https://orcid.org/0000-0001-6382-1365
Abstract
Female gametophyte (FG) is crucial for reproduction in flowering plants. Arabidopsis thaliana produces Polygonum-type FGs, which consist of an egg cell, two synergid cells, three antipodal cells and a central cell. Egg cell and central cell are the two female gametes that give rise to the embryo and surrounding endosperm, respectively, after fertilization. During the development of a FG, a single megaspore produced by meiosis undergoes three rounds of mitosis to produce an eight-nucleate cell. A seven-celled FG is formed after cellularization. The central cell initially contains two polar nuclei that fuse during female gametogenesis to form the secondary nucleus. In this study, we developed a gene induction system for analyzing the functions of various genes in developing Arabidopsis FGs. This system allows transgene expression in developing FGs using the heat-inducible Cre-loxP recombination system and FG-specific embryo sac 2 (ES2) promoter. Efficient gene induction was achieved in FGs by incubating flower buds and isolated pistils at 35°C for short periods of time (1–5 min). Gene induction was also induced in developing FGs by heat treatment of isolated ovules using the infrared laser-evoked gene operator (IR-LEGO) system. Expression of a dominant-negative mutant of Sad1/UNC84 (SUN) proteins in developing FGs using the gene induction system developed in this study caused defects in polar nuclear fusion, indicating the roles of SUN proteins in this process. This strategy represents a new tool for analyzing the functions of genes in FG development and FG functions.
Introduction
Female gametophyte (FG) plays an essential role in the reproductive processes of flowering plants. Most flowering plants, including Arabidopsis thaliana, produce Polygonum-type FGs, comprising an egg cell, a central cell, two synergid cells and three antipodal cells (Drews and Koltunow 2011). Post-fertilization, the egg cell and central cell give rise to the embryo and endosperm, respectively, in seeds. The synergid cells play central roles in pollen tube guidance and reception (Dresselhaus et al. 2016, Higashiyama and Yang 2017).
The development of a FG within a developing ovule consists of two phases, megasporogenesis and megagametogenesis. During megasporogenesis, four haploid megaspores are produced after meiosis. During megagametogenesis, three of the four megaspores degenerate, whereas one surviving megaspore develops into the mature FG. The megaspore containing a single haploid nucleus (stage FG1) undergoes three rounds of mitosis. After the second mitosis, a four-nucleate cell, with two nuclei at each pole (stage FG4), is produced. Cellularization starts after the third mitosis, resulting in the formation of a seven-celled and eight-nucleate FG (stage FG5). The central cell contains two polar nuclei that migrate toward the center of the FG and fuse to form the secondary nucleus before fertilization (stage FG6) in A. thaliana and other species (Christensen et al. 1998, Yadegari and Drews 2004, Drews and Koltunow 2011).
Genes required for megagametogenesis have been identified in large-scale screenings for FG mutants (Christensen et al. 1998, Pagnussat et al. 2005, Brukhin et al. 2011). Genes required for FG functions also have been identified in screenings of genes expressed specifically in FGs (Kasahara et al. 2005, Yu et al. 2005, Johnston et al. 2007, Steffen et al. 2007, Wang et al. 2010, Wuest et al. 2010, Sprunck et al. 2012, Takeuchi and Higashiyama 2012). Phenotypic analyses of mutants identified in these screenings or those obtained using reverse genetics approaches have been used to reveal the role of the identified genes in FG development and FG functions. Techniques that induce gene expression by different external factors have been used as powerful tools for the analyses of various gene functions in plants (Moore et al. 2006). In this study, we developed a gene expression system based on heat-inducible Cre-loxP recombination in developing FGs as a new tool for analyzing the functions of genes in FG development and FG functions.
Cre-loxP-mediated site-specific DNA recombination has been demonstrated in Arabidopsis (Russell et al. 1992). The heat-inducible gene induction system has been used for mosaic analysis (Sieburth et al. 1998), conditional gene knockout (Nagahara et al. 2015) and transgene induction in somatic cells (Ogawa et al. 2015). We showed efficient induction of transgene expression in developing FGs using a FG-specific promoter. In the gene induction system developed in this study, we expressed a dominant-negative mutant of Sad1/UNC84 (SUN) proteins in developing FGs. SUN proteins are located in the inner nuclear membrane and contain a conserved SUN domain at the C-terminus (Zhou et al. 2015a). The SUN domain resides in the perinuclear space and interacts with the Klarsicht/ANC-1/Syne homology (KASH) proteins spanning the outer nuclear membrane. While the N-terminal nucleoplasmic domain of SUN proteins associates with a variety of nuclear components, the N-terminal cytoplasmic domain of KASH proteins functions in interaction with cytoskeletal elements. These SUN–KASH protein pairs, also known as the linker of nucleoskeleton and cytoskeleton (LINC) complex, mechanically couple nuclear and cytoplasmic structures. These LINC complexes are involved in various processes such as nuclear positioning and nuclear movement (Zhou et al. 2015a, Meier 2016).
SUN proteins are present in animals, fungi, green plants and other eukaryotes. The Arabidopsis genome encodes five SUN proteins, which are classified into two types: SUN1 and SUN2 are classical SUN proteins with a SUN domain at the C-terminus; SUN3, SUN4 and SUN5 are classified as mid-SUN proteins harboring an internal SUN domain (Graumann et al. 2014, Zhou et al. 2015a). SUN proteins are essential for pollen and embryo development and consequently for fertility, making it difficult to construct multiple sun gene knockout mutant lines (Graumann et al. 2014, Varas et al. 2015, Zhou et al. 2015b). Herein, we show that the expression of a dominant-negative mutant of SUN proteins causes defects in polar nuclear fusion, indicating the roles of SUN proteins in this process. Thus, our gene induction system could serve as a new tool for the analysis of gene function in plant reproduction.
Results
Development of a FG-specific gene induction system
We developed a gene induction system that allows FG-specific expression of a target gene by heat treatment. Our gene induction system was based on the heat-inducible Cre-loxP recombination and FG-specific Arabidopsis embryo sac 2 (ES2) promoter (At1g26795) (Yu et al. 2005, Pagnussat et al. 2007) and utilized two gene constructs (Fig. 1A). The driver construct contained the coding sequence of Cre recombinase placed downstream of the Arabidopsis heat shock protein 18.2 (HSP18.2) gene promoter (Ogawa et al. 2015). The effector cassette contained the coding sequence of histone H2B gene fused to the green fluorescent protein (GFP) gene (H2B-GFP) flanked by loxP sites on either side, followed by the nopaline synthase terminator (NosT); this cassette was placed between the ES2 promoter and the target gene. Heat treatment of FG for a short period was expected to induce Cre-loxP site-specific recombination and result in the removal of the H2B-GFP-NosT cassette. Continued expression of the target gene from the ES2 promoter was expected to take place after the removal of heat. To analyze gene induction using this system, we generated transgenic lines expressing a fusion of the mitochondrial targeting sequence pCOXIV with GFP (pCOXIV-GFP) (Maruyama et al. 2015) as the target gene.
A FG-specific gene induction system using Cre-loxP-based site-specific recombination. (A) Schematic representation of the gene induction system. A sequence containing the pES2 driving the coding sequence of histone H2B-GFP fusion (H2B-GFP) and NosT, flanked by two loxP sequences on either side in the same orientation, followed by the coding sequence of yeast cytochrome c oxidase subunit 4 (pCOXIV) fused to GFP (pCOXIV-GFP). This construct was introduced into the HS-Cre line expressing the Cre recombinase under the control of the Arabidopsis HSP18.2 promoter (pHSP18.2 Cre). Heat treatment of ovules or pistils of transgenic plants at 35°C for approximately 5 min induced Cre-loxP recombination, resulting in the expression of pCOXIV-GFP from the ES2 promoter. (B–D) Images of a transgenic FG6 ovule. (E–G) Images of a FG6 ovule isolated from a pistil that was heated at 35°C for 5 min and further incubated at 22°C for 16 h. GFP fluorescence (B and E, green), chlorophyll autofluorescence (C and F, magenta) and merged (D and G) images are shown. en, egg nucleus; syn, synergid nuclei; sn, secondary nucleus; an, antipodal cell nuclei. Scale bar = 25 μm.
Analysis of FGs of transgenic plants by confocal laser scanning microscopy (CLSM) showed FG-specific expression of H2B-GFP from the ES2 promoter. Nuclei of the egg cell, central cell, two synergid cells and three antipodal cells were observed in FGs at the FG6 stage (Fig. 1B–D). The ES2 promoter-mediated expression of H2B-GFP was observed in the nuclei of developing FGs at the FG4 and FG5 stages (Supplementary Fig. S1), suggesting that gene induction is possible not only in mature FGs but also in developing FGs at the FG4 stage. We did not observe GFP signals in the mitochondria of FGs without heat treatment (ovules of five pistils), indicating that basal expression from the HSP18.2 promoter is tightly repressed in FGs.
Expression from the HSP18.2 promoter is induced by heating plants at 35–37°C (Takahashi et al. 1992, Ogawa et al. 2015). We heated pistils of transgenic plants by immersing whole inflorescences in water at 35°C for 5 min. Ovules were dissected from the pistils at 16 h after heat treatment and analyzed using CLSM. Punctate GFP signals were observed in FG cells (Fig. 1E). Tubular GFP signals, which are characteristic of mitochondria staining, were observed at higher magnifications (Supplementary Fig. S2), indicating expression of pCOXIV-GFP from the ES2 promoter. These results also indicate that the expression of Cre protein following heat treatment led to recombination in FG cells. Despite the deletion of the H2B-GFP cassette by Cre-loxP site-specific recombination, GFP signal was observed in the nucleus, probably because of the stability of the H2B-GFP fusion protein.
Short and mild heat treatment of pistils and inflorescences results in efficient gene induction in FGs
We analyzed the temperature dependency of gene induction in the heat-inducible system. Pistils were isolated from transgenic flowers at the late stage 12, immersed in ovule culture medium (Gooh et al. 2015) and then heated using a thermal cycler. Pistils were further incubated at 22°C, fixed and cleared by ClearSee (Kurihara et al. 2015). The expression of pCOXIV-GFP in pistils was analyzed using CLSM. To determine the temperature dependency of gene induction, we heated the pistils to 30–40°C for 1 min. Heat treatment at 35°C was the most efficient (Fig. 2A); at this temperature, 42% ± 13% of ovules (from an average of seven pistils) showed pCOXIV-GFP fluorescence in the FG. We, therefore, used 35°C as the temperature for heat-inducible gene expression in further experiments.
Dependency of gene induction on the temperature and duration of heat treatment. (A) Pistils isolated from transgenic flowers at late stage 12 were heated at the indicated temperature for 1 min using a thermal cycler and then incubated at 22°C for 16 h. The percentage of ovules containing a pCOXIV-GFP-positive FG was scored for each pistil. Data represent mean ± standard deviation (n ≥ 6 for each temperature). (B) Transgenic flowers at late stage 12 were emasculated and heated by immersing a whole inflorescence in water at 35°C for the indicated times. After heat treatment, plants were incubated at 22°C for 16 h. The percentage of ovules containing a pCOXIV-GFP-positive FG was scored for each pistil. Data represent mean ± standard deviation (n = 8 for each temperature).
Next, we analyzed the dependency of gene induction on the duration of heat treatment using flower buds. Flowers at the late stage 12 were emasculated, and whole inflorescences were immersed in water at 35°C for 1–20 min. At 16 h post-heat treatment, pistils were isolated from the emasculated flowers, fixed and analyzed using CLSM, as described above. The efficiency of gene induction showed no increase with the duration of heat treatment (Fig. 2B), as observed in isolated pistils. These results indicate that a short (∼5 min) heat treatment at 35°C is sufficient for achieving efficient gene induction in this system. When we analyzed pistils 6 h after heat treatment, 8% ± 3% of ovules (from an average of five pistils) showed pCOXIV-GFP fluorescence in the FG, suggesting that gene induction started shortly after heat treatment.
Induction of Cre-loxP recombination in isolated ovules
To control heat-inducible Cre-loxP recombination in a FG of isolated ovules, the expression of Cre protein was induced by laser irradiation using the infrared laser-evoked gene operator (IR-LEGO) system (Kamei et al. 2009). Because the rise in tissue temperature depends on the power of infrared laser irradiation and the depth and size of the target cell, we estimated that the heat shock response may be induced by heating the FGs at 25–40°C using approximately 10 mW irradiation for 1 min (Kamei et al. 2009, Nakano et al. 2017). The duration of irradiation was selected as 1 min because Kamei et al. (2009) showed that irradiation using low laser power for 1 min resulted in more efficient gene induction (90% of irradiated cells) in Caenorhabditis elegans than irradiation using high laser power for 1 s (60% of irradiated cells). However, in the case of animals, fast movement of target cells makes it difficult to perform irradiation for 1 min. Therefore, 1 s irradiation has been widely used in IR-LEGO of animals (Deguchi et al. 2009, Shimada et al. 2013, Kawasumi-Kita et al. 2015, Hasugata et al. 2018). An irradiation time of 1 s was also used for the gene expression/deletion system of Marchantia polymorpha (Nishihama et al. 2016). Assuming that isolated ovules would be less mobile, we used 1 min irradiation in this study. Isolated ovules containing a developing FG at the FG5 stage were placed in a glass-bottom dish containing ovule culture medium, and the region containing the FG was irradiated with a laser power of 8 mW. The expression of pCOXIV-GFP in the irradiated FG was observed post-irradiation (Fig. 3A). Expression of pCOXIV-GFP can be achieved shortly after irradiation because punctate GFP signals were detected in the irradiated FG at 6.5 h post-irradiation (Supplementary Fig. S3). These results indicate that the IR-LEGO system can be used for heat-induced gene expression in a FG of isolated ovules.
Targeted induction of gene expression in developing FGs using the IR-LEGO system. (A) An example of gene induction using the IR-LEGO system. An ovule containing a FG at the FG5 stage was isolated from a transgenic plant and placed in a glass-bottom dish containing the ovule culture medium. An infrared (1,480 nm) laser at 8 mW was used in a microscope-based irradiation system to irradiate the FG for 1 min. The ovule was cultured at 22°C for 19 h after irradiation and analyzed using CLSM. GFP fluorescence (green), chlorophyll autofluorescence (magenta) and merged images are shown. Scale bar = 25 μm. (B) Effect of laser power on gene induction and cell damage in FGs. Ovules containing a FG at the FG5 stage were irradiated with an infrared laser at 7–10 mW for 1 min. Gene induction and cell damage were scored at 16 h after irradiation. Circles indicate the percentage of FGs showing pCOXIV-GFP expression. Squares indicate the percentage of damaged FGs showing no or low nuclear GFP signals. The number of trials is shown in the upper margin.
The expression of pCOXIV-GFP varied with the power of irradiation. No expression of pCOXIV-GFP was observed when FGs were irradiated at 7 mW for 1 min (Fig. 3B; n = 7). Irradiation of FGs at 8 or 9 mW resulted in the expression of pCOXIV-GFP in the FG of 64% (n = 14) or 58% (n = 12) of ovules, respectively, at 16 h post-irradiation. In some instances, laser-damaged FGs containing no or reduced nuclear GFP signals were observed. The percentage of ovules containing a damaged FG was 7% with a laser power of 8 mW. However, the percentage of ovules containing a damaged FG increased significantly with the increase in laser power. At 10 mW, the expression of pCOXIV-GFP was observed in the FG of 50% (n = 16) of ovules; however, the remaining 50% of ovules contained a damaged FG. At 11 mW, irradiation for 1 min could not be achieved, as ovules moved because of the convection of culture medium due to heating. We, therefore, did not analyze gene induction at a laser power of 11 mW or higher.
Expression of a dominant-negative mutant of SUN proteins in developing FG causes defects in the fusion of polar nuclei
We used the FG-specific gene induction system for the expression of a dominant-negative mutant of SUN proteins in FGs. The dominant-negative approach is especially effective for the functional analysis of proteins encoded by multiple genes such as actin (Kawashima et al. 2014). We chose SUN proteins (Zhou et al. 2015a) as targets because (i) these proteins are encoded by multiple genes and are potentially involved in nuclear fusion and (ii) a dominant-negative mutant of SUN proteins was available (Zhou et al. 2015b).
We previously reported that, in budding yeast (Saccharomyces cerevisiae), the SUN protein Mps3/Nep98 is required for nuclear fusion (karyogamy) during mating (Nishikawa et al. 2003). The mechanisms of nuclear fusion in yeast (karyogamy) and Arabidopsis (polar nuclear fusion) are similar. Both processes proceed without nuclear envelope breakdown; nuclear fusion is achieved by sequential fusions of the outer and inner nuclear membranes. Studies report the involvement of BiP, a molecular chaperone HSP70 in the endoplasmic reticulum (ER), and its regulatory partners, J-protein family co-chaperones (J-proteins), in these processes (Rose et al. 1989, Nishikawa and Endo 1997, Brizzio et al. 1999, Maruyama et al. 2010, Maruyama et al. 2014). The SUN protein Mps3/Nep98 interacts with Jem1, a J-protein required for karyogamy (Nishikawa and Endo 1997, Nishikawa et al. 2003). This prompted us to question the role of SUN proteins in polar nuclear fusion.
A dominant-negative mutant of SUN proteins (SUNDN) has been reported previously (Zhou et al. 2015b). The SUNDN comprises an N-terminal ER targeting signal of 2S albumin, a monomeric red fluorescent protein (tagRFP), the lumenal domain of Arabidopsis SUN2 protein including the coiled-coil and SUN domains and a C-terminal ER retention signal (HDEL) (Fig. 4A). The expression of SUNDN in Arabidopsis pollen causes the delocalization of the KASH protein, WIP1 and its binding partner, WIT1, from the envelope of the vegetative nucleus, causing defects in nuclear movement in the pollen tube (Zhou et al. 2015b).
Expression of a dominant-negative mutant of SUN protein in developing FGs inhibits polar nuclear fusion. (A) Schematic representations of Arabidopsis SUN2 and SUNDN proteins. Numbers above SUN2 indicate the first and last amino acid residues of SUN2. SUNDN comprised the signal sequence of Arabidopsis 2S albumin (SS) fused to tagRFP and the lumenal domain of SUN2 containing the SUN domain (SUN) and coiled-coil (CC) domain tagged with the C-terminal ER retention signal (GGGHHHHHHDEL). TM, transmembrane region of SUN2. (B) Representative images of an ovule after the induction of SUNDN in the developing FG. Transgenic flowers at late stage 12, expressing SUNDN from the ES2 promoter using heat-inducible Cre-loxP-based site-specific recombination, were emasculated and heated by immersing the whole inflorescence in water at 35°C for 5 min. Ovules were dissected from the heat-treated pistils at 16 h post-heat treatment and analyzed by CLSM. GFP fluorescence (green), tagRFP fluorescence (magenta) and merged images are shown. Upn, unfused polar nuclei; syn, synergid nuclei; cc, central cell; ec, egg cell; sc, synergid cells. The egg nucleus was out of focus and therefore not visible. Scale bar = 25 μm. (C, D) Whole inflorescences were heated at 35°C for 5 min three times (at 0, 16 and 24 h relative to the first heat treatment). Early to middle stage 12 flowers in the first heat treatment were emasculated at 20 h and fixed at 64 h after the first heat treatment. Control experiments without heat treatment were performed in parallel. Three or two independent transgenic lines expressing SUNDN or SUNDN-Mut (a mutant version of SUNDN containing mutations in the KASH-binding region), respectively, in the FG were analyzed. Panel (C) shows percentages of tagRFP-positive FGs in a pistil. Panel (D) shows percentages of FGs containing unfused polar nuclei among tagRFP-positive and -negative FGs in a pistil. Statistical differences were calculated using the Tukey-Kramer method. P < 0.05 is indicated by different letters. No polar nuclear fusion was observed in FGs in pistils without heat treatment (n ≥ 7 for each line). Data represent mean ± standard deviation (n ≥ 7).
We generated transgenic plants expressing heat-inducible SUNDN in developing FGs using our gene induction system. Early to middle stage 12 flowers of three independent transgenic lines were heated at 35°C for 5 min; this treatment was repeated three times at 0, 16 and 24 h relative to the first heat treatment. At 40 h after the third heat treatment, pistils were fixed, cleared by ClearSee and analyzed using CLSM. Fluorescence signals of tagRFP were observed in the FG of 65–80% of ovules in heat-treated pistils (Fig. 4B, C), indicating the induction of SUNDN by heat treatment. Repeated heat treatment resulted in a higher rate of gene induction than the single heat treatment (Fig. 2B). A strong tagRFP signal was observed in the central cell, while weak tagRFP signals were observed in the egg cell and synergid cells (Fig. 4B). No tagRFP signal was observed in ovules of transgenic lines without heat treatment (Fig. 4C). Ovules showing tagRFP signal contained unfused polar nuclei in close contact (Fig. 4B). Defective polar nuclear fusion was observed in approximately 45% of tagRFP-positive ovules but not in tagRFP-negative ovules (Fig. 4D). No polar nuclear fusion defect was observed in pistils without heat treatment, suggesting a correlation between the polar nuclear fusion defect and SUNDN expression.
Two point mutations in the SUN domain (H434A and Y438F) impair the KASH-binding activity of SUN2 (Zhou et al. 2014). The SUNDN protein containing the double mutation (SUNDN-Mut) did not interact with KASH proteins or WPP domain-interacting proteins (WIPs; WIP1, WIP2 and WIP3) and lost the dominant-negative activity of SUNDN (Zhou et al. 2015b). We also generated transgenic lines expressing heat-inducible SUNDN-Mut in a FG. The efficiency of SUNDN-Mut induction in FGs by heat treatment was similar to that of SUNDN induction (Fig. 4C). However, unlike SUNDN, the expression of SUNDN-Mut did not cause the polar nuclear fusion defect. Only approximately 7% of tagRFP-positive ovules contained unfused polar nuclei in the two independent transgenic lines (Fig. 4D). No polar nuclear fusion defect was observed in tagRFP-negative ovules. These results indicate that the short repetitive heat treatment at 35°C did not cause the polar nuclear fusion defect. Instead, the polar nuclear fusion defect was caused by the expression of SUNDN retaining its KASH-binding activity. Since SUN proteins function together with KASH proteins in various processes, our results strongly suggest the involvement of SUN proteins in polar nuclear fusion.
Discussion
In this study, we developed a system for gene induction in Arabidopsis FGs by combining heat-inducible Cre-loxP site-specific recombination (Ogawa et al. 2015) with the FG-specific ES2 promoter. Cell type-specific gene induction can be achieved using driver constructs that allow tissue-specific expression of transactivators (Schürholz et al. 2018). Chemically inducible gene expression has been widely used for transgene induction in Arabidopsis using steroid hormone-based transactivators (Moore et al. 2006). This technique has also been used for the analysis of gene function in FGs. Kawashima et al. (2014) previously used this technique for the induction of a dominant-negative mutant of actin in the central cell of mature FGs, and demonstrated the role of F-actin in fertilization. In this study, efficient and stable gene expression could be induced in developing FGs using a short heat treatment at 35°C. Thus, our gene induction system provides a new tool that complements the chemically inducible systems.
The use of pCOXIV-GFP as a target gene resulted in no detectable Cre-loxP recombination in FGs in plants grown at 22°C, indicating that gene expression from the HSP18.2 promoter is tightly repressed in the absence of heat treatment. The heat shock response of the HSP18.2 promoter was observed in FGs, similar to that observed in diploid tissues and the pollen tube (Takahashi et al. 1992). We observed efficient Cre-loxP recombination after heat treatment of flowers for a short period of time (1–5 min), indicating that the amount of Cre protein expressed in a cell following a short heat treatment was sufficient for the induction of recombination between the two loxP sites. The efficiency of Cre-loxP recombination was the highest when pistils were heated at 35°C (Fig. 2A), which is consistent with the results of Takahashi et al. (1992), where heat shock induction of the GUS gene from the HSP18.2 promoter was maximal at 35°C.
Our gene induction system could also be used for heat-inducible gene expression in FGs of isolated ovules. We showed efficient induction of Cre-loxP recombination in FGs of isolated ovules using the IR-LEGO technique. It was possible to select a specific FG for heat application using this technique. Thus, our gene induction system could be used to induce a target gene in a FG at a defined developmental stage. In combination with live-imaging analyses of FG development (Gooh et al. 2015), this targeted gene induction technique represents a new strategy for the analysis of stage-specific gene function in FG development.
Additionally, we showed that the expression of SUNDN, a dominant-negative mutant of SUN proteins, in developing FGs inhibited polar nuclear fusion, suggesting the involvement of SUN proteins in this process. Not all tagRFP-positive ovules showed defects in polar nuclear fusion: approximately 45% of tagRFP-positive ovules contained unfused polar nuclei (Fig. 4D). This was probably due to the relatively long length of the FG5 stage in female gametogenesis. FG5 ovules are found in flowers in the period between mid-stage 12 to stage 13 (Christensen et al. 1997), whose duration is expected to be >1 d (Smyth et al. 1990). It is possible that the polar nuclear fusion process does not proceed synchronously in a pistil. At the onset of SUNDN expression, a fraction of FGs in a pistil probably had proceeded to steps that do not require SUN protein functions. Nevertheless, ovules from untreated pistils and tagRFP-negative ovules from heat-treated pistils did not show the polar nuclear fusion defect, indicating a correlation between the polar nuclear fusion defect and SUNDN expression. Although we observed the tagRFP signal in the egg cell and synergid cells, the tagRFP signal was strongest in the central cell (Fig. 4B), probably because expression from the ES2 promoter is predominant in the central cell in mature FGs (Steffen et al. 2007; reported as the DD9 promoter in this literature).
A polar nuclear fusion defect, observed in FGs expressing SUNDN, appeared to be at the nuclear membrane fusion step. In all FGs showing the polar nuclear fusion defect, unfused polar nuclei were in close contact in the central cell (Fig. 4B), indicating nuclear movement. Our results thus suggest the roles of SUN proteins in nuclear membrane fusion. The yeast SUN protein, Mps3/Nep98, is also required for nuclear fusion during mating. Nuclear fusion during yeast mating can be dissected into two processes: nuclear congression in yeast zygotes and nuclear membrane fusion (Kurihara et al. 1994). In zygotes lacking Mps3, nuclear congression is not affected; two haploid nuclei are in close contact but do not fuse (Rogers and Rose 2015). This indicates the roles of Mps3 in nuclear membrane fusion. Our results suggest that the function of SUN proteins in nuclear membrane fusion during reproductive processes is conserved between yeast and plants.
Our results also indicate the importance of the KASH-binding activity of SUNDN for the inhibition of polar nuclear fusion. Expression of SUNDN-Mut in developing FGs did not cause the polar nuclear fusion defect (Fig. 4D), suggesting the involvement of SUN–KASH interactions during polar nuclear fusion. Recent analyses have identified three WIPs, four SUN-interacting nuclear envelope proteins (SINEs) and TIK as KASH proteins in Arabidopsis (Zhou et al. 2015a). However, the role of these KASH proteins in female gametogenesis has not yet been reported. Interestingly, Zhou and Meier (2014) reported that WIPs and their binding partners, WIT1 and WIT2, play important roles in male fertility. The wip1 wip2 wip3 triple mutant and wit1 wit2 double mutant plants show male fertility defects, which are associated with impaired pollen tube reception. Moreover, these mutants show impaired movement of the vegetative nucleus in the pollen tube and inefficient delivery of sperm cells. SUN–KASH interactions appear to be important for nuclear migration in pollen tubes. Expression of SUNDN in pollen tubes also results in male fertility defects similar to those observed in wip and wit mutants (Zhou et al. 2015b). By contrast, we did not observe defects in the movement of polar nuclei in FGs expressing SUNDN. These data suggest that SUN–KASH interactions are not required for the movement of polar nuclei. Alternatively, it is possible that defects in polar nuclear movement were not visible because of the short travel distance of polar nuclei. Detailed analyses in combination with live imaging will help determine whether the expression of SUNDN is required for the movement of polar nuclei.
We showed that the gene induction system developed in this study could be used for the analysis of gene function in FG development. To analyze the role of SUN proteins in polar nuclear fusion, we used the SUNDN protein. Similar analyses would be possible for genes involved in other processes of FG development, provided dominant-negative mutants of these genes are available. Because efficient gene induction is possible in planta, the gene induction system developed in this study could be used for the analysis of gene function after FG development (e.g. during fertilization) by inducing the expression of a dominant-negative mutant in mature FGs. Analysis of stage-specific gene function in developing FGs is possible using the IR-LEGO system, as discussed above. Another possible application of this gene induction system potentially includes stage-specific complementation of loss-of-function mutants defective in FG development. Our heat-inducible Cre-loxP recombination system is not limited to the analysis of gene function in FG. Various cell type-specific promoters are available in Arabidopsis for the expression of transgenes (Schürholz et al. 2018). By replacing the ES2 promoter with these promoters, cell type-specific induction of gene expression is possible in somatic cells. Region-specific induction of gene expression by IR-LEGO has also been reported in Arabidopsis and Marchantia (Deguchi et al. 2009, Nishihama et al. 2016). Targeted induction of a gene expression in a specific type of cell or tissue is a promising new tool for the analysis of tissue-specific gene functions in plants at the single cell level.
Materials and Methods
Plant materials and growth conditions
The heat shock promoter (HS)-Cre line of A. thaliana (Ogawa et al. 2015), provided by Dr. Taku Takahashi at the Okayama University, Okayama, Japan, was used in this study. Seeds of this line were surface sterilized and sown on soil or Murashige and Skoog (MS) medium (Wako, Osaka, Japan) containing 0.7% agar and 1% sucrose. Plants were grown at 22°C under continuous light.
Plasmid construction and plant transformation
Primers used for plasmid construction are listed in Supplementary Table S1. A 1.5 kb DNA fragment containing the loxP-HISTONE H2B-NosT-loxP cassette was PCR amplified from pDME100 using primers loxP-H2BF-infusion/loxP-H2BR-infusion2. The plasmid pDME100 was generated by cloning the 1.5 kb HISTONE H2B-NosT-loxP fragment, which was amplified from pRPS5A::H2B-GFP (Adachi et al. 2011) using primers pENTR_H2B_F and loxP-NOSter_R, into the pENTR/D-TOPO vector (Life Technologies, Carlsbad, CA, USA). The amplified fragment was inserted into the XbaI site of pGWB501 (Nakagawa et al. 2007) using the In-Fusion HD Cloning Kit (Takara Bio USA, Mountain View, CA, USA), thus generating pSW1. A 1.1 kb DNA fragment containing the ES2 promoter, which was PCR amplified from the genomic DNA of Arabidopsis plants using primers 1g26795-F and 1g26795-R, was cloned into the HindIII site of pSW1 using the In-Fusion HD Cloning Kit, thus generating pSW9. A 0.8-kb DNA fragment containing pCOXIV-GFP (Maruyama et al. 2015) was amplified by PCR using primers pCOXIV-F and GFP-R and cloned into the pENTR/D-TOPO vector, generating pSW11. The fusion gene for SUNDN was constructed as follows. A DNA fragment corresponding to amino acid residues 129–455 of SUN2, containing the C-terminal ER retention signal, was amplified from a cDNA clone RAFL19-33-G08 (provided by RIKEN BioResource Center) using primers Sun2LmF and Sun2LmR and then cloned into the pENTR/D-TOPO vector to generate pSNA115. A DNA fragment corresponding to the signal sequence (amino acid residues 1–40) of Arabidopsis 2S1 was PCR amplified from the genomic DNA of Arabidopsis plants using primers 2S1F and 2S1RFPR. A DNA fragment corresponding to amino acid residues 1–237 of tagRFP was amplified from pTagRFP-C (Evrogen) using primers tagRFPCDSF and tagRFPCDSR. A DNA fragment corresponding to amino acid residues 129–455 of SUN2 with the C-terminal ER retention signal was amplified from pSNA115 using primers RFPSUN2F and SUN2HDELR. All three amplified DNA fragments were ligated using the In-Fusion HD Cloning Kit and then cloned into the pENTR/D-TOPO vector, generating pSNA118. The SUNDN-Mut sequence contained H434A and Y438F point mutations in SUN2 (Zhou et al. 2015b). These mutations were introduced by PCR amplification of pSNA118 using primers SUN2dMut-Fw and SUN2dMut-Rv, followed by self-ligation. The constructs for pCOXIV-GFP, SUNDN and SUNDN-Mut were introduced into pSW9 using LR clonase II (Life Technologies).
To generate transgenic lines, Agrobacterium tumefaciens strain GV3101 was introduced into Arabidopsis plants using the floral-dip method (Clough and Bent 1998). Transgenic plants were selected on MS agar plates containing 50 μg/ml hygromycin and subsequently transferred to soil. Transgenic plants in the T3 generation that were homozygous for the transgene were used.
Heat treatment of pistils
Heat treatment of isolated pistils was performed as described below. A pistil isolated from a flower at late stage 12 was placed in a 0.2 ml PCR tube containing 50 μl of ovule culture medium (Gooh et al. 2015). The sample was heated in a thermal cycler (Dice Gradient, Takara Bio, Shiga, Japan) without a heated lid. The heated pistil was incubated at 22°C under dark conditions. Heat treatment was also applied by immersing the emasculated pistils or flower buds in water at 35°C. Plants were then brought back to the growth room at 22°C and incubated under continuous light.
Irradiation of ovules using the IR-LEGO system
Ovules containing FGs at the FG4 or FG5 stage were dissected from pistils of transgenic plants and mounted in a multi-well glass-bottom dish (D141400, Matsunami Glass, Osaka, Japan) containing 400 μl of ovule culture medium. Irradiation using an infrared (1,480 nm) laser was performed as described previously (Deguchi et al. 2009, Shimada et al. 2013, Hasugata et al. 2018). An IR-LEGO 1000 system (Sigma-Koki, Saitama, Japan) equipped with a custom-made 40× objective lens (UAPO340 40×/0.90 UV; Olympus, Tokyo, Japan) was used for laser irradiation. Ovules were irradiated at 7–10 mW for 1 min and then incubated at 22°C under dark conditions. The large central vacuole in developing FGs was used as a target for irradiation.
Microscopy
Ovules were dissected from pistils, mounted in a multi-well glass bottom dish containing 400 μl of ovule culture medium and analyzed using CLSM. Ovules intact in pistils were analyzed using CLSM after fixation and clearing with ClearSee (Kurihara et al. 2015). A Leica TCS-SP8 confocal microscope fitted with a 20× multi-immersion objective lens (PL APO CS2 20×/0.75 IMM CORR HC; Leica Microsystems, Mannheim, Germany) or a 40× or 63× water-immersion objective lens (PL APO CS2 40×/1.10 W CORR HCX, PL APO CS2 63×/1.20 W CORR HC; Leica Microsystems) was used for CLSM. For GFP fluorescence, images of fluorescence at 495–540 nm were captured after excitation at 488 nm with a solid-state laser. For tagRFP and chlorophyll fluorescence, images of fluorescence at 560–650 nm and 562–700 nm, respectively, were captured after excitation with a 552 nm solid-state laser.
To create maximum intensity projection images, the acquired images were processed using the LASX software (Leica Microsystems). These images were then processed using Adobe Photoshop CC (Adobe Systems Inc., San Jose, CA, USA) for publication.
Funding
The Ministry of Education, Culture, Sports, Science and Technology of Japan Grant-in-Aid for Scientific Research on Innovative Areas [25120711, 17H05837 and 19H04857 to S.N.], Grant-in-Aid for Scientific Research [16K07394 and 19K06704 to S.N.], Grant-in-Aid for Challenging Exploratory Research [26650094 to S.N.], Grant-in-Aid for Challenging Research (Pioneering) [17H06258 to Y.K.] and NIBB Collaborative Research Program for Integrative Imaging [16-152 and 18-517 to S.N.]. Japan Society for Promotion of Science KAKENHI Grant-in-Aid for Scientific Research on Innovative Areas—Advanced Bioimaging Support [ABiS, JP16H06280].
Acknowledgments
We thank Drs. Daisuke Maruyama, Tsuyoshi Nakagawa and Taku Takahashi for providing plant materials. We also thank Ms. Misako Saida for assistance with the IR-LEGO system and Dr. Daisuke Maruyama for critically reading the article.
Disclosures
The authors have no conflicts of interest to declare.
References
Author notes
Dukhyun Hwang, Satomi Wada and Azusa Takahashi authors contributed equally to this work.
