A GFP-mouse talin fusion protein labels plant actin filamentsin vivoand visualizes the actin cytoskeleton in growing pollen tubes


*For correspondence (fax +1 212 3278327; e-mail chua@rockvax.rockefeller.edu).


The C-terminus of mouse talin (amino acids 2345–2541) is responsible for all of the protein’s f-actin binding capacity. Unlike full-length talin, the C-terminal f-actin binding domain is unable to nucleate actin polymerization. We have found that transient and stable expression of the talin actin-binding domain fused to the C-terminus of the green fluorescent protein (GFP-mTn) can visualize the actin cytoskeleton in different types of living plant cells without affecting cell morphology or function. Transiently expressed GFP-mTn co-localized with rhodamine-phalloidin in permeabilized tobacco BY-2 suspension cells, showing that the fusion protein can specifically label the plant actin cytoskeleton. Constitutive expression of GFP-mTn in transgenicArabidopsis thalianaplants visualized actin filaments in all examined tissues with no apparent effects on plant morphology or development at any stage during the life cycle. This demonstrates that in a number of different cell types GFP-mTn can serve as a non-invasive marker for the actin cytoskeleton. Confocal imaging of GFP-mTn labeled actin filaments was employed to reveal novel information on thein vivoorganization of the actin cytoskeleton in transiently transformed, normally elongating tobacco pollen tubes.


The actin cytoskeleton is a very dynamic structure which is involved in a variety of cellular activities. In plants, actin filaments are presumed to play essential roles in many important processes including cell division, cell elongation, establishment of cytoplasmic organization, cytoplasmic streaming, pathogen response, tropisms, and pollen tube growth. However, the precise function of the actin cytoskeleton in these processes has not been well characterized to date. Observation of the dynamic behavior of the actin cytoskeleton in living plant cells is expected to advance considerably our understanding of actin function in vivo. Microinjection of fluorescently labeled phalloidin has provided important insights into in vivo actin dynamics in some specific types of plant cells (Cleary 1995;Kim et al. 1995;Miller et al. 1996;Schmit & Lambert 1990;Zhang et al. 1993). Since this technique is invasive and only applicable to certain cells, we have set out to construct a GFP fusion protein that allows non-invasive visualization of the actin cytoskeleton in any living plant cell.

GFP fusion proteins have previously been used to observe actin localization in cells of Dictyostelium discoideum (Aizawa et al. 1997;Neujahr et al. 1997;Pang et al. 1998;Westphal et al. 1997), yeast (Doyle & Botstein 1996), Drosophila melanogaster (Edwards et al. 1997), and mammalian organisms (Ballestrem et al. 1998;Fischer et al. 1998;Ludin & Matus 1998). GFP directly fused to actin was used for most of these studies. Expression of GFP-actin fusion proteins potentially labels both filamentous actin (factin) as well as pools of monomeric actin (g-actin). Labeling of only a subpopulation of filamentous actin structures by GFP-actin has been reported (Doyle & Botstein 1996). Expression of GFP-actin can severely affect morphology and function of D. discoideum cells (Aizawa et al. 1997). Similarly, we have found that GFP-actin expression labels filamentous structures in tobacco BY-2 cells, but completely inhibits tobacco pollen tube growth (P. Spielhofer, B. Kost and N.-H. Chua, unpublished observations).

Experiments with GFP fused to full-length or truncated actin binding proteins have also been described. In yeast, GFP fused to two different actin-binding proteins, Abp1p and Sac6p, was found to bind to the same structures that were visualized by GFP-actin (Doyle & Botstein 1996). Inducible expression of GFP fused to a C-terminal fragment of moesin was employed to observe accumulation of filamentous actin structures in different tissues of D. melanogaster embryos (Edwards et al. 1997). Sites of f-actin accumulation were also visualized in D. discoideum cells using GFP fused to the actin binding domain of ABP-120 (Pang et al. 1998).

The C-terminus of mouse talin (197 amino acids) contains a number of highly conserved motifs that are found in the f-actin binding domain of a diverse group of actin binding proteins. The talin f-actin binding domain binds filamentous actin with a higher affinity than the full-length protein but, in contrast to the latter, does not nucleate actin polymerization (McCann & Craig 1997). The domain therefore appears to be an ideal tag for directing attached GFP to actin filaments. In this report, we show that expression of GFP fused to the mouse talin f-actin binding domain (GFP-mTn) can be used to non-invasively visualize the entire filamentous actin network in a number of different types of living plant cells, including growing pollen tubes.

The current understanding of the mechanism of pollen tube tip growth is limited (Cai et al. 1997;Taylor & Hepler 1997). It is clear, however, that this process is strictly dependent on a functional actin cytoskeleton. Drugs that interfere with the equilibrium between f- and g-actin, such as cytochalasins or latrunculins, inhibit pollen tube growth at very low concentrations (Cai et al. 1997;Taylor & Hepler 1997; B. Kost, P. Spielhofer and N.-H. Chua, unpublished observations).

The structure and function of the pollen tube actin cytoskeleton is still a matter for debate (Cai et al. 1997;Taylor & Hepler 1997). In a number of earlier studies, based on immunocytochemistry or on phalloidin labeling of chemically fixed or permeabilized pollen tubes, a dense filamentous actin network was observed at the very pollen tube tip (reviewed in Derksen et al. 1995). This tip-localized actin network was proposed to directly mediate pollen tube tip growth similar to the cortical actin cytoskeleton in cells showing amoeboid movement (Derksen et al. 1995;Steer & Steer 1989). Tip-localized actin filaments have also been suggested to guide secretory vesicles to apical membrane docking sites (Derksen et al. 1995). More recent experiments have shown that pollen tube ultrastructure is very sensitive to chemical fixation and permeabilization (Doris & Steer 1996;He & Wetzstein 1995). Physical fixation by rapid freeze-freeze substitution followed by electron microscopy as well as microinjection of rhodamine-phalloidin have been employed to re-examine the actin cytoskeleton in lily pollen tubes (Miller et al. 1996). In this study, no dense actin network was found at the pollen tube tip.

Our experiments confirm the absence of a tip-localized actin network and, in addition, provide novel, detailed information on the structure of the actin cytoskeleton in growing pollen tubes. We observed a helical arrangement of cortical actin filaments in the pollen tube tip cytoplasm and an actin ring around the organelle exclusion zone at the pollen tube apex. Functional implications of these observations are discussed.

Results and discussion

Visualization of the entire actin cytoskeleton in BY-2 cells by transient GFP-mTn expression

The C-terminus of a GFP version optimized for use as a marker in plant cells (see Experimental procedures) was fused in frame to the C-terminal 197 amino acids of mouse talin that constitute the protein’s f-actin binding domain (McCann & Craig 1997). Particle bombardment was employed to transiently express the resulting fusion protein (GFP-mTn) under the control of the CaMV 35S promoter in tobacco BY-2 suspension cells. Confocal imaging of GFP-fluorescence emitted from transformed cells revealed that GFP-mTn localized to a dense filamentous network (Fig. 1a,b), which very closely resembles the actin cytoskeleton in BY-2 cells (Hasezawa et al. 1989;Katsuta et al. 1990) and in other plant cells (Seagull et al. 1987), as visualized by rhodamine-phalloidin staining after permeabilization. GFP-mTn labeled filaments in living BY-2 cells were randomly arranged in the cytoplasm at the cell cortex and around the nucleus (Fig. 1a,b). They also followed trans-vacuolar cytoplasmic strands, which are known to be stabilized by actin filaments (Staiger et al. 1994). Weak, diffuse GFP-mTn fluorescence but no labeling of filaments was observed in the nucleus. GFP-mTn fluorescence was absent from vacuoles and nucleoli (Fig. 1b).

Figure 1.

Transient expression of GFP-mTn (a–c) and of untagged GFP (d–f) in living tobacco BY-2 suspension cells.

(a and d) Projections of serial confocal optical sections (a, 4× line averaging; d, 2× line averaging) showing the distribution of GFP fluorescence in entire cells.

(b and e) Single confocal optical sections through nuclei. Note the presence of bright filamentous structures in trans-vacuolar cytoplasmic strands and around the nucleus in (b).

(c and f) Transmitted light reference images. Size and shape of individual BY-2 suspension cells as well as number and size of nucleoli contained in their nuclei can vary considerably.

V, vacuoles; N, nuclei; n, nucleoli; nv, nucleolar vacuole; CS, cytoplasmic strands; P, gold particle that carried DNA into the cell. Scale bars: 100 μm.

Control BY-2 cells that expressed GFP without the mouse talin tag showed evenly distributed fluorescence throughout the cytoplasm (Fig. 1d,e). No fluorescence was found in vacuoles. GFP, which is small enough to freely diffuse through nuclear pores (Grebenok et al. 1997), slightly accumulated in the nucleus, as it does in other cell types (Haseloff et al. 1997). GFP fluorescence was absent from the nucleolar matrix but could be observed in nucleolar vacuoles (Fig. 1e). Under the conditions used to image transient expression of GFP-mTn or GFP, untransformed BY-2 cells did not emit detectable fluorescence (Fig. 1; data not shown).

To confirm that GFP-mTn binds to and visualizes actin filaments, BY-2 cells expressing the fusion protein were permeabilized and co-stained with rhodamine-phalloidin. Phalloidin is a fungal toxin that acts by stabilizing actin filaments to which it binds very specifically. Fluorescent conjugates of phalloidin are commonly use to visualize the actin cytoskeleton in permeabilized cells. At saturating concentrations, fluorescent phalloidin is presumed to visualize all actin filaments present in stained cells (Cooper 1987). Patterns of green GFP-fluorescence and red rhodamine-phalloidin-fluorescence emitted from double-labeled BY-2 cells were essentially identical (Fig. 2), showing that GFP-mTn binds to the entire BY-2 cell actin cytoskeleton in a specific manner. Adjacent cells that did not express GFP-mTn showed rhodamine-phalloidin labeling similar to double-labeled cells (Fig. 2b,c; data not shown), indicating that GFP-mTn expression does not alter f-actin organization.

Figure 2.

Co-localization of transiently expressed GFP-mTn and of rhodamine-phalloidin in a permeabilized BY-2 cell.

Artificially colored projections of serial confocal optical sections (4× line averaging) are shown.

(a) GFP-mTn fluorescence (excitation, 488 nm; emission, 525–565 nm [green]).

(b) Rhodamine-phalloidin fluorescence (excitation, 543 nm; emission, ≥ 570 nm [red]).

(c) Superimposition of (a) and (b). Yellow color indicates areas where the two images overlap and where the two markers co-localize.

(d) Transmitted light reference image. Scale bar: 25 μm.

Transient expression of GFP-mTn had no apparent effect on the morphology or viability of BY-2 cells. Cells appeared unaffected (Fig. 1c; data not shown) even after confocal imaging of GFP-mTn labeled actin filaments, which could potentially cause phototoxic effects (Hepler & Gunning 1998). GFP-mTn levels were high enough to allow visualization of the actin cytoskeleton in single cells as early as 3 h after particle bombardment. Several hours later, pairs of GFP-mTn expressing cells were regularly observed, which obviously originated from targeted cells that had divided (see cover photograph). This indicates that GFP-mTn does not inhibit mitosis or cytokinesis, both processes which are presumed to depend on f-actin function (Hepler & Gunning 1998;Lloyd 1989;Nagata et al. 1992).

Constitutive, non-invasive labeling of actin filaments by GFP-mTn in transgenic Arabidopsis thaliana plants

Transgenic A. thaliana plants that constitutively expressed GFP-mTn under the control of the 35S promoter were generated. GFP-mTn labeling of the actin cytoskeleton was observed in all examined tissues of etiolated transgenic T2 seedlings, including epidermal as well as cortical cell layers of roots, hypocotyls and cotyledons (Fig. 3). Control seedlings expressing untagged GFP from the 35S promoter showed evenly distributed cytoplasmic and nuclear fluorescence (data not shown). No fluorescence emission was detected from untransformed seedlings under the conditions used for GFP-imaging (data not shown).

Figure 3.

The actin cytoskeleton in cells of different tissues of etiolated transgenic T2 Arabidopsis thaliana seedlings as visualized by stable, constitutive expression of GFP-mTn and confocal microscopy.

Projections of serial confocal optical sections are shown.

(a) Epidermal cells of a cotyledon (8× line averaging). Scale bar: 100 μm.

(b) Stoma guard cells in the epidermal cell layer of a cotyledon (4× line averaging). Scale bar: 10 μm.

(c) Epidermal and cortical hypocotyl cells (8× line averaging). Scale bar: 100 μm.

(d) Epidermal and cortical root cells (4× line averaging). Scale bar: 50 μm.

At the T2 stage, transgenic lines have gone through each phase of the plant life cycle at least once. GFP-mTn expression did not have any apparent effects on plant morphology or development. These results demonstrate that GFP-mTn expression can visualize the actin cytoskeleton in different types of plant cells in a completely non-invasive manner. The transgenic GFP-mTn expressing A. thaliana lines we have established allow in vivo examination of the dynamic behavior and function of the actin cytoskeleton during a variety of actin-dependent processes in different tissues throughout plant development.

Non-invasive visualization of the tobacco pollen tube actin cytoskeleton by transient GFP-mTn expression

GFP-mTn was also employed to visualize the actin cytoskeleton in growing tobacco pollen tubes, which is difficult to observe using conventional techniques (Cai et al. 1997;Taylor & Hepler 1997). Tobacco pollen was plated on a solid culture medium and bombarded with a vector containing the GFP-mTn sequence fused to the LAT52 promoter which, unlike the 35S promoter, confers strong expression in pollen (Twell et al. 1989;Twell et al. 1990). The culture medium used had been carefully optimized to allow normal pollen tube growth in vitro (Read et al. 1993a, 1993b). On this medium, tobacco pollen tubes elongate for more than 48 h, grow to a final length of at least 15 mm, show normal tip morphology (Fig. 4 m) and cytoplasmic streaming, form callose plugs at regular intervals, and support mitotic division of the generative cell into two sperm cells (Read et al. 1993a, 1993b; B. Kost, P. Spielhofer and N.-H. Chua, unpublished observations). Labeling of actin filaments in growing pollen tubes derived from successfully bombarded pollen grains was observed after only 2 h and remained visible for at least 16 h. Very brightly fluorescent tubes that apparently expressed GFP-mTn at far above average levels contained thick actin cables, showed slow cytoplasmic streaming, and tended to cease growth prematurely. In contrast to BY-2 cells and A. thaliana plants, growing tobacco pollen tubes were apparently affected by very high levels of GFP-mTn expression. Pollen tube growth is not only extraordinarily sensitive to GFP-mTn expression, but also to actin depolymerizing drugs (Cai et al. 1997;Taylor & Hepler 1997). CytochalasinD and LatrunculinB completely inhibit tobacco pollen germination and tube growth at low concentrations (5 μm and 10 nm, respectively; B. Kost, P. Spielhofer and N.-H. Chua, unpublished observations). These findings suggest that pollen tube elongation is strictly dependent on actin function.

Figure 4.

The actin cytoskeleton in the tip of a growing tobacco pollen tube as visualized by transient expression of GFP-mTn and confocal microscopy.

This particular pollen tube was growing at an average rate (3.25 μm min–1) after confocal imaging.

(a–l) Serial confocal optical sections at a step size of 1 μm (8× line averaging) giving an overview over the entire pollen tube actin cytoskeleton. (a) Cell cortex facing the microscope lens. (l) Cell cortex facing the opposite direction.

(m) Transmitted light reference image.

VG, vegetative nucleus (generative cell is not visible); OEZ, organelle exclusion zone; arrow heads, actin ring around the organelle exclusion zone. Scale bar: 25 μm.

However, tip morphology (Fig. 4m) and cytoplasmic streaming (data not shown) of more moderately fluorescent tubes, such as the one shown in Fig. 4, were absolutely normal, even after imaging the actin cytoskeleton by serial confocal optical sectioning. The growth rate after confocal imaging of a number of such pollen tubes, which all displayed actin labeling similar to the one shown in Fig. 4, was determined. These pollen tubes elongated at the same average rate as untransformed pollen tubes (Fig. 5), which demonstrates that transient expression of GFP-mTn and confocal imaging also allows non-invasive visualization of the actin cytoskeleton in growing pollen tubes. Untagged GFP transiently expressed under the control of the LAT52 promoter was evenly distributed in the pollen tube cytoplasm, accumulated slightly in the nucleus and was excluded from the generative cell as well as from vacuoles (data not shown). Pollen tubes that did not express GFP were non-fluorescent (data not shown).

Figure 5.

GFP-mTn expression and confocal imaging have no effect on tobacco pollen tube growth.

GFP-mTn, average growth rate of GFP-mTn expressing pollen tubes after confocal imaging. The growth rate of 21 tubes with actin labeling similar to the one shown in Fig. 4 was measured. Control, average growth rate of 21 untransformed pollen tubes in different areas of the same plates. Error bars: 95% confidence interval.

Confocal imaging of GFP-mTn fluorescence resulted in clear images of the in vivo organization of the tobacco pollen tube actin cytoskeleton (Fig. 4), which revealed a number of details that were not observed in rhodamine-phalloidin microinjected lily pollen tubes (Miller et al. 1996). Serial confocal optical sections at a step size of 1 μm through a representative GFP-mTn expressing tobacco pollen tube are shown in Fig. 4. Long and relatively thick actin bundles were abundant in the pollen tube cytoplasm but did not extend into a 20 μm long region at the very tube tip. In the cell cortex, parallel actin bundles were arranged in a helical pattern (Fig. 4a,b,k,l), whereas actin cables in central regions were generally straight and longitudinally oriented (Fig. 4d–h). Pollen tube actin bundles appeared to be relatively immobile. During the time it took to complete a serial confocal scan (about 90 sec), actin filaments did not detectably change position. Some actin filaments were found to be associated with the vegetative nucleus (Fig. 4e–h) and the generative cell (data not shown). However, we did not detect a dense F-actin network around these structures as was occasionally observed following antibody or phalloidin staining of actin in permeabilized pollen tubes. F-actin networks have been proposed to be involved in maintaining shape as well as positioning in the tube cytoplasm of vegetative nuclei and generative cells (Derksen et al. 1995;Pierson & Cresti 1992).

Actin labeling in permeabilized cells has also provided evidence suggesting the presence of a dense actin network at the very pollen tube tip (Derksen et al. 1995). We did not find any indication of such a network, which is in accordance with what was observed in rhodaminephalloidin microinjected, living lily pollen tubes (Miller et al. 1996). Rather, imaging of GFP-mTn fluorescence revealed that the very pollen tube tip (a 20 μm long region at the end of the pollen tube) contains only sparse and fine actin filaments (Fig. 4c–h). No fluorescence was detected in the organelle exclusion zone at the extreme tube apex (Fig. 4d), which is thought to exclusively contain secretory vesicles (Derksen et al. 1995;Taylor & Hepler 1997), indicating that this region is devoid of actin bundles. This does not exclude, however, that individual, very fine actin filaments are occasionally present in the organelle exclusion zone, as was observed using sophisticated electron microscopical techniques (Miller et al. 1996). Interestingly, we regularly observed a ring of bright fluorescence around the apical organelle exclusion zone (Fig. 4b–h, arrow heads). This ring may correspond to a site where actin bundles are attached to the plasma membrane. Possible membrane attachment sites have been found close to the tip in rhodamine-phalloidin microinjected lily pollen tubes (Miller et al. 1996). However, since most actin bundles in tobacco pollen tubes do not extend into the proximity of the observed actin ring, other explanations for its function appear more likely. The ring actually marks the boundary between the regular cytoplasm and the organelle exclusion zone. It could mediate the maintenance of the distinction between the two different cytoplasmic domains by forming a physical barrier for all organelles that are larger than secretory vesicles. A similar function has been proposed for the cortical actin network in chromaffin cells, which is thought to regulate exocytosis by physically blocking access of secretory granules to the plasma membrane in the uninduced state (Aunis 1998).

The actin cytoskeleton at the pollen tube tip has been suggested to guide secretory vesicles to apical membrane docking sites and to drive pollen tube elongation by a mechanism related to the actin-mediated movement of amoeboid cells (Derksen et al. 1995;Steer & Steer 1989). F-actin organization in pollen tubes, as observed by confocal imaging of GFP-mTn fluorescence, is not in agreement with these postulated actin functions, which would require the presence of a dense actin network immediately underlying the plasma membrane at the extreme pollen tube apex. Instead, filamentous actin was found to be relatively sparse in pollen tube tips and to be essentially absent from the very tube apex. We propose that filamentous actin is essential for pollen tube growth because it is required for the prominent cytoplasmic streaming observed in these cells (Cai et al. 1997;Derksen et al. 1995;Taylor & Hepler 1997). It is well established that what is perceived as cytoplasmic streaming, is actually myosin dependent movement of cell organelles along actin filaments (Williamson 1993). Conceivably, organelle movement is necessary for sustained transport of secretory vesicles, which deliver cell membrane and cell wall material required for growth, to the extending pollen tube tip. The abundance and the arrangement of filamentous actin observed in the cytoplasm of GFP-mTn expressing pollen tubes is consistent with a function of actin filaments as tracks for organelle movement. In addition to its role in cytoplasmic streaming, filamentous actin in the form of a ring around the organelle exclusion zone might be required to maintain proper cytoplasmic organization at the pollen tube tip.


We have constructed a GFP-fusion protein that can serve as a marker for non-invasive visualization of the actin cytoskeleton in different types of living plants cells and most probably also in cells of other organisms. Transgenic A. thaliana lines with a constitutively visible actin cytoskeleton have been established, which can be used to study dynamics and function of the actin cytoskeleton in diverse processes and tissues throughout the plant life cycle. Finally, in vivo labeling of actin filaments using the marker we have developed was employed to visualize previously unknown features of the structure of the pollen tube actin cytoskeleton and to contribute to a better understanding of the mechanism of pollen tube tip growth.

Experimental procedures


Using PCR amplification and standard cloning techniques, a version of GFP optimized for high expression and easy detectability in plant cells was constructed based on mGFP4 (Haseloff et al. 1997). The mGFP4 sequence encodes the wild-type GFP protein (Prasher et al. 1992) but contains silent mutations that disrupt a cryptic intron which interferes with effective expression of the wild-type GFP gene in plant cells (Haseloff et al. 1997). Non-silent mutations were introduced at positions 65 (Ser to Thr; shifts excitation maximum from 395 nm to 490 nm and increases brightness;Heim et al. 1995) and 64 (Phe to Leu; further increases the brightness of GFP-S65T;Cormack et al. 1996). Using silent mutagenesis, restriction sites were added to or eliminated from the GFP sequence to facilitate subsequent cloning steps. To create a fusion between GFP and the mouse talin f-actin binding domain (GFP-mTn), the stop codon at the 3′ end of the GFP sequence was replaced by an oligonucleotide encoding a 5× Gly Ala linker followed by a multiple cloning site (MCS). The mouse talin f-actin binding sequence was PCR amplified from pGST-Tn2345–2541 (McCann & Craig 1997) and inserted in frame with GFP into the MCS. The amino acid sequence of the peptide connecting GFP and the mouse talin f-actin binding domain was: SGAGAGAGAGAG. Whenever possible, oligonucleotides used as PCR primers or inserted into the coding sequence were designed to optimize codon usage for gene expression in A. thaliana. Cloning products were sequenced to confirm error-free oligonucleotide synthesis and PCR amplification.

Modified GFP and GFP-mTn coding sequences were cloned between the NOS poly A+ signal and either the 35S promoter (both derived from pBI121;Jefferson et al. 1987; CLONTECH Laboratories Inc, Palo Alto, CA, USA) or the LAT52 promoter (derived from pLAT52–7;Twell et al. 1990). Vectors obtained by inserting the resulting expression cassettes into pUCAP (van Engelen et al. 1995) were used for transient transformation experiments. For Agrobacterium mediated A. thaliana transformation, GFP and GFP-mTn expression cassettes were cloned into the binary vector pZP202 (Hajdukiewicz et al. 1994), into which a bar (bialaphos resistance;D’Halluin et al. 1992) expression cassette had been inserted. The bar expression cassette was constructed by fusing a fragment containing the bar coding sequence under the control of the NOS promoter (derived from pGPTV-BAR;Becker et al. 1992) to the pea rbcS-E9 poly A+ signal (Coruzzi et al. 1984). The nucleotide sequences of all constructs are available upon request.

Particle bombardment

BY-2 cells (Nagata et al. 1992) were grown in 50 ml NT medium in 250 ml Erlenmayer flasks as described in Newman et al. (1993) and weekly subcultured by transferring 0.5 ml stationary culture to fresh medium. Cells contained in 4 ml of 3–4-day-old suspension cultures were vacuum filtrated onto filter paper circles (#HAWP 047 00; Millipore, Bedford, MA, USA). Filter paper circles covered with cells were transferred upside-up (cells facing upwards, filter paper in direct contact to the medium) onto 3 ml NT medium solidified with 0.8% agarose in 5.5 cm Petri dishes. Air trapped between filter paper and medium was removed by pressing slightly on the filter paper. Plated cells were kept in the dark at 25°C and bombarded after 12–24 h.

Mature anthers collected from flowering Nicotiana tabacum cultivar SR 1 plants, which were kept in growth chambers under standard conditions (16 h illumination per day at 23°C; night temperature: 20°C), were transferred into PT medium (1 mm CaCl2, 1 mm KCl, 0.8 mm MgSO4, 1.6 mm H3BO3, 30 μm CuSO4, 0.03% casein acid-hydrolysate, 5% sucrose, 12.5% PEG-6000, 10 mg l–1 rifampicin 0.3% MES, pH 5.9, filter sterilized;Read et al. 1993a, 1993b). Pollen was suspended by vortexing and the pollen suspension was filtered through a 1 mm grid to remove anthers. Pollen contained in 4 ml of cleared suspension, representing the amount of pollen obtained from 4 tobacco flowers, was vacuum filtrated onto filter paper circles (#HAWP 047 00; Millipore, Bedford, MA, USA). Filter paper circles covered with pollen were transferred upside-down (pollen facing downwards, in direct contact to medium) onto 3 ml PT medium solidified with 0.25% phytagel (#P-8169; Sigma, St. Louis, MO, USA) in 5.5 cm Petri dishes. After air trapped between filter paper and medium had been removed, the filter paper circles were immediately lifted off the medium and discarded. Essentially, all pollen grains remained on the surface of the culture medium. Pollen was bombarded as soon as possible after plating (generally after 5–10 min).

Particle bombardment was performed using a helium-driven particle accelerator (PDS-1000/He; BIO-RAD, Hercules, CA, USA) with all basic adjustments set according to the manufacturer’s recommendations. Particles with a diameter of 1.6 μm were coated with expression constructs and prepared for bombardment according to the manufacturer’s protocol. Plated BY-2 cells or pollen grains were placed under the stopping screen at a distance of 8 cm and bombarded in a vacuum of 28 inches of mercury using a helium pressure of 1100 psi to accelerate the macrocarrier. Bombarded cells were kept at 25°C in the dark for 2–18 h before analysis.

A. thaliana transformation

Transgenic A. thaliana lines were produced by Agrobacterium vacuum-infiltration of adult plants essentially as described by Bechtold et al. (1993), with some modifications. Agrobacterium strains carrying binary vectors were grown to an OD600 of 1.0 and resuspended in a solution containing 5% sucrose, 10 mm MgCl2 and 0.005% Silwet L-77 (#VIS-01; LEHLE SEEDS, Round Rock, TX, USA). A. thaliana ecotype Landsberg erecta plants were vacuum infiltrated with boiling Agrobacterium suspension for 5 min. T1 seeds obtained from vacuum infiltrated plants were vernalized, sterilized and plated on MS medium containing 10 mg l–1 phosphinothricin (#45520; Riedel-de Haën AG, Seelze, Germany). Resistant plants were transferred to soil and grown under standard conditions (20°C, 16 illumination per day) in an incubator. T2 seeds obtained from such plants were collected for analysis.

Sample preparation for microscopy

Bombarded BY-2 cells were collected from the filter paper and suspended in NT medium. Small drops of the resulting suspension were mounted between slides and coverslips.

Transgenic A. thaliana T2 seeds were plated on MS medium containing 10 mg l–1 phosphinothricin (Riedel-de Haën AG, Seelze, Germany). Resistant seedlings were grown for 5 days in the dark at 25°C. Seedlings or parts of seedlings were mounted in water between slides and coverslips.

Squares of solid culture medium covered with pollen tubes grown from bombarded grains were cut out and flipped upside-down (pollen tubes in direct contact with the coverslip) onto a cover slip.

Rhodamine-phalloidin staining

BY-2 cells were collected from the filter paper on the culture medium and suspended in a solution with a pH adjusted to 6.8 that contained 5 mm MgCl2, 5 mm EGTA, 50 mm PIPES, 0.05% non-idet P40, 10% DMSO, 10% Citifluor (#19470; TED PELLA Inc, Redding, CA, USA) and 150 nm rhodamine-phalloidin (#R-415; Molecular Probes, Eugene, OR, USA). Small drops of the resulting suspension were mounted between slides and coverslips. Samples were analyzed under the microscope immediately after preparation.

Confocal microscopy

Samples were examined using a 100× oil immersion lens (#440285) under a Zeiss LSM410 inverted confocal microscope (Carl Zeiss Inc., Thornwood, NY, USA) equipped with an external argon laser (#910740) and dual internal helium-neon lasers (#4525749901). GFP fluorescence was imaged using excitation with the 488 nm line of the argon laser, a 510 nm beam splitter (#446434) and a 515–565 band pass emission filter (#4679949904). The helium-neon laser providing excitation at 543 nm, a 488/543 nm beam splitter (#4525859901) and a 570 nm long pass emission filter (#467922) were employed for rhodamine-phalloidin imaging. Serial confocal optical sections were taken at a step size of 1 μm. Line averaging as indicated in the Fig. 1(e)gends was employed to improve image quality. Imaging of GFP and rhodamine fluorescence emission from double-labeled cells was performed sequentially. Transmitted light reference images were taken after fluorescence imaging using differential interference contrast (Nomarski) optics and helium-neon laser illumination at 543 nm. Pollen tube growth rates were determined with the help of LSM410 image analysis functions by measuring the distance between the positions of the very apex of a particular pollen tube on two overlaid transmitted light images that have been taken at a time interval of 2 min. LSM410 3D reconstruction functions were employed to compute projections of serial confocal sections. Images were contrast enhanced and artificially colored as required using image processing software (Photoshop®; Adobe Systems Inc., Mountain View, CA, USA) and printed on a Fujix Pictrogaphy 3000 (Fuji Photo Film Co. Ltd, Tokyo, Japan) printer.


We would like thank Drs Richard McCann, Susan Craig, Jim Haseloff, Sheila McCormick, Pal Maliga and Willem Stiekema for providing cDNAs, promoters and vectors; Yang-Sun Chan for help with plasmid construction; and Drs Diana Colgan and Ueli Klahre for critical reading of the manuscript. This work was supported in part by a DOE grant (DOE94ER20143) to N.H.C. P.S. and B.K. were supported by the Swiss National Science Foundation (grants #823 A-050394 and #823 A-046686, respectively).