Four Closely-related RING-type E3 Ligases, APD1–4, are Involved in Pollen Mitosis II Regulation in Arabidopsis

Authors

  • Guo Luo,

    1. State Key Laboratory for Protein and Plant Gene Research, College of Life Sciences, Peking-Tsinghua Center of Life Sciences, Peking University, Beijing 100871, China
    Search for more papers by this author
  • Hongya Gu,

    1. State Key Laboratory for Protein and Plant Gene Research, College of Life Sciences, Peking-Tsinghua Center of Life Sciences, Peking University, Beijing 100871, China
    2. The National Plant Gene Research Center, Beijing 100101, China
    Search for more papers by this author
  • Jingjing Liu,

    Corresponding author
    1. State Key Laboratory for Protein and Plant Gene Research, College of Life Sciences, Peking-Tsinghua Center of Life Sciences, Peking University, Beijing 100871, China
    Search for more papers by this author
  • Li-Jia Qu

    Corresponding author
    1. State Key Laboratory for Protein and Plant Gene Research, College of Life Sciences, Peking-Tsinghua Center of Life Sciences, Peking University, Beijing 100871, China
    2. The National Plant Gene Research Center, Beijing 100101, China
    Search for more papers by this author

Tel: +86 10 6275 3018; Fax: +86 10 6275 3339; E-mail: liujingjing@pku.edu.cn; qulj@pku.edu.cn

Abstract

Ubiquitination of proteins is one of the critical regulatory mechanisms in eukaryotes. In higher plants, protein ubiquitination plays an essential role in many biological processes, including hormone signaling, photomorphogenesis, and pathogen defense. However, the roles of protein ubiquitination in the reproductive process are not clear. In this study, we identified four plant-specific RING-finger genes designated Aberrant Pollen Development 1 (APD1) to APD4, as regulators of pollen mitosis II (PMII) in Arabidopsis thaliana (L.). The apd1 apd2 double mutant showed a significantly increased percentage of bicellular-like pollen at the mature pollen stage. Further downregulation of the APD3 and APD4 transcripts in apd1 apd2 by RNA interference (RNAi) resulted in more severe abnormal bicellular-like pollen phenotypes than in apd1 apd2, suggesting that cell division was defective in male gametogenesis. All of the four genes were expressed in multiple stages at different levels during male gametophyte development. Confocal analysis using green florescence fusion proteins (GFP) GFP-APD1 and GFP-APD2 showed that APDs are associated with intracellular membranes. Furthermore, APD2 had E2-dependent E3 ligase activity in vitro, and five APD2-interacting proteins were identified. Our results suggest that these four genes may be involved, redundantly, in regulating the PMII process during male gametogenesis.

Introduction

Gametogenesis, an essential process for plant sexual reproduction, plays a critical role in the alternation of generations in higher plants. In Arabidopsis thaliana, male gametophyte development is initiated from an archesporial cell, which gives rise to primary sporogenous cells (PSC) and primary parietal cells (PPC) through periclinal divisions. PSCs further differentiate into pollen mother cells (PMCs), which undergo a round of meiosis to produce a tetrad of microspores, which are then released by enzymatic hydrolysis of callose. The free microspores undergo the first asymmetric mitosis, pollen mitosis I (PMI), producing one large vegetative cell and one small generative cell embedded in the cytoplasm of the vegetative cell. The vegetative cell does not divide further but ultimately produces the pollen tube. The generative cell undergoes the second symmetric mitosis (PMII), giving rise to two sperm cells, which are transported through the pollen tube into an ovule for double fertilization (Sanders et al. 1999; McCormick 2004; Liu and Qu 2008).

PMII is a unique process involving asymmetric cell division. In order to clarify the molecular mechanism regulating PMII, mutants that exhibit PMII abnormalities have been identified, mapped, and characterized (Hou et al. 2010). Generative cells in duo1 pollen completed the DNA synthesis phase but failed to enter PMII (Durbarry et al. 2005; Rotman et al. 2005). DUO1 was shown to encode a MYB transcription factor to activate a germline-specific regulon that is essential for sperm cell differentiation (Brownfield et al. 2009a; Borg et al. 2011). The generative cells of duo2 entered PMII but were arrested at prometaphase, leading to the formation of bicellular pollen at anthesis (Durbarry et al. 2005). However, the DUO2 gene has not yet been characterized. Generative cells in duo3–1 pollen either failed to divide into two sperm cells or delayed germ cell division in PMII (Brownfield et al. 2009b). DUO3 has been shown to encode a nuclear protein conserved in land plants and related to GON-4, which is required for normal cell cycle progression in specific lineages during gonadogenesis in Caenorhabditis elegans (Feldman et al. 1997; Brownfield et al. 2009b). A-type cyclin-dependent kinase;1 (CDKA;1) is essential for generative cell division, and its loss of function has been shown to produce bicellular pollen containing a single sperm-like cell and a vegetative cell in anthesis (Imajuku et al. 2001; Verkest et al. 2005; Iwakawa et al. 2006; Dissmeyer et al. 2007). The fbl17 mutant displayed similar pollen phenotypes to those of the cdka;1 mutant. FBL17 formed an SKP1-Cullin1-F-box (SCF) E3 ubiquitin ligase complex (SCFFBL17) that targeted the cyclin-dependent kinase inhibitors Kip-related protein 6 (KRP6) and KRP7 for proteasome-dependent degradation (Kim et al. 2008; Gusti et al. 2009). KRP6 was also found to be degraded by two RING-type E3 ligases RHF1a and RHF2a, and disruption of the RHF1a/2a functions produced some bicellular phenotypes in mature pollen (Liu et al. 2008). Identification of these PMII mutants and characterization of the corresponding genes facilitate our understanding of the PMII process. Although the molecular mechanisms regulating PMII are too complicated to be clarified yet, characterization of fbl17 and rhf1a rhf2a mutants demonstrates that protein ubiquitination is important in order for PMII to proceed.

Protein ubiquitination is a universal biochemical enzymatic reaction, and plays very essential roles in diverse cellular processes in eukaryotes (Pickart 2001; Weissman 2001; Deshaies and Joazeiro 2009). In the ubiquitination reaction, the covalent attachment of ubiquitin (ub) to a target protein depends on three enzymes (Deshaies and Joazeiro 2009). First, ubiquitin is activated by the ubiquitin-activating enzyme (E1) in the presence of adenosine triphosphate (ATP), and then transferred to the ubiquitin-conjugating enzyme (E2) through the formation of a E2-Ub thioester bond. E3 ubiquitin ligase is responsible for substrate-specific recognition, and a ubiquitin is covalently added to a lysine residue of the substrate through the interaction between E2 and E3 ubiquitin ligase. Mono-ubiquitination or multiple mono-ubiquitination primarily serve to regulate protein function, while polyubiquitination typically mediates degradation of targeted proteins through the 26S proteasome (Deshaies and Joazeiro 2009). There are over 1400 E3 ligase genes in the A. thaliana genome (Smalle and Vierstra 2004; Stone et al. 2005). Previous studies showed that E3 ligases participate in diverse cellular processes, for example, hormone signaling, photomorphogenesis, pathogen defense, and gametogenesis (Yamamoto et al. 1998; Osterlund et al. 2000; Yamamoto et al. 2001; Xie et al. 2002; Devoto et al. 2003; Zhang et al. 2007; Kim et al. 2008; Liu et al. 2008; Gusti et al. 2009). Many E3 ligase genes are highly expressed in male gametogenesis (Honys and Twell 2003; Qin et al. 2009), but to date, only two E3 ligases, RHF1a/2a and SCFFBL17, have been functionally characterized (Kim et al. 2008; Liu et al. 2008; Gusti et al. 2009). Hence, the roles of other E3 ligase genes highly expressed in pollen require further investigation.

In this study, we identified four RING-finger E3 ligase genes, APD1–4, which are expressed in pollen at multiple stages. All four APDs carry the predicted transmembrane domains and are localized in intracellular membranes in pollen tubes. The apd1 apd2 double mutant exhibited an increased proportion of bicellular-like pollen at the mature pollen stage. Further knocking down the expression of APD3 and APD4 in apd1 apd2 by RNA interference (RNAi) resulted in more severe bicellular-like pollen phenotypes than apd1 apd2. APD2 was able to mediate E2-dependent protein ubiquitination in vitro. Using yeast two-hybrid screening, five APD2-interacting proteins were identified. Our data show that the APDs may participate in the regulation of PMII in A. thaliana.

Results

APDs are plant-specific RING-finger protein genes

The whole A. thaliana genome was predicted to encode 514 RING-type E3 ligases (data not shown). To identify the RING-finger genes that play critical roles in the A. thaliana reproductive process, we examined the expression profiles of all the RING-finger genes (https://www.genevestigator.com/gv/) and selected those genes with a high and/or specific expression pattern in pollen and sperm cells for further analysis. We found a cluster of four genes that were of particular interest: AT2G38185 (APD1), AT5G01450 (APD2), AT2G38220 (APD3) and AT2G38195 (APD4) (Figure 1A). According to the microarray data, APD1 was ubiquitously expressed except for in sperm cells and pollen, while APD2 was specifically expressed in sperm cells and pollen (data not shown). Furthermore, using the Simple Modular Architecture Research Tool (SMART) (http://smart.embl-heidelberg.de/), all four of these APDs were predicted to contain two transmembrane domains and one C-terminal RING-finger domain (Figure 1A, B). The sequence alignment showed that APD1, 2, 3 and 4 were highly conserved (Figure 1B), especially in the C-terminal RING-finger domain. APD3 is greatly similar to APD4 (96.5% identical at the amino acid level), whereas the similarity among the others ranges from 32.5% to 57.5%. APD1, APD4 and APD3 were tandemly located on chromosome 2 (Figure 1C), implying that they might be from recent duplication events. Basic Local Alignment Search Tool (BLAST) search results showed that the APDs were plant-specific proteins (Figure S1), suggesting that the APDs might be involved in plant-specific processes.

Figure 1.

Protein domains and sequence alignment of APDs in Arabidopsis thaliana.
(A) Phylogenetic relationship between APD1&2&3&4 and schematic diagram of domain organization in APDs. 
(B) Sequence alignment of four A. thaliana APDs. Two transmembrane domains are indicated by orange boxes, and one RING-finger domain is indicated in the blue box. Conserved amino acid residues of the RING-finger domain are indicated by the symbol *. 
(C) The gene loci of the APD1&3&4 genes in chromosome 2.

APDs were expressed with some overlaps in inflorescence

To analyze the expression patterns of the four APDs, we cloned their genomic DNA, including their promoter regions, into a destination vector harboring a β-glucuronidase (GUS) gene to generate pAPDs::APDs-GUS constructs, which were then introduced into wild type A. thaliana, respectively (Figure 2A). More than 10 individual transgenic lines were obtained for each construct, and the T2 generation of the transgenic plants were used for GUS staining analysis. In 10-d-old seedlings, obvious GUS activity was detected in the shoot apical meristem (SAM) and root tip regions for all APDs, and the GUS activity in APD2 transgenic lines was weaker than that in the other three APD transgenic lines (Figure 2B, E, H, K). The detection of high GUS activity in the SAM and root tip regions, where cell division is most vigorous, suggests a close link between the functions of these APD E3 ligases and mitotic cell division. In inflorescence, APD1 was highly expressed in the pistil and in seeds, but weakly expressed during male gametogenesis (Figure 2C, D); APD2 was specifically observed in mature pollen (Figure 2F, G). APD3 was expressed only in young floral buds and seeds (Figure 2I, J), and APD4 was detected in young floral buds, carpels and seeds (Figure 2L, M). The overlapping expression pattern of the APDs in inflorescence suggests that the four APDs may play redundant roles in inflorescence.

Figure 2.

Expression patterns of APDs revealed by GUS (β-glucuronidase) reporter lines.
(A) Diagram of the pAPDs::APDs-GUS constructs. Start codon ATG was as indicated. 
(B–D) Expression pattern of APD1 in seedlings (B), inflorescence (C), and young siliques (D). 
(E–G) Expression pattern of APD2 in seedlings (E) and inflorescence (F). (G) is the magnification of (F). 
(H–J) Expression pattern of APD3 in seedlings (H), inflorescence (I), and young siliques (J). 
(K–M) Expression pattern of APD4 in seedlings (K), inflorescence (L), and young seeds (M). 
Bar = 100 μm in (G), and bar = 1 mm in others.

APD1 and APD2 are associated with intracellular membranes

To analyze the subcellular localization of APDs, we cloned the full-length coding sequences (CDS) of APD1 and APD2, and fused them with the green fluorescent protein (GFP) gene driven by a LAT52 promoter (Twell et al. 1990). LAT52::APDs-GFP was then introduced into wild type A. thaliana, respectively (Figure 3A, B). Both APD1-GFP and APD2-GFP fusion protein fluorescence signals were observed as intracellular membrane sheets in the cytoplasm of the vegetative cells in pollen from transgenic lines (Figure 3C, D), but not in the nuclei of the sperm cells and the vegetative cells. In an in vitro assay, the GFP fluorescence signal was mainly observed in the tonoplast and endosomes in the germinating pollen tubes (Figure 3E, F), suggesting that APD proteins are associated with the intracellular membranes, and may participate in endomembrane-related processes.

Figure 3.

Subcellular localization of APD2 and APD3 in mature pollen and in vitro germinating pollen tubes.
(A, B) Diagram of the LAT52::APD1/2-GFP constructs. Start codon ATG was as indicated. The orange boxes indicate transmembrane domains. Blue boxes indicate RING-finger domains. Black boxes indicate the coding region of genes. 
(C, D) APD1/2-GFP fluorescence in mature pollen. DAPI staining indicates the nuclear positions in pollen. 
(E, F) APD1/2-GFP fluorescence in in vitro germinating pollen tubes. Bar = 10 μm.

Both APD2 and APD3 were highly expressed during male gametogenesis

To determine from which developmental stage APDs were expressed during male gametophyte development, the constructs pAPDs::APDs-GFP were generated for APD1, APD2, and APD3 (Figure 4A–C), and were introduced into wild type A. thaliana, respectively. The APD-GFP signals were observed at multiple stages during male gametogenesis. Consistent with the result that very weak GUS activity was detected in floral buds and pollen from pAPD1::APD1-GUS transgenic lines (Figure 2), a very weak florescence signal was detected for APD1 from the late bicellular pollen stage to the mature pollen stage (Figure 4D). An APD2-GFP fluorescence signal was clearly detected from the early bicellular pollen stage to the mature pollen stage, and the signal in vegetative cells increased with the development of pollen (Figure 4E). No obvious fluorescence signal was detected for APD3 in the microspore, the early bicellular pollen stage, or the middle bicellular pollen stage (Figure 4F). Instead, the APD3-GFP florescence signal became stronger at the late bicellular pollen stage, and gradually strengthened from the tricellular pollen stage to the mature pollen stage (Figure 4F). These observations suggest that APD1, APD2 and APD3 may be involved in male gametogenesis.

Figure 4.

Expression patterns of APD1&2&3 during male gametogenesis.
(A–C) Diagram of the pAPD1/2/3::APD1/2/3-GFP constructs. Start codon ATG was as indicated. 
(D–F) APD1 (D)/2 (E)/3 (F)-GFP fluorescence during pollen development. DAPI staining indicates the nuclear positions in pollen. Bar = 10 μm.

PMII was blocked in male gametogenesis in apd1 apd2 and higher mutants

To analyze the possible function of APD1 and APD2, we requested from the Arabidopsis Biological Research Center (ABRC) their T-DNA insertion homozygous mutants, apd1 (SALK_076267) with a T-DNA inserted in the third exon of APD1, and apd2 (SALK_102963) with a T-DNA inserted in the sixth exon of APD2 (Figure 5A). Reverse transcription-polymerase chain reaction (RT-PCR) analysis showed that no full-length transcripts of APD1 or APD2 were detected in apd1 and apd2, respectively (Figure 5B). We then crossed apd1 with apd2 to obtain the apd1 apd2 double mutant. No obvious defects were observed in apd1, apd2 or apd1 apd2 during the vegetative developmental stage. However, when staining pollen with 4′6′-diamidino-2-phenylindole dihydrochloride (DAPI), we found that there were significantly more bicellular-like pollen grains in the apd1 apd2 double mutant than in the wild type at the mature pollen stage (Figure 5C). In the bicellular-like pollen at the dehiscence stage in apd1 apd2, the vegetative cells developed normally, while the other one cell was larger and brighter than the wild type sperm cell, which probably resulted from more highly condensed chromatin and/or blocked cell division (Figure 5C). The fact that the percentage of bicellular-like pollen at the mature pollen stage was higher in apd1 apd2 than in either apd1 or apd2 (Figure 5E) suggests that APD1 and APD2 may be functionally redundant in PMII during male gametogenesis.

Figure 5.

Pollen development in wild type, apd mutants, and RNAi transgenic lines at the anthesis stage revealed by DAPI staining.
(A) Schematic diagram of the APDs gene structures and the T-DNA insertion sites in APD1&2. Start codon ATG and stop codon TAA, and primers used in reverse transcription-polymerase chain reaction (RT-PCR) were as indicated. 
(B) Transcript expression levels of APD1 and APD2 in apd1, apd2 and apd1 apd2 mutants. TUBULIN2 was used as an internal control. 
(C) Mature bicellular-like pollen phenotypes at the anthesis stage in wild type, apd mutants, and RNAi transgenic lines. Bar = 10 μm. 
(D) Diagram of the pDUO1::APD3-RNAi construct. The APD3 CDS region used for the RNAi construct was as indicated. 
(E) Statistical analysis of the percentage of bicellular-like pollen at the mature pollen stage in wild type, apd1, apd2, apd1 apd2 and four RNAi transgenic lines. The three stars on top of the columns indicate a significant difference from the wild type. Asterisks indicate statistically significant P values at P < 0.001. 
(F) Expression level of APDs in four independent apd2 apd3; pDUO1::APD3-RNAi transgenic lines. The primers used in the RT-PCR are indicated in (A).

At the mature pollen stage, although the percentage of bicellular-like pollen in the apd1 apd2 double mutant was significantly higher than in the wild type, the bicellular-like pollen population was not very large compared with that of the tricellular pollen in apd1 apd2. The low percentage of bicellular-like pollen in apd1 apd2 may result from the functional redundancy between APD1/2 and APD3/4 that were tandemly located with APD1 in the chromosome (Figure 1C). To test this hypothesis, we generated an APD3-RNAi construct driven by a DUO1 promoter, which was specifically expressed in the generative cells and sperm cells (Durbarry et al. 2005; Rotman et al. 2005; Brownfield et al. 2009a; Borg et al. 2011) (Figure 5D), and introduced it into apd1 apd2 plants. As a result, we found more bicellular-like pollen in different independent lines. In addition, we observed more abnormal tricellular-like pollen in some lines, in which two sperm cell-like nuclei displayed a dramatic difference in both size and brightness (Figure 5C). Statistical analysis on four independent transgenic lines showed that the percentages of bicellular-like pollen in these APD3-RNAi/apd1 apd2 lines were 4.33%, 6.40%, 4.55% and 26.44% respectively, and higher than 2% in apd1 apd2 (Figure 5E, Table 1). RT-PCR analysis showed that the RNA transcript level of APDs correlated well with the percentage of bicellular-like pollen in the four RNAi/apd1 apd2 lines (Figure 5F), suggesting that APD1, 2, 3 and 4 play redundant but essential roles in the process of PMII during male gametogenesis.

Table 1.  Statistical analysis of bicellular-like pollen from wild type, apd mutants, and RNAi transgenic lines at the mature pollen stage
PollenWild type apd1 apd2 apd1 apd2 apd1 apd2; pDUO1::APD3-RNAi
Line-1Line-2Line -6Line-18
  1. NA, not applicable.

No. of counted pollen72786176725451044169065722160
No. of bicellular-like pollen25506499185826571
Percentage of bicellular-like pollen (%)0.340.810.881.944.336.404.5526.44
P-valueNA2.47E-048.71E-057.21E-222.99E-265.03E-662.08E-33∼0

APD2 possessed E2-dependent E3 ligase activity in vitro

To test whether APDs have E3 ligase activity, we expressed in Escherichia coli and purified N-terminal tagged maltose binding protein (MBP) MPB-APD2 protein for an E3 ligase activity assay. Wheat (Triticum aestivum) ubiquitin-activating E1, human ubiquitin-conjugating E2 (Ubch5b), and His-ub, with a 6 × His tag fused at its N-terminus, were also expressed in E. coli and purified for the assay. As shown in Figure 6A, in the presence of E1, E2 and His-ub, a polyubiquitination signal was observed by western blot using an anti-His antibody, indicating that MBP-APD2 was able to catalyze self-ubiquitination. Furthermore, in the absence of E1 or E2, no polyubiquitination ladder was detected, suggesting that the E3 ligase activity of APD2 was E1 and E2-dependent (Figure 6A).

Figure 6.

E3 ubiquitin activity of APD2 and interacting proteins of APD2 revealed by yeast two-hybrid screening.
(A) E3 ubiquitin ligase activity of APD2. Maltose binding protein (MBP)-APD2 fusion proteins were tested for E3 activity in the presence of E1 (from wheat), E2 (UBCh5b), and 6 × His tag ubiquitin. The numbers on the left indicate the molecular weight of marker proteins in kilodaltons (kD). MBP protein itself was used as a negative control in the assay. An anti-His antibody was used to detect the 6 × His tagged ubiquitin. 
(B) Several interactors of APD2 were identified through yeast two-hybrid screening.

Five APD2-interacting proteins were found by yeast two-hybrid screening

To identify the possible targets of APD2 which are associated with intracellular membranes,we performed yeast two-hybrid (YTH) screening with a DUALmembrane system. Out of 5 × 106 clones, we found five APD2-interacting proteins (Figure 6B). Two of these APD2-interacting proteins, AT1G75630 and AT2G25610, which are parts of subunit C of the V-type ATPase that catalyzes ATP hydrolysis to transport protons, were preferentially expressed in sperm cells and mature pollen (data not shown) (https://www.genevestigator.com/gv/). Furthermore, a Rab subfamily homolog E1b (AT4G20360), which is involved in vesicle trafficking, was also able to interact with APD2. The other two APD2-interacting proteins include an S-ribonuclease binding protein (SBP) family protein (AT1G10650) that was specifically expressed in pollen, and a pollen Ole e 1 allergen and extensin family member (AT1G78040), both of which are functionally uncharacterized. Identification of these possible APD-interactors will help clarify our understanding of the biological roles of APDs in PMII during male gametogenesis.

Discussion

Double fertilization is specific to angiosperms, and plays an essential role in plant reproduction and life cycle alternation. It depends strictly on the fusion events between two sperm cells in germinating pollen with an egg cell and a central cell, respectively. The two sperm cells were produced by generative cells through PMII in A. thaliana. Among the known E3 ligases that are involved in PMII, cytoplasmic RHF1a/RHF2a and SCFFBL17 have been reported to target the protein degradation of cell division inhibitor KRP6 (Kim et al. 2008; Liu et al. 2008; Gusti et al. 2009). In this study, we identified four RING-finger E3 ligase genes, loss-of-function of which resulted in mature bicellular-like pollen. This suggests that PMII has been blocked and the subsequent double fertilization might be disrupted. Because the four APDs contain predicted transmembrane domains and APD1 and 2 are localized on intracellular membranes, we speculate that APDs may be involved in pollen development regulation through a different cellular mechanism.

Because the gene loci of APD1, APD4 and APD3 are tandemly arranged on chromosome 2, it is difficult to obtain triple or quadruple homozygous mutants through crossing. Although we obtained the APD3-RNAi transgenic lines, which exhibited more severe pollen phenotypes than did apd1 apd2, the remaining transcripts of APD3 and APD4 were still detected (Figure 5F), which is consistent with the fact that the APD3-RNAi transgenic plants were fertile and viable. In addition to the four APDs, there was another APD member, APD5, with a distinct signal peptide at its N-terminus and only one transmembrane domain (Figure S1), suggesting that APD5 might have a different subcellular localization and may be involved in different biological processes. This hypothesis is deserving of future investigation.

APD1 was predicted to interact with a member of the glutamate receptor-like channels (GLRs) family, and with TEOSINTE BRANCHED, CYCLOIDEA AND PCF (TCP) 14 in the Arabidopsis Interactions Viewer (http://bar.utoronto.ca/interactions). GLRs facilitate Ca2+ influx across the plasma membrane, modulate apical [Ca2+]cyt gradient, and consequently affect pollen tube growth and morphogenesis (Michard et al. 2011). The transcription factor TCP14 has been shown to regulate seed germination and embryonic growth (Tatematsu et al. 2008; Kieffer et al. 2011; Rueda-Romero et al. 2012). These interactions need to be confirmed in planta. Unfortunately, GLR members and TCP14 were not among the proteins identified to interact with APD2. The APD2-interacting proteins screened through YTH may participate in the hydrolysis of ATPase, vesicle trafficking, hypersensitivity reaction, and/or zinc ion binding. For example, as a member of the pollen allergen and extensin family, Ole e 1 is the major allergen in olive (Olea europaea L.) pollen, and is possibly associated with pollen hydration and pollen tube growth (Alche et al. 2004). Another olive allergen protein, Ole e 9, was the first 1, 3-β-glucanase detected in pollen tissue (Huecas et al. 2001). However, the function of pollen Ole e 1 in A. thaliana has not yet been clarified. In A. thaliana, there are 28 members in the pollen Ole e 1 allergen and extensin family. Most of these are predicted to be localized on the endomembrane, and only one of them, arabinogalactan-protein 30 (AGP30), had been functionally characterized (van Hengel and Roberts 2003). AGP30 was reported as an arabinogalactan-protein in the cell walls of the primary root, and plays a role in root regeneration and seed germination. Analyzing the relationship between APD2 and the members of the pollen allergen and extensin family would elucidate their functions in A. thaliana. In addition, one Rab GTPase was identified to interact with APD2. The Rab subfamily of small GTPases consists of 57 members in A. thaliana, most of which are involved in vesicle trafficking (Vernoud et al. 2003). The interaction between APD2 and the Rab homologue E1b implies a possible link between E3 ligases and small GTPases in plants, which was previously reported only in animals (Wilkins et al. 2004; Moren et al. 2005; Jura et al. 2006; Schwamborn et al. 2007; Berthold et al. 2008; Kim et al. 2009; Kawabe et al. 2010).

In conclusion, we provide evidence to show that the four APDs, APD1, 2, 3 and 4, play important but redundant roles in the PMII process during male gametogenesis. Identification of five putative APD-interactors will help elucidate the biological roles of APDs in PMII in the future.

Materials and Methods

Plant materials and growth conditions

Arabidopsis thaliana (L.) plants in this study were all of the Columbia-0 ecotype and were grown at 22 °C under 16 h light/8 h dark conditions. The apd1 (SALK_076267) mutant and the apd2 (SALK_102963) mutant were obtained from the Arabidopsis Biological Resource Center (ABRC). The apd1 apd2 mutant was obtained by genetic crosses. Seeds were surface-sterilized with 15% (v/v) sodium hypochlorite, and washed five times with sterile water. Sterile seeds were plated on 1/2 Murashige and Skoog (MS) medium plus 1% sucrose and 0.65% agar. The pH of the medium was adjusted to 5.8 with 1M KOH. Plates were placed in darkness for 2 to 4 d at 4 °C, and then transferred to a chamber at 22 °C under a 16 h light/8 h dark photoperiod. After 7–9 d, seedlings were sowed in soil and placed in a green house at 22 °C under a 16 h light/8 h dark photoperiod. Plants were transformed with Agrobacterium tumefaciens (GV3101) using a standard floral dipping method (Clough and Bent 1998). Transformants were selected on 1/2 MS medium containing 50 μg/mL kanamycin or phosphinotricine (PPT).

Genotype determination

Homozygous mutants were identified through PCR from genomic DNA and further analyzed by DNA sequencing to confirm the insertion of T-DNA (Hou et al. 2011). The primer pairs of 076267-F1 (5′-CTTATCTGGTTCTTTGGTTTGTTTC-3′) and 076267-R1 (5′-AATCTTTTAGGTTTTTCGTTGTGC-3′), and the primer pairs of T-DNA left border primer LBb1.3 (5′-ATTTTGCCGATTTCGGAAC-3′) and 076267-R1 were used to amplify the APD1 wild type allele and the apd1 mutant allele, respectively. The primer pairs of 102963-F1 (5′-GGTCACCTTCATTGTTGTTTTCT-3′) and 102963-R1 (5′-ACTGGTTCGTAAGTTTTATTCTCG-3′), and the primer pairs of LBb1.3 and 102963-R1 were used to amplify the APD2 wild type allele and the apd2 mutant allele, respectively. Because there were two neighboring inverted T-DNAs in the apd2 mutant, either of the LBb1.3 pairs and the forward or reverse primers could be used in genotyping analysis.

Total RNA isolation and RT-PCR analysis

Total RNA was isolated from the whole inflorescence using a TransZol Up regent (TransGen, Beijing, China) and then digested with RNase-free DNaseI (Takara, Tokyo, Japan) according to the manufacturer's instructions. RNA concentration was measured by spectrophotometer, and 5 μg total RNA was used to synthesize cDNA with a M-MLV RTasecDNA Synthesis Kit (Takara) according to the manufacturer's instructions. To examine the expression level of APDs, RT-PCR was performed (Luo et al. 2011). The primers were used as follows: for APD1, F1, 5′-AGCATTAAGGTTAAAGTGCTTGATT-3′ and R1, 5′-ATGTGAGATAGATAACACTCGGGAT-3′; APD1-FL-F1, 5′-ATGGATTCAGCTCCTTCTCC-3′ and APD1-FL-R1, 5′-GACATTGAAGATCTTCTTGACCTTC-3′; for APD2, F2, 5′-TTGCATACATAGTTTGCATGGG-3′ and R2, 5′-GTCAAAATAGAACTGTAGGACGAGC-3′; F3, 5′-GCTACCTTCTGTATCTTTGTGTCG-3′ and R3, 5′-TCCCTTGTTCCCTAATTCTTCC-3′; APD2-FL-F1, 5′-CAAACCTAAGGTTAGGGTTTTACC-3′ and APD2-FL-R1, 5′-CGGATATGGTAATGCGAGGA-3′; for APD3, F4, 5′-GAGGAAGGACATAGAGGTGGAA-3′ and R4, 5′-ATGGATACTCCCTGACCCAGT-3′; for APD4, F5, 5′-GCTAACTTGAAGAGGAAGGACG-3′ and R5, 5′-ATGGATACTCCCTGACCCAGT-3′. TUBULIN2 (for β-tubulin) was used as an internal control, and the primer pairs of TUB2-F1 (5′-GTTCTCGATGTTGTTCGTAAG-3′) and TUB2-R1 (5′-TGTAAGGCTCAACCACAGTAT-3′) were used.

Vector construction

Most vectors were generated using single Gateway recombination (Invitrogen, Carlsbad, CA, USA) as previously described (Brownfield et al. 2009a). Some entries were generated by TOPO cloning (Invitrogen) according to the manufacturer's instructions. For the constructs of pAPD1/2::APD1/2-GUS, the destination vector pCB308R was used (Lei et al. 2007). The primers were used as follows: for APD1, 2G38185-pag-B1F1, 5′-AAACTCGTGTCATTTTCCTCACT-3′ and 2G38185-pag-B2R1, 5′-CACTGTATAGATTCGCTTCACTTTC-3′; for APD2, 5G01450-pag-B1F1, 5′-CCCTTAGATATGATGGTGTACCTTT-3′ and 5G01450-pag-B2R1, 5′-GACATTGAAGATCTTCTTGACCTTC-3′. For the constructs of pAPD3/4::APD3/4-GUS/GFP, the destination vector pKGWFS7 was used. The primers were used as follows: for APD3, 2G38220-pag-B1F1, 5′-TTCTCCGTGAAAGTTGGTCTC-3′ and 2G38220-nostop-B2R1, 5′-AGCTGTATAGATTCGCTTCACATGT-3′; for APD4, 2G38195-pag-B1F1, 5′-TGCAAGAAGCTGATGGGAGTT-3′ and 2G38195-nostop-B2R1, 5′-AGCTGTATAGATTCGCTTCACAT-3′. For the constructs of LAT52::APD1/2-GFP, the destination vector pK7FWGLAT52 was used. The cDNA of APD1 was amplified from wild type A. thaliana through RT-PCR with the primers APD1-CDS-F1 (5′-TTCTCCTCGTCACAGATTCTCAG-3′) and APD1-CDS-R1 (5′-CACTGTATAGATTCGCTTCACTTTC-3′). The cDNA of APD2 was amplified from wild type A. thaliana through RT-PCR with the primers APD2-CDS-F1 (5′-CAAACCTAAGGTTAGGGTTTTACC-3′) and APD2-CDS-R1 (5′-GACATTGAAGATCTTCTTGACCTTC-3′). The vector pK7FWGLAT52 was modified from pK7FWG2 through replacement of the 35S promoter region with a fragment of the LAT52 promoter. The primers LAT52-F1 (5′-TCAGGAGCTCTTGAGGAATGATCGATTCTGG-3′) and LAT52-R1 (5′-CCATACTAGTGAATTTTTTTTTTGGTGTGTG-3′) were used to amplify the LAT52 promoter sequence. For the construct of pDUO1::APD3-RNAi, the destination vector pCB2004D was used (Lei et al. 2007). The primers pDUO1-F1 (5′-AGAAGAGCTCGTCCGAAGTTTCCCTCTTGG-3′) and pDUO1-R1 (5′-TTTTACTAGTCGCTAATCGATCTCTCTCTCG-3′) were used to amplify the DUO1 promoter region. pCB2004D-pDUO1 was constructed through a ligation of the SacI/SpeI-digested fragment from PCR-amplified DUO1 promoter and the SacI/SpeI-digested vector pCB2004D. The primers RNAi-F1 (5′-CTGTTGAGCTTGTCCTGCTT-3′) and RNAi-R1 (5′-GTTGCAGAAGTGTATAGCTACCA-3′) were used for amplification of the APD3-RNAi CDS region.

GUS assays

Histochemical GUS assays were performed as previously reported (Qin et al. 2005; Zhang et al. 2011). Imaging was taken using a M205C FA stereo microscope (Leica, Germany).

DAPI staining

Detached anthers were collected from open flowers with forceps and mounted directly into DAPI solution (0.1 M sodium phosphate, pH 7.0, 1 mM ethylenediaminetetraacetic acid (EDTA), 0.1% Triton X-100, and 0.5 mg/mL DAPI) on the microscope slide. The pollen was scattered by gently poking the anthers back and forth. The anthers were then removed. Developmental analysis was performed as previously described (Borg et al. 2011; Wang et al. 2012).

Confocal microscopy analysis and imaging

Pollen was sampled and mounted in DAPI solution (Wang et al. 2012). Fluorescence and imaging were performed using a TCS SPE for confocal laser scanning microscopy (CLSM) (Leica, Germany). Images were processed using Adobe Photoshop CS2 software (Adobe Systems).

Yeast two-hybrid screening

The APD2 two-hybrid screening was performed using the DUALmembrane kit 3 (Dualsystems) according to a previously-reported protocol (Ou et al. 2011). The cDNA of APD2 was amplified with the primers 5G01450-SfiI-F1 (5′-TTTGGCCATTACGGCCATGTCTCTACCGGATTCTCTCC-3′) and 5G01450-SfiI-R1 (5′-TTTGGCCGCCTCGGCCTTAGACATTGAAGATCTTCTTGACC-3′), and cloned into the EcoRV site of pBluescript SK+, designated as pBS-APD2-SfiI. The bait construct of pBT3-N-APD2 was constructed by the ligation of the SfiI-digested fragment from pBS-APD2-SfiI and the SfiI-digested vector pBT3-N. The bait construct of pMetYCgate-APD2 was constructed as previously described (Obrdlik et al. 2004). The primer pairs of 5G01450-y2h-F1 (5′-ATGTCTCTACCGGATTCTCTCC-3′) and 5G01450-y2h-R1 (5′-GACATTGAAGATCTTCTTGACCTTC-3′) were used for the amplification of the APD2 CDS region for the construction of pMetYCgate-APD2.

E3 ligase activity assay in vitro

The cDNA of APD2 was amplified with the primers 5G01450-EXP-F1 (5′-CCGGAATTCATGTCTCTACCGGATTCTCTC-3′) and 5G01450-EXP-R1 (5′-CGCGTCGACTTAGACATTGAAGATCTTCTT-3′) and cloned into the EcoRV site of pBluescript SK+, designated as pBS-APD2-EXP. The entire APD2 open reading frame region was cloned into the pMAL-c2x vector (New England Biolabs, Ipswich, MA, USA) by the ligation of the EcoRI/SalI-digested fragment from pBS-APD2-EXP and the EcoRI/SalI-digested vector pMAL-c2x. The fusion proteins were purified, and the in vitro E3 ligase assay was performed as previously reported (Xie et al. 2002; Zhang et al. 2007).

Bioinformatics analysis

The non-redundant protein database was searched using the full length of APD2 as the query through the online Basic Local Alignment Search Tool (BLAST), with default settings (http://blast.ncbi.nlm.nih.gov/). According to the prediction of the APD2 protein structure, 31 protein sequences were selected including APD1, 2, 3 and 4 from the BLAST results. Through sequence alignment, a phylogenetic tree was constructed using the neighbor-joining method, and the proteins from Physcomitrella patens subsp. Patens were made to be the root. The domain organization of each protein was illustrated with Adobe Illustrator CS2 software (Adobe Systems). The APD1, 2, 3 and 4 sequence alignment was performed using ClustalX software (v.1.83). The phylogenetic tree was constructed using MEGA software (v.5.05) with neighbor-joining statistical method. Some selected options were as follows: Test of Phylogeny, Bootstrap Method; No. of Bootstrap Replications, 1000; Model/Method, P-distance; and Gaps/Missing Data Treatment, Pairwise Deletion.

(Co-Editor: Chun-Ming Liu)

Acknowledgements

We would like to thank Dr. Qi Xie (Institute of Genetics and Developmental Biology, the Chinese Academy of Sciences) for the DUALmembrane yeast two-hybrid system, and Professor Chengbin Xiang (School of Life Sciences, University of Science and Technology of China) for providing pCB308R and pCB2004D. This work was supported by the National Key Basic Research Program of the People's Republic of China (2011CB915402), and the National Science Foundation of China (30970270 to JJL).

Ancillary