The egg cell is preferentially fertilized in Arabidopsis double fertilization^FA

Abstract

In flowering plants (angiosperms), fertilization of the egg cell by one sperm cell produces an embryo, whereas fusion of a second sperm cell with the central cell generates the endosperm. In most angiosperms like Arabidopsis, a pollen grain contains two isomorphic sperm cells required for this double fertilization process. A long-standing unsolved question is whether the two fertilization events have any preference. A tool to address this question is the usage of the cyclin-dependent kinase a1 (cdka;1) mutant pollen, which produces a single sperm-like cell (SLC). Here, we first adopt a complementation-based fluorescence-labeling method to successfully separate and collect cdka;1 mutant pollen containing a single SLC. Single-cell RNA-sequencing analysis revealed that cdka;1 SLCs show a gene expression profile highly similar to that of sperm cells and not to the generative cell, precursor of the two sperm cells. Pollination assays using a limited number of cdka;1 mutant pollen revealed that in 98.2% of the ovules, single fertilization of the egg cell occurred. Pollination of pistils with excessive cdka;1 mutant pollen allowed the delivery of a second SLC via fertilization recovery, which fertilized the central cell, resulting in 20.7% double-fertilized ovules. This indicates that cdka;1 SLCs are able to fertilize both the egg and the central cell. Taken together, our findings have answered a long-standing question and support that preferential fertilization of the egg cell is evident in Arabidopsis.

INTRODUCTION

In angiosperms, sexual reproduction results in double fertilization, in which two sperm cells fuse with the egg cell and the central cell to form an embryo and an endosperm, respectively (Dresselhaus et al., 2016; Sprunck, 2020). In Arabidopsis thaliana, a pollen mother cell undergoes meiosis to form a tetrad of microspores, each of which undergoes two rounds of mitotic division to produce a tri-cellular pollen (Berger and Twell, 2011; Hafidh and Honys, 2021). In Pollen Mitosis I (PMI), the haploid microspore divides asymmetrically to generate a larger vegetative cell and a smaller generative cell. In Pollen Mitosis II (PMII), the generative cell divides symmetrically to produce two sperm cells. During female gametogenesis, the haploid functional megaspore undergoes three consecutive nuclear divisions to ultimately produce a female gametophyte containing two female gametes (egg cell and central cell), two synergid cells and three antipodal cells (Yadegari and Drews, 2004; Yang et al., 2010). For double fertilization, two immobile sperm cells are transported by each pollen tube and released into the ovule containing the female gametophyte (Zhang et al., 2017; Zhong and Qu, 2019).

Double fertilization is an innovation in angiosperms, in which fertilization of the egg cell is highly conserved in all plants, whereas fertilization of the central cell is evolutionarily younger and specific to angiosperms. Therefore, one of the outstanding questions is whether there is preference between the two sperm cells and the female gametes, which includes: (i) whether each sperm cell is predetermined to target a preferential female partner; and (ii) whether there is priority between the egg cell and central cell fertilization events. As for the first aspect, except for Plumbago zeylanica in which the two dimorphic sperm cells have a predetermined female gamete as their fertilization target (Russell, 1985), most angiosperms like Arabidopsis generate two isomorphic sperm cells, which seem to function equivalently (Hamamura et al., 2011a). The eostre and rbr1 mutants, for example, produce two egg cells and both can be targeted by one of the two sperm cells, so that two embryos are formed (Pagnussat et al., 2007; Ingouff et al., 2009). Live-cell imaging using photo-convertible reporters demonstrated that the two sperm cells in Arabidopsis are each able to fuse with either of the two female gametes (Hamamura et al., 2011b).

With regard to the second aspect of fertilization priority between egg cell and central cell, live-cell imaging revealed no preferences for the order of the two fertilization events (Hamamura et al., 2011b). However, it was later reported that fusion of the egg cell usually occurs earlier than sperm cell-central cell fusion (Denninger et al., 2014; Hamamura et al., 2014). These partly contrasting findings did not solve the issue of preferential fertilization, largely due to the limits of imaging resolution in discerning two almost simultaneous events, which could be best clarified by studying fusion of the two female gametes after releasing only a single male gamete in each ovule. Fortunately, such mutants that generate a single sperm-like cell (SLC) in one pollen grain were identified already, but reports using these mutants were equally confusing. For example, it was reported that SLCs of multicopy suppressor of ira1 (msi1) and cyclin-dependent kinase a1 (cdka;1) caused single fertilizations with either the egg or the central cell at an equal frequency (Chen et al., 2008; Aw et al., 2010). However, it was also reported that the egg cell was predominantly targeted for single fertilization both in the cdka;1/+ (Iwakawa et al., 2006; Nowack et al., 2006) and in the fbl17 mutant (Kim et al., 2008). Even more confusing was the usage of translation inhibition in sperm cells leading to the production of a single SLC that preferentially fertilized the central cell (Frank and Johnson, 2009).

Before clear statements can be made, we believe a few key points need to be well addressed. First, sperm cell functionality, that is, capacity to fertilize, should be validated. For example, pollen of the cdka;1 but not of the duo pollen 1 (duo1) mutant expressed sperm cell marker genes and were able to fertilize (Rotman et al., 2005; Iwakawa et al., 2006; Nowack et al., 2006; Brownfield et al., 2009). Therefore, duo1 and similar mutants cannot be used to test preferential fertilization. Second, caution should be taken in genetic analyses when mutants can only be maintained as a heterozygote, because if mutant gametophytes are mixed with wild-type ones, it will be difficult to clarify the direct correlation between the parental genotypes and fertilization outcomes. For example, heterozygous msi1/+ plants produce 6% SLC pollen, resulting in less than 1% single fertilization events (Chen et al., 2008). Therefore, it is critical to explicitly identify mutant pollen from those of the wild type to obtain feasible statistics. Third, gamete fusion failure-induced fertilization recovery leads to secondary pollen tube attraction and thus the release of additional sperm cells that may cause miscalculation of the fertilization events and misinterpretations (Beale et al., 2012; Kasahara et al., 2012; Zhong et al., 2022).

In this study, we have addressed all these key points. First, we employed a complementation-based fluorescence-labeling method to successfully identify and separate cdka;1 mutant pollen grains. Second, we adopted single-cell RNA-sequencing analysis to verify the identity of the cdka;1 SLCs. Third, we conducted pollination assays with either limited or excessive cdka;1 mutant pollen grains to detect single or double fertilization events. Our results demonstrate that preferential fertilization of the egg cell is evident in Arabidopsis.

RESULTS AND DISCUSSION

Separation, collection, and single-cell RNA-seq analysis of cdka;1 pollen

For many essential genes, homozygous loss-of-function mutants cannot be obtained. To solve this problem, two new strategies are recently available. One is to directly obtain and study the homozygous albeit not heritable mutant plants via clustered regularly interspaced palindromic repeats (CRISPR)/CRISPR-associated nuclease 9 (Cas9) technology. This strategy has been successfully applied, for example, for the sperm cell-specific HAP2 gene essential for gamete fusion (Zhong et al., 2022). In the other strategy, gametophytes carrying the mutated gene are distinguished by fluorescence labels. This approach has been employed to successfully separate and collect mutant pollen of vacuolar protein sorting-associated protein 41 (VPS41) and VPS18, two essential proteins involved in endomembrane trafficking (Hao et al., 2016; Zhong et al., 2020; Hou et al., 2021).

As CDKA;1 regulates cell division of gametophytic and vegetative cells (Iwakawa et al., 2006; Nowack et al., 2006, 2012) and no homozygotes can be obtained via CRISPR/Cas9 technology, we thus employed the complementation-based fluorescence-labeling method to obtain cdka;1 pollen from the heterozygotes. The rationale of this method is that homozygous mutants are complemented by the heterozygous expression of a gene-of-interest-GFP (green fluorescent protein) fusion protein, to grow normally until meiosis, after which the mutated gametophytes and complemented gametophytes develop separately, so that the mutated ones can be specified by the absence of GFP expression. To realize this goal, we transformed a construct containing a wild-type copy of CDKA;1 coupled with a pollen-specific fluorescence cassette pCDKA;1:CDKA;1-pLAT52:GFP into the cdka;1+/− heterozygous mutant (Figure 1A). In the complementary line of cdka;1−/− carrying pCDKA;1:CDKA;1-pLAT52:GFP+/−, apparently mutant pollen grains showing GFP expression were fully rescued, whereas the ones without GFP expression (GFP−) were cdka;1 homozygous (Figure 1B). The complementary line was further transformed with the sperm nucleus marker pHTR10:HTR10-RFP. To populate a collection of cdka;1 pollen, we manually picked out the ones with GFP− and red fluorescent protein (RFP)+ signals, as described in previous reports (Hao et al., 2016; Zhong et al., 2020). In contrast to the wild-type pollen grain 98.8 ± 0.3% of which contains two sperm cells (n = 335), 99.7 ± 0.5% of these cdka;1 pollen grains contained a single SLC (n = 314), as indicated by the sperm cell reporter as well as 4′,6-diamidino-2-phenylindole (DAPI) staining (Figure 1C, D, F). Similarly, only one SLC was observed in 100% semi-in vivo grown cdka;1 pollen tubes (n = 196), compared with two sperm cells in the wild-type pollen tubes (n = 192) (Figure 1E, F). Furthermore, Alexander staining assays showed that cdka;1 pollen grains were comparably viable as those from wild type (Figure S1).

Figure 1. Isolation of cyclin-dependent kinase a1 (cdka;1) pollen grains containing a sperm-like cell (SLC) highly similar to sperm cells.

(A) An overview of the complementation-based fluorescence-labeling method to obtain cdka;1 pollen. (B) Complemented pollen grains show green fluorescent protein (GFP) signal from the complementation construct while cdka;1 pollen grains lack GFP signals. Sperm cells and SLCs are labeled by pHTR10:HTR10-RFP. Scale bars, 50 μm. (C) Manually collected cdka;1 pollen grain lacks GFP signal and contains an SLC as indicated by the red fluorescent protein (RFP) signal. Scale bars, 50 μm. (D) 4′,6-diamidino-2-phenylindole (DAPI) staining shows that cdka;1 pollen grain contains only one SLC nucleus in contrast to two sperm cell nuclei in the wild type ones. Arrows indicate sperm cells and SLC, respectively. Scale bars, 5 μm. (E) Sperm (like) cell nuclei in wild-type and cdka;1 pollen tube. Scale bars, 5 μm. (F) Quantification of sperm (like) cell numbers in pollen grains (PG) and pollen tubes (PT) of wild type or cdka;1. Data are mean values ± SD. (G) Principal component analysis (PCA) of single-cell RNA-seq data shows that cdka;1 SLCs cluster with sperm cells (SCs) and not with generative cells (GCs). (H) Expression level of representative genes related to male germline development in cdka;1 SLCs, SC, and GCs. SLC, sperm-like cell; TPM, transcripts per million mapped reads. See Table S1 for more details.

To further clarify sperm cell identification of cdka;1 SLCs at the molecular level, we adopted single-cell RNA-sequencing technology (Tang et al., 2009; Ryu et al., 2021; Mo and Jiao, 2022). We manually separated and collected 10 individual cdka;1 SLCs and 17 wild-type sperm cells that were released from semi-in vivo grown pollen tubes. In parallel, we also isolated 10 generative cells from wild-type pollen. All individual cells were subjected to Smart-seq2 analysis (Picelli et al., 2014). There were 3,808, 3,708, and 8,704 genes detected (transcripts per million mapped reads > 0.1) for cdka;1 SLCs, wild-type sperm cells and generative cells, respectively. This indicated that the single-cell manipulation/RNA-seq analysis method applied had high and comparable sensitivity as previously reported fluorescence-activated cell sorting-based single-cell sequencing method (Misra et al., 2019) or the bulk RNA-seq method (Borges et al., 2008). Unsupervised hierarchal clustering and principal component analysis revealed that cdka;1 SLCs are closely clustered with sperm cells (Figures 1G, S2). In particular, genes with essential sperm cell functions such as GEX2, HAP2/GCS1 and DMP9 showed comparable expression levels in cdka;1 SLCs and in the wild-type sperm cells, some of which, including HTR10, DAZ1 and DMP9, exhibited a distinct expression pattern in generative cells (Figure 1H; Table S1). Thus, the overall clustering and gene expression pattern demonstrated that the cdka;1 SLCs have substantially acquired gametic cell specification, and can be used for preferential fertilization tests.

Single cdka;1 sperm cell-like cells preferentially fertilize the egg cell

Next, we used these manually separated and collected cdka;1 mutant pollen grains to investigate preferential fertilization. We first conducted limited pollination in which each ovule would be targeted by only a single pollen tube due to the polytubey block established at the septum (Zhong et al., 2022). In our growth conditions, the average number of ovules in one silique was 55.5 ± 2.8 (n = 10) (Figure S3). Therefore, 50 pollen grains, roughly equating to the number of ovules, were used for limited pollination so that only one single cdka;1 SLC is allowed to be released into one ovule for fertilization, maximizing the opportunity to evaluate preferential fertilization. To clearly visualize successful gamete fusion, 50 cdka;1 or wild-type pollen grains expressing pHTR10:HTR10-RFP were applied at pistils expressing the dual reporter line pEC1.1:tagRFP-RemA/pDD65:GFP, in which the egg cell and the central cell were labeled by red and green fluorescence, respectively (Cyprys et al., 2019). In the control, two wild-type sperm cell nuclei were observed at 7–9 h after pollination (HAP), one inside the egg cell and the other one inside the central cell (Figure 2A, B), indicating completion of double fertilization. Strikingly, in the ovules receiving cdka;1 pollen tubes, 99.5 ± 1.4% (n = 313) SLC nuclei were visible at 7–9 HAP inside the egg cell, while only 0.5 ± 1.4% SLC nuclei were observed inside the central cell (Figure 2A, B). This indicates that almost all cdka;1 SLCs fused with the egg cell.

Figure 2. Cyclin-dependent kinase a1 (cdka;1) sperm-like cells (SLCs) almost exclusively fertilize the egg cell.

(A, B) Representative images and statistical analysis of male-female gamete fusion events at 7–9 h after limited pollination using wild-type (n = 218) and cdka;1 (n = 313) pollen, respectively. White arrow heads indicate sperm and SLC nuclei labeled by pHTR10:HTR10-RFP, respectively. Scale bars, 20 μm. (C, D) Representative images and statistical analysis of seed development at 2 d after limited pollination using wild-type (n = 779) and cdka;1 (n = 730) pollen, respectively. Black arrows indicate the embryo, black arrowheads indicate endosperm and black asterisks indicate autonomously proliferated central cell nuclei. Em, embryo development; En, endosperm development; Em + En: development of both, embryo and endosperm. Scale bars, 50 μm. (E) Seed development at 3 d after limited pollination using wild-type and cdka;1 pollen, respectively. Embryo (black arrow) and endosperm (black arrowhead) are indicated. Scale bars, 50 μm. (F) Seed development in siliques after limited pollination using wild-type and cdka;1 pollen, respectively. cdka;1 pollinated pistils lack mature seeds. Scale bars, 500 μm. (G) At 7–9 h after excessive pollination using cdka;1 pollen, two SLCs fuse with egg cell and central cell, respectively. White arrow heads indicate SLC nuclei labeled by pHTR10:HTR10-RFP. Scale bar, 20 μm. (H) Both, embryo (black arrow) and endosperm (black arrow head) are formed after excessive pollination using cdka;1 pollen. Scale bar, 50 μm. (I) Silique development after excessive pollination using wild-type and cdka;1 pollen, respectively. White asterisks indicate seeds produced by cdka;1 pollination. Scale bars, 500 μm. (J) Statistical analysis of seed set after limited (n = 543 seeds, 10 siliques) and excessive (n = 534 seeds, 10 siliques) pollination using cdka;1 pollen. (K) Seeds from cdka;1 excessive pollination germinate and produce seedlings similar to the wild type. Scale bar, 1 cm. Data are mean values ± SD. ***P < 0.01.

To further confirm that the cdka;1 SLCs indeed successfully fertilized the egg cell, we examined seed development in the siliques. In contrast to wild-type siliques in which always both, an embryo and endosperm, were observed in all examined ovules (n = 779) at 2 d after pollination (DAP), only one single embryo was observed in 98.2 ± 1.7% of the ovules (n = 730) in cdka;1 mutant pollen-pollinated pistils (Figure 2C, D). Endosperm development was only observed in 0.9 ± 1.5% of ovules (Figure 2D). Similar to the wild type, zygotes resulting from single fertilization of the egg cell elongated and divided asymmetrically (Figure 2C) indicating that zygote activation was normal. Although the central cell was not fertilized in cdka;1 pollen-pollinated ovules, its nucleus divided autonomously for a few rounds and the multinucleated central cell collapsed before the embryo reached the globular stage at 3 DAP (Figure 2E). This observation was consistent with previous reports (Nowack et al., 2006, 2007). Mature seeds were never obtained when a limited number of cdka;1 pollen was applied (Figure 2F). These findings clearly demonstrate that the SLCs from the cdka;1 mutant pollen preferentially fertilize the egg cell.

To further clarify that the observed preferential fertilization of the egg cell was not caused by the inability of cdka;1 SLCs to fertilize the central cell, we conducted the same pollination experiments with excessive cdka;1 pollen. Our rationale was that secondary pollen tubes would be attracted due to the fertilization recovery system (Beale et al., 2012; Kasahara et al., 2012; Maruyama et al., 2013) to release additional SLCs for fertilization of the central cell. If cdka;1 SLC is defective in fertilizing the central cell, it should be equally difficult for the second cdka;1 SLC (delivered via fertilization recovery) to fertilize the central cell. As a result, the fertilization frequencies observed in the excessive pollination assays should be similar to those in the limited pollination assays. When 150 pollen grains (almost three times the number of ovules per silique) were applied, secondary cdka;1 pollen tubes were attracted in 36.6 ± 11.7% ovules (n = 533, Figure S4). The cdka;1 SLC nuclei were observed both in the egg cell and in the central cell in fertilized ovules at 7–9 HAP (Figure 2G). This indicates that the cdka;1 SLC not only fuses with the egg cell, but are also capable to fertilize the central cell. Ultimately, in siliques pollinated with excessive cdka;1 pollen, we obtained 20.7 ± 7.1% (n = 534) seeds, in which both embryo and endosperm developed normally (Figure 2H). This was significantly higher than the seed set ratio of 1.1 ± 0.9% (n = 543) in the limited pollination assays (Figure 2I, J). The resulting seeds could germinate and grow normally (Figure 2K), further confirming that both the embryo and endosperm are normal, and that cdka;1 SLCs are functional male gametes.

The fact that the egg cell/central cell fertilization ratios are not proportionate between the limited and excessive pollination assays indicates that the fertilization of the two female gametes by the SLCs are not statistically independent events. On the contrary, this discrepancy can be best explained if the fertilization of the central cell is only possible after the egg cell is fertilized. We further noticed that the central cell fertilization efficiency (20.7%) is lower than the entry efficiency of additional pollen tubes (36.6%). We hypothesized that the fertilization efficiency of fertilization recovery-triggered secondary pollen tubes was lower than that of the first pollen tubes. To test this hypothesis, we first pollinated the wild-type pistil with 60 hap2 (Zhong et al., 2022) pollen grains and then with 90 wild-type pollen grains for fertilization recovery. Among the ovules targeted by a second wild-type pollen tube, only 70.2 ± 9.2% were fertilized (n = 669) (Figure S5A, B), suggesting that fertilization recovery can only partially rescue the failures of the first fertilization. This result suggests that the microenvironment in the embryo sac probably is already not the best for a second fertilization after the failure of the first fertilization. Regarding the previously reported observation of additional proliferative SLCs in the female gametophyte using cdka;1 heterozygous mutants (Aw et al., 2010), we assume that it might have resulted from the fertilization recovery system, because, in our limited pollination assays, the cdka;1 SLC did not divide but remained as a single cell until it fused with the egg cell (Figure 2A, B). Taken together, our results indicate that, although the cdka;1 SLC is able to fertilize both female gametes, it preferentially fertilizes the egg cell in Arabidopsis.

In this study, we provide evidence to clarify several long-standing and controversial questions in plant reproduction. The first scientific question is about the cell identity of the cdka;1 SLC. A thorough examination of cdka;1 SLC cell fate was difficult in previous studies (Iwakawa et al., 2006; Nowack et al., 2006; Aw et al., 2010). Our single-cell RNA-seq analysis now reveals that cdka;1 SLCs contain a gene expression pattern highly similar to that of sperm cells. More importantly, genetic analysis showed that cdka;1 SLCs were capable of fertilizing either female gamete. These two lines of evidence jointly warrant that the fertilization event by cdka;1 SLC is a genuine fertilization event. Different from cdka;1 SLCs, SLCs from mutants such as duo1, although they are also formed after defective pollen mitosis II, were unable to fertilize the female gametes (Rotman et al., 2005; Brownfield et al., 2009). Future comparisons between these SLCs will help to identify new genes involved in sperm cell-specification and cell cycle-related function.

The second scientific question we addressed is about the long-lasting controversial question, preferential fertilization, that is, the egg cell fertilization is prioritized during double fertilization. Our pollen labeling-collecting strategy and precise control of the number as well as genotypes of the pollen donor eliminated the disturbance from the wild-type pollen or that from the fertilization recovery, which had hampered the accurate quantification of the fertilization results in previous studies (Iwakawa et al., 2006; Nowack et al., 2006; Aw et al., 2010).

Fertilization of the egg cell by a sperm is an ancient and evolutionarily conserved process in sexual reproduction involving cell fusion facilitators (Sprunck, 2020). Fertilization of the central cell evolved much later and is restricted to the angiosperms (Sharma et al., 2021), thus cell fusion might be less specific. This hypothesis is supported by the observation that the block to polyspermy (fusion of multiple sperm cells with a female gamete) is more stringent in the egg cell than in the central cell (Scott et al., 2008; Nagahara et al., 2020; Tekleyohans and Groß-Hardt, 2020). The absence of preferential fertilization toward the egg cell would raise the risk of fertilizing the central cell with both sperm cells in plant reproduction.

Finally, this study further raised a third scientific question about the source of the molecular factors controlling preferential fertilization and supports previous assumptions. Upon pollen tube burst, two sperm cells are discharged to the space between the egg cell and the central cell, and their positions are rectified to ensure that each sperm cell adheres to one female gamete (Huang et al., 2015), implying that communications occur among the female and male gametes. After gamete adhesion, plasmogamy proceeds. Notably, in hap2/gcs1 mutant, in which fusion with both female gametes fail, the two sperm cells are more frequently attached to the egg cell (Mori et al., 2014). In the dmp8 dmp9 mutant where egg-sperm plasmogamy was affected, unfused sperm cells were still more frequently attached to the egg cell as a pair (Cyprys et al., 2019). These findings imply preferential adhesion of sperm cells to the egg cell and possible communications among gametes before or upon the initiation of plasmogamy. Because the two sperm cells seem to be functionally equivalent to each other (Ingouff et al., 2007; Hamamura et al., 2011b), it is reasonable to predict that preferential fertilization is controlled by factors at the egg cell surface. For instance, egg cell-specific proteins, such as EC1 peptides that play roles in gamete attachment and fusion (Sprunck et al., 2012; Cyprys et al., 2019) might also be involved in promoting preferential fertilization of the egg cell. Other factors that function inequivalently in egg-sperm and central cell-sperm fertilizations, such as DMP8/DMP9 (Takahashi et al., 2018; Cyprys et al., 2019) and GLAUCE (Ngo et al., 2007; Leshem et al., 2012), are further candidates regulating preferential fertilization. The cdka;1 SLC system reported here is a very valuable tool to be included in future studies of fertilization mechanisms and associated gene function analyses.

In conclusion, we have shown that the single SLC from cdka;1 mutant pollen displays a sperm cell-specific gene repertoire, has the ability to fertilize both female gametes, and that the egg cell is preferentially fertilized. These findings provide new insights into our understanding of how double fertilization is implemented in angiosperms.

Supplementary Material

supplemental info

Materials and Methods

Figure S1. Viability of cyclin-dependent kinase a1 (cdka;1) pollen is similar to wild-type ones

Figure S2. Single-cell RNA-seq shows that cyclin-dependent kinase a1 (cdka;1) sperm-like cells (SLCs) cluster with sperm cells rather than with generative cells

Figure S3. The number of ovules in wild-type pistils

Figure S4. Ovules targeted by cyclin-dependent kinase a1 (cdka;1) pollen tubes attract secondary pollen tubes in excessive pollination assays due to fertilization recovery

Figure S5. Fertilization recovery by wild-type pollen tubes partially rescues the fertilization failure after hap2 pollination

Table S1. Single-cell RNA-seq data of representative genes functioning in the male gametes

Table S2. Primers used in this study

Biographies

Ling Li

Li-Jia Qu