This Article Abstract Full Text (PDF) All Versions of this Article: 18/11/3015    most recent tpc.106.042044v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (6) Citing Articles via Google Scholar Google Scholar Articles by Jung, K.-H. Articles by An, G. Search for Related Content PubMed PubMed Citation Articles by Jung, K.-H. Articles by An, G. Agricola Articles by Jung, K.-H. Articles by An, G.

Wax-deficient anther1 Is Involved in Cuticle and Wax Production in Rice Anther Walls and Is Required for Pollen Development^[W]

^a National Research Laboratory of Plant Functional Genomics, Division of Molecular and Life Sciences, Pohang University of Science and Technology, Pohang 790-784, Republic of Korea
^b Institute of Cellular and Molecular Botany, University of Bonn, D-53115 Bonn, Germany
^c Department of Molecular Genetics, Max Planck Institute for Breeding Research, D-50829 Köln, Germany
^d Center for Plant Intracellular Trafficking and Division of Molecular and Life Sciences, Pohang University of Science and Technology, Pohang 790-784, Republic of Korea
^e Functional Genomic Center, Pohang University of Science and Technology, Pohang 790-784, Republic of Korea

² To whom correspondence should be addressed. E-mail genean{at}postech.ac.kr; fax 82-54-279-0659.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
In vegetative leaf tissues, cuticles including cuticular waxes are important for protection against nonstomatal water loss and pathogen infection as well as for adaptations to environmental stress. However, their roles in the anther wall are rarely studied. The innermost layer of the anther wall (the tapetum) is essential for generating male gametes. Here, we report the characterization of a T-DNA insertional mutant in the Wax-deficient anther1 (Wda1) gene of rice (Oryza sativa), which shows significant defects in the biosynthesis of very-long-chain fatty acids in both layers. This gene is strongly expressed in the epidermal cells of anthers. Scanning electron microscopy analyses showed that epicuticular wax crystals were absent in the outer layer of the anther and that microspore development was severely retarded and finally disrupted as a result of defective pollen exine formation in the mutant anthers. These biochemical and developmental defects in tapetum found in wda1 mutants are earlier events than those in other male-sterile mutants, which showed defects of lipidic molecules in exine. Our findings provide new insights into the biochemical and developmental aspects of the role of waxes in microspore exine development in the tapetum as well as the role of epicuticular waxes in anther expansion.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Waxes are major constituents of the cuticle, a hydrophobic barrier covering the aerial portions of land plants. The primary role of epicuticular waxes is protection from water loss (Riederer and Schreiber, 2001), but they are also believed to play important roles in plant defenses against bacterial and fungal pathogens (Jenks et al., 1994). Plant waxes predominantly comprise alcohols, aldehydes, ketones, alkanes, and esters derived from very-long-chain fatty acids (VLCFAs), ranging in chain length from C20 to C34 (Kunst and Samuels, 2003). Core biosynthetic activities to extend the acyl chain to the desired length are the sequential extensions performed by fatty acid elongation (FAE) complexes, involving four enzymatic reactions. Several rate-limiting FAE condensing enzymes exist, including ß-KETO ACYL-COA SYNTHASE1 (KCS1), ECERIFERUM6 (CER6), FIDDLEHEAD, and Le CER6, which have been implicated in wax biosynthesis (Todd et al., 1999; Fiebig et al., 2000; Pruitt et al., 2000; Vogg et al., 2004). Molecular characterization of the maize (Zea mays) glossy8 (gl8) mutant has resulted in the identification of another component of the FAE complex, which catalyzes 3-ketoacyl reductase reaction (Xu et al., 1997, 2002; Dietrich et al., 2005). Disruption of the Arabidopsis thaliana Enoyl-CoA reductase (ECR) gene, which is another FAE component, causes a reduction in the cuticle wax load (Zheng et al., 2005). VLCFAs are converted to other wax components through either an acyl reduction pathway or a decarbonylation pathway. The former gives rise to primary alcohols and wax esters, and the latter leads to the formation of aldehydes, alkanes, secondary alcohols, and ketones (Millar et al., 1999; Kunst and Samuels, 2003). In the garden pea (Pisum sativum), alcohol formation from VLCFA precursors is performed by FATTY ACYL-COA REDUCTASE (FAR) (Vioque and Kolattukudy, 1997). The FAR-related Arabidopsis gene, MALE STERILITY2 (MS2), encodes a tapetum-specific protein essential for pollen fertility (Aarts et al., 1997). Biochemical studies of TRITICUM AESTIVUM ANTHER1, a wheat (Triticum aestivum) anther-specific homolog, supports the notion that MS2 might act as a FAR in the formation of fatty alcohols required for wax ester production in the tapetum (Wang et al., 2002).

Stem wax analysis of the cer1 mutant has shown a strong reduction in products of the decarbonylation pathway (e.g., alkanes, secondary alcohols, and ketones) accompanied by an increase in aldehydes (Aarts et al., 1995). CER1 shares considerable sequence similarity with maize GL1, which is proposed to be a membrane-bound protein affecting wax biosynthesis (Hansen et al., 1997). CER1 and GL1 appear to perform a different role in cuticular wax biosynthesis. Further work is needed to resolve the functional identities of these proteins.

Cuticular waxes are embedded and overlaid on a polyester matrix of cutin (Nawrath, 2003). The combination of cutin, waxes, and possibly polysaccharides forms the cuticle (Jeffree, 1996; Kolattukudy, 1996). Cutin plays an important role in protecting organisms from water loss, UV irradiation, frost damage, and attacks from pathogens and insects. Cutin is predominantly built from C16 and C18 aliphatics, typically ester-linked via carboxy and -hydroxy groups of individual fatty acid derivatives. These monomers frequently contain additional hydroxyl and epoxy groups in the midchain positions. Although a biosynthesis pathway of cutin monomers has been proposed, based on the analysis of the cutin composition of cell surfaces fed with radiolabeled cutin monomer precursors, very little is known about this unique plant process (Kolattukudy, 1996).

In angiosperm species, the anther wall has four layers: outermost epidermis, endothecium, middle layer, and innermost tapetum. Whereas the epidermis is important in protecting an organism from various environmental stresses and pathogen attacks, the tapetum plays a critical role in pollen development and maturation. Pollen grains are covered by lipid-derived structures that are essential for pollen dispersal, pollen–stigma communication, and pollen rehydration (Piffanelli et al., 1997, 1998). The pollen wall comprises three layers: pollen coat, outer exine layer, and inner intine layer. The highly sculpted outer layer exine is largely composed of sporopollenin, which is resistant to degradation and withstands acetolysis (Scott, 1994). Sporopollenin is made up of polymers derived from VLCFAs and their derivatives plus more modest amounts of oxygenated aromatic rings and phenylpropanoids (Guilford et al., 1988; Wehling et al., 1989; Wiermann and Gubatz, 1992; Wilmesmeier et al., 1993; Ahlers et al., 1999; Meuter-Gerhards et al., 1999).

Functional and biochemical roles of the exine have been studied using mutants defective in pollen wall development and exine deposition, such as defective in exine formation1 (dex1), ms2, no exine formation1, faceless pollen1, and cer6 mutants in Arabidopsis (Aarts et al., 1997; Fiebig et al., 2000; Paxson-Sowders et al., 2001; Ariizumi et al., 2003, 2004; Kunst and Samuels, 2003). In cereal crops, wheat and rice (Oryza sativa) RAFTIN proteins are localized at orbicule and exine and are essential for the maturation process of pollen development, but their biochemical characteristics have not been determined (Wang et al., 2003).

Here, we describe a male-sterile mutant of rice, wax-deficient anther1 (wda1), that was generated through T-DNA insertional mutation (Jeon et al., 2000; Jung et al., 2003, 2005; Lee et al., 2003, 2004). This mutant shows morphological changes in its pollen and anther walls as well as alterations in its wax composition. Wda1 encodes a protein with high similarity to proteins involved in wax production (Aarts et al., 1995; Hansen et al., 1997; Chen et al., 2003). Our biochemical analysis of the wda1 anthers demonstrated that the levels of very-long-chain alkanes, alkenes, fatty acids, and primary alcohols are severely reduced in mutant anthers. This study revealed the role of VLCFAs in the development of the tapetum and the epidermal layer in the anther wall.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Isolation of a Male-Sterile Mutant from T-DNA–Tagged Lines
We previously reported the generation of T-DNA insertional lines in Japonica rice (Jeong et al., 2002; Jung et al., 2003, 2005). The T-DNA contains a ß-glucuronidase (GUS) reporter gene next to the T-DNA border. Its insertion within a gene can generate a fusion between the GUS reporter gene and an endogenous target gene. This feature allows us to identify a group of genes that are expressed in a particular tissue or organ. Among the 270 lines that show GUS activity in their anthers, 15 manifest male-sterile phenotypes that cosegregate with GUS (Jung et al., 2005).

Here, we report the detailed analysis of line 0-208-29. In the T2 progeny, the GUS-positive:GUS-negative ratio was 3:1. Although approximately two-thirds of the GUS-positive plants were fertile, the remaining one-third were sterile as a result of the lack of mature pollen grains. All of the GUS-negative plants were normal. T3 progeny from the GUS-positive fertile plants had the same 3:1 ratio, again, with one-third of the GUS-positive plants being male-sterile (Figure 1A ). The mutant anthers were small and white because they lacked mature pollen grains (Figure 1B). This sterile phenotype segregated as a single recessive allele (fertile:sterile = 142:49; ² = 0.02 for 3:1) (Figures 1A and 1B). We named this mutant wax-deficient anther1 (wda1).

View larger version (68K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 1. Phenotypes of Wild-Type, wda1 Knockout, and Wda1 Underexpressed Plants. (A) Comparison of wild-type (left) and wda1 knockout (right) plants after bolting. (B) Comparison among wild-type (left), wda1 (middle), and wda1 as-3 (right) anthers at the pollen mitosis stage. Bars = 1 mm. (C) Transcript levels of Wda1 antisense plants. Wda1 transcript levels were examined by semiquantitative RT-PCR using p3 and p4 primers (see Figure 2A). Ubiquitin extension protein1 (Ubq1) was used as a control. as-1, partial fertile plant; as-2 and as-3, sterile plants; as-4 and as-5, fertile plants; WT, wild-type plant.

View larger version (76K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 2. Scheme of the Wda1 Gene, Multiple Alignments, and Phylogenic Analysis of WDA1-Related Proteins. (A) Wda1 gene and relative insertion position of the T-DNA. Closed boxes represent exons, and connecting lines represent introns. The ATG start codon and TGA stop codon are indicated. The insertion position of the T-DNA in line 0-208-29 (gray box) is indicated in the first intron of Wda1. The number 1 indicates the starting nucleotide of translation; –88 indicates the starting point of the 5' untranslated region located 88 bp upstream from ATG; 5426 indicates the genomic nucleotide length from ATG to TGA; and 5605 indicates the end point of the 3' untranslated region located downstream at 5605 bp from ATG. Numbers represent start and end points of each exon. p1 and p2 are primers for the construction of Wda1 underexpressed plants, and p3 and p4 are primers for semiquantitative RT-PCR analysis of the Wda1 transcript. GUS, ß-glucuronidase; hph, hygromycin resistance marker for the selection of T-DNA insertion; nt, nucleotides. Bar = 0.5 kb. (B) Multiple alignments of plant proteins showing high similarity with the WDA1 protein. WDA1 (top line) was aligned with three rice proteins (LOC_Os02g56920, LOC_Os02g40780, and Os Wax2-like [LOC_Os09g25850]) and four Arabidopsis proteins (CER1, CER1-like, At2g37700, and WAX2). Black background indicates amino acid residues that are >70% conserved, and gray background indicates amino acids that are >30% conserved. Asterisks indicate perfectly matched amino acids, colons indicate highly conserved amino acids, and dots indicate low-matched amino acids among these eight proteins. The identity and similarity of the aligned proteins for WDA1 are shown at the end of the alignments. The dotted box indicates transient signal peptides for membrane targeting; the three black lines indicate His-rich motifs (HX3HH) for putative zinc binding; and the gray line indicates a desaturase domain. (C) Phylogenic analysis of proteins similar to WDA1. Phylogenic analysis by MEGA 2.0 was performed using the neighbor-joining tree with 1000 replicates; the handling gap option was pair-wise deletion. Multiple alignments used for this analysis are shown in Supplemental Table 5 online.

The predicted protein encoded by the tagged gene contains 621 amino acid residues with a molecular mass of 71.2 kD and a pI of 8.24. The protein has 60% identity with rice LOC_Os02g40780, which is the most similar ortholog to Arabidopsis CER1, 56% with the Arabidopsis CER1 protein (accession number BAA11024), 51% with the Arabidopsis CER1-like protein (accession number CAA65199), 51% with rice LOC_Os02g56920, and 34% with maize GL1 (accession number AY505017).

Multiple alignments were performed for the seven plant proteins with high similarity to the WDA1 protein (Figure 2B). A functional-domain analysis with Pfam 7.0 (http://www.sanger.ac.uk/Software/Pfam) indicated that the region between the 40th and 258th amino acid residues forms the sterol desaturase domain (gray lines in Figure 2B), which is present in C-5 sterol desaturase and C-4 sterol methyl oxidase. Members containing this conserved region are involved in the biosynthesis of plant cuticular waxes (Aarts et al., 1995; Chen et al., 2003). The prediction of a signal peptide with SignalP 3.0 (http://www.cbs.dtu.dk/services/SignalP/) and PSORT (http://psort.nibb.ac.jp/form.html) suggests that WDA1 comprises a membrane-targeting signal at the N terminus and a putative cleavage site between the 31st and 32nd amino acids. Amino acid sequence analysis using the DAS Transmembrane Prediction server (http://www.sbc.su.se/miklos/DAS/) classified WDA1 as an integral membrane protein containing five transmembrane domains in the N-terminal region. The WDA1 protein has one His-rich motif, HX3HH, starting at the 146th amino acid plus two HX2HH motifs beginning at positions 157 and 252 in the deduced amino acid sequence (black lines in Figure 2B). Taton et al. (2000) have demonstrated by site-directed mutagenesis that the His residues in His-rich motifs are essential for the catalysis of sterol desaturases, and they proposed that the ligands are divalent metal ions.

Phylogenic analysis of the 14 proteins with high similarities to WDA1 showed that they can be grouped into two clades (Figure 2C). WDA1 groups with CER1 and CER1-like proteins, whereas the WAX2-like proteins form the second clade (Chen et al., 2003). Three rice proteins (LOC_Os02g08230, LOC_Os06g44300, and LOC_Os09g25850) belong to the second clade. Arabidopsis CER1 functions in the biosynthesis of stem wax and pollen coat tryphine alkanes (Aarts et al., 1995). The Arabidopsis CER1-like protein (accession number CAA65199) is a flower-preferential isozyme of CER1, but its functional defects have not been identified (Aarts et al., 1995). Our phylogenic analysis suggests that WDA1 is likely involved in VLCFA metabolism.

Downregulation of Wda1 Causes Male Sterility
To confirm whether the male-sterile phenotype was attributable to mutation in the Wda1 gene, we generated transgenic plants that underexpressed Wda1. The 653-bp N-terminal region of the Wda1 cDNA was amplified by the p1 and p2 primers (Figure 2A), and the PCR product was placed under the control of the maize ubiquitin promoter in the antisense direction. The construct was introduced into rice calli by the Agrobacterium tumefaciens–mediated transformation method (Lee et al., 1999). Among the 74 transgenic plants that grew to maturity, 17 were completely sterile and 32 showed partial fertility. The remaining plants were of normal fertility. RT-PCR analysis of Wda1 transcript levels in the transgenic plants demonstrated that the fertile lines (as-4 and as-5 in Figure 1C) contained significant amounts of transcript, whereas the sterile lines (as-2 and as-3 in Figures 1B and 1C) contained low levels. The partial sterile line (as-1 in Figure 1C) contained an intermediate amount. Close examination of the Wda1 as-3 antisense plants revealed small anthers and defective pollen grains, as had been observed with the wda1 knockout anthers (Figure 1B). These experiments confirmed that mutations in Wda1 cause anther defects and male-sterile phenotypes.

Wda1 Transcript Is Abundant during Flowering Stages
Because the GUS gene located in T-DNA was fused to Wda1, the level and pattern of expression of the fusion gene could be assayed for GUS activity. High levels of GUS activity were observed in spikelets from the premeiosis stage to the heading stage (Figures 3A and 3K ). GUS activity was also prominent in the outer and inner epidermis (Figure 3H) and trichomes (Figure 3G) of the palea/lemma, lodicules (Figures 3C, 3E, 3F, and 3I), stigmas (Figure 3C), and anthers (Figures 3C, 3D, 3I, and 3J). Cross sections of heterozygotic anthers at the vacuolated pollen stage revealed that GUS was expressed strongly in the epidermis but only weakly in the tapetum, endothecium, and connective tissue (Figure 3J). In vegetative tissue, GUS activity was mostly absent except at collar regions (between the leaf sheath and the leaf blade) and the base of shoots (see Supplemental Figure 1 online).

View larger version (92K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 3. Expression Profiles of Wda1 by GUS Assay and RT-PCR. (A) Temporal and spatial expression patterns of the Wda1-GUS fusion product in heterozygous spikelets at various flowering stages. Sample 1, premeiosis; sample 2, meiosis; sample 3, tetrads; sample 4, young microspore; sample 5, vacuolated pollen; sample 6, early pollen mitosis; sample 7, late pollen mitosis. (B) Photograph taken after removal of the palea and half of the lemma from the spikelet at the tetrad stage. (C) Photograph taken after removal of the palea and half of the lemma from the spikelet at the vacuolated pollen stage. (D) Anther at early pollen mitosis. (E) Palea and lodicule at early pollen mitosis. (F) Other side of the palea and lodicule in (E). (G) Magnified image of sample 7 in (A). (H) Cross section of a palea at the vacuolated pollen stage (sample 5 in [A]). (I) Cross section of the lower portion of a spikelet at the vacuolated pollen stage (sample 5 in [A]). (J) Cross section of an anther at the vacuolated pollen stage (sample 5 in [A]). (K) Temporal expression patterns of Wda1 in shoots and roots of 7-d-old seedlings, various stages of panicles, developing seeds at 6 d after pollination, and flag leaves. (L) Spatial and temporal expression patterns of Wda1 in developing wild-type anthers at five different stages (meiosis, young microspore, vacuolated pollen, pollen mitosis, and heading), palea/lemma, and lodicules/ovaries. Ubq1 was used as a control. Data are representative of three independent experiments. Photographs in (H) to (J) were taken under dark-field microscopy. A, anther; AF, anther filament; C, carpel; Co, cortex; CT, cortical tissue; En, endothecial cell layer; Ep, epidermal cell layer; FL, flag leaf; G, glumes; H, heading; IE, inner epidermis; Le, lemma; Lo, lodicule; M, meiosis; O, ovary; OE, outer epidermis; P, pollen; Pa, palea; PM, pollen mitosis; S, shoot of seedling at 7 d after germination; Se, developing seeds at 6 d after pollination; St, stigma; T, tapetal cell layer; Tr, trichome; V, vacuolated pollen; Va, vascular bundle; wda1, wda1 homozyous anther; Y, young microspore. Bars = 1 mm in (A) to (F) and 20 µm in (G) to (J).

Characterization of wda1 Mutants
To further characterize the phenotypic alterations of mutant anthers, we performed thin-section analyses for wild-type and wda1 anthers and delineated anther development into six stages based on stages of pollen formation (Figure 4 ). During the premeiosis stage (Figure 4A), the archespores divided into secondary parietal cells and sporogenous cells. Then, those secondary cells divided into inner (tapetal) layer cells and middle layer cells. In the wda1 anthers, their sporogenous cells, endothecium, middle layer, and inner layer appeared normal at premeiosis (Figure 4G).

View larger version (62K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 4. Light Microscopic Analyses of Developing Anthers. Cross sections of wild-type ([A] to [F]) and wda1 ([G] to [K]) anthers at premeiotic stage ([A] and [G]), meiosis stage ([B] and [H]), young microspore stage ([C] and [I]), vacuolated pollen stage ([D] and [J]), early pollen mitosis stage ([E] and [K]), and late pollen mitosis stage (F). (L) shows a cross section of a wda1 as-2 anther at the early pollen mitosis stage. BP, binucleated pollen; DMC, degenerated meiocyte; DMY, degenerated young microspore; En, endothecium cell layer; Ep, epidermis cell layer; M, middle layer; MC, meiocyte; SC, sporogenous cell; T, tapetal cell layer; TP, trinucleated pollen; UP, uninucleated pollen; YM, young microspore. Bars = 10 µm.

View larger version (146K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 5. Transmission Electron Microscopy Analyses in Developing Anther Walls. (A) to (D) Cross sections of wild-type anther walls at meiosis (A), tetrad (B), young microspore (C), and early pollen mitosis (D) stages. (E), (F), and (H) Cross sections of wda1 anther walls at meiosis (E), young microspore (F), and early pollen mitosis (H) stages. (G) Cross section of a wda1 as-2 anther wall at the early pollen mitosis stage. (I) to (L) Magnified images of the peritapetal region in (A) to (D). (Q) to (T) Magnified images of the outer epidermal region in (A) to (D). (M) to (P) Magnified images of the peritapetal region in (E) to (H). (U) to (X) Magnified images of the outer epidermal region in (E) to (H). The white box in (A) indicates the peritapetal region, and the dotted black box indicates the outer epidermal region. Ag, aggregates; C, cuticle; CLB, cytoplasmic lipid body; CW, cell wall; DP, degenerated pollen; En, endothecium cell layer; Ep, epidermis cell layer; L, lipid body; M, middle layer; Mit, mitochondrion; Nu, nucleus; Or, orbicule; P, pollen; Pl, plastid; PM, plasma membrane; PT, peritapetal region; T, tapetal cell layer; Ve, vesicle; wda1, wda1 homozygous anther; wda1 as-2, wda1 antisense-2 anther; WT, wild-type segregant. Bars = 5 µm in (A) to (P) and 0.2 µm (Q) to (X).

At the young microspore stage (Figure 4C), the microspores were developed and exine deposition continued in wild-type anthers. Sporopollenins were deposited to microspores to form the bacular and nexine of exine (see Supplemental Figure 2 online). The peritapetal region contained thick orbicules for the transfer of lipids and proteins from the tapetum. In addition, cytoplasmic lipid bodies became visible (Figures 5C and 5K). However, the outer region of the anther epidermis was not much different from that in previous stages, except that a thin and darkly stained region developed in the outermost layer of the anther at the young microspore stage (Figures 5Q to 5S). By contrast, the wda1 tapetum did not produce orbicules at the peritapetal region but showed some aggregated lipidic molecules (Figures 5F and 5N). Another characteristic of the mutant anthers was that their middle layers did not degenerate (Figures 5F and 5H). At the vacuolated pollen stage (Figure 4D), the microspores finished their exine deposition, started to deposit pollen coats into the tectum of exines, and began to enrich the pollen cytoplasm (see Supplemental Figure 2 online).

At the pollen mitosis stage in wild-type anthers (Figures 4E and 4F), the uninucleate pollen developed into trinucleate pollen through two mitotic divisions. Simultaneously, starches, proteins, lipids, and other nutrients enriched the pollen cytoplasm. Tapetal cells were completely degenerated by this stage, and maturation of the pollen wall finished. The endothecial cell layers finally disrupted to release mature pollen grains. Transmission electron microscopy analysis at the early pollen mitosis stage showed that the peritapetal region contained orbicules loaded with thick lipidic molecules (Figure 5L). The outermost epidermal layer had hair-like extensions from the cuticle layer, and numerous lipid bodies appeared in the epidermis (Figures 5C and 5D). However, we did not observe further development in the cuticle layer of the wda1 epidermis (Figure 5X) because of a lack of lipid bodies with dark-stained lipidic molecules (Figure 5H). The wda1 anthers were also severely shrunken at this stage, and most pollen grains degenerated in the anther locules (Figures 4J and 4K). In particular, the wda1 tapetum accumulated irregularly shaped vesicles surrounded by dark-stained lipidic molecules; many degenerating mitochondria also aggregated around these vesicles (Figures 5H and 5P).

At the young microspore stage in wild-type tapeta, we observed regular patterns of lipidic molecules embedded in small vesicles, which were expected to be fused to the orbicules (Figure 5K). Moreover, we found no components of the degenerated tapeta at the pollen mitosis stage (Figures 5D and 5L). The wda1 anthers had a similar phenotype after the vacuolated pollen stage. Mutant anthers did not dehisce because their tapeta were not degenerated (Figure 5H).

Close observation of anthers of the wda1 antisense plants (as-2) at the early pollen mitosis stage showed an abnormally thick aggregation of sporopollenin in the pollen walls as well as undeveloped pollen cytoplasm (Figure 4L; see Supplemental Figure 2 online). Orbicules were generated similar to those in the wild type, although most were partially destroyed and, likely, incomplete (Figure 5O). In the epidermal layer of as-2, lipid bodies were abnormally incorporated in the cell wall and an abnormal cuticle layer was formed (Figures 5G and 5W). These antisense plants confirmed that the reduction in Wda1 expression causes the abnormal development of anthers and microspores.

These observations suggest that phenotypic alterations in wda1 probably result from blocked biosynthesis of lipidic molecules in the anther walls and, consequently, their hindered transference from the anther walls to the developing microspores and the cuticular layer of the anther.

wda1 Lacks Wax Crystals and Is Affected in Expansion and the Morphology of Anther Epidermal Cells
Our wda1 knockout and antisense anthers were much shorter than those of the wild-type segregants (Figure 1B). However, this reduction in length is not a common phenomenon in male-sterile plants (Aarts et al., 1997; Ariizumi et al., 2004; Jung et al., 2005). Because Wda1 is expressed strongly in the anther epidermis, we examined the surface structure of wda1 anthers by scanning electron microscopy. Mutant anther walls appeared smooth compared with wild-type walls (Figures 6B, 6C, 6G, and 6H ). Close examination of the wild-type anthers showed surfaces covered by epicuticular wax crystals, primarily linear and plate-shaped (Figure 6F). By contrast, the wda1 knockout anthers contained no wax crystals (Figure 6G). The as-2 anthers had an intermediate level of plate-shaped epicuticular wax crystals (Figure 6H). In addition, the epidermal cells in wda1 knockout and as-2 anthers were significantly shorter than those in the wild type (Figure 6F to 6H). However, the cell width of wda1 was not much different from that of the wild type. These observations are also consistent with recent reports that mutations in the Arabidopsis Long-Chain Acyl-CoA Synthetase gene (At1g49430) and the Arabidopsis ECR gene (At3g55360), both involved in wax biosynthesis, affect the expansion of leaf epidermal cells (Schnurr et al., 2004; Zheng et al., 2005).

View larger version (138K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 6. Scanning Electron Microscopy Analysis of Wild-Type and wda1 Mutant Plants. (A) Wild-type anther at the early pollen mitosis stage. (B) wda1 anther at the early pollen mitosis stage. (C) wda1 as-2 anther at the early pollen mitosis stage. (D) Wild-type pollen at the early pollen mitosis age. (E) wda1 pollen at the early pollen mitosis stage. (F) Outer epidermal region of a wild-type anther at the early pollen mitosis stage. (G) Outer epidermal region of a wda1 anther at the early pollen mitosis stage. (H) Outer epidermal region of a wda1 as-2 anther at the early pollen mitosis stage. GP, germination pore. Bars = 100 µm in (A) to (C), 5 µm in (D) and (E), and 10 µm in (F) to (H).

Major Components of Waxes Are Reduced in wda1 Anthers
From the previous scanning electron microscopy analysis, we observed defective epicuticular waxes in wda1 anthers. Using gas–liquid chromatography, we analyzed the composition and quantity of chloroform-extractable cuticular waxes in wild-type anthers (Table 1 , Figure 7A ). In contrast with leaf waxes, which were composed predominantly of long-chain alcohols and aldehydes (see Supplemental Figure 3 online), the major wax monomers in wild-type anthers were alkenes and alkanes (Table 1), with longer chains dominating in the former (C29, C31, and C33) compared with the latter (C25, C27, and C29) (Figure 7A; see Supplemental Table 1 online). These products of the decarbonylation pathway represented 87.29% of the total wax load, compared with 11.06% fatty acids and 1.65% alcohols as products of acyl reduction (Table 1).

View larger version (14K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 7. Wax and Cutin Constituents of Wild-Type and wda1 Mutant Anthers. (A) Wax analysis of wild-type and wda1 whole anthers. (B) Cutin analysis of wild-type and wda1 whole anthers. White bars represent levels of each compound in the wda1 mutant, and black bars represent them in the wild type. Error bars indicate SD (n = 4). Compound names are abbreviated as follows: C18 acid, octadecanoic acid; C20 acid, eicosanoic acid; C16 -OH acid, 16-hydroxy-hexadecanoic acid; C18 (1) -OH acid, 18-hydroxy-octadecanoic acid; C16 diacid, hexadecane-1,16-dioic acid; C18 (1) diacid, octadecene-1,18-dioic acid; C16 9/10,16-diOH acid, 9(10),16-dihydroxy-hexadecanoic acid; C18 9/10,18-diOH acid, 9(10),18-dihydroxy-octadecanoic acid; C18 epoxy -OH acid, 9-epoxy-18-hydroxy-octadecanoic acid; URCM, unknown rice cutin monomer. Acids were analyzed as methyl esters, and hydroxyl groups were analyzed as trimethylsilyl esters.

Consistent with the lack of wax crystals observed by scanning electron microscopy (Figure 6G), wax loads on wda1 anthers were reduced to 23.34% of the wild-type level (Table 1). Similar levels of reduction in wax amounts have been observed in other mutants with waxless phenotypes (Jenks et al., 1995). Even so, the strongest effects of the wda1 mutation on wax constituents were apparent in the unsaturated alkenes; the saturated alkanes, acids, and alcohols were also reduced (Table 1, Figure 7A). However, we were unable to observe any specific reduction in certain compound classes or for compounds of a certain chain length, indicating that the Wda1 gene product is involved in the biosynthesis of all anther waxes. Consistent with the weak expression of Wda1 in the vegetative organs, no significant qualitative or quantitative differences were seen in the leaf wax monomer composition of the wild-type and wda1 seedlings (see Supplemental Figure 3 online).

Cutin Characteristic Monomers Are Reduced in wda1 Anther Polyester
Consistent with Wda1 expression in anther epidermal cells (Figure 3J), wda1 anthers were defective in their epidermal cell wall formation, especially in the outermost portion of the cell wall (Figure 6G), which is typically characterized by the presence of a cuticle (i.e., the continuous layer of cutin secreted by the epidermis). Therefore, we wanted to investigate the role of Wda1 in cuticle formation by analyzing the aliphatic monomer composition of wda1 cutin compared with the wild type. However, traditional methods using polysaccharide hydrolases to isolate cuticles for subsequent depolymerization and analysis cannot be applied on the microscopic scale of anthers and pollen. Alternative approaches have now been demonstrated with the transesterification of polyester from delipidated tissue, resulting in a monomer composition representative of cutin (Bonaventure et al., 2004; Franke et al., 2005).

Analysis of the depolymerization products after transesterification revealed that rice anther polyester was composed predominantly of C16 and C18 oxygenated fatty acid derivatives. Of these, the most abundant compounds identified in the methanolysate were the cutin characteristic monomers 9(10),16-dihydroxy-hexadecanoic acid, 18-hydroxy-octadecanoic acid, 9-epoxy-18-hydroxy-octadecanoic acid, and 16-hydroxy-hexadecanoic acid (Figure 7B; see Supplemental Figure 5 online). Minor compounds included 9(10),18-dihydroxy-octadecanoic acid, octadecene-1,18-dioic acid, hexadecane-1,16-dioic acid, octadecanoic acid, and eicosanoic acid. Another three compounds with gas chromatography (GC) retention times different from those of the identified monomers were released from the anthers by transesterification. These compounds had mass spectra compatible with hydroxylated fatty acids (see Supplemental Figure 5 online), indicating that they represent unknown rice cutin monomers (URCM1, URCM2, and URCM3). Because of the lack of available standards and very little understanding of the cutin monomer composition in grasses, the exact structures of these unknown rice cutin monomers remain to be determined.

The analysis of the aliphatic monomer composition of wda1 anther methanolysates revealed that the wda1 anther polyester is strongly reduced in cutin characteristic monomers. Compared with the wild type, all detected cutin monomers are decreased by at least 50%, indicating that Wda1 is also involved in the pathways providing aliphatic precursors for proper cuticle formation.

In contrast with our analysis of whole anthers (Figure 7), comparison of the transesterification products of pollenless anther walls and isolated pollen (see Supplemental Figure 4 online) showed that almost no cutin monomers can be released from pollen by transesterification. This finding suggests that these functionalized aliphatics are incorporated into the highly resistant sporopollenin, which is required for proper exine formation.

Mutation in the Wda1 Gene Affects the Expression of Rice Raftin Genes
To compare the functional roles of Wda1 with those of other genes involved in pollen wall formation, we examined their expression levels in mutant anthers. Os Raftin1 and Os Raftin2 are highly similar to wheat RAFTIN1, a gene that plays important roles in exine maturation. RNA-induced gene-silenced lines of wheat RAFTIN1 or Os Raftin1 in rice showed male-sterile phenotypes attributable to arrested tapetal degeneration at the vacuolated pollen stage. Mutant anthers are nondehiscent and 10 to 15% shorter (Wang et al., 2003). These phenotypes are similar to those of Wda1. The gene product appears to function in transporting various molecules from the tapetum to pollen walls via orbicules (Wang et al., 2003). Our analyses showed that both rice Raftin genes were strongly expressed in the anthers (Figure 8 ). Whereas Os Raftin1 was expressed throughout anther development, Os Raftin2 was preferentially expressed at the early stages. In wda1 anthers, Os Raftin2 transcript was nearly undetectable, whereas that of Os Raftin1 was reduced significantly. Therefore, the defect in wax production affected the expression of both rice Raftin genes and pollen fertility.

View larger version (40K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 8. Effect of the wda1 Mutation on Exine-Related Genes. Semiquantitative RT-PCR analyses of exine-related genes (Os Raftin1, Os Raftin2, and Os Dex1). RNA samples were prepared from wda1 anthers and from various stage anthers, palea/lemma, and lodicules/ovaries from wild-type spikelets. Ubq1 served as a control. Bands shown are RT-PCR products. The experiment was repeated at least three times, and data from one representative experiment are shown. H, heading; Le, lemma; Lo, lodicule; M, meiosis; O, ovary; Pa, palea; PM, pollen mitosis; Y, young microspore; V, vacuolated pollen.

To find other components of the VLCFA biosynthesis genes in rice anthers, we analyzed lipid biosynthesis–related genes from previously reported microarray data (Kunst and Samuels, 2003; Jung et al., 2005) using the program for gene ontology (GO; http://www.geneontology.org/GO.slims.shtml) and clusters of orthologous groups of proteins (COG; http://www.ncbi.nlm.nih.gov/COG/). These analyses identified several anther-preferential genes that potentially encode acyl-CoA synthases, enoyl-CoA hydratase, ATP binding cassette (ABC) transporters, acyl carrier proteins, fatty acyl-CoA elongase, and acyl-CoA reductases, all of which might be involved in wax biosynthesis (see Supplemental Table 2 online) (Aarts et al., 1997; Hooker et al., 2002; Kunst and Samuels, 2003; Pighin et al., 2004; Zheng et al., 2005). Among these, expression of an acyl-CoA synthase gene and an acyl-CoA reductase gene was downregulated in wda1 anthers (see Supplemental Figure 6 online). Interestingly, the 3-ketoacyl-CoA synthase gene, which is expected to have a major role in the VLCFA biosynthesis pathway, showed strong expression in anthers. By contrast, fatty acyl-CoA reductase genes that are involved in the acyl reduction pathway, a minor portion of VLCFA biosynthesis in anthers, showed weak expression even though the expression level was examined after 35 cycles via semiquantitative RT-PCR (see Supplemental Figure 6 and Supplemental Table 2 online) (Aarts et al., 1997).

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
In this study, we identified a T-DNA insertion mutant in Wda1, a rice homolog of Arabidopsis CER1. CER1 and homologous genes, such as WAX2 and GL1, are involved in cuticle formation or wax deposition (Jenks et al., 1995; Hansen et al., 1997; Chen et al., 2003). Although conditional male sterility was observed in stem wax–deficient Arabidopsis mutants, the functional studies related to CER1 and homologous genes were focused on biochemical or developmental roles in leaves or stems (Aarts et al., 1995; Fiebig et al., 2000). Interestingly, the wda1 mutant exhibits developmental and biochemical phenotypes that reveal the functioning of Wda1 in the epidermal layer and tapetum of the anther wall, the orbicule in the peritapetal region and pollen development, and the biosynthesis of VLCFA derivatives required for cuticle formation and wax deposition.

Wda1 Is Preferentially Expressed in Epidermis of Various Floral Organs, including Anthers
The Wda1 gene was identified as a Wda1-GUS fusion in the line that showed strong GUS expression in various floral organs, especially the anther walls, trichomes, inner and outer epidermis of the palea/lemma, and surfaces of stigmas and lodicules (Figure 3). The GUS expression patterns suggest that Wda1 functions in the outer cell layers of the floral organs (Figures 3H and 3J). However, its reduced expression causes phenotypic alterations only in the anthers and not in other floral organs in which it is expressed. In those organs, such as the palea/lemma, other genes might compensate for mutations in the Wda1 gene. Two genes—LOC_Os02g40780 and LOC_Os02g56920—that are highly similar to Wda1 are present in the rice genome. Compared with Wda1, both are expressed only weakly in flowers but strongly in the vegetative tissues (Figures 3K and 3L; see Supplemental Table 3 online). Their expression may be sufficient to compensate for the wda1 defects in most vegetative and most floral organs. In vegetative organs, Wda1 is only weakly expressed and can be detected only in collar regions and basal parts around the shoot apical meristem (see Supplemental Figure 1 online). Perhaps wda1 defects in these areas may not be fully compensated for by other genes, thereby exhibiting the smaller and less tiller phenotypes (Figure 1A). Even though there were slight defects of growth in wda1 plants, Wda1 is primarily required for the developmental processes involved in anther and pollen maturation. Publicly available expression data (UniGene Cluster: http://www.ncbi.nlm.nih.gov/UniGene/UGOrg.cgi?TAXID=4530) indicate that Wda1 ESTs were predominant in the panicles but that ESTs of LOC_Os02g40780 were present in various organs (e.g., callus, leaf/stem, and flower) (see Supplemental Table 3 online). ESTs of another Wda1-related gene, LOC_Os02g56920, were abundant in both vegetative organs and flowers. This UniGene Cluster information supports the notion that Wda1 expression is panicle-preferential, whereas its related genes are more widely expressed in both vegetative and reproductive stages.

WDA1 Functions in the Biosynthesis of Various VLCFA Precursors for Wax and Cutin Monomers
The importance of VLCFA biosynthesis in plants for fertility and development in reproductive organs, such as anthers and pollen, has been reported for wax-deficient mutants or mutants in lipid biosynthesis in the dicotyledonous Brassicaceae, including Arabidopsis (Aarts et al., 1997; Hernandez-Pinzon et al., 1999; Fiebig et al., 2000; Ariizumi et al., 2003, 2004). However, related studies are rare in monocotyledonous plants (Wang et al., 2002). Based on the phylogenic analysis (Figure 2C), a biochemical function similar to that of Arabidopsis CER1, proposed to be involved in the decarbonylation pathway of stem wax biosynthesis (Aarts et al., 1995), could be expected for WDA1. However, WAX2 and GL1, other proteins similar to WDA1, are suggested to be involved in transporting various lipidic molecules or acting as a membrane-bound receptor, respectively (Jenks et al., 1995; Hansen et al., 1997; Chen et al., 2003; Kunst and Samuels, 2003), leaving the biochemical function of CER1-like proteins still unclear.

Analysis of the chloroform-extractable cuticular waxes of rice wild-type anthers and pollen grains showed that the major wax components in anthers and pollen are very-long-chain alkenes and alkanes (Figure 7A; see Supplemental Figure 4 online), similar to those found for maize pollen (Wilmesmeier and Wiermann, 1995). In wda1 anthers, these lipid levels are reduced to 23.34% (Table 1). These lipid defects and the observation of GUS activity in the anther wall indicate that WDA1 should have a significant role in the generation or secretion of very-long-chain aliphatic molecules, including the precursors for sporopollenin in anther walls. Because of the very pronounced reduction of unsaturated alkenes in wda1, it is tempting to speculate that WDA1 acts as a desaturase in the pathways to very-long-chain aliphatics, consistent with the three N-terminal His-rich motifs conserved in WDA1 and other desaturases (Taton et al., 2000). However, because saturated alkanes are also reduced significantly in wda1, a less specific function of WDA1 is more likely. In addition, a catalytic function of WDA1 specific for intermediates of the decarbonylation pathway, as proposed for CER1 (Aarts et al., 1995), is unlikely, because very-long-chain (C26 and C28) fatty acids and alcohols, products of the acyl reduction pathway, are also reduced significantly in wda1 anthers (Figure 7A). Therefore, WDA1 is potentially involved in the general processes of VLCFA biosynthesis. The latter hypothesis is supported by the fact that the amount of all cutin monomers is reduced severely in the methanolysate of wda1 anther polyester (Figure 7B). The decrease in monomers of various substance classes and monomers of different chain lengths strongly indicates that WDA1 participates in the modification of more than one aliphatic precursor. This is consistent with the broad range of activity observed in lipid metabolic enzymes, such as lipid desaturases, which have been shown to also function as hydroxylases (Broadwater et al., 2002). In addition, subcellular localization analysis using green fluorescent protein showed that WDA1 is targeted to the endoplasmic reticulum (see Supplemental Figure 7 online), similar to ECR, CER6, and GL8, which are part of the VLCFA elongase complex (Xu et al., 2002; Kunst and Samuels, 2003; Zheng et al., 2005). The desaturase motif in WDA1 would be consistent with a function for WDA1 as a dehydratase in an anther-specific fatty acid elongation complex for general lipid metabolism.

WDA1 Is Involved in the Regulation of Genes Related to Lipid Biosynthesis and Pollen Wall Development
Figures 5 and 6 demonstrate that WDA1 is involved in pollen and anther development, possibly via effects on the synthesis or diffusion of signals, such as the lipidic signaling molecules (Aharoni et al., 2004) that direct pollen and anther development. This is indicated by the fact that genes potentially involved in lipid metabolism and transfer are downregulated in wda1 anthers. Of them, we identified LOC_Os03g26600, a putative component of the rice fatty acid elongase complex, encoding KCS, which shows strong expression in rice anthers (Kunst and Samuels, 2003; Jung et al., 2005) (see Supplemental Table 2 online). It is significantly downregulated in the wda1 anther (see Supplemental Figure 6 online). This KCS has 61% similarity to Arabidopsis KCS1, which is involved in wax biosynthesis (Todd et al., 1999). Downregulation of KCS in wda1 anthers may also contribute to the overall differences in wax components between wild-type and mutant anthers. We also found that the Os MS2 gene, with 73% similarity to At MS2, a key component of the acyl reduction pathway in Arabidopsis (Aarts et al., 1997), is preferentially expressed in the anther and is severely reduced in the wda1 anther (see Supplemental Figure 6 online). Also indicative for Wda1 functioning at the molecular level are effects on rice Raftin genes that are homologs of the wheat RAFTIN1 gene located at orbicules and exines, which are supposed to have a guiding role in the proper fixation of sporopollenin polymers in the exine (Wang et al., 2003). Os Raftin2 is preferentially expressed during early pollen development, whereas Os Raftin1 functions at all stages in the wild-type anthers (Figure 8). The severe downregulation of both putative rice orbicule marker genes and the absence of orbicules and cytoplasmic lipid bodies in wda1 anthers imply that the wda1 mutant possibly affects the transfer of sporopollenin from the tapetum to the pollen walls via these organelles. Whether its effect on the expression of other lipid-related genes is direct (e.g., synthesis of signaling molecules) or indirect (e.g., downregulation of lipid-related genes by a feedback mechanism attributable to the accumulation of lipid precursors) can be clarified only by evaluating the biochemical function of WDA1. To date, all of our attempts have failed to heterologously express a functional WDA1 protein and determine catalytic activity.

WDA1 Is Required for Cell Expansion and Early Anther Development
Unlike most cer mutants or other VLCFA metabolism–related mutants with reduced fertility (Aarts et al., 1995; Fiebig et al., 2000; Pighin et al., 2004; Zheng et al., 2005), the major phenotypes for our wda1 mutant were observed in the rice anther wall, epidermis, and tapetum and included severe morphological abnormalities and smaller anthers. As a result, wda1 exhibited complete male sterility (Figure 1). A detailed examination of epidermal cells in the developing anther walls demonstrated that wda1 and wda1 as-2 cells are approximately four and two times smaller than wild-type cells, respectively, which is closely correlated with a similar reduction in anther size (Figure 6). In particular, expansion of the anther epidermal cell after the meiosis of male gametes did not proceed in wda1 anthers. This early developmental defect was the likely cause of reduced organ sizes, similar to what has been reported for leaves from Arabidopsis cer10 mutants (Zheng et al., 2005). It seems that this defect resulted from the blocked transport of VLCFA aliphatic moieties from the epidermal cell to the epicuticular wax layer via the lipid bodies (Figures 5H and 5X). Recently, CER5 was reported to encode an ABC transporter. It is thought to be involved in the transfer of cuticle wax constituents from epidermal cell walls to epicuticular or intracuticular wax layers. This kind of transporter is possibly involved in the formation of an epicuticular wax layer in the anther wall (Pighin et al., 2004), such as a rice putative CER5-like protein (e.g., LOC_Os10g35180, an ABC transporter) that is strongly expressed in anther through previous rice 60k oligomicroarray experiments (http://www.ggbio.com/) (see Supplemental Table 2 online). However, molecular and cellular mechanisms that regulate plant cell expansion during development still remain poorly understood, despite recent results from Zheng et al. (2005).

Developmental Roles for Wda1 in the Rice Anther Wall
Before meiosis, the development of meiocytes and the four layers of the anther walls are nearly normal in wda1 anthers. Plastid accumulation is also normal within the tapetum. During the young microspore stage in wild-type anthers, orbicules surrounded by thick lipidic molecules are easily distinguished at the peritapetal layer, and cytoplasmic lipid bodies occur in the locule. Exine of the pollen walls is formed by the deposition of molecules that are transferred from the developing tapetum via these orbicules and cytoplasmic lipid bodies (Piffanelli et al., 1998; Wang et al., 2003). By contrast, the tapetal layers of wda1 mutants do not contain any orbicules or cytoplasmic lipid bodies, causing exine formation to fail.

The outer epidermal layer begins to accumulate epicuticular waxes after the vacuolated pollen stage. Generally, the cuticle consists of these epicuticular waxes and cutins in the epidermal layer. It is important for overcoming various environmental stresses, especially uncontrolled water loss. The wda1 anthers, with defects in both components, are severely shrunken by dehydration and are easily damaged by physical forces (data not shown). This suggests that Wda1 has a major role in enhancing the physical protection of anthers by generating epicuticular waxes and cutins. In addition, there is a difference between the growth patterns of wild-type and wda1 anthers, and consequently, there is a difference in overall lipid content, resulting in a severe phenotype. To test this hypothesis, we examined the wax contents of another pollen-defect mutant (Os Dex1 knockout) that shows an exine-defect phenotype similar to Arabidopsis dex1. Unlike wda1, Os dex1 anthers did not display significant changes in surface lipid content (data not shown). Also, sequence analysis clearly shows that WDA1 is a lipid metabolic enzyme. However, we cannot exclude other possibilities that lipid content is reduced by the mutation.

This study reports on a mutant defective in the formation of cuticular waxes and cutins of anther walls. Therefore, this material will be valuable for studying the role of waxes in anther development and in response to various types of environmental stress.

	METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Plant Growth
Seeds of wild-type rice (Oryza sativa cv Japonica) and wda1 were germinated on MSO medium containing 0.44% Murashige and Skoog basal salts, 3% sucrose, 0.2% Phytagel, and 0.55 mM myo-inositol (Sigma-Aldrich). The seedlings were grown for 1 week at 28°C under continuous light, then transplanted to soil in the greenhouse and raised to maturity.

Phylogenetic Analysis
Sequences similar to that of WDA1 were searched through BLAST in the GenBank protein database (http://www.ncbi.nlm.nih.gov/BLAST/), and the results were inspected manually. Sequences were aligned with ClustalW version 1.81, followed by manual alignment. Trees were constructed on conserved positions of the alignment by clustered protein sequences from plants with the neighbor-joining algorithm as implemented in MEGA 2.1 (Kumar et al., 2001) with pairwise deletion for gap filling. To test inferred phylogeny, we used bootstraps with 1000 bootstrap replicates. The tree was represented in the traditional rectangular form.

Histochemical GUS Assays and Microscopic Analyses
Histochemical GUS staining was performed as described by Jefferson et al. (1987) and Dai et al. (1996), except for the addition of 20% methanol to the staining solution. For light microscopic analysis, the tissues were fixed in a solution containing 50% ethanol, 5% acetic acid, and 3.7% formaldehyde, then embedded in Paraplast (Sigma-Aldrich) and Technovit 8100 resin (Heraeus). The samples were sectioned to 3- to 16-µm thickness with a microtome (Leica) and observed with a microscope (Nikon) using bright- and dark-field illumination.

Transmission Electron Microscopic Analysis
Spikelets and anthers of wild-type, wda1, and wda1 as-2 mutant plants were sampled at various stages of development and fixed for 4 h in cacodylate buffer, pH 7.2, that contained 2% paraformaldehyde (Sigma-Aldrich) and 2% glutaraldehyde (Sigma-Aldrich). They were then rinsed with the same buffer and postfixed for 1 h in cacodylate buffer containing 1% osmium tetroxide (Pelco International). After dehydration, the specimens were embedded in London Resin White (London Resin). Ultrathin sections (40 to 60 nm thick) collected on uncoated nickel grids (300 mesh) were stained with 4% uranyl acetate and examined at 60 to 80 kV with a JEOL 1200 transmission electron microscope (JEOL).

Scanning Electron Microscopic Analyses
Spikelets of wild-type, wda1, and wda1 as-2 plants were harvested, embedded for 4 h at 4°C in 0.1 M sodium phosphate buffer solution containing 3% glutaraldehyde, pH 6.8, rinsed three times in 0.1 M phosphate buffer, pH 6.8, and further fixed in 0.1 M sodium phosphate buffer solution containing 2% osmium tetroxide at 4°C overnight. They were rinsed again in the same buffer, dehydrated with an acetone series from 30 to 100% and then three times with 100% acetone, exchanged three times with isoamyl acetate (Sigma-Aldrich), processed for critical point drying using liquid CO₂, and gold-coated (10-nm thickness) in a SCD040 sputter coater (Balzers Union). The specimens were then examined in a LEO1450vp (Zeiss) with an accelerating voltage of 15 kV.

Semiquantitative RT-PCR and DNA Gel Blot Analyses
Total RNAs were isolated using Tri reagent (Molecular Research Center). The mRNAs were acquired with a PolyATtract mRNA isolation system (Promega). For first-strand cDNA synthesis, 200 ng of mRNA was reverse-transcribed in a total volume of 100 µL that contained 10 ng of oligo(dT)_12-18 primer, 2.5 mM deoxynucleotide triphosphate, and 200 units of Moloney murine leukemia virus reverse transcriptase (Promega) in a reaction buffer. PCR was performed in a 50-µL solution containing a 1-µL aliquot of the cDNA reaction, 0.2 µM of gene-specific primers, 10 mM deoxynucleotide triphosphates, 1 unit of ExTaq DNA polymerase (Takara), and reaction buffer. The reaction included an initial 5-min denaturation at 94°C, followed by 21 to 35 cycles of PCR (94°C for 45 min, 60°C for 45 min, and 72°C for 1 min), and a final 10 min at 72°C. Afterward, 20 µL of the reaction mixture was separated on a 1.2% agarose gel, blotted onto a nylon membrane, and hybridized with ³²P-labeled gene-specific probes. The specificity of the RT-PCR primers was verified by sequencing the PCR products. The primers used for RT-PCR are described in Supplemental Table 4 online.

Generating Wda1 Antisense Plants
To generate Wda1 underexpressed lines (wda1 as-1 to as-5), the 653-bp N-terminal region of the Wda1 cDNA was amplified by p1 (5'-ACAAACCCAGGACTCTTCACG-3') and p2 (5'-GTAATGTTTGGAGCCTCGAGG-3') primers (Figure 2A), and the PCR product was placed under the maize (Zea mays) ubiquitin promoter in the antisense direction. The construct was then placed in a binary Ti plasmid vector and introduced to rice calli via Agrobacterium tumefaciens–mediated transformation (Hiei et al., 1997; Lee et al., 1999).

Analysis of Anther and Leaf Waxes
Waxes were analyzed basically as described previously (Hauke and Schreiber, 1998). Two to 4 mg of freeze-dried anthers was submersed in 700 µL of chloroform for 1 min. The resulting chloroform extract was spiked with 10 µg of tetracosane (Fluka) as an internal standard and then transferred to a new vial. Any remaining insoluble anther material was used for our polyester analysis (see below). The solvent was evaporated under a nitrogen stream, and compounds containing free hydroxyl and carboxyl groups were converted to their trimethylsilyl ethers and esters with 20 µL of bis-(N,N-trimethylsilyl)-tri-fluoroacetamide (Machery-Nagel) in 20 µL of pyridine for 40 min at 70°C before gas chromatography–mass spectrometry (GC-MS) analysis. Monomers were identified from their electron ionization–mass spectrometry spectra (70 eV, m/z 50 to 700) after capillary GC (30-m x 0.32-mm x 0.1-µm film thickness [DB-1; J&