Association of Arabidopsis type-II ROPs with the plasma membrane requires a conserved C-terminal sequence motif and a proximal polybasic domain

Authors


*(fax +972 3 6406933; e-mail shauly@tauex.tau.ac.il).

Summary

Plant ROPs (or RACs) are soluble Ras-related small GTPases that are attached to cell membranes by virtue of the post-translational lipid modifications of prenylation and S-acylation. ROPs (RACs) are subdivided into two major subgroups called type-I and type-II. Whereas type-I ROPs terminate with a conserved CaaL box and undergo prenylation, type-II ROPs undergo S-acylation on two or three C-terminal cysteines. In the present work we determined the sequence requirement for association of Arabidopsis type-II ROPs with the plasma membrane. We identified a conserved sequence motif, designated the GC-CG box, in which the modified cysteines are flanked by glycines. The GC-CG box cysteines are separated by five to six mostly non-polar residues. Deletion of this sequence or the introduction of mutations that change its nature disrupted the association of ROPs with the membrane. Mutations that changed the GC-CG box glycines to alanines also interfered with membrane association. Deletion of a polybasic domain proximal to the GC-CG box disrupted the plasma membrane association of AtROP10. A green fluorescent protein fusion protein containing the C-terminal 25 residues of AtROP10, including its polybasic domain and GC-CG box, was primarily associated with the plasma membrane but a similar fusion protein lacking the polybasic domain was exclusively localized in the soluble fraction. These data provide evidence for the minimal sequence required for plasma membrane association of type-II ROPs in Arabidopsis and other plant species.

Introduction

Rho GTPases are molecular switches regulating diverse signaling cascades (Etienne-Manneville and Hall, 2002; Hall, 1998). Rhos are subdivided into seven subfamilies designated Rho, Rac, Cdc42, Rnd, Miro, Rho-BTB and a plant-specific group designated ROP or RAC (Christensen et al., 2003; Hall, 1998; Sorokina and Chernoff, 2005; Winge et al., 2000; Yang, 2002). Plant ROPs (RACs) act in multiple signaling pathways, including organization of the actin cytoskeleton, Ca2+, phosphoinositide and reactive oxygen species signaling (Agrawal et al., 2003; Kost et al., 1999; Li et al., 2001; Molendijk et al., 2001; Yang, 2002). ROPs (RACs) are soluble proteins that localize and function at the plasma membrane by virtue of post-translational lipid modifications (Fu et al., 2002; Ivanchenko et al., 2000; Kost et al., 1999; Lavy et al., 2002; Li et al., 1999; Molendijk et al., 2001).

Based on differences in gene structure and the sequences of their C-terminal hypervariable domains, the plant ROPs (RACs) were subdivided into two major subgroups designated type-I and type-II. All type-II ROPs (RACs) have an additional exon at the 3′ end of the gene, probably resulting from insertion of an intron into an ancestral ROP (RAC) (Christensen et al., 2003; Winge et al., 1997). Similar to other members of the Ras superfamily of small GTP-binding proteins, ROPs (RACs) undergo post-translational lipid modifications at their C-terminals. Whereas type-I ROPs (RACs) probably undergo prenylation, type-II ROPs (RACs) undergo S-acylation but not prenylation (Lavy et al., 2002).

The majority of Ras superfamily small GTPases are soluble proteins that act in the plasma and endomembranes (Apolloni et al., 2000; Choy et al., 1999; Michaelson et al., 2001). Membrane attachment of almost all members of the Ras superfamily proteins depends on isoprenylation of C-terminal cysteine residues and concomitant proteolysis and carboxy methylation (Boyartchuk et al., 1997; Bracha et al., 2002; Choy et al., 1999; Clarke, 1992; Hancock et al., 1990; Michaelson et al., 2001; Rodriguez-Concepcion et al., 2000). Prenylation involves the covalent attachment via thioether linkage of either farnesylpyrophosphate (FPP) or geranylgeranylpyrophosphate (GGPP) isoprenoid intermediates. The prenylated cysteines are part of a conserved C-terminal sequence motif designated as the CaaX box or double cysteine motifs in Rab proteins (Maurer-Stroh et al., 2003; Reiss et al., 1991; Schafer et al., 1990; Seabra et al., 1992).

Prenylation promotes attachment of proteins to membranes, but is insufficient to provide stable membrane association. Stabilization of membrane association is achieved by the positive charges of polybasic domains located proximal to the CaaX box in some proteins (Hancock et al., 1990) or an additional lipid modification (S-acylation) of internal cysteine residues in other members of the Ras superfamily (Hancock et al., 1989, 1990; Linder and Deschenes, 2003; Peitzsch et al., 1995; Shahinian and Silvius, 1995). The plant type-II ROPs and the human Cdc42-related Wrch-1 are unique in the Ras superfamily of signaling small GTPases in that they undergo S-acylation without prenylation (Berzat et al., 2005; Lavy et al., 2002).

S-acylation is the attachment of palmitate (C16:0) or other acyl lipids to cysteine residues through a thioester linkage (Linder and Deschenes, 2003, 2004; Smotrys and Linder, 2004). Due to its reversibility, S-acylation has attracted much attention as a mechanism for modulating signaling by regulation of plasma membrane localization, attraction to lipid rafts and protein–protein interactions. Unlike prenylation, no canonical consensus sequences for S-acylation have been identified, S-acylation depending instead on either additional lipid modifications or specific sequences (Smotrys and Linder, 2004).

We have previously demonstrated that plasma membrane attachment of the three Arabidopsis type-II ROPs (RACs), namely AtROP9 (AtRAC7), AtROP10 (AtRAC8) and AtROP11 (AtRAC10), depends on two or more C-terminal cysteines (Lavy et al., 2002). Plasma membrane attachment in all three AtROPs (9, 10 and 11) was sensitive to the S-acylation inhibitor 2-bromopalmitate. In vitro, palmitoylation of AtROP10 (AtRAC8) depended on two C-terminal cysteines and was sensitive to 2-bromopalmitate (Lavy et al., 2002). Similar to Arabidopsis, two maize (Zea mays) type-II ROPs (RACs), ZmROP6 and ZmROP7, which do not terminate with canonical CaaX boxes, were localized to the plasma membrane pending on two or more cysteine residues at their C-terminals (Ivanchenko et al., 2000).

The C-terminal hypervariable domains of type-II ROPs (RACs) from different plant species are structurally conserved (Figure 1). In this study, mutational analysis was used to reveal sequence requirements for plasma membrane attachment of type-II ROPs (RACs), using green fluorescent protein (GFP) fusion proteins. We took advantage of the fact that the GFP moiety does not alter the localization or function of ROPs (RACs).

Figure 1.

 The C-terminal hypervariable domain of type-II ROPs.
(a) A sequence alignment of hypervariable domains of type-II ROPs from different plant species and the Arabidopsis type-I ROP AtROP6 (AtRAC3). The conserved GC and CG pairs are highlighted in black and the polybasic domain is underlined.
(b) An alignment of conserved GC-CG boxes of type-II ROPs. At, Arabidopsis (Arabidopsis thaliana); Os, rice (Oryza sativa); Zm, corn (Zea mays); Nt, tobacco (Nicotiana tabacum); Tv, spiderwort (Tradescantia virginiana); Ga, tree cotton (Gossypium arboreum); Ta, wheat (Triticum aestivum); Hv, barley (Hordeum vulgare). TUC, tentative unique contig sequences [Plant GDB and Christensen et al. (2003)]. Accession numbers for all sequences are given in Experimental procedures.

Results

Sequence conservation of hypervariable domains of plant type-II ROPs

Sequence comparison of the hypervariable domains of type-II ROPs (RACs) from a variety of dicotyledonous and monocot plant species revealed several common features (Figure 1a). Two putatively S-acylated cysteines are flanked by glycine residues in Arabidopsis AtROP9, -10 and -11, rice (Oryza sativa) OsRAC2, maize ZmROP6 and -7, tobacco (Nicotiana tabacum) NtRAC4, spiderwort (Tradescantia virginiana) TvROP1, cotton tree (Gossypium arboreum) and wheat (Triticum aestivum) ROPs (Figure 1a,b). In other type-II ROPs (RACs) there is at least one GC or CG pair (Figure 1a). The upstream GC pair is conserved between type-II and some type-I ROPs (RACs).

The cysteines in the GC-CG sequence motif are separated by five or six residues, referred to as X1–X5/6. Most of the X1--5/6 residues are non-polar and/or hydrophobic (Figure 1a,b). Due to its conservation, we designated the entire GC-X1–X5/6-CG domain as the ‘GC-CG box’.

A subgroup of monocot ROPs (RACs), referred to as non-GC-CG box ROPs, lacks the downstream CG pair. In these ROPs, the sequence adjacent to the GC pair is non-polar, as in the GC-CG box. Two members of this subgroup have an additional cysteine residue (OsROP4, ZmROP5). Other members retained the C-terminal glycine and contain a proximal short basic motif [ZmROP1, ZmROP8, OsRAC3, barley (Hordeum vulgare) and wheat ROPs (RACs)].

OsRAC1 and ZmROP3 are divergent amongst the presented ROPs (RACs), having only the CG pair but lacking the upstream GC pair and with shorter polybasic domains. Notably, the amino acids that are immediately adjacent to the cysteines are aliphatic and aromatic and there are two arginines upstream of the CG pair, which may facilitate interaction with the membrane.

With the exception of AtROP9, type-II ROPs (RACs) do not have a canonical CaaX box motif. Similar to AtROP10, ZmROP1, ZmROP8, OsRAC3, barley and wheat ROPs terminate with CXXX motifs with a lysine at the a2 position, which has been shown to inhibit prenylation. The other ROPs (RACs), have either two or seven amino acids downstream of the most C-terminal cysteine. The polybasic domains are located adjacent to the upstream GC pair.

The sequence conservation of the GC-CG boxes and polybasic domains suggested that they might be required for association of type-II ROPs (RACs) with the plasma membrane.

Subcellular localization of C-terminal sequence deletion mutants

To examine whether an intact GC-CG box is required for membrane association of type-II ROPs (RACs), the X1–X5/6 residues separating cysteines 196 and 203 in AtROP9 and 199 and 205 in AtROP10 were deleted (Figure 2a). In turn, the corresponding GFP fusion proteins were transiently expressed in Nicotiana benthamiana plants (Figure 2c).

Figure 2.

 Subcellular localization of AtROP9, AtROP10 and their C-terminal deletion mutants.
(a) Schematic representations of GFP-AtROP9 and GFP-AtROP10, their GC-CG box sequences and deletion mutants. Numbers denote amino acid positions. Gray shading denotes the polybasic domain and black the GC-CG box.
(b) A schematic representation of a signal cross-section profile. The profiles are measures of signal intensities along lines drawn on images of projection stack images of multiple confocal sections, across cells through the nuclei, using ImageJ software package.
(c) Left panels are projection stacks (projection) and the right panels are cross-section profiles of projection stacks of multiple confocal scans, drawn as described in (b). AtROP10: subcellular localization of wild-type GFP–AtROP10. Note that the signal is only detected at the periphery of cells. The cross-section profile has a baseline close to zero and two narrow peaks formed by the intersection of the drawn line with the cell membrane. Atrop10Δ200--204: subcellular localization of GFP–Atrop10Δ200--204 in which the entire domain between Cys199 and Cys205 was deleted. Note the fluorescence in nuclei and cytoplasmic strands and the fuzzy line at the periphery of the cells. Notably, the cross-section profile has a higher baseline level, a large peak corresponding to the nucleus and many small peaks along the profile, corresponding to the cytoplasmic strands. Atrop9Δ197--202: subcellular localization of GFP–Atrop9mΔ197--201 in which the entire domain between Cys196 and Cys202 of AtROP9 was deleted. The projection stack and cross-section profile are similar to those of Atrop10Δ200--204. GFP: subcellular localization of free GFP. The projection stack and cross-section profiles are similar to the ROP9 and AtROP10 deletion mutants. PM, plasma membrane; CS, cytoplasmic strands; N, nucleus. Bars are 20 μm.
A color version of this figure is available online as Figure S2.

We have previously demonstrated, using confocal imaging, plasmolysis, co-localization with a membrane marker and protein immunoblots of soluble and insoluble cell fractions and of proteins separated by membrane floatation centrifugation, that type-II ROPs (RACs) are localized in the plasma membrane (Bloch et al., 2005; Lavy et al., 2002). Following expression, GFP–AtROP10 (GFP–AtRAC8) was detected as a thin line of fluorescence circumventing the cells (Figure 2c, AtROP10 projection) [the GFP–AtROP9 (GFP–AtRAC7) control was omitted as we have previously demonstrated that it has identical subcellular localization to GFP–AtROP10 (Lavy et al., 2002)]. To assess the subcellular signal distribution, signal intensities along a line crossing the cell's projection stack images (Figure 2b) were determined. The signal cross-section profile of GFP–AtROP10 appeared as two side-peaks (Figure 2c, AtROP10 cross-section profile). For comparison, the fluorescence distribution of free GFP was markedly different (Figure 2c, GFP). In the projection image the signal was detected in the cell periphery as a diffuse rather than a sharp line, in nuclei and in cytoplasmic strands. This distribution was reflected in the cross-section profile. The side-peaks are wider, reflecting the diffuse rather than sharp thin lines at the cell periphery, the wide peak in the middle represents a nucleus and the smaller peaks represent cytoplasmic strands.

The subcellular distribution of GFP–Atrop10Δ200--204 and GFP–Atrop9Δ197--202 was similar to that of GFP. Cells expressing these mutants had fluorescing nuclei, cytoplasmic strands and diffuse rather than sharp and thin lines of fluorescence at the cell periphery (Figure 2c, projection and cross-section profile). These results suggested that the GFP–Atrop9Δ197--202 and GFP–Atrop10Δ200--204 deletion mutants, which contained the GC-CG box cysteines but lacked the X1–X5/6 residues between them, were no longer attached to the plasma membrane. More experiments were carried out to substantiate this finding.

During plasmolysis, the plasma membrane is detached from the cell wall and the cytoplasm shrinks, forming patches. When labeled with a fluorescent probe, such as GFP, plasmolysis causes the plasma membrane to appear as thin fluorescing strands detached from the cell wall and the cytoplasm as small patches of fluorescence. Following plasmolysis, the distribution of GFP–AtROP10 reflected its plasma membrane localization (Figure 3a, AtROP10) (Bloch et al., 2005; Lavy et al., 2002). The fluorescence appeared as thin lines and no patches were apparent (some plasma membrane remained attached to the cell wall even after plasmolysis, which therefore remained weakly labeled with GFP; Figure 3a, arrows). In contrast, many patches and only weakly labeled strands were visible following plasmolysis of cells expressing GFP–AtROP10Δ200--204, GFP–AtROP9Δ197--202 and non-fused GFP (Figure 3a, Atrop10Δ200--204, GFP–Atrop9Δ197--202 and GFP) (some cytoplasm remained pressed against the cell wall after plasmolysis, which, therefore, remained labeled with GFP; Figure 3a, arrows). These results were in agreement with the findings presented in Figure 2 and strongly suggested that the deletion of the GC-CG box X1–X5/6 residues compromised association of AtROP9 and AtROP10 with the plasma membrane.

Figure 3.

 Subcellular localization of AtROP9, AtROP10 and their deletion mutants as determined by plasmolysis and protein immunoblots.
(a) Following plasmolysis GFP–AtROP10 was detected in thin fluorescent strands typical of the plasma membrane (AtROP10). In contrast, GFP–Atrop10Δ200--204 (Atrop10Δ200--204), GFP–Atrop9Δ197--202 (Atrop9Δ197--202) and GFP were mostly detected in patches, typical of the cytoplasm. PM, plasma membrane; C, cytoplasm; arrowheads, cell wall detached membrane and cytoplasm; arrows, cell wall pressed plasma membrane and cytoplasm. Bars are 20 μm.
(b) Immunoblot of protein fractions separated by centrifugation at 100 000 g. Wild-type GFP–AtROP10 was detected in the insoluble pellet (P). In contrast, the GFP–AtROP10Δ200--204 deletion mutant, in which the X1–X5 residues of the GC-CG box were deleted, was only detected in the soluble fraction (S). Numbers at the left of the panels designate molecular mass in kDa.
A color version of this figure is available online as Figure S3.

To further substantiate the results presented in Figures 2 and 3(a), protein extracts were prepared from cells expressing either GFP–AtROP10 or GFP–Atrop10Δ200--204. The extracts were in turn separated into soluble and insoluble fractions (Figure 3b). Following centrifugal separation at 100 000 g, GFP–AtROP10 was only detected in the insoluble pellet, while GFP–Atrop10Δ200--204 was detected exclusively in the soluble fraction (Figure 3b).

The data presented in Figures 2 and 3 have demonstrated that the X1–X5/6 residues between the cysteines in the GC-CG box were required for the association of AtROP9 and AtROP10 with the plasma membrane. We then raised two questions: (i) is the sequence composition of the X1–X5/6 residues important for association of ROPs (RACs) with the plasma membrane or are these amino acids required just for spacing between the cysteines? (ii) Are the glycines important for membrane association? The next series of experiments was carried out to answer these questions.

Amino acid composition of the GC-CG box and association of ROPs with the membrane

Because the results obtained with AtROP10 and AtROP9 were similar, further analysis was continued with AtROP10 alone. To examine the importance of Gly198 and Gly206 for membrane attachment, they were mutated to alanines (Figure 4a), presenting a relatively small change in side group from hydrogen to methyl. Mutating either Gly198 or Gly206 resulted in a slight reduction in membrane association (Figure 4b, Atrop10mA198 and Atrop10mA206). Much of the proteins was still detected as a thin sharp line of fluorescence at the cell periphery. The cross-section profiles of the Atrop10mA198 and Atrop10mA206 mutants depicted the medium sized peaks due to nuclei and very small ‘cytoplasmic’ peaks. The presence of the ‘cytoplasmic’ peaks demonstrated the sensitivity of the cross-section profile method since the signal was barely visible to the eye in the projection stack images. A much bigger change occurred in the Atrop10mA198+206 double mutant, in which both Gly198 and Gly206 were mutated to alanines (Figure 4b, Atrop10mA198+206). The fluorescence from both nuclei and cytoplasm was stronger (Figure 4b,c, Atrop10mA198+206). The signal at the cell periphery, however, still appeared as a thin line of fluorescence and was depicted as sharp peaks in cross-section profiles, suggesting that part of the protein was still associated with the plasma membrane.

Figure 4.

 Sequence requirements for the association of AtROP10 with the plasma membrane.
Point mutations were introduced in the GC-CG box to assess the importance of single residues to plasma membrane attachment. Subcellular localization of GFP fusion proteins was visualized by projection stacks of multiple confocal images (projection) and signal cross-section profiles prepared as described in Figure 2(a).
(a) Schematic representations of GFP–AtROP10 and the point mutations in the GC-CG box. Mutant amino acids are underlined.
(b) Atrop10mA198: a small reduction in plasma membrane localization was detected in GFP–Atrop10mA198 in which Gly198 was mutated into alanine. Note the sharp bands at the cell periphery and the fluorescing nuclei. In the cross-section profile, narrow peaks formed at the intersection of the drawn line with the plasma membrane, a wide peak corresponding to the nucleus and very small peaks corresponding to cytoplasmic strands appeared. The baseline of the graph was higher than for GFP–AtROP10 (Figure 2). Atrop10mA206: GFP–Atrop10mA206 in which Gly206 was changed to alanine shows similar subcellular localization to GFP–Atrop10mA198. The cytoplasmic peaks detected in the cross-section profile are a little bit more pronounced, suggesting a slightly stronger effect of this mutation compared with the Gly198 to alanine mutation. Atrop10mA198+206: highly fluorescing nuclei and pronounced cytoplasmic peaks were detected in GFP–Atrop10mA198+206 in which both Gly198 and Gly206 were mutated to alanines. The fluorescence in the cell periphery appeared as sharp bands in the projection image and as narrow peaks in the cross-section profile, suggesting that some of the protein was still attached to the plasma membrane. Atrop10mP202: GFP–Atrop10mP202 in which Asn202 was mutated to proline was detected in the plasma membrane, nuclei and cytoplasm. Both the projection image and the cross-section profile suggested that this mutation had a stronger effect on plasma membrane localization than the glycine single mutants. PM, plasma membrane; N, nucleus; CS, cytoplasmic strands. Bars are 20 μm.
(c) A quantitative analysis of mutant distribution between the plasma membrane and soluble fractions. Bars represent average ratios between fluorescent intensities in membrane and nuclei as quantified in 50 cells expressing GFP–AtROP10 mutants, as labeled in the figure. The Y-axis scale represents the fluorescence intensity ratio of the membrane/nucleus and is therefore dimensionless. Error bars represent the standard error. Groups were significantly divergent from each other, P ≤ 0.0001 (anova).
A color version of this figure is available online as Figure S4.

Position 202 in AtROP10 is occupied by a conserved asparagine residue (Figure 1). We assumed that Asn202 may play a structural function and therefore mutated it to proline (Figure 4a), a non-polar group that is known to disrupt helical structures. Following expression, GFP–Atrop10mP202 was distributed between the membrane nuclei and cytoplasm (Figure 4b, Atrop10mP202).

Previously, we have demonstrated that mutants of AtROP9, -10 and -11 (AtRAC7, -8 and -10) that could not be modified by lipids accumulated in nuclei as a result of their polybasic domains acting as nuclear localization signals (Lavy et al., 2002). Thus, the ratio of fluorescence measured in membrane versus nuclei presented a quantitative assessment of the degree of plasma membrane association (Figure 4c). Statistical analysis (anova) indicated that the differences between the mutants were significant (P ≤ 0.0001). The quantitative analysis confirmed the results presented in Figure 4(b). The single glycine mutations had a weak effect; with two to two-and-a-half times more protein in the membrane versus the nuclei. Mutating the two glycines strongly disrupted association of their respective mutants with the membrane, and the proteins were almost equally distributed between membrane and nuclei (Figure 4c). Similarly, substitution of Asn202 to proline disrupted plasma membrane association more strongly than the single glycine mutations and in a similar manner to the double glycine mutant.

To substantiate the data presented in Figure 4, the subcellular distribution of the GFP fusion mutant AtROP10 proteins was examined by plasmolysis and immunoblots of soluble and insoluble protein fractions (Figure 5a,b). Following plasmolysis of all mutants, fluorescence was detected as thin membrane strands, cytoplasmic patches and nuclei (Figure 5a). (Even after plasmolysis, some membrane and cytoplasm was pressed against cell walls that therefore remained weakly labeled with GFP.) The fluorescent patches were less apparent in GFP–Atrop10mA198, indicating that this mutation had only a minor effect on association of AtROP10 with the plasma membrane. The protein immunoblots reflected the fluorescent images (Figure 5b): following precipitation at 100 000 g all four mutant proteins were detected in both the soluble and insoluble fractions. In both Atrop10mA198 and Atrop10mA206 single mutants a bigger protein fraction was detected in the insoluble pellet. In contrast, the differences between the soluble and insoluble fractions of GFP–Atrop10mA198+206 and GFP–Atrop10mP202 were less apparent. Interestingly, proteins recovered from soluble fractions migrated faster on SDS gels (Figure 5b), possibly reflecting differences in their lipid modification status.

Figure 5.

 Subcellular localization of AtROP10 C-terminal point mutants as determined by plasmolysis and protein immunoblots.
(a) Plasmolysis of leaves expressing the designated GFP–AtROP10 mutants. In all mutants fluorescence was detected in strands, typical of plasma membrane, and patches, typical of cytoplasm (cell walls were weakly labeled with GFP even after plasmolysis). The cytoplasmic patches detected in the GFP–Atrop10mA198-expressing cells were less pronounced, confirming the weaker effect of this mutation on association with the plasma membrane. Arrowhead, cell wall detached membranes and cytoplasm; arrows, cell wall pressed membranes and cytoplasm. Bars are 20 μm.
(b) Immunoblots of soluble (S) and insoluble (P) protein fractions separated by centrifugation at 100 000 g. Mutant proteins are designated on the right of each panel. The numbers on the left of each panel designate molecular mass in kDa.
A color version of this figure is available online as Figure S5.

The results presented in Figures 4 and 5 indicate that the amino acid composition of the GC-CG box is important for the association of ROPs (RACs) with the plasma membrane. However, in all these point mutants association with the plasma membrane was not completely compromised, as it had been in the deletion mutants. This suggested that either the X1–X5/6 sequence was not altered enough to perturb association of ROPs (RACs) with the plasma membrane or that spacing between the cysteines in the GC-CG box is also a determining factor in the subcellular localization of these proteins.

To discriminate between the two options, mutations were made that maintained the distance of five amino acids between the GC-CG box cysteines but at the same time changed the sequence properties. The X1–X5 residues of the GC-CG box, LSNIL, were changed to REDER to create the Atrop10mREDER200--204 mutant. These mutations changed the nature of the sequence from non-polar to charged (Figure 6a). Following expression, the Atrop10REDER200--204 mutant was dispersed throughout the cytoplasm and nuclei and was not attached to the membrane, as confirmed by the projection cross-section profile, plasmolysis and protein immunoblot (Figure 6b). These results strongly suggested that the sequence composition of the GC-CG box is a determining factor affecting association of ROPs with the plasma membrane.

Figure 6.

 An AtROP10 mutant with a charged GC-CG box is not localized in the plasma membrane.
(a) A schematic representation of the GFP–Atrop10mREDER200--204 mutants in which the X1–X5 residues (LSNIL) in the GC-CG box between Cys199 and Cys206 of AtROP10 were mutated to REDER.
(b) The GFP–Atrop10mREDER200--204 mutant (Atrop10m REDER200--204) was dispersed throughout the cytoplasm and nuclei and was not found in the membrane. Note the cytoplasmic strands and nuclei in the projection stack image (projection) and the peak distribution in the cross-section profile. Patches of GFP fluorescence were detected following plasmolysis. Protein immunoblot (immunoblot) confirmed that Atrop10mREDER200--204 was localized in the soluble supernatant (S) fraction and was absent from the insoluble pellet (P). Bars are 20 μm.
A color version of this figure is available online as Figure S6.

The function of the polybasic and GC-CG box domains

The results presented in Figures 2–6 indicate that the GC-CG box of AtROP10 is essential for its association with the plasma membrane. To examine whether this sequence could act as a plasma membrane targeting/association domain, facilitating S-acylation, it was fused to GFP and the resulting fusion protein, designated GFP–10C198--207 (Figure 7a) was expressed in N. benthamiana plants. Similar to GFP, GFP–10C198--207 was not associated with the membrane but accumulated in the cytoplasm and nuclei. This localization was confirmed by projection stack images and signal cross-section profiles (Figure 7b), plasmolysis (Figure 7c) and protein immunoblots (Figure 7d). Thus, the GC-CG box is required, but is not sufficient for, association of ROPs (RACs) with the plasma membrane.

Figure 7.

 The GC-CG box of AtROP10 is insufficient for association with the plasma membrane.
GFP–10C198--207, in which the last 10 residues of AtROP10 were fused to GFP, was localized in the cytoplasm and nuclei.
(a) A schematic representation of the GFP–10C198--207 fusion protein.
(b) A projection stack of multiple confocal sections (left panel) and a signal intensity cross-section (right panel). Fluorescence was detected in cytoplasmic strands and nuclei.
(c) Following plasmolysis GFP–10C198--207 was detected in nuclei and cytoplasmic patches. N, nucleus; CS, cytoplasmic strands. Bars are 20 μm.
(d) An immunoblot of GFP–10C198--207 protein extracts separated by centrifugation at 100 000 g. The GFP–10C198--207 protein was detected only in the soluble supernatant (S) but not in the insoluble pellet (P). The number on the left denotes molecular mass in kDa.
A color version of this figure is available online as Figure S7.

The question remained: what is the minimal sequence that is required and sufficient for association of AtROP10 [and other type-II ROPs (RACs)] with the plasma membrane? In several proteins, polybasic domains were shown to facilitate interaction with the plasma membrane. The following experiments were designed to assess the importance of the polybasic domain and to identify the minimal sequence required for association of AtROP10 with the plasma membrane (Figure 8).

Figure 8.

 Identification of the AtROP10 plasma membrane association domain.
(a) Schematic representations of GFP–10PBC183--207 in which the AtROP10 polybasic and C-terminal domains starting from amino acid 183 were fused to GFP, and of the GFP–AtROP10Δ183--197 mutant, which lacks the polybasic domain from amino acids 183 to 197. The numbers represent amino acid positions in AtROP10.
(b)–(d) Subcellular localization of GFP–10PBC183--207 and GFP–Atrop10Δ183--197. The GFP–Atrop10Δ183--197 mutant was localized exclusively in the soluble fraction, while GFP–10PBC183--207 was primarily localized in the membrane.
(b) Projection stacks of multiple confocal sections and signal cross-section profiles.
(c) Localization following plasmolysis. N, nucleus; PM, plasma membrane; CS, cytoplasmic strands; arrowheads, cell wall detached membrane and cytoplasm; arrows, cell wall pressed plasma membrane and cytoplasm. Bars are 20 μm.
(d) A protein immunoblot of soluble (S) and insoluble (P) fractions following separation by centrifugation at 100 000 g. Numbers at the sides denote the molecular mass in kDa.
A color version of this figure is available online as Figure S8.

Residues 183–197, comprising the polybasic domain, were deleted to create GFP–AtROP10Δ183--197 (Figure 8a). Following expression in N. benthamina leaf epidermal cells, GFP–AtROP10Δ183--197 was localized in cytoplasm and nuclei (Figure 8b–d), similar to GFP. Fluorescing nuclei and cytoplasmic strands were detected in projection stacks and signal cross-section profiles (Figure 8b) and cytoplasmic patches were detected following plasmolysis (Figure 8c). In protein immunoblots, GFP–AtROP10Δ183--197 was detected in the soluble fraction but not in the insoluble pellet (Figure 8d). These data indicate that the polybasic domain of AtROP10 is required for association of this protein with the membrane. The results also suggest that GFP–10C197--207 is not associated with the plasma membrane because it lacks a polybasic domain.

To determine whether the polybasic domain together with the GC-CG box comprise a minimal plasma membrane association domain, residues 183 to 207 of AtROP10 were fused to GFP to create GFP–10PBC183--207 (Figure 8a). Following its expression in N. benthamiana, GFP–10PBC183--207 was primarily localized in the plasma membrane and only a small fraction was detected in nuclei (Figure 8b,c). In protein immunoblots, the majority of the protein was detected in the insoluble pellet fraction and a much fainter band was detected in the 100 000 g soluble fraction (Figure 8d). We suspect that the small fraction of protein detected in nuclei resulted from the relatively small size of the fusion protein, or it may reflect higher expression levels. The results in Figures 2–8 demonstrate that the polybasic domain together with the GC-CG box are required and sufficient for association of AtROP10, and most probably other type-II ROPs (RACs), with the plasma membrane.

Discussion

In a previous work we demonstrated that Arabidopsis type-II ROPs (RACs) are attached to the plasma membrane through S-acylation of two or more cysteine residues at their C-terminal hypervariable domains (Lavy et al., 2002).

In this work, the analysis of requirements for the association of type-II ROPs (RACs) with the plasma membrane has been extended to include sequences proximal to the S-acylated cysteines. Sequence comparison of type-II ROPs (RACs) from monocots and dicotyledonous plants revealed a conserved non-polar domain located next to the putatively modified cysteines (Figure 1). Deletion of this conserved non-polar domain or substitution of its non-polar residues with polar amino acids completely compromised association of the ROP (RAC) with the plasma membrane (Figures 2–6). In addition, substitution of two conserved glycine residues adjacent to the modified cysteines or introduction of the helix-breaking amino acid proline partially disrupted membrane association (Figures 4 and 5). The GFP–10C198--207 fusion protein, harboring the last 10 residues of AtROP10, remained in the soluble fraction, indicating that this C-terminal domain is required for, but is not sufficient to promote, association with the plasma membrane. However, GFP–PBC10C183--207 fusion protein, that includes the polybasic domain in addition to the last 10 residues of AtROP10, was primarily localized in the plasma membrane. In agreement, the GFP–AtROP10Δ183--197 mutant that lacks the polybasic domain was exclusively detected in the soluble fraction (Figure 8). These data identified the association domain of AtROP10, and most probably other type-II ROPs (RACs), with the plasma membrane. In agreement, a fusion protein comprising the C-terminal 30 amino acids of the maize type-II ROP, ZmROP7, fused to GFP was also primarily localized to the plasma membrane (Ivanchenko et al., 2000), indicating that the results obtained with AtROP10 could be extended to other plant type-II ROPs.

The structure of the GC-CG box

Our results suggest that there may be some structural requirements for S-acylation and membrane association of type-II ROPs (RACs). The secondary structure of the GC-CG domain could not be predicted by standard algorithms. However, the enrichment in non-polar and hydrophobic residues and the effect of substituting the asparagine to proline suggest that the X1–X5/6 sequence between the cysteines may form an α-helical domain. Both sequence and mutational analyses showed that the glycine residues play and important role in membrane association of type-II ROPs (RACs), possibly by facilitating the S-acylation. Glycine is the amino acid with the smallest side group, suggesting that there might be a specific structure required around the cysteines to facilitate S-acylation and membrane association of type-II ROPs (RACs).

Membrane association of plant ROPs (RACs)

Most if not all plant ROPs (RACs) act at the plasma membrane (Bloch et al., 2005; Lavy et al., 2002; Yang, 2002). The association of ROPs (RACs) with the membrane occurs via two mechanisms. In vitro prenylation assays suggest that type-I ROPs (RACs) are prenylated, preferentially by protein geranylgeranyltransferase-I (PGGT-I ), with affinities at the range of 10−7 m (N. Sorek and SY, unpublished data). Following prenylation, proteins undergo CaaX proteolysis and methylation in the endomembrane system (Bracha et al., 2002; Rodriguez-Concepcion et al., 2000). It is still unknown whether CaaX processing is required for targeting to the plasma membrane and whether type-I ROPs are modified by additional lipid modifications.

Many type-II ROPs (RACs) lack C-terminal CXXX motifs (Figure 1), and in those that have it the CXXX is not a functional prenylation substrate (Lavy et al., 2002). Furthermore, the polybasic domains in type-II ROPs (RACs) are localized at a distance from the CXXX motif, and are therefore unlikely to promote prenylation. The accumulation of GFP–Atrop10mREDER200--204 in nuclei and cytoplasm (Figure 6b) exemplifies this principle. In this mutant, the REDER sequence is localized right next to Cys205, which is part of a C-terminal CGKN sequence motif. The CGKN sequence is not a prenylation CaaX box and the substitution of the LSNIL non-polar residues with REDER creates a polar domain that interferes with an S-acylation-dependent membrane attachment mechanism. Interestingly, AtROP10 mutant plants are ABA hypersensitive (Zheng et al., 2002). The prenylation-independent membrane localization of AtROP10 indicates that its function is probably not affected in the ABA-hypersensitive farnesyltransferase β-subunit mutant era1 (Cutler et al., 1996).

Thus, association of ROPs (RACs) with the plasma membrane reflects differential affinities toward different systems that are modulated by the combinatorial structure of the hypervariable domains. In type-I ROPs (RACs), CaaL boxes and adjacent polybasic domains probably promote prenylation. In type-II ROPs (RACs), cysteines, adjacent glycines, a non-polar domain (the GC-CG box) and an upstream polybasic domain promote membrane association by a different mechanism that involves S-acylation.

It is currently unknown where in the cell S-acylation of type-II ROPs (RACs) takes place. Following either transient or stable transformation, type-II ROPs (RACs) were only found in the plasma membrane (Bloch et al., 2005; Ivanchenko et al., 2000; Lavy et al., 2002). GFP–AtROP11 (GFP–AtRAC10) was not detected in intracellular vesicles and endomembranes following treatment of GFP–AtROP11 (GFP–AtRAC10) transgenic plants with the secretory pathway inhibitor brefeldin A (BFA) or the actin-disrupting drugs latrunculin b or cytochalsin D that inhibit vesicle trafficking (Bloch et al., 2005). By themselves, these data do not prove that type-II ROPs (RACs) are not targeted through, or at least modified in, the endomembrane system, since ROP (RAC) activity disrupts membrane vesicle cycling (Bloch et al., 2005). Future mutant analysis and identification of ROP (RAC) S-acyl transferases should assist in elucidating subcellular targeting of type-II ROPs (RACs). The GFP–10PBC183--207 fusion protein, developed in this study, could be used in the future as a membrane marker to study the mechanisms of targeting and membrane association of type-II ROPs (RACs) independently of their effects on the cell.

Conservation of the plasma membrane association domain in type-II ROPs

The structural conservation of the C-terminal domain in type-II ROPs (RACs) may imply a conserved membrane association mechanism, unique to plants. Interestingly, all the dicotyledonous type-II ROPs (RACs) with known sequences belong to the GC-CG box subgroup, whereas in monocot grasses there are both GC-CG box and non-GC-CG box type-II ROPs (RACs) (Figure 1). It is tempting to speculate that the GC-CG box proteins represent the common ancestors of type-II ROPs (RACs) and that the non-GC-CG box subgroup have diverged from it during evolution in some plant species. Indeed, the majority of ROPs (RACs) in monocots are type-II, suggesting greater evolution of this subclass in this group of plants. In contrast, in dicotyledonous species the majority of ROPs (RACs) are type-I. More sequence information would be required to validate this assumption.

Experimental procedures

Site-directed mutagenesis and plasmid construction

All plasmids are listed in Table S1. Plasmid numbers refer to the numbering in Table S1. Cloning of AtROP10 (AtRAC8) and AtROP9 (AtRAC7) has been described previously (Lavy et al., 2002). Mutant GFP fusion proteins were prepared using either direct PCR (method 1) or with a QuikChange mutagenesis kit (Stratagene, La Jolla, CA, USA) (method 2). The DNA templates used for mutagenesis are specified in Table S3. When using method 1 (direct PCR), fragments containing mutant genes were subcloned into pGEM (Promega, Madison, WI, USA). In turn, mutant pGEM–AtROP plasmids were digested with SacI and subcloned in frame into pGFP (Table S1). For expression in plants, pGFP–AtROP plasmids were digested using HindIII. The resulting cassettes, containing a cauliflower mosaic virus (CaMV) 35S promoter, the gene of interest and a nitric oxide synthase (NOS) transcriptional terminator, were subcloned into pCAMBIA 2300. Primers are listed in Table S2. The primer combinations and PCR method used for mutagenesis are listed in Table S3. All clones were sequenced to verify that no PCR-generated errors had been introduced. (Plasmid pGFP-MRC was previously described in Rodriguez-Concepcion et al., 1999).

Plant transformation

Agrobacterium tumefaciens-mediated transformation of N. benthamiana plants was carried out as previously described (Bracha et al., 2002; Lavy et al., 2002).

Plasmolysis

Plasmolysis was carried out by incubating leaves in 0.8 m NaCl for 5–10 min. In turn, leaf sections were mounted on microscope slides in the plasmolysis solution.

Protein immunoblots

Protein extract from GFP–AtROP10 wild-type and mutant plants was prepared by homogenization in buffer 1 [50 mm HEPES-KOH (pH 7.5), 10% sucrose, 50 mm NaCl, 5 mm MgCl2, 1 mm 2-mercaptoethanol, plant protease inhibitor mix, 2 mm phenylmethanesulfonyl flouride (PMSF)]. To precipitate insoluble material, extracts were centrifuged for 10 min at 4°C at 25 000 g using a Sigma (Sigma, Buckinghamshire, UK) 4K15 centrifuge with a 12130-H rotor. The resulting supernatant was collected and centrifuged again at 100 000 g for 1 h at 4°C using Beckman (Beckman Coulter, Fullerton, CA) ultracentrifuge with TLA-100 rotor. The resulting pellet was incubated at 4°C for 30 min in Buffer 2 [50 mm HEPES-KOH (pH 7.5), 10% sucrose, 50 mm NaCl, 5 mm MgCl2, 1 mm 2-mercaptoethanol, 0.5% SDS, 1% Triton X-100, plant protease inhibitor mix, 2 mm PMSF]. The extract was centrifuged at 25 000 g for 15 min at 4°C using a Sigma 4K15 centrifuge with a 12130-H rotor. Supernatant was collected for further analyses. Proteins were resolved by SDS-PAGE and transferred onto nitrocellulose membranes (Schleicher and Schuell, Dassel, Germany). Membranes were incubated with mouse anti-GFP monoclonal antibodies (Covance, catalog no. MMS-118P, 1:8000, Babco, Berkeley, CA, USA) and in turn were washed and incubated with blotting-grade goat anti-mouse horseradish peroxidase-conjugated secondary antibodies (Bio-Rad, Hercules, CA, USA). Detection was performed using EZ-ECL (Biological Industries, Bet Haemek, Israel).

Confocal imaging

Confocal imaging was performed using a Leica TCS-SL confocal laser scanning confocal microscope (Leica, Wetzlar, Germany). Excitation was performed with an argon laser 488 nm channel using a 500 nm beam splitter. Emission was detected with spectral detector set between 510 and 550 nm. Image analysis was performed with a TCS (Leica), LSM Image browser (Zeiss, Jena, Germany), ImageJ (http://rsb.info.nih.gov/ij/) and Photoshop 7.0 (Adobe, Mountain View, CA, USA). The images shown in Figures 2–8 are of representative cells. All experiments were repeated at least three times, each time with different plants. In each experiment large population of cells were examined. Fluorescence emission spectra were determined using the λ-scan mode of the confocal laser scanning microscope. The spectra were compared with the emission spectra of GFP to ensure that measured signals originated from GFP and not from autofluorescence of the tissue (Figure S1).

Quantification of fluorescence intensity

To avoid saturation, images were prepared using low laser intensities and detector gains. The average fluorescence intensities of membranes and nuclei were measured with ImageJ software using the following methodology: for averaging signal intensity per pixel on the membrane, a line was drawn on in-focus membrane sections (on average 80 pixels). Similarly, average signal intensities in nuclei were measured by drawing a circle around an in-focus nuclear region. To calculate the relative signal intensities in the membrane versus the nucleus for each cell, the average signal intensities were divided by one another. A total of 50 cells were sampled for each mutant. All experiments were repeated at least three times, each time with different plants.

Signal cross-section plots

Signal cross-sections of projection stack images were made with ImageJ software. Lines were drawn through cells and pixel intensities along these lines were displayed as a two-dimensional graph. The X-axis represents distances along the line. The Y-axis is a 256 gray scale of pixel intensities. The plots shown in Figures 2, 4 and 6–8 are of representative cells.

Miscellaneous: All chemicals were purchased from Sigma (St Louis, MO, USA) unless otherwise specified.

All chemicals were purchased from Sigma (St Louis, MO, USA) unless otherwise specified.

Acknowledgements

This research was supported by ISF grant no. 399/03 and the Teva Corporation research prize to S.Y. M.L. was a recipient of an Israel Ministry of Science Eshkol fellowship for doctoral students.

Gene accession numbers: AtROP6-At4g35020; AtROP9-At4g28950; AtROP10-At3g48040; AtROP11-At5g62880; NtRAC4-AJ496228; TvROP1-AF239751; OsRAC1-AB029508; OsRAC2-AB029509; OsRAC3-AB029510; OsROP4-AF380335; ZmROP1-AF126052; ZmROP3-AF126054; ZmROP5-AJ278666; ZmROP6-AJ278665; ZmROP7-AJ278664; ZmROP8-AF376054; Tree cotton-GaTUC650; wheat-TaTUC20419; wheat-TaTUC4441; Barley-HvTUC10103.

Ancillary