Inositol polyphosphate 5-phosphatase-controlled Ins(1,4,5)P₃/Ca²⁺ is crucial for maintaining pollen dormancy and regulating early germination of pollen

Pollen dormancy and germination are crucial for reproductive growth of flowering plants. In Arabidopsis, mature pollen grains are released from anthers in a partially dehydrated and dormant state. Once they fall onto stigma, rehydration of desiccated pollen grains breaks dormancy and turns on pollen germination (McCormick, 2004). Inappropriate pollen germination will inevitably affect plant fertility, and thus studies on pollen dormancy and germination are of great significance to agricultural production. Precocious pollen germination in anthers was first observed in the Arabidopsis gametophytic mutant raring-to-go (rtg), in which pollen grain development is affected as early as the bicellular stage and affected pollen grains are unable to desiccate completely (Johnson and McCormick, 2001). In addition, rtg pollen tubes were dramatically elongated within anthers under high humidity conditions. These findings indicate the importance of dehydration/rehydration in pollen dormancy and germination; to date, dehydration is considered to be the main factor in maintaining pollen dormancy and preventing precocious germination.

Abscisic acid (ABA) promotes and maintains seed dormancy (Finkelstein et al., 2008). Deficiencies in ABA synthesis or signaling frequently result in reduced dormancy, seed precocious germination or vivipary (Koornneef and Karssen, 1994; McCarty, 1995; Bewley, 1997). Supportive evidence comes from the studies of maize vp (viviparous) (Neill et al., 1986), tomato sitiens (Hilhorst and Downie, 1996), rice phs (pre-harvest sprouting) (Fang et al., 2008), and Arabidopsis ABA-deficient (aba) and ABA-insensitive (abi) mutants (Koornneef et al., 1989). Moreover, recent studies show that the ABA content in the lily pollen is increased before anthesis, which correlates with the pollen dehydration stage (Hsu et al., 2010). A unique desiccation-associated ABA signaling transduction has been reported, through which a Rop (Rho GTPase of plants) gene is regulated during the stage of pollen dehydration (Hsu et al., 2010). These results indicate the importance of ABA-mediated dehydration for pollen dormancy.

Both hydration/humidity and Ca²⁺ are essential for pollen germination. In Arabidopsis, pollen germination starts with the rehydration of pollen grains on the stigma. During this process, Ca²⁺ flows into the pollen grains and triggers cytoplasmic reorganization. Subsequently, a cytoplasmic gradient of Ca²⁺ is formed with the highest [Ca²⁺] at the germination site, which is crucial for polar tip growth of prospective pollen tubes (Heslop-Harrison and Heslop-Harrison, 1992a; Heslop-Harrison and Heslop-Harrison, 1992b; Franklin-Tong, 1999). Dynamic analysis of cytosolic [Ca²⁺] ([Ca²⁺]_cyt) in Arabidopsis pollen grains reveals an increase in [Ca²⁺]_cyt at the potential germination site soon after rehydration; this high Ca²⁺ concentration is maintained until the onset of pollen germination (Iwano et al., 2004). In hydrated pollen of Nicotiana tabacum L., a high level of calmodulin (CaM) is consistently present in the region of the germination apertures and is associated with the plasma membrane of the germination bubble and with the onset of germination (Tirlapur et al., 1994). High environmental humidity and exogenous Ca²⁺ are also requisites for pollen germination in vitro (Taylor et al., 1997).

Inositol polyphosphate 5-phosphatase (5PT), a key enzyme of the phosphatidylinositol (PI) signaling pathway, functions in growth, development and responses to stress in plant, yeast and mammals (Stevenson et al., 2000; Astle et al., 2006; Xue et al., 2007; Xue et al., 2009). 5PT removes the phosphate at the 5-position of the inositol ring of specifically phosphorylated (5-position) phosphoinositides [PtdIns(4,5)P₂, PtdIns(3,4,5)P₃ and/or PtdIns(3,5)P₂], and inositol 1,4,5-trisphosphate [Ins(1,4,5)P₃] or inositol 1,3,4,5-tetrakisphosphate [Ins(1,3,4,5)P₄].

There are 15 5PTs in Arabidopsis (Berdy et al., 2001; Ercetin and Gillaspy, 2004; Zhong and Ye, 2004) that have diverse roles in plant growth and development, and phytohormone signaling. 5PT13 modulates cotyledon vein development through the regulation of auxin homeostasis (Lin et al., 2005). FRAGILE FIBER 3 (FRA3) and COTYLEDON VASCULAR PATTERN 2 (CVP2) are two 5PTs containing the 5PTase domain. FRA3 is required for secondary wall synthesis and actin organization in fiber cells (Zhong et al., 2004), and CVP2-mediated Ins(1,4,5)P₃ signal transduction is essential for closed venation patterns of Arabidopsis foliar organs (Carland and Nelson, 2004). In addition, 5PTs participate in seedling growth (Gunesekera et al., 2007; Ercetin et al., 2008) and blue-light signaling via the regulation of Ins(1,4,5)P₃/[Ca²⁺]_cyt (Chen et al., 2008), and are important regulators of Ins(1,4,5)P₃-mediated ABA signaling (Sanchez et al., 2001; Burnette et al., 2003; Lin et al., 2005; Gunesekera et al., 2007). However, no 5PTs have been reported to function in the regulation of reproductive development.

Here, through the functional characterization of Arabidopsis 5PT12 and 5PT13, we show that the increase of 5PT-controlled Ins(1,4,5)P₃/Ca²⁺ is sufficient for breaking pollen dormancy and triggering early germination, and further confirm that, independently of dehydration, 5PT-controlled Ins(1,4,5)P₃/Ca²⁺ is crucial for pollen dormancy.

The Arabidopsis Columbia ecotype was used. The 5pt13 mutant has been described by Lin et al. (Lin et al., 2005). Seedlings were grown in a phytotron at 22°C with a 16-hour light/8-hour dark photoperiod.

The 5pt12 mutant was obtained from the Arabidopsis Biological Resource Center (SALK_065920, http://www.arabidopsis.org/). 5pt12 mutant seeds were screened on MS medium containing 30 mg/l kanamycin for segregation ratio analysis and a PCR-based approach was employed to confirm the T-DNA insertion and to identify the homozygous mutant. Genotyping primers 5PT12-F (5′-CACAGCCTCGACGACATTC-3′), 5PT12-R (5′-TTGCCAAGCTAGACCTTCC-3′) and LBa1 (5′-TGGTTCACGTAGTGGG CCATCG-3′) were used.

The 5pt14 mutant was obtained from the Arabidopsis Biological Resource Center (GK-309D03, http://www.arabidopsis.org/). Genotyping primers 5PT14-F (5′-GGTGGTGTAAGAGGATGGT-3′), 5PT14-R (5′-CGGCACTTCCTTCAACCC-3′) and GK-LB (5′-ATATTGACCATCATACTCATTGC-3′) were used to confirm the T-DNA insertion and to identify the homozygous mutant.

qRT-PCR was performed to test the expression patterns of 5PT12, 5PT13 and 5PT14 in various tissues, the transcription levels of 5PT12 and 5PT14 in mutant or wild-type plants, and the expression of ABA-responsive genes in seedlings or flower tissues.

For analyzing the expression pattern of 5PT12, total RNAs were extracted from 4-day-old seedlings, roots, 4-week-old rosette leaves, flowers and stems. For analyzing the expression pattern of 5PT12, 5PT13 and 5PT14 at different floral stages, total RNAs were extracted from flowers at stages 12, 13 and 14. For testing the transcriptional levels of 5PT12 and 5PT14 in mutant or wild-type plants, total RNAs were extracted from flowers of 4-week-old plants. For examining the expressions of ABA-responsive genes in flower tissues or seedlings, flowers of 4-week-old plants at floral stages 12-14 or an earlier stage were collected and submerged in ABA (100 μM) solution for 0, 1 or 3 hours, and 8-day-old wild-type and 5pt12 seedlings were submerged in ABA (100 μM) solution for 0, 0.5, 1 or 2 hours.

Total RNAs were extracted using Trizol reagent (Invitrogen) and then reversely transcribed according to the manufacturer’s instructions (ReverTra Ace kit, Toyobo). RNA samples were incubated for 30 minutes at 37°C with RNase-free DNase I (Takara), and then precipitated for further analysis.

RotorGene RG-3000A (Corbett Research) and a SYBR green detection kit (SYBR Green Realtime PCR Master Mix kit; TOYOBO) were used, with a program of denaturation (5 minutes, 95°C), 40 cycles of denaturation (95°C, 10 seconds), annealing (55°C, 15 seconds) and elongation (72°C, 20 seconds). Fluorescence is monitored during each annealing and denaturation phase. The product amounts were determined by the method of comparative Delta-Delta Ct (Fleige et al., 2006). Primers used were as follows: 5PT12 (5′-AGAATCGGTAGGAGGAAACG-3′ and 5′-TTCTCAGATTCTTCCTCACC-3′), 5PT13 (5′-CCGAGGACAAAATCTAAGTAACG-3′ and 5′-CAGCGCCGGTGCTTGGAATAG-3′), 5PT14 (5′-GGTGGTGTAAGAGGATGGT-3′ and 5′-CGGCACTTCCTTCAACCC-3′), ABA-responsive genes Rd22 (5′-TTACCAAACACTCCCATTCC-3′ and 5′-CGTACACCTCCCTTTCCAAC-3′), COR47 (5′-GAACAAGCCTAGTGTCATCG-3′ and 5′-TCTTGTCGTGGTGTCCTGG-3′), KIN2 (5′-ACCAACAAGAATGCCTTCCA-3′ and 5′-ACTGCCGCATCCGATATACT-3′) and COR15 (5′-CTCAGTTCGTCGTCGTTTC-3′ and 5′-CATCTGCTAATGCCTCTT T-3′). Arabidopsis ACTIN gene (At5g09810) was amplified with primers (5′-CCGGTATTGTGCTCGATTCTG-3′ and 5′-TTCCCGTTCTGCGGTAGTGG-3′) and used as positive internal control. Primers of 5PT13 have been previously described (Chen et al., 2008). All the experiments were repeated three times.

The genomic DNA region in front of the translation initiation site ATG was PCR amplified with primers 5PT12p-1 (5′-CCCAAGCTTTAAGGCAGTGTCTGTCGAGG-3′, added HindIII site underlined) and 5PT12p-2 (5′-CGGGATCCGGAAGAGAGAATGAGAGAAATG-3′, added BamHI site underlined). The amplified 1216 bp DNA fragment was confirmed by sequencing and then subcloned into modified pCAMBIA1301 (Liu et al., 2003), resulting in the construct containing 5PT12 promoter-GUS (β-glucuronidase) fusion. The resultant construct was transferred into Agrobacterium tumefaciens strain GV3101 and then transformed into Arabidopsis using the floral dip method. T1 transgenic seedlings were screened on MS medium supplemented with 30 mg/l hygromycin and the T-DNA integration was confirmed by PCR analysis using primers annealing to the hygromycin resistance gene (hyg-s, 5′-GCTTCTGCGGGCGATTTGTGT-3′ and hyg-a, 5′-GGTCGCGGAGGCTATGGATGC-3′). Seedlings of T3 homozygous plants were analyzed for GUS activities and observed using a SMZ 800 stereoscope (Nikon).

The transgenic seedling harboring a 5PT13 promoter-GUS reporter gene construct has been described previously (Lin et al., 2005).

A 7.4 kb fragment of genomic DNA containing the full-length 5PT12 gene and the promoter region was cut from a BAC clone F6E13 (from ABRC, http://www.arabidopsis.org) and subcloned into modified pCAMBIA1301 vector (Liu et al., 2003). The resultant construct was transferred into the 5pt12 mutant by Agrobacterium-mediated transformation through the floral dipping method. Complemented expression of 5PT12 was confirmed through qRT-PCR using primers (5′-AGTACTTCTCCCACCACTTG-3′ and 5′-TCTGAACATGACCACAAATCG-3′). T2 homozygous plants were used for complementation studies.

To generate 5pt12 5pt13 double mutant by genetic cross, 5pt13 homozygotes (carrying Basta-resistant gene) were used as female parent, and 5pt12 homozygotes (carrying kanamycin-resistant gene) were used as male parent. The cross was carried out by removing the petals, sepals and androecia from large green buds in 5pt13 plants, followed by artificial fecundation with the 5pt12 pollen grains on the next day. F1 cross progenies were selected on kanamycin-containing MS plates and positive plants were used for harvesting F2 seeds. F2 generation plants were used for PCR analysis to confirm the T-DNA insertions in 5PT12 gene loci and 5PT13 gene loci. Identified homozygotes were grown for one additional generation and harvested seeds were used for further experiments.

Surface-sterilized seeds of wild-type or 5pt12 plants were sown onto MS medium containing different concentrations of ABA (0, 0.1, 0.3 and 0.5 μM). Following stratification in 4°C for 2 days, seeds were grown at 22°C for 7 days and germination rates were calculated. One-hundred seeds were used for each line, and all the assays were performed in triplicate. Assays were repeated at least twice.

Dehiscent anthers of wild-type and 5pt12 plants were harvested and fixed by FAA, followed by dehydration with a graded ethanol series. Specimens were dried at crucial points and coated with gold before examination with a Scanning Electron Microscope (JSM-6360LV, Tokyo, Japan).

Dehydrated pollen grains were stripped out from dehiscent anthers of wild-type and 5pt12 plants and covered with type N immersion oil to retain their original shapes. Image J software was used to measure major and minor axes of dehydrated pollen grains, and the ratio of minor axes/major axes was calculated. More than 50 pollen grains of each genotype were analyzed, and the assays were repeated twice.

Flower buds prior to opening were harvested, and dispersed pollen grains were treated with FDA (fluorescein diacetate, diluted with acetone to a concentration of 100 mg/l) for 30 minutes, and then observed in ultraviolet light with a microscope (Leica).

Pollen harvested from open flowers was treated with DAPI solution (final concentration of 2 mg/l dissolved in 7% sucrose, w/v) for 5 minutes and imaged with a confocal laser scanning microscope (FITC488, Zeiss LSM500).

Open flowers from wild-type and 5pt12 mutant plants were dehydrated in a desiccator at room temperature for at least 2 hours. Pollen germination was performed as described by Li et al. (Li et al., 1999) and 2 mM Ca²⁺ (with an equal molar ratio of CaCl₂ and Ca[NO₃]₂) was added into basal medium to obtain the optimal pollen germination rate. Soaked filter paper was used to provide environment humidity. After incubation for 6 hours at 25°C, germinated pollen were observed and photographed.

To examine the effect of extracellular Ca²⁺ on pollen germination, pollen was cultured in basal agar medium supplemented with different concentrations of Ca²⁺ [with an equal molar ratio of CaCl₂ and Ca(NO₃)₂]. For the humidity assay, dry filter paper was used to provide a coarse gradient of environmental humidity conditions by adding different volumes of distilled water onto filter papers. 0, 2.3, 4.6 or 7 ml of distilled water was added to five pieces of dry filter paper to provide very low (∼0-10%), low (∼25-35%), middle (∼55-65%) or saturation (high humidity, ∼85-95%). Pollen grains (>200) collected from more than three individual plants were used for each genotype, and each assay was performed in triplicate. Assays were repeated at least twice.

Different concentrations of Ca²⁺ [with an equal molar ratio of CaCl₂ and Ca(NO₃)₂] were applied to anthers at floral stage 11 to examine the pollen germination in anthers. BAPTA (2 mM, Sigma-Aldrich) was used to block the cytosolic Ca²⁺ in anthers of 5pt12, 5pt13 and 5pt12 5pt13 mutants. At late floral stage 12, Ca²⁺- or BAPTA-treated anthers were harvested for pollen tube staining by DAB assay.

Dehiscent anthers or dispersed pollen grains from wild-type and 5pt12, 5pt13, 5pt12 5pt13, 5pt14 and 5pt12 plants with complementary expression of 5PT12 were treated with 0.1% decolorized Aniline Blue (DAB) for 5 minutes followed by observation in ultraviolet light with a microscope (Leica).

Anthers at floral stage 12 or anthers at earlier stage (pollen were too small to collect) were collected for ABA measurement using liquid chromatography-mass spectrometry (LC-MS) (Welsch et al., 2008).

Stamens at floral stages 12-14 from wild-type, 5pt12, 5pt13, 5pt12 5pt13 and 5pt12 plants with complementary expression of 5PT12 were harvested for measuring Ins(1,4,5)P₃ content. Extraction of Ins(1,4,5)P₃ was performed according to a previously published method (Burnette et al., 2003). Ins(1,4,5)P₃ content was measured using the D-myo-inositol-1,4,5-trisphosphate [³H] assay kit (GE healthcare) and determined from a standard curve generated with known amounts of Ins(1,4,5)P₃.

For inositol phosphates (IPs) control assay, flower tissues at floral stages 12-14 of wild-type, 5pt12, 5pt13, and 5pt12 5pt13 plants were collected and used for the assay. A mixture of unlabelled PtdIns(4,5)P₂, Ins(1,3,4)P₃ and InsP₆ were added into assays. As carried out previously (Perera et al., 2008), PtdIns(4,5)P₂, InsP₆ and Ins(1,3,4)P₃ were added, respectively, at 18×, 10× and 1× the concentration of Ins(1,4,5)P₃ found in wild-type stamen tissues. Assays were performed in triplicate and experiments were repeated twice.

Anthers at floral stages 11-13 from wild-type, 5pt12, 5pt13 and 5pt12 5pt13 plants were harvested, following addition of 1 ml 100% HNO₃ and kept overnight. The samples were kept at 99°C for 1 hour and diluted to 14 ml with ultra-pure water. Samples were analyzed using an ELAN DRC-e ICP-MS system (PerkinElmer) and Multi-Element calibration Standard-2A (Agilent) samples were used as standards. Assays were performed in triplicate.

Microspores or pollen grains at different floral stages from wild-type, 5pt12, 5pt13, and 5pt12 5pt13 plants were stained with 20 μM Indo-1 (57180, Sigma) on slides in the dark. Two hours after loading and washing of the dye, fluorescence was observed with a confocal SMZ510 under UV light excitation (364-nm excision, and 400- to 435-nm and 480 nm emission) and imaged. [Ca²⁺]_cyt was pseudocolored according to the scale. Pseudocolor ratio images of the [Ca²⁺]_cyt distribution was calculated with Olympus Fluoview Ver. 1.7a Viewer tool (Olympus) to measure the fluorescence intensity. White indicates approximately twice the [Ca²⁺]_cyt of that indicated by the blue and 1.5 times the [Ca²⁺]_cyt indicated by the yellow. For the root hair plasmolysis, 7-day-old seedlings of wild-type, 5pt12, 5pt13, and 5pt12 5pt13 plants were stained with Indo-1, followed by treatment with 500 mM mannitol for 10 minutes.

A pollen protoplast staining assay was performed as described previously (Wu et al., 2011). Pollen grains at different floral stages from wild-type, 5pt12, 5pt13, and 5pt12 5pt13 plants were harvested for the assay.

Sequence data from this article can be found in the GenBank/EMBL data libraries under the following accession numbers: 5PT12 (At2g43900), 5PT13 (At1g05630), 5PT14 (At2g31830), KIN2 (At5g15970), COR15 (At2g42540), RD22 (At5g25610) and COR47 (At1g20440).

Our previous studies have shown that inositol polyphosphate 5-phosphatase 13 (5PT13) plays crucial roles in auxin-regulated cotyledon vein development (Lin et al., 2005), blue light-mediated photomorphogenesis (Chen et al., 2008) and root gravitropism (Wang et al., 2009). To further characterize the 5PT functions in Arabidopsis, we focused on 5PT12, a close paralog of 5PT13 (because the mutant of the closest paralog, 5PT14, was not available when we started). Based on domain prediction, 5PT12 has a conserved 5-phosphatase domain (5PTase domain), which is required for 5-phosphatase activity towards inositol phosphates and phosphoinositides, and three WD40 repeats (supplementary material Fig. S1A). Phylogenetic analysis of 15 Arabidopsis 5PTs and representative yeast and mammalian 5PTs showed that 5PT12 is most closely related to 5PT13, 5PT14 and FRA3 (supplementary material Fig. S1B).

The gene expression pattern of 5PT12 in various tissues was analyzed using quantitative real-time RT-PCR (qRT-PCR) and the results showed that 5PT12 was mostly expressed in leaves and flowers, whereas it was only weakly expressed in roots, stem and young seedlings (Fig. 1A). To investigate the detailed expression pattern, a construct containing the 1.2 kb 5PT12 promoter region fused to the Escherichia coli β-glucuronidase (GUS)-coding gene was generated and transformed into Arabidopsis. Consistent results from five independent transgenic lines showed that 5PT12 is transcribed in cotyledon tips and the hydathode of leaves (Fig. 1Ba,b).

Interestingly, 5PT12 is preferentially expressed in pollen grains at floral stages (Smyth et al., 1990; Sanders et al., 1999) 12 to 14 (Fig. 1Bc-f). At stage 13, the filaments have elongated to the height of the stigma, mature pollen and pollen tubes (Fig. 1Bg,h), suggesting a potential role for 5PT12 in pollen development and germination. qRT-PCR analysis indicated the gradually increased expression of 5PT12 as floral development proceeded (from floral stage 12 to 14, Fig. 1C), which is consistent with the microarray data (https://www.genevestigator.ethz.ch/) showing that the 5PT12 gene is largely expressed in anthers, with high transcript abundance in tri-cellular pollen and mature pollen, and low abundance in uni-cellular microspore and bi-cellular pollen. Furthermore, 5PT12, 5PT13, 5PT14 and 5PT5 are also highly expressed in pollen grains, based on the microarray data (supplementary material Table S1; qRT-PCR analysis confirmed relatively less expression of 5PT13 in comparison with 5PT12, Fig. 1C). As the 5pt14 mutant was not available and 5PT5 is not a close paralog to 5PT12, 5PT13 and 5PT14, we first focused on the function of 5PT12 in pollen development.

A putative T-DNA insertion mutant of 5PT12 was obtained from SALK, and PCR analysis confirmed the T-DNA insertion at the first exon of the 5PT12 (Fig. 1E). The homozygote was characterized and designated as a 5pt12 mutant. Both forward and reverse gene-specific genotyping primers could amplify the PCR product when combined with T-DNA border primer LBa1 (Fig. 1E), indicating that two tandem T-DNA insertions are located back-to-back in 5PT12. This was further confirmed by the corresponding T-DNA flanking sequences. Back-cross between 5pt12 and wild-type plants was performed (supplementary material Fig. S2A) and segregation analysis of F2 progenies of a single 5PT12/5pt12 F1 plants by PCR amplification of NPTII (selectable marker located in T-DNA for resistance to kanamycin) showed that 30 out of 41 progenies carried NPTII (supplementary material Fig. S2B). In addition, scoring for kanamycin (Kan) resistance showed that 212 F2 seedlings were Kan resistant and 72 were Kan sensitive. A chi-square test for both assays confirmed the 3:1 segregation ratio. These results validated that the two T-DNA insertions are closely linked without segregation. Transcription analysis by qRT-PCR showed that the 5PT12 gene was completely disrupted (Fig. 1F), indicating that 5pt12 is a knockout mutant.

Owing to the predominant expression of 5PT12 in pollen grains at floral stages 12-14, in mature pollen and in pollen tubes, the function of 5PT12 in pollen development and germination was specifically examined. Phenotypic observation showed that, unlike wild-type dehiscent anthers, in which very few pollen grains remained (floral stage 13), many more pollen grains were left in the 5pt12 dehiscent anthers, and they had tube-like structures (Fig. 1G). Pollen grains were washed off and tube-like structures could be easily distinguished (Fig. 1H). Further staining with decolored Aniline Blue (DAB), a widely used dye for pollen tube observations (Escobar-Restrepo et al., 2007), confirmed that the tube-like structures are indeed pollen tubes (Fig. 1I).

The precocious pollen germination rates of 5PT12/5pt12 heterozygotes were 20-30% and those of 5pt12 homozygotes reached 40-70% in comparison with the rate of 1-3% in wild-type anthers (Fig. 1J), indicating that 5PT12 deficiency resulted in precocious pollen germination within the anther. This is a very interesting finding, because, in Arabidopsis, only two mutants (raring-to-go and callose synthase 9) and one transgenic line overexpressing callose synthase 5 have been reported so far to show a similar phenotype with unknown mechanism (Johnson and McCormick, 2001; Xie et al., 2010).

The link between T-DNA insertion in 5PT12 gene and the precocious germination phenotype was further tested using randomly selected 40 progenies of one single 5pt12 plant. PCR analysis using the genotyping primers confirmed that all the 40 progenies were homozygotes (supplementary material Fig. S2C) and all exhibited precocious pollen germination at a rate of 40-70%. Complementation studies were performed by transferring a 7.4 kb fragment of genomic DNA containing the entire 5PT12 gene and promoter region into the 5pt12 mutant (Fig. 1F). The results showed that the Aniline Blue fluorescence was barely detectable in complementation lines (Fig. 1I) and the precocious pollen germination rates of complementation lines (T2 homozygous lines) were also similar to those in wild type (Fig. 1J), confirming that 5PT12 deficiency is responsible for precocious pollen germination.

To investigate whether there are developmental defects in 5pt12 that cause precocious pollen germination, pollen development of 5pt12 was examined. The observations showed the normal development of 5pt12 pollen, and the DAPI staining assay revealed that precociously germinated pollen grains within 5pt12 anthers have regular tri-cellular structures, similar to wild-type pollen grains (Fig. 2A, left panel). Further analysis of pollen viability with FDA staining showed that 5pt12 pollen grains were indistinguishable from wild type (Fig. 2A, right panel). Besides, mature pollen grains of 5pt12 that did not germinate precociously within the anther could germinate well in vitro (Fig. 2B), and the seed set rate of 5pt12 was not affected, despite the precocious pollen germination (at the rate of 40-70%, Fig. 2C). All these results confirm the normal development of 5pt12 pollen grains.

Previous studies have demonstrated the significance of dehydration, which is probably mediated by ABA signaling, in pollen dormancy and germination (Hsu et al., 2010). Our observations showed that precociously germinated pollen grains in 5pt12 anthers keep their dehydrated shape (Fig. 3A). The axis lengths of pollen grains from dehiscent anthers were further measured and there was no statistical difference in the ratio of minor axis:major axis between wild-type and 5pt12 pollen, suggesting that the mature pollen grains of 5pt12, including the precocious germinated ones, undergo normal dehydration process.

Deficiencies in ABA synthesis or signaling lead to seed precocious germination or vivipary, and pollen dehydration is probably mediated by ABA signaling. As disruption or overexpression of several characterized Arabidopsis 5PTs has been reported to alter ABA responses, we therefore tested whether 5pt12 has altered ABA content or ABA signaling. Wild-type or 5pt12 anthers at different floral stages were collected for ABA measurements. There was no significant difference in ABA content between wild type and 5pt12 either at floral stage 12 or at earlier stages (Fig. 3B). Further analysis of the ABA-induced expression of ABA-responsive genes (KIN2, COR15, RD22 and COR47) in seedlings showed a much enhanced induction of these genes under 5PT12 deficiency (Fig. 3C), indicating stimulated ABA signaling in 5pt12 mutant. Consistently, the 5pt12 mutant was more sensitive to exogenous ABA in seed germination assays (Fig. 3D). The expression patterns of ABA-responsive genes in floral tissues were also tested. However, ABA treatment could not induce uniform and consistent expression of all the tested ABA-responsive genes, although some expression was induced at specific floral stages (supplementary material Fig. S3). The reason might be that ABA cannot effectively penetrate into the pollen grains owing to thick anther walls (no similar ABA treatment has been reported previously). Taken together, these results indicate unaffected ABA synthesis and stimulated ABA signaling in the 5pt12 mutant; thus, precocious pollen germination seen in 5pt12 mutants is not due to altered ABA responses.

We also examined the germination rate of pollen (at early floral stage 12) in response to different humidity conditions. The germination rate of wild-type stage 12 pollen grains increased by 2.8-fold (from 18% to 52%) as the environmental humidity rose, whereas that of 5pt12 early stage 12 pollen grains (at this time the pollen grains are not germinated) showed a moderate increase of 1.6-fold (from 46% to 76%, Fig. 3E). Compared with wild-type pollen grains, 5pt12 pollen grains showed a higher germination rate under low humidity conditions and a significantly higher germination percentage at high environmental humidity (76% versus 45% of wild type). This ruled out the possibility that the altered environmental humidity caused the precocious pollen germination in 5pt12 mutant.

Taken together, these results strongly suggested that the precocious pollen germination in the 5pt12 mutant is not caused by alterations in dehydration/rehydration or by the ABA response, and indicated the existence of another key regulator in pollen dormancy and germination, which is possibly triggered in the 5pt12 mutant. Ca²⁺ is the potential candidate, based on the fact that Ca²⁺ is crucial for pollen development and germination, and that Ins(1,4,5)P₃ stimulates the cytosolic Ca²⁺.

Based on microarray data (https://www.genevestigator.ethz.ch/), 5PT13, a close paralog of 5PT12, is weakly expressed in pollen grains. qPT-PCR analysis confirmed the expression of 5PT13 at floral stages 12 to 14 (Fig. 1C), with a similar pattern to that of 5PT12, and a detailed expression analysis by promoter-reporter gene fusion studies shows that the 5PT13 gene is largely expressed in pollen grains at floral stages 13 and 14, as well as in mature pollen and pollen tubes (Fig. 1D). Although the expression level of 5PT13 is lower than 5PT12 at floral stage 12 to 14, especially at floral stage 12, the 5PT13 knockout mutant (5pt13) also exhibits precocious pollen germination in dehiscent anthers (Fig. 4A,B) at a low incidence. Double mutant 5pt12 5pt13 has a much higher incidence of precocious pollen germination (∼60-70%) than 5pt12 or 5pt13 single mutants (Fig. 4C). In addition, abnormal pollen tubes were occasionally observed in precociously germinated pollen grains, including more than one pollen tube formed in one pollen grain or aberrant pollen tube structures (Fig. 4D).

Previous studies have shown that in vitro, both 5PT12 and 5PT13 exhibit phosphatase activity towards only Ins(1,4,5)P₃, rather than towards inositol lipids (Zhong and Ye, 2004). Thus, Ins(1,4,5)P₃ is suggested to be involved in the regulation of pollen dormancy and germination. Ins(1,4,5)P₃ receptor binding protein was employed specifically to determine the intrinsic Ins(1,4,5)P₃ content by using the D-myo-inositol-1,4,5-trisphosphate [³H] assay kit. The basal level of Ins(1,4,5)P₃ was significantly increased in 5pt12 anthers and was even more increased in 5pt12 5pt13 double mutants in comparison with wild type, while the complementary expression of 5PT12 restored Ins(1,4,5)P₃ levels (Fig. 5A). Basal Ins(1,4,5)P₃ levels corresponded to the precocious germination incidences in each genotype.

It has been reported that the basal level of InsP₂ or InsP₆ is higher than Ins(1,4,5)P₃ in Arabidopsis (Perera et al., 2008). To exclude the potential influence caused by non-specific binding, unlabelled PtdIns(4,5)P₂, InsP₆ and Ins(1,3,4)P₃ [Ins(1,4,5)P₃ stereoisomer] were added into the samples and the results showed that the addition of these inositol phosphates and phospholipids did not affect the measurement (Fig. 5B), consistent with the specificity described by the manufacturer (the cross-reactivity for other inositol phosphates is lower than 1%, http://www.gehealthcare.com/lifesciences). These results, thus, indicate a close correlation between Ins(1,4,5)P₃ content and precocious pollen germination, and indicate that Ins(1,4,5)P₃ is involved in controlling pollen grain dormancy in anthers.

As Ca²⁺ is crucial for pollen germination on stigma or in vitro, total Ca²⁺ contents in anthers of wild type, 5pt12, 5pt13 and 5pt12 5pt13 double mutant were measured using ICP-MS (inductively coupled plasma mass spectrometric analysis) (Chen et al., 2003). The results showed that the Ca²⁺content in 5pt12 mutant anthers (stage 13) was higher than that in wild-type anthers, and the 5pt12 5pt13 double mutant had an even higher Ca²⁺ content in stage 12 and 13 anthers (Fig. 5C). No significant difference in Ca²⁺ content was detected between the mutants and wild-type stage 11 anthers, where 5PT12 and 5PT13 have very low expression.

Although the Ins(1,4,5)P₃ receptor has not been identified in higher plants, it is assumed that Ins(1,4,5)P₃ can stimulate the Ca²⁺ release from the internal Ca²⁺ store. The concentration and distribution of [Ca²⁺]_cyt in 5pt12, 5pt13 or 5pt12 5pt13 pollen were further detected by Indo-1 (an indicator of [Ca²⁺]_cyt) staining. [Ca²⁺]_cyt was gradually increased along with the pollen development, and reached its highest level in mature tricellular pollen grains at floral stage 12 (Fig. 5D). At floral stage 9, no differences in Indo-1-indicated [Ca²⁺]_cyt were detected between wild type and 5pt mutants, whereas, at stage 12, significantly higher [Ca²⁺]_cyt was detected in 5pt12 pollen grains, consistent with the expression pattern of 5PT12. Furthermore, the distribution of [Ca²⁺]_cyt showed a diffuse pattern rather than accumulation in germination furrows (high concentration of [Ca²⁺]_cyt is accumulated in germination furrows of wild-type pollen grains for subsequent germination of pollen tubes, Fig. 5D). The alterations of [Ca²⁺]_cyt were much more significant in the 5pt12 5pt13 double mutant (Fig. 5D).

To rule out the possible effect of pollen coat on Indo-1 staining, pollen protoplast was used for detecting Indo-1-indicated [Ca²⁺]_cyt and similar results were obtained (Fig. 5E). Observation of plasmolyzed root hair cells stained with Indo-1 further indicated that Indo-1 fluorescence came from cytoplasm rather than the cell wall (supplementary material Fig. S4), confirming that Ca²⁺ associated with the pollen coat or cell wall is not labeled by Indo-1.

These results indicate that the Ins(1,4,5)P₃-mediated [Ca²⁺]_cyt increase is responsible for the precocious pollen germination in 5pt12 and 5pt12 5pt13 mutants, and suggest that 5PTs and Ins(1,4,5)P₃/Ca²⁺ are crucial for maintaining pollen dormancy.

Although Ca²⁺ is known to be required for pollen germination and pollen tube growth, the effect of Ca²⁺ on pollen dormancy has not been reported before. We therefore tested the influence of exogenous Ca²⁺ on pollen grains at floral late stage 11. [Ca²⁺]_cyt in 5pt12, 5pt13 and 5pt12 5pt13 pollen at floral late stage 11 was only slightly higher than that in wild-type pollen (Fig. 5D), and all these mutants did not show the precocious pollen germination within anthers at this stage. However, in vitro pollen germination assays showed that under 100% humidity, wild-type pollen at floral late stage 11 could germinate in a low percentage (∼26%, Fig. 6A), indicating the low germination potential of pollen at late floral stage 11. As exogenous Ca²⁺ concentration rose, the precocious germination rate of wild-type pollen increased, especially at concentrations of 1 or 1.5 mM (Fig. 6A), indicating that exogenous Ca²⁺ can break the dormancy and promote precocious germination of wild-type pollen at late floral stage 11 in a dose-dependent manner. As expected, 5pt12 and 5pt12 5pt13 mutants showed higher precocious pollen germination rates than wild type when supplemented with 1 or 1.5 mM Ca²⁺ (Fig. 6A), consistent with the slightly increased [Ca²⁺]_cyt in pollen of these mutants at floral late stage 11. There was no difference in precocious germination rate under high concentrations of exogenous Ca²⁺ (2 mM, supplementary material Fig. S5).

At floral stage 12, pollen grains are in enclosed anthers. As pollen dehydration is supposed to occur at late stage 12, the inside of the anther should have very low environmental humidity. Therefore, we further tested the effect of Ca²⁺ on pollen germination in very low relative humidity (0-10%). In the absence of additional Ca²⁺, wild-type pollen grains (floral stage 12) germinated at a very low rate, whereas supplementation with external Ca²⁺ significantly stimulated the germination (4 mM Ca²⁺ produces a 44% germination rate, Fig. 6B). This indicated that [Ca²⁺]_cyt, independently of hydration/humidity, can break pollen dormancy and stimulate precocious pollen germination, which mimicked the precocious pollen germination of 5PTs-deficient mutants to some extent.

To further demonstrate that Ca²⁺ is crucial for precocious pollen germination in vivo, exogenous Ca²⁺ was applied to wild-type anthers and was found to stimulate precocious pollen germination, which could reach to 35% with application of 6 mM Ca²⁺ (Fig. 6C). Moreover, the precocious germination rates in anthers of 5pt12, 5pt13 and 5pt12 5pt13 mutant were decreased when applying the [Ca²⁺]_cyt inhibitor BAPTA (Fig. 6D), confirming the role of [Ca²⁺]_cyt in pollen dormancy and germination. These results indicate that the increase of Ins(1,4,5)P₃/Ca²⁺ in 5PTs-deficient pollen grains is responsible for precocious pollen germination and confirm that 5PT-controlled Ins(1,4,5)P₃/Ca²⁺ is crucial for maintaining pollen dormancy.

After maturation, pollen grains undergo dehydration and are then released from anthers in a partially dehydrated and dormant state. Pollen germination on the stigma starts with the rehydration that is subject to stringently spatiotemporal control (Heslop-Harrison, 1979). Inappropriate dehydration/rehydration results in premature or earlier germination within the anther, or germination on the wrong surface (Lolle and Cheung, 1993; Lolle et al., 1998). Therefore, pollen dehydration is considered to be the major determinant that controls pollen dormancy. Here, our study first reveal that, in addition to dehydration, 5PTs and Ins(1,4,5)P₃/Ca²⁺ play crucial roles in pollen dormancy and uncover a novel mechanism for controlling it.

Deficiency of 5PT12, 5PT13 leads to precocious pollen germination within anthers, suggesting the essential roles of these 5PTs in pollen dormancy. However, the precociously germinated pollen grains in 5pt12 anthers could dehydrate normally and pollens of the 5pt12 mutant is dehydrated before release from the anthers (Fig. 3A), suggesting that humidity or altered sensitivity to environmental humidity are not responsible for precocious germination, As the 5pt12 mutant exhibits hypersensitive ABA responses, further excluding the influence of dehydration/hydration or ABA signaling in altered pollen dormancy and germination of the 5pt12 mutant.

As speculated, the basal levels of Ins(1,4,5)P₃, the common substrate of 5PT12 and 5PT13, are elevated significantly in the 5pt12 or 5pt12 5pt13 mutant, and an increased [Ca²⁺]_cyt concentration is detected in pollen grains of the 5pt12 mutant or 5pt12 5pt13 double mutant (more significantly) at floral stage 12. Previous studies have shown that a 1.6-fold increase in Ins(1,4,5)P₃ did affect the seedling growth (Ercetin et al., 2008) and here an approximate doubling of Ins(1,4,5)P₃ in the 5pt12 5pt13 double mutant is sufficient to lead to precocious pollen germination. In addition, preventing the increase in basal Ins(1,4,5)P₃ levels in the 5pt12 mutant by complementary expression of 5PT12 resulted in the normal pollen germination rate, further confirming the role of increased Ins(1,4,5)P₃ levels in precocious pollen germination. No significantly increase in Ins(1,4,5)P₃ or total Ca²⁺ content was detected in 5pt13 mutant anthers, which may be because whole anthers were used as the material for Ins(1,4,5)P₃ and total Ca²⁺ measurement (because pollen grains are too small to obtain enough material for measurement), and the minor alterations in 5pt13 mutant pollen grains might be covered up. However, the increased Ca²⁺ in the 5pt13 mutant pollen grains can be clearly observed by the Indo-1 staining assay, which is consistent with precocious pollen germination in the 5pt13 mutant. These findings indicate that precocious pollen germination, a higher basal level of Ins(1,4,5)P₃ and increased [Ca²⁺]_cyt in pollen grains are closely correlated.

A more diffuse [Ca²⁺]_cyt distribution pattern was observed in the pollen of the 5pt12 5pt13 double mutant (distinguished from the aggregated pattern in wild-type pollen grains prior to germination on stigma) (Fig. 5D,E), which may explain the emergence of more than one pollen tube. The concentration and distribution of [Ca²⁺]_cyt is supposed to be stringently controlled during pollen dormancy and 5PTs/Ins(1,4,5)P₃ may be the regulators.

In vitro pollen germination assays showed that exogenous Ca²⁺ could break dormancy and stimulate the early germination of wild-type pollen grains at floral late stage 11, which confirms the crucial roles of Ca²⁺ in pollen dormancy. In addition, exogenous Ca²⁺ can break dormancy and trigger early germination of wild-type pollen grains at floral stage 12 under very low humidity (Fig. 6B). BAPTA treatment resulted in the decreased pollen germination rates in anther of 5pt12, 5pt13 and 5pt12 5pt13 mutants, confirming that the precocious pollen germination in 5PT mutants is caused by increased [Ca²⁺]_cyt (Fig. 6D). In addition, 5PT14, a close paralog of 5PT12 and 5PT13, is also expressed at floral stage 12 to 14, and the expression level of 5PT14 is higher than 5PT12 and 5PT13 at these stages (Fig. 1C, supplementary material Fig. S6A). Recently, we identified a T-DNA insertion mutant of 5PT14, which also exhibited precocious pollen germination in dehiscent anthers (∼15-30%) (supplementary material Fig. S6B,C,D). Although all 5PT12, 5PT13 and 5PT14 are expressed in pollen grain, each single mutant exhibits the phenotype of precocious pollen germination and the 5pt12 5pt13 double mutant with much higher percentage of the of precocious pollen germination, suggesting the dose effect of 5PTs in regulating the pollen germination. A further study with mutation of all 5PT12, 5PT13 and 5PT14 is on the way.

Take together, these findings indicate that 5PT-controlled Ins(1,4,5)P₃/Ca²⁺ has a crucial role in maintaining pollen dormancy and the increase of 5PT-controlled Ins(1,4,5)P₃/Ca²⁺ is sufficient to break pollen dormancy and trigger early germination, independent of hydration/humidity. This study sheds light on the mechanism that controls pollen dormancy and germination (Fig. 6E). In wild-type plants, 5PT12, 5PT13 and 5PT14 are highly expressed in tri-cellular pollen at floral stage 12 in order to restrain the level of Ins(1,4,5)P₃/[Ca²⁺]_cyt in tri-cellular pollen. The pollen at stage 12 is then maintained in a dormant state, and prevented from precocious germinating. At late stage 12, tri-cellular pollen undergoes dehydration to further maintain the dormant state. Once the dehydrated pollen grains fall on pistils and rehydrate, Ca²⁺ will flow into pollen grains and aggregates in the germination pore, breaking dormancy and stimulating pollen early germination. Deficiency of 5PT genes will result in the increased intrinsic Ins(1,4,5)P₃/[Ca²⁺]_cyt in tri-cellular pollen at floral stage 12, which sufficiently breaks the pollen dormancy and results in the precocious germination of pollen grains in anthers.

Ins(1,4,5)P₃ can be converted into InsP₆ (inositol hexakisphosphate) and InsP₆ was reported to mobilize intracellular Ca²⁺ stores (Lemtiri-Chlieh et al., 2003) in guard cells. Although there is no evidence that InsP₆ is involved in breaking pollen dormancy, it is possible that the InsP₆ content in 5PT mutants is also increased and, to some extent, contributes to the alteration of [Ca²⁺]_cyt in 5PT mutants.

Here, we report a striking phenotype in Arabidopsis: precocious pollen germination within anthers at an incidence of 40-70%, caused by 5PT12 deficiency. Pollen grains in wild-type Arabidopsis do not germinate until falling onto stigma. In fact, pollen grains in almost all the flowering plants will not germinate prior to pollination. The only exceptions are some cleistogamous species, in which pollen germinates inside the anther, and the pollen tubes extend via the anther stomium to reach to the stigma for fertilization (Lord, 1981; Trent, 1942), e.g. Cardamine chenopodifolia (Lord, 1981). It is noteworthy that C. chenopodifolia is a Cruciferae plant, similar to Arabidopsis. In species in which the anther sac fails to open, germinated pollen tubes go through the anther walls to the stigma (Hanson, 1943; Lord, 1981). However, little is known about the genetic mechanism. Although a similar phenotype has been reported in the raring-to-go (rtg) mutant, the pollen grains of which do not desiccate completely and thus germinate precociously (Johnson and McCormick, 2001), the responsible gene and genetic mechanism are still unknown. In our study, 5pt12 and 5pt12 5pt13 mutants showed precocious pollen germination within anthers, similar to these cleistogamous plants, providing convincing evidence of the crucial roles of 5PTs and Ins(1,4,5)P₃/[Ca²⁺]_cyt in pollen dormancy, and revealing a novel mechanism for keeping pollen dormancy. These results provide insights into the mechanisms of pollen germination in such cleistogamous species. It is speculated that Ins(1,4,5)P₃/[Ca²⁺]_cyt might also function in breaking dormancy and stimulating pollen germination in these cleistogamous plants, and 5PTs or other enzymes controlling Ins(1,4,5)P₃ metabolism in the pollen of these species are expressed at low levels or are deficient, which promotes pollen germination in anthers.

Increased Ins(1,4,5)P₃/Ca²⁺ is sufficient to break pollen dormancy and trigger early germination, suggesting an engineering prospect of 5PTs in altering dormancy. The mechanism controlling seed dormancy is very likely to be similar to pollen dormancy and thereby 5PTs may be applied to alter seed dormancy. Recently, two lines with altered expression of callose synthases were reported to show precocious pollen germination, but at a very low incidence (4-8%) (Xie et al., 2010). However, in contrast to the 5pt12 mutant, all the reported lines, including rtg mutant, have abnormal pollen development as early as the bi-cellular stage. Thus, 5PT12 may be a better candidate for genetic engineering in crops for improving fertility and sterility, especially under environmental stress. For cereal crops, seed dormancy before harvest is a desired trait. In some cultivars, mature grains germinate on maternal plants following high temperature or rains, causing large economic losses. This phenomenon is characterized as pre-harvest sprouting or vivipary. It is worth testing whether the vivipary phenotype could be somehow rescued by overexpression of 5PT12. On the contrary, some crops, such as barley, exhibit excessive dormancy at harvest that prevent grains from rapid and uniform germination, which is required in the malting process. Using a transgenic strategy (using antisense fragment of 5PT12 under the control of seed-specific promoter) to specifically silence 5PT orthologs in mature seeds of these crops might inhibit seed dormancy. Thus, our research provides possibilities of genetic engineering for improving crop traits and agricultural production.