ABORTED MICROSPORES Acts as a Master Regulator of Pollen Wall Formation in Arabidopsis.

The bHLH protein, AMS, acts as a master transcriptional regulator directly modulating the expression of 23 genes related to pollen wall development and sporopollenin biosynthesis in Arabidopsis. These genes are implicated in the separation of microspore mother cells, dissolution of the callose layer of tetrads, and subsequent sporopollenin biosynthesis and pollen coat formation. Mature pollen is covered by durable cell walls, principally composed of sporopollenin, an evolutionary conserved, highly resilient, but not fully characterized, biopolymer of aliphatic and aromatic components. Here, we report that ABORTED MICROSPORES (AMS) acts as a master regulator coordinating pollen wall development and sporopollenin biosynthesis in Arabidopsis thaliana. Genome-wide coexpression analysis revealed 98 candidate genes with specific expression in the anther and 70 that showed reduced expression in ams. Among these 70 members, we showed that AMS can directly regulate 23 genes implicated in callose dissociation, fatty acids elongation, formation of phenolic compounds, and lipidic transport putatively involved in sporopollenin precursor synthesis. Consistently, ams mutants showed defective microspore release, a lack of sporopollenin deposition, and a dramatic reduction in total phenolic compounds and cutin monomers. The functional importance of the AMS pathway was further demonstrated by the observation of impaired pollen wall architecture in plant lines with reduced expression of several AMS targets: the abundant pollen coat protein extracellular lipases (EXL5 and EXL6), and CYP98A8 and CYP98A9, which are enzymes required for the production of phenolic precursors. These findings demonstrate the central role of AMS in coordinating sporopollenin biosynthesis and the secretion of materials for pollen wall patterning.


INTRODUCTION
The pollen wall, a multilayer specialized cell wall surrounding the pollen grains, not only provides mechanical protection for male gametophytes from desiccation, environmental stresses, and microbial attacks, but is also essential for various aspects of pollination, including pollen adhesion, hydration, and germination (Scott et al., 2004). The outer layer, called exine, is composed principally of sporopollenin, a highly resistant biopolymer derived from fatty acids, phenylpropanoids, phenolics, and traces of carotenoids (Ahlers et al., 1999). Exine morphology varies among species (Blackmore et al., 2007), and this complex, and sometimes species-specific patterning, combined with its durability, enable its use in paleontological and forensic analyses. Sporopollenin is highly conserved across distant phylogenetic groups of embryophytes, fossilized green algae, and higher plants, suggesting that it has remained evolutionary conserved since the initial colonization of land by embryophytes. Despite this, due to its insolubility and chemical resilience, its biochemistry and biosynthesis still remain largely uncharacterized (Blackmore et al., 2007).
The anther, the higher plant male reproductive organ, comprises four somatic cell layers surrounding the sporogenous cells, which will subsequently form the mature pollen (Scott et al., 2004). The tapetum, the innermost layer, serves as a nutritive tissue that provides metabolites, nutrients, and lipidic sporopollenin precursors for pollen development. Tapetal cells are known to transcribe genes involved in pollen wall biosynthesis and patterning, and in the secretion of the pollen coat, or tryphine, which contains flavonoids, phenolamides, carotenoids, lipids, and proteins that cover the surface and cavities of the exine (Jiang et al., 2013).
In higher plants, a number of conserved transcription factors have been shown to be associated with tapetal function and pollen development (Wilson and Zhang, 2009;Zhang et al., 2011); however, the mechanisms that regulate pollen wall biosynthesis and secretion remain elusive. ABORTED MICRO-SPORE (AMS), a basic loop-helix-loop (bHLH) tapetum-specific transcriptional factor, has been shown to affect the expression of genes involved in the transport of lipids, flavonol accumulation, substrate oxidation, methyl modification, and pectin dynamics (Xu et al., 2010). The ams mutant displays abnormally enlarged tapetal cells and aborted microspore development (Sorensen et al., 2003;Xu et al., 2010).
In this work, we uncover the central role that AMS plays in coordinating the biosynthesis and secretion of materials for pollen wall biosynthesis. Using genome-wide transcriptional analysis, combined with biochemical and functional validation, we show that AMS acts as a principal coordinator of pollen wall formation by directly regulating target genes associated with separation of the microspore mother cell, microspore compartmentalization, and callose dissociation within the tetrad, sporopollenin precursor synthesis (fatty acid elongation and hydroxylation and synthesis of phenolic compounds) and transport, as well as pollen coat formation.

Genome-Wide Identification of Candidate Genes Required for Pollen Wall Development
To understand the precise molecular mechanism underlying pollen wall formation in this study, we used coexpression analysis to show that 98 genes are closely associated with pollen wall development. We determined the expression profiles ( Figure 1A) of previously characterized genes involved in callose wall formation, primexine formation, or sporopollenin biosynthesis (Ariizumi and Toriyama, 2011) (Supplemental Figure 1 and Supplemental Table 1) to generate a list of 16 genes that are specifically expressed during various stages of pollen development ( Figure 1A). These 16 genes were then used as the query for a genome-wide coexpression analysis to identify additional genes associated with pollen wall development from public microarray data sets. Using a cutoff threshold of 0.6 for pairwise Pearson correlation coefficients in the public expression database ATTED-II, a total of 251 genes with a frequency of occurrence of 1059 times were identified as coexpressed with the 16 "guide genes," among them, 43.8% (110/251) genes appeared only once with one guide gene, while 98 genes were present more than four times with 16 guide genes ( Figure 1C). Furthermore, their expression pattern was confirmed as consistent with the guide genes using the Arabidopsis eFP browser ( Figure 1B). Additionally, 11 genes with functional relevance, i.e., b-Ketoacyl-CoA synthases (KCSs; KCS7, 15, and 21), Lipid Transfer Protein (LTP), Extra Cellular Lipases (EXLs; EXL4, 5, and 6), Anther-specific protein6 (A6), CYP450 encoding genes (CYP703A2), AMS, and AtbHLH089, were further validated by quantitative RT-PCR (qRT-PCR) as highly expressed during stage 9 of anther development ( Figure 1D). Anther staging in Arabidopsis refers to Sanders et al. (1999).

AMS Acts as a Key Transcriptional Regulator in Pollen Wall Patterning
Among the 98 genes associated with pollen wall development were six transcription factors (bHLH089, ANAC025, WUS/ WUSCHEL, SPL/SPOROCYTELESS, MYB99, and AMS) possibly associated with specifying/determining tapetal fate and development. Zhang et al. (2006) reported that SPL, EMS1, and DYT1 might regulate anther development via the expression of AMS and thereby indirectly affect pollen wall formation (Zhang et al., 2006). We compared the expression of these 98 genes in the microarray data of spl (Zhang et al., 2006), dyt1 (Feng et al., 2012), ams (Xu et al., 2010), myb80 (Phan et al., 2011), ms1 (Yang et al., 2007), and ems1/exs (Canales et al., 2002;Zhao et al., 2002) and showed that the expression of most of these 98 candidate genes had been changed in spl (73 downregulated), ems1 (69 downregulated), ams (70 downregulated), dyt1 (44 downregulated), and ms1 (52 downregulated; 13 upregulated), while the expression of only a few genes was changed in myb80 (nine downregulated; five upregulated) (Figures 2A and 2B). Given the earlier role of SPL, EMS1/EXS, and DYT1 during anther development, our analyses support the key role of AMS in pollen wall development (Figures 2A and 2B). Previously, we reported that AMS affects the expression of 549 genes in ams buds, and 13 directly regulated genes are involved in the transport of lipids, oligopeptides, and ions, fatty acid synthesis and metabolism, flavonol accumulation, substrate oxidation, methyl modification, and pectin dynamics (Xu et al., 2010). Interestingly, seven of the 13 genes were included in these 98 candidate genes (Table 1).

ams Mutants Have Defects in Microspore Compartmentalization and Callose Dissolution
The ams mutant failed to produce functional pollen; after pollen mother cell meiosis, abnormal tetrads were frequently observed ( Figure 2D) and the tapetal cells were highly vacuolated ( Figures  2J to 2N), in contrast to the condensed cytoplasm of wild-type tapetum ( Figures 2I to 2M). Compartmentalization of the microsporocyte during meiosis and initial primexine wall patterning occurs within a specialized callose (b-1,3-glucan) cell wall, which is subsequently degraded by tapetal b-1,3-glucanase(s) (callase) (Stieglitz, 1977). The callose wall was clearly visible in the wild type ( Figure 2C), while only a weakly staining, ambiguous boundary was visible between the ams microspores (Figure 2D). Reduced callose accumulation occurred in the ams anther locule ( Figure 2P), which subsequently showed delayed, abnormal breakdown ( Figures 2R and 2T).
After the formation of free microspores, wild-type tapetal cells actively secrete sporopollenin precursors onto the primexine (a microfibrillar matrix consisting mainly of cellulose) on the outer surface of microspore. Ultimately, the pollen grains form an exine with a distinct tectum, bacula, and nexine (Figures 2E and 2G). However, although ams mutant microspores formed a primexine, no obvious accumulation of lipidic sporopollenin precursors was seen, with eventual collapse of microspores occurring ( Figures 2F and 2H). Active lipid synthesis was evident in the wild-type tapetum, which appeared to be condensed and degenerated, with disintegration of the cell membrane and accumulation of lipidic tapetosomes and elaioplasts ( Figure 2M). In contrast, ams tapetal cells were severely swollen, with large vacuoles and few lipidic tapetosomes and elaioplasts ( Figure 2N).

AMS Controls the Synthesis of Lipidic and Phenolic Components and Flavonoids for Pollen Wall Formation
The defective pollen wall and absence of obvious tapetosomes and elaioplasts in the ams tapetal cells suggest the ams mutant may have defects in synthesizing chemical precursors of pollen wall development. Gas chromatography-mass spectrometry and gas chromatography-flame ionization detection (GC-FID) were used to measure the lipidic compositions of cuticular waxes, cutin monomers, and total soluble lipids in buds at the tetrad and early microspore stages. GC-FID analysis indicated (A) Sixteen guide genes were selected from the 26 genes that have been demonstrated to be involved in pollen wall formation by forward genetics analysis. (B) Using coexpression analysis, 98 candidate genes were shown to be specifically expressed during pollen wall development. (C) About 43.8% of the genes (110/251) only matched one guide gene during the coexpression analysis, and the average frequency of each gene among 98 genes matching guide genes was more than 4. (D) Expression analysis of 11 selected genes by qRT-PCR analysis. Expression profiles of At1g71160 (KCS7), At3g52160 (KCS15), At5g49070 (KCS21), At1g75910 (EXL4), At1g75920 (EXL5), At1g75930 (EXL6), At1g06170 (bHLH089), At1g01280 (CYP703A2), At5g62080 (LTP), At2g16910 (AMS), and At4g14080 (MEE48/A6) as determined by qRT-PCR using total RNA from shoot (15 d), root (15 d), stem (15 d), leaf (15 d), seed, and anther samples. The relative amount of amplified product was normalized against At-ACT2 transcripts. Each value represents the average of two or more biologic replicate experiments in which each value is the mean of three separate qRT-PCR reactions (6SE). that total wax in the ams buds increased by 12.67% (Figure 3; Supplemental Figure 2). In particular, C29 alkane was significantly increased (P < 0.05) (Supplemental Figure 2A). Whereas total cutin decreased by 15.8% (Figure 3; Supplemental Table  2), mainly due to the significant reduction in the levels of acids (C18, C18:1, C20, and C24), 2-hydroxy-fatty acids (C22, C24, C24:1, C25:1, C26, and C26:1) and fatty di-acids (C18 and C18: 2) (Supplemental Figure 2B).
Phenol amides and flavonoids are major components of the pollen coat and possibly also the pollen wall . Total flavonoid content (TFC) and total phenol content (TPC), determined using NaNO 2 -Al (NO 3 ) 3 and Folin-Ciocalteau reagent analysis, respectively, were reduced in ams buds (30.6 and 34.4%) (Figure 3), indicating that AMS also regulates the biosynthetic pathways for lipidic, phenolic, and flavonoid compounds in the pollen wall and coat. This is supported by transcriptomic analysis that suggested that 21 genes implicated in lipid acyl metabolism, six in secondary metabolism and 21 miscellaneous enzyme encoding (MISC) genes are downregulated (Supplemental Figures 3C and 3D) in ams buds. Thirteen of these 21 genes were previously shown to be involved in pollen wall development ( Figure 3).

AMS Is Directly Associated with the Promoters of Genes Related to Pollen Wall Development
Analysis of the 1-kb upstream promoter region of the putative AMS target genes showed significant enrichment for E-box binding motifs (CANNTG) (Xu et al., 2010), with 17 genes containing more than five E-box motifs in this region and 25 genes containing $3 motifs, suggesting direct regulation by AMS. To confirm this, we employed quantitative chromatin immunoprecipitation PCR (qChIP-PCR) analysis using an AMS-specific antibody (Xu et al., 2010) and observed that the promoter regions of 17 genes, including A6, QUARTET3 (QRT3), EXL4, EXL5, EXL6, KCS7, and KCS15, two GLYCINE-RICH PROTEIN encoding genes (GRP18 and GRP19), four CYP 450 encoding genes (CYP703A2, CYP704B1, CYP98A8, and CYP98A9), At5g62080 (LIPID TRANSFER PROTEIN [LTP]), At1g06990 (GDSL-like Lipase), PKSB/LAP5, and TKPR1/DRL1, from the 55 well-annotated genes from the set of 70 genes showing reduced expression in ams, were enriched by AMS ( Figure 4A). Furthermore, electrophoretic mobility shift assay (EMSA) confirmed the binding of AMS to the promoter of LTP (At5g62080), EXL5, KCS15, and CYP703A2 ( Figure 4B). Moreover, analysis using the GeneCAT tool suggested that AMS and the coexpressed genes form a connective regulatory network ( Figure 4C). Among the genes showing coexpression with AMS were seven genes, including KCS21, WBC27, LTP12, and CYP86C3, that are putatively associated with lipid transport, fatty acid synthesis and metabolism, substrate oxidation, methyl modification, and pectin dynamics, and were previously identified as direct targets of AMS by qChIP-PCR (Xu et al., 2010). Six of the 13 previously identified direct targets of AMS reported by Xu et al. (2010) do not show a significant correlation of expression with AMS, as revealed by our coexpression analysis. Furthermore, available expression data from the websites indicate that these six genes (Supplemental Figure 4C) are expressed in other tissues as well as anthers, suggesting that they may be regulated by additional transcription factors besides AMS. Thus, we did not include these six genes in this study, even though we cannot exclude the possibility that they may directly regulate the formation of the pollen wall. Therefore, together with the previously identified seven targets associated with pollen wall development, AMS may directly regulate the expression of 23 of the 98 genes identified as linked to pollen wall development (Table 1), suggesting that AMS forms a key regulatory hub in pollen wall biosynthesis ( Figure 4C, Table 1).

AMS Regulates Dissociation of the Pollen Mother Cell Wall
The callose wall surrounding the microspore is subjected to enzymatic digestion by b-1,3-glucanase secreted from the tapetum after the tetrad stage. A6 is thought to encode a callase that plays a key role in callose dissolution during microspore release in Arabidopsis (Hird et al., 1993). In this study, we showed that AMS may determine the configuration of microspores in tetrad and callose degeneration by regulating the expression of A6 (Figures 2O to 2T). The callose layer is thought to serve as a temporary wall for microspores that facilitates microspore separation during meiosis and supports primexine formation. Additionally, it may provide both the sugar source for microspore development and a stress factor that acts to compress and flatten the upper ends of the probacula for tectum formation (Heslop-Harrison, 1968;Knox and Heslop-Harrison, 1970).
Consistent with the major role of AMS in regulating callose wall formation and degradation, and subsequent pollen wall formation, no obvious change in the expression of most primexine formation-associated genes was seen in our previous ams microarray analysis (Xu et al., 2010) (Supplemental Table 1).

Synthesis of Phenolic Compounds by CYP98A8 and CYP98A9 Is Essential for Pollen Wall Patterning
Phenol amides are major components of the pollen coat of all higher plants, but their function is still unknown. Phenolic compounds also are reported to be components of the pollen wall (Scott et al., 2004). Arabidopsis CYP98A8 and CYP98A9 were shown to catalyze oxygenation of phenol-amides, generating major pollen phenolic constituents ). qChIP-PCR and expression analysis revealed that CYP98A8 and its paralog CYP98A9 were directly regulated by AMS ( Figure  4A) . To further document the role of AMS in pollen development, null cyp98A8/CYP98A9 RNAi double mutants from the T-DNA insertion line of CYP98A8 and RNA interference (RNAi) line of CYP98A9 were analyzed ) ( Figure 5A). No expression of the CYP98A8 transcript and a very dramatic reduction of the CYP98A9 transcript were observed in the homozygous null cyp98A8/CYP98A9 RNAi line ( Figure 5B). I 2 -KI staining revealed that ;20% of its pollen grains were collapsed with impaired exine architecture. Furthermore, transmission electron microscopy (TEM) analysis indicated an abnormal exine layer with irregular baculum and tectum, and disordered tryphine deposition ( Figures 5D and 5E). AMS thus also impacts pollen wall formation by regulating the biosynthesis of phenolic compounds.

AMS Regulates the Synthesis of Pollen Coat Composition
Hydration of desiccated pollen grains on the stigma is critical for pollination and is closely regulated by proteins and lipids of the pollen coat and stigma cuticle. qChIP-PCR analysis revealed that two pollen oleosin GRP-encoding genes (GRP18 and GRP19) and three EXL family genes (EXL4, EXL5, and EXL6) are directly regulated by AMS (Xu et al., 2010) ( Figure 4A). Arabidopsis includes six clustered EXLs with high similarity at the coding sequence level (Supplemental Figure 5G), which are assumed to be generated from a common ancestor (Mayfield et al., 2001). The eFP browser tool and hierarchical clustering analysis indicate that EXL4, EXL5, and EXL6 are specifically expressed in flowers during pollen wall formation (Supplemental Figure 5F). EXL4 expression was previously shown to be highly specific to anthers and was not detectable in microspores, The wax, cutin, and total soluble lipid amounts expressed as dry weight (mg/mg) in the wild type and ams flower buds. The TPC and TFC amounts relative to standard substance content in each of 100 buds. Error bars indicate SD (n = 5). Some genes (shown in blue on pathway) had been reported as being involved in these metabolic processes; these genes are down in ams, which lead to changes of the corresponding compounds.
[See online article for color version of this figure.] pollen, or other plant tissues (Zimmermann et al., 2004). Promoterb-glucuronidase and RNA in situ hybridization analyses indicate that EXL5 and EXL6 are expressed in the tapetum and that the protein is present in the tapetum and pollen coat (Supplemental Figure 5C). This suggests that AMS is also a regulator of the expression of the pollen coat proteins that mediate the early contact between pollen grains and the stigma that are required for successful adhesion, hydration, and germination. Null exl5 and exl6 mutants ( Figure 5A), as indicated by qRT-PCR from the T-DNA insertion line of EXL5 (+403, second exon) and EXL6 (2121, promoter), displayed a partial lack of exine patterning, resulting in a smooth outer surface ( Figure 5D; Supplemental Figure 5E). Furthermore, TEM analysis showed that exl5 and exl6 pollen exhibited an irregular baculum and tectum, and disordered tryphine deposition ( Figure 5E).
EXLs contain a predicted family II lipase domain and perform acyl transfer reactions in extracellular environments (Upton and Buckley, 1995). EXL4 is enzymatically active during pollen development as the pollen coat is deposited onto pollen grains, creating the complement of lipids required for pollen hydration (Updegraff et al., 2009). Esterase activity, as measured using pollen grains from ams, wild type/col, exl5, and exl6 using p-nitrophenyl butyrate as a substrate (Updegraff et al., 2009) showed that exl5 and exl6 pollen grains had significantly reduced esterase activity (Figures 5F and 5G). As expected, ams had low esterase activity due to the lack of pollen wall ( Figures  5F and 5G). (A) qChIP-PCR analysis revealed the direct association of AMS with the promoters of 17 genes. DNA recovered from immunoprecipitation was used as templates for PCR using primer pairs (the double-headed arrows) spanning the putative AMS binding sequences in the upstream regions of candidate genes. Stars indicate the validation of the binding activity by AMS using EMSA. Data presented here represent the mean of five biological replicates. (B) EMSA of the binding of recombinant AMS onto the promoter fragments. DNA oligomers containing a consensus AMS recognition sequence found in the upstream regions of each of the candidate genes were labeled with digoxin and used as probes.
(C) A network composed of AMS and correlated genes that are putatively involved in microspore compartmentalization in the tetrad, callose degeneration, lipid metabolism, and pollen exine formation and are direct targets of AMS, as determined using the AtGenExpress visualization tool on the GeneCAT website (http://genecat.mpg.de). (B) qRT-PCR of CYP98A8 and CYP98A9 expression in the wild type, ms null mutant, and null cyp98A8/CYP98A9 RNAi buds. Expression was normalized to ACTIN7 and presented relative to wild-type expression levels. Error bars represent SD of three biological replicates. (C) qRT-PCR analysis of the expression of EXL4, EXl5, and EXl6 in the wild-type, ams, exl5, and exl6 buds during anther development at stage 8. Expression was normalized to ACTIN7 and presented relative to wild-type expression levels with three biological replicates; error bars represent SD. (D) Analysis of the outer surface structure of pollen grains from the wild type, the null cyp98A8/CYP98A9 RNAi double mutant, and exl5 and exl6 mutants by scanning electron microscopy. Bars = 10 mm. (E) TEM analysis of the pollen wall from the wild type, the null cyp98A8/CYP98A9 RNAi double mutants, and the exl5 and exl6 mutants. Bars = 500 mm. (F) and (G) The standard curve of lipase activity (F) and lipase activity of pollen grains of ams, exl5, exl6, and the wild type (G). The values were obtained under A 420 wavelength measured, and assays were performed with three biological replicates.

DISCUSSION
Biochemically, sporopollenin is recognized as one of the most resistant biopolymers in nature because it is highly resistant to prolonged desiccation and various stresses as well being insoluble in the strongest acids, bases, and oxidizers. A wealth of biochemical characterization attempts suggest that sporopollenin consists of complex biopolymers derived mainly from longchain fatty acids and phenolic compounds (Scott, 1994;Bubert et al., 2002) and that sporopollenin has been preserved unchanged for >400 million years ago (Wiermann and Gubatz, 1992). An Arabidopsis fatty acid elongase gene, KCS1, shares a high degree of sequence identity to FATTY ACID ELON-GASE1, which encodes a 3-ketoacyl-CoA synthase. KCS1 is essential for wax biosynthesis via catalyzing very-long-chain fatty acid synthesis in vegetative tissues (Todd et al., 1999). We reveal here that AMS is able to regulate the transcription of genes involved in fatty acid elongation such as KCS7, KCS15, and KCS21 ( Figure 4A, Table 1). Furthermore, in situ analysis indicates that KCS15 and At5g62080 (LTP) are expressed in tapetal cells, similar to AMS (Supplemental Figures 4A and 4B).
In higher plants, cytochromes P450 (P450s) play essential crucial roles in both primary metabolism and a wide variety of specialized metabolic processes, including the pathways for production of the precursors of biopolymers, such as oxygenated fatty acid derivatives and lignin monomers (Schuler and Werck-Reichhart, 2003;Pinot and Beisson, 2011). CYP703As and CYP704Bs belong to P450 families specific to land plants under high purifying selection . CYP703s and CYP704Bs from Arabidopsis and rice are specifically expressed in the tapetal cells and microspores (Morant et al., 2007;, and when mutated cause distorted pollen exine and malesterile phenotypes. Recombinant CYP703A2 in yeast cells catalyzes the in-chain hydroxylation of medium-chain saturated fatty (Morant et al., 2007). In contrast, heterologously expressed Arabidopsis CYP704B1 and rice CYP704B2 in yeast catalyze the v-hydroxylation of long-chain fatty acids (Dobritsa et al., 2009;. CYP86C3 belongs to the CYP86C subfamily, which has four members in Arabidopsis; however, the biological functions of this subfamily remain unclear. Recently, recombinant CYP86C3 was shown to act as a short-chain fatty acid hydroxylase with substrates of lauric acid (C12:0), myristic acid (C14:0), palmitic acid (C16:0), and myristoleicacid (C14:1) (Kai et al., 2009). In this study, we found that AMS directly regulates the transcription of CYP703A2, CYP704B1, and CYP86C3 (Table 1, Figure 4A), suggesting that AMS is a critical regulator of the hydroxylation of mid-chain and long-chain fatty acids during pollen exine biosynthesis. Consistent with this, we observed a reduction of hydroxylated lipidic monomers in ams buds (Figure 3).
Similarly, Arabidopsis CYP98A8 and CYP98A9 were shown to be expressed in tapetal cells and to be required for oxygenated phenolamide formation . We provide additional evidence that these genes are regulated by AMS and contribute to pollen wall formation. In Arabidopsis, PKSA and PKSB encode plant-type III polyketide synthases, which catalyze the condensation of malonyl-CoA units with mid-chain and v-hydroxylated fatty acyl-CoAs to generate tri-and tetra-ketide a-pyrone compounds. PKSB/LAP5 and TKPR1/DRL1 encode enzymes that produce hydroxylated a-pyronepolyketide compounds, which are thought to serve as sporopollenin monomers (Grienenberger et al., 2010;Kim et al., 2010). PKSA and PKSB are transiently expressed specifically in tapetal cells, and the pksa pksb mutant lacks an obvious exine, causing complete male sterility. Furthermore, a-pyronepolyketides formed by PKSA and PKSB can be reduced by two tapetally expressed oxido-reductases, TKPR1/2. These form hydroxylated a-pyrone compounds that serve as sporopollenin precursors. Interestingly, in this study, we observed that AMS is capable of triggering the formation of sporopollenin precursors, hydroxylated a-pyrone, and phenolic compounds by directly binding to the promoters of PKSB, TKPR1, CYP98A8, and CYP98A9 ( Figures 4A and 6).
Emerging evidence suggests that pollen coat EXLs in combination with GRP play a role in the initial steps of pollination, namely, hydration on the stigma (Updegraff et al., 2009). Previous reports revealed that six tandem duplicated genes (GRP14, encode the pollen coat GRPs that are mainly expressed in the tapetum and pollen coat (Alves Ferreira et al., 1997;Mayfield et al., 2001;Kim et al., 2003). Furthermore, it has been reported that GRP17 is required for the rapid initiation of hydration on the stigma (Mayfield and Preuss, 2000). Mutations of one GRP result in no obvious phenotype, indicating the possible functional redundancy of these genes in Arabidopsis (Mayfield et al., 2001). In this study, we showed that AMS directly regulates the expression of pollen coat protein, EXLs, and GRPs, which are implicated in the pollen exine and pollen coat formation and subsequent pollination (Figures 4 and 6).
In short, these data support a diverse and critical role for AMS during pollen development, which includes a direct transcriptional regulatory role in the separation of microspore mother cells, dissolution of the callose layer of tetrads, and subsequent sporopollenin biosynthesis and pollen coat formation ( Figure 6). An AMS ortholog (TDR) is functionally conserved in rice (Li et al., 2006), further suggesting that the biochemical regulation of sporopollenin production may be transcriptionally regulated across both monocot and dicot plants via a limited number of key transcriptional regulators. This work provides insight into the transcriptional control of the synthesis of the durable pollen wall in plants.

Plant Material
Seeds of the SALK insertional mutant lines SALK_152147, SALK_057114, and SALK_020241C (SIGnal; Alonso et al., 2003) and Columbia were obtained from the Nottingham Arabidopsis Stock Centre. Phenotypic analysis of the SALK_152147 mutant confirmed that the pollen development defects seen in the ams mutant (Sorensen et al., 2003;Xu et al., 2010) were also observed in the ams SALK knockout. Phenotypic analysis was performed using the null cyp98A8/CYP98A9 RNAi double mutants derived from a T-DNA insertion line of CYP98A8 and an RNAi line. The SALK_057114 and SALK_020241C mutants of EXL5 and EXL6 were confirmed to have pollen development defects. Primers for RT-PCR and genotyping are listed in Supplemental Table 3.

Coexpression Analysis in Arabidopsis thaliana and Cluster Analysis
The Arabidopsis genes in the manually curated abstracts/full text articles that contain experimental evidence for pollen wall biosynthesis genes were used as bait genes to query the ATTED-II database (Obayashi et al., 2009). The expression pattern analysis was performed using the Arabidopsis eFP browser (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi) (Winter et al., 2007). Microarray data analysis employed Genesis (1.7.5) software (Sturn et al., 2002). The Web-based Classification SuperViewer program (http://bar. utoronto.ca/ntools/cgi-bin/ntools_classification_superviewer.cgi) was used to search for differential distributions of Gene Ontology and biological terms within the correlated genes.

Chromatin Immunoprecipitation and qRT-PCR
Experimental plant material and preparation of specific antibodies were as described by Xu et al. (2010). For qChIP-PCR and RT-PCR, oligonucleotides used for the experiments are listed in Supplemental Table 3. Quantification involved normalization of each immune precipitation sample, Ct subtraction of the Ct of the input control in the same immune precipitation to obtain DCt values, and 2 (2DCt) as the percentage of input. All samples were run at least in duplicate.
I 2 -KI Staining, Scanning Electron Microscopy, and TEM Analysis I 2 -KI and callose staining and scanning electron microscopy were performed as described previously (Chen et al., 2011), and the samples treated with adjusted dehydration and fixation as described by Xu et al. (2010).

Analysis of Bud Waxes, Internal Lipids, and Cutin
Wax, internal lipid and cutin extraction, and gas chromatography-mass spectrometry and GC-FID analyses were performed as described previously (Chen et al., 2011;Shi et al., 2011) with slight modifications. Small buds (form the tetrad stage to early microspore stage) were collected; each sample contained ;2 to 3 g dry matter from five individual plants.

Determination of TPC (Folin-Ciocalteau Assay) and TFC of Arabidopsis Buds
Arabidopsis buds (0.5 g) were weighed in an EP tube and extracted with 500 mL of 75% (v/v) ethanol at room temperature for 24 h; then, the extracts were sonicated for 2 h. After extraction, the mixture was centrifuged at 10,000g for 10 min to give the supernatant. The crude extracts were diluted 10 times with 75% (w/w) ethanol before the assay. Total phenolics were determined using Folin-Ciocalteau reagents with some modification (Singleton and Rossi, 1965). Reagent blank using distilled deionized water was also prepared. TPC was quantified using a calibration curve obtained from measuring the absorbance of known concentrations of ferulic acid standard solution. The result was calculated as ferulic acid equivalent per 1 g Arabidopsis buds and reported as mean value 6 SD. All samples were analyzed in triplicate.
The determination of flavonoids was performed according to the colorimetric assay of Kim et al. (2003). Distilled deionized water was used as a reagent blank. TFC was calculated by extrapolating the absorbance of the reaction mixture using a standard curve of rutin. The experiment was repeated thrice and the TFC was expressed as equivalent to rutin (RE) in milligrams/grams of Arabidopsis.

In Situ Hybridization
RNA hybridization and immunological detection of the hybridized probes were performed according to the protocol of Kouchi and Hata (1993). All primers used to amplify probes for in situ hybridization are listed in Supplemental Table 3.

EMSA
The recombinant GST-AMS protein was prepared as previously described by Xu et al. (2010). The DNA fragments containing the E-box of the regulatory region of target genes were generated using PCR amplification with the primers listed in Supplemental Table 3.

Supplemental Data
The following materials are available in the online version of this article.