Cell Wall Reactive Proteins in the Coat and Wall of Maize Pollen

The surface of a pollen grain consists of an outermost coat and an underlying wall. In maize (Zea mays L.), the pollen coat contains two major proteins derived from the adjacent tapetum cells in the anthers. One of the proteins is a 35-kDa endoxylanase (Wu, S. S. H., Suen, D. F., Chang, H. C., and Huang, A. H. C. (2002) J. Biol. Chem. 277, 49055–49064). The other protein of 70 kDa was purified to homogeneity and shown to be a β-glucanase. Its gene sequence and the developmental pattern of its mRNA differ from those of the known β-glucanases that hydrolyze the callose wall of the microspore tetrad. Mature pollen placed in a liquid medium released about nine major proteins. These proteins were partially sequenced and identified via GenBank™ data bases, and some had not been previously reported to be in pollen. They appear to have wall-loosening, structural, and enzymatic functions. A novel pollen wall-bound protein of 17 kDa has a unique pattern of cysteine distribution in its sequence (six tandem repeats of CX3CX10–15) that could chelate cations and form signal-receiving finger motifs. These pollen-released proteins were synthesized in the pollen interior, and their mRNA increased during pollen maturation and germination. They were localized mainly in the pollen tube wall. The pollen shell was isolated and found to contain no detectable proteins. We suggest that the pollen-coat β-glucanase and xylanase hydrolyze the stigma wall for pollen tube entry and that the pollen secrete proteins to loosen or become new wall constituents of the tube and to break the wall of the transmitting track for tube advance.

In plants, sexual reproduction is initiated when the pollen grain (male component) lands on the stigma of the style (female component) in the flowers (for reviews, see Refs. [1][2][3]. The pollen produces a tube that penetrates the stigma and grows through the wall of adjacent cells in the transmitting track in the style to reach the ovary, where the tube delivers the male gametes to fertilize the eggs. Pollen tube penetration of the stigma and advancement in the style are critical steps in sexual reproduction; yet biochemical information on this process is limited. The surface of a pollen grain makes the initial contact with the stigma. It includes an outermost coat and an underlying wall (1)(2)(3). The coat consists of lipids and proteins, which are initially synthesized and accumulated in the tapetum cells that enclose the pollen locule in the anthers. Upon lysis of the tapetum cells, the accumulated lipids and proteins are discharged onto the microspores (maturing pollen), forming the bulk of the pollen coat. In some species, the coat also contains minor proteins involved in self-incompatibility, which are synthesized in the pollen interior (4). The underlying wall consists of sporopollenin and other polymers embedded with proteins; it is derived from the tapetum and the pollen interior (3).
The pollen coat consists of lipids and proteins, and its composition varies, depending on the species. The coat in insect-or self-pollinating species, of which Brassica and Arabidopsis are the best studied (5-7), is thick; steryl esters and very non-polar lipids are the major lipids, and oleosins are the predominant proteins. The lipids are for waterproofing, whereas the amphipathic oleosins may act as a wick for water uptake to initiate germination.
The pollen coat in wind-pollinating species is thin (3,8), and major biochemical studies have been carried out only with maize (8,9). Maize pollen coat contains undefined neutral lipids and a few proteins of 25, 35, and 70 kDa. Only the 35-kDa protein, which is the most abundant, has been characterized and shown to be an active endoxylanase. It is initially synthesized as a large, inactive precursor in the tapetum and then activated after cleavage by a serine protease at both the N and C termini. The precursor or the mature 35-kDa xylanase does not have an apparent signal sequence that could target the protein to specific organelles; the protein is presumed to be present in the cytosol and is released to the anther locule and then the pollen surface after lysis of the tapetum cells. After landing on the stigma, the pollen would release the xylanase to hydrolyze the stigma wall for the entry of the pollen tube into the transmitting track. The characteristics of the other coat proteins of 70 and 25 kDa are presented in the current report.
Imbibed pollen quickly releases proteins that were present in the coat or the wall, or are secreted from the pollen interior. These proteins include expansin, extensin, polygalacturonase, trypsin inhibitors, and a few others (2, 10 -12). Most of them are also categorized as pollen allergens because of their allergenic properties (13). Some of these proteins were shown to have their mRNA present in mature pollen. The relative amounts of these individual proteins in the whole pollen are unknown. It is also unclear whether these proteins are present in the coat or the wall of the mature or germinated pollen and thus whether they will exert their functions on the stigma or inside the style. Another uncertainty is whether these proteins and their mRNA are present continuously in the pollen or pollen tube after germination. Finally, although these proteins have been individually studied in diverse species, their possible collaborative actions in one species has not been examined. All of these uncertainties are addressed in this report.
We have used maize as an example of wind-pollinating species to characterize the major pollen proteins present in the coat and released from the wall or interior of mature and germinated pollen. The coat proteins include a xylanase and a ␤-glucanase derived from the adjacent tapetum cells. The pollen-released proteins represent about 10% of the total pollen proteins. They include many wall-reactive constituents and are synthesized in the pollen interior and secreted to the wall and the exterior. The actions of the coat and pollen-released proteins would allow the advance of the pollen tube on the stigma and through the style, respectively. Several of the proteins in the coat and released from the pollen are novel or have not been previously reported as being present in the pollen.

EXPERIMENTAL PROCEDURES
Plant Materials-Maize (Zea mays L., B73) plants were grown in a greenhouse, and fresh pollen and anthers were collected as described (9). Anthers of five developmental stages were selected on the basis of the following criteria (8): At stage 1, the tassel was still embedded in the shoot apex. The anthers filled up about one-third of each floret. Each microspore mother cell had undergone meiosis to produce a tetrad of microspores, which were still encased within a callose wall. At stage 2, the upper portion of the tassel had protruded from the shoot apex. The anthers filled up about one-half of the floret. Young microspores had been released from the dissolved callose wall, and the outer pollen wall (exine) had been synthesized. At stage 3, the tassel had protruded completely out of the top of the plant. The anthers filled up about two thirds of the floret. The microspores had become larger and contained multiple small vacuoles. The first mitosis had occurred, and the microspores were binucleate. At stage 4, the anthers filled up the floret completely. Second mitosis in the microspores had occurred, and the microspores were trinucleate. At stage 5, the tassel was yellow. Some of the florets on the tassel were open, and the pollen was ready to be released.
Pollen Germination-A sample of 5 mg of freshly collected pollen was placed in 1.5 ml of liquid germination medium containing 100 ppm Ca(NO 3 ) 2, 10 ppm H 3 BO 3 , 37.5 ppm lysine, 5 ppm cystine, 0.05 ppm glutamic acid, and 15% (w/v) sucrose (14). More than 80% of the pollen germinated, and the length of the pollen tube was about 0.5 and 4 times the diameter of the pollen after 10 and 30 min, respectively.
Preparation of an Anther Wall Fraction and a Microspore Fraction-Each anther of developmental stage 3 was placed in a drop of solution containing 0.05 M sodium acetate, pH 5.2, and 0.4 M sucrose on a dish and sliced open longitudinally with a scalpel under a dissecting microscope. The microspores were gently scraped from the anther wall. The microspores and the anther wall, which still retained about 10% of the original microspores, were collected separately. Each of the two fractions from 10 anthers were combined and subjected to RNA extraction.
Preparation of a Total Pollen Fraction, a Pollen-released Protein Fraction, an Interior Fraction, and a Pollen Shell Fraction-Fresh pollen was homogenized rigorously in 0.01 M sodium acetate, pH 5.2, with a mortar and pestle to yield a total protein fraction. A sample of 20 mg of fresh pollen was placed in a 1.5-ml microcentrifuge tube containing 1 ml of 0.01 M sodium acetate, pH 5.2 or the germination medium described in a preceding paragraph. The preparation was shaken gently for 10, 20, or 60 min and centrifuged at 8,000 ϫ g for 3 min. The supernatant was collected as the pollen-released protein fraction. The pellet was homogenized gently in 0.2 ml of 0.01 M sodium acetate, pH 5.2 with a small mortar and pestle, and the homogenate was referred to as the pollen interior fraction. The homogenate was put onto a sucrose gradient containing, from top to bottom, 0.5 ml each of 1, 1.5, and 1.9 M sucrose in 0.05 M HEPE-NaOH, pH 7.5, and centrifuged at 12,000 ϫ g for 30 min. The interface materials between the 1 and 1.5 M sucrose solution were collected. The sample was mixed with three volumes of 0.05 M HEPE-NaOH, pH 7.5, and centrifuged at 8,000 ϫ g for 3 min. The pellet was resuspended with a small volume of 0.05 M HEPE-NaOH, pH 7.5, and was referred to as the pollen shell fraction.
Extraction of a Coat Fraction by Diethyl Ether from Pollen and Partition of Its Proteins into an Aqueous Medium for Enzymatic Studies-Freshly collected pollen was mixed with diethyl ether (1 g of pollen/10 ml of ether) for 1 min in a capped tube by repeated inversions.
The ether layer was collected after centrifugation for 10 min at 8,000 ϫ g. The ether preparation was reduced to 1 ml under a stream of nitrogen and used as the coat fraction for SDS-PAGE. Also, the proteins in this coat fraction were partitioned into an equal volume of 0.05 M sodium acetate, pH 5.2 for fast protein liquid chromatography (FPLC). 1 Cation Exchange FPLC-The procedure was as described earlier (8). All solutions contained 0.05 M sodium acetate, pH 5.2. The aqueous pollen coat sample (ϳ30-g proteins in 3 ml) described in the preceding paragraph was filtered through a 0.2-m syringe filter and applied to a pre-equilibrated Mono S HR 5/5 FPLC column (Amersham Biosciences). Solutions of 5 ml of 0 M NaCl, 6 ml of 0.25 M NaCl, 15 ml of a linear gradient of 0.25-0.75 M NaCl, 5 ml of 1 M NaCl, and 5 ml of 2 M NaCl were applied successively to the column. Chromatographic fractions of 0.5 ml each were collected and analyzed for protein constituents by SDS-PAGE and ␤-glucanase and xylanase activities. The fractions containing the purified 70-and 35-kDa proteins had the peak activities of ␤-glucanase and xylanase, respectively and were retained for additional analyses of enzyme activities.
Enzyme Activity Assay-␤-glucanase activity was measured by monitoring the appearance of reducing ends from substrates at time intervals. Laminarin from Laminarin digitata (Sigma) was used as the substrate. The ␤-glucanase reaction mixture of 0.3 ml contained the purified 70-kDa protein fraction and 0.3 mg laminarin in a 0.05 M buffer, which included sodium acetate, pH 4.0, 5.0, and 5.5; succinate-NaOH, pH 5.5, and 6; phosphate-NaOH, pH 6, 7, 8, and 8.5; CHES-NaOH, pH 8.5, 9, and 10. The reaction was allowed to proceed at 30°C and terminated by the addition of 0.9 ml of p-hydroxybenzoic acid hydrazide reagent and then heating (15). After the mixture was cooled, the absorbance of the reaction mixture was read at 410 nm with a spectrophotometer. The enzyme activity was monitored at four time intervals within a 6-h period to ensure linearity of the reaction. The activities of the purified coat ␤-glucanase fraction, the purified coat xylanase fraction, and Aspergillus niger ␤-glucanase (EC 3.2.1.4, from Sigma) on various substrates (xylan, lichenan, carboxymethylcellulose, and polygalacturonic acid; Sigma) in 0.05 M sodium acetate, pH 5.2, were assayed as described for laminarin. Activities of the various enzyme preparations on 0.15 mg of p-nitrophenyl ␤-D-glucopyranoside (4-NPG; Sigma) were assayed in 0.3 ml of 0.05 M sodium acetate, pH 5.2, at 30°C. The reaction was terminated by the addition of 0.6 ml of 4% (w/v) Na 2 CO 3 (16), and the absorbance of the reaction mixture was read at 410 nm with a spectrophotometer.
SDS-PAGE of Proteins for Separation, Antibody Preparation, and Peptide Microsequencing-Acrylamide (12.5%, w/v) SDS-PAGE was performed as described (5). The proteins in the gels were stained with Coomassie Blue. For antibody preparation, gel slices containing the 10and 17-kDa proteins from the pollen-released and interior fractions, respectively, were cut from the gel and used to produce antibodies in chicken (5). For microsequencing, fractionated proteins in a gel were transferred to a polyvinyldene fluoride membrane (Millipore Corp., Bedford, MA) (9) and sequenced at their N termini at the Genomic Institute at the University of California, Riverside (Riverside, CA). Alternatively, the proteins in the gel were subjected to trypsin digestion (for the 14- Analysis of Nucleotide and Amino Acid Sequences-We used the NCBI (www.ncbi.nlm.gov) and TIGR (www.tigr.org) data bases to search for EST nucleotide sequences or protein sequences. The GCG Program (gcg.ucr.edu) was used to compare nucleotide sequences or amino acid sequences, to construct phylogenetic trees and to translate nucleotide sequences into amino acid sequences. Subcellular locations of proteins were analyzed by use of the PSORT Program in ExPASy (us.expasy.org).
RNA Blot Hybridization-Each sample of 30 g of total RNA was fractionated with the use of a 1.2% formaldehyde gel by electrophoresis. The gel was equilibrated in 10ϫ SSC, pH 7.0 for 20 min. After equilibration, the RNAs were blotted onto a Hybond-N membrane (Amersham Biosciences). The RNA-blotted membrane was prehybridized at 65°C in potassium phosphate, pH 7.2, 7% SDS, 1% bovine serum albumin, and 0.01 M EDTA, pH 8.0 for 4 h; hybridized with 32 P-labeled probes (preceding paragraph) overnight; and then washed with 2ϫ SSC, 0.1% SDS for 20 min; 1ϫ SSC, 0.1% SDS for 20 min; and 0.1ϫ SSC, 0.1% SDS for 20 min, all at 65°C.
5Ј-Rapid Amplification of cDNA Ends (5Ј-RACE)-It was performed with the use of a 5Ј-RACE system (Invitrogen) according to the provided instructions. The first-strand cDNA was synthesized with a primer, 5Ј-GTTGTGGGGGAAGAT-3Ј. Primary and nested PCR products were synthesized with the primers, 5Ј-GGCGTTGTAGACGTCGTTGTT-3Ј and 5Ј-CCGTGCACGGCGTCGATGC-3Ј, respectively. The above three primers were designed on the basis of a maize gDNA clone, BZ402366, which is highly similar to the 5Ј-terminus of a rice EST clone, TC81322.
The sequence of the new maize gene, termed ZmGLA3 (registered as GenBank TM AY344632), was assembled from the sequenced 5Ј-RACE product, maize EST TC163747 and maize gDNA BZ402366, BZ533772, BZ530441, and BZ533777.
Microscopy-Samples of pollen before and after protein release were observed under a Zeiss Axiophot microscope and then photographed. For immunofluorescence microscopy, germinated pollen grains having different tube lengths in the germination medium were fixed by mixing with 1 volume of 2ϫ fixation solution (1ϫ: 4% (w/v) paraformaldehyde, 50 mM PIPES buffer, pH 6.9, 2 mM MgSO 4 ), and 15% (w/v) sucrose). The preparation was placed at room temperature for 1 h. The pollen grains were transferred to a 1ϫ fixation solution (omitting sucrose) at room temperature for 1 h. After fixation, the pollen grains were washed with PBS (10 mM phosphate-buffered saline, pH 7.4, 138 mM NaCl, 2.7 mM KCl) three times for 15 min each. The pollen grains were blocked with 3% (w/v) nonfat dry milk in PBS at room temperature for 2 h. The blocked pollen grains were allowed to react with chicken antibodies against the maize pollen 10-or 17-kDa protein (1:100 dilution) in PBS containing 1% (w/v) nonfat dry milk at 4°C overnight. After being washed with PBST (PBS containing 0.05% (v/v) Triton X-100) 3 times for 10 min each, the pollen grains were incubated with Cyanine 3-conjugated donkey secondary antibodies against chicken IgG or IgY (Jackson Immuno Research Lab., West Grove, PA) at room temperature for 2-3 h. After being washed with PBST three times for 10 min each, the pollen grains were mounted in an antifade reagent from SlowFade Antifade Kit (Molecular Probes, Eugene, OR) and viewed under a Leica SP2 UV confocal microscope.

Mature Maize Pollen Was Separated into Fractions of Distinct Structures or
Origins-Mature pollen was separated into four distinct fractions, the coat, a pollen-released protein fraction, the interior, and the shell. These fractions were analyzed for their protein constituents by SDS-PAGE.
A coat fraction was obtained by washing the mature pollen with diethyl ether. SDS-PAGE revealed three visible protein bands, which represented only a very small proportion of the total pollen proteins (Fig. 1A). N-terminal sequencing and FPLC revealed that the 70-and 35-kDa protein bands each represented one protein (to be described) and that the 25-kDa protein band contained at least two fragments of different proteins via a search of GenBank TM protein data bases. These fragments were not studied further.
Fractions of pollen-released proteins and pollen interior proteins were obtained by the following procedure. Mature pollen was shaken gently in a liquid medium of 10 mM sodium acetate, pH 5.2 for 20 min, during which most of the pollen (about 95%) did not burst (Fig. 1, B and C). The proteins released from and those retained in the pollen were analyzed by SDS-PAGE. The released proteins in the designated pollen-released protein fraction represented about 10% of the total pollen proteins and were separated into several sharp protein bands on the gel (Fig.  1A). The proteins retained in the pollen in the designated pollen interior fraction were separated into many bands. The pollen-released and interior proteins resolved in the gel were mostly non-overlapping, which indicates the selectivity of the separation procedure. After the almost complete liberation of the pollen-released proteins, the pollen still retained its interior density as observed by light microscopy (Fig. 1, B and C). Both the pollen-released and interior fractions contained some of the coat proteins, which are not visible from the stained SDS-PAGE gel because of their relatively minute amounts (Fig. 1A) but were revealed by immunoblotting with antibodies against the 35-kDa proteins (data not shown) A pollen shell fraction, representing the pollen after the pollen-released and interior proteins had been removed, was obtained by gently grinding the interior fraction and subjecting the ground materials to sucrose gradient centrifugation. A clean fraction of individual pollen shells was obtained (Fig. 1D). The aperture on each pollen shell was visible by high magnification microscopy (inset in Fig. 1D). No proteins were detected in this shell fraction (Fig. 1A). Thus, no abundant proteins were tightly associated with the pollen wall.
The Coat 70-kDa Protein Is an Active ␤-Glucanase Whose Gene/Protein Has Not Been Previously Studied-Of the two coat proteins, the 35-kDa protein has been shown to be an endoxylanase (9). We purified the 70-kDa protein to homogeneity from the coat fraction by FPLC ( Fig. 2A) and subjected it to trypsin digestion. One trypsin fragment was sequenced. The sequence of this fragment, YFVGSVLSGG, closely matches those of proteins encoded by two maize genes (9 of 10 residues, ␤-glucanase ExoI, termed ZmGLA1 for convenience in the current report; and 7 of 10 residues, ␤-glucanase ExoII, termed ZmGLA2 in the current report) (Fig. 3A); the former protein was studied in relation to its ␤-glucanase activity in young shoots (19). It also closely matches those of proteins encoded by two barley genes (9 of 10 residues, ␤-glucanase ExoI, HvGLA1; and 7 of 10 residues, ␤-glucanase ExoII, HvGLA2); both barley proteins were studied in relation to their ␤-glucanase activities in seeds (20). It completely matches a protein encoded by a full-length rice EST sequence (10 of 10 residues, OsGLA3) derived from maturing flowers. We used the rice EST sequence to search for and obtain two related, short maize gene sequences (BZ402366 and TC163747). On the basis of these rice and maize gene sequences, we designed primers for RT-PCR and obtained the 5Ј-sequence of a maize cDNA; this new gene, termed ZmGLA3, encodes the coat 70-kDa protein (see next section).
We explored whether the coat ␤-glucanase was synthesized in the tapetum or the pollen interior. Individual anthers were dissected into a pure microspore fraction and an anther wall fraction, which contained all of the outer anther cells, the inner tapetum cells and about 10% of the original microspores. RT-PCR analyses revealed that the transcripts of ZmGLA3 encoding the coat ␤-glucanase and the gene encoding the pollenreleased 10-kDa protein (to be described) were present exclusively in the anther wall fraction and the microspore fraction, respectively (Fig. 3B). Apparently, the coat ␤-glucanase is similar to the coat xylanase (Ref. 9 and Fig. 3B) in that they are synthesized in the tapetum cells and released into the locule and then onto the microspore surface.
The sequence of the nascent coat ␤-glucanase, as deduced from the gene sequence, contains a putative N-terminal signal peptide (Fig. 3A). The enzyme is apparently secreted from the tapetum cells via the endoplasmic reticulum (ER)-vesicle-exterior secretory pathway. This release mechanism is different from that for the xylanase, which is apparently a cytosolic protein that is released from the tapetum cells after cell death and activated by extensive proteolysis (9).
The Pollen Coat ␤-Glucanase Is Different from the Well Known ␤-Glucanase That Hydrolyzes the Callose Wall of the Microspore Tetrad-During early microsporogenesis, the tapetum secretes ␤-glucanases that hydrolyze the callose wall of the microspore tetrad to produce solitary microspores (21). These biochemical findings were initially obtained from large lily flowers. The peaking of the enzyme activities during anther development coincided with the formation of solitary microspores, and thereafter the enzyme activities disappeared. In subsequent studies, genes encoding ␤-glucanase in anthers or tapetum in several species were cloned (22) and assumed to be those encoding the ␤-glucanase that hydrolyzed the callose wall of the microspore tetrad.
We examined whether the coat ␤-glucanase is the same enzyme that hydrolyzes the callose wall of the microspore tet- The enzymes are from maize (ZmGLA1 from ␤-glucanase ExoI derived from the DNA sequence of AF225411; ZmGLA2, from ␤-glucanase ExoII, AF064707; ZmGLA3, our registered gene AY344632), barley (HvGLA1 from ␤-glucanase ExoI, AF102868; HvGLA2 from ␤-glucanase ExoII, U46003; HvGLA3 (a partial EST), assembled from BU995244, BU995937, and BQ468423) and rice (OsGLA1, TC124353; OsGLA2, TC113312; OsGLA3, TC81322). HvGLA1, HvGLA2 and ZmGLA1 have been shown to be exo-␤-glucanases; for convenience, all the cereal enzymes are termed ␤-glucanase (GLA) in this figure. The shaded sequences were those used in the current study to identify the coat protein ␤-glucanase to ZmGLA3; specifically, a valine (in bold) in lieu of isoleucine is present in OsGLA3 and ZmGLA3. The rad. Developing anthers were divided into five stages (see "Experimental Procedures"). Microspores were in the tetrad phase in stage 1 and the solitary phase in stage 2. The transcript of the gene (ZmGLA3) encoding the coat ␤-glucanase was absent in stage 1, appeared in stage 2, peaked at stage 3, and declined in stage 4 (Fig. 3C). Clearly, the coat ␤-glucanase is not the enzyme that hydrolyzes the callose wall of the microspore tetrad. In contrast, transcripts of the other two known maize ␤-glucanase genes (ZmGLA1 and ZmGLA2) were highest in stage 1 and disappeared gradually thereafter. It is likely that ZmGLA1 and ZmGLA2 encode the recognized ␤-glucanases that hydrolyze the callose wall of the microspore tetrad. Nevertheless, the possibility of other genes encoding unknown ␤-glucanases for callose hydrolysis cannot be excluded. We conclude that the coat ␤-glucanase is different from those hydrolyzing the callose wall of microspore tetrads in the anthers. The coat ␤-glucanase could, along with the coat xylanase (9), hydrolyze the stigma wall during pollen germination and tube growth.
A phylogenetic tree of related ␤-glucanases in cereals available in GenBank TM was constructed on the basis of amino acid sequences (Fig. 3D). These ␤-glucanases presumably can hydrolyze 1,3:1,4-␤-glucan. They can be divided into three groups, each consisting of a maize ␤-glucanase. Members of the group that includes the maize coat ␤-glucanase have not been previously characterized, whereas those in the other two groups have been studied (the maize ExoI (ZmGLA1) in seedling shoots, Ref. 19  The current results show that the ␤-glucanases in the latter two groups are present not only in vegetative organs but also in anthers. The finding that the three maize ␤-glucanases scatter in the tree indicates that their genes are not products of recent gene divergence but encode specific ␤-glucanases for different physiological functions. Several studied ␤-glucanases in dicots are relatively dissimilar to the cereal ␤-glucanases and are not included in the phylogenetic tree.
Proteins Were Released from the Pollen and Tube During Germination-We studied the liberation of the proteins of the pollen-released protein fraction (Fig. 1A). We allowed mature pollen to germinate in a liquid germination medium (of a pH of ϳ5.0) and then analyzed the medium by SDS-PAGE for proteins that had been released from the pollen. Many specific proteins were present in the medium after a 10-min incubation (Fig. 4). At the end of this incubation, the length of the tube reached about half the diameter of the pollen grain. Attempts to analyze the released proteins after a longer time of incubation were unsuccessful because the longer tubes broke during removal of the grains to recover the medium. We switched to a medium of 10 mM sodium acetate, pH 5.2, in which the pollen did not germinate, and only about 5% of them burst after incubation for up to 1 h. Thus, we could recover most of the pollen-released proteins for detailed analysis. The pattern of proteins released by the pollen in this medium after a 10-min incubation was similar to that in the germination medium (Fig.  4). After a 60-min incubation, most of the readily released proteins were recovered in the medium, and these proteins were distinct from those retained in the pollen grains. These released proteins pre-existed in the mature pollen, because the sum of the released proteins and the retained proteins matched the total proteins of mature pollen. The released proteins represented about 10% of the total proteins, as estimated from the stain intensity of all the proteins on the gel (Figs. 1 and 4). Light microscopy revealed that their removal did not affect the appearance of the remaining pollen grains (Fig. 1, B and C). The proteins are mostly secretory proteins because their nas-cent proteins contained a removal N-terminal ER targeting sequence (see next section). They should be present in secretory vesicles in the cytoplasm and/or in the wall of mature pollen.

The Pollen-released Proteins Could Loosen or Form New Tube Walls or Loosen or Hydrolyze the Transmitting Track
Wall-After separation by SDS-PAGE (Fig. 4), the major pollen-released proteins, or their fragments after trypsin digestion, were subjected to N-terminal sequencing. The proteins were identified via the GenBank TM or TIGR data bases according to amino acid sequences. Their identities suggest that they could loosen and form new tube walls for tube elongation or loosen and hydrolyze the transmitting track wall for tube advance. We briefly describe the proteins.
The 10-and 35-kDa proteins are identified as expansin-like allergen and ␤-expansin, respectively (Ref. 10 and Table I). These two specific ␤-expansins have not been previously reported, even though several pollen expansins of maize and other cereals are known (24,25). The 35-kDa ␤-expansin has the binding and catalytic domains of an expansin, which catalyzes the break-and-reunion of non-covalent bonds in the cell wall (10). It may loosen the wall of the tube or the transmitting track for tube advance. The 10-kDa expansin-like allergen has only the binding domain (in the allergen group II/III, Ref. 13) and may retard cellulose crystallization at the tube tip or the transmitting track for tube advance (26).
The 14-and 23-kDa proteins are identified as pollen allergens described in maize (27) and Lolium (28), respectively ( Table I). The 14-kDa allergen is also known as a profilin (27). Profilins are known to be allergens released from pollen of several cereals (13). The possible function of profilin, which is abundant in maize pollen (Fig. 1), is puzzling. Profilin is supposed to be involved in cytoskeleton organization inside the cell. Unlike all other pollen-released proteins we studied (Table  I), the nascent 14-kDa profilin does not have a putative Nterminal ER-targeting signal peptide. Thus, the protein is supposed to be cytosolic and not released to the exterior. It is likely that profilins can somehow be easily released from pollen and perform a yet-to-be-determined function on the tube wall or in the transmitting track. The 23-kDa allergen is also known to be related to trypsin inhibitors in the amino acid sequence, except that the inhibitor active site in the allergen is absent (28). Its function in the pollen as a released protein is unknown.
A 20-kDa protein is identified as a wall-modulating protein previously studied as an inhibitor of pectin methylesterase (29) or inhibitor of invertase (30) in non-pollen tissues ( Table I). The common features of these two enzymes include the presence of a sugar-binding motif and the reversibility of the enzymic reactions. The 20-kDa maize pollen protein released from germinated pollen may modulate the activities of these two enzymes or other related enzymes in the tube wall or the transmitting track wall and facilitate tube elongation. Alternatively, its sugar-binding motif may retard cellulose crystallization at the tube tip and enhance tube growth (26).
The 25-and 38-kDa proteins are tentatively identified as wall structural proteins called extensins (Table I), which have motifs for binding among themselves and to sugar residues of the wall (31)(32)(33). The 25-kDa protein could be a hydroxyproline-rich chimeric extensin (33) because of its similarity to an Arabidopsis hydroxyproline-rich chimeric extensin (7 of 9 sequenced residues identical (NP_200917)), which was so identified via a studied Chlamydomonas protein (34). The 38-kDa protein could be a glycine-rich extensin because of its similarity to a bacteria Ca 2ϩ -binding protein (8 of 10 sequenced residues identical (NP_437273, Ref. 35)), which in turn is similar to an Arabidopsis extensin (AAK07681, Ref. 36). Maize pollen-synthesized extensins of higher molecular weights have been reported (37).
Finally, a 47-kDa protein is identified as a polygalacturonase ( Table I). The gene encoding this exact maize polygalacturonase has been described (11). Polygalacturonate is a major component of pectin in the cell wall, even though monocot cell walls do not have abundant pectin. The polygalacturonase could hydrolyze the pectin between adjacent cells in the transmitting track and facilitate tube advance.
A Novel, Abundant, and Potentially Cation-chelating Protein with Three Consecutive Finger Motifs Is Associated with the Wall of the Pollen and Pollen Tube-An abundant 17-kDa protein, representing 1-2% of the total pollen proteins, was not readily released from the mature pollen in a mildly acidic medium (Figs. 1 and 4). It is identified (14 of 14-sequenced residues) as a protein encoded by a maize EST (TC84818) derived from mixed anthers and pollen (Fig. 5A). The maize protein is closely related to a sorghum protein encoded by an EST (Fig. 5A). The maize EST sequence has 1,153 nt and an open reading frame that encodes a protein of 192 residues. The N-terminal 25 residues of this protein are predicted (via PSORT in ExPASy) to represent a removable ER-targeted sequence. In our study, 30 residues had been removed from the N terminus of the nascent protein to produce the mature protein (Fig. 5A), which has 162 residues and 17,814 daltons.
Other than having a removable N-terminal, ER-targeting signal peptide, the 17-kDa protein has no other recognized organelle-targeting sequences, transmembrane segment, or glycosylation site. The mature protein is hydrophilic and has a predicted pI of 8.8. Although it is a secretory protein and present as the processed form in mature pollen (Fig. 1), it remained associated with the pollen after our washing/germination procedure in a pH 5.2 medium (the presumed pH in cell wall) that removed many other pollen-released proteins in their entirety (Figs. 1 and 4). This association was not tight, because the protein was absent in the pollen shell fraction that was isolated after gentle grinding and gradient centrifugation (Fig. 1A). In germinated pollen, the protein was associated with the wall of the protruded tube (Fig. 5C).
A striking feature of the sequence of the 17-kDa protein is the presence of 14 cysteine residues in a unique pattern not found in any other protein. The N terminus of the mature protein has a 21-residue sequence that possesses no cysteine but many hydrophilic residues (Fig. 5A). This sequence is followed by six tandem repeats of CX 3 CX 10 -15 (C for cysteine and X for any residue other than cysteine) and then by a seventh slightly modified CX 3 CX 10 -15 repeat at the C terminus. The six tandem repeats are also present in a protein derived from a sorghum EST sequence, which does not have the seventh repeat (Fig. 5A). Two such tandem repeats could form a finger loop motif that brings the four cysteine residues to proximity via binding of a divalent cation (Fig. 5B). Thus, the 17-kDa protein could form three consecutive finger motifs. Finger motifs binding to zinc are present in steroid hormone receptors (a single finger, with four cysteines) and transcription factors TABLE I Major proteins extracted from maize pollen coat (the first two) and released from mature and germinated pollen (the remaining nine) The proteins are listed according to their size, and their resolution by SDS-PAGE can be found in Fig. 1   High stringent washing was applied to the radioactive blot so that only RNAs from the designated genes were observed (see "Experimental Procedures"). The genes encode tapetum-synthesized coat 35-and 70-kDa proteins and microspore-synthesized released proteins (listed and named in Table I). Ethidium bromide-stained 25 S and 16 S rRNA in the gel reveal that equal amounts of RNA in each sample were used. Chicken antibodies against the protein and cyanine 3-linked donkey secondary antibodies against chicken IgY were used. Germinated pollen grains having different tube lengths were observed under a fluorescence microscope for fluorescence (circled 1, 3, and 5) or in a bright field (circled 2, 4, and 6). Germinated pollen grain in circled 5 and 6 was subjected to similar treatments but without chicken antibodies. Scale bar represents 40 m. under specific incubation and analysis conditions remains to be elucidated.
We speculate the functions of the abundant 17-kDa protein. The sequence of six tandem repeats of CX 3 CX 10 -15 is flanked by 21 residues at the N terminus and 33 residues at the C terminus. The N-terminal 21-residue sequence does not conform to any known functional motif in other proteins. The C-terminal 33-residue sequence contains a slightly modified CX 3 CX 10 -15 repeat that is absent in the sorghum analog (Fig. 5A). It is likely that the functioning of the 17-kDa protein rests largely on its six tandem repeats of CX 3 CX 10 -15 . These repeats could form three consecutive finger motifs that could chelate cations and act as a signal receptor. Divalent cations such as calcium are present in the cell wall; they chelate polysaccharides and maintain the cell wall structure. The 17-kDa protein could chelate divalent cations in the cell wall and thus increase wall fluidity of the pollen tube wall for tube extension, or of the transmitting track and thus soften resistance for tube advancement. Or, the three consecutive finger motifs, which are known to be versatile in specific binding to proteins (38), could interact with signal proteins in the transmitting track to guide the tube to the ovules. It is also possible that the cysteine residues form numerous sulfide linkages to generate a unique protein structure for a yet-to-be-determined function.
The Appearance of the Transcripts of the Coat Proteins in the Tapetum Preceded That of the Pollen-released Proteins in the Microspores During Anther Development-The transcripts of the coat xylanase (9) and ␤-glucanase (Fig. 3B) were present in the tapetum and not in the microspores of the anthers. The levels of the transcripts of both enzymes in anthers peaked at stages 2 to 3 of development (Figs. 3C and 6). In contrast, the transcripts of the pollen-released proteins were present in the microspore and not the tapetum, as exemplified in the transcript of the 10-kDa expansin being localized in the microspores (Fig. 3B) and shown in their presence in mature and germinated pollen (Fig. 6). The levels of some of these pollenreleased protein transcripts, especially the 10-kDa expansin transcript, remained fairly similar throughout development from stage 4 to germinated pollen (Fig. 6); it is unknown whether these transcripts in the mature and germinated pollen were those synthesized in the maturing microspores or were newly synthesized after pollen maturation. The levels of some other transcripts, especially the 23-kDa trypsin inhibitor and the 35-kDa ␤-expansin transcripts, continued to increase in mature and germinated pollen; apparently new mRNA were synthesized continuously.
The Pollen-released Proteins, as Exemplified by the 17-kDa Protein and 10-kDa Expansin-like Allergen, Were Present on the Wall of the Pollen Tube Tip-Immunofluorescence microscopy of germinated pollen grains having different tube lengths located the 17-kDa protein mainly in the pollen tube wall (Fig. 5). It also showed the presence of the 10-kDa allergen in both the wall and the tube tip (Fig. 7). In this in vitro analysis, a large proportion of the 10-kDa allergen would have been released from the pollen after 10 to 30 min germination (Figs. 1 and 4). Therefore, what we observed (Fig. 7) was the expansin-like allergen still in the wall and that stored or newly synthesized in the cytoplasm of the tube tip to be released to the exterior. CONCLUSIONS Although many proteins present on or released from mature pollen have been reported, their whereabouts on the pollen and their origins and functions are far from clear. We separated the two external domains, the coat and the wall (plus the released proteins), of a pollen grain and analyzed their protein origins, constituents, and physiological functions.
The coat has two proteins, xylanase (9) and ␤-glucanase, which are active enzymes synthesized in the tapetum at the same time during anther development. The two enzymes are transferred from the tapetum to the coat via different mechanisms for an unknown reason. Regardless, the two enzymes likely hydrolyze the wall of the stigma so that the pollen tube can enter the transmitting track of the style. The hydrolysis may be aided by the 10-kDa expansin-like allergen from the wall domain. This possibility is raised because the abundant allergen is released quickly from the pollen in a liquid medium (Fig. 4), and its mRNA accumulates earlier in the microspores than those of other released proteins (Fig. 6). The pollen-released proteins are synthesized in the microspore and pollen interior, and their synthesis in the anthers takes place after that of the coat proteins. They are abundant, representing about 10% of the total pollen proteins. The Nterminal ER-targeting signals in their nascent precursors have been removed, and they should be present in the wall of mature pollen and/or in secretory vesicles in the cytoplasm. In germinated pollen in vivo, they should be slowly released to the exterior. Most appear to interact with the wall of the pollen tube or the transmitting track of the style for advancement of the tube. The novel 17-kDa protein could act as a receptor to interact with signals in the style, perhaps to guide the tube to the ovules. We studied only the abundant pollen-released proteins. Minor proteins for possible signaling and other functions may exist and need to be studied.