Synergid Cell Death in Arabidopsis is Triggered Following Direct Interaction With the Pollen Tube

During angiosperm reproduction, one of the two synergid cells within the female gametophyte undergoes cell death prior to fertilization. The pollen tube enters the female gametophyte by growing into the synergid cell that undergoes cell death and releases its two sperm cells within the degenerating synergid cytoplasm to effect double fertilization. In Arabidopsis and many other species, synergid cell death is dependent upon pollination. However, the mechanism by which the pollen tube causes synergid cell death is not understood. As a first step toward understanding this mechanism, we defined the temporal relationship between pollen tube arrival at the female gametophyte and synergid cell death in Arabidopsis. Using confocal laser scanning microscopy, light microscopy, transmission electron microscopy, and real-time observation of these two events in vitro , we demonstrate that synergid cell death initiates after the pollen tube arrives at the female gametophyte but before pollen tube discharge. Our results support a model in which a signaling cascade triggered by pollen tube-synergid cell contact induces synergid cell death in Arabidopsis.


INTRODUCTION
During the angiosperm fertilization process, a pollen tube grows into one of the female gametophyte's two synergid cells. The synergid cell that the pollen tube grows into undergoes cell death, either before or upon entry of the pollen tube into this cell. The pollen tube then ceases growth and releases its contents, including the two sperm cells, into the degenerating synergid cytoplasm.
Ultimately, the two sperm cells migrate to and fuse with the egg cell and central cell to effect double fertilization (Lord and Russell, 2002;Weterings and Russell, 2004).
Synergid cell death may facilitate several steps of the angiosperm fertilization process. First, synergid cell death may be required for pollen tube entry into the female gametophyte (Russell, 1993;Higashiyama, 2002). Second, the degenerated state may be required for tube growth cessation and release of pollen tube contents (van Went and Willemse, 1984). Third, synergid degeneration is accompanied by a cytoskeletal reorganization that is thought to facilitate migration of the sperm cells to the egg and central cells (Russell, 1993(Russell, , 1996Fu et al., 2000). Finally, it is likely that breakdown of the synergid membrane is required to provide the sperm cells direct access to the fertilization targets (Russell, 1993).
Little is known about the molecular processes that regulate and mediate synergid cell death. In many species, synergid cell death is pollinationdependent (van Went and Willemse, 1984;Willemse and van Went, 1984;Russell, 1992), indicating that it is not a part of the megagametogenesis developmental program and suggesting that pollen tubes induce a physiological cell death program (i.e., programmed cell death) (An and You, 2004). Pollen tubes could induce synergid programmed cell death through either a diffusible or a contact-mediated (i.e., pollen tube-synergid cell contact) signal. Alternatively, pollen tubes may cause cell death through physical disruption of the synergid cell. For example, pollen tube penetration and/or discharge may increase the physical pressure within the synergid cell and result in rupture of this cell (van 5 Went and Willemse, 1984;Willemse and van Went, 1984;Russell, 1992;Higashiyama, 2002).
Several of these models can be distinguished by precisely defining the temporal relationships among synergid cell death, pollen tube arrival at the female gametophyte, and pollen tube discharge. For example, cell death before pollen tube-synergid cell contact would suggest that the pollen tube induces synergid programmed cell death through a long-range, diffusible signal.
Alternatively, cell death upon pollen tube-synergid cell contact would suggest that a contact-mediated signal induces programmed cell death in the synergid cell.
Finally, cell death after pollen tube discharge would suggest that synergid cell death occurs through physical rupture caused by pollen tube penetration and/or discharge.
The relationship between pollen tube arrival and synergid cell death has been determined in many species and these observations are variable. Synergid cell death occurs before arrival of the pollen tube at the female gametophyte in some species (Jensen and Fisher, 1968;Cass and Jensen, 1970;Maze and Lin, 1975;Mogensen, 1978;Mogensen and Suthar, 1979;Wilms, 1981;Dute et al., 1989;Kuroiwa, 1989;Russell et al., 1990;Yan et al., 1991;Huang et al., 1993;Huang and Russell, 1994) and after arrival in other species (van der Pluijm, 1964;Schulz and Jensen, 1968;van Went, 1970;Newcomb, 1973). Similarly, observations of the cell death process are variable. For example, synergid degeneration appears to occur via programmed cell death in wheat (An and You, 2004) but by physical disruption due to pollen tube discharge in Torenia (Higashiyama et al., 2000). These variable observations might result from differences among species in the induction of synergid cell death. Alternatively, the variation may be due to technical differences; for example, the detection of synergid degeneration has been shown to be fixation dependent (Fisher and Jensen, 1969 (Faure et al., 2002); however, these experiments did not distinguish between cell death occurrence shortly before or soon after pollen tube arrival. Thus, although these data appear to eliminate cell death induction through a very long-range diffusible signal (e.g., from stigma), they do not distinguish among the other possibilities including induction by a short-range diffusible signal, induction following pollen tube-synergid cell contact, or physical disruption following pollen tube penetration and/or discharge.
Recently, two Arabidopsis female gametophyte mutants, gfa2 (Christensen et al., 2002) and sirene (srn) (Rotman et al., 2003), affected in synergid cell death have been reported. gfa2 and srn synergid cells appear normal and attract wild-type pollen tubes, but fail to undergo cell death following pollination. With gfa2, the precise spatial relationship between the pollen tube and the synergid cell has not been determined. With srn, wild-type pollen tubes enter the female gametophyte and grow between the egg and synergid cells, but fail to enter the synergid cell, cease growth, and release their contents (Rotman et al., 2003). Although these observations indicate that pollen tube-synergid cell contact per se is not sufficient to cause cell death, they do not eliminate the possibility that cell death is caused by physical rupture of the synergid cell following pollen tube discharge. Furthermore, these observations do not eliminate a model in which the pollen tube induces a physiological cell death program since it is possible that the srn mutation affects a signal transduction molecule.
In this paper, we defined the temporal relationship between pollen tube arrival at the female gametophyte and synergid cell death in Arabidopsis using light microscopy and transmission electron microscopy of fixed material, as well as real-time imaging of these two events in an in vitro pollen tube growth assay.
We showed that the synergid cell initiates cell death after the pollen tube arrives at the female gametophyte but before pollen tube discharge. suggest that the pollen tube triggers cell death by directly interacting with the synergid cell and inducing a physiological cell death program.

Time Course of Pollen Tube Arrival at the Ovule in Arabidopsis
As a first step toward determining the temporal relationship between pollen tube growth and synergid cell death in Arabidopsis, we established the timing of pollen tube arrival at each ovule within the pistil. We performed controlled pollinations (described in Materials and Methods), collected pistils at 2 to 10 hours after pollination, stained the pollen tubes with congo red, and used confocal laser scanning microscopy (CLSM) to score the number of ovules containing a pollen tube in its microyple (Palanivelu et al., 2003). To determine whether pollen tubes arrive at the ovules closest to the stigma significantly before those farthest from the stigma, we divided the pistils into three sections, top (the five ovule rows closest to the stigma; rows 1-5), middle (the next five ovule rows; rows 6-10), and bottom (the five ovule rows closest to the pedicle; rows 11-15), and scored the ovules according to position within the pistil. Figure 1A summarizes the analysis of 1,594 ovules and shows that pollen tubes reach the top ovule rows by four to six hours after pollination, the middle ovule rows by six to ten hours after pollination, and the bottom ovule rows at over eight hours after pollination.
These data indicate that pollen tubes reach ovules in a temporal series, first at the top ovule rows and later at the bottom ovule rows, and that the ovules closest to the stigma receive pollen tubes approximately six hours before those closest to the pedicle. To determine whether synergid cell death occurs significantly before or at about the same time as pollen tube arrival at the ovule in Arabidopsis, we performed controlled pollinations, waited 2 to 10 hours, and scored synergid cell death based on its morphology using CLSM (Christensen et al., 1997). Female gametophytes were placed into one of three categories: female gametophytes with two intact synergid cells (Figure 2A), female gametophytes with a slightly degenerate synergid cell (Figures 2B), and female gametophytes with a highly degenerate synergid cell ( Figure 2C). Slightly degenerate synergid cells had slightly higher autofluorescence in the cytoplasm, an irregular nucleus, and an intact vacuole ( Figure 2B), and highly degenerate synergid cells were extremely autofluorescent and had neither an observable nucleus nor a vacuole ( Figure   2C).
Because pollen tubes reach the ovules along the pistil at different times ( Figure 1A), we limited our analysis to those at the top of the pistil. Figure 1B summarizes the analysis of 629 ovules and shows that synergid cell death was first detected in the top ovule rows at four hours after pollination. At 4-8 hours after pollination, the percentage of female gametophytes undergoing synergid cell death ( Figure 1B) was lower than the percentage of ovules with a pollen tube in its micropyle ( Figure 1A). These results suggest that synergid cell death occurs after entry of the pollen tube into the ovule micropyle.
To validate these observations, we scored pollen tube position and synergid cell death within the same ovules. We performed controlled pollinations, collected pistils at four to six hours after pollination, fixed and cleared the tissue, and used DIC microscopy to observe these two events. We scored six pistils and 39 ovules. Of these, 30 ovules had two intact synergid cells ( Figures 3A and 3B) and nine ovules had a degenerating synergid cell ( Figure 3D). All of the ovules containing a degenerating synergid cell also had a pollen tube in the micropyle. Most significantly, six of the ovules containing two intact synergid cells had a pollen tube in the micropyle ( Figures 3A to 3C).
These observations suggest that that synergid cell death occurs after the pollen tube enters the ovule micropyle in Arabidopsis.

Arabidopsis
After entering the ovule micropyle, the pollen tube penetrates through the integuments and then enters the female gametophyte. Although the methods used above revealed that synergid cell death occurs after the pollen tube enters the micropyle, they however did not allow us to define the position of the pollen tube within an ovule at the time of synergid cell death. To more closely define this spatial relationship, we used two microscopic methods that allowed us to score pollen tube position and synergid cell death within the same ovule. We performed controlled pollinations, waited four to eight hours, fixed ovules attached to the placenta (from rows 1-5), embedded the ovules in Spurr's resin, and analyzed plastic sections using light microscopy ( Figure 4) and transmission electron microscopy ( Figure 5). Using these methods, we scored 35 ovules. Of these, 19 ovules had a degenerating synergid cell ( Figures 4B and 5B); all of these also had a pollen tube in the degenerating synergid cell ( Figure 5B). In addition, 16 ovules had two intact synergid cells (Figures 4A and 5A); in five of these, a pollen tube could be observed within the female gametophyte (Figures 4C and 4D) or contacting the filiform apparatus ( Figure 5C). These results suggest that the synergid cell death in Arabidopsis is not initiated until after the pollen tube arrives at the synergid cell.

Observations of Pollen Tube Growth and Synergid Cell Death In Vitro
To confirm the temporal relationship between synergid cell death and pollen tube arrival at the female gametophyte, we performed live-imaging of pollen tube growth and synergid cell death using an Arabidopsis in vitro assay. We previously showed that this system reflects much of in vivo pollen tube behavior indicating that these ovules survived our assay and image capture conditions and remained with intact synergids at the end of the experiment. Loss of GFP in just one synergid cell, as with pollen-induced cell death described below, was never observed.
We next monitored synergid degeneration as a consequence of interaction with DsRed-tagged pollen tubes. We followed 59 ovules that had intact synergids over the course of the experiment. Of these, 31 ovules had a pollen tube within 100 µm of the micropyle and 28 ovules had no pollen tube near the micropyle. Of the 31 ovules with a nearby pollen tube, 23 were penetrated by a pollen tube and eight were not. Of the 23 ovules penetrated by a pollen tube, 18 exhibited synergid degeneration, as evidenced by loss of GFP or change in shape in one of the two synergid cells ( Figure 6, Figure 7 and Videos S1-4 degeneration is a specific event that initiates upon interaction with a pollen tube, which reflects this aspect of in vivo pollen tube-synergid interaction. To determine whether synergid degeneration occurs before or after pollen tube arrival, we scored the temporal relationship between these two events. In all 18 ovules that underwent synergid degeneration, we were able to score both pollen tube arrival and synergid degeneration. With each ovule, we scored (i) the time point at which the dsRed-tagged tube tip first overlapped with the GFPtagged synergid cell, representing the time point of pollen tube arrival (e.g., Video S1, 120 minutes frame) and (ii)  In 13 of these 18 ovules, we were able to score both pollen tube discharge and synergid degeneration. For these measurements, we used the time point at which the dsRed signal spread out explosively from the pollen tube tip to represent the time point of pollen tube discharge (e.g., Video S2, 350 minutes frame) and synergid degeneration was scored as described above. In all 13 cases, pollen tube discharge occurred after synergid degeneration. On average, pollen tube discharge occurred 102 ± 69 (s.d.) minutes after synergid degeneration was first detected. Together, these data suggest that the interaction between a pollen tube and a synergid cell occurs in the following order: (1) pollen tube arrival at the female gametophyte, (2) pollen tube growth around the synergid cell, (3) synergid degeneration, and (4) pollen tube discharge.

Tube Discharge in Arabidopsis
In Arabidopsis, synergid cell death does not occur in the absence of pollination (Christensen et al., 1997). A dependence of synergid cell death on pollination has also been reported for many other species (van Went and Willemse, 1984; Willemse and van Went, 1984;Russell, 1992). These observations suggest that pollen tube growth within a pistil is responsible for initiation of the synergid cell death process in these species. However, the lack of information about the position of the pollen tube at the time of synergid cell death has limited our understanding of the mechanism by which the pollen tube triggers synergid cell death in Arabidopsis.
In the present study, we used a variety of microscopic methods to clarify the temporal relationship between synergid cell death and pollen tube growth in Arabidopsis. Using CLSM (Figures 1 and 2), light microscopy ( Figures 3 and 4) TEM ( Figure 5), and real-time observation of pollen tubes growing in vitro ( Figures 6 and 7), we show that synergid cell death occurs after arrival of the pollen tube at the female gametophyte. Synergid cell death after arrival of the pollen tube at the female gametophyte has also been reported in other species (Russell, 1992) including Torenia fournieri (Higashiyama et al., 2000), Capsella bursa-pastoris (Schulz and Jensen, 1968), sunflower (Newcomb, 1973), and petunia (van Went, 1970).
Using an in vitro assay, we were able to make real-time observations of pollen tube-synergid cell interaction within the same ovule by labeling pollen tubes with dsRed and synergid cells with GFP. In this assay, the pollen tube reaches the synergid cell before synergid degeneration, consistent with our in vivo observations using light microscopy ( Figures 4C and 4D) and TEM ( Figure   5C). The pollen tube then continues to grow and extend around the synergid cell for ~174 minutes before synergid disintegration is first detected. Although not commented on previously, prior analysis of pollen tube growth in the ovule has shown continued pollen tube growth near a synergid cell after arriving at the female gametophyte and prior to discharge (e.g., movie 5, supplemental data, Rotman et al., 2003). Finally, at ~100 minutes after synergid degeneration is first detected, the pollen tube discharges its contents. In a Torenia in vitro pollen tube guidance assay, it was found that the pollen tube discharges at about the same time as synergid rupture (Higashiyama et al., 2000).
These observations support the proposal that synergid cell death is required for cessation of pollen tube growth and pollen tube discharge (van Went and Willemse, 1984;Higashiyama, 2002

Early Steps of Synergid Cell Death in Arabidopsis
Using CLSM, we observed many (>100) ovules that contained one synergid cell exhibiting slightly higher autofluorescence in the cytoplasm, an irregular nucleus, and an intact vacuole ( Figure 2B). The proportion of the synergid cells exhibiting this morphology was highest at early time points and progressively lower at later time points (Figure 1B), suggesting that these synergid cells are at an early stage of cell death. All other degenerating synergid cells observed resembled those in Figure 2C, suggesting that the cell death process progresses rapidly to a highly degenerated state. These observations suggest that the early steps of synergid cell death include breakdown of the nucleus and a biochemical change in the cytoplasm that results in elevated autofluorescence by CLSM.
We also analyzed degenerating synergid cells at these early time points using TEM. Consistent with the CLSM analysis, the cytoplasm exhibited elevated electron density ( Figure 5B). However, in contrast to the CLSM analysis, nuclei were not observed and the vacuoles were fragmented (data not shown). These differences are likely due to the harsher fixation methods During the cell death process, we observed cytoplasmic material in the narrow space between the chalazal end of the egg cell and the micropylar end of the central cell. This was consistently observed in our CLSM ( Figure 2C), light microscopy ( Figure 4B), and TEM (data not shown) images. Cytoplasmic material between the egg and central cells has been reported previously in several other species (van Went and Cresti, 1988;Huang et al., 1993;Huang and Russell, 1994;Higashiyama et al., 2000) and is thought to be the contents of the pollen tube (Higashiyama, 2002). It is likely, therefore, that the sperm cells fuse with the fertilization targets in this region (Russell, 1992).
At approximately 10-20 minutes after pollen tube discharge, we consistently observed movement of cytoplasmic contents (i.e., concentration of the GFP signal) to the micropylar end of the degenerating synergid cell in our in vitro assay (e.g., see Video S2, between time points 350 and 360 minutes). It is likely that this movement represents collapse of the synergid cell during the final stages of cell death, as has been observed in TEM analysis of embryo sacs in several other species (Jensen and Fisher, 1968;Cass and Jensen, 1970;Mogensen, 1972;Dute et al., 1989;Huang and Russell, 1992).

Arabidopsis
Our observations clarify the relationship between pollen tube guidance and the ovule micropyle, they, however, left open the possibility that synergid degeneration might be required for the final steps of pollen tube guidance, i.e., pollen tube growth from the micropyle to the female gametophyte (Jensen, 1974;Weterings and Russell, 2004). In the present study, we show that synergid degeneration occurs after the pollen tube arrives at the female gametophyte.
These observations firmly establish that synergid cell death is not a prerequisite for any aspect of pollen tube guidance in Arabidopsis.

Arabidopsis
In many other species examined, synergid cell death occurs before arrival of the pollen tube at the female gametophyte (Jensen and Fisher, 1968;Cass and Jensen, 1970;Maze and Lin, 1975;Mogensen, 1978;Mogensen and Suthar, 1979;Wilms, 1981;Dute et al., 1989;Kuroiwa, 1989;Russell et al., 1990;Yan et al., 1991;Huang et al., 1993;Huang and Russell, 1994). These observations suggest that synergid cell death is induced by a long-range diffusible signal in these species. However, our observations that the synergid cell directly interacts with the pollen tube before it degenerates suggest that cell death in Arabidopsis is not induced by a long-range diffusible signal.
Initiation of synergid cell death after arrival of the pollen tube raises several possible means by which degeneration is caused in Arabidopsis. First, the pollen tube may induce a physiological cell death program by a contactmediated (i.e., pollen tube-synergid cell contact) signal. Alternatively, pollen tube penetration and/or discharge may trigger mechanical breakdown of the synergid cell (van Went and Willemse, 1984;Willemse and van Went, 1984;Russell, 1992 result from mechanical breakdown in Arabidopsis. First, we observed many ovules containing two intact synergids and in which the pollen tube is in contact with the synergid cell ( Figures 4C, 4D, and 5C). Second, in our in vitro assay, synergid degeneration does not initiate until ~174 minutes after the pollen tube arrives at the female gametophyte and during this time, the pollen tube continues to extend, grow around, and interact with the synergid cell. Third, in our in vitro assay, we did not observe any instance where pollen tube discharge occurred before initiation of synergid degeneration. Instead, synergid degeneration initiated ~100 minutes before pollen tube discharge. Finally, in srn mutants, even though the synergid cells come in contact with wild-type pollen tubes for an extended period of time, synergid degeneration fails to occur (Rotman et al., 2003). Together, these observations suggest very strongly that mechanical impact resulting from pollen tube penetration and/or pollen tube discharge does not cause synergid cell death in Arabidopsis.
Observations in several other species support the conclusion that mechanical breakdown is not the cause of synergid cell death. In both cotton (Jensen and Fisher, 1968) and barley (Cass and Jensen, 1970), instances of the pollen tube entering the persistent synergid have been reported. In Plumbago zeylanica, which does not have synergid cells, the pollen tube contacts the egg cell for several minutes and discharges its contents adjacent to the egg cell, but these events do not cause egg cell death (Russell, 1982(Russell, , 1983. These observations suggest that pollen tube penetration and pollen tube discharge per se are not sufficient to cause the synergid to undergo cell death. The data presented here suggest that in Arabidopsis, the pollen tube and the synergid cell interact in the following series of steps: (1)  synergid cell could respond to factors released by the pollen tube, as with cell death in response to fungal pathogens (Ellis et al., 2006). We recently have identified a number of female gametophyte mutants that attract pollen tubes but that fail to undergo synergid cell death (L. S.-N, M. F. Portereiko, and G. N. D., unpublished). Analysis of these mutants should lead to molecular dissection of the pathway by which the pollen tube induces synergid cell death.

Plant Material and Growth Conditions
We Plants used for in vitro assays were germinated directly on the soil.

Controlled Pollinations
The experiments discussed in this paper required pollen tube growth to female gametophytes of the same developmental stage. However, female gametophyte development within a pistil is not perfectly synchronous (Christensen et al., 1997

Confocal Microscopy Analysis of Pollen Tube Growth
Congo red staining was performed as previously described (Palanivelu et al., 2003). We emasculated flowers at stage12c (Christensen et al., 1997), waited 24 hours, and pollinated with wild type pollen. Flowers were collected at 2 hours after pollination, 4 hours after pollination, 6 hours after pollination, 8 hours after pollination, and 10 hours after pollination. The sepals, petals and stamens were removed from the isolated flowers and pistils were placed on double-sided tape (Scotch). Cuts were made on both sides of the pistil replum using a 30.5-gauge syringe to expose the ovules to the Congo red stain. The pistils then were transferred to a microscope slide with a drop of 0.4% of Congo red solution for staining. The pistil was divided into three sections for ovule scoring. The top section included the first five rows of ovules (rows 1 to 5), the middle section included the next five rows (rows 6 to 10), and the bottom section included the last five rows (rows 11 to 15). We performed three independent experiments per time point, and in each experiment, we scored the ovules within 10 pistils (five plants and two pistils per plant). Ovules were analyzed on a Zeiss LSM 510 confocal microscope. The Congo red dye was excited with a HeNe laser at a wavelength of 543 nm. Emission was detected between 585 nm and 650 nm.

Confocal Microscopy Analysis of Synergid Cell Degeneration
Pistils were fixed and mounted for Confocal Laser Scanning Microscopy as previously described (Christensen et al., 1997) with the modification that we used a Zeiss LSM 510 microscope. We emasculated flowers at stage 12c the plant; removed the sepals, petals and stamens from the isolated flowers; and placed the pistils on double-sided tape (Scotch) on a microscope slide. We then made cuts on both sides of the pistil replum using a 30.5-gauge syringe to expose the ovules to fixative. The cuts spanned the entire length of the pistil.
The pistils were fixed in a solution of 4% Glutaraldehyde and 12.5 mM cacodylate buffer (pH 6.9) for 4 hrs at room temperature, dehydrated in a graded ethanol series (10%, 20%, 40%, 60%, 80%, 95%, for 10 min each), and incubated in 100% ethanol overnight at room temperature, and then cleared in a 2:1 mixture of benzyl benzoate:benzyl alcohol for 20 minutes at room temperature. The cleared pistils were mounted in a drop of immersion oil onto microscope slides. We performed three independent experiments per time point, and in each experiment, we scored the ovules within 10 pistils (five plants and two pistils per plant). Observations were made using a Zeiss LSM 510 confocal microscope. Ovules were excited with a HeNe laser at a wavelength of 543 nm.

Differential Interference Contrast Analysis of Ovules
We emasculated flowers at stage 12c (Christensen et al., 1997), waited 24 to 40 hours, and pollinated with wild-type pollen. Pistils were harvested 4 and 6 hours after pollination. The tissues were processed as described above for the CLSM analysis. The tissue was then mounted on microscope slides in a drop of immersion oil. The ovules were observed with a Zeiss Axioplan microscope using DIC optics. Images were obtained with an Axiocam MRm digital camera (Carl Zeiss, Jena, Germany) and processed using Adobe Photoshop 7 and Illustrator 10.

In vitro Pollen Tube Targeting Assay and Fluorescence Microscopy
In vitro pollen tube targeting assays were performed essentially as described previously (Palanivelu and Preuss, 2006