistils were stuck onto double-sided tape (Scotch; 3M, St Paul, MN, USA) and dissected under a dissecting microscope (SMZ 645; Nikon, Tokyo, Japan) equipped with a cold light source (KL200; Schott, Mainz, Germany), as described by Christensen et al. (1997). Pistils were then fixed, dehydrated and cleared (see Christensen et al., 1997). Cleared pistils were mounted in immersion oil (518 N; Carl Zeiss, Jena, Germany) under a glass coverslip (No. 0; Chance Propper; West Midlands, UK) sealed with fingernail polish. Observations of cleared samples were all made using a LSM 510 Confocal LASER Scanning system on an Axiovert 135 microscope, equipped with a × 63 objective lens (Plan Apochromat, oil immersion; Carl Zeiss). Excitation light at 543 nm was provided by a helium-neon LASER. Fluorescence light was selected using a HFT 543 dichroic beamsplitter and a band pass BP 560–615 emission filter. Under these conditions, nucleoli are brightly autofluorescent. The surrounding structures and cytoplasm can also be visualized due to some weaker autofluorescence. Typically, optical sections of 1.4–1.8 µm were obtained with a pinhole diameter of 200–250 µm. Images were acquired and processed using Zeiss LSM 510 version 2.0 software.
Pistils crossed with pollen from the promLAT52::EGFP line were collected at precise times, from 3 up to 10 HA. Each pistil was stuck onto double-sided tape and micro-dissected in order to remove the two valves and collect the septum with lines of ovules still attached to it. This structure was mounted in a solution of 10% sucrose, 0.01% boric acid, and 3 mm Ca(NO3)2, 4H2O, under a glass coverslip sealed with fingernail polish. Synergid cells penetrated by pollen tubes were then scored under a dissecting microscope (MZ12, Leica, Wetzlar, Germany) equipped for GFP detection and with a × 1.6 zoom (Plan Apo; Leica). Further imaging of the structures was done on a Nikon microscope (Eclipse E600) equipped for GFP detection and with a CCD camera (monochrome Spot RT; Diagnostic Instruments, Burroughs, MI, USA). We used × 20 and × 40 Plan Fluor DIC M objective lenses (Nikon). Images were acquired and processed using the Advanced 3.2.4 software for Windows, from Diagnostic Instruments. We also used Confocal LASER Scanning Microscope described above. Observations were made with a × 40 objective lens (Plan Apochromat, oil immersion; Carl Zeiss). Excitation light at 488 nm was provided by a argon-krypton LASER. Fluorescence light was selected using a HFT 488 dichroic beamsplitter and a band pass BP 505–550 emission filter. Optical sections of 2–4 µm were obtained. Images were acquired and processed using Carl Zeiss LSM 510 version 2.0 software.
The authors are very grateful to A. Cheung (University of Massachusetts, Amherst, MA, USA) for the generous gift of seeds from the promLAT52::EGFP line. We thank F. Berger for stimulating discussions. We also thank him, A. Chaboud, A. Cheung and C. Scutt for comments on the manuscript, A. Lacroix for taking care of the plants, and F. Rozier for technical help. J.-E.F. was supported by the French Centre National de la Recherche Scientifique (CNRS).