Nt-RhoGDI2 regulates Rac/Rop signaling and polar cell growth in tobacco pollen tubes

Authors


*(fax +49 6221 54 5859; e-mail bkost@hip.uni-hd.de).

Summary

Rac/Rop-type Rho-family small GTPases accumulate at the plasma membrane in the tip of pollen tubes and control the polar growth of these cells. Nt-RhoGDI2, a homolog of guanine nucleotide dissociation inhibitors (GDIs) regulating Rho signaling in animals and yeast, is co-expressed with the Rac/Rop GTPase Nt-Rac5 specifically in tobacco (Nicotiana tabacum) pollen tubes. The two proteins interact with each other in yeast two-hybrid assays, preferentially when Nt-Rac5 is prenylated. Transient over-expression of Nt-Rac5 and Nt-RhoGDI2 depolarized or inhibited tobacco pollen tube growth, respectively. Interestingly, pollen tubes over-expressing both proteins grew normally, demonstrating that the two proteins functionally interact in vivo. Nt-RhoGDI2 was localized to the pollen tube cytoplasm and effectively transferred co-over-expressed YFP–Nt-Rac5 fusion proteins from the plasma membrane to this compartment. A single amino acid exchange (R69A), which abolished binding to Nt-RhoGDI2, caused Nt-Rac5 to be mis-localized to the flanks of pollen tubes and strongly compromised its ability to depolarize pollen tube growth upon over-expression. Based on these observations, we propose that Nt-RhoGDI2-mediated recycling of Nt-Rac5 from the flanks of the tip to the apex has an essential function in the maintenance of polarized Rac/Rop signaling and cell expansion in pollen tubes. Similar mechanisms may generally play a role in the polarized accumulation of Rho GTPases in specific membrane domains, an important process whose regulation has not been well characterized in any cell type to date.

Introduction

Pollen tubes grow rapidly through flower tissues and mediate fertilization by transporting sperm cells to ovules (Bedinger et al., 1994). Vegetative pollen tube cells elongate by tip growth, during which polarized apical secretion results in unidirectional cell expansion (Hepler et al., 2001). Because pollen tubes grow well in culture and are amenable to experimental manipulation, they are widely employed as a model system to investigate polar cell expansion in plants. Evidence accumulated during the past decade demonstrates that Rac/Rop GTPases, plant homologs of the Rho family of small GTPases, associate with the plasma membrane specifically in the tip of pollen tubes and are key regulators of the polar growth of these cells (Kost et al., 1999; Li et al., 1999; Zheng and Yang, 2000). However, the molecular mechanisms that control the localization and activity of pollen tube Rac/Rop are not well understood.

Rho-family GTPases constitute a group of highly conserved signaling proteins that have key functions in the regulation of cellular polarization and polar cell growth in yeast and animals (Etienne-Manneville and Hall, 2002). Rho GTPases act as molecular switches that transduce signals in the GTP-bound conformation by interacting with a variety of effectors, and are inactive after GTP hydrolysis. Inter-conversion between active GTP-bound and inactive GDP-bound forms of Rho GTPases is regulated by GTPase-activating proteins (RhoGAPs), which stimulate GTP hydrolysis, and guanine nucleotide exchange factors (RhoGEFs), which catalyze GDP release to promote GTP binding. Point mutations can be introduced into the particularly highly conserved nucleotide binding domain of Rho GTPases to generate constitutively active or dominant negative mutant versions of these proteins. Constitutively active Rho GTPases display little GTPase activity even in the presence of RhoGAPs, and show a significantly increased association with GTP in vivo and in vitro (Read et al., 2000; Trahey and McCormick, 1987). In contrast, dominant negative forms of these proteins do not bind nucleotides well and preferentially assume a conformation that mimics the nucleotide-free transition state in which Rho GTPases interact most strongly with RhoGEFs. Over-expressed dominant negative mutant Rho GTPases are thought to inhibit endogenous Rho signaling by forming inactive complexes with available RhoGEFs (Feig, 1999).

Most Rho GTPases are post-translationally modified by the attachment of an isoprenoid lipid to the cysteine residue within a C-terminal CAAX box. Prenylation is responsible for the association of Rho GTPases with membranes, which is thought to be essential for the activity of these proteins (Wennerberg and Der, 2004). Membrane-bound Rho GTPases are typically in equilibrium with a cytoplasmic pool that is maintained by another group of regulatory proteins, the guanine nucleotide dissociation inhibitors (RhoGDIs; Sasaki and Takai, 1998). A key biochemical activity of these proteins is their ability to remove prenylated Rho GTPases from membranes and to form cytoplasmic heterodimers with them (Michaelson et al., 2001; Read et al., 2000). Consistent with these activities, RhoGDIs are most commonly thought to act as negative regulators of Rho signaling in vivo (Etienne-Manneville and Hall, 2002; Olofsson, 1999; Sasaki and Takai, 1998), although recent studies have indicated that the translocation of activated Rho GTPases to specific membrane domains (Del Pozo et al., 2002) and the ability of these proteins to control certain cellular processes may depend on RhoGDI interactions (Lin et al., 2003). Membrane-associated RhoGDI dissociation factors (RhoGDFs), which promote membrane translocation of Rho GTPases by destabilizing the interaction of these proteins with RhoGDIs (DerMardirossian and Bokoch, 2005), may be involved in the RhoGDI-mediated positive control of Rho signaling.

Plants contain large gene families encoding Rac/Rop GTPases (e.g. 11 members in Arabidopsis). Some of these proteins are specifically expressed in pollen tubes and control the growth of these cells, whereas others regulate various aspects of cellular behavior in different cell types (Gu et al., 2004). Plant Rac/Rop GTPases seem to function in a very similar manner to their yeast and animal homologs, but form a distinct subgroup of the Rho family of small GTPases based on amino acid sequence comparison (Valster et al., 2000), and are controlled in part by plant-specific regulatory mechanisms. Sequences showing homology to genes encoding Dbl-family RhoGEFs, key regulators of Rho GTPases in other organisms, are absent from the Arabidopsis and rice genomes (Valster et al., 2000). A novel plant-specific family of proteins unrelated to previously characterized RhoGEFs has recently been demonstrated to act as GEFs of Rac/Rop GTPases (Berken et al., 2005; Gu et al., 2006). Plant homologs of animal and yeast RhoGAPs or RhoGDIs have been identified, but display unique structural features (e.g. CRIB domains in plant RhoGAPs, Wu et al., 2000; N-terminal extension of plant RhoGDIs, Bischoff et al., 2000; see below).

Although the molecular and cellular functions of plant RhoGAPs (Wu et al., 2000), RhoGEFs (Gu et al., 2006; Kaothien et al., 2005) and RhoGDIs (Carol et al., 2005) have begun to be investigated, they are not well understood to date. Arabidopsis mutants disrupted in a gene encoding a RhoGDI homolog (At-RhoGDI1/SCN1, At3g07880) display defects in restricting sites of Rac/Rop accumulation and cell expansion in root epidermal cells that normally form single root hairs (Carol et al., 2005). An N-terminally truncated version of this Arabidopsis RhoGDI homolog (missing the first 43 amino acids) was found to interact in yeast two-hybrid assays with wild-type and constitutively active, but not with dominant negative, forms of the Arabidopsis Rac/Rop GTPases AtRop4 and AtRop6 (Bischoff et al., 2000). Expression of the same Arabidopsis RhoGDI isoform in tobacco (Nicotiana tabacum) pollen tubes prevented the formation of transversal actin bands induced by co-expression of the Arabidopsis Rac/Rop GTPase At-Rop1 (Fu et al., 2001), and reduced ballooning at the tip caused by over-expression of the tobacco Rac/Rop GTPase Nt-Rac1 (Chen et al., 2003). When expressed in tobacco protoplasts, this Arabidopsis RhoGDI isoform interfered with the activation of an auxin-responsive promoter by over-expressed Nt-Rac1, and with the accumulation of Nt-Rac1 in the cell cortex (Tao et al., 2002).

A more detailed functional characterization of plant RhoGDIs and other regulators of Rac/Rop GTPases is clearly essential to understand how Rac/Rop signaling controls cell behavior. Because different isoforms of Rac/Rop GTPases and their regulators are likely to have distinct features, it is important to investigate functional interactions between isoforms that display overlapping expression patterns, and to perform in vivo experiments using cell types in which these isoforms are co-expressed during normal development. Pollen tubes, in which activated Rac/Rop GTPases specifically accumulate in a restricted apical domain of the plasma membrane (Zheng and Yang, 2000), represent an excellent system to investigate regulatory mechanisms controlling the activity of these proteins, potentially in a spatially resolved manner. Polarized activation of Rho family GTPases does not only occur in pollen tubes, but is also essential for fibroblast migration and yeast cell expansion, for example (Etienne-Manneville and Hall, 2002). Investigation of the control of Rac/Rop signaling in pollen tubes can potentially provide significant insights into the molecular mechanisms underlying this process, which are currently not well understood in any cell type (Etienne-Manneville, 2004).

Here, we present a functional characterization of Nt-RhoGDI2, which was identified in a yeast two-hybrid screen for proteins interacting with the pollen tube Rac/Rop GTPase Nt-Rac5. Nt-RhoGDI2 and Nt-Rac5 are both highly and specifically expressed in elongating tobacco pollen tubes. Analysis of interactions of Nt-RhoGDI2 with wild-type or mutant forms of Nt-Rac5 in two-hybrid assays and in co-over-expression experiments has demonstrated that Nt-RhoGDI2 is a key regulator of Nt-Rac5 signaling and polar growth in pollen tubes. Our data strongly suggest that Nt-RhoGDI2-mediated recycling of Nt-Rac5 from the flanks to the apex of the pollen tube tip controls the intracellular distribution and activity of this GTPase.

Results

Identification of Nt-RhoGDI2 as an interaction partner of Nt-Rac5

The tobacco Rac/Rop GTPase Nt-Rac5 (AJ250174) has been reported to be specifically expressed at high levels in pollen (Kieffer et al., 2000), and is very closely related (93.9–95.4% identical amino acids) to Rac/Rop GTPases known to control pollen tube growth in other plant species (At-Rop1, At-Rop5/Rac2; Ps-Rop1; Gu et al., 2004). We have cloned the coding sequence of Nt-Rac5 by RT-RCR using RNA from tobacco pollen tubes grown in vitro as a template. Transcripts with the expected length of 1.4 kb hybridizing under stringent conditions to the Nt-Rac5 coding sequence were found to accumulate to high levels specifically in tobacco flowers, pollen grains and pollen tubes (Figure 1a). The effects of over-expressing wild-type or mutant Nt-Rac5 on tobacco pollen tube growth, as well as the intracellular localization of these different forms of Nt-Rac5 fused to YFP, were essentially the same as those observed with other pollen tube Rac/Rop GTPases (see Figures 5a and 7) (Chen et al., 2003; Kost et al., 1999; Li et al., 1999). Together, these data establish that Nt-Rac5 regulates polar pollen tube growth in a similar manner to its closest homologs.

Figure 1.

 Analysis of gene expression patterns by Northern blotting.
Aliquots of 10 μg (a) or 5 μg (b) total RNA per lane were separated by gel electrophoresis (lower panels), blotted and hybridized with DIG-labeled PCR fragments corresponding to the Nt-Rac5 (a) or Nt-RhoGDI2 (b) coding sequences (upper panels). Groups of lanes (separated by vertical lines) shown in (a) have been rearranged to facilitate comparison with (b). R, root; S, stem; L, leaf; Fb, flower bud; Fm, mature flower; P, mature pollen; PT, pollen tube; *, 1.4 kb; °, 1 kb.

Figure 5.

 Morphology of tobacco pollen tubes transiently over-expressing wild-type or mutant Nt-Rac5, in the absence or presence of co-over-expressed Nt-RhoGDI2.
Microscopic analysis of pollen tubes co-expressing 24 h after gene transfer different forms of Nt-Rac5 (a) with GUS and YFP, or (b) with Nt-RhoGDI2 and YFP. Upper panels in (a) and (b): high-magnification (40× lens) epi-fluorescence and transmitted light images (scale bar: 50 μm). Lower panels in (a) and (b): low-magnification (5× lens) epi-fluorescence images (scale bar: 500 μm). g, pollen grain; t, pollen tube tip.

Figure 7.

 Intracellular localization of transiently expressed YFP fused to wild-type or mutant Nt-Rac5 in tobacco pollen tubes in the presence or absence of co-over-expressed Nt-RhoGDI2.
Single confocal optical sections through the tips of weakly fluorescent pollen tubes co-expressing wild-type or mutant YFP:Nt-Rac5 with GUS (left column) or Nt-RhoGDI2 (right column) at low levels. Images were taken 4–6 h after gene transfer, before significant growth depolarization was induced, and represent at least 20 similar images collected in three independent experiments. In all cases, central sections through pollen tube tips lying flat on the cover-slip surface were imaged (scale bar: 10 μm). WT, Nt-Rac5; G15V, Nt-Rac5G15V; T20N, Nt-Rac5T20N; R69A, Nt-Rac5R69A; arrow heads, weak association of YFP:Nt-Rac5T20N and YFP:Nt-Rac5R69A with the plasma membrane at the tip. Serial confocal sections through pollen tubes expressing YFP:Nt-Rac5 or YFP:Nt-Rac5T20N are shown in Figure S4.

To identify proteins involved in Rac/Rop signaling in pollen tubes, a cDNA library representing genes expressed in tobacco pollen tubes 3 h after germination was screened for sequences encoding polypeptides that show yeast two-hybrid interaction with constitutively active Nt-Rac5 (Nt-Rac5G15V) carrying an additional point mutation (C194S) that disrupts the C-terminal CAAX domain and prevents prenylation. A screen of approximately 3 × 105 yeast transformants resulted in the selection of six cDNA clones encoding pollen tube polypeptides that showed specific interactions with Nt-Rac5 (unpublished data). These polypeptides were similar to hypothetical proteins (n = 3), protein kinases (n = 2) or RhoGDIs (n = 1). The RhoGDI cDNA clone coded for an N-terminally truncated polypeptide of 223 amino acids with a lysine residue at position one. Colony hybridization screening of the tobacco pollen tube library described above using a probe derived from the partial RhoGDI cDNA resulted in the identification of a corresponding full-length cDNA clone (DQ416769). This cDNA clone contained stop codons in all reading frames upstream of an open reading frame coding for a 235 amino acid protein, called Nt-RhoGDI2 hereafter.

Nt-RhoGDI2 interacts specifically with wild-type and constitutively active Nt-Rac5 in two-hybrid assays

In yeast two-hybrid assays, full length Nt-RhoGDI2 fused to the activation domain of the GAL4 transcription factor (GAL4-AD) interacted specifically with Nt-Rac5 and constitutively active Nt-Rac5G15V fused to the GAL4 DNA binding domain (GAL4-DB). In contrast, no two-hybrid interaction was detected between Nt-RhoGDI2 and dominant negative Nt-Rac5 (Nt-Rac5T20N, Figure 2 and Figure S1). In these assays, growth of yeast co-transformants on medium without histidine was analyzed, which indicates reporter gene expression resulting from two-hybrid interactions. Co-transformants expressing the GAL4-AD:Nt-RhoGDI2 fusion together with free GAL4-DB, or free GAL4-AD together with wild-type or mutant Nt-Rac5 fused to GAL4-DB, did not grow on histidine-free medium.

Figure 2.

 Yeast two-hybrid interactions between Nt-RhoGDI2 and Nt-Rac5.
Yeast transformants co-expressing different forms of Nt-Rac5 (G15V, constitutively active; T20N, dominant negative) with an intact CAAX domain fused to the DNA binding domain of the GAL4 transcription factor (GAL4 DB), together with Nt-RhoGDI2 fused to the GAL4 activation domain (GAL4 AD), plated on histidine-containing medium and on histidine-free medium. Serving as negative controls are transformants co-expressing different forms of GAL4 DB:Nt-Rac5 or GAL4 AD:Nt-RhoGDI2, along with free GAL4 AD or GAL4 DB, respectively. Growth of yeast transformants on histidine-free medium indicates reporter gene activation resulting from two-hybrid interactions between Nt-RhoGDI2 and Nt-Rac5.

Data presented in Figure 2 show two-hybrid interactions between Nt-RhoGDI2 and wild-type or mutant forms of Nt-Rac5 with an unmodified C-terminal CAAX domain, which are presumably prenylated when expressed in yeast (Read et al., 2000). Prenylation is expected to interfere with nuclear targeting, which is required for the detection of two-hybrid interactions. On the other hand, a large proportion of the free energy of RhoGDI binding to Rho GTPases is thought to be provided by interactions between a hydrophobic pocket at the C-terminus of RhoGDIs and the prenyl tail attached to Rho GTPases (Hoffman et al., 2000; Scheffzek et al., 2000). The strength of yeast two-hybrid interactions between Nt-RhoGDI2 and different forms of prenylated or unprenylated (C194S) Nt-Rac5 was estimated based on the growth of yeast co-transformants on medium containing the histidine biosynthesis inhibitor 3-amino-1,2,4-triazole (3-AT) at different concentrations (Aguilar et al., 1997). In these experiments, Nt-RhoGDI2 interacted most strongly with prenylated wild-type Nt-Rac5 (Figure S1).

Nt-RhoGDI2 is closely related to well characterized RhoGDIs, and is specifically expressed at high levels in pollen and in pollen tubes

Nt-RhoGDI2 shows significant similarity (Figure S2) to bovine Bt-RhoGDI1 (S12121, 26.7% identical amino acids) and yeast Sc-RDI1 (BAA06499, 27.1% identical amino acids), two RhoGDIs sharing 34.9% identical amino acids that have been extensively characterized structurally (Hoffman et al., 2000) and/or biochemically (Masuda et al., 1994; Sasaki et al., 1993).

Probing Northern blots under stringent conditions with the Nt-RhoGDI2 coding region demonstrated that hybridizing transcripts with the expected length of 1 kb were detectable in flowers, and accumulated to high levels specifically in pollen and in pollen tubes (Figure 1b). The data presented in Figure 1 indicate that Nt-Rac5 and Nt-RhoGDI2 show substantial co-expression specifically in tobacco pollen and pollen tubes. Together with the significant binding between Nt-Rac5 and Nt-RhoGDI2 observed in two-hybrid assays (Figure 2 and Figure S1), this indicated that the two proteins may functionally interact during pollen tube elongation.

Transiently over-expressed Nt-RhoGDI2 inhibits tobacco pollen tube growth and accumulates in the cytoplasm

Cultured tobacco pollen tubes transiently expressing Nt-RhoGDI2 under the control of the Lat52 promoter (Twell et al., 1991) following gene transfer into germinating pollen by particle bombardment remained much shorter compared to control pollen tubes expressing the non-invasive marker protein β-glucuronidase (GUS) under the control of the same promoter (Figure 3). YFP co-expression, also under the control of the Lat52 promoter, was employed to identify transformed pollen tubes. As indicated by YFP fluorescence, the expression level of introduced genes constantly increased during the first 24 h after particle bombardment, and varied considerably between individual pollen tubes. Analysis of pollen tubes readily detectable by epi-fluorescence microscopy using short (<500 msec) exposure times showed that Nt-RhoGDI2-expressing pollen tubes were less than half as long as GUS-expressing pollen tubes 9 h after gene transfer (Figure 3b), and, in contrast to the latter, did not continue to elongate after this time (Figure 3a).

Figure 3.

 Effects of transient over-expression of Nt-RhoGDI2 and YFP:Nt-RhoGDI2 on tobacco pollen tube growth.
(a) Microscopic analysis of pollen tubes co-expressing the indicated proteins 24 h after gene transfer. Upper panels: high-magnification (40× lens) epi-fluorescence and transmitted light images (scale bar: 50 μm). Lower panels: low-magnification (5× lens) epi-fluorescence images (scale bar: 500 μm). g, pollen grain; t, pollen tube tip.
(b) Statistical analysis of pollen tube length 9 h after gene transfer (representative example of three independent data sets). GDI2, Nt-RhoGDI2; error bars: 95% CI (n ≥ 42).

When co-expressed with GUS, Nt-RhoGDI2 fused to the C-terminus (Figure 3), or the N-terminus (unpublished data), of YFP had the same effects on pollen tube growth as co-expression of free Nt-RhoGDI2 and YFP (Figure 3), indicating that both YFP fusions were functional. Neither fusion protein affected pollen tube growth when transiently expressed at minimal levels, at which they were barely detectable by low magnification epi-fluorescence microscopy using short (<500 msec) exposure times (Figure 4a). Confocal analysis of weakly fluorescent pollen tubes expressing Nt-RhoGDI2 fused to the C-terminus (Figure 4b), or the N-terminus (unpublished data), of YFP at minimal levels demonstrated that both fusion proteins were diffusely distributed throughout the pollen tube cytoplasm, similar to transiently expressed free YFP (Figure 4b). Nt-RhoGDI2 fusion proteins expressed at minimal or at higher levels (based on fluorescence intensity) were never found to accumulate at the plasma membrane, not even in the presence of co-over-expressed Nt-Rac5 (unpublished data). In accordance with these observations, fractionation of tobacco pollen tube extracts by high-speed centrifugation showed that endogenously expressed Nt-RhoGDI2 accumulated in membrane-depleted cytoplasm (Figure 4c). The data shown in Figures 3 and 4 are consistent with models of RhoGDI function presented in the literature (Etienne-Manneville and Hall, 2002), which propose that RhoGDIs are cytoplasmic proteins that act as negative regulators of Rho signaling.

Figure 4.

 Intracellular localization of Nt-RhoGDI2 in tobacco pollen tubes.
(a) Low-magnification (5× lens) epi-fluorescence images of pollen tubes expressing YFP:Nt-RhoGDI2 at moderate (upper panels) or minimal (lower panels) levels 6 h after gene transfer (scale bar: 500 μm). Moderate expression inhibited pollen tube growth, caused image saturation after 5 sec of exposure and was optimally imaged with an exposure time of 250 msec. Minimal expression did not affect pollen tube elongation and was visible after 5 sec of exposure, but barely detectable with an exposure time of 250 msec (representative example of 15 similar data sets obtained in three independent experiments).
(b) Single confocal optical sections through the tips of pollen tubes expressing YFP:Nt-RhoGDI2 or YFP at minimal levels 6 h after gene transfer (each representing at least 20 similar images collected in three independent experiments; scale bar: 10 μm). Central sections through pollen tubes lying flat on the cover-slip surface were imaged.
(c) Immunoblot showing accumulation of endogenously expressed Nt-RhoGDI2 (detected using a polyclonal antibody prepared against this protein) in the cytoplasmic fraction (S100k) of tobacco pollen tube extracts, which was separated from membrane fractions (P10k, P100k) by centrifugation. S, supernatant; P, pellet; 10k/100k, 10 000/100 000 g.

Nt-RhoGDI2 and Nt-Rac5 neutralize each other's over-expression effects

Transient over-expression of Nt-Rac5 depolarized pollen tube growth and resulted in the formation of large balloons instead of elongating tips (Figures 5a and 6). The same effects in a more pronounced form were obtained when Nt-Rac5G15V was expressed (Figures 5a and 6). In contrast, Nt-Rac5T20N inhibited pollen tube growth without causing comparable ballooning (Figures 5a and 6). These effects were essentially the same as those observed upon over-expression of wild-type or mutant forms of related pollen tube Rac/Rop GTPases (Chen et al., 2003; Kost et al., 1999; Li et al., 1999).

Figure 6.

 Growth of tobacco pollen tubes transiently over-expressing wild-type or mutant Nt-Rac5, in the absence or presence of co-over-expressed Nt-RhoGDI2.
Statistical analysis of pollen tube length 7 h after gene transfer (representative example of three independent data sets). GDI2, Nt-RhoGDI2; WT, Nt-Rac5; G15V, Nt-Rac5G15V; T20N, Nt-Rac5T20N; R69A, Nt-Rac5R69A; error bars: 95% CI (n ≥ 44).

The pollen tubes shown in Figure 5(a) expressed wild-type or mutant Nt-Rac5 along with YFP and GUS, all under the control of the Lat52 promoter, following bombardment with particles coated with equal amounts of three different expression constructs (see Experimental procedures). To analyze effects of Nt-RhoGDI2 co-over-expression with different forms of Nt-Rac5, a plasmid conferring Lat52-controlled expression of Nt-RhoGDI2 was used instead of the GUS construct (Figure 5b). Interestingly, pollen tubes co-over-expressing Nt-Rac5 and Nt-RhoGDI2 reached a similar length to control pollen tubes expressing only marker proteins 7 h after gene transfer (Figure 6). Apart from some tip-swelling occasionally detected after prolonged culture (24 h, unpublished data), they generally showed a normal morphology (Figure 5b) and displayed neither the massive ballooning nor the reduced growth that were observed when Nt-Rac5 (Figure 5a) or Nt-RhoGDI2 (Figure 3), respectively, were over-expressed individually. These results demonstrate that Nt-RhoGDI2 and Nt-Rac5 functionally interact in tobacco pollen tubes, and indicate, together with the two-hybrid data shown in Figure 3 and Figure S1, that the two proteins may form an inactive complex in the pollen tube cytoplasm.

This interpretation, which is consistent with models of RhoGDI function proposed in the animal and yeast literature (Etienne-Manneville and Hall, 2002), was supported by titration experiments (Figure S3). In these experiments, pollen tubes were bombarded with particles coated with plasmid mixtures containing Nt-Rac5 and Nt-RhoGDI2 expression vectors at different ratios. The combined amount of both expression vectors was kept constant. Normal pollen tube growth was obtained at ratios of roughly 1:1. In contrast, bombardment with an excess of Nt-Rac5 vector (2:1, 3:1 or 5:1) or NtRhoGDI2 vector (1:2, 1:3 or 1:5) induced ballooning or inhibited growth, respectively (Figure S3).

Nt-RhoGDI2 had no effect on the ballooning induced by co-expressed Nt-Rac5G15V (Figures 5b and 6), conceivably because this Nt-Rac5 mutant interacts weakly with Nt-RhoGDI2, as indicated by two-hybrid assays (see Figure S1), and has a higher potential to depolarize pollen tube growth than Nt-Rac5 (Figures 5a and 6). The length of pollen tubes expressing Nt-Rac5T20N was further reduced upon co-expression of Nt-RhoGDI2 (Figures 5 and 6), which is consistent with the expectation that each of the two proteins inhibits pollen tube growth by a different mechanism when over-expressed. Nt-Rac5T20N is likely to inhibit putative Rac/Rop GEFs by forming inactive complexes with them (Feig, 1999), whereas Nt-RhoGDI2 appears to do the same to endogenous Rac/Rop GTPases.

Nt-RhoGDI2 differentially affects the intracellular localization of wild-type and mutant Nt-Rac5

As observed on confocal optical sections through weakly fluorescent elongating pollen tubes, Nt-Rac5 or Nt-Rac5G15V fused to the C-terminus of YFP (YFP:Nt-Rac5/Nt-Rac5G15V) accumulated at the plasma membrane at the tip of these cells when transiently expressed at low levels 4–8 h after gene transfer, before significant growth depolarization was induced (Figure 7, left column, and Figure S4). Wild-type and constitutively active forms of a related pollen tube Rac/Rop GTPase (At-Rac2/Rop5) fused to YFP showed essentially the same intracellular distribution (Kost et al., 1999).

Co-expression of Nt-RhoGDI2 prevented YFP:Nt-Rac5 from detectably accumulating at the plasma membrane (Figure 7, right column), which confirms that Nt-RhoGDI2 functionally interacts with Nt-Rac5 in vivo and demonstrates that it can translocate associated Rho GTPases from the plasma membrane to the cytoplasm. This ability is characteristic of all RhoGDIs, and is thought to be essential for the regulation of Rho signaling by these proteins (Etienne-Manneville and Hall, 2002). YFP:Nt-Rac5G15V was only partially removed from the plasma membrane by co-over-expressed Nt-RhoGDI2 (Figure 7), which is consistent with the comparably weak interaction between Nt-RhoGDI2 and Nt-Rac5G15V detected in two-hybrid assays (Figure S1), as well as with the absence of an effect of Nt-RhoGDI2 on ballooning induced by Nt-Rac5G15V in co-over-expression experiments (Figures 5 and 6).

Particularly informative was the analysis of the intracellular localization of Nt-Rac5T20N fused to the C-terminus of YFP, both in the absence and in the presence of co-over-expressed Nt-RhoGDI2, with which this form of Rac5 does not detectably interact in two-hybrid assays (Figure 2 and Figure S1). In the absence of Nt-RhoGDI2 over-expression, YFP:Nt-Rac5T20N clearly displayed a much stronger overall association with the plasma membrane than YFP:Nt-Rac5 and YFP:Nt-Rac5G15V (Figure 7 and Figure S4), indicating that endogenous pollen tube RhoGDI activity partially transferred YFP:Nt-Rac5 and YFP:Nt-Rac5G15V, but not YFP:Nt-Rac5T20N, from the plasma membrane to the cytoplasm. Under these conditions, in contrast to YFP:Nt-Rac5 and YFP:Nt-Rac5G15V, YFP:Nt-Rac5T20N did not significantly accumulate at the plasma membrane in the pollen tube apex (Figure 7, arrow head, and Figure S4), where endogenously expressed pollen tube Rac/Rop GTPases are presumed to be localized (Zheng and Yang, 2000). These observations suggest the exciting possibility that RhoGDI interaction may promote tip localization of pollen tube Rac/Rop GTPases.

Co-over-expressed Nt-RhoGDI2 failed to effectively transfer YFP:Nt-Rac5T20N to the cytoplasm, but detectably reduced the overall plasma membrane association of this fusion protein and induced its partial redistribution to the pollen tube tip (Figure 7). Residual low-affinity binding of Nt-RhoGDI2 to Nt-Rac5T20N, which is beyond the detection limit of two-hybrid assays (Figure 2 and Figure S1), may have been responsible for these effects. However, it cannot be excluded that the strong inhibition of pollen tube growth resulting from Nt-RhoGDI2 co-over-expression indirectly affected YFP:Nt-Rac5T20N localization in these experiments.

Nt-Rac5R69A is impaired in its ability to interact with Nt-RhoGDI2, to accumulate at the pollen tube tip and to affect pollen tube growth when over-expressed

To clarify the role of interactions with Nt-RhoGDI2 in the control of the intracellular localization of Nt-Rac5, we have replaced the arginine at position 69 of Nt-Rac5 by alanine using site-directed mutagenesis. Corresponding mutations have been shown to specifically disrupt the interaction of animal Rho GTPases with RhoGDIs, without affecting their binding to effectors or other regulators (Gibson and Wilson-Delfosse, 2001; Lin et al., 2003). As demonstrated using yeast two-hybrid assays, the R69A mutation abolished the interaction of Nt-Rac5 with Nt-RhoGDI2, but did not significantly affect binding of Nt-Rac5 to Nt-RhoGAP, Nt-Kinase1 or Nt-Ric1, three putative Nt-Rac5 regulators or effectors that we are currently functionally characterizing (Figure S5). Together with the observation that recombinant Nt-Rac5 and Nt-Rac5R69A displayed identical in vitro GTPase activities (Figure S5), this demonstrates that the R69A mutation disrupts binding of Nt-Rac5 to Nt-RhoGDI2 without affecting other key biochemical functions of this protein.

Interestingly, transiently over-expressed Nt-Rac5R69A failed to induce the massive ballooning at the pollen tube tip that is induced by over-expressed Nt-Rac5 (Figure 5a). Nt-Rac5R69A over-expression had no effect on the length of tobacco pollen tubes 7 h after gene transfer (Figure 6), and generally did not affect pollen tube morphology (Figure 5a), although some tip-swelling was occasionally observed after prolonged culture (24 h, unpublished data). These data strongly suggest that the activation of Nt-Rac5 in pollen tubes requires interaction with Nt-RhoGDI2, and that Nt-RhoGDI2 promotes rather than downregulates Rac/Rop signaling in these cells. In contrast to NtRac5, Nt-Rac5R69A did not prevent the inhibition of pollen tube growth resulting from transiently co-over-expressed Nt-RhoGDI2 (Figures 5b and 6), which confirmed that the R69A mutation disrupted the functional interaction between Nt-Rac5 and Nt-RhoGDI2 in living pollen tubes.

Both in the presence and in the absence of co-over-expressed Nt-RhoGDI2, the intracellular localization in pollen tubes of transiently expressed Nt-Rac5R69A fused to the C-terminus of YFP was indistinguishable from that of YFP fused to Nt-Rac5T20N (Figure 7), which does not detectably interact with Nt-RhoGDI2 in yeast two-hybrid assays either (Figure 2 and Figure S1). As shown in Figure 7 (left column), YFP:Nt-Rac5R69A strongly associated with the plasma membrane in the shank and in the flanks of the tip, but not in the apex (arrow head), of normally growing pollen tubes that did not co-over-express Nt-RhoGDI2. This strongly suggests that Nt-RhoGDI2 interaction is required for the tip localization of wild-type Nt-Rac5. Nt-Rac5R69A is not only mis-localized but also fails to depolarize pollen tube growth when over-expressed (see above), which indicates that Nt-RhoGDI2-dependent transport of Nt-Rac5 to the tip may be essential for activation of this protein.

As discussed above, the intracellular distributions of YFP:Nt-Rac5T20N and YFP:Nt-Rac5R69A in pollen tubes co-over-expressing Nt-RhoGDI2 are consistent with low-affinity interactions between these fusion proteins and Nt-RhoGDI2, which are not detectable in two-hybrid assays. However, it is possible that these distributions were indirectly affected by the strong inhibition of pollen tube growth resulting from Nt-RhoGDI2 co-over-expression.

Discussion

Functional interactions between Nt-RhoGDI2 and Nt-Rac5 play an import role in the control of tobacco pollen tube growth

Several lines of evidence demonstrate that Nt-Rac5 and Nt-RhoGDI2 are specifically co-expressed in tobacco pollen tubes. The Nt-Rac5 coding sequence was cloned by RT-PCR using RNA isolated from tobacco pollen tubes as a template. Colony hybridization screening of a tobacco pollen cDNA library using the Nt-Rac5 coding sequence as a probe resulted in the identification of two cDNA clones encoding Nt-Rac5 and a novel, highly similar pollen tube Rac/Rop GTPase, respectively (U. Klahre and B. Kost, unpublished data). cDNAs encoding Nt-RhoGDI2 were identified in the same pollen tube library using yeast-two hybrid and colony hybridization screening. Transcripts hybridizing to the Nt-Rac5 or the Nt-RhoGDI2 coding sequences were detectable in flowers containing developing or mature pollen, and accumulated to high levels specifically in pollen and in pollen tubes (Figure 1). Similar expression patterns are displayed by other genes with important functions during pollen tube growth (e.g. Chen et al., 2002; Cheung et al., 2002). Mature pollen grains are known to store large amounts of transcripts of such genes to prepare for rapid protein production upon germination (Becker et al., 2003; Honys and Twell, 2003).

Nt-Rac5 and Nt-RhoGDI2 dramatically affected the growth of tobacco pollen tubes when over-expressed individually, in the case of Nt-Rac5 in a manner that is characteristic of pollen tube Rac/Rop GTPases. In transient over-expression experiments, pollen tube elongation was depolarized by Nt-Rac5 or Nt-Rac5G15V, and strongly inhibited by Nt-Rac5T20N or Nt-RhoGDI2, indicating that Nt-Rac5 and Nt-RhoGDI2 both have important functions in the control of this process. Interestingly, the two proteins showed significant interaction with each other in yeast two-hybrid assays (Figure 2 and Figure S1) and functionally interacted in vivo (Figures 5–7 and Figure S3). When Nt-RhoGDI2 was co-over-expressed with wild-type Nt-Rac5, with which it interacts most strongly (Figure 7 and Figure S1), under conditions that are expected to result in similar expression levels of the two proteins, they prevented each other from affecting pollen tube growth (Figures 5 and 6 and Figure S3). Rac/Rop signaling apparently remained unperturbed and pollen tubes were able to elongate normally, conceivably because excess Rac/Rop and RhoGDI proteins were sequestered into inactive cytoplasmic heterodimers, without disturbing the critical equilibrium between the activities of these proteins. A minor increase or decrease of the ratio between Nt-Rac5 and Nt-RhoGDI2 over-expression levels in titration experiments depolarized or inhibited pollen tube growth, respectively, hinting at a rather delicate nature of this equilibrium (Figure S3). When co-expressed in normally elongating pollen tubes with Nt-RhoGDI2, YFP:Nt-Rac5 did not detectably accumulate at the plasma membrane in the apex (Figure 7), although the presence of Rac/Rop GTPases at this location is thought to be essential for pollen tube elongation (Zheng and Yang, 2000). Conceivably, excess YFP:Nt-Rac5 forming an inactive complex with Nt-RhoGDI2 in the cytoplasm (see above) has prevented the detection of low-level accumulation of this fusion protein at the plasma membrane.

Together, these data establish that functional interactions between Nt-RhoGDI2 and Nt-Rac5 play an important role in the control of tobacco pollen tube growth. This conclusion is strongly supported by the observation that the R69A mutation, which disrupted the interaction of Nt-Rac5 with Nt-RhoGDI2 in two-hybrid assays (Figure S5) and in co-over-expression experiments (Figures 5–7), dramatically reduced the ability of this protein to depolarize pollen tube growth upon transient over-expression.

Nt-RhoGDI2 can translocate Nt-Rac5 from the pollen tube plasma membrane to the cytoplasm

In accordance with the considerable sequence conservation between well-characterized animal or yeast RhoGDIs and Nt-RhoGDI2, these proteins share a number of important features. They interact with nucleotide-bound, but not with nucleotide-free, Rho GTPases (Figure 2 and Figure S1) (Gibson and Wilson-Delfosse, 2001; Michaelson et al., 2001; Strassheim et al., 2000). Curiously, animal RhoGDIs have been reported in some cases to bind with a strong preference to GDP-bound Rho GTPases (Ueda et al., 1990), whereas other studies have indicated that they interact equally well with GDP- and GTP-bound forms of these proteins (Nomanbhoy and Cerione, 1996). In vitro interaction assays using nucleotide-loaded recombinant Nt-Rac5 would be needed to firmly establish possible differences in the affinities with which this protein in the GDP- or GTP-bound conformations binds to Nt-RhoGDI2. Such assays could not be performed, because they require significant amounts of prenylated recombinant Nt-Rac5 (Nomanbhoy and Cerione, 1996), which we were unable to generate using various methods that work for animal and yeast homologs (over-expression in yeast, in vitro translation in reticulocyte extracts, prenylation using cellular extracts or purified geranylgeranyl transferase). However, our data show that Nt-RhoGDI2 interacts more strongly with Nt-Rac5 than with Nt-Rac5G15V in two-hybrid assays, which is consistent with preferential binding of Nt-RhoGDI2 to GDP-bound Nt-Rac5 (Figure S1).

The interaction of animal RhoGDIs (Gibson and Wilson-Delfosse, 2001; Hori et al., 1991; Nomanbhoy and Cerione, 1996), as well as of Nt-RhoGDI2 (Figure S1), with their target Rho GTPases is promoted by prenylation of the latter. RhoGDIs, including Nt-RhoGDI2, are localized to the cytoplasm (Figure 4) (Koch et al., 1997; Ueda et al., 1990), can remove target Rho GTPases from membranes and are able to retain these proteins in the cytoplasm (Figure 7) (Isomura et al., 1991; Michaelson et al., 2001; Read et al., 2000), presumably by forming heterodimers with them. Consistent with these activities and the notion that membrane localization of Rho GTPases is considered essential for their activity (Wennerberg and Der, 2004), over-expression of RhoGDIs was generally found to inhibit cellular processes that depend on Rho signaling (e.g. Li et al., 2003; Masuda et al., 1994). The inhibition of pollen tube elongation by over-expressed Nt-RhoGDI2 (Figure 3) represents a typical example of this effect. Based on the observations summarized in this paragraph, RhoGDIs are most commonly thought of as negative regulators of Rho GTPases, which remove inactive GDP-bound forms of these proteins from the plasma membrane and maintain them in this state in the cytoplasm (Etienne-Manneville and Hall, 2002; Olofsson, 1999; Sasaki and Takai, 1998).

Nt-RhoGDI2 interaction is essential for the normal localization and activity of Nt-Rac5

Recent observations have indicated that RhoGTPase signaling in some cell types is promoted rather than downregulated by RhoGDIs. RhoGDIs that appear to strongly interact with GTP-bound Rho GTPases have been proposed to have a function in translocating these proteins in the activated form to specific membrane domains (Del Pozo et al., 2002). A mammalian Rho GTPase carrying a mutation corresponding to the R69A exchange in Nt-Rac5 was specifically disrupted in its ability to interact with RhoGDI and failed to induce the same effects as its wild-type counterpart when over-expressed in cell cultures, indicating that the functions of this GTPase depend at least in part on RhoGDI interaction (Lin et al., 2003).

We have demonstrated that Nt-Rac5R69A does not significantly interact with Nt-RhoGDI2 in yeast two-hybrid assays (Figure S5a) or in co-over-expression experiments (Figures 5b, 6 and 7), but shows the same GTPase activity as Nt-Rac5 (Figure S5b) and displays normal two-hybrid interactions with three different putative Nt-Rac5 regulators or effectors (Figure S5a). Interestingly, the R69A mutation caused Nt-Rac5 to show enhanced overall membrane association (Figure 7), to accumulate at the plasma membrane in the shank and in the flanks of the pollen tube tip, but not in the apex (Figure 7), and to be significantly compromised in its capability to induce growth depolarization upon over-expression (Figure 5a). These data strongly suggest that interaction with Nt-RhoGDI2 is essential for the normal localization and activity of Nt-Rac5 at the tip of tobacco pollen tubes.

Consistent with the results summarized above and with all our other data, we propose that Nt-RhoGDI2 mediates recycling of inactive Nt-Rac5 from the flanks of the tip to the pollen tube apex, where this GTPase is activated (Figure 8). Our data provide compelling evidence for a key role of this mechanism in the maintenance of the spatially restricted Rac/Rop signaling that controls polarized cell expansion at the tip of tobacco pollen tubes. Although the recently observed accumulation of Rac/Rop GEFs at the pollen tube apex (Gu et al., 2006) strongly supports the model shown in Figure 8, the proposed association of RhoGDF and RhoGAP activities with the apex and the flanks of the tip, respectively, remains to be demonstrated.

Figure 8.

 Nt-RhoGDI2-mediated recycling is required for normal Nt-Rac5 localization and activity in tobacco pollen tubes.
Nt-RhoGDI2 is proposed to transfer inactive GDP-bound Nt-Rac5 from the plasma membrane at the flanks of the pollen tube tip to the cytoplasm, and to be essential for the transport of this GTPase back to the apex, where it is re-inserted into the plasma membrane and turned into the GTP-bound active form by nucleotide exchange. Asterisks indicate hypothetical activities whose association with the pollen tube plasma membrane at the indicated locations remains to be demonstrated.

In accordance with a key function of GDI-mediated recycling in the control of Rac/Rop localization and activity, it has recently been shown that an Arabidopsis GDI homolog (At-RhoGDI1/SCN1, At3g07880) is essential for the spatial restriction of Rac/Rop-dependent root hair growth (Carol et al., 2005), a process that is closely related to pollen tube elongation (Hepler et al., 2001). From individual root epidermal cells of mutant plants disrupted in the At-RhoGDI1/scn1 gene, instead of a single highly elongated unbranched root hair, several short root hairs emerge, which often split at the tip to form multiple growth sites. As may be expected from the disruption of a GDI-dependent Rac/Rop recycling mechanism similar to that illustrated in Figure 8, enhanced and delocalized accumulation of a Rac/Rop GTPase at the plasma membrane is associated with the cell expansion defects in the root epidermis of At-RhoGDI1/scn1 mutants.

RhoGDI-mediated recycling of Rho GTPases may specifically be required in elongating pollen tubes and root hairs, in which massive apical secretion is expected to cause a constant retrograde flow of plasma membrane material away from the tip. This assumption is consistent with the observation that pollen tube RhoGDIs are highly conserved among each other, but diverge more significantly from sporophytic and non-plant isoforms (Figure S2). However, accumulation of activated Rho GTPases in specific plasma membrane domains plays a crucial role in polarizing not only pollen tubes but also yeast and animal cells (Etienne-Manneville and Hall, 2002). It is possible that Rho GTPase recycling by RhoGDIs is also involved in the polarization of Rho signaling in these cells, a process that is not well understood to date (Etienne-Manneville, 2004).

Experimental procedures

Plant material

As a source of pollen, tobacco (N. tabacum cv. Petit Havana SR1) plants were grown from seeds at monthly intervals and maintained in a greenhouse.

cDNA library construction and cloning by RT-PCR

Total RNA was isolated using repeated phenol–chloroform extraction and selective lithium chloride precipitation based on a published protocol (Soni and Murray, 1994) from tobacco pollen tubes that were grown in liquid culture medium (Read et al., 1993) for 3 h, collected by centrifugation (700 g) and washed with buffer (0.4 m mannitol, 50 mm Tris–Cl pH 6.0). Freezing or grinding were not required for RNA extraction. PolyA+ RNA was prepared by oligo(dT)-cellulose (#27-5543-0227; Amersham Biosciences, Piscataway, NJ, USA) affinity purification. cDNA generated from 5 μg polyA+ RNA using a cDNA synthesis kit (#200400; Stratagene, La Jolla, CA, USA) was cloned into the yeast two-hybrid prey vector pGAD-GH (BD Biosciences-Clontech, Mountain View, CA, USA) restricted with EcoRI and XhoI. The resulting primary cDNA library consisted of 1.3 × 106Escherichia coli clones, of which more than 90% contained plasmids with cDNA inserts that were in roughly one-third of the cases larger than 1 kb. The primary library was amplified for 2 h in liquid LB medium before plasmid purification using a large-scale kit (#12181; Qiagen, Valencia, CA, USA).

A cDNA corresponding to the coding sequence of Nt-Rac5 was amplified from total pollen tube RNA prepared as described above using a one-step reverse transcription PCR kit (#210210; Qiagen). The nucleotide sequence of the amplification product was identical to the Nt-Rac5 sequence reported in the literature (Kieffer et al., 2000).

cDNA isolation by colony hybridization

Colony hybridization was performed using DIG (digoxigenin)-labeled probes as described in the DIG user manual supplied by Roche Inc. (Basel, Switzerland). Approximately 100 000 colonies of the cDNA library described above were screened on large agar plates and re-screened using this procedure. Membranes were hybridized as described below for RNA blots.

Recombinant DNA construction

Site-directed mutagenesis of the Nt-Rac5 coding sequence was performed using a PCR-based mutagenesis kit (#200519 quick-change; Stratagene), or regular PCR with primers carrying mis-matches. Mutagenized fragments were sub-cloned into expression vectors, and sequenced to confirm the absence of PCR errors as well as the presence of introduced mutations.

Standard recombinant DNA methodology (Sambrook and Russell, 2001) was employed to clone coding regions of cDNAs encoding wild-type or mutant Nt-Rac5, or Nt-RhoGDI2, into the multiple cloning sites (MCS) of various expression vectors: (i) a pUCAP-based vector (van Engelen et al., 1995; Kost et al., 1998) containing an MCS between a Lat52 promoter (Twell et al., 1991) and a nos polyA+ signal for transient expression in pollen tubes, (ii) the same vector containing, before or after a modified MCS, a sequence encoding EYFP (BD Biosciences-Clontech) fused at the C- or the N-terminus, respectively, to a 5 × Gly–Ala linker for transient expression of YFP fusion proteins in pollen tubes, (iii) pGBK-T7 (BD Biosciences-Clontech) to express the DNA binding domain of the GAL4 transcription factor fused to different forms of Nt-Rac5 in yeast cells as a bait in two-hybrid screens and assays, (iv) pGAD-GH (BD Biosciences-Clontech) to express the activation domain of the GAL4 transcription factor fused to Nt-RhoGDI2 as prey in yeast for two-hybrid assays, and (v) pGEX-4T (Amersham Biosciences) to express GST (glutathione S-transferase) fusion proteins in E. coli for purification. Expression constructs containing the coding sequences of EYFP, or β-glucuronidase (Kost et al., 1999), between the Lat52 promoter and the nos polyA+ signal were also generated based on the pUCAP-derived vector described above for transient expression in pollen tubes. PCR-amplified fragments introduced into expression vectors, as well as junctions between sequences linked to express translational fusions, were sequenced to confirm the absence of PCR errors and the generation of in-frame fusions, respectively.

RNA isolation and blotting

RNA employed in blotting experiments was isolated using Trizol according to the manufacturer's recommendations (Invitrogen, Carlsbad, CA, USA). RNA blotting was essentially performed as described by Sambrook and Russell (2001). Aliquots of 5–10 μg RNA per lane were loaded on a denaturing agarose gel (1.5% agarose, 20 mm MOPS, 10 mm Na-acetate, 1 mm EDTA, pH 7), electrophoresed in MOPS buffer and blotted onto Duralon-UV membranes (Stratagene). RNA was cross-linked to membranes using a Stratalinker (Stratagene). Probes were prepared using DIG-nucleotides according to the manufacturer's method (Roche Inc.) and hybridized in 47% formamide, 540 mm NaCl, 30 mm Na-phosphate, 3 mm EDTA, 1% SDS, 0.005% each of BSA, Ficoll and polyvinyl-pyrrolidone at 42°C overnight. Membranes were rinsed once in 2× SSC at room temperature followed by three washes in 0.1% SDS, 0.1× SSC at 55°C. DIG-labeled probes were detected according to the manufacturer's recommendation using CDP-star as a substrate for AP-linked anti-DIG antibodies (Roche Inc.). Blots were exposed to Hyperfilm ECL (Amersham Biosciences).

Yeast two-hybrid screen and assays

Yeast two-hybrid screening was performed using materials and protocols provided by the ‘Matchmaker’ GAL4 system (manual PT3061-1; BD Biosciences-Clontech). Saccharomyces cerevisiae HF7c cells containing intact bait constructs (Nt-Rac5G15V/C194S coding sequence in pGBK-T7, see above), as verified by restriction analysis and partial sequencing of plasmid isolated from these cells, were transformed using a large-scale lithium acetate method with 200 μg plasmid purified from a tobacco pollen tube cDNA library (cDNA inserts in pGAD-GH; prey constructs, see above). Transformed cells were plated on medium lacking histidine (#4027-012 and #4530-122; MP Biomedicals, Irvine, CA, USA) to screen for two-hybrid interactions. Simultaneous plating of a small aliquot of transformed cells on medium supplemented with histidine (#24842; Serva, Heidelberg, Germany) indicated the total number of co-transformants containing bait and prey constructs screened. Prey constructs were purified from yeast colonies appearing on histidine-free medium 3–14 days after gene transfer and amplified in E. coli. Specific interactions between Nt-Rac5 and polypeptides encoded by cDNA inserts in purified prey constructs were verified using two-hybrid assays.

To perform yeast two-hybrid assays, prey constructs (pGAD-GH with cDNA inserts) were simultaneously co-transformed into HF7c cells with different bait constructs (pGBK-T7 containing cDNAs encoding wild-type or mutant Nt-Rac5) using a small-scale lithium acetate method (manual PT3061-1; BD Biosciences-Clontech). All bait and prey constructs were also co-transformed with empty pGAD-GH and pGBK-T7, respectively, to generate negative control samples. Equal volumes of each batch of co-transformed cells were plated on histidine-containing medium to determine co-transformation efficiencies, and on histidine-free medium supplemented with 0, 1 or 2 mm 3-AT (3-amino-1,2,4-triacole; #A-8056, Sigma, St Louis, MO, USA) to detect two-hybrid interactions. If co-transformation of all samples was successful (several hundred yeast colonies visible on histidine-containing medium 3 days after gene transfer), specific two-hybrid interactions were demonstrated by growth on histidine-free medium of HF7c cells transformed with bait and prey constructs, and by the absence of growth on this medium of HF7c cells containing only bait or prey constructs along with empty pGAD-GH or pGBK-T7, respectively. To obtain the data shown in Figure 2 and Figures S1 and S5, HF7c co-transformants growing on histidine-containing medium (several colonies from each plate) were transferred to 5 ml liquid histidine-containing medium and cultured. After 48 h, 10 μl aliquots of each culture were plated on media with and without histidine, the latter supplemented with 0, 1 or 2 mm 3-AT. Plates were incubated for 3 days before photographs were taken.

Transient gene expression

Expression vectors were transferred into tobacco pollen grains germinating on solid culture medium (Read et al., 1993) by particle bombardment using a helium-driven particle accelerator (PDS-1000/He; Bio-Rad, Hercules, CA, USA) as previously described (Kost et al., 1998). When two or three plasmids were co-transformed, respectively, particles were coated with 5 μg (2.5 μg of each plasmid) or 6 μg (2 μg of each plasmid) plasmid DNA (unless stated otherwise: see titration experiments). All expression vectors used ranged in size between 4.3 and 5.3 kb.

Microscopy and image analysis

At the indicated times after gene transfer, transiently transformed fluorescent pollen tubes were transferred as previously described (Kost et al., 1998) onto cover slips for microscopic analysis. Epi-fluorescence and transmitted light images were recorded using an inverted microscope (DM IRB; Leica, Bensheim, Germany) equipped with DIC (differential interference contrast) optics, a 100 W mercury lamp, an FITC filter block (excitation: 450–490 nm, dichroic: 510 nm, emission: 515 long pass; I3 S, Leica), 5× and 40× lenses (N PLAN 5×/0.12 and HCX PL FL L 40 ×/0.6, Leica), and a digital camera (DFC350FX R2, Leica). Unless stated otherwise, epi-fluorescence images shown and/or employed to analyze pollen tube length (using publicly available image analysis software: imagej; http://rsb.info.nih.gov/ij/) were taken with exposure times shorter than 500 msec. A laser scanning microscope (#1220004 LSM510Meta; Zeiss, Jena, Germany) and a 100 ×/1,45 NA oil immersion lens (#1084514; Zeiss) were employed for confocal analysis. YFP fluorescence excited with the 514 nm line of an argon laser was imaged through a 405/514 nm dichroic mirror and a 530–600 nm band pass emission filter. Epi-fluorescence and confocal images were contrast-enhanced by adjusting brightness and gamma settings using image-processing software (photoshop; Adobe Systems Inc., San Jose, CA, USA).

GTPase assays

Recombinant wild-type and mutant versions of Nt-Rac5 fused to the C-terminus of GST were purified from E. coli BL-21 using standard procedures (Sambrook and Russell, 2001) and assayed for GTPase activity essentially as described previously (Self and Hall, 1995). In brief, subsequent to preloading recombinant fusion proteins with [γ-32P]GTP (Hartmann Analytic, Braunschweig, Germany), radioactivity remaining associated with these proteins was measured after different periods of incubation in assay buffer.

Cell fractionation and immunoblotting

Preparation and fractionation of pollen tube extracts was performed according to the method described by Potocky et al. (2003) with minor modifications. Pollen of 50 flowers was transferred to 10 ml culture medium (Read et al., 1993). After 5 h, pollen tubes were collected by vacuum filtration, ground in liquid nitrogen and resuspended in 600 μl homogenization buffer (0.25 m sucrose, 3 mm EDTA, 5 mm DTT, 70 mm Tris–Mes pH 8.0) containing protease inhibitors (Serva). Extracts were centrifuged sequentially at 3000 g for 5 min, at 10 000 g for 5 min, and at 100 000 g for 60 min to separate cytoplasmic and various membrane fractions. Protein concentrations were determined with Bradford solution (Bio-Rad) to ensure equal loading. Blots were probed with an antibody generated against a GST:Nt-RhoGDI2 fusion protein.

Acknowledgements

The authors would like to thank Katja Piiper for excellent technical support. Funding was received from the German Research Council (DFG; KO 2278) and the state of Baden-Württemberg (Forschungsschwerpunktprogramm).

Ancillary