- Split View
-
Views
-
Cite
Cite
Mostafa Aboulela, Tsuyoshi Nakagawa, Akinobu Oshima, Kohji Nishimura, Yuji Tanaka, The Arabidopsis COPII components, AtSEC23A and AtSEC23D, are essential for pollen wall development and exine patterning, Journal of Experimental Botany, Volume 69, Issue 7, 16 March 2018, Pages 1615–1633, https://doi.org/10.1093/jxb/ery015
- Share Icon Share
Abstract
The specialized multilayered pollen wall plays multiple roles to ensure normal microspore development. The major components of the pollen wall (e.g. sporopollenin and lipidic precursors) are provided from the tapetum. Material export from the endoplasmic reticulum (ER) is mediated by coat protein complex II (COPII) vesicles. The Arabidopsis thaliana genome encodes seven homologs of SEC23, a COPII component. However, the functional importance of this diversity remains elusive. Here, we analyzed knockout and knockdown lines for AtSEC23A and AtSEC23D, two of the A. thaliana SEC23 homologs, respectively. Single atsec23a and atsec23d mutant plants, despite normal fertility, showed an impaired exine pattern. Double atsec23ad mutant plants were semi-sterile and exhibited developmental defects in pollen and tapetal cells. Pollen grains of atsec23ad had defective exine and intine, and showed signs of cell degeneration. Moreover, the development of tapetal cells was altered, with structural abnormalities in organelles. AtSEC23A and AtSEC23D exhibited the characteristic localization pattern of COPII proteins and were highly expressed in the tapetum. Our work suggests that AtSEC23A and AtSEC23D may organize pollen wall development and exine patterning by regulating ER export of lipids and proteins necessary for pollen wall formation. Also, our results shed light on the functional heterogeneity of SEC23 homologs.
Introduction
The multilayered pollen wall provides structural and physical support for the microspore cytoplasm, helps the male gametes survive severe environmental conditions, and facilitates pollination and male–female interaction processes (Scott et al., 2004). In Arabidopsis thaliana, the pollen wall has an architecturally complex structure comprised of three layers: the inner pectocellulosic-based intine, the outer sporopollenin-based exine, and a lipid-based pollen coat covering the exine. The exine is divided into two layers: nexine and sexine. Further, sexine is subdivided into two structures: pillar-like bacula and roof-like tectum (Ariizumi and Toriyama, 2011). The bacula and the tectum are responsible for the characteristic and taxon-specific architecture of the exine, which is reticulate in A. thaliana.
Pollen wall development relies largely on the adjacent sporophytic anther layer, the tapetum, which provides the developing microspores with metabolites, enzymes, nutrients, and structural components necessary to build up the outer exine (Pacini et al., 1985; Blackmore et al., 2007). During the early stages of pollen development, tapetal cells secrete β-1,3-glucanase (callase) to degrade the callose wall around tetrads and release the microspores (Stieglitz, 1977; Verma and Hong, 2001). Then, tapetal cells immediately synthesize lipidic sporopollenin precursors as major components of the exine wall (Quilichini et al., 2014). Finally, during the late stages of pollen development, tapetal cells synthesize and store lipidic materials of the pollen coat in two specialized organelles, the plastid-derived elaioplasts and the endoplasmic reticulum (ER)-derived tapetosomes. Premature or delayed differentiation of tapetal cells, as well as alteration of their internal structure, usually results in defective pollen development and reduced fertility (Kawanabe et al., 2006; Ariizumi and Toriyama, 2011; Liu and Fan, 2013).
The coat protein complex II (COPII)-mediated transport from the ER to the Golgi apparatus represents the first step of the secretory pathway (Schekman and Orci, 1996). Newly synthesized lipids and proteins exit the ER through COPII-coated vesicles heading to the Golgi apparatus for further modification and sorting before reaching their final destinations (Vitale and Denecke, 1999; Foresti and Denecke, 2008). The COPII-coated vesicle formation requires the sequential recruitment of the five cytosolic components, SAR1, SEC23/24, and SEC13/31 (Matsuoka et al., 1998; Lord et al., 2013). In this process, SEC23 interacts with SAR1, SEC24, and SEC31, and functions as a SAR1-specific GTPase-activating protein (GAP) and in cargo recognition with SEC24 (Kuehn et al., 1998; Mancias and Goldberg, 2007; Fromme et al., 2008).
Previous studies have revealed the importance of some conserved amino acids in SEC23. The Arg722 of the yeast SEC23 is necessary for its GAP activity (Bi et al., 2002). In human, a missense mutation at the conserved Phe382 in SEC23A leads to a reduction of its affinity for SEC31 and causes cranio-lenticulo-sutural dysplasia disease (Boyadjiev et al., 2006; Fromme et al., 2007). In A. thaliana, an amino acid substitution in AtSEC23A at a conserved aspartic acid has been reported as being essential for the unique interaction with AtSAR1 (Zeng et al., 2015).
Many studies demonstrate the involvement of COPII components in regulating plant growth and development. In A. thaliana, we previously reported that both AtSEC24B and AtSEC24C were redundantly involved in male and female gametophyte development (Tanaka et al., 2013). AtSEC24A is essential for male fertility (Conger et al., 2011), ER–Golgi integrity (Faso et al., 2009; Nakano et al., 2009), and maintaining cell size patterning in sepals (Qu et al., 2014). The specific interaction between AtSAR1 and AtSEC23A is required for their function in ER export of proteins (Zeng et al., 2015). More recently, AtSEC31B was reported to be required for pollen wall development, probably by regulating the early secretory pathway of tapetal cells (Zhao et al., 2016).
Although SEC23 is an essential component of COPII vesicle formation, its involvement in regulating plant growth and development has yet to be elucidated. Moreover, the functional differences among the SEC23 homologs remain to be determined. In this study, we identified and characterized two A. thaliana SEC23 gene homologs, AtSEC23A and AtSEC23D. Both AtSEC23A and AtSEC23D exhibited the characteristic COPII localization at ER exit sites (ERESs) and are required for proper pollen wall formation, exine patterning, and tapetum development. Our results indicate that ER export of proteins in the early secretory pathway of tapetal cells is a key factor for pollen wall development. Also, this work points to the functional diversity of SEC23 homologs in A. thaliana.
Materials and methods
Plant materials, growth condition, and transformation
Arabidopsis thaliana T-DNA insertion lines (Col-0 background) and the quartet1 (qrt1-2) mutant were obtained from the Arabidopsis Biological Resource Center (ABRC), Ohio State University, USA. Arabidopsis thaliana ecotype Col-0 was used as the wild type. The seeds were germinated on a Murashige and Skoog agar medium at 22 °C under continuous light. After 10–14 d, seedlings were transplanted to Jiffy-7 (Jiffy Preforma Production K. K, Yokohama, Japan) and grown at 22 °C under long-day conditions (16 h light/8 h dark) or under continuous light. Transformation of A. thaliana was performed by the floral dip method (Clough and Bent, 1998) or the floral inoculating method (Narusaka et al., 2010), and transgenic plants were screened on Murashige and Skoog agar medium containing 100 mg l−1 Cefotax (Chugai Pharmaceutical, Tokyo, Japan) and appropriate antibiotics (20 mg l−1 hygromycin B or 30 mg l−1 kanamycin).
Genotyping analysis
Genomic DNAs were extracted from leaves following Edwards et al. (1991). The T-DNA insertions were analyzed by PCR genotyping with the specific primers GN-AtSEC23A-F and GN-AtSEC23A-R for AtSEC23A or RT-AtSEC23D-F and GN-AtSEC23D-R for AtSEC23D. The T-DNA-LB was used as a common primer for T-DNA detection. The primers are listed in Supplementary Table S1 at JXB online.
Reverse Transcription PCR (RT-PCR) analysis
Total RNAs were extracted from various tissues of wild-type plants using the RNeasy Mini Kit (Qiagen, Tokyo Japan), and used as templates to synthesize cDNAs using ReverTraAce (TOYOBO, Osaka, Japan), according to the manufacturer’s instructions. The ACTIN2 gene (At3g18780) was used as an internal reference. RT-PCR was performed with 0.2 mg of cDNAs using KOD-Plus-Neo DNA polymerase (TOYOBO) for 26 cycles. The specific primers RT-AtSEC23A-F, RT-AtSEC23A-R, RT-AtSEC23D-F, RT-AtSEC23D-R, ACT2-F, and ACT2-R used for RT-PCR are listed in Supplementary Table S1.
Preparation of promoter and ORF entry clones
All entry clones shown in Table 1 were prepared with the following method. Promoter and ORF fragments were amplified to add attB sequences to their 5' and 3' ends by one or two PCRs with templates and the forward and reverse primers indicated in Table 1. The amplified DNA fragments were introduced into the vectors designated in Table 1 by BP reactions, following the manufacturer’s instructions (Thermo Fisher Scientific, Kanagawa, Japan), to construct the entry clones.
Entry clone . | Cloned DNA fragment . | Template . | Forward primera . | Reverse primera . | Vector . |
---|---|---|---|---|---|
pDONR201-PAtSEC23A (attL1-PAtSEC23A-attL2) | AtSEC23A promoter (–1996 to –1)b | A. thaliana Col genomic DNA | PAtSEC23A-attB1 | PAtSEC23A-attB2 | pDONR201 |
pDONR P4-P1R-PAtSEC23A (attL4-PAtSEC23A-attR1) | AtSEC23A promoter (–1996 to –1)b | PAtSEC23A-attB4 | PAtSEC23A-attB1r | pDONR P4-P1R | |
pDONR201-PAtSEC23D (attL1-PAtSEC23D-attL2) | AtSEC23D promoter (–2126 to –1)b | PAtSEC23D-attB1 | PAtSEC23D-attB2 | pDONR201 | |
pDONR P4-P1R-PAtSEC23D (attL4-PAtSEC23D-attR1) | AtSEC23D promoter (–2126 to –1)b | PAtSEC23D-attB4 | PAtSEC23D-attB1r | pDONR P4-P1R | |
pDONR P4-P1R-Pnos (attL4-Pnos-attR1) | Nopaline synthase promoter | pGWB 401 | Pnos-attB4 | Pnos-attB1r | |
pDONR201-AtSEC23A (attL1-AtSEC23A-attL2) | AtSEC23A ORFc | pda 01836d | AtSEC23A-attB1 (the first PCR) and attB1 adaptor (the second PCR) | AtSEC23A-attB2 (the first PCR) and attB2 adaptor (the second PCR) | pDONR201 |
pDONR201-AtSEC23D (attL1-AtSEC23D-attL2) | AtSEC23D ORFc | pda 04203d | AtSEC23D-attB1 (the first PCR) and attB1 adaptor (the second PCR) | AtSEC23D-attB2 (the first PCR) and attB2 adaptor (the second PCR) |
Entry clone . | Cloned DNA fragment . | Template . | Forward primera . | Reverse primera . | Vector . |
---|---|---|---|---|---|
pDONR201-PAtSEC23A (attL1-PAtSEC23A-attL2) | AtSEC23A promoter (–1996 to –1)b | A. thaliana Col genomic DNA | PAtSEC23A-attB1 | PAtSEC23A-attB2 | pDONR201 |
pDONR P4-P1R-PAtSEC23A (attL4-PAtSEC23A-attR1) | AtSEC23A promoter (–1996 to –1)b | PAtSEC23A-attB4 | PAtSEC23A-attB1r | pDONR P4-P1R | |
pDONR201-PAtSEC23D (attL1-PAtSEC23D-attL2) | AtSEC23D promoter (–2126 to –1)b | PAtSEC23D-attB1 | PAtSEC23D-attB2 | pDONR201 | |
pDONR P4-P1R-PAtSEC23D (attL4-PAtSEC23D-attR1) | AtSEC23D promoter (–2126 to –1)b | PAtSEC23D-attB4 | PAtSEC23D-attB1r | pDONR P4-P1R | |
pDONR P4-P1R-Pnos (attL4-Pnos-attR1) | Nopaline synthase promoter | pGWB 401 | Pnos-attB4 | Pnos-attB1r | |
pDONR201-AtSEC23A (attL1-AtSEC23A-attL2) | AtSEC23A ORFc | pda 01836d | AtSEC23A-attB1 (the first PCR) and attB1 adaptor (the second PCR) | AtSEC23A-attB2 (the first PCR) and attB2 adaptor (the second PCR) | pDONR201 |
pDONR201-AtSEC23D (attL1-AtSEC23D-attL2) | AtSEC23D ORFc | pda 04203d | AtSEC23D-attB1 (the first PCR) and attB1 adaptor (the second PCR) | AtSEC23D-attB2 (the first PCR) and attB2 adaptor (the second PCR) |
a Sequences of forward and reverse primers are listed in Supplementary Table S1.
b A of the initiation codon is +1.
c Nucleotide sequences of ORFs correspond to the region from the translation initiation site to the last amino acid of AtSEC23A and AtSEC23D.
d Arabidopsis full-length cDNA clones, pda01836 and pda04203, were obtained from RIKEN BRC (RIKEN, Tsukuba, Japan).
Entry clone . | Cloned DNA fragment . | Template . | Forward primera . | Reverse primera . | Vector . |
---|---|---|---|---|---|
pDONR201-PAtSEC23A (attL1-PAtSEC23A-attL2) | AtSEC23A promoter (–1996 to –1)b | A. thaliana Col genomic DNA | PAtSEC23A-attB1 | PAtSEC23A-attB2 | pDONR201 |
pDONR P4-P1R-PAtSEC23A (attL4-PAtSEC23A-attR1) | AtSEC23A promoter (–1996 to –1)b | PAtSEC23A-attB4 | PAtSEC23A-attB1r | pDONR P4-P1R | |
pDONR201-PAtSEC23D (attL1-PAtSEC23D-attL2) | AtSEC23D promoter (–2126 to –1)b | PAtSEC23D-attB1 | PAtSEC23D-attB2 | pDONR201 | |
pDONR P4-P1R-PAtSEC23D (attL4-PAtSEC23D-attR1) | AtSEC23D promoter (–2126 to –1)b | PAtSEC23D-attB4 | PAtSEC23D-attB1r | pDONR P4-P1R | |
pDONR P4-P1R-Pnos (attL4-Pnos-attR1) | Nopaline synthase promoter | pGWB 401 | Pnos-attB4 | Pnos-attB1r | |
pDONR201-AtSEC23A (attL1-AtSEC23A-attL2) | AtSEC23A ORFc | pda 01836d | AtSEC23A-attB1 (the first PCR) and attB1 adaptor (the second PCR) | AtSEC23A-attB2 (the first PCR) and attB2 adaptor (the second PCR) | pDONR201 |
pDONR201-AtSEC23D (attL1-AtSEC23D-attL2) | AtSEC23D ORFc | pda 04203d | AtSEC23D-attB1 (the first PCR) and attB1 adaptor (the second PCR) | AtSEC23D-attB2 (the first PCR) and attB2 adaptor (the second PCR) |
Entry clone . | Cloned DNA fragment . | Template . | Forward primera . | Reverse primera . | Vector . |
---|---|---|---|---|---|
pDONR201-PAtSEC23A (attL1-PAtSEC23A-attL2) | AtSEC23A promoter (–1996 to –1)b | A. thaliana Col genomic DNA | PAtSEC23A-attB1 | PAtSEC23A-attB2 | pDONR201 |
pDONR P4-P1R-PAtSEC23A (attL4-PAtSEC23A-attR1) | AtSEC23A promoter (–1996 to –1)b | PAtSEC23A-attB4 | PAtSEC23A-attB1r | pDONR P4-P1R | |
pDONR201-PAtSEC23D (attL1-PAtSEC23D-attL2) | AtSEC23D promoter (–2126 to –1)b | PAtSEC23D-attB1 | PAtSEC23D-attB2 | pDONR201 | |
pDONR P4-P1R-PAtSEC23D (attL4-PAtSEC23D-attR1) | AtSEC23D promoter (–2126 to –1)b | PAtSEC23D-attB4 | PAtSEC23D-attB1r | pDONR P4-P1R | |
pDONR P4-P1R-Pnos (attL4-Pnos-attR1) | Nopaline synthase promoter | pGWB 401 | Pnos-attB4 | Pnos-attB1r | |
pDONR201-AtSEC23A (attL1-AtSEC23A-attL2) | AtSEC23A ORFc | pda 01836d | AtSEC23A-attB1 (the first PCR) and attB1 adaptor (the second PCR) | AtSEC23A-attB2 (the first PCR) and attB2 adaptor (the second PCR) | pDONR201 |
pDONR201-AtSEC23D (attL1-AtSEC23D-attL2) | AtSEC23D ORFc | pda 04203d | AtSEC23D-attB1 (the first PCR) and attB1 adaptor (the second PCR) | AtSEC23D-attB2 (the first PCR) and attB2 adaptor (the second PCR) |
a Sequences of forward and reverse primers are listed in Supplementary Table S1.
b A of the initiation codon is +1.
c Nucleotide sequences of ORFs correspond to the region from the translation initiation site to the last amino acid of AtSEC23A and AtSEC23D.
d Arabidopsis full-length cDNA clones, pda01836 and pda04203, were obtained from RIKEN BRC (RIKEN, Tsukuba, Japan).
Preparation of green fluorescent protein (GFP) fusion constructs for complementation and expression analyses
The pDONRP4-P1R-PAtSEC23A and pDONR201-AtSEC23A entry clones were subjected to LR reaction with R4pGWB550 (Nakagawa et al., 2008) to generate PAtSEC23A:AtSEC23A-G3GFP. PAtSEC23D:AtSEC23D-G3GFP was prepared by the same procedure using pDONRP4-P1R-PAtSEC23D and pDONR201-AtSEC23D entry clones. In this study, we used G3GFP (Kawakami and Watanabe, 1997), a brighter variant of GFP with S65A/Y145F mutations.
Staining and semi-thin sectioning
Alexander’s (Alexander, 1969) and DAPI stainings were performed as described in Tanaka et al. (2013). For Alexander’s staining, anthers were observed by a BZ-X710 All-in-One fluorescence microscope (KEYENCE, Osaka, Japan). DAPI fluorescence was detected using a BX51 fluorescence microscope (Olympus, Tokyo, Japan) equipped with a UV mirror unit. For the aniline blue staining, buds at the tetrad stage were squeezed into a drop of aniline blue solution (100 mg l−1 in 50 mM potassium phosphate buffer, pH 7.5) on a slide and observed using a BX51 fluorescence microscope. For the auramine O staining, pollen grains were immersed in a drop of auramine O solution (0.001% auramine O in 50 mM Tris–HCl, pH 7.5; Dobritsa et al., 2010) on a slide and viewed by confocal microscopy as mentioned below.
Floral buds at different developmental stages were fixed in 4% paraformaldehyde solution, dehydrated in a graded ethanol series, embedded in Technovit 7100 resin (Heraeus Kulzer, Wehrheim, Germany), and sectioned (2–3 µm) by an RV-240 rotary microtome (Yamato Kohki Industrial, Saitama, Japan). Semi-thin sections were stained with a toluidine blue solution (1% toluidine blue, 1% sodium borate) and viewed by the All-in-One fluorescence microscope.
Promoter:β-glucuronidase (GUS) assay
For the promoter:GUS assays, pDONR201-PAtSEC23A and pDONR201-PAtSEC23D entry clones were applied for LR reactions with pGWB233 (Hino et al., 2011) to make PAtSEC23A:GUS and PAtSEC23D:GUS. These constructs were used for transformation of wild-type A. thaliana. T2 or T3 lines were stained and examined following Nakamura et al. (2012).
In vitro pollen germination
Pollen grains of at least six recently opened flowers were tested for germination in vitro according to Boavida and McCormick (2007). Images were captured using an SZX16 stereo-microscope (Olympus). Germination rates were calculated by counting at least 500 pollen grains for each sample with ImageJ (http://rsbweb.nih.gov/ij/).
Transient co-localization analyses in Nicotiana benthamiana leaves
The R4 dual-site Gateway cloning system (Aboulela et al., 2017) was used to express two genes in the co-localization analyses. The promoter entry clone pDONRP4-P1R-Pnos and the ORF entry clones pDONR201-AtSEC23A, pDONR201-AtSEC23D, pDONR201-AtSEC24A (Tanaka et al., 2013), and pDONR201-SYP31 (Tanaka et al., 2013) were used for preparation of the two-gene constructs with R4pDD650-MD8 and R4pGWB6459-MD8 according to Aboulela et al. (2017); these were Pnos:AtSEC24A-TagRFP–Pnos:AtSEC23A-G3GFP, Pnos:AtSEC24A-TagRFP–Pnos:AtSEC23D-G3GFP, Pnos: SYP31-TagRFP–Pnos:AtSEC23A-G3GFP, Pnos:SYP31-TagRFP–Pnos:AtSEC23D-G3GFP, and Pnos:AtSEC23A-TagRFP–Pnos:AtSEC 23D-G3GFP.
Transient expression of each two-gene construct in N. benthamiana was performed according to the agroinfiltration method described in Shah et al. (2013) with some modifications. Agrobacterium tumefaciens strain EHA101 was used and the infiltration medium contained 150 mM acetosyringone.
Confocal microscopy
The fluorescence of G3GFP and TagRFP (red fluorescent protein) was viewed with a TCS SP5 confocal laser scanning microscope (CLSM) (Leica Microsystems, Wetzlar, Germany) using an HCX IRAPO L 25.0 × 0.95 water-immersion objective lens. In observations of anthers, settings described in Huang et al. (2013) were applied to distinguish the G3GFP signals from the chlorophyll autofluorescence in anthers. In the co-localization experiments in N. benthamiana, G3GFP and TagRFP were excited with the argon (488 nm) or the helium–neon (543 nm) laser lines, and their fluorescence was captured through a band path of 500–530 nm or 555–615 nm, respectively. The fluorescence of auramine O was detected using the fluorescein isothiocyanate settings and an HCX PLAPO 100.0 × 1.40-0.70 oil-immersion objective lens.
Electron microscopy
SEM observations were performed as described in Tanaka et al. (2013) using either an S-4800 field emission scanning electron microscope (Hitachi High-Tech, Tokyo, Japan) or a TM3000 miniscope (Hitachi High-Tech).
Floral buds at appropriate developmental stages were fixed with 2.5% glutaraldehyde in 50 mM sodium phosphate buffer (pH 7.2). The specimens were treated with 1.5% osmium tetroxide in 100 mM cacodylate buffer (pH 7.4) for 90 min and embedded in EPON 812 (TAAB, Berkshire, UK). Ultrathin sections (70 nm) were double-stained with 3% uranyl acetate and lead citrate then observed with a JEM-1400 transmission electron microscope (JEOL, Tokyo, Japan).
Results
AtSEC23A and AtSEC23D are divergent from other Arabidopsis SEC23 homologs
Within the A. thaliana genome, there are five homologs for SAR1, seven for SEC23, three for SEC24, and two for each of SEC12, SEC13, SEC16, and SEC31 (Robinson et al., 2007; Chung et al., 2016). The seven SEC23 homologs, At4g01810, At1g05520, At2g21630, At2g27460 At3g23660, At4g14160, and At5g43670, were named AtSEC23A, AtSEC23B, AtSEC23C, AtSEC23D, AtSEC23E, AtSEC23F, and AtSEC23G, respectively, following Chung et al. (2016). Phylogenetic analysis of SEC23 homologs in yeast, green algae, plants, and mammals showed that five of the A. thaliana SEC23 homologs were clustered in a single clade with rice and green algae homologs (Fig. 1A). While both AtSEC23A and AtSEC23D were divergent from other AtSEC23 proteins, each was separated in a distinct clade with only homologs of rice. AtSEC23A and AtSEC23D showed no significant identity with other AtSEC23s (Supplementary Fig. S1).
In Saccharomyces cerevisiae, ScSEC23 includes five domains, zinc finger, trunk, β-barrel, all-helical, and gelsolin-like domains (Bi et al., 2002). These five domains are well conserved within AtSEC23s, except for AtSEC23D which is missing the β-barrel and gelsolin-like domains (Fig. 1B). Sequence alignment of AtSEC23s and ScSEC23 revealed that only AtSEC23A and AtSEC23D have an amino acid substitution at the conserved aspartate residue in the trunk domain [Asp351 of ScSEC23 to Cys484 of AtSEC23A (Zeng et al., 2015) or Glu337 of AtSEC23D] (Supplementary Fig. S1). Interestingly, AtSEC23D is also missing Phe366 [equivalent to Phe382 of human SEC23A (Boyadjiev et al., 2006; Fromme et al., 2007)] and the catalytic Arg706 [Arg722 of ScSEC23 (Bi et al., 2002)], which are conserved in ScSEC23 and other AtSEC23s (Supplementary Fig. S1). These results indicated that both AtSEC23A and AtSEC23D are distinct from other AtSEC23s.
atsec23a and atsec23d single mutants show defects in exine patterning
To understand the functions of AtSEC23s during plant development, we prepared homozygous T-DNA insertion lines of the seven AtSEC23 genes and screened them for developmental abnormalities. Among them all, we found abnormalities in the pollen wall structure of insertion lines of AtSEC23A (At4g01810) and AtSEC23D (At2g27460) (Fig. 2; Supplementary Fig. S2). These T-DNA insertion lines, SALK_021996 and SALK_012411 (Alonso et al., 2003), were named atsec23a and atsec23d, respectively, and used for further experiments. In atsec23a and atsec23d, T-DNAs were inserted in the first intron located at the 5'-untranslated region and in the last (14th) exon of AtSEC23A and AtSEC23D, respectively (Fig. 2A). RT-PCR analysis revealed that atsec23a and atsec23d were knockdown and knockout mutants, respectively (Fig. 2B). In contrast to wild-type pollen grains which had an exine with the characteristic net-like structure (Fig. 2C), the pollen grains of atsec23a and atsec23d showed an abnormal exine phenotype represented by a partial loss of the net-like structure and incomplete tectum formation in some areas (Fig. 2D, E), which was also shown by auramine O (exine-specific fluorescent dye) stain (Fig. 2G–I). A complementation test was conducted by introducing PAtSEC23A:AtSEC23A-G3GFP and PAtSEC23D:AtSEC23D-G3GFP into atsec23a and atsec23d, respectively. Several complemented lines showing normal exine patterning were obtained (Supplementary Fig. S2F, G). These results indicated that the depletion of AtSEC23A or AtSEC23D leads to defects in the exine patterning.
To analyze functional redundancy of AtSEC23A and AtSEC23D, we generated the double mutant atsec23ad by cross-fertilization. Most pollen grains in the atsec23ad became collapsed and flattened (Fig. 2F, J), and tended to aggregate as a bulk of remnant materials strongly adhesive to anther walls failing to be released (Fig. 3F). Many granules of different shape and size were irregularly scattered on the pollen surface (Fig. 2F). These granules showed auramine O fluorescence, indicating that they are composed of a material similar to sporopollenin (Fig. 2J). Therefore, these results suggested that AtSEC23A and AtSEC23D contribute to exine formation and sporopollenin deposition.
The atsec23ad double mutant shows a semi-sterile phenotype with impaired function of male gametophytes
Because atsec23ad showed severe pollen defects, we analyzed the seed production after self-pollination. The wild type, atsec23a, and atsec23d contained elongated siliques with an almost full set of seeds (Fig. 3A–C), and the total seed yield was ~5000 seeds per plant (Fig. 3D). In contrast, the atsec23ad mutant showed a semi-sterile phenotype with <100 seeds per plant (Fig. 3D). Most siliques were short and devoid of any seeds, and only a few siliques contained a reduced set of developing seeds (Fig. 3A, C).
Next, we observed open flowers. Stamens of atsec23a, atsec23d, and atsec23ad were comparable in height with those of the wild type (Fig. 3E), and their anthers showed normal dehiscence (Fig. 3F). In the wild type, atsec23a, and atsec23d, an abundant amount of pollen grains was visible on anthers and adhered to stigmas, whereas no pollen grains appeared on atsec23ad anthers or stigmas (Fig. 3E). To determine whether atsec23ad also has defects in the female gametophyte, we cross-pollinated stigmas of atsec23ad with wild-type pollen grains. Normal elongated siliques with an almost full set of seeds indistinguishable from that of the wild type were produced (Supplementary Fig. S3), indicating that female gametophytes function normally in atsec23ad. The insufficiency of both AtSEC23A and AtSEC23D is linked to defects in male gametophytes and results in the semi-sterile phenotype.
The atsec23a single mutant exhibits an impaired pollen germination
To confirm whether the semi-sterility in atsec23ad was caused by an abnormal pollen production, we examined the viability of pollen grains by Alexander’s and DAPI stainings. The wild-type, atsec23a, and atsec23d anthers contained purple-stained (viable) pollen grains (Fig. 4A–C). In contrast, most anthers of the atsec23ad mutant contained green-stained (dead) pollen grains (Fig. 4D), and only a few purple-stained pollen grains were observed in some anthers (Supplementary Fig. S4A), consistent with the production of few seeds in atsec23ad. DAPI staining showed that the wild type, atsec23a, and atsec23d generated pollen grains containing two sperm nuclei and one vegetative nucleus, indicating normal development and male mitosis (Fig. 4E–G). In contrast, the majority of pollen grains generated in atsec23ad were smaller and collapsed with no DAPI signals; however, only a few pollen grains possessed the three nuclei (Fig. 4H), consistent with Alexander’s staining results.
To examine whether the phenotype of atsec23a, atsec23d, and atsec23ad is attributed to gametophytic or sporophytic defects, we performed genetic analyses using heterozygous lines. Almost all pollen grains generated from heterozygous lines of both +/atsec23a and +/atsec23d were indistinguishable from the wild-type pollen grains (Supplementary Fig. S5A–F), indicating that the incomplete exine pattern was caused by defects in sporophytic tissues. Furthermore, we observed pollen grains produced by a heterozygous line of atsec23a in the homozygous atsec23d and qrt1-2 background (+/atsec23a, atsec23d/atsec23d, qrt1-2/qrt1-2). qrt1-2 causes incomplete separation of microspores in tetrads, enabling precise segregation analysis of pollen grains derived from a single microsporocyte (Preuss et al., 1994). As revealed by SEM and Alexander’s staining, although the exine pattern showed the phenotype of atsec23d/atsec23d, most pollen tetrads had four uncollapsed pollen grains similar to those from qrt1-2 (Supplementary Fig. S5G–O), indicating that the failure of pollen development in atsec23ad was due to a sporophytic aberration. These results suggested that both AtSEC23A and AtSEC23D function sporophytically.
Next, we performed an in vitro pollen germination assay. Pollen grains of atsec23d showed a germination rate of 82% comparable with the wild-type pollen grains (80.5%), while only 41.6% of atsec23a pollen grains could germinate (Fig. 4I–K, N). The quantitative analysis of pollen germination for atsec23ad was difficult because most pollen grains were collapsed and adhesive to anther walls. However, after forcing pollen grains to separate from anther walls using forceps, very few germinated pollen grains were seen on the germination medium or within the anther (Fig. 4M). Complemented atsec23a lines carrying PAtSEC23A:AtSEC23A-G3GFP showed similar pollen germination rates to that of the wild type (Fig. 4L, N). These results indicated that pollen germination was normal in atsec23d but impaired in atsec23a, and also showed the presence of a few functional pollen grains in atsec23ad, which agreed with the few seeds obtained from atsec23ad. Therefore, these findings indicated that atsec23a produces viable pollen grains that normally undergo pollen mitosis but with a reduced germination.
AtSEC23A and AtSEC23D are highly expressed in the tapetum
Expression of AtSEC23s was anticipated to be universally distributed throughout the whole plant by GENEVESTIGATOR (Robinson et al., 2007). In this study, we analyzed the expression pattern of AtSEC23A and AtSEC23D in detail. By RT-PCR, AtSEC23A was expressed almost equally in all investigated organs except siliques, which showed a weak expression (Fig. 5A). AtSEC23D was expressed in all investigated organs but was most abundant in closed floral buds, and weak in siliques (Fig. 5A). In the promoter:GUS assay, GUS activity in transgenic plants carrying PAtSEC23A:GUS was widely detected in flower parts including sepals, petals, filaments, anther walls, mature pollen grains, pollen tubes, and young siliques (Fig. 5F–J), but not in mature siliques (Fig. 5K). In 5-day-old seedlings, GUS activity was detected in the whole seedling except the hypocotyl (Fig. 5B). In 14-day-old seedlings, GUS activity was observed throughout the whole plant including roots, cotyledons, true leaves, trichomes, and leaf primordia (Fig. 5C–E). In transgenic plants carrying PAtSEC23D:GUS, the GUS activity was detected mainly in floral buds and open flowers (Fig. 5P, Q), mature pollen grains, pollen tubes (Fig. 5Q–S), and fertilized ovules (Fig. 5T), but was not observed in other floral parts including sepal, petal, filaments, anther walls (Fig. 5P–R), and mature siliques (Fig. 5U). In 5-day-old seedlings, no expression was detected (Fig. 5L). In 14-day-old seedlings, only a negligible GUS staining was observed in roots and leaf primordia (Fig. 5M–O).
Next, we followed the expression closely in anthers. GUS signals were observed predominantly in tapetal cells at uninucleate and bicellular stages in PAtSEC23A:GUS and PAtSEC23D:GUS transgenic plants; meanwhile, no or few signals were detected in developing microspores (Fig. 6A, B, D, E). At the tricellular stage, when tapetum degeneration has completed, strong GUS signals were observed in mature pollen grains (Fig. 6C, F). We further traced the expression of both genes in tapetal cells of complemented lines, atsec23a carrying PAtSEC23A:AtSEC23A-G3GFP and atsec23d carrying PAtSEC23D:AtSEC23D-G3GFP. In both cases, GFP fluorescence was detected mainly in the tapetum at the uninucleate stage (Fig. 6G–J). Taken together, although AtSEC23A and AtSEC23D show different expression patterns in the whole plant, both are highly expressed in the tapetum.
AtSEC23A and AtSEC23D are localized to the cytoplasm and ERESs
AtSEC23A has shown ERES localization when co-expressed with SAR1A in A. thaliana protoplast (Zeng et al., 2015). To examine the subcellular localization of AtSEC23D, we transiently co-expressed AtSEC23D–G3GFP with AtSEC23A–TagRFP in N. benthamiana leaves. AtSEC23D–G3GFP exhibited a similar localization pattern to AtSEC23A–TagRFP; both were localized to the cytoplasm, with the existence of many dot-like structures (Fig. 7A; Supplementary Fig. S6A). To confirm that these dot-like structures were ERESs, we used the ERES marker AtSEC24A (Hanton et al., 2007; Wei and Wang, 2008) and the cis-Golgi marker SYP31, which is transported as a cargo to the Golgi apparatus via the secretory pathways (Bubeck et al., 2008). AtSEC24A–TagRFP was co-localized with AtSEC23A–G3GFP and AtSEC23D–G3GFP, and the dot-like structures overlapped with the ERESs labeled by AtSEC24A–TagRFP (Fig. 7B, C; Supplementary Fig. S6B, C). Furthermore, a time-lapse analysis showed that these GFP and RFP signals moved together as a single unit in the cytoplasm (Supplementary Movies S1, S2), confirming that these dots were ERESs. In addition, RFP signal of SYP31–TagRFP was detected in close proximity to the ERESs to which AtSEC23A–G3GFP and AtSEC23D–G3GFP were localized (Supplementary Fig. S7A, B). These results indicated that both AtSEC23A and AtSEC23D exhibit the characteristic localization of COPII, suggesting that they contribute to COPII formation and ER–Golgi vesicle transport.
The significant defects in microspore development of the atsec23ad double mutant appear at the late uninucleate stage
To determine the precise timing of pollen defects, we examined callose wall deposition on tetrads by aniline blue staining. Tetrads of atsec23a, atsec23d, and atsec23ad showed levels of fluorescence comparable with that of the wild type (Supplementary Fig. S8), suggesting that the defects were later than this stage. Next, we analyzed the microspore development from the tetrad to the tricellular stage by semi-thin sectioning. The wild type, atsec23a, atsec23d, and atsec23ad showed no obvious differences in microspore development at the tetrad stage (Fig. 8A–D); however, tapetum was more vacuolated in atsec23ad than in the wild type, atsec23a, and atsec23d (Fig. 8D). At the early uninucleate stage, microspores were normally released from the tetrads in wild-type, atsec23a, atsec23d, and atsec23ad anthers (Fig. 8E–H). At the late uninucleate stage, microspores and tapetal cells of atsec23a and atsec23d showed comparable development with that of the wild type (Fig. 8I–K). In contrast, most microspores in atsec23ad had irregular shapes with abnormal exine walls that mostly detached from the microspore cells (Fig. 8L). At the bicellular stage, microspore development proceeded in wild-type, atsec23a, and atsec23d anthers with no detectable differences (Fig. 8M–O). In contrast, the defects of atsec23ad became more obvious, with a majority of pollen grains misshapen and shrunken, and displaying signs of degeneration (Fig. 8P). Many pollen grains have shown a clear separation of the vegetative and the generative cells with detached exine walls (Fig. 8P; arrowheads), while others had no visible exine walls (Fig. 8P; arrows). The tapetum degeneration was initiated normally in the atsec23a and atsec23d as in wild-type anthers (Fig. 8M–O), while metabolically active swollen tapetal cells with many vacuoles were evident in the atsec23ad anthers, indicating a delayed programmed cell death (Fig. 8P). At the tricellular stage, the pollen mitosis II was completed, and numerous tricellular pollen grains were found in wild-type, atsec23a, and atsec23d anther locules (Fig. 8Q–S). In contrast, only some remnants of tapetal cells, pollen materials, and empty pollen walls were adherent to locule walls in most of the atsec23ad anthers (Fig. 8T), and only a few intact pollen grains were occasionally seen in some anther locules (Supplementary Fig. S4B). These observations indicated that atsec23a and atsec23d showed no detectable defects in this analysis, whereas atsec23ad mutants were defective in exine wall formation, which directly affected the whole process of pollen development and subsequently the seed yield.
Exine and intine formation is impaired in the atsec23ad double mutant
We further analyzed the ultrastructure of microspores and tapetal cells by TEM. In the wild type at the tetrad stage, microspores enveloped in callose wall showed successful plasma membrane undulation and primexine synthesis, followed by probacula deposition (Fig. 9A). Similar processes were observed in atsec23a, atsec23d, and atsec23ad (Fig. 9B–D). At the late uninucleate stage, the wild type, atsec23a, and atsec23d developed a well-defined bacula and tectum, and a thin layer of nexine (Fig. 9E–G). In addition, a primary intine was seen at this stage (Fig. 9E–G; black arrows). In atsec23ad microspores, such typical exine structures were completely missing. Instead, the microspores were surrounded by a fragmented nexine-like layer and many electron-dense semi-spherical sporopollenin-like aggregations (Fig. 9H). The cytoplasm of most microspores showed signs of degeneration with detached pollen walls. At the bicellular stage, wild-type, atsec23a, and atsec23d microspores continued their development with an expanded bacula and thicker tectum (Fig. 9I–K). In addition, the intine became more obvious at this stage (Fig. 9I–K; arrows). In contrast, the atsec23ad microspores showed more damaged pollen walls in which the sporopollenin-like aggregations increased in size and fused together with clear separation from the microspore plasma membrane (Fig. 9L). Some pollen grains were seen as naked cells with no walls (Fig. 9L;Supplementary Fig. S4C; yellow stars), while others had only empty walls without microspore cells (Supplementary Fig. S4C; yellow asterisks), consistent with semi-thin sectioning observations (Fig. 8P). Moreover, the nexine and intine were indistinguishable in the atsec23ad microspores (Fig. 9L). The inner structure of atsec23a and atsec23d microspores was comparable with that of the wild type (Fig. 9J, K). In contrast, most cellular inclusions of atsec23ad microspores degenerated and only a few distinguishable organelles remained (e.g. plastids and mitochondria) (Fig. 9L). Interestingly, the plastids in the incompletely degenerated atsec23ad pollen grains showed profound alteration in their morphologies compared with those of the wild type, atsec23a, and atsec23d (Fig. 9L; arrows). At the tricellular stage, mature pollen walls were completed by deposition of pollen coat materials in the wild-type, atsec23a, and atsec23d pollen grains (Fig. 9M–O). At this stage, wild-type and atsec23d pollen grains developed a thin uniformly distributed intine around the plasma membrane, while this layer was thicker and less electron-dense in atsec23a pollen grains (Fig. 9M–O; Supplementary Fig. S9A–C). Most pollen grains in the atsec23ad anthers degenerated, and flattened pollen shells, remnants of tapetum, and dead pollen materials were left inside the locules (Fig. 9P). Anthers of atsec23ad also contained few undegenerated pollen grains with incomplete pollen walls. Some of them had walls composed of sporopollenin-like aggregations with little pollen coat materials which contained numerous rod-shaped electron-lucent structures (Fig. 9P), while others had walls consisting of elongated bacula that were not covered by tectum and with many depositions of pollen coat materials (Supplementary Fig. S9D, E). These depositions were less electron dense and less compact than those of the wild type, atsec23a, and atsec23d, and contained many electron-lucent vesicle-like structures of different sizes. This suggested incomplete pollen coat depositions and a possible defect in the development of tapetal cells, the main source of pollen coat materials. These results indicated that the depletion of both AtSEC23A and AtSEC23D results in defective pollen wall development and consequently leads to pollen degeneration.
Loss of AtSEC23A and AtSEC23D causes abnormalities in tapetum development
The sporophytic effect on pollen phenotype, apparent defects in microspores at the time of formation of significant exine walls, and the incomplete pollen coat in atsec23ad suggested the possibility of dysfunction of tapetal cells. Thus, we performed TEM observation focused on the endomembrane system and organelles in tapetal cells during different developmental stages. At the tetrad stage, tapetal cells in the wild type, atsec23a, and atsec23d showed several small and few large vacuoles, and contained elongated stacks of rough ER with clear ribosomes and well-defined Golgi bodies (Fig. 10A–C). In contrast, tapetal cells of atsec23ad had smaller vacuoles and clusters of tiny vesicles, and contained rough ER surrounded by less clear ribosomes and abnormal Golgi stacks swollen at their ends and connected with numerous irregular vesicles (Fig. 10D). The tapetum at the late uninucleate stage is characterized by the formation of two specialized storage organelles, the elaioplast and tapetosome. Well-developed elaioplasts with numerous electron-lucent plastoglobules and clusters of small electron-dense tapetosomes were evident in tapetal cells of the wild type, atsec23a, and atsec23d (Fig. 10E–G). In contrast, in atsec23ad tapetal cells, undifferentiated proplastids (precursors of elaioplasts) were observed and the formation of tapetosomes was not yet initiated (Fig. 10H), which indicated a delay in tapetum development. In addition, several sporopollenin-like aggregations were deposited on the middle layer that faces the tapetum (Fig. 10H; arrows), coinciding with the clear defects in microspore wall development (Fig. 9H). At the bicellular stage in the wild type, atsec23a, and atsec23d, tapetal cells continued to develop more abundant mature elaioplasts with definite membranes and fully developed tapetosomes (Fig. 10I–K). The tapetosomes became larger and contained numerous fragmented electron-dense structures. In contrast, the atsec23ad tapetal cells had elaioplasts without definite membranes consisting of fewer and larger plastoglobules (Fig. 10L). In addition, the tapetosomes lost their integrity and had unclear fragmented inner structures. Moreover, deposited sporopollenin-like aggregations on the middle layer became more abundant (Fig. 10L; arrows). At the tricellular stage, tapetal cells completely degenerated in the wild type, atsec23a, and atsec23d (Fig. 10M–O). However, the remnants of tapetal cells were still observed in atsec23ad anthers (Fig. 10P), indicating a delay in tapetum degeneration. These findings suggested that lack of AtSEC23A and AtSEC23D causes structural abnormalities and an incomplete development of tapetal cells.
Discussion
The involvement of SEC23 proteins in regulating plant growth and development, and the functional differences among them have remained unknown. In the present study, we revealed that out of the seven A. thaliana SEC23 homologs, AtSEC23A and AtSEC23D have divergent primary structures from that of other AtSEC23 genes and play essential roles in pollen wall formation, exine patterning, and tapetum development. We also provided evidence for their partially different functions in intine formation and pollen germination.
Knockout of AtSEC23D caused a defect in the microspore exine but did not affect the intine (Fig. 9). Pollen grains of atsec23d exhibited less sporopollenin deposition, incomplete tectum formation, and disrupted reticulate architecture (Fig. 2). In contrast, knockdown of AtSEC23A caused defects in both exine and intine (Fig. 9). The exine of atsec23a pollen grains was also lacking the reticulate architecture and had wide areas with incomplete tectum formation and less sporopollenin deposition (Fig. 2), whereas the intine was thicker and less electron dense (Fig. 9; Supplementary Fig. S9). Moreover, pollen grains of atsec23a showed an impaired germination rate (Fig. 4J, N). Despite the low pollen germination rate, the atsec23a mutant was normally fertile, which may be due to the sufficient amount of functional pollen grains required for fertilization. Mutations in AtSEC24A, AtSEC24B, and AtSEC31B, other members of COPII, have also been shown to cause compromised pollen germination (Conger et al., 2011; Tanaka et al., 2013; Zhao et al., 2016). During pollen germination, the growing pollen tube is encased only by the intine (Chebli et al., 2012). Possibly, the thickened intine in atsec23a is responsible for the impaired pollen germination rate by increasing the required pressure for pollen germination. In Brassica campestris, knockdown of BcMF8 (putative arabinogalactan protein gene) and double knockdown of BcMF26a and BcMF26b (polygalacturonase genes) causes reduced pollen germination and retarded pollen tube growth as a result of remarkable thickening of the intine (Lin et al., 2014; Lyu et al., 2015). In atsec23a, AtSEC23D had the main function in pollen development. However, since AtSEC23D lacks the β-barrel and gelsolin-like domains, it may not be able to complement some of the functions of AtSEC23A including the ER export of such proteins as BcMF8 and BcMF26s, explaining the compromised pollen germination in atsec23a.
The frequent microspore degeneration and leakage observed in atsec23ad locules were linked with loss of the intact pollen wall. Pollen wall integrity must be maintained to ensure normal pollen development. Mutations in genes that affect pollen wall integrity, such as MALE STERILITY1 (MS1), MS2, CALLOSE SYNTHASE5 (CALS5), CYTOCHROME P450 (CYP703A2), RUPTURED POLLEN GRAIN1 (RPG1), DEFECTIVE IN EXINE PATTERNING1 (DEX1), NO EXINE FORMATION1 (NEF1), TRANSIENT DEFECTIVE EXINE (TDE1), ATP-binding cassette transporter G26 (ABCG26), and the COPII component AtSEC31B, have resulted in collapsed pollen grains with degenerated cytoplasm (Aarts et al., 1997; Paxson-Sowders et al., 1997; Wilson et al., 2001; Ariizumi et al., 2004, 2008; Dong et al., 2005; Morant et al., 2007; Guan et al., 2008; Choi et al., 2011; Paxson-Sowders et al., 2001; Zhao et al., 2016). Interestingly, the mutants dex1, nef1, tde1, abcg26, and atsec31b have accumulated sporopollenin-like aggregations on the middle layer similar to the atsec23ad mutant. These sporopollenin-like aggregations indicate an unsuccessful sporopollenin deposition and polymerization onto the microspore plasma membrane. One explanation is that the sporopollenin produced by tapetal cells in atsec23ad lacked components required for correct deposition and polymerization. In the ms1 mutant, the collapsed microspores that tended to stick together were thought to be a result of an unusual chemical composition of pollen wall materials (Vizcay-Barrena and Wilson, 2006). The tendency of atsec23ad microspores to aggregate and stick together may support the above-mentioned explanation. Another possibility is that AtSEC23A and AtSEC23D may mediate transport of signaling proteins necessary for the co-ordination of sporopollenin deposition and polymerization onto the microspore plasma membrane. It was suggested that ABCG26 may transport some signaling molecules required for the co-ordination of exine formation and sporopollenin polymerization on the developing microspore wall (Quilichini et al., 2010). The study of the tde1 mutant has suggested the existence of factors controlling the exine patterning and being potentially active at the uninucleate microspore stage (Ariizumi et al., 2008).
Pollen wall development relies greatly on tapetum development. TEM analysis has revealed the occurrence of morphological abnormalities in organelles of tapetal cells (e.g. ER, Golgi, elaioplasts, and tapetosomes), and a delay in tapetum development and degeneration (Fig. 10). The endomembrane system in atsec23ad tapetal cells was highly affected. Lack of AtSEC23A and AtSEC23D affects the efficiency of COPII assembly, which may ultimately cause aberrant ER and Golgi structures and lead to the appearance of clusters of the tiny vesicles. Several studies have shown that maintaining the integrity of the ER and Golgi apparatus is associated with successful assembly of COPII proteins. For example, a mutation of SAR1 inhibits ER to Golgi trafficking and affects the Golgi integrity (Takeuchi et al., 2000; Ward et al., 2001). A missense mutation in AtSEC24A has resulted in defective ER and Golgi structures and has led to the formation of different sized vesicular clusters (Faso et al., 2009).
In addition to the defective ER and Golgi, elaioplasts and tapetosomes in the atsec23ad tapetal cells exhibited a delayed development and structural abnormalities (Fig. 10). Tapetosomes originate from the ER and are enriched with triacylglycerols, flavonoids, and oleosins (Huang et al., 2013), while elaioplasts develop from proplastids and are mainly filled with steryl esters and polar lipids (Piffanelli et al., 1998; Wu et al., 1999). It was suggested that the ER and plastids may exchange lipids at contact sites of their membrane (Andersson et al., 2007; Fan et al., 2015; Li et al., 2016). The structural abnormalities and the delayed development of elaioplasts and tapetosomes may be a consequence of defective ER structures or a defective lipid exchange between the ER and plastids. Another possibility is that lack of AtSEC23A and AtSEC23D causes less efficient COPII assembly at the ERESs and disrupts transport of unidentified components required for normal elaioplast and tapetosome development. Knockout lines of AtSEC31B also showed a retarded development of elaioplasts and tapetosomes, with ultrastructural abnormalities (Zhao et al., 2016). However, the mechanisms by which COPII components affect the formation of elaioplasts and tapetosomes require further investigation.
The atsec23ad mutant showed abnormal pollen coat depositions which were less electron dense and less compact than those of the wild type and abnormally contained many electron-lucent vesicle-like and rod-shaped structures (Fig. 9P; Supplementary Fig. S9). Similar abnormal depositions were previously observed in atsec31b and abcg9 abcg31 mutants (Choi et al., 2014; Zhao et al., 2016). These abnormal depositions may be a direct result of the abnormal development of elaioplasts and tapetosomes in tapetal cells and/or the delay in tapetum degeneration. Alternatively, it may be due to less efficient ER export of proteins that participate in pollen coat formation. Recently, it was shown that ABCG9, which is involved in the transport of lipidic components required for pollen coat formation (Choi et al., 2014), is probably transported through the ER–Golgi route (Zhao et al., 2016).
In summary, our results showed that AtSEC23A and AtSEC23D are required for proper pollen wall development and exine patterning possibly by mediating efficient ER export of some essential proteins/enzymes participating in pollen wall formation. Our work provides a direct link between the early secretory pathway in tapetal cells and sporopollenin deposition/exine patterning. Moreover, our results indicated that AtSEC23A and AtSEC23D may not fulfill exactly the same functions; AtSEC23D functions only sporophytically in the tapetum to participate in exine formation, whereas AtSEC23A redundantly functions with AtSEC23D in exine formation, and additionally is involved in intine formation and pollen germination. Thus, we propose that AtSEC23A and AtSEC23D may have differing substrate preferences, providing evidence of functional diversity of SEC23 proteins in A. thaliana. Identifying the specific substrate and the interaction partners of AtSEC23A and AtSEC23D will be the basis for a future study.
Supplementary data
Supplementary data are available at JXB online.
Table S1. Oligonucleotides used in this study.
Fig. S1. Multiple sequence alignment of SEC23 family proteins in yeast and A. thaliana.
Fig. S2. Pollen morphology in T-DNA insertion lines for the remaining AtSEC23 genes and in complemented atsec23a and atsec23d lines.
Fig. S3. Normal functionality of the female gametophyte of atsec23ad.
Fig. S4. Phenotypes in the double mutant atsec23ad.
Fig. S5. Sporophytic control of AtSEC23A and AtSEC23D.
Fig. S6. Quantitative co-localization analyses of AtSEC23A and AtSEC23D.
Fig. S7 Co-localization analyses of AtSEC23A and AtSEC23D in N. benthamiana leaf epidermal cells.
Fig. S8. Aniline blue staining of microspores at the tetrad stage.
Fig. S9. Abnormal thickening of the intine of atsec23a pollen grains and defective wall development in atsec23ad pollen grains.
Movie S1. Time-lapse confocal imaging of N. benthamiana leaf epidermal cells co-expressing AtSEC23A–G3GFP and the ERES marker AtSEC24A–TagRFP.
Movie S2. Time-lapse confocal imaging of N. benthamiana leaf epidermal cells co-expressing AtSEC23D–G3GFP and the ERES marker AtSEC24A–TagRFP.
Abbreviations:
- ABRC
Arabidopsis Biological Resource Center
- CLSM
confocal laser scanning microscope
- COPII
coat protein complex II
- ER
endoplasmic reticulum
- ERES
ER exit site
- GUS
β-glucuronidase.
Acknowledgements
We thank the ABRC for the seeds of A. thaliana T-DNA insertion lines and qrt1-2 mutant, RIKEN BRC for Arabidopsis full-length cDNA clones developed by the plant genome project of the RIKEN Genomic Sciences Center, and Kanae Koike (the TEM Service, Center for Gene Science, Hiroshima University) for ultrastructural analyses. This work was supported by a KAKENHI grant from the Japan Society for the Promotion of Science (JSPS) [Grant-in-Aid for Scientific Research (C) No. JP15K07109 to TN].
References
Comments