LlFH1-mediated interaction between actin fringe and exocytic vesicles is involved in pollen tube tip growth

Transcription

1 Research LlFH1-mediated interaction between actin fringe and exocytic vesicles is involved in pollen tube tip growth Shanwei Li*, Huaijian Dong*, Weike Pei, Chaonan Liu, Sha Zhang, Tiantian Sun, Xiuhua Xue and Haiyun Ren Key Laboratory of Cell Proliferation and Regulation Biology of Ministry of Education and College of Life Science, Beijing Normal University, Beijing , China Author for correspondence: Haiyun Ren Tel: Received: 31 August 2016 Accepted: 16 November 2016 doi: /nph Key words: actin fringe, exocytic vesicle, formin, lily (Lilium longiflorum), pollen tube, profilin, tip growth. Summary Pollen tube tip growth is an extreme form of polarized cell growth, which requires polarized exocytosis based on a dynamic actin cytoskeleton. However, the molecular basis for the connection between actin filaments and exocytic vesicles is unclear. Here, we identified a Lilium longiflorum pollen-specific formin (LlFH1) and found that it localized at the apical vesicles and plasma membrane (PM). Overexpression of LlFH1 induced excessive actin cables in the tube tip region, and downregulation of LlFH1 eliminated the actin fringe. Fluorescence recovery after photobleaching (FRAP) analysis revealed that LlFH1- labeled exocytic vesicles exhibited an initial accumulation at the shoulder of the apex and coincided with the leading edge of the actin fringe. Time-lapse analysis suggested that nascent actin filaments followed the emergence of the apical vesicles, implying that LlFH1 at apical vesicles could initiate actin polymerization. Biochemical assays showed that LlFH1 FH1FH2 could nucleate actin polymerization, but then capped the actin filament at the barbed end and inhibited its elongation. However, in the presence of lily profilins, LlFH1 FH1FH2 could accelerate barbed-end actin elongation. In addition, LlFH1 FH1FH2 was able to bundle actin filaments. Thus, we propose that LlFH1 and profilin coordinate the interaction between the actin fringe and exocytic vesicle trafficking during pollen tube growth of lily. Introduction Pollen tube growth is a fundamental process in sexual reproduction in plants (Johnson & Preuss, 2002; Franklin-Tong, 2010). In addition, the process interests cell biologists at large because the growth is fast and tip-focused, representing a mode of polar growth commonly found in eukaryotic cells. The regulation and maintenance of tip growth require the coordination of many cellular events, the most important elements of which are the cell cytoskeleton system and the vesicle transport system (Hepler et al., 2001). The actin cytoskeleton is essential for pollen tube tip growth. In the shank of the tube, numerous longitudinal actin cables or bundles are polarized and responsible for cytoplasmic streaming (Lenartowska & Michalska, 2008). However, in the tip region, different kinds of dynamic actin structure have been observed, depending on the plant species. For example, in the lily pollen tube, densely arranged actin filaments are largely located in the cortical region to form a structure called the actin fringe (Lovy- Wheeler et al., 2005; Kroeger et al., 2009). However, the subapical collar or mesh in the tobacco pollen tube is closer to the tip and denser in the middle than in the fringe seen in the lily pollen tube (Fu et al., 2001; Chen et al., 2002; Vidali et al., 2009a; *These authors contributed equally to this work. Cheung et al., 2010). In addition, pollen tube elongation relies on continuous exocytic and endocytic activities to deliver and recycle membrane. The endocytic and exocytic vesicles, plus the excess secretory vesicles recycled rearward at the tip of the pollen tube, form the inverted cone of vesicles within the clear zone (Parton et al., 2001; Bove et al., 2008). It is generally accepted that the apical and apical flank membrane domain defined by the inverted cone is the prominent region for exocytosis (Samaj et al., 2006; Campanoni & Blatt, 2007). Although essential for understanding the tip growth of the pollen tube, the precise exocytic site is still controversial. There are two possible spatial patterns of exocytosis: one occurs at the extreme tip (Lee et al., 2008; Cheung et al., 2010; Wang H et al., 2013; Rounds et al., 2014) and the other takes place in the plasma membrane (PM) at the shoulder (subapex) of the pollen tube (Bove et al., 2008; Zonia & Munnik, 2008; Kroeger et al., 2009; Cheung et al., 2010). Although, dynamic actin polymerization at the tip is assumed to be necessary for the delivery of the membrane to the site of expansion in the form of exocytic vesicles, the molecular mechanism underlying the connection between actin filaments and exocytic vesicles has not been described (Hormanseder et al., 2005; Lee et al., 2008; Kroeger et al., 2009). In pollen tubes, the configuration and dynamics of the actin cytoskeleton are regulated spatially and temporally by numerous actin-binding proteins (ABPs), including actin 745

2 746 Research Phytologist monomer-binding proteins, nucleation proteins, capping proteins, severing proteins and bundling proteins (Ren & Xiang, 2007; Cheung & Wu, 2008; Staiger et al., 2010). To date, numerous ABPs regulating longitudinal actin cables in the shank have been well identified and characterized in pollen tubes (Yokota et al., 1999, 2003; Wang et al., 2008; Ye et al., 2009; Papuga et al., 2010; Wu et al., 2010; Zhang et al., 2010). However, the formation and regulation of the actin fringe at the subapex by ABPs have been poorly studied. Some obvious candidates, for example profilin (Kovar et al., 2000), gelsolin/villin (Yokota et al., 2005; Xiang et al., 2007) and actin depolymerizing factor (ADF) (Feijo et al., 1999; Chen et al., 2002; Lovy-Wheeler et al., 2006), may play a particular role in the disassembly of the actin fringe. ADF and the actin fringe are located in a cytoplasmic alkaline region, and thus the existing actin fringe has been suggested to be depolymerized by the severing activity of ADF under high ph conditions (Feijo et al., 1999; Chen et al., 2002; Lovy-Wheeler et al., 2006). In the region of the tip-focused Ca 2+ gradient, it is reasonable to predict that gelsolin/villin would fragment the existing actin fringe (Yokota et al., 2005; Xiang et al., 2007). In addition to depolymerization, the actin fringe needs to be maintained and polymerized to keep up with the advancing tip. Several ABPs have been mentioned recently to be involved in the maintenance of the actin fringe in the subapex of the pollen tube, including fimbrin, villin and profilin (Wu et al., 2010; Su et al., 2012; Qu et al., 2013; Liu et al., 2015). Although Arabidopsis FH5 has been reported to participate in the nucleation of the actin assembly from the subapical PM for the subapical actin structure in tobacco pollen tubes (Cheung et al., 2010), it is still unknown where and by which mechanism the actin fringe initiates in lily. Previous studies have shown that profilin sequesters the majority of actin monomers in plant cells and that the actin/profilin complex represents the physiologically relevant source of actin monomers in plants (Wang et al., 2005; Chaudhry et al., 2007). Recently, one class II formin AtFH14 from Arabidopsis has been demonstrated to accelerate actin elongation in the presence of plant profilin (Zhang et al., 2016). However, it is unclear whether pollen-specific class I formin interacts with profilin to promote the rate of actin elongation. In this study, we identified a pollen-specific class I formin (LlFH1) from Lilium longiflorum and provided evidence for the coordination of LlFH1 with the actin fringe and exocytic vesicles. In addition, we demonstrated that LlFH1 could nucleate the actin assembly and accelerate the elongation rate of the actin filament in the presence of lily profilins. Our results also suggested that LlFH1 plays an important role in the delivery of exocytic vesicles to the shoulder of the apex during the growth of lily pollen tubes. Materials and Methods Plant materials, pollen tube germination and chemicals Pollen grains of Lilium longiflorum were collected from dehydrated anthers and used immediately for experiments or frozen in liquid N 2 and stored at 80 C. There are no differences in the germination rate and tube morphology between fresh and frozen pollen grains (Sommer et al., 2008). The pollen grains were germinated in a germination medium (GM) containing 1.6 mm H 3 BO 3, 1 mm KNO 3, 1 mm Ca(NO 3 ) 2, 1.6 mm MgSO 4 and 10% sucrose, ph 5.6. Single-cell PCR A modified method from Zhou et al. (2005) was used to perform single-cell PCR. Under an inverted fluorescence microscope (Axio Observer D1; Carl Zeiss) with a numerical aperture (NA) objective, one pollen grain or pollen tube with a green fluorescent protein (GFP) signal was transferred into 4 ll of lysis buffer of a thin-walled tube (Axygen, Tewksbury, MA, USA) with a micropipette (Eppendorf, Hamburg, Germany). The lysis buffer contained 50 mm dithiothreitol (DTT) and 10 units ll 1 RNAase inhibitor. The 4 ll of lysis buffer containing the pollen grain or tube was immediately frozen in liquid nitrogen. The content of each grain or tube was released by quick thawing before the reverse transcription-polymerase chain reaction (RT- PCR) experiment. To avoid RNAase contamination, all solutions were prepared with diethylpyrocarbonate (DEPC) water. DNase I (RNase-free; Promega) was used to rule out the presence of genomic DNA before reverse transcription. The first-strand cdna synthesis was performed using MLV reverse transcriptase (Promega) in a final volume of 10 ll; 1 ll of the whole 10-ll reaction from reverse transcription was used for PCR amplification. The quantitative PCR primer sets were LlFH1-QRT-F/ LlFH1-QRT-F for LlFH1 and LlEF1a-F/LlEF1a-R for LlEF1a (Supporting Information Table S1). Fluorescence recovery after photobleaching (FRAP) analysis LlFH1-GFP FRAP was performed on a confocal fluorescence microscope (LSM 510; Carl Zeiss) with a NA oil immersion lens by iterations of the 488-nm laser line at 100% emission strength in the region of interest. The fluorescence recovery was recorded with 20% of laser power using LSM 510 Release v.4.2 SP1 software (Carl Zeiss). For FM4-64 FRAP, a modified method described by Bove et al. (2008) was used. Membranes were fluorescently labeled by addition to the GM of FM4-64 (Molecular Probes, Invitrogen, Carlsbad, CA, USA) to a final concentration of 5 lm. After 15 min of labeling, pollen tubes were washed five times until no characteristic recovery of the overall membrane signal and resuspended in GM, mixed with an equal volume of 1.4% (w/v) low melting point agarose (type VII) in GM, and then FRAP was performed on a confocal fluorescence microscope (LSM 700; Carl Zeiss) with a oil immersion lens. Fluorescence in the region of interest was bleached by the 555-nm laser line at 100% emission strength for 100 iterations. Fluorescence recovery was recorded with 20% of laser power using ZEN 2009 Light Edition software (Carl Zeiss).

3 Phytologist Research 747 Fluorescence microscopy visualization of actin filaments and bundles Actin (1 lm, 20% Oregon green-labeled) alone or together with profilin (3 lm) was prepolymerized to actin filaments (F-actin) with 1 9 KMEI buffer (50 mm KCl, 1 mm MgCl 2, 1mM EGTA and 10 mm imidazole (ph 7.0)). To test the bundling activity of LlFH1 FH1FH2 and LlFH1 FH2, 1 lm of prepolymerized F-actin was incubated with different concentrations of LlFH1 proteins at room temperature for 30 min. All of the polymerized F-actin was diluted to 50 nm with fluorescence buffer containing 10 mm imidazole (ph 7.0), 50 mm KCl, 1 mm MgCl 2, 100 mm DTT, 15 mg ml 1 glucose, 20 mg ml 1 catalase, 100 mg ml 1 glucose oxidase and 0.5% methylcellulose. Actin filaments were observed by a TIRF 3 microscope (Carl Zeiss) equipped with a 9100/1.46 NA oil objective. Total internal reflection fluorescence microscopy (TIRFM) Microscope flow cells were prepared as described previously (Breitsprecher et al., 2012; Winkelman et al., 2014). TIRFM images were collected at s intervals with an ixon EMCCD camera (Andor Technology, Belfast, Northern Ireland, UK) using Laser TIRF 3 (Carl Zeiss). Mg-ATP-actin (15 50% Oregongreen actin, lm) was mixed with LlFH1 proteins (with or without profilin) and 2 9 TIRF buffer (20 mm imidazole (ph 7.4), 100 mm KCl, 0.4 mm ATP, 10 mm DTT, 2 mm MgCl 2, 2 mm EGTA, 30 mm glucose, 200 lgml 1 glucose oxidase, 40 lgml 1 catalase and 1% methylcellulose (4000 cp)) and transferred to a flow cell for imaging at room temperature. Methods S1 S11 can be found in Supporting Information. Results LlFH1 plays an important role in pollen tube tip growth We isolated a formin homolog protein from lily (L. longiflorum) pollen, named LlFH1 (GenBank accession no. KY221827). LlFH1 had two typical carboxyl-terminal domains of formins: the proline-rich formin homology-1 (FH1; amino acids ) domain, and the conserved formin homology-2 (FH2; amino acids ) domain (Fig. 1a). In addition to the FH1FH2 unit, LlFH1 contained two typical amino-terminal domains of class I formins: signal peptide (SP) and transmembrane (TM) domain (Fig. 1a). The phylogenetic analysis showed that LlFH1 is a member of branch c in class I (Fig. 1b). Most members in branch c are abundant or specific in pollen (Cvrckova et al., 2004; Ye et al., 2009; Cheung et al., 2010). RT-PCR analysis indicated that LlFH1 transcripts are specifically expressed in pollen grains and tubes (Fig. 1c). Quantitative RT-PCR analysis revealed that LlFH1 expression is decreased during pollen germination and tube elongation (Fig. 1d). This result implies that LlFH1 mrna is likely to be presynthesized and stored in matured pollen grains, but utilized during pollen germination and tube elongation, which is common for mrnas in male gametophyte development (McCormick, 2004). These data Fig. 1 Cloning and characterization of LlFH1. Schematic representation of domain organization of LlFH1 protein. The full-length LlFH1 contains 896 amino acids (aa). The amino-terminus of LlFH1 has a putative signal peptide (SP) followed by a predicted transmembrane (TM) domain. In the carboxylterminus, the LlFH1 protein features the two typical domains of formins: the formin homology-1 (FH1) domain, which is rich in proline, and the formin homology-2 (FH2) domain. LlFH1 (in box) is a plant formin in branch c of class I. The unrooted neighbor-joining phylogenetic tree was constructed using alignments of FH2 domains from Arabidopsis thaliana (At) and Oryza sativa (Os). Mouse mdia1 and yeast Bni1 were used as the outgroup sequences. (Bootstrap = 1000.) (c, d) Spatial and temporal expression analyses of LlFH1 transcripts in different lily (Lilium longiflorum) organs (c) and germination stages of pollen (d). Error bars represent SE, n 3.

4 748 Research Phytologist suggest that LlFH1 is a pollen-specific class I formin, and potentially plays an important role during pollen germination and growth. To study the function of LlFH1, we performed transient overexpression of LlFH1 in lily pollen. GFP was co-transformed to help identify the overexpressed pollen tubes. Within the first 3 h, the elongation of pollen tubes overexpressing LlFH1 was not significantly different from that of untreated and GFP- pollen tubes (Fig. 2a1,a2,b). However, overexpression of LlFH1 significantly retarded pollen tube elongation after 5 h of incubation (Fig. 2a3, a5,b) and caused tip swelling after 8 h (Fig. 2a4,a6,c). Meanwhile, we employed RNA interference (RNAi) to down-regulate LlFH1 expression. The empty hairpin vector hprnai was used as a control. The germination rate of pollen (Fig. 2f) and the average length of pollen tubes (Fig. 2d,g) were significantly reduced when the expression of LlFH1 was decreased, which was dependent on the level of LlFH1 suppression (Fig. 2e g). Taken together, the results suggest that LlFH1 is a key regulator of lily pollen tube tip growth. LlFH1 is targeted to the apical vesicle and PM In order to investigate how LlFH1 regulates tip growth, we examined the distribution and dynamics of LlFH1 by expressing LlFH1-GFP in pollen tubes. Compared with the control GFP signals in the elongating pollen tube (Fig. 3a), LlFH1- GFP signals were found to be specifically localized at the PM and the apical inverted-cone region, where densely packed transport vesicles almost exclusively occur (Fig. 3b; Movie S1; Parton et al., 2001). This distribution of LlFH1-GFP was similar to the distinct apical staining pattern of a membrane marker, FM4-64, in the pollen tube (Fig. 3c; Parton et al., 2001). When the pollen tube expressing LlFH1-GFP was stained with FM4-64, the two signals overlapped (Fig. 3d). Even in the plasmolyzed pollen tube, the signal recoveries of LlFH1-GFP and FM4-64 were synchronous (Fig. S1). In timelapse confocal images, the tip-localized LlFH1-GFP signals remained concentrated at the elongating tip region (Fig. 3e,f), and the position of the LlFH1-GFP signals was associated with (d) (e) (f) (c) (g) Fig. 2 LlFH1 plays an important role in pollen tube tip growth. (a c) Overexpression of LlFH1 arrests pollen tube growth and causes pollen tube swelling in lily (Lilium longiflorum). Pollen tubes transformed with free green fluorescent protein (GFP) (a1) and LlFH1 (a2) after 3 h of incubation in vitro. Pollen tubes transformed with free GFP (a3) and LlFH1 (a5) after 5 h of incubation. Pollen tubes transformed with free GFP (a6) and LlFH1 (a4) after 8 h of incubation. Bars: (a1 a3, a5) 100 lm; (a4, a6) 50 lm. Overexpression of LlFH1 has no significant effect on pollen tube elongation after 3 h of germination (n 200, P = 0.78), but impairs pollen tube growth after 5 h of incubation (n 200, P < 0.01) compared with the untreated (UT) or GFPtransformed pollen tubes. The columns represent the lengths of pollen tubes. (c) Overexpression of LlFH1 causes pollen tube tip swelling when cultured for 8h(n 200, P < 0.01). The columns represent the percentage of swollen tips (left) and the tip width of pollen tubes (right). Error bars indicate SE. (d g) Downregulation of LlFH1 expression inhibits pollen tube growth and germination rate. (d) Pollen tubes transformed with the empty hairpin vector control hprnai after 4 h (d1) or 6 h (d2) of incubation in vitro. Pollen tubes transformed with 5 lg LlFH1-RNAi after 4 h (d3) or 6 h (d4) of incubation. Pollen tubes transformed with 10 lg LlFH1-RNAi after 4 h (d5) or 6 h (d6) of incubation. Bar, 100 lm in (d1 d6). (e) Quantitative reverse transcription-polymerase chain reaction (RT-PCR) of single pollen demonstrates the down-regulation of LlFH1 expression in LlFH1-RNAi-transformed pollen tubes (n = 10, P < 0.01) compared with the untreated (UT) and hprnai-transformed pollen tubes. LlEF1a was used as an internal control. (f) Down-regulation of LlFH1 inhibits pollen germination (n 100, P < 0.01). (g) Down-regulation of LlFH1 reduces tube length (n 100, P < 0.01). Error bars indicate SE.

5 Phytologist Research 749 the direction of pollen tube growth (Fig. 3g). These results suggest that LlFH1 is localized in the apical vesicle and PM in lily pollen tubes. To further investigate which domain targets LlFH1 to the apical vesicle and PM, we constructed variable truncated protein of LlFH1 (Fig. S2a). As shown in Fig. S2, the pollen tube expressing Nter-GFP showed strong signals in the apical inverted cone and PM, which was identical with the full-length LlFH1 localization. The other mutant construct including DNter-GFP, FH1FH2-GFP and FH2-GFP showed an even cytosolic distribution throughout transformed pollen tubes. These data demonstrate that the N-terminal domain is essential for the localization of LlFH1. LlFH1-labeled exocytic vesicles predominantly occur at the shoulder of the apex in lily pollen tubes To further characterize the dynamics of LlFH1 in growing pollen tubes, we photobleached LlFH1-GFP signals in the apical inverted-cone region and a section of lateral PM, respectively. Then, we monitored the recovery process of the fluorescence signal in the bleached region. Following the bleaching of the inverted-cone region, the fluorescence signal began to reappear at the PM of the shoulder of the apex within 3 s (Fig. 4a, arrowheads) and reached c. 90% of the original intensity in 120 s (Fig. 4a,c; Movie S2). The result indicates that the shoulder of the apex is the predominant exocytic site in lily pollen tubes, as proposed previously (Bove et al., 2008). However, when the lateral membranes were bleached, the fluorescent signal only reverted to 20% of the original level in 180 s (Fig. 4b,c; Movie S3). We also found that the fluorescent signal gradually recovered from both sides towards the middle of the lateral membranes (Movie S3), indicating that the much slower recovery of LlFH1-GFP at the lateral membranes results from lateral diffusion of PM-localized LlFH1-GFP, rather than from diffusion of cytosolic LlFH1-GFP. These results demonstrate that the vesicle-localized LlFH1 is targeted to the vesicle membrane and then fuses into the PM via the exocytic pathway. Furthermore, we used receptor-like kinase (RLK) -GFP, which was found to target to the PM via exocytosis in tobacco pollen tubes, to monitor the exocytic activity in lily pollen tubes (Lee et al., 2008). By contrast with the extreme tip PM localization in the tobacco pollen tube (Lee et al., 2008), RLK-GFP was targeted to the shoulder of the apex of the lily pollen tube (Fig. 4d, arrowheads; Movie S4), strongly supporting the notion that the shoulder of the apex is the predominant exocytic site in lily pollen tube growth. To analyze the relationships between the dynamics of vesiclelocalized LlFH1 and the cytoskeleton systems, we employed a microfilament-specific inhibitor, latrunculin B (LatB), and a microtubule-specific inhibitor, oryzalin, respectively. The drug treatment assay indicated that the distribution and dynamics of LlFH1 in the inverted cone were not dependent on the microtubules (Fig. 4f), but on the actin cytoskeleton (Fig. 4e). After treatment with LatB, the residual LlFH1-GFP signals were gradually retained at the shoulder of the apex (Fig. 4e, arrowheads), and finally just retained evenly at the PM with growth cessation, further supporting the hypothesis that the shoulder is the major site at which exocytosis occurs. LlFH1 is required for actin fringe formation Considering that the distribution of LlFH1 correlates with the actin cytoskeleton, we examined the effect of LlFH1 on the actin cytoskeleton in the pollen tube tip. In control pollen tubes, a clear actin fringe structure, consisting of a palisade of short parallel cables, encircled the tube s subapex (Fig. 5a,d; Movie S5). However, in the pollen tube overexpressing LlFH1, an invertedcone-shaped structure filled with numerous short actin cables (c) (d) (e) (f) (g) Fig. 3 Distribution and dynamics of LlFH1 in elongating pollen tubes. (a d) Co-localization of LlFH1-GFP with FM4-64 in lily (Lilium longiflorum) pollen tubes transformed or stained with free green fluorescent protein (GFP), LlFH1-GFP, free FM4-64 dye (c) and LlFH1-GFP, with subsequent FM4-64 dye uptake (d). (e, f) Time-lapse confocal images (e) of a growing lily pollen tube expressing LlFH1-GFP and quantitative analysis (f) of the signal intensity of tip-localized LlFH1-GFP (circle in e) show that the signals remain concentrated at the tip during pollen tube growth. (g) Time-lapse confocal images show the relationship between the apical position of LlFH1-GFP and the pollen tube growth direction. Lines indicate the initial direction of the pollen tube, and arrows indicate the changing direction of the pollen tube. Bar, 10 lm. All confocal images show the medial section.

6 750 Research Phytologist (c) (d) (e) (f) Fig. 4 Exocytosis predominantly occurs at the shoulder of the apex in lily pollen tubes. (a c) Fluorescence recovery after photobleaching (FRAP) analysis of LlFH1-GFP in growing pollen tubes of lily (Lilium longiflorum). Photobleaching was performed and recovery was analyzed in the inverted cone region and lateral plasma membrane (PM; b) of the pollen tube. The numbers in each image indicate the time points after photobleaching. The bleached areas are marked by the boxes. Fluorescence recovery (c) was measured by calculating the mean green fluorescent protein (GFP) signal intensity in the inverted cone and lateral PM. (d) Localization of receptor-like kinase (RLK)-GFP in a lily pollen tube. (e, f) The effect of drug treatment on the localization of LlFH1-GFP. Transformed pollen tubes expressing LlFH1-GFP were treated with 2.5 nm latrunculin B (LatB) (e) or 10 nm oryzalin (f) for the indicated times. Bar, 10 lm. Error bars indicate SE, n = 3. All confocal images show the medial section. appeared at the tip region (Fig. 5b,d; Movie S5). The excessive short actin cables were only confined to the subapex, but did not enter the apex dome (Fig. 5b, arrowheads). Contrary to the actin configuration in the LlFH1-overexpressing pollen tube, the actin fringe structure became fuzzy or disrupted in the tip of the LlFH1-RNAi pollen tube (Fig. 5c,d; Movie S5). A similar actin fringe was also observed in live pollen tubes expressing Lifeact- GFP, an actin probe that does not appear to affect the dynamics of actin cables (Fig. S3; Vidali et al., 2009a). We also employed GFP-fABD2, a probe used to clearly visualize thick longitudinal actin cables (Ye et al., 2009), to label the actin cytoskeleton in the shank of pollen tubes. The results indicated that the longitudinal actin cables showed no significant differences in either LlFH1- overexpressing or LlFH1-RNAi pollen tubes compared with those in control pollen tubes (Fig. S4). To dissect the effect of each LlFH1 domain on the actin fringe, we constructed a series of domain truncations of LlFH1 (Fig. S5a). Among the truncations examined, FH1FH2 induced the excessive assembly of actin cables in the subapex, similar to that of full-length LlFH1 overexpression (Fig. S5b). Pollen tubes overexpressing FH2 (Fig. S5c) or DFH2 (Fig. S5d) exhibited the normal actin fringe structure in the subapex, similar to that of the control. DFH1 caused a fuzzy or disrupted subapical actin fringe, similar to that induced by LlFH1-RNAi (Fig. S5e), suggesting that DFH1 is a dominant-negative mutant. These results indicate that both FH1 and FH2 domains are required for LlFH1 activity. Taken together, our results suggest that LlFH1 plays an indispensable role in the formation and maintenance of the actin fringe in the subapex. Actin fringe and exocytic vesicles are spatially associated with each other The apical vesicle localization of LlFH1 and the excessive assembly of subapical actin cables induced by LlFH1 suggest that LlFH1 plays a bridging role between the structure of the subapical actin filaments and the tip vesicle system in lily pollen tubes. To better understand the correlation between the two systems, we first performed FM4-64 FRAP analysis on Lifeact-GFP-expressed pollen tubes. To specifically observe the exocytic vesicles, the excess FM4-64 dye was removed from the medium before FRAP analysis (Bove et al., 2008). The result showed that FM4-64 signal was preferentially recovered at the leading edge of the actin fringe (Fig. 6a, arrowheads; Movie S6), suggesting that the actin fringe was spatially associated with the exocytic vesicles. Then, we co-expressed LlFH1-GFP and Lifeact-mRFP in the pollen tube to observe vesicle organization and actin arrangement simultaneously. In the co-transformed pollen tubes, both of the signals formed similar inverted cones, but actin cables displayed by LifeactmRFP were behind the apical vesicles labeled by LlFH1-GFP (Fig. 6b; Movie S7), further indicating that the actin network was spatially associated with and likely to be assembled from the vesicles.

7 Phytologist Research 751 LlFH1 mediates the interaction between actin polymerization and vesicle trafficking in the tip of pollen tubes To analyze the interaction of tip vesicle trafficking with subapical actin organization during pollen tube polarity morphogenesis (c) (d) Fig. 5 LlFH1 is required for actin fringe formation. The control pollen tube of lily (Lilium longiflorum) shows the fine actin fringe labeled with rhodamine phalloidin at the subapex. Pollen tube overexpressing LlFH1 generates excessive actin cables at the subapex. Arrowheads mark the apex of the pollen tube. (c) Down-regulation of LlFH1 expression induces a fuzzy or disrupted actin fringe at the subapex of the LlFH1-RNAi pollen tube. (d) Quantitative analysis of (a c). The areas within 20 lm from the tip (circle) were measured by calculating the mean fluorescence intensity (n 20, P < 0.01). Bars, 10 lm. Error bars indicate SE. and tip growth, we treated pollen tubes co-transformed with LlFH1-GFP and Lifeact-mRFP using a final concentration of 2.5 nm LatB for min according to Cheung et al. (2010), which can just disrupt the subapical actin filament and vesicle organization (Fig. 7a), and then lead to growth inhibition and tip swelling (Cardenas et al., 2005; Hormanseder et al., 2005). The pollen tube growth gradually recovered as LatB was washed out. Early during recovery (Fig. 7d, left), vesicles (Fig. 7a, arrows) reappeared earlier than nascent actin filaments (Fig. 7a, arrowheads; Movie S8). Then, the actin filaments and the vesicles showed similar localization (Fig. 7a). Furthermore, they were in highly dynamic transformation between disassembly and reassembly (Movie S8). These results imply the possibility that LlFH1 initiates actin filament assembly from the vesicles. During the re-establishment of pollen tube polarity (Fig. 7d, middle), both vesicles (Fig. 7b, arrows) and actin filaments (Fig. 7b, arrowheads) reappeared at the same location near the regrowth site, and accumulated increasingly to form an apical vesicle dome, followed by actin networks with the recovery of growth (Movie S9). As the pollen tube grew (Fig. 7d, right), vesicles began to expand to the subapical region from the dome region, and formed an obvious accumulation area at the shoulder of the apex (Fig. 7c, arrows). Meanwhile, the actin filaments withdrew to the subapical region from the position closer to the apex (Fig. 7c, arrowheads). Finally, the vesicles and actin filaments recovered to form an inverted-cone shape. These results suggest that LlFH1-mediated synchronous coordination of vesicles and actin filaments is essential for the establishment of polarity and the growth of pollen tubes. LlFH1 nucleates actin polymerization To investigate the effects of LlFH1 on actin dynamics, we generated two recombinant proteins containing the FH1FH2 Fig. 6 Actin filaments and vesicles are spatially associated with each other. The recovery of FM4-64 (vesicle) after fluorescence recovery after photobleaching (FRAP) in the growing lily (Lilium longiflorum) pollen tube transformed with Lifeact-GFP (actin). The FM4-64 signals are preferentially recovered at the leading edge (arrowhead) of the actin fringe after FRAP. The bleached areas are marked by the box. A live pollen tube expressing LlFH1-GFP (vesicle) and Lifeact-mRFP (actin). Bars, 10 lm. DIC, differential interference contrast.

8 752 Research Phytologist domain (LlFH1 FH1FH2) and the FH2 domain (LlFH1 FH2) of LlFH1, respectively (Fig. S6a,b). Purified fusion proteins were used to determine whether they affect actin polymerization in vitro. We first examined the effects of LlFH1 FH1FH2 and LlFH1 FH2 on actin nucleation by the pyrene actin assay. As shown in Fig. 8, LlFH1 FH1FH2 eliminates the nucleation step of actin assembly in a dose-dependent manner, corresponding to active nucleation. However, LlFH1 FH2 has barely detectable nucleation activity (data not shown), similar to the feature of AtFH3, AtFH5 and AtFH19 (Ingouff et al., 2005; Ye et al., 2009; Zheng et al., 2012). These results indicate that the FH1 domain is necessary for actin filament nucleation. The concentration of new barbed ends is strongly dependent on the presence of LlFH1 FH1FH2 (Fig. 8b). Therefore, we employed TIRFM to visualize actin polymerization directly, and found that the actin filaments in the field of view become shorter and more abundant in the presence of LlFH1 FH1FH2 (Fig. 8c,d), suggesting that LlFH1 promotes actin nucleation and functions as a negative regulator of elongation. By contrast, the presence of LlFH1 FH2 did not increase or shorten actin filaments (Fig. S7a,f). (c) (d) Fig. 7 LlFH1 mediates the interaction between actin polymerization and vesicle trafficking at the tip of pollen tubes. Reassembly of vesicles and actin filaments on washout of latrunculin B (LatB) in a live lily (Lilium longiflorum) pollen tube expressing LlFH1-GFP (vesicle) and Lifeact-mRFP (actin). The dynamics of vesicles and actin filaments during polarity re-establishment of the pollen tube on washout of LatB. (c) The dynamics of vesicles and actin filaments during the later growth recovery on washout of LatB. Arrows in (a c) mark the vesicles and arrowheads mark the actin filaments. DIC, differential interference contrast. (d) The different growth rates of pollen tubes at the indicated points from, and (c) show the recovery progress on washout of LatB. Bars, 10 lm.

9 Phytologist Research 753 Previous studies have shown that the majority of actin monomers are bound by a high concentration of profilin in plant cells (Wang et al., 2005; Chaudhry et al., 2007). We thus decided to test whether LlFH1 nucleates actin assembly from the actin/ profilin complex. As shown in Fig. 8(e f), LlFH1 FH1FH2 exhibits an efficient nucleation activity using actin bound to Lilium profilin 1 (LlPRF1). Similar results were obtained when we replaced LlPRF1 with LlPRF2 or LlPRF3 (Fig. S6c,d). These data indicate that LlFH1 can effectively nucleate actin polymerization from the actin/profilin complex. LlFH1 attaches to the filament barbed end and inhibits elongation A general feature of formins is their binding to the barbed end of actin filaments. To ascertain barbed-end dynamics in the presence of LlFH1, a seeded actin filament elongation assay was performed. We observed that LlFH1 FH1FH2 inhibited actin polymerization rates in a dose-dependent manner, indicating that it binds to and caps the barbed end of actin filaments (Fig. 9a,b). To further confirm the capping activity of LlFH1, we performed (c) (d) (e) (f) Fig. 8 LlFH1 FH1FH2 nucleates actin polymerization from monomeric actin or actin/profilin complex. Time course of actin polymerization in the presence of LlFH1 FH1FH2 monitored by pyrene fluorescence. Different concentrations of LlFH1 FH1FH2 were added to 2 lm actin (10% pyrene labeled) before initiation of polymerization. au, arbitrary units. Nucleation efficiency of LlFH1 FH1FH2. (c, d) Micrographs of actin filaments in the presence of LlFH1 FH1FH2. Actin alone (c), actin and 0.5 lm LlFH1 FH1FH2 (d). Bar, 2 lm. (e) Time course of actin polymerization in the presence of LlFH1 FH1FH2 and lily (Lilium longiflorum) profilin1 (LlPRF1) monitored by pyrene fluorescence. From bottom to top: actin + profilin; actin + profilin lm LlFH1 FH1FH2; actin alone; and actin lm LlFH1 FH1FH2. (f) Time course of actin polymerization in the presence of LlFH1 FH1FH2 and LlPRF1 monitored by pyrene fluorescence. From bottom to top: actin + profilin; actin + profilin lm LlFH1 FH1FH2; actin + profilin lm LlFH1 FH1FH2; and actin + profilin lm LlFH1 FH1FH2.

10 754 Research Phytologist dilution-mediated actin filament depolymerization assays. As shown in Fig. 9(c), the initial rate of depolymerization decreased with increasing LlFH1 FH1FH2 concentration, in agreement with a barbed-end capping activity. These results indicate that LlFH1 FH1FH2 could bind to the barbed end of the actin filament and prevent actin elongation and depolymerization at the barbed end. To directly observe the effect of LlFH1 on the barbed end of individual actin filaments, we used a TIRFM assay to visualize actin elongation. Actin elongation could be tested in the presence of actin alone, or with additional LlFH1. We observed that the elongation rate of the actin filament was reduced in the presence of LlFH1 FH1FH2 compared with actin alone (Fig. S7a e), and that LlFH1 FH1FH2 slowed down the elongation rate in a dosedependent manner (Fig. 9d). By contrast, the presence of LlFH1 FH2 did not inhibit the elongation of actin filaments (Figs 9d, S7f). Plant profilin accelerates the rate of LlFH1 FH1FH2- mediated elongation A number of studies have demonstrated that profilin increases the rate of actin filament elongation associated with formin FH1FH2 domains (Kovar et al., 2006; Vidali et al., 2009b; Zhang et al., 2016). We cloned three profilin isoforms from the mature pollen of Lilium, which have 76 88% identity. LlPRF1, LlPRF2, LlPRF3 and human profilin (HPRF) were chosen to investigate the effect of the interaction of profilin and LlFH1 on the regulation of the elongation rate of actin filaments. Using TIRFM, we determined the rate of elongation of actin filament bound to LlFH1 FH1FH2 in the presence of different lily profilin isoforms. The elongation rate is c. 0.6 subunits s 1 for actin in the presence of LlFH1 FH1FH2 (Fig. 10c,e; Movie S10). By contrast, when LlPRF1 was added to the system, (c) (d) Fig. 9 LlFH1 FH1FH2 inhibits elongation and depolymerization at the barbed ends of actin filaments. Kinetics of elongation of the barbed end of actin filaments in the presence of LlFH1 FH1FH2. Preformed actin filament seeds (0.8 lm) were incubated with various concentrations of LlFH1 FH1FH2 (from top to bottom: 0, 0.1, 0.2, 0.3, 0.4, 0.5 and 0.6 lm) before addition of 1.0 lm pyrene actin monomers. au, arbitrary units. Variation in the initial rate of elongation as a function of LlFH1 FH1FH2 concentration. (c) Kinetics of actin filament depolymerization in the presence of LlFH1 FH1FH2. LlFH1 FH1FH2 was incubated with 5 lm actin filaments (60% pyrene-labeled) for 5 min before dilution of the solution 25-fold in G-buffer. (d) Average elongation rates of filaments with or without LlFH1 FH1FH2 or LlFH1 FH2 on the barbed end at different concentrations (Error bars represent SE, n>30, N 3).

11 Phytologist Research 755 the elongation rate was c. 3.6 subunits s 1 for the actin/profilin complex in the presence of LlFH1 FH1FH2 (Fig. 10d,e; Movie S10), indicating that LlPRF1 increases the elongation rate of actin filaments in association with LlFH1 FH1FH2. For LlPRF2 and LlPRF3, the rates of LlFH1 FH1FH2-mediated elongation were c. 1.9 subunits s 1 and c. 2.0 subunits s 1, respectively (Fig. 10e; Movie S10). These data demonstrate that lily profilins increase the rate of elongation of actin filaments associated with LlFH1 FH1FH2 at different levels. We also measured the LlFH1 FH1FH2-mediated elongation rate with HPRF. As with AtFH14, the growth rate was hardly altered in the presence or absence of HPRF (Fig. 10e; Movie S10). Therefore, as the rate of formin-mediated actin elongation is affected by the source of profilin, this further supports the conclusion that the interaction of profilin with formin is isoform and species dependent (Zhang et al., 2016). LlFH1 induces bundling of actin filaments It has been demonstrated that several plant formins, such as AtFH1, AtFH8, AtFH14, AtFH16 and OsFH5, bundle actin filaments (Michelot et al., 2005; Li et al., 2010; Xue et al., 2011; Yang et al., 2011; Zhang et al., 2011, 2016; Wang J et al., 2013). To determine the binding activity of LlFH1 to actin filaments, we visualized actin filaments directly using fluorescence microscopy. The actin filaments were scattered individually in the absence of LlFH1 FH1FH2 (Fig. 11c). However, in the presence of LlFH1 FH1FH2, obvious actin bundles could be observed (Fig. 11d). By contrast, LlFH1 FH2 could not bundle actin filaments (Fig. S8e). To examine further the effect of LlFH1 FH1FH2 on the bundling of actin filaments, a low-speed cosedimentation assay was employed. This assay provided data in agreement with microscopic observations. Most of the actins (c) (d) Fig. 10 Effect of profilin on barbed-end actin assembly in the presence of LlFH1 FH1FH2. (a d) Total internal reflection fluorescence micrographs of actin (0.7 lm, 30% Oregon green-labeled) assembled in the presence of LlFH1 FH1FH2 and LlPRF1. Actin alone, actin lm LlPRF1, actin lm LlFH1 FH1FH2 (c), actin lm LlFH1 FH1FH lm LlPRF1 (d). Arrows indicate the elongating barbed end of the actin filament. Bar, 5 lm. (e) Average elongation rates of actin filaments in the presence of LlFH1 FH1FH2 and/or profilin (Error bars represent SE, n>30, N = 3). (e)

12 756 Research Phytologist stayed in the supernatant in the absence of LlFH1 FH1FH2 (Fig. 11a,b). By contrast, the addition of LlFH1 FH1FH2 increased the amount of actin in the pellet in a dose-dependent manner (Fig. 11a,b). These data suggest that LlFH1 FH1FH2 can bundle actin filaments. As it is well known that the majority of actin monomer is buffered by profilin in plant cells (Wang et al., 2005; Chaudhry et al., 2007), the effects of LlFH1 on the assembly of the actin/ profilin complex were also examined. In the presence of profilin, actin could only polymerize into actin filaments (Figs 11e, S8a, c). When LlFH1 FH1FH2 was added to the polymerization system containing actin and profilin, actin filaments were organized into thick actin bundles (Figs 11f, S8b,d). However, actin bundles were not observed when LlFH1 FH2 was added (Fig. S8f). Discussion LlFH1 regulates microfilament organization in the lily pollen tube In this study, we have shown that LlFH1 FH1FH2 nucleates actin assembly from monomers. However, LlFH1 FH2 does not have actin nucleation activity, in contrast with AFH1, AtFH8, AtFH14 and OsFH5 (Michelot et al., 2005; Yi et al., 2005; Yang et al., 2011; Zhang et al., 2011, 2016), but is similar to AFH3, AtFH5 and AtFH19, whose nucleation activity requires the presence of the FH1 domain (Ingouff et al., 2005; Ye et al., 2009; Zheng et al., 2012). The FH1 domain of LlFH1 may contribute to the dimer formation of FH2, which has been shown to be critical for its nucleation (Xu et al., 2004). LlFH1 can also nucleate actin assembly from profilin-bound actin monomers. As the majority of actin is sequestered by profilin in pollen tubes (Wang et al., 2005; Chaudhry et al., 2007), our finding shows that LlFH1 could nucleate actin assembly from the actin/profilin complex, indicating that LlFH1 is a functional actin nucleator in vivo. In agreement with this hypothesis, overexpression of fulllength LlFH1 or the deleted mutant FH1FH2 induces excessive subapical actin cables, and down-regulation of LlFH1 eliminates the actin fringe. LlFH1 FH1FH2 can bind to the barbed ends of actin filaments and can decrease the elongation rate of actin filaments, indicating that LlFH1 has a capping activity. This property depends on the presence of the FH1 domain, which is similar to AFH3 and AtFH19 (Ye et al., 2009; Zheng et al., 2012). LlFH1 FH1FH2 can also bundle actin filaments in the absence or presence of plant profilins in vitro. Similar to AFH1 (Michelot et al., 2005), whose bundling activity requires the FH1 domain, we can distinguish LlFH1 from AtFH8, AtFH14, AtFH16 and OsFH5, whose FH2 domain alone is sufficient to bundle actin filaments (Li et al., 2010; Xue et al., 2011; Yang et al., 2011; Zhang et al., 2011; Wang J et al., 2013). The results of both the capping and (c) (d) (e) (f) Fig. 11 LlFH1 FH1FH2 bundles actin filaments in vitro. Low-speed co-sedimentation assays to determine the bundling activity of LlFH1 FH1FH2. Lane 1, actin alone; lanes 2 8, actin plus 0.25, 0.5, 1, 2, 4, 6 or 8 lm LlFH1 FH1FH2, respectively; lane 9, 8 lm LlFH1 FH1FH2 alone. Percentage of actin filaments recovered in the low-speed pellet as a function of the concentration of LlFH1 FH1FH2. Error bars represent SE, n = 3. (c f) Total internal reflection fluorescence micrographs of actin bundles mediated by LlFH1 FH1FH2 with or without profilin. Actin filaments alone (c), actin lm LlFH1 FH1FH2 (d), actin + LlPRF1 (e), actin + LlPRF lm LlFH1 FH1FH2 (f). Bar, 5 lm.

13 Phytologist Research 757 bundling activity of LlFH1 demonstrate that LlFH1 may function by stabilizing actin filaments. Dilution-mediated actin depolymerization assays confirm this speculation. Moreover, the downregulation of LlFH1 eliminates the actin fringe in pollen tubes, further suggesting that LlFH1 plays a role in the stabilization of actin filaments in vivo. Our work has demonstrated that LlFH1 FH1FH2 can increase the elongation rate of actin assembly from the barbed end in the presence of profilin, and that the FH1 domain is essential for the interaction between LlFH1 and actin/profilin. Moreover, this interaction plays a key role for LlFH1 to nucleate actin/profilin and regulate actin/profilin to the barbed end. Therefore, we conclude that profilin may be a regulator that promotes the assembly of the barbed end of LlFH1 nucleated filaments. In vivo, the overexpression of LlFH1 FH1FH2 induces the excessive assembly of actin cables in the subapex of the pollen tube. However, pollen tubes overexpressing LlFH1 FH2 exhibit the normal actin fringe in the subapex. Considering that the coordination of formin with profilin is mediated by the FH1 domain of formin and that the majority of actin exists as actin/profilin in pollen, we propose that profilin maintains the actin monomer pool, thereby favoring formin-mediated actin assembly to control the subapical actin structure in the pollen tube. Given that LlPRF1, LlPRF2 and LlPRF3 are expressed in the pollen/pollen tube, LlFH1 may work together with these profilins to form and regulate the actin structure in the subapical region of the lily pollen tube. LlFH1 serves as a linking protein between the actin fringe and exocytic vesicles in lily pollen As they are essential elements for pollen tube tip growth, actin filaments and vesicles are the focus of study. Most direct evidence on contact between the two systems is mainly derived from studies in the shank of the pollen tube. The longitudinally oriented actin cables provide the main tracks, and myosins drive the movement of Golgi-derived vesicles and other cellular components towards the growth site via cytoplasmic streaming (Smith & Oppenheimer, 2005; Cai & Cresti, 2009). In spite of no direct evidence, the results in the shank make it likely that the actin fringe at the subapex is the final track conveying vesicles specifically to the site of growth by an actin myosin mechanism. However, the short arranged actin fringe is more dynamic than, and thus necessarily distinct from, the longitudinal actin cables that guide cytoplasmic streaming (Vidali & Hepler, 2001). It is estimated that a 5 10-µm-wide fringe would completely turn over every s (Lovy-Wheeler et al., 2005). Meanwhile, the condition surrounding the actin fringe is more sensitive than that in the shank, such as the high Ca 2+ gradient and ion concentration (Kovar et al., 2000). It has been reported that plant myosins are activated by calmodulin binding and inactivated by Ca 2+ -induced calmodulin dissociation (Yokota et al., 1999; Vidali & Hepler, 2001). Thus, it is reasonable to presume that there exists another interaction between the actin fringe and vesicle transport in addition to the universal actin myosin link. As an important actin nucleation factor in plant cells, formin is a potential candidate for the acceleration of actin nucleation (Blanchoin & Staiger, 2010). Because of the presence of the SP and TM domain, the common membrane structure-associated localization of plant class I formins has been convincingly demonstrated by GFP fusion protein localization from Arabidopsis formins 1, 4, 5, 6 and 8 (Cheung & Wu, 2004; Favery et al., 2004; Deeks et al., 2005; Ingouff et al., 2005; Cheung et al., 2010; Martiniere et al., 2011). Among the formins mentioned above, AFH1 and AFH5 have been investigated in the pollen tube. AFH1 locates at the pollen tube PM and initiates actin assembly from the PM (Cheung & Wu, 2004). Apical-localized AFH5 mediates the assembly of the subapical actin structure, and provides actin filaments for vesicle trafficking in pollen tubes. The overexpression of AFH5 induces a basket-shaped subapical actin structure in the pollen tube (Cheung et al., 2010). Similar to AFH5, LlFH1 is targeted to the apical vesicles in lily pollen tubes, whereas the overexpression of LlFH1 induces excessive actin cables that form a solid inverted cone coinciding with LlFH1-localized vesicles. Pollen tube expressing LlFH1 FH1FH2 also induces excessive actin cables in the subapex, although it distributes diffusely in the pollen tube. We assume that this might be a result of certain conditions in the subapex, such as ph or Ca 2+ concentration, as it is known that the ph in the actin fringe area is alkaline and there is a tipfocused Ca 2+ gradient present in growing pollen tubes (for a review, see Cheung & Wu, 2008; Hepler, 2016). Similar to this result, the supernumerary actin cables are especially evident in the apex and subapex of pollen tube overexpressing [FH1 + FH2] of AFH1, although [FH1 + FH2]:GFP shows an even distribution throughout the transformed tube (Cheung & Wu, 2004). In addition, DFH1 of LlFH1 leads to a disrupted subapical actin fringe; we believe that this is probably because DFH1 competes with the native LlFH1. In our study, we also found that the exocytic vesicles preferentially accumulate at the leading edge of the actin fringe. Furthermore, the polymerization of nascent actin filaments follows the emergence of the apical vesicles during the recovery from LatB treatment. These results raise an interesting possibility that LlFH1-mediated actin polymerization generates a propelling force for exocytic vesicle movement towards the shoulder of the pollen tube tip. In support of this hypothesis, LlFH1 is able to nucleate actin assembly and to accelerate the elongation rate of actin filaments in the presence of pollen-expressed lily profilin in vitro. In addition, this hypothesis is consistent with the fact that several animal formins initiate the assembly of actin filaments from the surface of beads and thus push the beads forwards, simulating the movement of vesicles in vitro, indicating that direct actin assembly-promoted vesicle movement is feasible in addition to along the actin filament track (Kozlov & Bershadsky, 2004; Romero et al., 2004;Barkoet al., 2010). Thus, LlFH1 may be an intermediate protein that links the actin fringe and exocytic vesicles, which is important for the delivery of exocytic vesicles to the shoulder of the apex during the growth of lily pollen tubes. There may exist two different exocytic patterns in tipgrowing pollen tubes Pollen tube elongation is the result of an equilibrium between exocytosis and endocytosis. To understand the molecular