The AtUBQ10 promoter is suitable for driving transient gene expression in pollen germ cells
To identify an effective promoter for driving the expression of Cas9 in pollen, we evaluated the activity of selected promoters in bombarded pollen based on fluorescent protein expression. Fluorescent proteins are driven under the control of cauliflower mosaic virus (CaMV) 35S, A. thaliana ribosomal protein S5A (AtRPS5A), and A. thaliana UBIQUITIN10 (AtUBQ10) promoters, which have previously been used as constitutive promoters (Liang et al. 2018; Nekrasov et al. 2013; Tsutsui and Higashiyama 2017), as well as the pollen vegetative cell-specific LAT52 promoter derived from S. lycopersicum (Eady et al. 1995). The plasmid DNA vectors containing these promoters, which were introduced into the pollen and leaves, are summarized in Table S1. Tricellular pollen is generally short-lived relative to bicellular pollen, which increases the difficulty of handling this type of pollen (Hoekstra and Bruinsma 1975). In the present study, we used four species with bicellular pollen to investigate the introduction of the aforementioned plasmid DNA vectors, namely N. benthamiana (tobacco), N. tabacum (tobacco), T. fournieri (torenia), and S. lycopersicum (tomato), for which the in vitro induction of pollen tube germination has been established (Liang et al. 2018; Okuda et al. 2009; Paungfoo-Lonhienne et al. 2010; Zhang et al. 2020). Biolistic delivery into leaves and pollen was examined twice, and gene introduction was evaluated based on the expression of fluorescence proteins. Accordingly, we observed that in the bombarded leaves of all examined species, fluorescent proteins were expressed under the control of each of the three constitutive promoters (Fig. 1), whereas no fluorescent proteins were detected in leaves containing the pollen-specific LAT52 promoter (Table 1). In the case of bombarded pollen, AtUBQ10 and AtRPS5A promoters drove H2B-fused fluorescent protein expression in the nuclei of all examined species (Fig. 1). However, in the pollen cytoplasm, the AtRPS5A promoter did not drive the expression of fluorescent protein in N. benthamiana and N. tabacum (Table 1, Fig. 1). These differences in fluorescent protein expression were thought to be owing to differences in promoter activity among species. Similarly, the 35S promoter was active in the pollen of N. benthamiana and N. tabacum, but not in that of torenia or tomato (Table 1). A promoter that controls constitutive gene expression in generative cells is most suitable, as these cells contain the genome that is transmitted to the next generation. Accordingly, we used the AtUBQ10 promoter to drive Cas9 in pollen transformed via particle bombardment. Furthermore, we used tobacco pollen in subsequent bombardment assays, as large amounts of pollen could be obtained from these plants.
Efficiency of gene delivery into generative cells by particle bombardment
In general, the efficiency with which genes are delivered by particle bombardment is typically as low as several percent (Wang and Jiang 2011). Given that genome editing occurs in only a fraction of the bombarded pollen, we initially investigated the efficiency with which genes were delivered into pollen. Plasmids carrying fluorescent protein-coding sequences driven under the control of the aforementioned promoters were introduced into the pollen of N. benthamiana by particle bombardment. We simultaneously introduced two types of plasmid DNA encoding mApple and H2B-mClover into the same pollen; accordingly, we observed mApple signals in the cytoplasm, whereas mClover signals were localized in the nuclei, indicating that subcellular structures in pollen and pollen tubes were labeled transiently (Fig. 2a). Staining of the nuclei of the wild-type pollen indicated a spindle-shaped generative cell (arrowhead in Fig. 2b) and vegetative nucleus (arrow in Fig. 2b). We also produced Agrobacterium-mediated transgenic N. benthamiana expressing H2B-mClover under the control of the AtUBQ10 promoter and detected the corresponding H2B-mClover signals in the nuclei of both vegetative and generative cells (Fig. 2c). When plasmid DNAs were introduced into pollen vegetative cells, only the vegetative nucleus was labeled (Fig. 2d), whereas when they were also introduced into generative cells, both nuclei were labeled (Fig. 2e). After pollen germination, the vegetative cell nucleus was localized in the tip, and the generative cell nucleus was located in the stained pollen tube (Fig. 2f). Additionally, the observed expression of delivered genes in the vegetative and generative cells of the pollen tubes indicated that the gold particles had penetrated the generative cell enclosed within the pollen (Fig. 2g). Interestingly, in some pollen tubes, we observed dot-like structures, which were assumed to be 0.6-μm gold particles (Fig. 2h). Furthermore, following bombardment, some pollen tubes were found to contain two sperm cells that were divided from the generative cell (Fig. 2i). Although nuclear signals of both vegetative and sperm cells were observed, fluorescence signals derived from the delivered plasmid DNA were detected only in the vegetative cells, indicating that gold particles were introduced only into the vegetative cells of pollen and that generative cell division occurred to produce two sperm cells.
To obtain an estimate of the efficiency with which genes are delivered into the vegetative and generative cells of pollen, we subsequently determined the ratio of transformed pollen grains to total pollen grains. The estimated percentages of the vegetative cells of transformed pollen showing expression of mApple in the cytoplasm and mClover in vegetative nuclei were 1.2% (18.3 ± 6.5 pollen grains) and 1.1% (16.0 ± 4.7 pollen grains), respectively (1479 ± 298.5 pollen grains per experiment, n = 3). This indicated that there was little difference in the efficiency of gene delivery and expression regardless of protein localization (i.e., cytoplasm or nucleus). However, we estimated that only 0.2% of the transformed pollen showed mClover signals in generative cell nuclei, and in such pollen, both vegetative and generative nuclei were labeled (3.5 ± 1.0 pollen grains). This indicates that approximately one-sixth of the delivered pollen was also introduced into the generative cells. Based on these observations, we concluded that particle bombardment is applicable for gene delivery into vegetative and generative cells.
Genome editing in pollen can be induced by exogenous CRISPR/Cas9 components
To facilitate CRISPR–Cas9-mediated genome editing, the introduced gene must initially be transcribed and translated, which results in a time lag before the effects can be detected. Therefore, we performed time-lapse imaging of the bombarded pollen to estimate the length of time required prior to the detection of transient expression of introduced DNA in vitro. We found that fluorescent signals of mClover encoded by the AtUBQ10p::H2B-mClover plasmid (sSNv28; Table S1) appeared at 2.5–3.5 h after bombardment, whereas those of tdTomato encoded by the AtUBQ10p::tdTomato plasmid (sSNv26; Table S1) appeared at 3.5–4.5 h (Fig. 3a, Video S1). In this regard, it has been established that pollen tube germination in N. benthamiana commences after 0.5–5 h on agarose media and thereafter grows for approximately 24 h (Paungfoo-Lonhienne et al. 2010). Accordingly, we decided to analyze CRISPR/Cas9 plasmid-mediated genome editing in N. benthamiana pollen 20–24 h post-bombardment. To investigate whether the biolistically delivered CRISPR/Cas9 system is functional in pollen, we introduced a plasmid DNA consisting of a human codon-optimized Cas9 gene and an A. thaliana U6.26 promoter (AtU6.26)-driven sgRNA cassette (Tsutsui and Higashiyama 2017). The functioning of the AtU6.26 promoter and sgRNA in N. benthamiana leaves was confirmed by transient expression via agro-infiltration (Nekrasov et al. 2013). We initially introduced the AtUBQ10p::Cas9/U6.26p::NbPDS3-sgRNA plasmid (sSNv21; Table S1) into N. benthamiana leaves by particle bombardment and then identified different mutation patterns based on sequence analysis (Fig. 3b–d and Fig. S1). Analysis of the sequences obtained from 116 clones derived from the PCR product shown in lanes 3 and 6 of the gel depicted in Fig. 3b revealed the presence of indels in 17 of these clones. These mutations could be grouped into five different types: 9-, 4-, and 2-bp deletions or 1-bp insertions of either T or A (Fig. 3d). The same plasmid was subsequently delivered into N. benthamiana pollen. Bulk pollen tubes with and without fluorescent signals (Fig. 3c) were collected, and genomic DNA was extracted. Sequence analysis of 168 clones derived from the PCR product revealed the presence of deletions or substitutions in 33 of these clones, which could be grouped into three different types: 1-bp deletion or substitutions of either C to T or T to C (Fig. 3d). No mutations were detected in genomic DNA extracted from pollen tubes bombarded without plasmid DNA. These results indicate that the transient expression of plasmid DNA, including CRISPR/Cas9, delivered via particle bombardment, induces genome editing in the leaves and pollen of N. benthamiana.
Visualization of fertilization process of bombarded pollen in vivo
To investigate whether bombarded pollen retains fertilization capacity in vivo, we used a nuclear-labeled marker line to investigate pollen tube germination and the delivery of generative cells. For bombardment, we used pollen derived from a UBQ10p::H2B-mClover transformant into which plasmid DNA containing the UBQ10p:H2B-tdTomato sequence (DKv277; Table S1) was introduced. Thereafter, we performed time-lapse imaging of pollen tube elongation for 18 h after bombardment. Time-lapse observation revealed the delivery of bombarded generative cells within the elongating pollen tube (Fig. 4a, Video S2). In some species, pollen is unable to germinate on the stigma when the pollen has initially been hydrated on the medium (Zuberi and Dickinson 1985). Thus, we investigated whether N. benthamiana and N. tabacum pollen that had been hydrated on the germination medium could germinate on the pistil. Following hydration, pollen was immediately used to pollinate the emasculated stigmas within 15 min. The pollinated pistils were then fixed and stained with aniline blue solution, and observations indicated that the hydrated pollen of both N. benthamiana and N. tabacum germinates on the respective pistils in vivo (Fig. 4b, c). To enable direct observations of bombarded pollen tube elongation, we conducted a semi-in vivo pollen tube growth assay (Palanivelu and Preuss 2006). Owing to the thin style, the pistils of N. benthamiana are difficult to dissect without causing damage; therefore, we used N. tabacum, which has thick hard pistils, for this assay. Wild-type pollen bombarded with a mixture of plasmid DNAs encoding fluorescent proteins driven by AtUBQ10 or 35S promoters was immediately used to pollinate emasculated pistils, and pollen tubes were observed, including those derived from bombarded pollen, emerging from the end of the cut style (Fig. 4d). Moreover, we found that the vegetative nucleus in the pollen tube was labeled and that the lengths of the emerging pollen tubes derived from bombarded pollen were similar to those of other pollen tubes that had germinated from non-bombarded pollen. Furthermore, the fertilization process of the bombarded pollen after pollen tube germination was observed in vivo (Fig. 5). The emasculated wild-type N. tabacum pistil was pollinated with bombarded pollen with plasmid DNA, including AtUBQ10p::sGFP and 35Sp::H2B-tdTomato sequences. Dissection of pistils at 24 h after pollination (Fig. 5a) revealed that the pollen tubes derived from bombarded pollen had elongated within the longitudinal section of the style in vivo (Fig. 5b). Moreover, we observed that some pollen tubes underwent cell division, resulting in the production of two sperm cells and a single vegetative cell within the pollen tube (black arrows in Fig. 5b). Additionally, the pistils were dissected 48 h after pollination. We observed that bombarded pollen elongated on the placenta in the ovary (Fig. 5c, d) and that some ovules had received pollen tubes derived from bombarded pollen grains 48 h after pollination (Fig. 5e). When bombardment was performed twice and two flowers were pollinated each, one, two, three, and five ovules showed fluorescence signals in each ovary at 48 h after pollination. Fluorescence signals derived from pollen tube cytoplasm reporters were detected as a large spot of fluorescence signals inside the ovules (Fig. 5f). Furthermore, some of the enlarged ovules that seemed to be fertilized remained fluorescent 72 h after pollination. The fluorescence of the cylindrical green sGFP and spot-shaped red H2B-tdTomato signals was observed inside an enlarged ovule (Fig. 5g). Interestingly, two red H2B-tdTomato fluorescence signals were occasionally observed inside ovules 72 h after pollination, which were suggested as the multiple nuclei derived from bombarded pollen, embryo, or endosperm (Fig. 5h). This indicated that the pollen tubes released their contents inside the ovule, which were detected as a convenient marker of targeted ovules (Palanivelu and Preuss 2006). These observations indicate that even in vivo, bombarded pollen germinates and gives rise to pollen tubes that undergo subsequent elongation and deliver their contents, including sperm cells, into the ovule.