Fertilization-Coupled Sperm Nuclear Fusion Is Required for Normal Endosperm Nuclear Proliferation

Abstract

Angiosperms exhibit double fertilization, a process in which one of the sperm cells released from the pollen tube fertilizes the egg, while the other sperm cell fertilizes the central cell, giving rise to the embryo and endosperm, respectively. We have previously reported two polar nuclear fusion-defective double knockout mutants of Arabidopsis thaliana immunoglobulin binding protein (BiP), a molecular chaperone of the heat shock protein 70 (Hsp70) localized in the endoplasmic reticulum (ER), (bip1 bip2) and its partner ER-resident J-proteins, ERdj3A and P58^IPK (erdj3a p58^ipk). These mutants are defective in the fusion of outer nuclear membrane and exhibit characteristic seed developmental defects after fertilization with wild-type pollen, which are accompanied by aberrant endosperm nuclear proliferation. In this study, we used time-lapse live-cell imaging analysis to determine the cause of aberrant endosperm nuclear division in these mutant seeds. We found that the central cell of bip1 bip2 or erdj3a p58^ipk double mutant female gametophytes was also defective in sperm nuclear fusion at fertilization. Sperm nuclear fusion was achieved after the onset of the first endosperm nuclear division. However, division of the condensed sperm nucleus resulted in aberrant endosperm nuclear divisions and delayed expression of paternally derived genes. By contrast, the other double knockout mutant, erdj3b p58^ipk, which is defective in the fusion of inner membrane of polar nuclei but does not show aberrant endosperm nuclear proliferation, was not defective in sperm nuclear fusion at fertilization. We thus propose that premitotic sperm nuclear fusion in the central cell is critical for normal endosperm nuclear proliferation.

Introduction

Nuclear fusion is essential for sexual reproduction in various organisms including plants and animals. During the life cycle of angiosperms, nuclear fusion occurs three times. Two of these nuclear fusion events occur during double fertilization, in which two sperm cells released from a pollen tube fertilize the egg and central cell, giving rise to the embryo and endosperm, respectively (van Went and Willemse 1984). After the fusion of a sperm cell with the egg or central cell, the sperm nucleus migrates to the female nucleus in an actin-dependent manner, resulting in nuclear membrane fusion (Kawashima et al. 2014, Ohnishi et al. 2014, Ohnishi and Okamoto 2017) and decondensation of the highly condensed sperm nuclear chromatin into the fused nucleus (Wagner et al. 1990, McCormick 1993, Ingouff et al. 2007, Kawashima et al. 2014). In maize (Zea mays) and rice (Oryza sativa) zygotes, sperm nuclear decondensation coincides with the activation of paternal genome (Scholten et al. 2002, Ohnishi et al. 2014).

The third nuclear fusion event takes place during female gametophyte development. Most angiosperms, including Arabidopsis thaliana, have a Polygonum type of female gametophyte, which consists of one egg cell, one central cell, two synergid cells and three antipodal cells. The central cell in many angiosperms, including Arabidopsis, initially contains two haploid polar nuclei, which fuse to produce a diploid secondary nucleus before fertilization (Yadegari and Drews 2004). Genes involved in the fusion of polar nuclei have been identified using genetic and molecular approaches; these include NFD1/RPL21M, GCD1 and RPS9M, all of which encode subunits of mitochondrial ribosomes (Portereiko et al. 2006, Wu et al. 2012, Lu et al. 2017). Female gametophytes of the gfa2 and fiona/syco mutants are also defective in polar nuclear fusion. The GFA2 gene encodes a J-domain-containing protein (J-protein) in the mitochondrial matrix (Christensen et al. 2002). J-proteins are a class of functional proteins that interact with heat shock protein 70 (Hsp70) through the well-conserved J-domain (Walsh et al. 2004, Bukau et al. 2006). The FIONA/SYCO gene encodes the cysteinyl t-RNA synthase in the mitochondrial matrix (Kägi et al. 2010).

We previously showed that the immunoglobulin binding protein (BiP), a molecular chaperone of the Hsp70 family of proteins in the endoplasmic reticulum (ER), is involved in the fusion of polar nuclei (Maruyama et al. 2010). The Arabidopsis genome harbors three BIP genes (BIP1–3) with redundant functions (Maruyama et al. 2014a). Female gametophytes of the bip1 bip2 double knockout mutant, which lacks two ubiquitously expressed BIP genes, contain two unfused polar nuclei in close contact. Hsp70 binds to and dissociates from its client proteins in an ATP-regulated cycle, and its chaperone function is regulated by partner proteins including J-proteins (Bukau et al. 2006). The observation that multiple J-proteins associate with Hsp70 may explain how Hsp70 achieves its versatile functions (Walsh et al. 2004). We previously identified a set of ER-resident J-proteins that function as partners for BiP in Arabidopsis (Yamamoto et al. 2008). We have also shown that three luminal ER-resident J-proteins, ERdj3A, ERdj3B and P58^IPK, function in the polar nuclear membrane fusion process. Female gametophytes of the double knockout mutants erdj3a p58^ipk and erdj3b p58^ipk show defects in the fusion of outer and inner membranes of polar nuclei, respectively. Therefore, P58^IPK/ERdj3A and P58^IPK/ERdj3B function as partners of BiP at distinct steps of the polar nuclear membrane fusion process (Maruyama et al. 2014b).

Female gametophytes of bip1 bip2 and erdj3a p58^ipk double mutants, which are defective in outer nuclear envelope fusion, are not defective in pollen tube guidance or gamete fusion when fertilized with wild-type (WT) pollen. However, after fertilization, >95% of seeds formed by fertilization of the mutant female gametophytes did not develop because of aberrant endosperm nuclear division (Maruyama et al. 2010, Maruyama et al. 2014b). Endosperm nuclei in bip1 bip2 mutant seeds display irregular size, abnormal ploidy and non-uniform distribution throughout the endosperm. Female gametophytes of the erdj3a p58^ipk double mutant also display similar aberrant endosperm nuclear proliferation. Although embryo development is initiated in mutant seeds, it is aborted at the globular and heart stages in bip1 bip2 and erdj3a p58^ipk double mutants, respectively. Defective seed development does not occur after the fertilization of erdj3b p58^ipk double mutant female gametophytes, which are defective in the inner nuclear envelope fusion (Maruyama et al. 2014b), indicating that defective polar nuclear fusion itself does not cause seed abortion.

Variation in seed development is observed after the fertilization of mutant female gametophytes defective in the fusion of polar nuclei. Seeds of the nfd1 mutant are defective in sperm nuclear fusion both in the egg and central cell; endosperm development in fertilized nfd1 mutant female gametophytes is not initiated even at 1 d after fertilization (Portereiko et al. 2006). The gcd1 and rps9m mutant female gametophytes are not defective in fertilization when fertilized with WT pollen; however, seed development in these mutants is arrested at an early stage of endosperm development (∼4 nuclei) with 1–2 cell zygotes (Wu et al. 2012, Lu et al. 2017). By contrast, fiona mutant female gametophytes are fully fertile when fertilized with WT pollen (Kägi et al. 2010). It remains unclear why aberrant endosperm nuclear division occurs after the fertilization of female gametophytes in bip1 bip2 and erdj3a p58^ipkdouble mutants.

In contrast to Arabidopsis, many angiosperms, including wheat (Triticum aestivum) and rice, produce female gametophytes containing unfused or partially fused polar nuclei in the central cell (You and Jensen 1985, Jones and Rost 1989). In these plant species, seed formation occurs normally despite incomplete polar nuclear fusion. In wheat, fusion between the sperm nucleus and one of the polar nuclei occurs in the central cell during fertilization, followed by the fusion of polar nuclei (Hoshikawa 1959). In maize, temporal analysis of double fertilization shows fusion of the sperm nucleus with either one of the polar nuclei or with the secondary nucleus, followed by normal endosperm development (Mòl et al. 1994).

To reveal why aberrant endosperm nuclear proliferation occurs after fertilization in bip1 bip2 and erdj3a p58^ipk mutants, we performed live-cell imaging analysis of nuclear fusion and division events in the endosperm. Our results showed that unfused polar nuclei fused during the first endosperm division in polar nuclear fusion-defective mutant female gametophytes including erdj3b p58^ipk, which does not show aberrant endosperm nuclear proliferation, suggesting that it was unlikely that aberrant endosperm nuclear proliferation was caused by the polar nuclear fusion defect. Unexpectedly, we found an unfused sperm nucleus in close contact with an unfused polar nucleus in the fertilized central cell of bip1 bip2 and erdj3a p58^ipk female gametophytes. In these fertilized mutant female gametophytes, triple nuclear fusion takes place during the first endosperm nuclear division, most likely via nuclear membrane breakdown and reconstitution, thus producing a triploid endosperm nucleus. However, we observed aberrant nuclear division as well as delayed expression of paternally derived genes after the fusion of highly condensed sperm nucleus in the resulting endosperm. Together, these data indicate roles of premitotic sperm nuclear fusion during fertilization in the activation of the paternal genome for normal endosperm nuclear proliferation and formation of the triploid endosperm genome.

Results

Endosperm development after fertilization of erdj3b p58^ipk mutant female gametophytes

First, we used time-lapse live-cell imaging to analyze nuclear dynamics during early endosperm development in fertilized erdj3b p58^ipk mutant ovules containing unfused polar nuclei but producing normal seeds. We pollinated erdj3b/erdj3b p58^ipk/+ pistils, which ubiquitously expressed histone H2B tagged with the tandem tomato (tdTomato) reporter gene under the control of RPS5A promoter (pRPS5A::H2B-tdTomato), with WT pollen, which expressed histone H2B tagged with the green fluorescent protein (GFP) gene driven by the same promoter (pRPS5A::H2B-GFP). Fertilized ovules were dissected from the pistils, and nuclear dynamics was analyzed using time-lapse imaging for 7–13.5 h after pollination (HAP). At 7 HAP, we observed two types of fertilized female gametophytes. One of these gametophytes contained one nucleus (n = 6), while the other contained two nuclei labeled with H2B-tdTomato in the central cell (n = 7); these corresponded to erdj3b and erdj3b p58^ipk female gametophytes, respectively (Fig. 1A, F). Time-lapse imaging showed that the erdj3b ovules exhibited normal endosperm development as observed in WT ovules (Maruyama et al. 2015). Two consecutive synchronized nuclear divisions in erdj3b ovules were observed during 6.5 h of observation (n = 6: Fig. 1A−E; Supplementary Movie S1, 3b). Although tdTomato fluorescence was dominant in the central cell of fertilized ovules at 7 HAP (Fig. 1A), the dividing endosperm nuclei exhibited both GFP and tdTomato fluorescence, indicating that nuclear fusion was successful during fertilization.

Fig. 1

Fusion of the polar nuclei occurs upon the first endosperm division in fertilized erdj3b p58^ipk female gametophytes that produce normal seeds. (A–E) The erdj3b/erdj3b p58^ipk/+ plants expressing HISTONE H2B-tdTomato from the RPS5A promoter were pollinated with WT pollen expressing HISTONE H2B-GFP from the RPS5A promoter. Panels show endosperm nuclear division in a seed produced by fertilization of an erdj3b female gametophyte. (F–J) Panels show endosperm nuclear division in a seed produced by fertilization of an erdj3b p58^ipk female gametophyte. Observation started 7 HAP. Time (h:min) from the start of observation is shown in each panel. syn, synergid nuclei; zn, zygote nucleus; en, endosperm nuclei; vgn, vegetative nucleus of pollen tube. Scale bars: 20 μm.

Fertilized female gametophytes of the erdj3b p58^ipk mutant contained two unfused polar nuclei labeled with H2B-tdTomato in the central cell (n = 7: Fig. 1F−J; Supplementary Movie S1, 3bp58). Two polar nuclei remained unfused at 7HAP, indicating that fertilization of the central cell with a WT sperm cell did not rescue the defective fusion of polar nuclei (Fig. 1F). Although erdj3b p58^ipk mutant female gametophytes are defective in polar nuclear fusion during the inner nuclear membrane fusion process (Maruyama et al. 2014b), they are not defective in fusion with a WT sperm nucleus. Sperm-derived GFP fluorescence was observed in dividing endosperm nuclei (Fig. 1H−J). However, sperm-derived GFP fluorescence was not visible in the central cell at 7 HAP, most likely because fusion had already occurred between the sperm nucleus and one of the unfused polar nuclei, which would have resulted in diffusion of GFP fluorescence into one of the unfused polar nuclei (Fig. 1F). Time-lapse imaging showed that two consecutive endosperm nuclear divisions occurred during 6.5 h of observation in all erdj3b p58^ipk ovules analyzed in this study (Fig. 1G−J). The unfused polar nuclei were observed to fuse during the first nuclear division, and tdTomato fluorescence coalesced to form an equatorial plane during metaphase of the first endosperm division (Supplementary Movie S1, 3b p58 at 01:40), and then segregated during anaphase. No retardation of nuclear division was observed in the endosperm of erdj3b p58^ipk ovules compared with that observed in fertilized erdj3b ovules.

Female gametophytes of erdj3a p58^ipk mutants are defective in sperm nuclear fusion in the central cell during fertilization

Next, we analyzed nuclear dynamics in the endosperm of erdj3a p58^ipk ovules fertilized with WT pollen. Live-cell imaging was performed as described above. Briefly, pistils of erdj3a/erdj3a p58^ipk/+ plants expressing pRPS5A::H2B-tdTomato were pollinated with pollen from WT transgenic plants expressing pRPS5A::H2B-GFP. At 7 HAP, we observed female gametophytes with one and two nuclei labeled with H2B-tdTomato in the central cell, corresponding to erdj3a and erdj3a p58^ipk female gametophytes, respectively (Fig. 2A, F, K). Time-lapse imaging showed two consecutive endosperm nuclear divisions in fertilized erdj3a female gametophytes during 6.5 h of observation (n = 5: Fig. 2A−E; Supplementary Movie S2, 3a). Strong GFP fluorescence was observed in the dividing endosperm nuclei (Fig. 2D, E), indicating that fusion between the sperm nucleus and secondary nucleus had already occurred at 7 HAP.

Fig. 2

Female gametophytes containing the erdj3a p58^ipk double mutation are defective in sperm nuclear fusion and subsequent nuclear divisions in the central cell. The erdj3a/erdj3a p58^ipk/+ plant expressing HISTONE H2B-tdTomato from the RPS5A promoter was pollinated with WT pollen expressing HISTONE H2B-GFP from the RPS5A promoter. (A–E) Panels show endosperm nuclear division in a seed produced by fertilization of an erdj3a female gametophyte. (F–O) Panels show endosperm nuclear division in a seed produced by fertilization of an erdj3a p58^ipk female gametophyte. Observation started 7 HAP. Time (h:min) from the start of observation is shown in each panel. syn, synergid nuclei; zn, zygote nucleus; en, endosperm nuclei; upn, unfused polar nuclei; usn, unfused sperm nucleus; vgn, vegetative nucleus of pollen tube; zn, zygote nucleus. Scale bars: 20 μm.

In contrast to the fertilized erdj3b p58^ipk female gametophytes, we observed punctate GFP fluorescence adjacent to the unfused polar nuclei in fertilized erdj3a p58^ipk female gametophytes (n = 8) at 7 HAP (Fig. 2F, K). Transmission electron microscopy of a fertilized erdj3a p58^ipk female gametophyte containing this punctate GFP signal showed the presence of one sperm nucleus in close contact with the central cell nuclei, which remained unfused (Fig. 3A−C). The unfused sperm nucleus was in close contact with one of the unfused polar nuclei; however, fusion of the nuclear membranes was not observed, even at the outer nuclear envelope (Fig. 3B). Additionally, fusion did not occur between unfused polar nuclei (Fig. 3C). These results indicate that erdj3a p58^ipk mutant female gametophytes are defective in fusion between the sperm and central cell nuclei during fertilization. The GFP and tdTomato fluorescence merged in egg cells of erdj3a p58^ipk ovules at 7 HAP (Fig. 2F, K), suggesting that fusion between the egg and sperm nucleus was normal.

Fig. 3

Electron micrographs of a fertilized erdj3a p58^ipk double mutant female gametophyte with an unfused sperm nucleus in the central cell. (A) Electron micrographs of the unfused sperm nucleus associated with two polar nuclei. The erdj3a/erdj3a p58^ipk/+ plant expressing HISTONE H2B-tdTomato from the RPS5A promoter was pollinated with pollen from WT plants carrying the pRPS5A::H2B-GFP. Ovules with a punctate GFP signal of the unfused sperm nucleus were selected and analyzed by electron microscopy. (B) and (C) are magnifications of boxes with white and black dotted lines in (A), respectively. Outer and inner nuclear envelopes of the sperm nucleus or polar nuclei are indicated by black and white arrowheads, respectively. upn1, unfused polar nucleus 1; upn2, unfused polar nucleus 2; usn, unfused sperm nucleus. Scale bars: in (A), 1 μm; in (B) and (C), 0.5 μm.

Aberrant endosperm division and delayed paternal genome activation in ovules defective in sperm nuclear fusion in the central cell during fertilization

Live-cell imaging revealed that endosperm entered the nuclear division cycle in fertilized erdj3a p58^ipk ovules. Two types of nuclear division events were observed in mutant ovules. In one type of nuclear division (four ovules: Fig. 2F−J; Supplementary Movie S2, 3a p58 #1), GFP and tdTomato fluorescence coalesced to form an equatorial plane-like image at metaphase (Fig. 2G). Subsequently, both fluorophores diffused to form an elongated sphere, suggesting that the nucleus entered anaphase; however, no nuclear division occurred during 6.5 h of observation (Fig. 2H−J). In the other type of nuclear division (four ovules: Fig. 2K−O; Supplementary Movie S2, 3a p58 #2), GFP and tdTomato fluorescence coalesced before the formation of the equatorial plane during metaphase. Strong GFP fluorescence was observed in the dividing nuclei, which indicates fusion between the sperm nucleus and unfused polar nuclei (Fig. 2N, O). In these ovules, only one endosperm nuclear division event was observed during 6.5 h of observation, indicating retardation of nuclear division. These results suggest significant retardation of nuclear division in the endosperm of fertilized erdj3a p58^ipk ovules, despite sperm nuclear fusion during the first endosperm nuclear division cycle.

We noticed retarded expression of the paternally derived pRPS5A::H2B-GFP construct in the developing endosperm of fertilized erdj3a p58^ipk ovules (Fig. 4). In fertilized erdj3a ovules, H2B-GFP expression in endosperm started to increase at 30 min after metaphase of the first nuclear division (Fig. 4A−F, M, open circles). By contrast, in fertilized erdj3a p58^ipk ovules, the induction of H2B-GFP expression in the endosperm was not observed even at 120 min after metaphase of the first nuclear division (Fig. 4G−L, M, closed circles). The temporal expression profile of pRPS5A::H2B-GFP was similar between the endosperm of fertilized erdj3b ovules and that of fertilized erdj3b p58^ipk ovules (Fig. 4N), suggesting that defects in polar nuclear fusion do not affect the expression of paternally derived genes. We also observed retarded expression of paternally derived pAGL62::AGL62-GFP construct after the fertilization of erdj3a p58^ipk ovules but not of erdj3a, erdj3b or erdj3b p58^ipk ovules (Fig. 4O, P;Supplementary Movie S3). These data suggest that defective sperm nuclear fusion in fertilized erdj3a p58^ipk ovules likely causes retarded expression of paternally derived genes in the developing endosperm.

Fig. 4

Defective sperm nuclear fusion in the central cell at fertilization retards expression of paternally derived genes in endosperm. The erdj3a/erdj3a p58^ipk/+ plant expressing HISTONE H2B-tdTomato from the RPS5A promoter was pollinated with WT pollen expressing HISTONE H2B-GFP from the RPS5A promoter (as in Fig. 2 A–P). (A–F) Panels show time-lapse images of GFP fluorescence in the endosperm of an erdj3a ovule (3a). (G–L) Panels show time-lapse images of GFP fluorescence in the endosperm of an erdj3a p58^ipk ovule (3a p58). Dotted lines show position of the endosperm nucleus indicated by HISTONE H2B-tdTomato. Numbers indicate time (min); 0 min corresponds to metaphase of the first endosperm division. Scale bars: 20 μm. (M) Quantitative analysis of mean fluorescent signals in the endosperm nuclei in erdj3a (open circles) and erdj3a p58^ipk (closed circles) ovules after fertilization (A–L). (N) Quantitative analysis similar to that in (M) was performed after fertilization of erdj3b (open circles) and erdj3b p58^ipk (closed circles) ovules with WT pollen expressing HISTONE H2B-GFP from the RPS5A promoter (as in Fig. 1). (O and P) Pistils of the erdj3a/erdj3a p58^ipk/+ (O) or the erdj3b/erdj3b p58^ipk/+ (P) were pollinated with pollen from WT plants expressing AGL62-GFP from its own promoter. Quantitative analyses similar to those in (M and N) were performed. Error bars in (M), (N), (O) and (P) represent mean and standard deviations (n = 3 or more ovules). Asterisks indicate significant difference between 3a and 3a p58 in each time point (P < 0.05, Student’s t-test).

Female gametophytes of bip1 bip2 mutants show defective sperm nuclear fusion during fertilization both in the egg and central cell

Female gametophytes of the bip1 bip2 mutant also showed defects in sperm nuclear fusion during fertilization. We performed live-cell imaging analyses after pollinating bip1/+ bip2/bip2 pistils expressing H2B-tdTomato with WT pollen expressing H2B-GFP, as described above. Out of 38 fertilized ovules analyzed, 13 ovules contained a female gametophyte with a secondary nucleus at 7 HAP, which corresponded to a bip2 mutant (Fig. 5A). These ovules did not show obvious defects in endosperm nuclear division (Fig. 5B–E; Supplementary Movie S4, bip2); one (three ovules) and two consecutive (10 ovules) nuclear divisions were observed. The remaining 25 ovules were of the genotype bip1 bip2 and contained unfused polar nuclei in the central cell at 7 HAP (Fig. 5F, K). Punctate GFP fluorescence was observed adjacent to the unfused polar nuclei in all fertilized bip1 bip2 female gametophytes, indicating that these female gametophytes were also defective in sperm nuclear fusion in the central cell during fertilization (Fig. 5F, K). Although all bip1 bip2 ovules entered the endosperm nuclear division cycle, nuclear division was significantly delayed. We observed one endosperm nuclear division in three ovules, but no ovule entered the second nuclear division. In 20 ovules, unfused polar nuclei fused with the sperm nucleus during metaphase and entered anaphase. However, the first endosperm nuclear division was not completed within 7.5 h of observation (Fig. 5G–J; Supplementary Movie S4, bip1 bip2 #1). Notably, in two ovules, we observed formation of two types of endosperm nuclei derived from each of the unfused polar nuclei (Fig. 5L–O; Supplementary Movie S4, bip1 bip2 #2). Unfused polar nuclei appeared to separate before metaphase (Fig. 5L). Each unfused polar nucleus formed a spindle simultaneously (Fig. 5M). However, subsequent nuclear division phenotypes were different. Punctate GFP fluorescence was observed adjacent to the polar nucleus on the micropylar side until the division cycle entered metaphase (Fig. 5K, M). Division of the micropylar side nucleus started but did not complete. By contrast, two consecutive nuclear divisions were observed in the other polar nucleus on the chalazal side (Fig. 5L–O; Supplementary Movie S4, bip1 bip2 #2). After the first endosperm nuclear division event, GFP signals were observed in all endosperm nuclei (Fig. 5O;Supplementary Movie S4, bip1 bip2 #2). In this ovule, the sperm nucleus probably fused with the polar nucleus on the micropylar side. Newly synthesized H2B-GFP was transported to all endosperm nuclei by diffusion in the coenocytic endosperm before cellularization. Alternatively, triple nuclear fusion took place in this ovule, but the fused sperm nucleus somehow only affected division of the polar nucleus on the micropylar side, which had been in close contact with the unfused sperm nucleus. These results support the idea that the observed aberrant endosperm nuclear proliferation was caused by sperm nuclear fusion during the first endosperm nuclear division.

Fig. 5

Retardation of sperm nuclear fusion was observed in two female gametes containing the bip1 bip2 double mutations. (A-O) The bip1/+ bip2/bip2 plants expressing HISTONE H2B-tdTomato from the RPS5A promoter were pollinated with WT pollen expressing HISTONE H2B-GFP from the RPS5A promoter. (A-E) Panels show endosperm nuclear division in a seed produced by fertilization of a bip1 female gametophyte. (F-O) Panels show retardations of sperm nuclear fusions and endosperm nuclear division in a seed produced by fertilization of a bip1 bip2 female gametophyte. (P-T) The bip1/+ bip2/bip2 plants expressing BiP1 from the central cell-specific DD65 promoter were pollinated with WT pollen expressing HISTONE H2B-GFP from the RPS5A promoter. Panels show endosperm nuclear division in a seed produced by fertilization of a bip1 bip2 female gametophyte carrying the pDD65::BIP1. Observation started 7 HAP. Time (h:min) from the start of observation is shown in each panel. syn, synergid nuclei; zn, zygote nucleus; en, endosperm nuclei; upn, unfused polar nuclei; uen, unfused egg nucleus; usn, unfused sperm nucleus; vgn, vegetative nucleus of pollen tube. Scale bars: 20 μm.

Additionally, we observed a punctate GFP signal in the egg cell of 24 out of 25 fertilized bip1 bip2 ovules (Fig. 5F, K), indicating failure of sperm nuclear fusion in the egg cell. Live-cell imaging showed that the sperm nucleus remained unfused during observation in 11 ovules. In the remaining 13 ovules, the sperm nucleus fused with the egg nucleus (Fig. 5G–J, L–O; Supplementary Movie S4, bip1 bip2 #1 and bip1 bip2 #2). These results indicate that the egg cell nucleus of bip1 bip2 female gametophytes is defective in sperm nuclear fusion during fertilization but retains its fusion activity.

Previously, we showed that the expression of BIP1 cDNA driven by the central cell-specific DD65 promoter in bip1 bip2 female gametophytes complements the polar nuclear fusion defect as well as seed abortion after fertilization (Maruyama et al. 2010). Live-cell imaging showed that BIP1 expression from the DD65 promoter rescued the sperm nuclear fusion defect of bip1 bip2 female gametophytes in the central cell but not in the egg cell. Pistils of bip1/+ bip2/bip2 plants that were hemizygous for the pDD65::BIP1 transgene were pollinated with WT pollen expressing H2B-GFP. One-fourth of the ovules in the pistils were expected to contain the bip1 bip2 double mutation and pDD65::BIP transgene. Out of 19 fertilized ovules, six contained unfused sperm nuclei both in the egg and central cell. Among these six ovules, three contained an unfused sperm nucleus only in the egg cell at 7 HAP (Fig. 5P), suggesting that these ovules were most likely bip1 bip2 pDD65::BIP1 ovules. Time-lapse imaging showed strong GFP fluorescence in the dividing endosperm nuclei, indicating that sperm nuclear fusion occurred in the central cell (Fig. 5Q–T; Supplementary Movie S5). These results suggest that rescue of the sperm nuclear fusion defect in the central cell is enough to ensure proper endosperm nuclear proliferation of bip1 bip2 female gametophytes after fertilization. The unfused sperm nucleus in the egg cell started to diffuse during the second endosperm nuclear division (Fig. 5T;Supplementary Movie S5), suggesting the initiation of sperm nuclear fusion or degradation of the unfused sperm nucleus.

Discussion

Double knockout mutations erdj3a p58^ipk and bip1 bip2 cause sperm nuclear fusion defect in female gametophytes during fertilization

Nuclear fusion is an essential process during fertilization in both plants and animals. Here, we showed that fertilization-coupled sperm nuclear fusion in the central cell is essential for normal endosperm nuclear proliferation in Arabidopsis. Polar nuclear fusion deficient erdj3a p58^ipk and bip1 bip2 female gametophytes were also defective in sperm nuclear fusion during fertilization (Figs. 2F, K, 3A, B, 5F, K). The sperm nuclear fusion defect of erdj3a p58^ipk and bip1 bip2 female gametophytes was observed when mutant female gametophytes were fertilized with WT pollen.

It is unlikely that the observed sperm nuclear fusion defect is due to the lack of BiP and its partner proteins in sperm cells. BiP functions in key cellular processes, such as protein translocation and protein folding, in various organisms including Arabidopsis. All three BIP genes are expressed in mature pollen. The bip1 bip2 bip3 triple mutation causes pollen lethality, indicating essential roles of BiP proteins in pollen (Maruyama et al. 2014a). Transcriptome analyses have revealed the expression of ERDJ3A gene in sperm cells (Borges et al. 2008). It is likely that BiP, together with ERdj3A and P58^IPK, functions in nuclear fusion by regulating the conformation or assembly of as yet unidentified proteins required for nuclear membrane fusion. It should be noted that karyogamy is defective in the yeast bip mutant when it is mated with WT yeast (Rose 1996).

Delayed sperm nuclear fusion and aberrant endosperm nuclear proliferation

Endosperm nuclear division starts in fertilized erdj3a p58^ipk and bip1 bip2 ovules without obvious delay, despite the sperm nuclear fusion defect (Figs. 2G, 5G, M). Sperm fusion most likely triggers the nuclear division in the central cell without sperm nuclear fusion, as reported previously (Aw et al. 2010, Zhao et al. 2018). The sperm nucleus fuses with polar nuclei during the first endosperm nuclear division in fertilized erdj3a p58^ipk and bip1 bip2 mutant ovules, thus producing triploid endosperm nuclei. However, the resulting endosperm shows aberrant proliferation.

Aberrant endosperm nuclear division observed in erdj3a p58^ipk and bip1 bip2 ovules is most likely due to the fusion of condensed sperm nucleus during the first nuclear division. The sperm nucleus contains highly condensed chromatin (Wagner et al. 1990, McCormick 1993), which must decondense before nuclear division starts. In this study, punctate H2B-GFP fluorescence was observed in the central cell of fertilized erdj3a p58^ipk and bip1 bip2 female gametophytes, indicating that sperm nuclear chromatin remained condensed (Figs. 2F, K, 5F, K). In Arabidopsis, first nuclear division starts approximately 2 h after sperm entry into the central cell (Faure et al. 2002). It is possible that chromosomes in the sperm nucleus did not have enough time to prepare for nuclear division because sperm chromatin remained condensed at the start of nuclear division (Figs. 2H, I, L, M, 5G, H, L, M). The failed sperm nuclear fusion during fertilization was also associated with a significant delay in the expression of paternally derived genes (Fig. 4M, O). Based on these data, we propose that fertilization-coupled premitotic sperm nuclear fusion during fertilization plays key roles in the activation of the paternal genome through nuclear decondensation and in the formation of triploid endosperm genome (Fig. 6, erdj3a p58^ipk and bip1 bip2).

Fig. 6

A schematic representation of nuclear fusion and division in the central cell after fertilization of WT, bip1 bip2, erdj3a p58^ipk and erdj3b p58^ipk ovules. Female and male nuclei are shown in magenta and green, respectively. Endosperm nuclei containing both parental genomes are colored with purple. Magenta and green bars represent female and male chromosomes at metaphase of the first mitotic division, respectively. Male chromosomes in endosperm of bip1 bip2 and erdj3a p58^ipk ovules remain condensed at this stage (indicated by a green-colored knotted lines). Sperm nuclear fusion occurs normally in endosperm of erdj3b p58^ipk ovules.

Failed sperm nuclear fusion in the fertilized central cell has also been observed upon the fertilization of WT female gametophytes with cdka;1 mutant sperm cells (Aw et al. 2010); however, the resulting seeds do not show aberrant proliferation or size of endosperm nuclei (Nowack et al. 2006, Nowack et al. 2007, Aw et al. 2010, Maruyama et al. 2010, Maruyama et al. 2014b), in contrast to fertilized erdj3a p58^ipk and bip1 bip2 ovules. Moreover, paternally derived genes were not expressed in the endosperm of seeds produced by fertilization with cdka;1 mutant sperm cells, suggesting the elimination of sperm nucleus at some point during fertilization (Aw et al. 2010). By contrast, in this study, the sperm nucleus was not eliminated from fertilized erdj3a p58^ipk and bip1 bip2 ovules, as evident from the expression of paternally derived genes in the developing endosperm (Figs. 2J, O, 4G–L, M, O 5J, O,).

While the sperm nuclear fusion defect of the erdj3a p58^ipk female gametophyte was specific to the central cell, bip1 bip2 female gametophytes showed significant delay in sperm nuclear fusion in the fertilized egg cell. We observed punctate GFP signal in the egg cell after the start of endosperm nuclear proliferation (Fig. 5I, N). We previously showed that early embryo development (∼2 d after pollination) occurs in fertilized bip1 bip2 ovules, similar to that in WT seeds (Maruyama et al. 2010). Consistent with this observation, we showed that sperm nuclear fusion in the egg cell of fertilized bip1 bip2 and bip1 bip2 pDD65::BIP1 female gametophytes was completed before the first zygotic division (Fig. 5J, O, T). It is likely that the delay of sperm nuclear fusion approximately 6 h did not affect the onset of embryo development, which is reported to start approximately 12 h after fertilization in Arabidopsis (Boisnard-Lorig et al. 2001).

In the mammalian zygote, fusion between male and female nuclei occurs during the first mitotic division by nuclear envelope breakdown and reassembly. During fertilization, the highly condensed sperm nucleus undergoes remodeling to form the male pronucleus, which fuses with the female pronucleus (McLay and Clarke 2003). Because the male pronucleus is decondensed and transcriptionally active, nuclear fusion during mitosis does not affect subsequent mitotic divisions. By contrast, sperm nuclear fusion is achieved without production of male pronucleus in angiosperms, including Arabidopsis. Chromatin decondensation in the male gamete is achieved by sperm nuclear fusion at fertilization both in the egg and central cells (Ingouff et al. 2007, Kawashima et al. 2014). In this study, we showed that sperm nuclear fusion during the first endosperm division resulted in aberrant endosperm nuclear proliferation (Figs. 2F-O, 5F–O); this suggests that premitotic sperm nuclear fusion plays an important role in the decondensation and activation of the paternal genome during plant reproduction.

Completion of polar nuclear fusion at the start of endosperm nuclear proliferation

Our live-cell imaging analyses indicate that fertilization or sperm nuclear fusion does not rescue defective polar nuclear fusion. The central cell of erdj3a p58^ipk, erdj3b p58^ipk and bip1 bip2 mutant ovules contained two unfused polar nuclei after fertilization with WT sperm cells, indicating that the introduction of WT genome is not sufficient to rescue the polar nuclear fusion defect. However, endosperm nuclear proliferation was observed in the fertilized mutant female gametophytes, irrespective of the success of the sperm nuclear fusion in the central cell. Polar nuclear fusion was achieved during the first endosperm nuclear division, most likely via breakdown and reassembly of the nuclear envelope, similar to that observed during the fusion of sperm and egg pronuclei in mammals (Longo 1985).

Punctate H2B-GFP signal was not detected in fertilized erdj3b p58^ipk ovules, unlike that observed in the fertilized erdj3a p58^ipk and bip1 bip2 ovules, due to sperm chromatin decondensation after the fusion of the sperm nucleus with one of the unfused polar nuclei (Fig. 1F). The erdj3b p58^ipk female gametophytes were fully fertile, which suggests that the establishment of triploid endosperm nucleus during the first endosperm nuclear division is sufficient for normal seed development, as long as sperm nuclear fusion occurs during fertilization (Fig. 6, erdj3b p58^ipk). Normal seed formation has been reported after the fertilization of fiona mutant female gametophytes containing unfused polar nuclei with WT pollen (Kägi et al. 2010), which is consistent with our results. Unfused polar nuclei are much larger than the sperm nucleus and are assumed to be in a decondensed state before the completion of nuclear fusion.

Using live-cell imaging, we showed that premitotic sperm nuclear fusion in the fertilized central cell is a prerequisite for normal endosperm nuclear proliferation. Sperm nuclear fusion results in the decondensation of sperm chromatin, which is essential for the activation and normal segregation of the paternal genome in the developing endosperm. Once the sperm nuclear fusion is successfully completed upon fertilization, normal endosperm development is initiated, irrespective of the completion of polar nuclear fusion. In the case of central cell containing unfused polar nuclei, the formation of triploid endosperm nucleus can be achieved during the first endosperm nuclear division. Similar mechanisms probably operate to produce the triploid endosperm after fertilization in species such as rice, wheat and maize, which produce female gametophytes with unfused polar nuclei in the central cell (You and Jensen 1985, Jones and Rost 1989, Mòl et al. 1994). This enables the production of seeds in various angiosperms with female gametophytes containing unfused polar nuclei.

Materials and Methods

Plant materials and growth conditions

Arabidopsis qrt1-2 (CS8846; ecotype Columbia) was used as the WT strain in this study. All plants used in this study were homozygous for the qrt1-2 allele. The bip1-4 (CS856879), bip2-1 (CS842467), erdj3a-1 (SALK_103280), erdj3b-1 (SALK_113364) and p58^ipk-1 (SALK_140273) mutant alleles have been described previously (Yamamoto et al. 2008, Maruyama et al. 2010). Transgenic lines harboring pRPS5A::H2B-GFP (Adachi et al. 2011) and pAGL62::AGL62-GFP (Kang et al. 2008) were provided by Dr. Daisuke Kurihara at the Nagoya University, Nagoya, Japan and Dr. Gary Drews at the University of Utah, Salt Lake City, UT, respectively. The bip1-4/+ bip2-1/bip2-1, erdj3a-1/erdj3a-1 p58^ipk-1/+ and erdj3b-1/erdj3b-1 p58^ipk-1/+ plants expressing H2B-tdTomato from the RPS5A promoter were constructed by introducing the pRPS5A::H2B-tdTomato construct (a gift from Dr. D. Kurihara; Adachi et al. 2011) using the floral-dip method (Clough and Bent 1998). The bip1-4/+ bip2-1/bip2-1 plant expressing BIP1 from the DD65 promoter was constructed as described previously (Maruyama et al. 2010). Seeds were surface-sterilized using chlorine gas, and then sown on soil or Murashige-Skoog (MS) medium (Wako, Osaka, Japan) containing 0.7% agar and 1% sucrose. Plants were grown at 22°C under continuous light.

Transmission electron microscopy

Flowers of erdj3a-1/erdj3a-1 p58^ipk-1/+ plants expressing H2B-tdTomato from the RPS5A promoter were emasculated at late stage 12 (Smyth et al. 1990) and pollinated with pollen expressing H2B-GFP at 1 d after emasculation. Ovules were dissected from the pistils at 7 HAP and placed on half-strength MS agar medium containing 5% sucrose and 1.5% NuSieve GTG agarose. Ovules containing two tdTomato-labeled polar nuclei and a punctate GFP signal emanating from the unfused sperm nucleus in the central cell were selected under a fluorescent microscope and subsequently fixed in a solution containing 2% glutaraldehyde, 2% paraformaldehyde and 50 mM sodium cacodylate (pH 7.4) for 3 d at 4°C. Tissue segments were washed in buffer and post-fixed for 8 h in 2% aqueous osmium tetroxide at 4°C. Tissues were then dehydrated in a graded ethanol series, transferred into propylene oxide, infiltrated and embedded in Quetol 651. A series of thin sections (80 nm) were prepared, stained with 2% aqueous uranyl acetate and lead citrate, and examined at 80 kV using a JEOL JEM 1200 EX electron microscope (JEOL Ltd., Tokyo, Japan).

Time-lapse live-cell imaging

Flowers were emasculated at late stage 12 and pollinated with WT pollen carrying pRPS5A::H2B-GFP or pAGL62::AGL62-GFP at 1 d after emasculation. Six HAP ovules were dissected from the pistils and mounted on a cover slip with half-strength MS medium containing 5% sucrose (pH adjusted to 5.7 with 1 N KOH). Time-lapse confocal imaging was started at 7 HAP using an inverted microscope (IX-81, Olympus, Tokyo, Japan) equipped with an automatically programmable XY stage (MD-XY30100T-Meta; Molecular Devices, Sunnyvale, CA, USA), a disk-scan confocal system (CSU-XI, Yokogawa Electric, Tokyo, Japan), 488 nm and 561 nm LD lasers (Sapphire, Coherent Inc., Santa Clara, CA, USA), and an EM-CCD camera (Evolve 512, Photometrics, Tucson, AZ, USA). Time-lapse images were acquired every 10 min with 8 z-planes at 1.5 μm intervals, with sequential exposures of 600 and 200 ms at excitation wavelengths of 488 and 561 nm, respectively, using a water-immersion objective lens (UApo 40 × W3/340; Olympus, Tokyo, Japan). Images were processed with MetaMorph Ver. 7.7.7.7.0. (Universal Imaging Corp., Downingtown, PA, USA) to create maximum intensity projection images. From the projection images, signal intensities of the endosperm nuclei were analyzed by ImageJ Ver. 1.43 u (http://imagej.nih.gov/ij/). Regions of interest (ROIs) were set on the primary endosperm nucleus or one of the endosperm nuclei located at the micropylar end based on their tdTomato signals. The means of the background green fluorescent intensities of the ovules were subtracted from the means of the green fluorescent intensities of the ROI at each time point, and the corrected signal intensities were normalized by using the signal in the primary endosperm nucleus at 30 min before metaphase. Means and standard deviations of the relative signal intensities of three or four ovules were plotted. Significant differences between mutant ovules were analyzed at each time point using Student’s t-test with StatPlus:mac (AnalystSoft Inc.—statistical analysis program for macOS. Version v6. See http://www.analystsoft.com/en/).

Image processing

ImageJ Ver. 1.43 u (http://imagej.nih.gov/ij/) was used for editing images and time-lapse movies. All other images were processed for publication using Adobe Photoshop CS Ver. 8.0.1 (Adobe Systems Inc., San Jose, CA, USA).

Funding

The Ministry of Education, Culture, Sports, Science and Technology of Japan Grants-in-Aid for Scientific Research on Priority Areas (no. 1685202 to S.N.); Grants-in-Aid for Scientific Research on Innovative Areas [grant numbers 23120512, 25120711, 17H05837 and 19H04857 to S.N., 17H05846 and 19H04869 to D.M., 16H06464 to T.H.]; Grants-in-Aid for Scientific Research [grant numbers 23570051, 16K07394 and 19K06704 to S.N.]; Grants-in-Aid for Exploratory Research [grant number 15K14541 to D.M.]; Grants-in-Aid for Young Scientists (A) [grant number 16H06173 to D.M.]; Grants-in-Aid for Scientific Research on Priority Areas [grant number 19058005 to T.E.], Japan Society for Promotion of Science KAKENHI Grant-in-Aid for Scientific Research on Innovative Areas—Advanced Bioimaging Support (ABiS, Grant Number JP16H06280) and Japan Science and Technology Agency [ERATO project JPMJER1004 to T.H.]. D.M. was supported by a fellowship from the Japan Society for the Promotion of Science [grant number 6526].

Acknowledgments

We thank Drs. Daisuke Kurihara and Gary Drews for providing plant materials, and Daichi Susaki for his assistance in statistical analysis.

Disclosures

The authors have no conflicts of interest to declare.