Biochemical and Immunocytological Characterizations of Arabidopsis Pollen Tube Cell Wall

During plant sexual reproduction, pollen germination and tube growth require a development under tight spatial and temporal control for the proper delivery of the sperm cells to the ovules. Pollen tubes are fast growing tip-polarized cells able to perceive multiple guiding signals emitted by the female organ. Adhesion of pollen tubes via cell wall molecules may be part of the battery of signals. In order to study these processes, we investigated the cell wall characteristics of in vitro -grown Arabidopsis thaliana pollen tubes using a combination of immunocytochemical and biochemical techniques. Results showed a well-defined localization of cell wall epitopes. Low esterified homogalacturonan epitopes were found mostly in the pollen tube wall back from the tip. Xyloglucan and arabinan from rhamnogalacturonan I epitopes were detected along the entire tube within the two wall layers and the outer wall layer, respectively. In contrast, highly esterified homogalacturonan and arabinogalactan-protein epitopes were found associated predominantly with the tip region. Chemical analysis of the pollen tube cell wall revealed an important content of arabinosyl residues (43%) originating mostly from (1 → 5)- α -L-arabinan, the side chains of rhamnogalacturonan I. Finally, MALDI-TOF MS analysis of endo -glucanase-sensitive xyloglucan showed mass spectra with two dominant oligosaccharides (XLXG/XXLG and XXFG), both being mono O acetylated, and accounting for over 68% of the total ion signals. These findings demonstrate that Arabidopsis pollen tube wall has its own characteristics compared to other cell types in the Arabidopsis sporophyte. These structural features are discussed in terms of pollen tube cell wall biosynthesis and growth dynamics.


INTRODUCTION
Fertilization of flowering plants requires the delivery of the two sperm cells, carried by a fast growing tip polarized pollen tube, to the egg cell. In plants with dry stigma and solid style such as Arabidopsis thaliana, this process begins with the deposition and specific adhesion of the pollen grains on the stigmatic tissue, subsequent hydration of the pollen grains and germination of pollen tubes (Palanivelu and Preuss, 2000). Pollen tubes invade the papillae cell wall of the stigma, enter the short style, and grow through the apoplast of the specialized transmitting tract (TT) cells that is filled with a nutrient-rich extracellular matrix (Kandasamy et al., 1994;Lennon et al., 1998). During this invasive growth, pollen tubes are guided to the ovules via signals that need to pass through the cell wall to reach their membrane-associated or intracelllular targets (Lord and Russell, 2002;Kim et al., 2003;McCormick and Yang, 2005;Boavida et al., 2005;Johnson and Lord, 2006). In plant species with wet stigma and hollow style such as lily, adhesion between the pollen tube wall and the TT epidermis extracellular matrix is important for the growth of the pollen tubes toward the ovules (Mollet et al., 2000;Park et al., 2000;Chae et al., 2007;Mollet et al., 2007). In addition of being the interface between the tube cells and the surroundings (female sporophyte or culture medium), the pollen tube wall also controls the cell shape, protects the generative cells and allows resistance against turgor pressure (Geitmann and Steer, 2006;Geitmann 2010).
Most of our knowledge on cell wall polymers of higher plants comes from investigations on vegetative organs in which cells have diffuse growth. The cell wall is mainly composed of polysaccharides (cellulose, hemicellulose, pectin and occasionally callose, depending on the tissue) and proteoglycans (e.g. extensin and arabinogalactan proteins) forming a complex network with processing enzymes.
Xyloglucan (XyG) is the major hemicellulosic polysaccharide of the primary wall of flowering plants. Classic XyG consists of a (1→4)-β-D-glucan backbone substituted with xylose, galactose-xylose or fucose-galactose-xylose motifs which correspond, according to the one letter code proposed by Fry et al. (1993), to X, L and F, respectively, G being the unsubtituted glucosyl residue of the glucan backbone. The main XyG fragments released after endo-glucanase treatment of the cell wall from wild type Arabidopsis vegetative organs are generally XXXG, XXLG/XLXG, XXFG and XLFG (Zablackis et al., 1995;Lerouxel et al., 2002;Nguema-Ona et al., 2006;Obel et al., 2009). In addition, O-acetylation of XyG can occur, most generally on the galactosyl residues, but its biological function is unknown (Cavalier et al., 2008). In the primary wall, XyG interacts with cellulose microfibrils via hydrogen bonds and participate in the control of cell expansion (Cosgrove, 1999).
Despite the importance of pollen tubes for the delivery of the sperm cells to the egg, little is known about the underlying molecular mechanisms that regulate the mechanical interaction of pollen tubes with female floral tissues. There are very scarce data concerning the different components of the pollen tube cell wall. Past approaches to characterize the pollen tube cell wall are limited to few plant species including Camellia (Nakamura and Suzuki, 1981), Lilium (Jauh and Lord, 1996;Mollet et al., 2002), Nicotiana (Rae et al.,1985;Li et al., 1995;Ferguson et al., 1998;Qin et al., 2007), Pinus (Derksen et al., 1999) and Zea (Rubinstein et al., 1995) and are mostly based on immunocytochemistry. These studies revealed that, depending on the species, the pollen tube cell wall contains epitopes that are found in the polymers described above including HGs with varying levels of methylesterification, AGPs, extensin-like proteins, and low amounts of cellulose. Unlike most other plant cells, callose, a (1→3)-β-glucan, is predominant and it is deposited in the wall 7 back from the tip. Moreover, it is deposited in regular intervals to form callose plugs that maintain the tube cell in the apical expanding region of the tube and separate the viable from the degenerating region of the tube (reviewed by Geitmann and Steer, 2006

In vitro pollen tube growth and β -glucan localization
In order to study the cell wall of Arabidopsis pollen tubes, sufficient material was necessary to perform the biochemical analysis. To achieve this, a large number of flowers, a good in vitro pollen germination rate with sufficient pollen tube length were required. As shown in Fig. 1A, high rates of pollen germination were obtained (67% ± 12) with an average of pollen tube length of 1248 µm ± 374 and a growth rate of 0.46 µm.min -1 ± 0.05 after 16h of culture in liquid medium. Moreover, 16h-old pollen tubes displayed typical tip organization ( Fig. 1A insert) without any plasmolysis event. These results showed that using these in vitro conditions, pollen tube length is approaching what we should expect in in vivo condition given the distance between stigma and ovules. By comparison, 6h-old pollen tubes were fourfold shorter with an average length of 307 µm ± 102 and an estimated growth rate of 1.6 µm.min -1 ± 0.17.

8
Cytochemical staining of β -glucans with aniline blue (for callose) and calcofluor white (mainly callose and cellulose) was performed in order to visualize the distribution of these β glucans along the tube wall and within the tube. Aniline blue staining was not detectable in the tip region and was mainly localized in the wall, back from the tip, and within the tube as callose plugs that are periodically synthesized ( Fig. 1B; Table I). In contrast, staining with calcofluor white was uniform along the whole tube and the tip. The bright regions within the tubes revealed several plugs (Fig. 1C). At the ultrastructural level, the Arabidopsis pollen tube wall back from the tip appeared as a bi-layered structure consisting of a fibrillar outer and a weakly electron-dense inner wall layer ( Fig. 2A). Immunogold localization of callose resulted in dense labeling in the inner wall layer back from the tip (Supplemental data, Fig. S1A).

Immunolocalization of Arabidopsis pollen tube cell wall epitopes
Immunolabeling pattern of pectin domains (Table I). Probing with JIM7 and JIM5 that recognize partially and highly methylesterified HGs revealed that JIM7 epitopes were dominantly localized in the tip region of the pollen tubes (Fig. 1D). Similar observations were made at tube tips emerging from pollen grains (Fig. 1D insert). In contrast, JIM5 staining was observed mostly back from the tip with brighter ring-like deposits presumably originating from successive temporary slow-downs in the growth of the tubes (Fig. 1E). LM8, specific for xylogalacturonan epitopes, uniformly labeled the whole pollen tube wall (Figs. 1F, 1F insert).
(1→5)-α-L-arabinan epitopes, localized with LM6 and LM13, were clearly detected in the entire pollen tube wall (Figs. 1G, 1I) with stronger fluorescent signal at the pollen tube tip (Figs. 1H,1J). A ring-like structure was also noticeable as a collar at the emergence of the pollen tube from the pollen grain (Fig. 1G). Immunogold labeling of epitopes recognized by LM6 was mostly found in the fibrillar outer wall layer (Figs. 2B-2C). A strong concentration of gold particles was also observed over densely stained vesicles possibly associated with the trans Golgi network (Fig. 2B). Epitopes recognized by LM6 were also detected in the intine wall of the pollen grain (Supplemental data, Fig. S1B).
Immunolabeling of xyloglucan motifs. To investigate the distribution of XyG, pollen tubes were probed with the MAbs LM15 and CCRC-M1 that recognize non-galactosylated (XXXG motif) and fucosylated XyG domains, respectively. As shown in Figs. 1K and 1L, both MAbs bound to the entire pollen tube wall.

9
TEM observations showed that fucosylated XyG epitopes probed with the MAb CCRC-M1 were mainly found in the inner pollen tube wall (Fig. 2D), whereas the nongalactosylated XyG epitopes tagged with LM15 were detected in both the inner and outer layers of the wall (Figs. 2E-2F; Table I). Immunolabeling of pollen grains with the LM15 and CCRC-M1 MAbs resulted in a strong detection of XyG epitopes, mostly in the intine wall and at the vicinity of the plasma membrane (Supplemental data, Figs. S1C-S1D).
Immunolabeling of arabinogalactan-proteins. Epitopes of AGPs recognized by the MAb LM2 were found slightly labeled throughout the pollen tube wall with a stronger signal in the tip region (Figs. 1M-1N; Table I). Similar results were observed with MAC207 whereas JIM13 did not label the tube cell wall (Table I).
Immunolabeling at fluorescent microscopy level was carried out on 6h and 16h-old pollen tubes and did not show any difference in labeling pattern between the two time points (data not shown). Controls for immunofluorescent labeling did not show any fluorescence of the pollen tubes and a weak autofluorescence of the pollen grain exine (data not shown).
Controls for immunogold detection did not exhibit any significant non specific gold particles on the resin, pollen tubes and pollen grains (Supplemental data, Figs. S1E-S1H). A summary of immunolocalization data on the Arabidopsis pollen tube wall is shown in Table I.

Contribution of pollen grain wall versus pollen tube wall for biochemical analysis
We estimated the relative abundance of the pollen tube cell wall versus the pollen grain cell wall by measuring the cell wall thickness based on MET observations, pollen germination rates and pollen tube length (Supporting Table SI

Monosaccharide composition of the Arabidopsis pollen tube wall
Monosaccharide analysis of Arabidopsis pollen tube cell wall (Table II) displayed a high level of glucose (19.8 mol%) originating probably mostly from callose and to a lesser extent from XyG polymers. The most abundant monosaccharide, arabinose (43.6 mol%), and galactose (8.4 mol%) may come from AGP glycans and/or the pectic RG-I domain. Indeed, both sugars composing the backbone of RG-I (i.e. rhamnose and galacturonic acid) were present in the extract in equimolar amounts (~5 mol%). The remaining galacturonic acid (~5 mol%) composes undoubtedly the pectic HG backbone. XyG polymers generally consist of Xyl and Fuc residues that are also found in this analysis although in low amounts (1.2 mol% Fuc and 6.2 mol% Xyl) suggesting that XyG is not a major cell wall component of pollen tube cell wall.
The sequential extraction of the Arabidopsis pollen tube cell wall with ammonium oxalate and KOH aimed at yielding pectin-and hemicellulose-enriched fractions. Both extracts were analyzed for monosaccharide composition by gas chromatography (GC) and linkage/substitution sites by GC-mass spectrometry (MS) ( Table II). The oxalate extract showed similar levels of uronic acid content by colorimetric assay (21.5 mol%) and GC (24.9 mol%) analysis. This extract was enriched in galacturonic acid (21.5 mol%) indicating the presence of the pectin domains HG and highly branched RG-I, as shown by the detection of 4.2 mol% of rhamnose, 24.7 mol% of arabinose and 18.3 mol% of galactose. These data are supported by the immuno-dot assay (Fig. 3) showing a strong interaction of the oxalate extract with JIM5 for partially methylesterified HG and LM6 for (1→5)-α-L-arabinan. This labeling was consistent with the presence of 28.2% of 5-Araf (Table II). Labeling with LM5 was weak which suggests a low amount of (1→4)-β-D-galactan. This observation was confirmed by the detection of only 1.5% of 4-linked galactosyl residues (Table II). Instead, galactosyl residues detected in the monosaccharide analysis originate probably from type II arabinogalactans as shown by the detection of 3-Gal, 6-Gal and 3,6-Gal (Table II).
In the KOH extract, levels of HG (galacturonic acid) and RG-I backbone (rhamnose and galacturonic acid) decreased significantly compared to the oxalate extract but the remaining RG-I domains were probably highly substituted by arabinans and/or arabinogalactans, as both arabinosyl and galactosyl residues counted for 37 mol% of the total sugars (Table II). The dot-blot immuno-assay (Fig. 3)  signal in the KOH extract with JIM5 compared to the oxalate extract indicating that KOH treatment had substantially removed methylester groups from the partially methylesterifed HG leaving low amount of non-esterified HG. In contrast, labeling for (1→5)-α-L-arabinan with LM6 was still strong (Fig. 3) and was confirmed with the detection of 14.6% of 5-Araf (Table II). The different components of XyG (i.e. Glc, Xyl, Gal and Fuc) were also present in the KOH extract (Table II), with a large increase in the glucose level (40.1 mol%) compared to the oxalate extract. As revealed by methylation analysis, the latter originated from XyG (2.2% 4,6-Glc), starch and/or amorphous cellulose (15.4% 4-Glc) but dominantly from callose with the presence of 3-Glc (27.6%) and possibly 2,3-Glc (7%) which may suggest that callose can be substituted on the C2 of the glucosyl residues (Table II). The MAb LM15 reacted strongly with both oxalate and KOH extracts ( Fig. 3) suggesting that XyG were present in both extracts.

Spectrometry (MS) analysis of pollen tube cell wall xyloglucan fragments
Analysis by MALDI-TOF MS of XyG fragments released after xyloglucan endoglucanase treatment of Arabidopsis pollen tube cell wall ( Fig. 4A) showed the typical ions, XXXG, XLXG/XXLG and XXFG (14.7%, 5.7% and 8.4%, respectively), found in XyG (Table III). However, these oligosaccharides were not the main motifs. Instead, the two principal ions identified (XLXG/XXLG and XXFG) present a shift of m/z = 42, characteristic of an O-acetyl group, branching presumably the galactosyl residue. These two O-acetylated fragments count for over 68% of the total oligosaccharide released. Overall, galactosylated and fucosylated xyloglucan oligosaccharides showed a relative abundance of 85% and 52%, respectively. By comparison, the main XyG fragment released after endo-glucanase treatment of Arabidopsis leaf cell wall was XXXG (Supplemental data, Fig. S3) and the total level of Oacetylated fragments was 33%. Moreover, based on the signal intensity, the ratio of Fuc to Xyl can be estimated to 1:5.7 similar to what was found in the compositional analysis of the pollen tube wall extract (1:5.2) which suggests that other xylose-containing polymers such as xylan may not be present or in very low amount in the pollen tube cell wall. Finally, monosaccharide analysis of the endo-glucanase-resistant pollen tube cell wall residue (Fig. degraded by the enzyme, perhaps due to their tight association with other cell wall components.

DISCUSSION
Since the AGI (2000), increasing genomic, transcriptomic and proteomic data on Arabidopsis have become available (Becker et al., 2003;Honys and Twell, 2004;Tung et al., 2005;Qin et al., 2009;Zou et al., 2009). The use of male and female gametophyte or sporophyte mutants has allowed major insight into pollen formation, pollen tube growth, guidance and signaling (Rhee et al., 2003;Dong et al., 2005;Escobar-Restrepo et al., 2007), but the biosynthesis, the remodeling and the overall role of the cell wall of the tip growing pollen tube cell is far from being fully understood. Using the in vitro pollen tube method developed by Boavida and McCormick (2007), we obtained high levels of pollen germination ranging from 55% to 79% with pollen tubes that can expand over 1.5 mm after 16h of culture.
Based on this, we decided to investigate one of the main components, the cell wall, of this tipgrowing cell by means of immunolocalization at light and electron microscopy levels and analytical polysaccharide biochemistry. Our results showed that (1)  main component of the inner wall layer but it is not detectable at the tip whereas cellulose is weakly detected (Ferguson et al., 1998).

Arabidopsis pollen tube wall contains a subset of AGPs recognized by LM2
We found that AGPs were labeled in the whole tube with a stronger signal at the tip with LM2 whereas JIM13 labeling was almost absent. In contrast, JIM13 labeled lily pollen tube at the tip (Mollet et al., 2002), uniformly or in a ring-like deposition along the tube of tobacco (Li et al., 1992, Qin et al., 2007. Pereira et al. (2006) also noted a lack of labeling of Arabidopsis pollen tube with JIM13 unlike MAC207, which labels uniformly the whole tube (Coimbra et al., 2008). These divergent data depend on the MAb used (despite the apparent identical epitope, β-D-GlcpA-(1→3)-α-D-GalpA-(1→2)-L-Rha, recognized by MAC207 and JIM13) and the species investigated. These differences may suggest that the carbohydrate moieties of AGPs are species and/or cell type specific (vegetative and sperm cells) or that the cell wall organization is different in these species which may alter the accessibility of the epitopes.

HG pectin domains are present in different location in Arabidopsis pollen tube wall
Immunofluorescent labeling with JIM5 and JIM7 shows a pattern similar to that observed in pollen tubes from species possessing a solid style such as the Solanaceae (potato, tobacco, petunia), Oleaceae (jasmine) and Poaceae (corn) (Li et al., 1994;Qin et al., 2007). A dominant localization of highly methylesterified HG, recognized by JIM7, was present at the tip and a periodic patterning of low methylesterified HG epitopes, labeled with JIM5, was visible behind the tip region. Pollen tube wall contains 10 mol% of GalA, half of it being presumably dedicated to HG building, which explains the weak labeling by JIM5 along the wall compared to other species (Jauh and Lord, 1996;Qin et al., 2007) even though reduced accessibility of the MAb to the epitopes may occur due to other polymers. In fact, the pectinenriched extract isolated from the pollen tube wall showed a strong interaction with JIM5 ( Fig. 3). Highly methylesterified HG and other pectic motifs are deposited at the growing tip by Golgi-derived vesicles and de-esterified via pectin methylesterases (PMEs) in the wall back from the tip, to allow calcium cross linking of the carboxyl groups (Geitmann and Steer, 2006). In the tobacco pollen tube, two PME isoforms have been localized in the Golgi, secretory vesicles and at the outer surface of the plasma membrane where they may deesterify HG (Li et al., 2002). Two Arabidopsis mutants, vanguard1 (Jiang et al., 2005) and Atppme1 (about 20%) of overall PME activity, reduced growth compared to wild type pollen tubes but different pollen tube phenotypes. Vanguard1 pollen tubes were unstable in vitro resulting in tip bursting and in vivo vanguard1 plants show reduced male fertility. In contrast, atppme1 pollen tubes showed in vitro branching pattern with numerous tips but did not show male sterility in vivo. These reports suggest that these mutant pollen tubes have modified cell wall mechanical properties, despite a lack of either immunolocalization or biochemical data showing a modification in the methylesterification level of the HG motif. It is noteworthy to point out, as suggested by Jiang et al. (2005), that these PMEs may also have a role in modifying the female TT wall to facilitate pollen tube progression through the female tissue.

Arabidposis pollen tube wall is enriched in pectin Arabinan
In addition to HG motifs, Arabidopsis pollen tube wall contains highly branched RG-I with arabinan. Our results based on immunolocalization using LM6 and LM13, both the hydrolysis method) mostly 5-linked and to a lesser extent 5,2-linked. Moreover, tobacco pollen tube wall displayed low level of galactosyl residues and uronic acids (Rae et al., 1985).
Similarly, Lilium, Camellia and Tulipa pollen tube wall showed also high levels of arabinosyl residues (Nakamura and Suzuki, 1981). Analysis of Golgi vesicles from Camellia pollen tubes were also enriched in arabinose, galactose and uronic acid (Hasegawa et al., 1998). All together, our data suggest that Arabidopsis pollen tube wall contains short RG-I backbone harboring long chains of (1→5)-α-L-arabinans in the outer wall layer. The significance of this common feature (high level of arabinan) between the pollen tubes originating from non- MAbs directed against two different XyG epitopes labeled the whole tube mostly in the outer wall layer and the in the tip. Oligosaccharide fragments containing these two epitopes (XXXG and XXFG) were also detected with MALDI-TOF MS after pollen tube cell wall digestion. Interestingly, XyG analysis from vegetative organs (leaf, stem or root), revealed consistently that the main fragments are XXXG and XXFG (Zablackis et al., 1995;Pauly et al., 2001;Lerouxel et al., 2002). Arabidopsis pollen tube XyG shows distinct mass spectra with two major fragments detected (XLXG/XXLG and XXFG) both being substituted by one acetyl group, presumably linked to the galactosyl residue. The signal intensity of these two fragments consistently accounted for over 68%. been considered to have large capacity of interacting with water (Fu et al., 2007). All together, acetyl groups of XyG may promote hydrophobic interactions with other molecules within the pollen tube cell wall but also with other female components of the TT extracellular matrix during the intrusive growth of pollen tubes. Moreover, acetylation may modulate water uptake due to the non-polar property of these groups, but this requires further investigation.
In summary, this study provides a solid foundation for the use of Arabidopsis pollen tubes in cell wall biology and for the investigation of the male gametophyte cell elongation with mutants impaired in proteins implicated in the biosynthesis of specific cell wall polymers

Plant materials
Arabidopsis thaliana ecotype Columbia seeds stored at 4°C were spread on the surface of a sterile soil and cultured in a growth chamber with a photoperiod of 16h light/8h dark cycle at 20°C during the light phase and at 16°C in the dark phase with 60% humidity and daily watering. Only recently open flowers were collected.

In vitro pollen tube growth
Pollen was grown in vitro in a liquid medium according to the method described by tweezers and the pollen suspension was then pelleted down at 3200g for 6 min. New GM (250 µL) was added to the pellet and pollen grains were transferred into glass vials (14 x 45 mm) and grown in a growth chamber, in the dark at 22°C. For large scale culture, pollen from 200 flowers was grown in glass vials (25 x 50 mm) in 1.25 mL of GM. Using this harvesting method, the number of pollen grains collected per flower was estimated to 291 ± 63.3 (n = 10 independent experiments). Pollen tubes were grown for 6h for TEM sample preparation, 6h and 16h for immunolabeling at fluorescent microscopy level and 16h for biochemical analysis. Before any further manipulation, pollen germination and pollen tube growth were assessed under an inverted microscope.

Immunolocalization of Arabidopsis pollen tube wall epitopes
Pollen tubes in GM were mixed v/v with a fixation medium containing 100 mM PIPES buffer, pH 6.9, 4 mM MgSO 4 7H 2 O, 4 mM EGTA, 10% (w/v) sucrose, 5% (w/v) formaldehyde and incubated for 90 min at room temperature. Pollen tubes were rinsed 3 times by centrifugation with 50 mM PIPES buffer, pH 6.9, 2 mM MgSO 4 7H 2 O, 2 mM EGTA and 3 times with phosphate-buffered saline (PBS) (100 mM potassium phosphate, 138 mM NaCl, 2.7 mM KCl, pH 7.4) supplemented or not with 3% fat free milk. Primary antibodies were diluted at 1/5 or 1/10 as previously described (Mollet et al., 2002) with PBS (with or without 3% milk). Pollen tubes were rinsed with the buffer and incubated overnight at 4°C in the dark with the secondary antibody combined with fluorescein isothiocyanate (FITC, Sigma) diluted at 1/50 with the appropriate buffer for 3h at 30°C. For JIM, LM and MAC antibody detection, goat anti-rat (immunoglobulin G, whole molecule)-FITC was used, for CCRC antibody detection, sheep anti-mouse (immunoglobulin G, whole molecule)-FITC was used. Controls were carried out by incubation of the pollen tubes with the secondary antibody only.

Microscope observation and acquisition of pollen tube pictures
Pollen tubes were observed under Nomarski differential interference contrast (DIC) optics or fluorescence illumination on a Leica DLMB microscope equipped with FITC (absorption, 485-520 nm; emission, 520-560 nm wavelength) or calcofluor white filter sets.
Pictures were acquired with a Leica DFC300FX camera.
Pollen count and germination rate were estimated with a Nageotte chamber. A pollen grain was considered germinated if the pollen tube length was greater than the pollen grain diameter. Pollen tube length was measured from the picture using the nih program ImageJ (Abramoff et al., 2004). At least 350 pollen tubes were measured in three independent experiments.

High pressure freezing/Freeze substitution (HPF/FS) sample preparation.
Centrifuged 6h-old pollen tubes were transferred into the cavity of gold cupules (200 μ m in depth and 1.2 mm in diameter) coated with soybean lecithin (100 mg.mL -1 in chloroform).
Excess medium was removed using a filter paper. Then, samples were frozen using a highpressure freezing EM-PACT (Leica), according to a maximum cooling rate of 10000°C/s, incoming pressure of 7.5 bars and working pressure of 4.8 bars. Cupules containing frozen samples were stored in liquid nitrogen until the freeze substitution procedure was initiated.
After high-pressure freezing, samples were transferred to a freeze substitution automate (EM-AFS, Leica) precooled to -140°C. Freeze substitution conditions followed a modified procedure from D. Studer (personal communication). Substitution media were composed of 2% osmium in anhydrous acetone. Samples were substituted at -90°C for 72h.
The sections were stained with 0.5% (w/v) uranyl acetate in methanol for 10 min in the dark, rapidly rinsed 10 times with water and stained with lead citrate for 10 min and briefly rinsed 10 times with water.
Grids were observed at 80 kV with a TEM (Tecnai 12 Bio-Twin, Philips, FEI) and pictures were acquired with an Erlangshen ES500W camera (Gatan).

Cell wall extraction of Arabidopsis pollen tubes
16h-old pollen tubes from pollen grains from 8500 flowers collected manually were pooled after addition of three volumes of EtOH 95% to the GM. This experiment was performed three times to allow three independent replicates. Pollen tubes were centrifuged at 5000g and rinsed three times with EtOH 70% to remove salts and sucrose from the GM. The insoluble material (3 x 12 mg) was ground and then treated three times with EtOH 70% at 70°C for 15 min followed by incubation with a mixture of chloroform:methanol (1:1, v/v).
The remaining insoluble material was then dried to yield the cell wall fraction (3 x 10 mg). As baseline control, cell wall from non-germinated pollen grains was extracted. A similar extraction method was applied to Arabidopsis mature rosette leaves to yield a leaf wall extract except that an acetone incubation was added after the chloroform:methanol treatment for pigment removal. This extract was used to compare the XyG fragments analyzed by MALDI-TOF MS with the pollen tube cell wall extract.
www.plantphysiol.org on June 1, 2018 -Published by Downloaded from Copyright © 2010 American Society of Plant Biologists. All rights reserved.

21
As described by Ray et al., (2004), a pectin enriched fraction was obtained by extraction with boiling ammonium oxalate 0.5% (w/v) for 1h. After centrifugation, the supernatant was dialyzed against ddH 2 O and freeze-dried. The ammonium oxalate insoluble residue was then treated with 4M KOH supplemented with 20 mM NaBH 4 at room temperature for 12h. The alkaline soluble fraction was acidified to pH 5.5 with acetic acid, dialyzed against water and freeze-dried, to yield a hemicellulose-enriched fraction.

Analysis of cell wall polysaccharides
Photometric assays. Total uronic acid and neutral sugars were estimated with the mhydroxydiphenyl assay (Blumenkrantz and Asboe-Hansen, 1973) and the phenol sulfuric method (Dubois et al., 1956) with galactose and galacturonic acid as standards. Estimation of uronic acid and neutral sugar contents were calculated after correction of the interference of galacturonic acid with the phenol sulfuric and galactose with the mHDP assays.
Gas chromatography mass spectrometry. Monosaccharide compositions of the total cell wall, pectin and hemicellulose enriched fractions were determined by gas liquid chromatography according to York et al. (1985) using inositol as internal standard. Briefly, each fraction (500 µg) was treated with 1 M methanolic-HCl at 80°C for 16h and the free monosaccharides were converted to their methyl glycosides. After silylation at 110°C for 20 min, samples were dried, dissolved in cyclohexane, and analyzed using a GC 3800 Varian gas chromatography system equipped with a DB1 capillary column and a flame ionization

MALDI-TOF mass spectrometry analysis of xyloglucan derived oligosaccharides.
MALDI-TOF mass spectra were acquired on a Voyager DE-Pro MALDI-TOF instrument (Applied Biosystems) equipped with a 337 nm nitrogen laser. Mass spectra were performed in the reflector delayed extraction mode using 2,5-dihydroxybenzoic acid (Sigma) as matrix. The matrix, freshly dissolved at 5 mg.mL -1 in a 70:30 acetonitrile/ 0.1% TFA, was mixed with the water solubilized oligosaccharides in a ratio 1:1 (v/v). These spectra were recorded in a positive mode, using an acceleration voltage of 20,000 V with a delay time of 100 ns and above 50% of the laser energy. They were externally calibrated using commercially available mixtures of peptides and proteins (Applied Biosystems). In this study, the spectra were calibrated using des-Arg 1 -bradykinin (904.4681 Da), angiotensin I (1296.6853), Glu 1fibrinopeptide B (1570.6774 Da), ACTH18-39 (2465.1989. Laser shots were accumulated for each spectrum in order to obtain an acceptable signal to noise ratio (sum of 10 spectra of 1000 shots per spectrum).

Supplemental data
Supplemental
www.plantphysiol.org on June 1, 2018 -Published by Downloaded from Copyright © 2010 American Society of Plant Biologists. All rights reserved.