Formation of self-organizing functionally distinct Rho of plants domains involves a reduced mobile population

Abstract

Rho family proteins are central to the regulation of cell polarity in eukaryotes. Rho of Plants-Guanyl nucleotide Exchange Factor (ROPGEF) can form self-organizing polar domains following co-expression with an Rho of Plants (ROP) and an ROP GTPase-Activating Protein (ROPGAP). Localization of ROPs in these domains has not been demonstrated, and the mechanisms underlying domain formation and function are not well understood. Here we show that six different ROPs form self-organizing domains when co-expressed with ROPGEF3 and GAP1 in Nicotiana benthamiana or Arabidopsis (Arabidopsis thaliana). Domain formation was associated with ROP–ROPGEF3 association, reduced ROP mobility, as revealed by time-lapse imaging and Fluorescence Recovery After Photobleaching beam size analysis, and was independent of Rho GTP Dissociation Inhibitor mediated recycling. The domain formation depended on the ROPs’ activation/inactivation cycles and interaction with anionic lipids via a C-terminal polybasic domain. Coexpression with the microtubule-associated protein ROP effector INTERACTOR OF CONSTITUTIVELY ACTIVE ROP 1 (ICR1) revealed differential function of the ROP domains in the ability to recruit ICR1. Taken together, the results reveal mechanisms underlying self-organizing ROP domain formation and function.

Plasma membrane self-organizing polarity domains of small GTP-binding proteins form upon their co-expression together with their activator and suppressor due to restriction of protein mobility.

Introduction

Cell polarity is a basic entity of living organisms, which affects their development and physiological function. In plants, cell polarization regulates the localization and spatial orientation of cell division and morphogenesis. Cell shape and orientation of cell divisions regulate organ shape and function. Hence, an analysis of cell polarity is fundamentally important both in terms of generating basic theory and understanding of how plants develop and respond to the environment, as well as in terms of applied aspects that aim to engineer polarity to control important traits (Wallner, 2020).

Central to the regulation of cell polarity are Rho of Plants (ROPs) small GTP-binding proteins (Feiguelman et al., 2018). ROPs cycle between GTP-bound “ON” and GDP-bound “OFF” states. In the GTP-bound state ROPs interact with effectors to initiate signaling (Feiguelman et al., 2018). Accumulation of active ROPs at the tip of root hairs (RHs), pollen tubes, trichomes, and differentiating metaxylem (MX) cells have been associated with specific spatial organization of the actin and microtubules (MTs) and is required for proper development and polar growth of these cells and organs (Fu et al., 2001; Jones et al., 2002; Gu et al., 2005; Hwang et al., 2010; Yanagisawa et al., 2015, 2018; Kang et al., 2017; Nakamura et al., 2017; Hirano et al., 2018). In leaf epidermis pavement cells, ROP activation status has been implicated in regulating actin dynamics and MTs orientation pattern at the cell wall; that pattern produces the polarized lobes and indentations (Fu et al., 2002, 2005, 2009; Lin et al., 2013; Majda et al., 2017). Indeed, several identified ROP effectors are actin and MTs binding proteins that influence growth. However, the control mechanisms are still poorly understood (Fu et al., 2005; Gu et al., 2005; Lavy et al., 2007; Uhrig et al., 2007; Basu et al., 2008; Mucha et al., 2010; Oda and Fukuda, 2012a; Lin et al., 2013; Stephan et al., 2014; Kang et al., 2017; Li et al., 2017; Feiguelman et al., 2018; Sugiyama et al., 2019).

Based on their amino acid sequences, ROPs have been divided into two major subgroups, designated types I and II (Winge et al., 2000). All type-I ROPs terminate with a canonical prenylation CaaX box motif in which the X residue is always a Leu and are preferentially geranylgeranylated (Sorek et al., 2011). Type-II ROPs terminate with a domain called the GC-CG box and are S-acylated but not prenylated (Lavy et al., 2002; Lavy and Yalovsky, 2006). Similar to Rho GTPases from yeast and animals, ROPs contain a Lys and Arg enriched domain proximal to the CaaX or GC-CG motifs, referred to as the Poly-Basic Region (PBR). Experiments in animal cells and plants showed that the PBRs promotes interaction with anionic lipids including phosphatidyl phosphoinositide phosphates (PIPs/PtdInsPs) and phosphatidyl serine (PS), stabilize the interaction of Rhos with the plasma membrane and their clustering into nanodomains (Heo et al., 2006; Lavy and Yalovsky, 2006; Remorino et al., 2017; Sartorel et al., 2018; Platre et al., 2019). It is highly likely that in type-I ROPs the PBRs also enhance prenylation by protein geranylgeranyl transferase-I (PGGT-I), given the structural and functional conservation of the plant PGGT-I (James et al., 1995; Caldelari et al., 2001).

ROPs are regulated by three major groups of proteins: Guanyl nucleotide Exchange Factors (GEFs; ROP-GEFs) that facilitate dissociation of GDP and enable rebinding of GTP; ROP-GTPase-Activating Proteins (GAP; ROP-GAPs), which enhance GTP hydrolysis and inactivation, and RhoGDIs (GDP Dissociation Inhibitors) that are required to cycle inactive and active ROPs from and to the plasma membrane and possibly to stabilize ROP in the cytosol (Feiguelman et al., 2018). Intriguingly, the ROP regulators have been shown to self-organize active ROP into specialized organelle domains at the plasma membrane and this organization translates into the spatial control of the cytoskeleton (Oda and Fukuda, 2012a; Yanagisawa et al., 2018; Denninger et al., 2019; Feiguelman and Yalovsky, 2020; Kulich et al., 2020).

There are two known families of ROP-GEFs in plants: a plant unique group called ROPGEFs that contain a catalytic Plant-specific ROP Nucleotide Exchanger (PRONE) domain that comprises a 14-member protein family in Arabidopsis (Arabidopsis thaliana; Berken et al., 2005) and a Dock180-related ROP-GEF, of which the sole representative in Arabidopsis is SPIKE1 (SPK1) (Basu et al., 2008). ROPGEFs were shown to function in RHs and pollen tubes tip and during pit formation in MX, implicating their function in tip growth and local polarization processes in differentiating cells (Kaothien et al., 2005; Gu et al., 2006; Zhang and Mccormick, 2007; Duan et al., 2010; Oda and Fukuda, 2012a; Chang et al., 2013).

GAPs accelerate the nucleotide hydrolysis rate of GTP-bound small GTPases and in combination with GEFs can promote small GTPase cycling between active and inactive states. GAPs are often considered as negative regulators of signaling pathways; however, any pathway that requires local and/or rapid small GTPase cycling requires GAP function. GAPs promote active domain formation in the plant cortex (Oda and Fukuda, 2012a). Recently, Armadillo Repeat Only (ARO) proteins have shown to interact with GAPs and ROPs and delimit active ROP domains in RH and trichomes (Feiguelman and Yalovsky, 2020; Kulich et al., 2020).

Remarkably, transient co-expression of a trio of ROP-GTPase, ROPGAP, and a PRONE-domain ROPGEF in Nicotiana benthamiana leaf epidermal cell results in spontaneous formation of domains where the GEF and likely active ROP are clustered (Oda and Fukuda, 2012a). The size and spacing of the active ROP domains are nonrandom, and concentration gradients of the GEF and GAP exist within the domains (Nagashima et al., 2018). As yet, however, only GEF but not ROPs was detected in the domains. Furthermore, the molecular basis for domain formation is not well understood.

Here we show that six different types I and II ROPs cluster in activity domains when coexpressed with a GEF and a GAP. Clustering in the domains is characterized by a low mobility fraction of the ROP, depends on activation/inactivation cycles and interaction with anionic lipids, and is independent of RhoGDI-mediated membrane dissociation. Within the domains, ROPs display differential ability to recruit the ROP effector Interactor of Constitutively active ROP 1 (ICR1), indicating that the formation of synthetic ROP domains provides a platform for analyzing ROP signaling as a well as for engineering cell morphogenesis.

Results

Formation of domains by different ROPs

For examining domain formation by different ROPs, three representative type-I ROPs: ROP2, ROP4, and ROP6 and the three type-II ROPs: ROP9, ROP10, and ROP11 were transiently expressed as green fluorescent protein (GFP) fusion proteins together with mCherry-tagged ROPGEF3 PRONE domain (mChGEF3p) and untagged ROPGAP1 in N. benthamiana leaf epidermis pavement cells (Figure 1). The localization and function of the same or similar ROP, GEF, and GAP fusion proteins have previously been validated using immuno-localization, biochemical, and genetic approaches (Molendijk et al., 2001; Fu et al., 2002, 2005; Jones et al., 2002; Lavy et al., 2002, 2007; Bloch et al., 2005; Carol et al., 2005; Gu et al., 2006; Klahre and Kost, 2006; Klahre et al., 2006; Lavy and Yalovsky, 2006; Duan et al., 2010; Sorek et al., 2010, 2011, 2017; Bloch et al., 2011; Oda and Fukuda, 2012a, 2013; Poraty-Gavra et al., 2013; Sun et al., 2015; Li et al., 2017; Hirano et al., 2018; Nagashima et al., 2018; Denninger et al., 2019; Platre et al., 2019, Kulich et al., 2020) and hence they can serve as reliable markers. Plasma membrane-localized domains in which each of the six ROPs was colocalized with mChGEF3p were observed with each one of the six ROPs (Figure 1, A–F). These data confirmed the earlier hypothesis that ROPs reside in the domains together with ROPGEFs (Oda and Fukuda, 2012a), and further showed that domain formation is not a unique feature of ROP11. Higher magnification images (Figure 1, G–L) showed that the domain is composed of smaller nanodomains. Quantification of the domain numbers showed that domain density was variable between experiments and statistically significant differences could be observed among ROP2, ROP4 and ROP6, and ROP10 (Figure 1M) (P < 0.05, analysis of variance (ANOVA) and Tukey Kramer honestly significant difference [HSD]). Measurements of domain and nanodomains sizes showed that domain vary in sizes between ∼10 and 50 µm² (Figure 1N), while nanodomains are an order of magnitude smaller with sizes mostly ˂1 µm² (Figure 1O). Colocalization analysis using both average overlap coefficient and Pearson correlation demonstrated that all the ROPs were colocalized with mChGEF3p in the domains (Figure 2, A–F; Tables 1;Supplemental Table S1). Similar colocalization analysis between GFPROP11 and mCherryGAP1 (mChGAP1) showed overlap coefficients ranging between 0.85 and 0.92 but negative Pearson correlation coefficient (R) values, indicating that mChGAP1 was localized outside the domains (Figure 2G;Supplemental Table S2). Importantly, the imaging of the GFPROPs and mChGEF3p was by single confocal scans, while the signal of mChGAP1 was only visible in maximum intensity scans of multiple confocal scans. Hence, the high overlap coefficients of GFPROP11 and mChGAP1 resulted from overlap between the GFP and mCherry signals from different confocal planes while the negative R value indicates that the proteins were not colocalized. Plasma membrane localization of the ROP–ROPGEF3p domains was examined by plasmolysis, which induces shrinkage of the cytoplasm and vacuole and detachment of the plasma membrane from the cell wall. Following plasmolysis, GFP:ROP2 and mChGEF3p domains were still observed, as well as ROP2-associated Hechtian strands which connect between the plasma membrane and the cell wall (Supplemental Figure S1A). Panel B of Supplemental Figure S1 is a schematic representation highlighting that imaging of the domains were carried out by single confocal scans taken from the upper surfaces of the cells (represented by a rectangle in the figure). The domains were either rounded, ring-shaped, or elongated (Supplemental Figure S2) and for each of the ROPs the number of domains with each shape was variable and different between cells and experiments (Supplemental Table S3).

Figure 1.

Formation of domains by ROPs and ROPGEF3 in N. benthamiana leaf epidermal cells. GFP-tagged ROP2 (A), ROP4 (B), ROP6 (C), ROP9 (D), ROP10 (E), and ROP11 (F) formed domains following transient expression with mCh-GEF3p when coexpressed with un-tagged GAP1 in N. benthamiana leaf epidermal cells. O/L, overlay of GFP and mCherry signals. G–L, High magnification images of domains: ROP2 (G), ROP4 (H), ROP6 (I), ROP9 (J), ROP10 (K), ROP11 (L). Representative domains are enclosed with circle, arrows designate nanodomains. All images are single confocal scan of the cell surface (see Supplemental Figure S1). Bars, 10 µm for (A)–(F), 1 µm for (G)–(L). M, Quantification of domain density in 50 µm² area. N and O, Domain and nanodomains areas in (µm)². Box plots represent the upper and lower quartile, whiskers are upper and lower extremes. The line in the boxes is the median. Letters denote significant differences between density and domain and nanodomains areas (P < 0.05, one-way ANOVA and Tukey’s HSD test). n = 10 in (M)–(O), respectively.

Figure 2.

Colocalization analysis of GFPROPs with mCherry:GEF3p or mCherryGAP1 in ROP domains. A–F, Colocalization of GFPROPs with mCherryGEF3p coexpressed with GAP1. Representative confocal images and scatter diagrams of each ROP (Green channel) and GEF3p (Cyan channel). G, Colocalization of GFP:ROP11 with mCherry:GAP1 expressed with GEF3p. O/L, overlay of GFP and mCherry signal. Bar is 10 µm for all panels. Red circles mark regions analyzed for colocalization. H, Scatter plots scale colors bar.

Table 1.

Average colocalization of ROPs and GEF3p in the domains

ROP Protein	Average Overlap Coefficient	Average Correlation R
ROP2	0.93	0.81
ROP4	0.89	0.77
ROP6	0.85	0.70
ROP9	0.82	0.74
ROP10	0.83	0.75
ROP11	0.91	0.80

Similar to the expression in the heterologous N. benthamiana system, transient expression of GFP fusion proteins of either ROP4, ROP6, ROP9, ROP10, and ROP11 together with mChGEF3p and GAP1 in Arabidopsis cotyledons resulted in the formation of plasma membrane domains in which either of the ROPs were colocalized with mcChGE3p (Figure 3). Similar to N. benthamiana (Figure 1, G–L, N, and O), the domains formed by either ROP4, ROP10, or ROP11, and ROPGEF3p were composed of clustered nanodomains (Figure 3, A, D, and E). This composition of the domains suggested they involve clustering at two levels: a molecular level at which ROP and ROPGEF_PRONE molecules cluster to generate micron-sized domains (nanodomains) and a higher-order level at which the individual nanodomains cluster to generate a domain. Taken together, the data show that both types I and II ROPs can generate domains together with ROPGEF3_PRONE when coexpressed together with GAP1. The similarity between the results obtained with N. benthamiana leaf epidermis pavement cells or Arabidopsis cotyledon indicated that either system could be used for the analysis of domain formation. Yet, since accumulation of ROPs in domains in a developmental context has not been shown it was important to verify that it does indeed take place.

Figure 3.

Formation of domains by ROPs and ROPGEF3 in Arabidopsis cotyledons. GFP-tagged ROP4 (A), ROP6 (B), ROP9 (C), ROP10 (D), and ROP11 (E) formed domains following transient expression with mChGEF3p when coexpressed with un-tagged GAP1 in Arabidopsis cotyledon epidermis. O/L, overlay of GFP and mCherry signals. Bars, 10 µm for all panels.

Accumulation of GFP:ROP11 in domains during tracheary elements differentiation

To visualize GFP:ROP11 in a developmental context we utilized in Vascular Cell Induction Culture System Using Arabidopsis Leaves (VISUAL) system in which cotyledon mesophyll cells are induced to dedifferentiate into tracheary elements using the Glycogen Synthase Kinase 3 inhibitor bikinin (Kondo et al., 2016). GFP:ROP11 expression was induced by estradiol using the synthetic transcription factor XVE (Zuo et al., 2000; Siligato et al., 2016). By placing XVE under regulation of the ROP11 promoter (ROP11p), GFP:ROP11 expression was induced only in cells and tissues where it is expressed in a developmental context. By using the estradiol–XVE induction system, both expression timing and levels could be controlled, avoiding potential adverse effects. The secondary cell wall (SCW) organization in the dedifferentiating tracheary elements was intermediary between the helical organization characteristic of protoxylem and the pitted arrangement of MX with rounded and elongated pits (Figure 4, A and B; red propidium iodide label). GFP:ROP11 (Figure 4, A–B, green label) accumulated in pitted areas between the SCW. Time-lapse imaging showed that GFP:ROP11 accumulated in domain-like forms which remained immobile (Figure 4C, arrowheads). The distribution of GFP:ROP11 in domains, when expressed under the ROP11 promoter in a developmental context, resembled the domain formation upon transient co-expression of ROPs with ROPGEF3_PRONE and ROPGAP1 (Figures 1–3), indicating that formation of the domain could be related to ROP immobility in these membrane regions. Furthermore, The GFPROP11 domains were composed of smaller nanodomains (Figure 4C, T = 0 the circle and T = 5, T = 10, and T = 20 the arrows).

Figure 4.

Distribution of GFP:ROP11 in dedifferentiating Arabidopsis tracheary elements. Expression of proROP11::GFP:ROP11 was induced by estradiol in bikinin-induced dedifferentiating tracheary elements. A, Group of differentiating cells highlighting the structure of the SCW pits. B, A whole cell. C, A 25 min time-lapse image series. T, time in minutes. Note the constant localization (immobility) of GFP:ROP11 (arrowheads). The circle surrounds a domain, and the arrow points to a nanodomain, respectively (C, T = 0 and T = 5, T = 10, and T = 20). Red, PI fluorescence; green, GFP fluorescence. Bars 10 µm for all panels.

ROP cycling between active and inactive states and their effects on domain formation

The ROP domains appeared as clustered nanodomains of ROPs and GEF3/GEF3p (Figures 1–4). Next, we examined the distribution of singly expressed GEF3p and of ROP11 coexpressed with GEF3p in the absence of GAP1 (Figure 5). Remarkably, confocal scans taken from the upper surface of the cells (Supplemental Figure S1B) revealed that mChGEF3p was localized in the plasma membrane in multiple dispersed nanodomains (Figure 5A). Following coexpression of GFP:ROP11 with mChGEF3p, the two proteins were colocalized and the nanodomains were more pronounced (Figure 5B). Quantification of nanodomain number in a 25 µm² area showed their number was not significantly different in the presence or absence or ROP11 (Figure 5C), suggesting that the ROP11-GEF3p nanodomains form by association of ROP11 with the GEF3p nanodomains. The size of the nanodomains was smaller than 1 µm² (Figure 5D), similar to the nanodomains which were part of ROP–ROPGEF3p domains when co-expressed with GAP1 (Figures 1–4). Similar results were observed upon coexpression of mChGEF3p with GFP:ROP6 but not with GAP1 (Supplemental Figures S3 and S4). Cell surface and mid cell confocal images showed that when either GAP1, ROP6, or ROP11 were expressed alone or coexpressed together, no nanodomains could be observed (Supplemental Figure S4). These results suggest that the domains are initiated by association of ROPs with GEF3 nanodomains and that GAPs, by suppressing ROP function, delimit the distribution of the nanodomains, resulting in the formation of discrete domains. The proposed function of GAPs in delimiting the ROP-GEF nanodomains predicts that ROP cycling between active and inactive states would be critical for the formation of domains.

Figure 5.

GEF3p forms nanodomains which are enhanced in the presence of ROP11. A, When expressed by itself mCherryGEF3p (mChGEF3p) formed nanodomains in the plasma membrane. B, Clusters on mCherryGEF3p and wild-type GFPROP11 plasma membrane nanodomains. The scale bar is 10 µm for all panels. C, Quantification of nanodomains density in 10 µm² area formed by expressing mCherryGEF3p alone or coexpressed with GFPROP11. D, Coexpressed GEF3P ROP11 and singly expressed GEF3p nanodomains areas µm². Box plots represent upper and lower quartiles, whiskers are upper and lower extreme. n = 8 in (C) and 10 in (D). The line in the boxes is the median. Letters above the bars denote statistical analysis (one-way ANOVA and Tukey’s HSD test, P > 0.05).

Next, we examined whether ROP recycling between active and inactive states is required for domain formation. To assess the effect of cycling between active and inactive states of ROPs, constitutively active (CA) GFP:rop11CA and GFP:rop2CA were coexpressed with Cyan Fluorescent Protein (CFP):GAP1 and mChGEF3p. The result was a formation of multiple rop11CA-GEF3p and rop2CA-GEF3p nanodomains, which were dispersed throughout the cell rather than being clustered in discrete domains (Figure 6, A and C). Importantly, nanodomains were not observed in cell surface images of either singly expressed GFP:rop11CA or GFP:rop2CA (Figure 6, B and D), indicating that nanodomain formation by these CA ROPs depended on the coexpression of GEF3p. Similar expression of WT GFP:ROP11 or GFP:ROP2 with CFP:GAP1 and mCherry:GEF3p resulted in domain formation (Figure 6, E and G), indicating that CFP:GAP1 function was similar to the untagged GAP1. Cell surface images showed that when singly expressed, neither GFP:ROP11 nor GFP:ROP2 formed domain or nanodomains (Figure 6, F and H), similar to their CA mutants. Quantification of domains formed by GFP:ROP11 and GFP:ROP2 versus nanodomains formed by GFP:rop11CA and GFP:rop2CA in a 25 µm² area showed that the numbers were significantly different (Figure 6, I and J). Importantly, while ROP11 and rop11CA, ROP2 and rop2CA form nanodomains together with GEF3p, only the wild-type ROP11 and ROP2 formed domains in the presence of GAP1 or CFP:GAP1.

Figure 6.

Formation of CA rop11CA-GEF3p nanodomains in the presence of GAP1. A, Multiple GFP:rop11CA and mChGEF3p nanodomains form when coexpressed with CFP:GAP1. O/L, overlay of mCherry and GFP signals. B, Singly expressed GFPROP11CA at the cell surface plane. C, Multiple GFP:rop2CA and mChGEF3p nanodomains form when coexpressed with CFP:GAP1. O/L, overlay of mCherry and GFP signals. D, Singly expressed GFPROP2CA at the cell surface plane. E, Wild-type GFPPROP11 and mCherryGEF3p form domain when coexpressed with CFP:GAP1. O/L, overlay of mCherry and GFP signals. F, Singly expressed GFPROP11 at the cell surface plane. G, Wild-type GFPPROP2 and mCherryGEF3p form domain when coexpressed with CFP:GAP1. O/L, overlay of mCherry and GFP signals. H, Singly expressed GFPROP2 at the cell surface plane. The scale bar is 10 µm for all panels. The scale bar is 10 µm for all panels. I, Quantification of the domains and nanodomains densities formed by wild-type GFP:ROP11 and CA GFP:rop11CA in a 50 µm² area. J, Quantification of the domains and nanodomains formed by wild-type GFP:ROP2 and CA GFP:rop2CA in a 50 µm² area. Box plots represent range from upper and lower quartile, whiskers are upper and lower extreme. The line in the boxes is the median. ^***P < 0.0001 (two-tailed Student’s t test). n = 10 in (I) and (J).

Similar nanodomains were formed upon expression of GFP:rop2CA or GFP:rop11CA and mChGEF3p without GAP1 (Figure 7, A and F), further indicating that the dispersed nanodomains required the presence of the activated ROP11 and GEF3p but did not depend on the function of the GAP. Similar to rop2CA and rop11CA, multiple dispersed nanodomains formed following coexpression of mChGEF3p with GFP fusion of rop2CA, rop4CA, rop6CA, rop9CA, and rop10CA (Figure 7, B–E). Importantly, no nanodomains or domains were observed following expression of GFProp2CA or GFP:rop11CA with mCherry:GAP1 without GEF3p (Figure 7, G and H). Quantification of nanodomains number showed their numbers were significantly different between rop2CA and rop6CA but were not significantly different between all other ropCAs (Figure 7I).

Figure 7.

Formation of ROP-ROPGEF3_PRONE clusters in the absence of GAP. GFP fusions of rop2CA (A), rop4CA (B), rop6CA (C), rop9CA (D), rop10CA (E), and rop11CA (F) were coexpressed with mChGEF3p without GAP1. All CA ROPs formed multiple clusters with mChGEF3p but not domains. G, When coexpressed just with mCherry:GAP1 (mChGAP1), GFP:rop2CA did not form domains. H, When coexpressed just with mCherry:GAP1 (mChGAP1), GFP:rop11CA did not form domains. Bars are 10 µm for all panels. I, Quantification of nanodomain densities formed by GFP:ropCAs coexpressed with mCherryGEF3p in a 10 µm². Box plots represent upper and lower quartile, whiskers are upper and lower extreme. n = 8–10. The line in the box is the median. Different letters indicate significant differences between the nanodomains number (ANOVA and Tukey’s HSD, P < 0.05 between 2CA and 6CA, P > 0.05 for all other samples).

The data presented in Figures 6 and 7 indicate that the formation of nanodomains likely results from the association of some ROPs with GEF3p and that GAPs delimit the nanodomains resulting in the formation of domains.

Dynamics of GFP:ROP11 in the ROP–GEF–GAP domains

The constant localization of GFP:ROP11 domains in the tracheary elements pits (Figure 4) prompted us to examine the interaction dynamics of GFP:ROP11 with the plasma membrane under various conditions using a quantitative approach. To this end we used Fluorescence Recovery After Photobleaching (FRAP) beam-size analysis (Henis et al., 2006), employing a fixed-localization Gaussian beam and continuous photon counting to follow fast recoveries ([Berkovich et al., 2011]; see “Materials and methods”). We have previously used this method successfully to characterize GFP:ROP6 plasma membrane interaction dynamics (Sorek et al., 2011, 2010). This method measures the relative contribution of lateral diffusion in the membrane and of membrane-cytoplasm exchange to the fluorescence recovery. It is based on the use of two different-magnification objectives, in our case ×63 or ×40, to generate two Gaussian laser beam sizes (Henis et al., 2006). The laser beam is focused to a very small spot, comprising ˂1% of the membrane area, which is illuminated by the beam, and only part of it is bleached (Figure 8A, white dot and arrow) such that there is no change in the fluorescence outside of the measurement region. The ratio between the areas illuminated by these beams, ω²(×40)/ω²(×63) (where ω is the Gaussian beam radius), was 2.28. When FRAP occurs by lateral diffusion, the characteristic fluorescence recovery time (τ) is proportional to the bleached area. Therefore, the ratio between the τ values measured with the two objectives, τ(×40)/τ(×63), should be equal to the ratio between the bleached areas (2.28). On the other hand, when FRAP takes place by exchange, τ reflects the chemical relaxation time, which is independent of the beam size. Under these conditions, τ(×40)/τ(×63) should equal 1 (Henis et al., 2006). When both processes contribute to the FRAP measurement, the one that occurs faster will contribute more, and the τ ratio will be intermediate between 1 and 2.28 (Henis et al., 2006). In both cases, the mobile fraction (R_f) on the FRAP timescale is derived from the extent of fluorescence recovery. The FRAP beam-size analysis was carried out on GFP:ROP11 or GFP:rop11CA transiently expressed in N. benthamiana alone or with various combinations of mChGEF3p and His₆:GAP1 (Figure 8;Supplemental Table S4 for the source data for the FRAP studies containing statistics). As shown in Figure 8, B and C, the R_f values of GFP:ROP11 were 0.75–0.82 when expressed alone or together with either GEF3p or GAP1 suggesting the existence of two sub-populations (a major one which is mobile, and another which is immobile on the FRAP timescale). The high R_f (∼1) of singly expressed GFP:rop11CA in the same expression system indicates that this is not an artifact of the system. Of note, R_f of both ROP proteins became significantly lower when coexpressed with GEF3p and GAP1 (with a stronger effect on GFP:ROP11). This suggests that a substantial fraction of the molecules becomes immobile, either due to lateral immobilization (e.g. for GFP:ROP11, whose recovery is dominated by lateral diffusion—see Figure 8E), or due to reduced exchange (e.g. for GFP:rop11CA, whose recovery is dominated by exchange—Figure 8E). This is consistent with the relative immobility of GFP:ROP11 domains in dedifferentiating tracheary elements (Figure 4). The specific effects of GEF3p and GAP1 on the membrane interactions of ROP11 and rop11CA are evident in Figure 8, D and E. The τ(×40)/τ(×63) ratio of GFP:ROP11 alone was significantly different from the 2.28 beam size ratio expected for FRAP by lateral diffusion, and was closer to the value of 1, which characterizes exchange. This suggests a high contribution of exchange and a relatively low affinity to the membrane. In the presence of GEF3p, the τ ratio was increased (from 1.25 to 1.55), suggesting enhanced membrane interactions; coexpression with GAP1 had an even stronger effect, increasing the τ ratio to 2.28, value of FRAP by lateral diffusion. This enabled to derive the lateral diffusion coefficient D, which under these conditions was 0.23 µm²/s (see Figure 8 legend), in the same range but about two-fold slower than the value measured earlier for ROP6 in Arabidopsis leaf epidermis cells (Sorek et al., 2010, 2011), suggesting mobility-restricting interactions with slower-diffusing assemblies. The difference is likely due to the different membrane anchor moieties of type-II (ROP11) and type-I (ROP6) ROPs. Of note, coexpression with both GEF3p and GAP1, which dramatically reduced R_f of GFP:ROP11 (Figure 8C), reduced the τ ratio of the mobile GFP:ROP11 population (comprising ∼40% of the molecules) to ∼1.4, an effect which most likely reflects the lateral immobility of the protein in the domains formed, leaving mostly exchange to be measured. Interestingly, GFP:rop11CA, which displayed a stronger association with the plasma membrane when expressed alone (a τ ratio close to 2.28, indicating negligible contribution by exchange), was shifted to a lower R_f (0.70–0.80) upon coexpression with GEF3p and GAP1, as well as to FRAP by pure exchange (τ ratio of ∼1.0). These results suggest that the membrane interactions of GFP:ROP11 and GFP:rop11CA are nonidentical, and are affected differently by GEF3p and GAP1, in line with the notion that the activation/inactivation cycle triggered by these proteins in wild-type ROP11 is important for its interactions with the domains. The results also demonstrated the transient expression assays were highly reproducible allowing for accurate quantification and statistical analysis.

Figure 8.

GAP1 and GEF3p alter the interactions of ROP11 with the plasma membrane. Nicotiana benthamiana leaves transiently expressing GFP:ROP11 (ROP11) or GFP:rop11CA (rop11CA), alone or with various combinations of mChGEF3p or His₆:GAP1 (GAP1) were subjected to FRAP studies at 22°C as described under “Materials and Methods”. The laser beam was focused on the plasma membrane in cases where there were no domains, or within a domain in cases where the GFP:ROP1 accumulated in domains, where most of the fluorescence was concentrated. A, An example of the beam location in a nanodomain is shown by a circle (pointed at by an arrow) representing a typical beam location. Bar 10 µm. B, Typical FRAP curves of ROP11 alone or together with GEF3p and GAP1 obtained using ×63 (smaller beam size) or ×40 (larger beam size) objectives. The time segments showing prebleach fluorescence intensity, the bleach point, and postbleach recovery are indicated by arrows. Solid lines are the best fit for a nonlinear regression analysis (Henis et al., 2006); the resulting fluorescence recovery time (τ) and mobile fraction (R_f) values are shown. C–E, FRAP beam-size analysis. Bars are means ± sem of multiple measurements (the number of independent measurements, each conducted on a different cell, is shown within the bars). The data depicted in (D) are also used to calculate the τ ratios shown in (E). The ratio of the beam sizes obtained with the two objectives, ω²(×40)/ω²(×63), was 2.28 ± 0.15 (n = 59). Thus, a τ(×40)/τ(×63) ratio similar to that value is expected for FRAP by lateral diffusion, while a ratio of 1 is expected for recovery by exchange (Henis et al., 2006). C and D, R_f values (C) and τ values (D). Asterisks indicate significant differences between the values measured for singly expressed ROP11 and ROP11 coexpressed with GEF3p, GAP1 or both, and between ROP11 and rop11CA, comparing separately the R_f or τ values measured with each laser beam size (**P = 0.0013; ***P = 0.0004; ****P < 0.0001; one-way ANOVA and Bonferroni post hoc test). In addition, rop11CA was compared with rop11CA coexpressed with GEF3p and GAP1 (****P < 0.0001; Student’s two-tailed t test). E, Bootstrap analysis (see “Materials and Methods”) using 1,000 bootstrap resampling values for both τ and beam-size values. Each τ(×40/τ(×63) value was compared to the 2.28 beam-size ratio expected for FRAP by lateral diffusion. The τ ratios of singly expressed ROP11, ROP11 coexpressed with GEF3p alone or together with GAP1, and rop11CA in the presence of GEF3p and GAP were significantly different from the 2.28 value expected for FRAP by lateral diffusion (****P < 0.0001; ******P < 10⁻¹⁰; *******P < 10⁻²⁰; bootstrap analysis and Student's two-tailed t test). On the other hand, for singly expressed rop11CA or ROP11 coexpressed with GAP1, the τ(×40)/τ(×63) ratios were not significantly different from the 2.28 beam-size ratio, indicating FRAP by lateral diffusion. In the cases where the τ ratio was similar to that expected for lateral diffusion (GFP:ROP11 with GAP1, and singly-expressed GFP:rop11CA), τ is the characteristic diffusion time, and the diffusion coefficient (D) can be calculated (D = ω²/4τ). This yielded D values of 0.23 µm²/s for ROP11 in the presence of GAP1, and 0.20 µm²/s for rop11CA expressed alone. The source data for the FRAP studies containing statistics (C–E) are provided in Supplemental Table S4.

The involvement of RhoGDI in domain formation

In yeast and animal cells RhoGDI-mediated cycling is critical for Rho GTPases polarization (Irazoqui et al., 2003; Johnson et al., 2011; Goryachev and Leda, 2017; Golding et al., 2019). In plants, ROP polarization in pollen tubes depends on RhoGDI and polar RH growth is compromised in RhoGDI mutants (Carol et al., 2005; Sun et al., 2015). To further establish that the formation of ROP domains depends on the stability of the interaction with the plasma membrane and reduced mobility rather than on RhoGDI-mediated cycling, we examined the interaction of different ROPs with RhoGDI. Studies on mammalian Cdc42 and Rac demonstrated that geranylgeranylation is required for the interaction between Rho proteins and RhoGDI (Hoffman et al., 2000; Scheffzek et al., 2000). Type-I ROPs terminate with a canonical geranylgeranylation CaaL box motif and in vivo geranylgeranylation of ROP6 has been demonstrated (Sorek et al., 2011). Based on their sequences and function, the plant RhoGDIs are structurally and functionally conserved (Carol et al., 2005; Sun et al., 2015). Type-II ROPs contain a C-terminal GCCG motif and are S-acylated but not prenylated (Lavy et al., 2002; Lavy and Yalovsky, 2006). It was therefore expected that type-I ROPs, but not type-II ROPs, would interact with RhoGDI. In yeast two-hybrid assays, RhoGDI1 (RhoGDI) interacted with ROP2, ROP4, and ROP6 type-I ROPs but not with ROP9, ROP10, and ROP11 type-II ROPs (Figure 9A).

Figure 9.

RhoGDI1 interacts with and extracts from the membrane type-I but not type-II ROPs. A, Yeast two hybrid assays showing interaction of GST:RhoGDI1 (GST:GDI) with ROP2, ROP4, and ROP6 but not with ROP9, ROP10, and ROP11. -LT: Leu Trp dropout medium; -LTH: Leu, Trp, His dropout medium, NC, Negative control (vector control [VC] of the Gal4BD vector used for ROP expression); VC (Gal4AD vector used for expression of RhoGDI1); numbers at the top denote dilution of the yeast culture. B, Extraction of ROPs from the membrane by GST:GDI or GST alone following transient expression in N. benthamiana. T, total protein extract; s, soluble protein extract; m, membrane protein extract. Note that ROP2, ROP4, and ROP6 were extracted to the soluble fraction while ROP9, ROP10, and ROP11 were not. C, Quantification of the bands in the soluble fraction from (B) as percentage of soluble and membrane fractions. D, Extraction of ROPs from the membrane of transgenic Arabidopsis plants with GST:GDI or GST. Labeling of fractions as in B. Note that both ROP6 and rop6CA were extracted to the soluble fraction by GST:GDI and not by GST whereas ROP10 and ROP11 were not solubilized by GST:GDI. E, Overlay images of domains formed by ROP2, ROP4, ROP6, ROP9, ROP10, and ROP11 GFP fusion proteins when coexpressed with mCherryGEF3p, CFPGAP1, and RhoGDI1. The domains resemble the domains formed in the absence of RhoGDI. Bars are 20 µm. See Supplemental Figures S5 and S6 for the full images in all channels.

Studies in mammalian cells showed that as part of their function, RhoGDIs extract Rho protein from the plasma membrane (Golding et al., 2019). Given the conservation of the plant RhoGDIs, testing the ability of RhoGDI1 to extract ROPs from the plasma membrane could therefore test their functionality on different ROPs. To test for ROPs’ extraction by RhoGDI1, GFP fusion proteins of ROP2, ROP4, ROP6, ROP9, ROP10, and ROP11 were transiently expressed in N. benthamiana. Total protein extracts (t) were fractionated into soluble (s) and insoluble membrane (m) fractions by centrifugation. In turn, the insoluble membrane fraction was incubated with recombinant Escheritia coli expressed Glutathione-S-Transferase (GST):RhoGDI1 fusion protein or GST alone and was fractionated again into soluble and insoluble membrane fractions. The proteins in each fraction were resolved by sodium dodecyl sulphate (SDS)–polyacrylamide gel electrophoresis and the ROP proteins were identified by immunoblots decorated with anti-GFP antibodies. The results demonstrated that the type-I ROPs: ROP2, ROP4, and ROP6 were extracted from the membrane by GST:RhoGDI1 but not by GST, while the type-II ROPS: ROP9, ROP10, and ROP11 were not (Figure 9, B and C). While in some type-II ROPs (in particular ROP10) a significant soluble fraction was identified prior to incubation with GST:RhoGDI, the ratio of ROP in soluble to insoluble fractions was not altered following incubation with GST:RhoGDI1 or GST. In contrast, ROP2, ROP4, and ROP6 extraction from the membrane was only detected with GST:RhoGDI1 and not with GST (Figure 9, B and C). Similar results were obtained with GFP fusion proteins of ROP6 rop6CA, ROP10, and ROP11, which were stably expressed in Arabidopsis. Namely, GST:RhoGDI1 (GDI) extracted ROP6 and rop6CA from the membrane, but not ROP10 and ROP11 (Figure 9D). Experiments have been repeated with similar data. Taken together, these data show that RhoGDI1 can interact with and extract from the membrane type-I but not type-II ROPs. These results indicated that the formation of plasma membrane ROP domains upon coexpression with GEF3p and GAP1 does not depend on RhoGDI.

To further examine the effect of RhoGDI on domain formation, the type-I ROP2, ROP4, ROP6, and the type-II ROP9, ROP10, and ROP11 were coexpressed with GEF3p, GAP1, and CFP:RhoGDI1. All six ROPs formed domains either in the absence or presence of CFP:RhoGDI1 (Figure 9E;Supplemental Figures S5 and S6). Hence, the formation of ROP domains does not require RhoGDI-dependent recycling. It is more likely that it depends on a feature of the ROP proteins that enables them to associate with additional proteins and lipid domains in the presence of GEF3p and GAP1, resulting in their immobilization in the domains.

The role of anionic lipid interaction in domain formation

All ROPs contain a C-terminal PBR which facilitates membrane interaction in type-II ROPs, and type-I ROPs are also prenylated by geranylgeranyl transferase-I (James et al., 1995; Caldelari et al., 2001; Lavy et al., 2002; Lavy and Yalovsky, 2006; see also Supplemental Figure S7). Studies on mammalian Rho family protein demonstrated that they interact with PIPs/PtdInsPs, anionic lipids via the hypervariable domain PBR (Heo et al., 2006). In Arabidopsis, it was shown that ROP6 signaling involves clustering in plasma membrane nanodomains which depends on interaction with PS through the hypervariable domain PBR (Platre et al., 2019). Recently, Nicotiana tabacum (tobacco) NtRAC5 has shown to interact with Arabidopsis PI4P4 5-Kinase 2 and in tobacco pollen tubes PIP5K5-EYFP and NtRAC5-RFP containing nanodomain were colocalized, implicating phosphatidylinositol4,5-bisphosphate (PtdIns(4,5)P2) in NtRAC5 clustering in nanodomains (Fratini et al., 2020). We examined the interaction of E. coli expressed recombinant ROP11 and ROP6 (Supplemental Figure S8) with lipid using “PIP Strips” (Echelon Biosciences, Salt Lake City, UT, USA). Both ROP11 and ROP6 interacted with monophosphate, diphosphate, and to a lesser extent with triphosphate PtdInsPs/PIPs, as well as with as with phosphatidic acid, PS, phosphatidyl ethanolamine, and lysophosphatidic acid (Figure 10, A and B). ROP11 and ROP6 mutants from which the PBR has been deleted (Supplemental Figure S7) did not interact with any of the lipids on the strips (Figure 10B; rop11D-PBR and rop6D-PBR). Hence both ROP11 and ROP6 can interact with anionic lipid via their C-terminal hypervariable domain PBR.

Figure 10.

ROP11 and ROP6 interaction with anionic lipids and domain formation. A, The PIP strip scheme displaying the order of the spotted lipids. Lipids with which ROP11 and ROP6 interact are highlighted in blue. B, Blots displaying the interaction with the PIP strip of E. coli expressed poly His₆-tagged ROP11, ROP6, and their mutants from which the polybasic region was deleted (rop11D-PBR and rop6D-PBR). Identification of the bound ROPs was carried out with anti-poly-His monoclonal Abs. Both ROP11 and ROP6 interacted with the same set of lipids while their PBR deleted mutant did not interact with any of the lipids. C, Small clusters detected following coexpression of GFP:rop11D-PBR with mChGEF3p and CFP:GAP1. D, Enlargement of the region highlighted by rectangle in (C). The scale bar is 10 µm for all panels. E, Quantification of nanodomains density in 10 × 10 µm area. Box plots represent upper and lower quartile, whiskers are upper and lower extreme. The line is the median. n = 10.

To examine whether the ability to interact with anionic lipids affects ROP domain formation, GFP:rop11D-PBR was coexpressed with mChGEF3p and CFP:GAP1. This resulted in GFP:rop11D-PBR localized in multiple nanodomains together with mChGEF3p rather than in discrete domains (Figure 10, C–E), similar to the distribution of the CA ROPs under such conditions. These data suggest that in addition to ROP cycling between active and inactive states, domain formation requires interaction of ROPs with anionic lipids in the membrane.

Recruitment of the ROP-binding protein ICR1 to domains

To further characterize the ROP domains, we examined the ability of different ROPs to recruit the ROP interactor ICR1 (Lavy et al., 2007). Untagged GEF3p and GAP1 were coexpressed with GFP-tagged ROPs and mCherry-tagged ICR1. The mCherry tag was placed at the C-terminal end of ICR1 to allow MTs binding (Hazak et al., 2019). Following expression with ROP2 and ROP4, the ICR1:mCherry MTs were organized around the ROP domains (Figure 11, A and B). The overlay images show that the GFP and mCherry probes were not colocalized, indicating that ICR1 was not recruited to the domains with either ROP2 or ROP4. Following expression with GFP:ROP6, ICR1:mCherry was still detected on MTs but also colocalized with ROP6 in the domains. The MTs organization was not disrupted (Figure 11C). This distribution of mCherry:ICR1 when expressed with GFP:ROP6 suggests that it exists in two populations, one associated with MTs and one with ROP6 domains. Colocalization analysis of the ROP6 domains showed that the average overlap and Pearson coefficient of GFP:ROP6 and ICR:mCherry were 0.94 and 0.86, respectively, indicating high degree of colocalization (see also Table 1). Remarkably, when ICR1:mCherry was coexpressed with either ROP9, ROP10, or ROP11, the MTs were organized around the domains and underwent substantial reorganization (Figure 11, D–F). Hence, rather than MT-bound and unbound populations of mCherry:ICR1 as in the case of the coexpression with GFP:ROP6, coexpression with ROP9/10/11 resulted in mCherry:ICR1 remaining associated with MTs even when recruited to the vicinity of the domains (compare Figure 11, C–F). Contrary to ROP6, the Pearson R values of the colocalization between ICR1mCherry and either GFP:ROP9/10/11 were low (Table 2) indicating that the proteins were not localized in the same plane, likely reflecting the localization of the ICR:mCherry on MTs.

Figure 11.

Recruitment of MTs-associated ICR1 to ROP domains. ICR1:mCherry (ICR1mCh) was coexpressed with untagged GEF3p and GAP1 and with GFP fusions of ROP2 (A), ROP4 (B), ROP6 (C), ROP9 (D), ROP10 (E), and ROP11 (F). ICR1mCh was not recruited to the domains formed by GFP:ROP2 and GFP:ROP4 (A and B) and was recruited to the domains formed by GFP:ROP6/ROP9/ROP10/ROP11 (C–F). While ICR1mCh was recruited to the GFP:ROP6 domains, no rearrangement of the MTs was detected (C). Recruitment of ICR1mCh to the GFP:ROP9/ROP10/ROP11 domains was associated with rearrangement of the MTs around the domains (D–F). O/L, overlay of the GFP and mCherry probes. The scale bars are 10 µm for all panels.

Table 2.

Average colocalization between domain localized ROPs and ICR1

ROP Protein	Overlap Coefficient	Correlation R
ROP6	0.94	0.87
ROP9	0.85	0.14
ROP10	0.91	0.15
ROP11	0.57	0.14

To further examine the differences between ROP6 and the type-II ROPs, ICR1:mCherry was coexpressed with GFP:ROP6 or GFP:ROP10, along with GEF3 and GAP1, followed by treating the leaves with 20 µM of the anti-MTs drug oryzalin. The oryzalin treatment eliminated ICR1:mCherry from ROP10 domains, while GFP:ROP10 accumulation in the domains was not affected (Figure 12A). In contrast, ICR1:mCherry was still associated with GFP:ROP6 domains following the oryzalin treatments (Figure 12B). MTs labeled with the MT reporter proteins GFP:TUA6 (GFPTUA) (Hazak et al., 2019) and monomeric Red Fluorescent Protein (mRFP) MT Binding Domain (MBD) fusion protein (mRFP–MBD) (Lipka et al., 2014) were not associated with and reorganized around ROP9, ROP10, and ROP11 domains (Figure 12C). Taken together, the data in Figure 12 confirmed that mCherry:ICR1 associated with ROP10 domains localized on MTs, while ICR1:mCherry which was associated withROP6 domains was not associated with MTs. Furthermore, the association of MTs with the type-II ROP domains requires the ROP interacting ICR1 and is not due to unspecific interaction of MTs with the ROP domains.

Figure 12.

The effect of MT destabilization on ICR1 recruitment to the ROP domains and the specificity toward ICR1. A and B, ICR1:mCherry (ICR1mCh) domains with GFP:ROP10, but not with GFP:ROP6, were disrupted by treatment with 20 µM oryzalin. A, Cortical MTs and ICR1mCh recruited to GFP:ROP10 domains disappeared while the GFP:ROP10 domains remained intact. B, ICR1mCh was still observed in GFP:ROP6 domains while cortical MTs almost disappeared. C, GFP:tubulin6 (GFPTUA) and mRFP:mRFPMBD MT markers were not recruited to mCherry:ROP9 (mChROP9), or GFP:ROP10 and GFP:ROP11 domains. In all panels, GAP1 and GEF3p were untagged. O/L, overlay of GFP and mCherry probes (A and B). The scale bars are 10 µm for all panels.

Discussion

Here we examined several fundamental aspects of self-organizing domains of ROP GTPases, namely: the ability of different types I and II ROPs to form domains, the molecular mechanisms involved in domain formation, and the characteristics of domains formed by different ROPs. As discussed below, the results also offer a synthetic biology approach for engineering cell structures.

ROPs and domain formation

All six ROPs tested in the current work formed domains. Previously, using a similar N. benthamiana transient expression system, the localization of ROPs in the domains has been inferred from the distribution of GEF4_PRONE/GEF7_PRONE, but the distributions of the ROPs themselves have not been reported (Oda and Fukuda, 2012a; Nagashima et al., 2018). The ability to observe the ROPs in the domains was critical for analysis of the mechanisms underlying domain formation. We have also tested domain formation with GEF4p and GEF7p but did not observe domains. Of note, when previously tested with ROP11 and GAP3, in another system, GEF3p did not form domains (Oda and Fukuda, 2012b). In addition, we have not observed substantial differences in the ability of different ROPs to form domains. Possibly, domain formation by different GEFs and ROPs is sensitive to the experimental conditions. In the future, it should be interesting to compare the different experimental setups used by different labs to study such domains. Importantly, the transient expression system which was extensively use in this study provided reproducible results which could be quantified and subjected to statistical examinations. Furthermore, the accumulation of ROPs in the domains following transient expression was similar to the distribution of ROP11 in dedifferentiating tracheary elements and the ability of RhoGDI1 to extract ROPs from the plasma membrane was the same whether they were transient expressed in N. benthamiana or stably expressed in Arabidopsis. In addition, the ROPs in the domains induced MT reorganization in an ICR1-dependent fashion, reflecting on their functionality. Taken together, the results from the different experiments indicate that the data obtained from the transient expression system is physiologically relevant.

The type-I ROP2/4/6 and the type-II ROP9/10/11 formed domains irrespective of their different C-terminal posttranslational lipid modifications and differences in the amino acid sequences in the G and hypervariable domains, yet the domains of different ROP types had different characteristics (Lavy et al., 2002; Lavy and Yalovsky, 2006; Sorek et al., 2011). In agreement, RhoGDI, which only interacts and functions with the geranylgeranylated type-I ROPs, was not required for domain formation. In the RhoGDI1 Arabidopsis mutant supercentipede1 (scn1), ROP2 accumulates in multiple RH initials (Carol et al., 2005), indicating that RhoGDI1 is required for localized positioning of ROP2 and possibly other type-I ROPs. It is therefore likely that RhoGDI-dependent type-I ROP positioning and the ROP–GEF3–GAP1 domain module are independent and may function additively to mediate ROP-dependent polarity. Interestingly, GEF3 appears early in RH initials prior to ROP2 (Denninger et al., 2019). Possibly, GEF3 and ROP2 form domains that are delimited by GAPs and AROs (Kulich et al., 2020), while RhoGDI1 (SCN1) is dependent recycling of ROP2 ensures formation of a single domain required for single tip growth of RHs. Combination of the two mechanisms is reminiscent of Cdc42 polarization in budding yeast, which depends on a positive feedback loop between Cdc42•GTP, the kinase PAK1, the scaffold protein Bem1 and the GEF Cdc24 on one hand, and on RhoGDI-mediated recycling of Cdc42•GDP on the other hand (Klunder et al., 2013; Woods and Lew, 2017).

The domains consist of colocalized ROP-GEF3p nanodomains. Imaging the plasma membrane at the upper surface revealed that singly expressed GEF3p formed nanodomains, which were enhanced upon coexpression with ROP11 or ROP6. CA GTP-bound ROPs formed dispersed nanodomains when coexpressed with GEF3p. Taken together, these results indicate that the domains are formed by association of ROPs and ROPGEFs which are delimited by the GAP. Such a mechanism for domain formation is in agreement with the report by Nagashima et al. (2018) which proposed that the domains are formed by clustering of GEF–ROP•GTP complexes. Given that dissociation of ROP•GTP from ROPGEFs is a basic feature of the GDP/GTP exchange mechanism (Berken et al., 2005; Thomas et al., 2007), formation of stable GEF–ROP•GTP complexes is not straightforward. One option which would require future studies could be that ROPGEFs and possibly ROPs function as homodimers enabling simultaneous association of ROP•GDP to one GEF subunit and release of ROP•GTP from the second subunit. Under the experimental conditions described in this study, association of ROP•GTP and GEF via a budding yeast-like mechanism, where Cdc42 and its GEF Cdc24 interact through an adaptor Bem1 (Kozubowski et al., 2008; Johnson et al., 2011), is less likely. The reason is that both the ROPs and the GEF3/GEF3p were in excess and for interaction with a third protein component it would need to be stoichiometrically abundant. The mechanisms that underlie nanodomain formation by GEF3 are not known and are a subject for future studies. Nor is it known whether other ROPGEFs form similar nanodomains. Interestingly, at the onset of RH development, GEF3 accumulates at the future RH imitation site (Denninger et al., 2019). Possibly, the local accumulation of GEF3 enables the following recruitment by ROP2 and the initiation of the RH polar growth.

Activation-dependent transient S-acylation and consequent partitioning into membrane microdomains are required for ROP6 signaling (Sorek et al., 2010). In addition, ROP6 interaction with PS is developmentally regulated and required for its function in auxin signaling (Platre et al., 2019). Similar to ROP6, the C-terminal PBR of ROP11 facilitates interaction with PtdnsPs and PS (this work) and is required for stabilizing the interaction with the plasma membrane (Lavy and Yalovsky, 2006). In tobacco, pollen tubes RAC5 localize in PtdInsPs nanodomains (Fratini et al., 2020). The results of this work further indicate that the ROP PBR is required for domain formation but not for association with GEF3 in the membrane. Thus, in addition to activation/inactivation cycles mediated by GEF and GAP, the clustering of at least ROP11 into domains requires interaction with anionic lipids in the plasma membrane. Unfortunately, similar analysis of type-I ROPs would be more challenging since the polybasic region is also required for geranylgeranylation by PGGT-I, making it difficult to discern between the effects of prenylation and binding to PtdInsPs on the membrane interactions.

The Interaction dynamics of ROP11 with the plasma membrane

Both the time-lapse imaging of GFP:ROP11 in dedifferentiating tracheary elements and the FRAP beam-size analysis studies indicated that the formation of the ROP domains is associated with reduced mobility of a major population of ROP11 in the domains, indicative of mobility-restricting interactions with relatively immobile structures. Some interactions with immobile elements are already evident by the presence of an immobile sub-population (18%–25%) of singly expressed GFP:ROP11 (Figure 8C). The mobile population, which comprises the majority of GFP:ROP11 molecules under these conditions, recovered in FRAP experiments mainly by exchange (the τ(×40)/τ(×63) ratio is close to 1), suggesting relatively weak association with the plasma membrane (Figure 8, D and E). Of note, coexpression of GFP:ROP11 with GAP1 alone, which does not induce domain formation by itself and is mostly left out of the domains formed when mChGEF3p is also expressed (Figure 2G;Supplemental Table S2), shifted the FRAP mechanism of GFP:ROP11 from exchange to pure lateral diffusion (Figure 8E, bar 3), but did not induce immobilization (Figure 8C). This indicates that the interaction of GAP1 with ROP11 stabilizes the membrane association of the latter, perhaps by inducing a conformation where the ROP11 PBR domain is more susceptible to binding to the negatively charged lipids of the membrane, slowing down ROP11 dissociation rate from the membrane and making lateral diffusion the faster and dominant process. This effect does not appear to depend on domain formation, as coexpression with GAP1 alone does not form domains. On the other hand, coexpression of mCh:GEF3p with GFP:ROP11, which induces nanodomains, has a much weaker effect than GAP1 on the membrane affinity of ROP11, shifting its recovery only modestly to include a contribution of lateral diffusion (Figure 8E, bar 2). This suggests that the presence of nanodomains is not sufficient to immobilize ROP11 molecules interacting with them, either due to relatively weak affinity in the absence of GAP1, or due to dynamic transient kinetics of the nanodomains. However, when GFP:ROP11 is coexpressed with both mChGEF3p and GAP1, the nanodomains cluster into larger domains. These domains appear to be laterally immobile, and the higher affinity of GFP:ROP11 to them would slow down to undetectable level both their exchange and lateral diffusion. This notion is supported by the large immobile fraction of GFP:ROP11 under these conditions (Figure 8C, bar pair 4) and the faster exchange measured under these conditions for the small fraction of GFP:ROP11 molecules that resides outside of the domains (Figure 8D, bar pair 4; Figure 8E, bar 4). The formation of the domains appears to depend on the activation/inactivation cycle of ROP11, which leads to enhanced interactions with the domain regions resulting in immobilization of the majority of the ROP11 molecules in the domains. The notion that the ROP activation/inactivation cycle is important for domain formation is in line with the observation that rop11CA, which is CA and thus cannot undergo activation/inactivation cycles, exhibits a much lower immobile population (20%–30%) in the presence of GEF3p and GAP1, and stops at the formation of nanodomains without proceeding to domains (Figures 6 and 7).

The stabilization of ROPs’ interaction with the plasma membrane may involve activation-dependent transient S-acylation of G-domain Cys residues (Sorek et al., 2010). In addition, conformational changes in the hypervariable domain of ROPs may facilitate the ROPs’ interaction with anionic lipids in the plasma membrane through their PBRs (Heo et al., 2006; Zhou and Hancock, 2018). In Arabidopsis, ROP6 nanoclustering in the plasma membrane was shown to involve its association with PS. Similarly, interaction with PS is required for nanoclustering of K-Ras, Rac1, and Cdc42 in animal and yeast cells (Zhou et al., 2015, 2017; Remorino et al., 2017; Sartorel et al., 2018). In animal cells, PS is highly abundant in the inner leaflet of the plasma membrane and has a significant immobile fraction (Leventis and Grinstein, 2010; Kay et al., 2012). Further clustering of PS is thought to occur by its interaction with PBR-containing proteins (Zhou and Hancock, 2018). In tobacco pollen tubes PtdIns(4,5)P2 was implicated in the nano clustering of NtRAC5 (Fratini et al., 2020). The dispersed nanodomains structures which were formed by GEF3p and the PS/PtdInsPs nonbinding mutant rop11D-PBR suggest that the clustering of ROP11-GEF3p nanodomains to form domains involves interaction with anionic lipids. Yet, the formation of the GEF3p-rop11D-PBR nanodomains appears to be independent of ROP11 interaction with PtdSer (Figure 10).

The nanoclustering of mammalian K-Ras depends on its combined interactions with the plasma membrane via the PBR (with PS) and by its prenylation by either farnesyl or geranylgeranyl (Zhou et al., 2017; Zhou and Hancock, 2018). Type-II ROPs are not prenylated and instead attach to the plasma membrane by S-acylation of 2–3 Cys residues (Lavy et al., 2002). It is therefore expected that the differences between types I and II ROPs in their membrane anchorage and distinct conformations in the vicinity of the membrane would result in different interactions with membrane lipids and associated protein structures. A dependence of ROPs domain formation in the presence of GEF3p and GAP1 on specific interactions of a given ROP with these proteins and other cellular components is highlighted by the differences between distinct ROP proteins in membrane interactions and domain formation. This may be the reason for the slower diffusion of GFP:ROP11 (D = 0.23 µm²/s) relative to ROP6 (0.59 µm²/s; Sorek et al., 2010), which suggests that ROP11 experiences more mobility-restricting interactions. The differences between ROPs become clearly evident when comparing distinct ROPs for the ability to associate with MT-associated proteins (ICR1). Here, the type-I ROP2 and ROP4 failed to recruit ICR1 to the domains formed in the presence of GEF3p and GAP1, while ROP6 was able to recruit part of the ICR1 population to the domains but did not alter MTs organization (Figure 11, A–C). On the other hand, the type-II ROPs (including ROP11) induced reorganization of the ICR1-associated MTs around the domains but the ICR1 signal was not colocalized with the ROP signal in the domains, suggesting that ICR1 remained associated with MTs even when recruited to the type-II ROP domains (Figure 11, D–F;Table 2). Thus, ROP11 in the domains interacts preferentially with MTs and their associated proteins, and these interactions could contribute to its immobilization in the domains.

The domain formation reported in this work was formed following overexpression of ROPs, GEF3p, and GAP1 in the same cells, which could raise concerns on the physiological relevance of the reported findings. However, as described above, the domains thus formed are similar to the previously reported GEF domains in dedifferentiating MX cells (Oda and Fukuda, 2012a; Nagashima et al., 2018) and the ROP11 domains formed in dedifferentiating tracheary elements (Figure 4). GEF3 accumulates in future RH initials and is required for localizing ROP2 (Denninger et al., 2019). Also, in RHs ARO-mediated GAP distribution is required for polar accumulation of ROP2 at the RH tip (Kulich et al., 2020). Similarly, in tobacco pollen tubes subapical distribution of GAP is required for polarizing NtRAC5 at the tip (Klahre and Kost, 2006; Sun et al., 2015). In trichomes, local accumulation of the GEF SPK1 is required for ROP activation at the tip (Yanagisawa et al., 2018). Thus, both in diffuse and tip growth colocalization of ROPs and GEFs and surrounding distribution of GAPs are required for maintaining localized ROP accumulation and growth. Hence, the synthetic domains formed by overexpression of ROPs, GEF3p, and GAP1 in N. benthamiana or Arabidopsis leaf epidermis pavement cells recapitulate developmental contexts and enable the analysis of mechanisms required for self-organizing ROP domain formation.

Characterization of ROP function using synthetic domain formation

To date, analysis of ROP function at the molecular level had been challenging. Gain of function mutants often displays similar phenotypes, while phenotypes of loss of function mutants may result from time and space of expression. Protein–protein interaction studies of ROPs with their effectors often fail to show specificities. For example, in vitro studies in yeast and in plant cells showed that ICR1 interacts with different ROPs without visible specificity (Lavy et al., 2007). Intriguingly, the differences observed here in the interaction of different ROPs in domains with ICR1 may highlight specific differences. Under the experimental condition used in this study, the MT-associated ICR1 was recruited to domains formed by type-II ROPs (ROP9/10/11), which resulted in local MT reorganization. ICR1 was not detected in the domains formed by type-I ROPs (ROP2/4), and ICR1 was recruited to ROP6 domains but it was not detected on MTs (Figures 11 and 12). These findings are in accord with the notion that the different membrane anchorage mode of types II and I ROPs, as reflected in ROP11 (type-II) showing slower lateral diffusion in the membrane than ROP6 (type-I) (Figure 8), is due to stronger mobility-restricting interactions experienced by the type-II ROPs. These results may suggest that ICR1 may function only with a subset of ROPs and that it may have MT-dependent and -independent functions. Testing different ROP effectors by coexpression with different ROPs, ROPGEFs, and ROPGAPs, may thus provide a very useful system for analyzing ROP function at the molecular level.

The reorganization of MTs around domains formed by type-II ROPs suggests that coexpression of ROPs, GEFs, GAPs, and ROP effectors may provide a synthetic platform for cell engineering. It should be possible to study how recruitment of different effectors locally changes cellular components and to accurately determine at the molecular level how ROPs regulate cell growth and morphogenesis.

Materials and methods

Molecular cloning

All plasmids and oligonucleotide primers used in this work are listed in Tables 3 and 4, respectively.

Table 3.

Plasmids used in this study

Name	Description
pSY2058	pB7-35S::GFP-ROP2
pSY2075	pB7-35S::GFP-ROP4
pSY815	pCAMBIA3300-35S::His6-GFP-ROP6
pSY130	pCAMBIA2300-35S::GFP-ROP9
pSY1752	pCAMBIA2300-35S::CFP-ROP9
pSY130	pCAMBIA2300-35S::GFP-ROP10
pSY132	pCAMBIA2300-35S::GFP-ROP11
pSY1776	pCAMBIA2300-35S::GFP-TUA6
pSY1005	pB7-35S::GFP-rop2CA-Q63L
PSY1006	pB7- 35S:GFP-:rop4CA-Q64L
pSY452	pK7 35S:: GFP ROP6 CA-Q64L
pSY1002	pB7-35S::GFP-rop9CA-G13V
pSY447	pCambia2300-35S::GFP-rop10CA-Q65L
pSY2515	pCAMBIA2300-35S::GFP-rop11CA-G13V
pSY688	pGBT-ROP2
pSY2002	pGBT-ROP4
pSY176	pGBT-ROP6
pSY2905	pGBT-ROP9
pSY177	pGBT-ROP10
pSY215	pGBT-ROP11
pSY804	pET28a-ROP6
pSY2079	pET28a-ROP6-deltaPBR
pSY172	pET28a-ROP11
pSY2091	pET28a-ROP11-deltaPBR
pSY2040	pB7-35S::CFP- GAP1
pSY2050	pB7-35S::HisGAP1
pSY2048	pB7-35S::mCherry-GEF3prone
pSY1001	pB7-35S::mCherry-GEF3fl
pSY2064	pB7-35S::HisGEF3prone
pSY480	pB7 -35S::ICRI-mCherry
pSY2532	pROP11:XVE-GFP-ROP11
pSY1714	pGEX-GST-RhoGDI1
	pGEX-4T-1
	pUBN:: mRFP- MBD
pSY1776	pCAMBIA3300-GFP-TUA6
pSY2902	pK7 35S::CFP:RhoGDI1
pSY179	pGAD-RhoGDI1
pSY2904	pK7 35S::His:RhoGDI1

Table 4.

Oligonucleotide primers used in this work

Name	Sequence 5′–3′	Target Gene
attB2F1-RhoGDI1	GTACAAAGTGGCCATGTCTTTGGTATCTGGAGCC	RhoGDI1
attB3R1-RhoGDI1	GTATAATAAAGTTGCTCAAAGCGCAGGCCATTC	RhoGDI1
attB2F2-RhoGDI1	GGGGACAGCTTTCTTGTACAAAGTGGCCATGTCTTTG	RhoGDI1
attB3R2-RhoGDI1	GGGGACAACTTTGTATAATAAAGTTGCTCAAAGCG	RhoGDI1
attB1F1-RhoGDI1	AAAGCAGGCTTGATGTCTTTGGTATCTGGAGCC	RhoGDI1
attB1R1-RhoGDI1	GGGGACAAGTTTGTACAAAAAAGCAGGCTTGATGTC	RhoGDI1
AtRhoGDI1 4-24	GCAAACTAGTTTCTTTGGTATCTGGAGCCAGG	RhoGDI1
AtRhoGDI1 703-723	GCACAAGCTTTCTAGATCAAAGCGCAGGCCATTCTTT	RhoGDI1
AtRhoGDI1 3-24	GTTGAATTCGAGCTCTCTTTGGTATCTGGAGCCAGG	RhoGDI1
AtRhoGDI1 703-723	GTTGAATTCGAGCTCTCAAAGCGCAGGCCATTCTTT	RhoGDI1
AtROP6 4-24-SacI	GGAAGAGCTCAGTGCTTCAAGGTTTATCAAG	AtROP6
AtROP6 577-597-Xho	CGAACTCGAGTCAGAGTATAGAACAACCTTT	AtROP6
AtROP6 4-24-SpeI	GGAAACTAGTTAGTGCTTCAAGGTTTATCAAG	AtROP6
AtROP9 4-24-SacI	CATATGGAGCTCAGTGCTTCGAAGTTCATAAAA	AtROP9
AtROP9 597-630-XbaI,XhoI	GCAGATCTAGATTAAGCAGCGGTGCTACCTCCACTGACAATACTC	AtROP9
AtROP9 4-24-SpeI	GCAAACTAGTTAGTGCTTCGAAGTTCATAAAA	AtROP9
AtROP10 4-24-SpeI	GCAAACTAGTTGCTTCGAGTGCTTCAAAATTC	AtROP10
AtROP11 4-24-SpeI	GCAAACTAGTTGCTTCAAGTGCTTCAAAGTTC	AtROP11
AtROP10 4-24-SacI-NdeI	GGAGCGAGCTCCATATGGCTTCGAGTGCTTCAAAATTC	AtROP10
AtROP10 607-627-XabI, XhoI	GCTCTAGACTCGAGTCAATTCTTCCCACTCAGAATGTT	AtROP10
AtROP11 4-24-SacI	CATATGGAGCTCGCTTCAAGTGCTTCAAAGTTC	AtROP11
AtROP11 628-648-XHO, XBAI	CCGCTCGAGTCTAGATCAATGCCGAGTCACTATCCT	AtROP11
attB2F-ROP2	GGGGACAGCTTTCTTGTACAAAGTGGCCATGGCGTCAAGGTTTATAAAGTGTGTGAC	AtROP2
attB3-ROP2R	GGGGACAACTTTGTATAATAAAGTTGTTCACAAGAACGCGCAACGG	AtROP2
attB2F-ROP4	GGGGACAGCTTTCTTGTACAAAGTGGCCAGTGCTTCGAGGTTTATAAAGTGTGTCA	AtROP4
attB3R-ROP4R	GGGGACAACTTTGTATAATAAAGTTGGTCACAAGAACACGCAGCGGTTC	AtROP4
attB2F-ROP6	GGGGACAGCTTTCTTGTACAAAGTGGCCAGTGCTTCAAGGTTTATCAAGTGTGTCA	AtROP6
attB3R-ROP6	GGGGACAACTTTGTATAATAAAGTTGTTCAGAGTATAGAACAACCTTTCTGAGATTTTCTCTTCTT	AtROP6
attB2F-GAP1	GGGGACAGCTTTCTTGTACAAAGTGGCCAGTGCTTCAAGGTTTATCAAGTGTGTCA	AtGAP1
attB3R-GAP1	GGGGACAACTTTGTATAATAAAGTTGTTCAGAGTATAGAACAACCTTTCTGAGATTTTCTCTTCTT	AtGAP1
attB2F-GEP3prone	GGGGACAGCTTTCTTGTACAAAGTGGCCCCAGAACTTGAGACAATGAAGGAAAGATTCG	AtGEF3prone
attB3R-GEF3prone	GGGGACAACTTTGTATAATAAAGTTGTTTATTCACTACCTCTCATGGTTTTGTCTACATAGAG	AtGEF3prone
attB1F-ICRI	GGGGACAAGTTTGTACAAAAAAGCAGGCTTGATGCCAAGACCAAGAGTTTCAGAGT	AtICRI
attB2R-ICR1	GGGGACCACTTTGTACAAGAAAGCTGGGTCCTTTTGCCCTTTCTTCCTCCACAACT	AtICRI
rop2CAF	TGGGATACTGCTGGTCTAGAGGACTACAACAGG	AtROP2
rop2CAR	CCTGTTGTAGTCCTCTAGACCAGCAGTATCCCA	AtROP2
rop4CAF	GTGGGATACAGCTGGTCTAGAAGACTATAACAGGT	AtROP4
rop2CAR	ACCTGTTATAGTCTTCTAGACCAGGTGTATCCCAC	AtROP4
rop6CAF	GGGATACTGCAGGGCTAGAGGACTACAATA	AtROP6
rop6CAR	CTATTGTAGTCCTCTAGCCCTGCAGTATCC	AtROP6
rop10CAF	GGACACTGCCGGGCTAGAAGACTATAACAG	AtROP10
rop10CAR	CTGTTATAGTCTTCTAGCCCGGCAGTGTCC	AtROP10
pGBT ROP2F	TGGATCCATGGCGTCAAGGTTTATAAAGTGTGTGAC	AtROP2
pGBT ROP2R	GGTCGACTCACAAGAACGCGCAACGGT	AtROP2
pGBT ROP4F	GTGGATCCCATGAGTGCTTCGAGGTTTATAAAGTGTGTC	AtROP4
pGBT ROP4R	GGTCGACTCACAAGAACACGCAGCGGT	AtROP4
pGBT ROP11F	CGAATTCGCTTCAAGTGCTTCAAAGTTCATCAAGTGT	AtROP11
pGBT ROP11R	TTGTCGACTCAATGCCGAGTCACTATCCTCCC	AtROP11
pET28a ROP6F	GCGGATCCAGTGCTTCAAGGTTTATCAAGTGTGTCAC	AtROP6
pET28a ROP6R	TTGTCGACTCAGAGTATAGAACAACCTTTCTGAGATTTTCTCTTC	AtROP6
pET28a ROP11F	CGAATTCGCTTCAAGTGCTTCAAAGTTCATCAAGTG	AtROP11
pET28a ROP11R	TTGTCGACTTCAATGCCGAGTCACTATCCTCCC	AtROP11

ROP6, ROP9, ROP10, ROP11, and Tubulin-6 (TUA6) were cloned downstream GFP into pGFP-MRC as previously described (Rodriguez-Concepcion et al., 1999; Lavy et al., 2002). The resulting cassettes, containing a Cauliflower Mosaic Virus 35S promoter, the gene of interest, and a nitric oxide synthase (NOS) transcriptional terminator, were subcloned into pCAMBIA2300 plant binary vector as previously described (Lavy et al., 2002). GFP:ROP2, GFP:ROP4, CFP:GAP1, His₆:GAP1, mCherry:GEF3_PRONE (mChGEF3p), His₆:GEF3_PRONE, and ICR1:mCherry were amplified using attB2F and attB3R oligonucleotide primers and subsequently subcloned into expression vectors pB7m34GW or pK7m34GW by Three Way Gateway according to the manufacturer instructions (Thermo Fisher Waltham, MA, USA). The expression cassette included the 35S promoter, fluorescent/His₆ tag gene, and NOS terminator.

CA ROP mutants were generated by site-directed mutagenesis with Pfu-Ultra DNA polymerase followed by digestion with DpnI to eliminate unwanted templates as previously described (Hazak et al., 2019). rop2CA, rop4CA, ROP6CA, and ROP10CA were generated by substituting Gln67 to Leu (Q67L) using the ropCA forward and reverse primers. rop9CA and rop11CA were generated by mutating Gly15 to Val (G15V) as previously described (Lavy et al., 2002). Oligonucleotide primer pairs were designed using Agilent QuikChange Primer Design tool (https://www.genomics.agilent.com/primerDesignProgram.jsp). For generating ROP6 DeltaPBR, nucleotides 543–569 were deleted from ROP6 CDS by PCR with oligonucleotide primers which excluded the PBR sequence (pET ROP6; Table 4). For ROP11 DeltaPBR, nucleotides 571–585 were deleted from ROP11 CDS by PCR with oligonucleotide primers which excluded the PBR sequence (pETROP11; Table 4).

For Yeast 2-hybrid, ROP2, ROP4, ROP6, ROP9, ROP10, and ROP11 were subcloned into pGBT9.BS, and RhoGDI1 into pGAD-GH by digestion with SpeI following amplification with appropriate oligonucleotide primers as listed in Table 4.

For protein expression in E. coli, RhoGDI1 was subcloned into pGEX (GE Healthcare, Sigma, Chicago, IL, USA; downstream of GST by digestion with EcoRI. ROP11, rop11:D-PBR (delta 570–588 bp), ROP6, and rop6D-PBR were subcloned into pET28a (Addgene) to generate poly-His (His₆) tagged proteins by digestion with SacI and XhoI.

Transient expression in N. benthamiana

Transient expression of tested proteins was performed by transforming Agrobacterium tumefaciens GV3101 pMP90 cells harboring the respective plasmids into the abaxial side of 3- to 4-week-old N. benthamiana leaf epidermis essentially as previously described (Lavy et al., 2002, 2007; Hazak et al., 2019) with the following modifications. Cultures were grown in LB media to obtain stationary cultures, which were diluted 1:200 in Induction Medium (10 mM MES pH 5.6, 0.5% w/v Glucose, 1.734 mM NaH₂PO₄, 5% 20 X AB Salts. The 20× AB salt mixtures contained 1.2 mM NH₄Cl, 1.2 mM MgSO₄, 2 mM KCl, 0.068 mM CaCl₂, 0.81 mM FeSO₄-7H₂O, and 0.2 mM acetosyringone). The cultures were incubated at 28°C for 6 h and adjusted to optical density (OD)₆₀₀ = 0.2 for expressing a single construct. Alternatively, a mixture of plasmids was obtained by adjusting each construct to OD₆₀₀ = 0.1 for multiple constructs infiltration. For expression of multiple plasmids or in cases where expression levels were too low for detection, Agrobacterium expressing the silencing suppressor protein p19 form tomato bushy stunt virus (Sainsbury and Lomonossoff, 2014) were co-transformed added at a dilution OD₆₀₀ of 0.05. Imaging was performed 48 h after infiltration.

Transient expression in Arabidopsis cotyledons

The Agrobacterium strains were grown as described above for expression in N. benthamiana except that the induction medium was supplemented with 0.4 mM acetosyringone and 2 mg/mL Rifampicin. The cultures were grown over night to OD₆₀₀ = 0.6–0.8. Four days old Arabidopsis (A. thaliana) seedlings, grown on 0.5× Murashige and Skoog (MS) medium were submerged with Agrobacterium culture under vacuum for 2 min for consecutive 3 times. Imaging was performed after 2–3 d.

MT stability

MTs dissociation was performed by incubation of transiently transformed N. bentamiana leaves with 20 µM oryzalin for 30 min.

Protein purification

His₆-tagged ROP11, rop11D-PBR, ROP6, and rop6D-PBR have expressed in BL21 CodonPlus DE3 RIL (Stratagene San Diego, CA, USA) E.coli cells. Cells were grown to OD₆₀₀ = 0.6. Protein expression was induced by 2 mM IPTG. Protein extraction and purification on Ni-nitrilotriacetic acid (NTA) columns were performed according to the manufacturer's (QIAGEN, Hilden, Germany) instructions. Cells were grown at 37°C to OD₆₀₀ = 0.5–0.7 before induction with 2 mM IPTG overnight at 16°C. In turn, cells were harvested by centrifugation at 5,000g for 15 min at 4°C and resuspended in binding buffer (50 mM NaPO₄ pH 8.0, 300 mM NaCl, 5% Glycerol, 30 mM imidazole, 2 mM β-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1% (v/v) protease inhibitor cocktail (Sigma-Aldrich St Louis, MO, USA). Resuspended cells were broken with micro fluidizer (Microfluidics, Newton, MA, USA) at 80 PSI and centrifuged at 20,000g for 45 min at 4°C. The supernatant was collected and loaded on Nickel-NTA column. Following loading, the column was eluted with 30 column volumes of binding buffer before elution with binding buffer supplemented with 200 mM imidazole. Eluted protein fractions were collected and dialyzed overnight against phosphate-buffered saline (PBS) (140 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄ pH 7.3).

GST:RhoGDI1 and GST were purified from E.coli BL21 CodonPlus DE3 RIL (Stratagene). Cells were grown at 37°C to OD₆₀₀ = 0.5–0.7 before induction with 2 mM IPTG overnight at 16°C. Cells were harvested by centrifugation at 5,000g for 15 min at 4°C, washed with PBS and resuspended in PBS containing 1 mM PMSF and 1% (v/v) protease inhibitor cocktail (Sigma-Aldrich). The cells were broken with micro fluidizer (Microfluidics, Newton, MA, USA) at 80 PSI and centrifuged at 20,000 × g for 45 min at 4°C and resuspended in PBS. The extracts were loaded on a GST-TRAP FF column (GE Healthcare, Amersham, UK), washed with PBS, and eluted with Glutathione Elution Buffer (10 mM glutathione, 50 mM Tris–HCl pH 8.0), followed by dialysis against dialysis buffer (50 mM Tris–HCl pH 7.5, 300 mM NaCl, 5 mM MgCl₂, 10% glycerol).

Lipid binding—PIP strips assays

PIP strips (Echelon Bioscience; Cat. P-601) were blocked for 1 h with PBST Blocking solution (PBS, 0.1% (v/v) Tween-20) supplemented with 3% fatty acid free fraction 5 Bovine Serum Albumin. In turn, 0.5 µg/mL of His₆:ROP11 or His₆:ROP11D-PBR were added and incubated for 1 h. The ROP11 solution was removed and the strips were washed 3 times with PBS before incubation with mouse anti-poly-His monoclonal antibodies. The strips were washed again with PBS as above and incubated with goat anti-mouse Horse Radish Peroxidase (HRP)-conjugated secondary antibodies. Detection was carried out with MicroChemi version 6.0, DNR Bioimaging Systems.

Extraction of ROPs from the membrane by RhoGDI1

ROPs (GFP:ROP2, GFP:ROP4, GFP:ROP6, GFP:ROP10, GFP:ROP9, and GFP:ROP11) were transiently expressed in N. benthamiana leaves. About 30 mg tissue were harvested, batch frozen in liquid nitrogen, grinded with pestle and mortal and resuspended in 900 μL of plant protein extraction buffer (50 mM Tris–HCl pH 7.5, 5 mM MgCl₂, 300 mM NaCl, 10% glycerol, 2 mM β-mercaptoethanol, 1% (v/v) plant protease inhibitor mix (Sigma-Aldrich), 1 mM PMSF). After incubation with shaking for 30 min at 4°C, the extracts were centrifuged (30 min, 20,000 × g, 4°C). The supernatant included soluble proteins, while the insoluble membrane pellets were incubated in 900 μL extraction buffer for 30 min. In turn, 50 μL of the membrane extracts were incubated with 1 μg GST or GST-RhoGDI1 for 1 h. The mixtures were then fractionated by centrifugation at 20,000 × g for 30 min and both the soluble and insoluble fractions were collected for further analysis. Analysis ROPs’ distribution was performed by protein immuno blots decorated with anti-GFP monoclonal antibodies and goat anti-mouse HRP-conjugated secondary antibodies. Detection was carried out with MicroChemi version 6.0.

Yeast two-hybrid assays

Yeast two-hybrid assays were performed with Saccharomyces cerevisiae strain PJ69-4a. ROPs (ROP2, ROP4, ROP6, ROP9, ROP10, or ROP11) were subcloned into Gal4 DNA Binding Domain (Gal4_BD) vector pGBT and RhoGDI into Gal4 Activation Domain (Gal4_AD) vector pGAD. The plasmids were co-transformed into yeast cells via a standard lithium acetate transformation protocol. Four decimal dilutions of colonies expressing both plasmids were grown in spots in medium lacking either leucine (L), tryptophan (T), or L, T, and histidine (H) supplemented with 1 mM 3-Amino-1,2,4-triazole (3AT). The plates were incubated at 28°C for 2–4 d.

FRAP beam-size analysis

FRAP beam-size analysis was carried out essentially as previously described (Henis et al., 2006; Sorek et al., 2010), using a fixed-localization Gaussian laser beam focused to a single small spot on the membrane for both prebleach, bleaching, and monitoring periods, measuring the fluorescence intensity continuously by photon counting (Berkovich et al., 2011). This method has a time resolution of a few millisecond, significantly better than FRAP of larger regions-of-interest in a confocal microscope, which employs nonsynchronous stepwise and time-consuming bleaching and monitoring (Berkovich et al., 2011). The improved time resolution is critical for fast recovery contributed by exchange. Nicotiana benthamiana leaves transiently expressing GFP fusions of ROP11 or rop11CA (alone, or coexpressed with mChGEF3p and His₆:GAP1 at various combinations) were subjected to FRAP measurements carried out at 22°C on the abaxial side of the leaves. The monitoring Argon ion laser beam (488 nm, 1.2 μW, Innova 70C, Coherent) was focused through the microscope (AxioImager.D1, Carl Zeiss MicroImaging) to a spot with Gaussian radius ω = 0.77 ± 0.03 μm (×63/1.4 NA oil-immersion objective) or 1.17 ± 0.05 μm (×40/1.2 NA water immersion objective), and FRAP experiments were conducted with each beam size. The ratio between the illuminated areas, ω²(×40)/ω²(×63), was 2.28 ± 0.15 (n = 59; standard error (sem) calculated using bootstrap analysis as described below). After a brief measurement of the monitoring intensity, a 5 mW pulse (5 ms for the ×63 objective, 15 ms for the ×40 objective) bleached 50%–70% of the fluorescence in the illuminated region. Fluorescence recovery was followed by the monitoring beam. The apparent characteristic fluorescence recovery time (τ) and the mobile fraction (R_f) were derived from the FRAP curves by nonlinear regression analysis, fitting a lateral diffusion process with a single τ value (Petersen et al., 1986). The significance of differences between sets of R_f or τ values measured with the same beam size was evaluated by one-way ANOVA and Bonferroni post-hoc test or Student’s t test (depending on whether multiple samples were compared or only a pair of samples). To compare the τ(×40)/τ(×63) ratio of GFP:ROP11 under different expression conditions with the ω²(×40)/ω²(×63) beam-size ratio (comparing a single pair of data sets), we employed two-tailed Student’s t test using bootstrap analysis with 1,000 bootstrap samples (Efron and Tibshirani, 1993), which is preferable for comparison between ratios.

ROP11 expression pattern in VISUAL

Arabidopsis plants stably expressing an estradiol-inducible GFP:ROP11 under regulation of the ROP11 promoter (ROP11p::XVE-GFP:ROP11) were induced to dedifferentiate into tracheary elements using the “VISUAL” system (Kondo et al., 2016). Briefly, 10–15 seedlings were grown in 10 mL liquid Growth Media (2.2 g/L MS Basal Medium, 10 g/L Sucrose and 0.5 g/L MES adjusted to pH to 5.7 using KOH) in 6-well plates at 25°C at long day regime (16-h light, 8-h dark) for 6 d until true leaves appear. The central part of the hypocotyl was cut and roots were removed. For induction of dedifferentiation, the aerial part of the seedlings was incubated in Induction Medium (2.2 g/L MS Basal Medium containing 50 g/L D(+)Glucose adjusted to pH 5.7 with KOH) supplemented with 1.2 mg/L 2,4D, 0.25 mg/L Kinetin, and 10 μM Bikinin in 12-well plates for additional 4 d. On Day 3 GFP:ROP11 expression was induced by 25-µM estradiol.

Confocal laser scanning microscopy

Laser scanning confocal microscopy imaging was performed using LSM 780-NLO confocal laser scanning microscope (Zeiss, Jena, Germany) with ×40 or ×100 water objectives. GFP was visualized by excitation with an argon ion laser at 514 nm. Emission was detected with a spectral detector set between 517 and 628 nm. CFP was visualized by excitation with an argon ion laser at 458 nm. Emission was detected with a spectral detector set between 463 and 488 nm. mCherry was visualized by excitation with a 561 nm diode laser. Emission was detected with a spectral detector set between 578 and 628 nm. Emission detection was carried out using GaASP detector with gains ranging between 1% and 5%. The images of ROPs-GAP1-GEF3p domains are surface images on plasma membrane plane, as demonstrated in Supplemental Figure S1. The images of ICR1-ROPs-GAP1-GEF3p and Visual system are Z stuck projections. Colocalization analysis was carried out by Zeiss Zen10 software. The analysis included an overlap coefficient based on Manders colocalization analysis (Manders et al., 1993) and R.

Accession numbers

AtROP2-AT1g20090; AtROP4-AT1g75840; AtROP6-AT4g35020; AtROP9-AT4g28950; AtROP10-AT3g48040; AtROP11-AT5g62880; AtROPGEF3-AT4G00460; AtGAP1-AT5G22400; AtICR1-AT1g17140; AtRhoGDI1-AT3G07880.1.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Plasma membrane localization of the ROP-ROPGEF domains.

Supplemental Figure S2. Representative domain shapes.

Supplemental Figure S3. Co-expression of GFPROP6 and mCherryGEF3p without GAP1.

Supplemental Figure S4. Neither domains nor nanodomains formed following expression mChGAP1 alone, GFP:ROP6 or GFP:ROP11 alone or coexpressed with mChGAP1.

Supplemental Figure S5. RhoGDI does not affect domain formation by type-I ROPs.

Supplemental Figure S6. RhoGDI does not affect domain formation by type-II ROPs.

Supplemental Figure S7. Amino acid sequences of ROP6, ROP11, and their D-PBR deletion mutants.

Supplemental Figure S8. Samples of purified poly his-tagged ROP11 (R11), rop11D-PBR (r11D-PBR), ROP6 (R6), and rop6D-PBR (r6D-PBR) which were used for PIP strip assay.

Supplemental Table S1. Colocalization essay of GFPROPs and mChGEF3p domains, coexpressed with GAP.

Supplemental Table S2. Colocalization essay of GFPROP11 and mChGAP1.

Supplemental Table S3. Percentage of different shaped ROPs GAP1 GEF3p domains.

Supplemental Table S4. Source data for Figures 8C, 4, D and E.

 

H.S.: investigation, data curation, formal analysis, methodology, validation, visualization, writing-review, and editing; E.B.: investigation, methodology, and visualization; D.B.: methodology; O.G.: investigation, data curation, formal analysis, and methodology; Y.H.: formal analysis, validation, writing-review, and editing; S.Y.: conceptualization, formal analysis, methodology, validation, supervision, project administration, funding acquisition, visualization, writing-original draft, writing-review, and editing.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/General-Instructions) is Shaul Yalovsky (shauly@tauex.tau.ac.il).