Expressing the diphtheria toxin A subunit from the HAP2(GCS1) promoter blocks sperm maturation and produces single sperm-like cells capable of fertilization.

After meiosis, the male germline of flowering plants undergoes two mitoses, producing two sperm that are carried within a pollen tube to an ovule. One sperm fuses with the egg to form the zygote and the other fuses with the central cell to form the primary endosperm. The mechanisms that control male germline development and gene expression, and ensure that sperm properly fuse with female gametes are just beginning to be understood. Expression of the potent translation inhibitor, diphtheria toxin A subunit, from the Arabidopsis (Arabidopsis thaliana) HAP2(GCS1) promoter blocked sperm development before the final cell division, resulting in pollen tubes that carried a single sperm-like cell rather than two sperm. These pollen tubes targeted ovules and fertilized either the egg or the central cell, producing seeds with either endosperm or an embryo, but not both. Endosperm-only seeds significantly outnumbered embryo-only seeds, suggesting that single sperm-like cells preferentially fuse with the central cell. These experiments show that de novo translation is required for completion of sperm development, that the HAP2(GCS1) promoter is very tightly controlled, and that disruption of gene expression can result in male germ cells with a bias for gamete fusion.

Flowering plants have a unique reproductive system called double fertilization. Two nonmotile sperm cells develop within a pollen grain and are delivered to female gametes by a pollen tube, a highly polar extension of the pollen grain that elongates by tip growth. The pollen tube is attracted to an ovule where it bursts and releases sperm. One sperm fertilizes the egg to produce the zygote, thereby initiating development of the embryo. The other sperm fuses with the central cell to produce the primary endosperm, which develops into endosperm, a tissue that provides nutritional support for the embryo as it develops into a seedling .
Male germline development occurs in the anther where microspore mother cells undergo meiosis to produce a tetrad of haploid microspores. Each microspore undergoes an asymmetric mitotic division during which a vegetative cell engulfs a generative cell. These cells have very different fates ; the vegetative cell does not divide again, while the generative cell undergoes DNA replication and divides once more to produce two haploid sperm cells within the vegetative cell cytoplasm. The vegetative cell and the two sperm cells each carry the same haploid genome and constitute the three-celled male gametophyte (pollen grain). One question raised by this developmental program and gamete delivery system is whether the two sperm cells are identical and interchangeable, or whether the two sperm have different fates and are programmed to fuse with one or the other female gamete.
In Plumbago zeylanica, pollen tubes carry morphologically distinct sperm that have distinct targets for fertilization (Russell and Cass, 1981;Russell, 1984). One sperm is larger than the other, contains many mitochondria but few or no plastids, is associated with the vegetative nucleus, and is more likely to fuse with the central cell. The smaller sperm contains many plastids but fewer mitochondria and is more likely to fuse with the egg (Russell, 1985). In maize (Zea mays), the egg is fertilized two to three times more frequently by sperm carrying B chromosomes than by sperm carrying only the A complement of chromosomes (Roman, 1948;Carlson, 1969;Faure et al., 2003). Proposed explanations for preferential fusion in these systems have included the order of sperm arrival at the female gametophyte, the morphology/size of the sperm, and differential expression of cell surface factors that mediate specific cellular gamete:gamete interactions leading one sperm to specifically target either the egg or the central cell for fusion (Roman, 1948;Russell, 1985;Spielman and Scott, 2008).
Flowering plant sperm, like mammalian sperm, have compact chromatin, a unique set of histones (Xu et al., 1999a;Okada et al., 2005), and were assumed to be transcriptionally quiescent. However, several genes have been described that are male germline specific (Xu et al., 1999a(Xu et al., , 1999bEngel et al., 2003Engel et al., , 2005Singh et al., 2003;Durbarry et al., 2005;Okada et al., 2005;Rotman et al., 2005) and microarray analysis shows that Arabidopsis (Arabidopsis thaliana) sperm possess a large, diverse, and cell-type-specific array of transcripts (Borges et al., 2008). While no genes have yet been identified that are differentially expressed between Arabidopsis sperm, the morphologically distinct P. zeylanica sperm express different mRNAs (Gou et al., 2009). These findings raise the possibility that flowering plant sperm may express a specific gene expression program, leading to differentiation and gamete fusion specificity.
To disrupt gene expression during male germline development, we expressed the translation blocking diphtheria toxin A (DTA) chain from the male germlinespecific HAP2(GCS1) (for HAPLESS2, GENERATIVE CELL SPECIFIC1; At4g11720) promoter (von Besser et al., 2006). We anticipated that this approach would result in sperm ablation and/or disruption of sperm gene expression. DTA blocks protein synthesis by ADP ribosylation of translation elongation factor 2 and has been expressed from various promoters in studies of animal and plant development to determine the consequences of loss of particular cells/tissues within the context of otherwise normal development (Breitman et al., 1987;Thorsness et al., 1993;Day et al., 1995;Twell, 1995;Singh et al., 2003). A single molecule of DTA is sufficient to completely block propagation of mouse cells in vitro (Yamaizumi et al., 1978). Therefore, DTA can reveal very weak promoter activity and has been used to refine understanding of expression patterns of sporophytic (Thorsness et al., 1991) and gametophytic (Thorsness et al., 1991;Twell, 1995;Singh et al., 2003) genes in flowering plants. Furthermore, when DTA is expressed without the diphtheria toxin B chain, it cannot be transported across the plasma membrane, and is therefore cell autonomous (Pappenheimer, 1977). This is a critical feature for our purpose because the sperm cells are housed within the vegetative cell cytoplasm and we wanted to perturb sperm gene expression without interfering directly with gene expression in the vegetative cell.
We show that male germline development is blocked just before the generative cell divides in transgenic plants expressing a HAP2(GCS1)promoter: DTA [HAP2(GCS1):DTA] construct. This result is consistent with recent data showing that DUO POLLEN1 (DUO1) and DUO3 are required to activate HAP2 (GCS1) transcription in the generative cell (Brownfield et al., 2009a(Brownfield et al., , 2009b. Furthermore, it demonstrates that de novo translation in the generative cell is necessary for completion of sperm development. HAP2(GCS1): DTA pollen tubes are functional and deliver single sperm-like cells (SSLCs) to female gametophytes, resulting in ovules that contain either an embryo or an endosperm, but not both products of double fertilization. We show that these are the products of fertilization and that endosperm-only ovules significantly outnumber embryo-only ovules. We conclude that expression of DTA in the male germline perturbs the normal sperm gene expression program and results in a population of SSLCs that preferentially fuse with the central cell. Recent analysis of Arabidopsis mutants that generate SSLCs have led to models suggesting either that the egg is the default target for sperm (Nowack et al., 2006), or that fusion is random (Chen et al., 2008). Our data indicate that the egg is not the default target for fertilization, and that blocking translation during germline development results in a SSLC that preferentially fertilizes the central cell.

Expression of HAP2(GCS1):DTA Generates Mature Pollen Grains Containing a SSLC
We transformed Arabidopsis quartet1 (qrt1) mutants, which release pollen grains as intact meiotic tetrads but do not otherwise affect pollen function (Preuss et al., 1994), with a HAP2(GCS1):DTA construct and recovered transgenic plants that had no obvious vegetative or floral developmental defects. However, all transgenic lines had a striking pollen defect; many mature pollen grains contained a vegetative nucleus and a single additional nucleus rather than the two sperm nuclei present in wild-type pollen grains (Fig.  1A). Subsequent analyses showed these single cells with bright-staining nuclei shared features with sperm cells (see below), so we called them SSLCs.
We determined the segregation of this defect in tetrads of three transgenic lines (Fig. 1B); for HAP2 (GCS1):DTA B7, the majority (92/106; Fig. 1, A and B) of tetrads comprised two normal pollen grains and two that contained a SSLC. This 2:2 segregation pattern in tetrads indicated that the plant was hemizygous for the transgene, the transgene caused a postmeiotic defect, and expressivity of the defect was complete in these tetrads. A fraction of tetrads (14/106; Fig. 1B) comprised three normal pollen grains and one containing a SSLC. The 3:1 (3 normal:1 defective) segregation observed in these tetrads indicated that the defect was not completely expressed in all pollen grains that carried HAP2(GCS1):DTA. The expressivity of pollen developmental defects was lower in two other lines analyzed [HAP2(GCS1):DTA B1 and B4; Fig. 1B].
In reciprocal crosses to wild type, HAP2(GCS1):DTA behaved like a mutation that disrupts the development and/or function of the male gametophyte. We used pollen from three independent hemizygous HAP2(GCS1):DTA transgenic lines to pollinate male  (Table I). T-DNAs that do not disrupt male gametophyte function are inherited by 50% of progeny in such crosses (Johnson et al., 2004;von Besser et al., 2006). This severe reduction in transmission through the male gametophyte indicated that the transgene disrupted pollen development and/or function. Data from two other transgenic lines were consistent with a reduction in transmission through the male gametophyte, however the expressivity of the defect was lower in these lines (Table I).
Transmission through the female gametophyte was also reduced in the lines with the greatest reduction in transmission through the male gametophyte (Table I). This result suggests that the HAP2(GCS1) promoter is active in female cells, consistent with a recent report that HAP2(GCS1) mRNA is present in ovule and silique samples (Borges et al., 2008). HAP2(GCS1) expression in sperm/minus gametes has been shown to be essential for fertilization in Arabidopsis (Mori et al., 2006;von Besser et al., 2006), Plasmodium (Hirai et al., 2008;Liu et al., 2008), and Chlamydomonas ; however, expression from female/ plus gametes is not required in these organisms, so the role of female-expressed HAP2(GCS1) is currently not known. Furthermore, DUO1, a critical regulator of HAP2(GCS1) expression (Brownfield et al., 2009a), is restricted to the male germline (Rotman et al., 2005), suggesting a distinct control mechanism for expression in female cells.

DNA Content of SSLCs
To determine the stage at which DTA expression disrupted male germline development, we measured the fluorescence intensity of 4#,6-diamidino-2-phenylindole (DAPI)-stained nuclei of SSLCs and normal sperm in HAP2(GCS1):DTA B7 and HAP2(GCS1):DTA B1 pollen. In both lines, the fluorescence intensity of Figure 1. Expression of HAP2(GCS1):DTA generates mature pollen grains containing a SSLC. A, DAPIstained pollen grains of qrt1 show a faint-staining vegetative nucleus and two sperm nuclei (arrowheads) in each of the four grains. Two of the four grains from a hemizygous HAP2(GCS1):DTA B7 plant have two sperm, the other two grains show a SSLC (arrows). B, Schematics represent the segregation of sperm developmental defects in pollen tetrads (vegetative nuclei not drawn). qrt1 tetrads are observed in which one of the pollen grains appears to contain a single sperm nucleus; this is an artifact of imaging overlapping sperm nuclei. C, Ratio of DAPI fluorescence intensity of SSLCs/sperm. Each triangle or rectangle represents the average relative fluorescence intensity of two SSLCs/four sperm nuclei in a single tetrad; black bars represent the mean. Table I. HAP2(GCS1):DTA displays distorted segregation Self-fertilized, Progeny of self-fertilization of primary transformants. Male: ms1 3 HAP2(GCS1):DTA, ms1 females were hand pollinated using HAP2(GCS1):DTA anthers. Female: HAP2(GCS1):DTA 3 qrt1, HAP2(GCS1):DTA females were emasculated and hand pollinated using qrt1 anthers. %Basta R = the percentage of Basta R F 1 progeny from the indicated cross. SSLCs was roughly twice that of normal sperm (Fig.  1C). We concluded that the generative cell, which normally divides to produce two sperm cells, completed S phase of the cell cycle but failed to complete mitosis in pollen grains that expressed HAP2(GCS1):DTA.

Expression of Sperm Identity Markers in SSLCs
To explore the identity of SSLCs and to assess the impact of the HAP2(GCS1):DTA construct on sperm gene expression, we analyzed four male germline markers in HAP2 (GCS1) Brownfield et al., 2009a).We analyzed hemizygous HAP2(GCS1):DTA plants that were hemizygous for DUO1:H2B:mRFP1. In this experiment, a maximum of approximately 50% of pollen with either normal sperm or a SSLC carried the reporter construct. We found that approximately 50% of pollen of either type expressed DUO1:H2B:mRFP1, indicating that almost all SSLCs that carried this reporter expressed it (Fig. 2, A and E). Consequently, we suggest that the DUO1:H2B:mRFP1 reporter construct is translated in SSLCs before DTA begins to block translation. Similarly, nearly all SSLCs expressed HTR10:HTR10:mRFP1 (Fig. 2, B and E). Microarray analysis of sperm showed that HTR10 transcripts were among the most abundant and accumulated to significantly higher levels than HAP2(GCS1) or AtGEX2 (Borges et al., 2008). We propose that the HTR10 promoter is sufficiently active before the onset of translation inhibition by DTA to allow the reporter to accumulate to detectable levels.
Analysis of hemizygous HAP2(GCS1):DTA pollen that was homozygous for HAP2(GCS1):HAP2(GCS1): YFP or AtGEX2:eGFP (100% of pollen carry the reporter) showed that many SSLCs expressed these markers, but a significant fraction did not (Fig. 2, C-E). The HAP2(GCS1) and AtGEX2 promoters are expected to be activated contemporaneously with the HAP2(GCS1):DTA construct. We propose that in some single-sperm like cells, detectable levels of HAP2 (GCS1):HAP2(GCS1):YFP or AtGEX2:eGFP were translated, whereas in other SSLCs translation of these markers was blocked by DTA. These data suggest that the effect of DTA on sperm gene expression is variable and that some SSLCs express the endogenous HAP2 (GCS1) and AtGEX2 genes, while others do not. HAP2 (GCS1) is essential for fertilization (Mori et al., 2006;von Besser et al., 2006;Hirai et al., 2008;Liu et al., 2008) and it, along with any other sperm-expressed genes required for gamete fusion, would have to be expressed for SSLCs to fertilize female gametes.  We germinated pollen from HAP2(GCS1):DTA B7 and HAP2(GCS1):DTA B1 hemizygous plants in vitro and stained them with DAPI to identify pollen tubes that carried SSLCs and those with two normal sperm (Fig. 3A). In both cases, the number of growing pollen tubes that contain a SSLC was roughly equivalent to the number containing two sperm (Fig. 3B). This indicated that pollen grains with a SSLC germinated at the same rate as wild-type pollen grains. Furthermore, after 6 h of pollen tube growth, the length of HAP2(GCS1):DTA B7 and HAP2(GCS1):DTA B4 pollen tubes with a SSLC was the same as that of pollen tubes with two sperm (Fig. 3C). Expression of DTA from the HAP2(GCS1) promoter does not affect pollen tube growth, suggesting that the effects of DTA are limited to the generative cell contained within the vegetative cell cytoplasm.
We previously proposed that sperm-expressed genes play a role in pollen tube guidance because hap2-1 mutant pollen tubes were less likely than their wild-type counterparts to target ovules in competitive hand pollinations of ms1 pistils (von Besser et al., 2006). Recently, duo3 pollen tubes, which are defective in expression of HAP2(GCS1) and other spermexpressed genes, were also shown to be less likely to target ovules than wild-type pollen tubes (Brownfield et al., 2009b). We could not determine whether HAP2 (GCS1):DTA expression caused pollen tube guidance defects in competitive crosses because the incomplete expressivity of DTA defects precluded quantitative analysis of ovule targeting.

SSLCs Fertilize the Central Cell More Often Than the Egg
We analyzed embryo and endosperm development in pistils of self-fertilized hemizygous HAP2(GCS1): DTA B7, HAP2(GCS1):DTA B1, and HAP2(GCS1):DTA B4 transgenic plants approximately 5 d after pollination. In each independent transgenic line, we observed three classes of developing seeds: (1) normal seeds containing a late globular-stage embryo and proliferating endosperm, (2) abnormal seeds in which endosperm proliferated, but there was no embryo, (3) abnormal seeds in which embryos initiated development, but endosperm was completely absent (Supplemental Fig. S1). We also observed a significant number of unfertilized ovules. We hypothesized that normal seeds were the result of fertilization by wild-type pollen produced by HAP2(GCS1):DTA hemizygous plants and that abnormal seeds were the result of fertilization of either the egg or the central cell by a pollen tube carrying a SSLC. Neither endosperm-only nor embryo-only seeds completed development; they resulted in aborted ovules that were obvious in developing siliques of self-fertilized HAP2(GCS1):DTA hemizygous plants (Supplemental Fig. S1).
In each of these transgenic lines, endosperm-only ovules significantly outnumbered embryo-only ovules, suggesting that SSLCs preferentially fertilize the central cell. We performed a quantitative analysis of seed development by pollinating ms1 pistils with pollen from homozygous HAP2(GCS1):DTA B7 transgenic plants to determine the relative frequency of embryo-and endosperm-only ovules and to analyze the development of these aberrant seeds over time. This rare homozygous line was likely the progeny of a pollen grain that carried HAP2(GCS1):DTA, but was not defective in generative cell division or expression of essential sperm genes (Figs. 1B and 2). As expected, seed production was significantly reduced in this line; however, it produced a small number of homozygous individuals following self-fertilization. We used this homozygous line for quantitative analysis of seed development because approximately 80% of its pollen carried SSLCs (Supplemental Fig. S2, A and B) and pollen tubes targeted approximately 90% of ms1 ovules following hand pollination (Supplemental Fig.  S2C). These features increased the number of single fertilization events that could be analyzed.
We analyzed ovule development 1.5, 3, and 4 d after pollination of ms1 (Fig. 4) and found that endospermonly ovules outnumbered embryo-only ovules at each time point when homozygous HAP2(GCS1):DTA B7 pollen was used ( Fig. 4; Table II). In total, we observed 73 endosperm-only ovules and eight embryo-only ovules, a ratio of 9:1. These classes of aberrant ovule development were not observed when qrt1 pollen was used to pollinate ms1 or when no pollen was applied to ms1 (Table II). Significantly, a central cell nucleus is clearly apparent in embryo-only ovules (Fig. 4, C, F, and I); this observation is consistent with the hypothesis that embryo-only ovules are the result of fertilization of the egg by a SSLC. The central cell remains unfertilized and does not proliferate autonomously (Fig. 4, C, F, and I). This observation distinguished SSLCs produced by expression of HAP2(GCS1):DTA from those of cell division cycle2 (cdc2) or f-box-like17 ( fbl17) mutants (Nowack et al., 2006;Kim et al., 2008;Gusti et al., 2009). In these cases, egg-only fertilization by a SSLC initiated division of the primary endosperm nucleus.
We analyzed 899 ovules across the three time points in pistils pollinated with homozygous HAP2(GCS1): DTA B7 pollen. In addition to the 81 single fertilization events observed, there were 32 products of double fertilization that contained an embryo and endosperm (Table II). These ovules were likely targeted by HAP2 (GCS1):DTA B7 pollen tubes that carried two sperm, indicating that DTA was either not expressed in these pollen, or that DTA expression did not affect generative cell division or expression of essential sperm genes. The majority of ovules were unfertilized (Table  II; S2C) and is likely due to lack of expression of HAP2(GCS1) (Fig. 2, C and E) and/or other essential sperm genes required for fertilization. It is likely that sperm gene expression is also disrupted in pollen grains that contain two sperm, because we observed two sperm in approximately 20% of pollen grains, but double fertilization was observed in only approximately 4% of ovules. The observed rates of double fertilization, egg-only fertilization, and central cellonly fertilization increased during the time course we analyzed (Table II). This suggests that fertilization events mediated by sperm and SSLCs carrying HAP2 (GCS1):DTA were delayed relative to wild type.

Initiation of Endosperm or Embryo Development by a SSLC Requires Fertilization
We considered the possibility that deposition of a SSLC in the female gametophyte initiated endosperm or embryo development through an autonomous pathway not requiring gamete fusion. Proliferation of endosperm without fertilization occurs in female gametophyte mutants with loss of function of either FERTILIZATION-INDEPENDENT SEED2, FERTILI-ZATION-INDEPENDENT ENDOSPERM, or MEDEA genes that normally repress endosperm development until fertilization (Grossniklaus et al., 1998;Luo et al., 1999;Ohad et al., 1999). To address the possibility that SSLCs promoted autonomous endosperm proliferation, we examined whether a SSLC could transmit KS117 endosperm marker (Sorensen et al., 2001) expression to a wild-type female. Embryo development and endosperm proliferation were apparent in ovules fertilized by wild-type or KS117 pollen (Fig. 5, A and C); GFP signal was only detectable in ovules fertilized by KS117 pollen (Fig. 5, D compared to B). We generated plants hemizygous for HAP2(GCS1):DTA B7 and KS117 and used these to pollinate ms1. Importantly, in these crosses, we observed endosperm-only ovules ( Fig. 5E) and products of double fertilization (Fig. 5G) that express the KS117 marker (Fig. 5, F and H), indicating that endosperm initiation resulted from fertilization by a SSLC and not autonomous endosperm proliferation.
To determine whether embryos developing in embryo-only ovules were the product of fertilization by a SSLC we used an RCN1promoter:GUS (RCN1: GUS) transgene as a paternal marker for embryo gene expression. RCN1 (for ROOTS CURL IN NPA1) is a ubiquitously expressed protein phosphatase 2A regulatory subunit (Deruere et al., 1999;Zhou et al., 2004;Blakeslee et al., 2008). Four days after pollination of ms1 pistils with pollen carrying RCN1:GUS, we detected GUS activity in globular embryos (Fig. 5J); GUS activity was not detected in ovules pollinated with qrt1 (Fig. 5I). When we pollinated ms1 with pollen hemizygous for both HAP2(GCS1):DTA B7 and RCN1:GUS we observed GUS expression in embryos resulting from double fertilization (Fig. 5K) and in embryo-only ovules (Fig. 5L), indicating that embryos developing without endosperm were the product of fertilization by a SSLC carrying the RCN1:GUS reporter gene.

HAP2(GCS1):DTA Transgenic Plants Offer a New Tool for the Dissection of Male Germline Development and Fertilization
We generated pollen grains that carry a SSLC, instead of two sperm, by expressing DTA from the HAP2 (GCS1) promoter in transgenic plants. SSLCs were capable of fertilization and produced significantly more endosperm-only ovules than embryo-only ovules, suggesting that SSLCs were biased toward fusing with the central cell rather than the egg. HAP2 (GCS1):DTA transgenic plants can be used to identify proteins that are translated immediately before or after the final cell division in male germline development, Complete ablation of the generative cell, a possible outcome of our experiment given that DTA leads to cell death in other systems, would have produced pollen grains (and/or tubes) with a vegetative nucleus but no generative or sperm cell(s). We did not observe this phenotype, which would have been detectable even at low frequencies. It is possible that the brief period of time between onset of HAP2(GCS1):DTA expression and release of sperm is insufficient for induction of cell death. Alternatively, the relatively mild and variable phenotypes associated with HAP2 (GCS1):DTA expression may be due to low and/or sporadic expression of the DTA transgene.
HAP2(GCS1):DTA expression did not affect pollen grain development (Fig. 1) or the ability of pollen tubes to grow in vitro (Fig. 3). In contrast, expression of DTA from either the lily generative cell1 promoter (Singh et al., 2003), or the vegetative nucleus-specific LAT52 promoter (Twell, 1995), resulted in pollen grain collapse. Since the pollen grain was not affected in HAP2 (GCS1):DTA plants, we conclude that the HAP2(GCS1) promoter is restricted to the generative cell and not active in the vegetative nucleus.  Loss of function of either DUO1 or DUO3 blocks HAP2 (GCS1) expression in the generative and sperm cells (Brownfield et al., 2009a(Brownfield et al., , 2009b. DUO1 expression is limited to the male germline (Rotman et al., 2005), while DUO3 is expressed in the male germline and broadly in sporophytic tissues (Brownfield et al., 2009b). The developmental defects caused by expression of HAP2(GCS1):DTA were limited to the male germline, consistent with the hypothesis that HAP2 (GCS1) promoter activity is restricted by the expression pattern of DUO1 (Brownfield et al., 2009a).
In addition to regulating HAP2(GCS1) expression, DUO1 and DUO3 have overlapping and distinct roles in controlling expression of other male germ-line expressed genes and in promoting generative cell mitosis. duo1 mutants do not express AtGEX2, HTR10, or the G2/M regulator AtCycB1;1 (for CYCLIN B1; Brownfield et al., 2009a). duo3 mutants are also defective in AtGEX2 expression, but express HTR10 and AtCycB1;1 (Brownfield et al., 2009b). The generative cells of both mutants complete S phase, but fail to enter M phase of the cell cycle (Brownfield et al., 2009a(Brownfield et al., , 2009b. In the case of duo1 mutants, failure to complete mitosis is likely due to lack of expression of AtCycB1;1. We propose that HAP2(GCS1):DTA expression in the generative cell blocks translation of AtCycB1;1 and/or other positive regulators of the generative cell mitosis, resulting in mature pollen grains with a SSLC with approximately twice the DNA content of wild-type sperm.

Regulation of Sperm Fusion Events during Double Fertilization
All Arabidopsis sperm-expressed genes studied thus far are expressed in both sperm (Xu et al., 1999a(Xu et al., , 1999bEngel et al., 2005;Rotman et al., 2005;Mori et al., 2006;von Besser et al., 2006;Ingouff et al., 2007), suggesting these cells do not differentiate from each other. Furthermore, when two wild-type sperm are delivered to a mutant female gametophyte that produces two eggs rather than one, both eggs can be fertilized, suggesting that both sperm have the capacity to fuse with an egg (Pagnussat et al., 2007;Ingouff et al., 2009). These data suggest that Arabidopsis sperm are essentially interchangeable and that either is capable of fertilizing the egg.
On the other hand, experiments analyzing the outcomes of single fertilization events mediated by pollen tubes carrying a SSLC suggest the potential for preferential fertilization. Loss of function of either of two critical cell cycle regulators, CDC2A (CDKA;1) or FBL17, blocks the generative cell cycle during S phase and produces a SSLC that preferentially fertilizes the egg (Iwakawa et al., 2006;Nowack et al., 2006;Kim et al., 2008;Gusti et al., 2009). Three explanations for egg-only fertilization were proposed: (1) The architecture of the female gametophyte dictates fertilization of the egg because sperm are deposited in one of two synergids that are adjacent to the egg cell.
(2) There is active signaling by the egg cell that attracts the first sperm cell released by the pollen tube. (3) Each of the two sperm cells has a specific, predetermined target for fertilization (Nowack et al., 2006).
In another study of single fertilization, the fusion target for the SSLC appeared to be random (Chen et al., 2008). Approximately 6% of pollen grains generated by Arabidopsis msi1/MSI1 (for MULTICOPY SUP-PRESSOR OF IRA1) mutants contain a SSLC due to pleiotropic defects in male germline development caused by loss of function of a complex that regulates chromatin assembly (Chen et al., 2008). Interestingly, an equal number of egg-only (18/3,600 ovules scored) and central cell-only (20/3,600 ovules scored) single fertilization events were observed.
SSLCs produced by msi1 or by expression of HAP2 (GCS1):DTA can bypass the egg and fertilize the central cell. Therefore, female gametophyte architecture or active signaling by the egg to attract the first sperm deposited can be ruled out as explanations for eggonly fertilization by cdc2a or fbl17 mutants. Analysis of SSLCs produced in msi1 mutants, which were equally likely to fuse with either female gamete, led to the conclusion that the two sperm produced by wild-type pollen are identical (Chen et al., 2008). Our results indicate that expression of a translation inhibitor during sperm development results in a SSLC that preferentially fuses with the central cell.
We propose that SSLCs produced by cdc2a, fbl17, msi1, or by expression of HAP2(GCS1):DTA show different preferences for fertilization because each of these perturbations has a different effect on male germline gene expression and development. cdc2a and fbl17 specifically block the generative cell cycle at S phase and produce SSLCs that fuse with the egg, but are not expected to directly affect gene expression in the male germline (Iwakawa et al., 2006;Nowack et al., 2006;Kim et al., 2008;Gusti et al., 2009). msi1 is pleiotropic, sometimes causing arrest immediately after meiosis or after the first pollen mitosis; therefore, it is possible that the population of SSLCs generated by this mutation is heterogeneous, hence the finding that they fertilize either the egg or the central cell (Chen et al., 2008). Expression of HAP2(GCS1):DTA blocks generative cell mitosis, but might also directly block translation of mRNAs involved in regulating fusion of the sperm with either the egg or central cell. Therefore, HAP2(GCS1):DTA SSLCs may fuse predominantly with central cells because they fail to translate a factor expressed by cdc2a and fbl17 single sperm cells that directs or enables sperm to fuse with the egg.
The mechanisms that ensure that one sperm fertilizes the egg and the other fertilizes the central cell are not known, but it is clear that such a mechanism is in place to prevent both sperm from fusing with either the egg or the central cell. Differentiation between the two sperm resulting in specific gamete fusion provides one potential mechanism. An alternative hypothesis is that fusion is random and a mechanism similar to the rapid block to polyspermy in animals (Wong and Wessel, 2006) prevents both sperm from fertilizing either of the female gametes Spielman and Scott, 2008). Our studies on pollen carrying a SSLC suggest that male germline development can be altered to produce a sperm cell with a specific fate and suggest the possibility of an underlying program of differential gene expression that could ensure that one sperm fuses with the egg and the other with the central cell. Support for this hypothesis will require identification of genes that differentially mark the two sperm and direct a specific fertilization preference. Comparison of SSLCs produced by cdc2a or fbl17 mutants (egg preference) and by expression of HAP2(GCS1):DTA (central cell preference), provides a path toward discovery of these sperm fusion specificity factors.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S3. Unfertilized ovules observed in quantitative analysis of early seed development.