A cysteine-rich receptor-like kinase NCRK and a pathogen-induced protein kinase RBK1 are Rop GTPase interactors


  • Arthur J. Molendijk,

    Corresponding author
    1. Institute for Biologie II/Botany, Faculty of Biology, Albert-Ludwigs-University Freiburg, Schänzlestrasse 1, 79104 Freiburg, Germany,
      (fax +49 761 203 2675; e-mail arthur.molendijk@biologie.uni-freiburg.de or klaus.palme@biologie.uni-freiburg.de).
    Search for more papers by this author
  • Benedetto Ruperti,

    1. Institute for Biologie II/Botany, Faculty of Biology, Albert-Ludwigs-University Freiburg, Schänzlestrasse 1, 79104 Freiburg, Germany,
    2. Dipartimento di Scienze Agrarie e Ambientali, University of Udine, via delle Scienze 208, 33100 Udine, Italy,
    Search for more papers by this author
  • Manoj K. Singh,

    1. Institute for Biologie II/Botany, Faculty of Biology, Albert-Ludwigs-University Freiburg, Schänzlestrasse 1, 79104 Freiburg, Germany,
    Search for more papers by this author
  • Alexander Dovzhenko,

    1. Institute for Biologie II/Botany, Faculty of Biology, Albert-Ludwigs-University Freiburg, Schänzlestrasse 1, 79104 Freiburg, Germany,
    Search for more papers by this author
  • Franck A. Ditengou,

    1. Institute for Biologie II/Botany, Faculty of Biology, Albert-Ludwigs-University Freiburg, Schänzlestrasse 1, 79104 Freiburg, Germany,
    Search for more papers by this author
  • Mattia Milia,

    1. Institute for Biologie II/Botany, Faculty of Biology, Albert-Ludwigs-University Freiburg, Schänzlestrasse 1, 79104 Freiburg, Germany,
    2. Dipartimento di Scienze Agrarie e Ambientali, University of Udine, via delle Scienze 208, 33100 Udine, Italy,
    Search for more papers by this author
  • Lore Westphal,

    1. Leibniz Institute of Plant Biochemistry, 06120 Halle, Germany,
    Search for more papers by this author
  • Sabine Rosahl,

    1. Leibniz Institute of Plant Biochemistry, 06120 Halle, Germany,
    Search for more papers by this author
  • Tim-Robert Soellick,

    1. Novartis Institutes for BioMedical Research, Biologics Center, 4002, Basel, Switzerland,
    Search for more papers by this author
  • Joachim Uhrig,

    1. Botanisches Institut, Cologne University, Gyrhofstrasse 15, 50931 Köln, Germany, and
    Search for more papers by this author
  • Lars Weingarten,

    1. Department of Molecular Immunology, Institute for Biologie III, University of Freiburg and Max-Planck-Institute for Immunobiology, Stübeweg 51, 79108, Freiburg, Germany
    Search for more papers by this author
    • Present address: Research Group Redoxregulation, German Cancer Research Center, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany.

  • Michael Huber,

    1. Department of Molecular Immunology, Institute for Biologie III, University of Freiburg and Max-Planck-Institute for Immunobiology, Stübeweg 51, 79108, Freiburg, Germany
    Search for more papers by this author
  • Klaus Palme

    Corresponding author
    1. Institute for Biologie II/Botany, Faculty of Biology, Albert-Ludwigs-University Freiburg, Schänzlestrasse 1, 79104 Freiburg, Germany,
      (fax +49 761 203 2675; e-mail arthur.molendijk@biologie.uni-freiburg.de or klaus.palme@biologie.uni-freiburg.de).
    Search for more papers by this author

(fax +49 761 203 2675; e-mail arthur.molendijk@biologie.uni-freiburg.de or klaus.palme@biologie.uni-freiburg.de).


In plants, Rop/Rac GTPases have emerged as central regulators of diverse signalling pathways in plant growth and pathogen defence. When active, they interact with a wide range of downstream effectors. Using yeast two-hybrid screening we have found three previously uncharacterized receptor-like protein kinases to be Rop GTPase-interacting molecules: a cysteine-rich receptor kinase, named NCRK, and two receptor-like cytosolic kinases from the Arabidopsis RLCK-VIb family, named RBK1 and RBK2. Uniquely for Rho-family small GTPases, plant Rop GTPases were found to interact directly with the protein kinase domains. Rop4 bound NCRK preferentially in the GTP-bound conformation as determined by flow cytometric fluorescence resonance energy transfer measurements in insect cells. The kinase RBK1 did not phosphorylate Rop4 in vitro, suggesting that the protein kinases are targets for Rop signalling. Bimolecular fluorescence complementation assays demonstrated that Rop4 interacted in vivo with NCRK and RBK1 at the plant plasma membrane. In Arabidopsis protoplasts, NCRK was hyperphosphorylated and partially co-localized with the small GTPase RabF2a in endosomes. Gene expression analysis indicated that the single-copy NCRK gene was relatively upregulated in vasculature, especially in developing tracheary elements. The seven Arabidopsis RLCK-VIb genes are ubiquitously expressed in plant development, and highly so in pollen, as in case of RBK2. We show that the developmental context of RBK1 gene expression is predominantly associated with vasculature and is also locally upregulated in leaves exposed to Phytophthora infestans and Botrytis cinerea pathogens. Our data indicate the existence of cross-talk between Rop GTPases and specific receptor-like kinases through direct molecular interaction.


Plant receptor-like kinases (RLKs) are divided into three large related groups of proteins, namely receptor kinases, receptor-like cytosolic kinases (RLCKs), which lack the extracellular receptor domains, and receptor-like proteins (RLPs), which lack a protein kinase domain (Shiu and Bleecker, 2003). The RLK gene families in Arabidopsis and rice have been classified according to protein kinase domain and extracellular receptor domain homologies, with 45 subfamilies in Arabidopsis (Shiu and Bleecker, 2001, 2003; Shiu et al., 2004). Most receptor kinases and RLPs have no known ligands, and regulation of receptor kinase activity and downstream signalling is known in detail for only a few RLKs. Receptor kinases are implicated in many aspects of plant biology including growth and development, self-incompatibility in pollination, plant–microbe interactions, and defence responses (Johnson and Ingram, 2005). Receptor-like cytosolic kinases and RLPs are thought to associate with partner receptor kinases to build functional complexes, and function in self-incompatibility (Murase et al., 2004), plant development and pathogen defence (Muto et al., 2004; Rathjen et al., 1999; Rowland et al., 2005; Veronese et al., 2006).

Rops constitute a plant-specific group of Rho small GTPases corresponding to the Rac/Cdc42 clade in yeast and animals, and are crucial for cell growth, cell morphogenesis and pathogen defence. Rops control cell polarity in tip-growing cells and lobe formation of leaf epidermal pavement cells through regulation of cytoskeletal dynamics (reviewed by Gu et al., 2003, 2004; Molendijk et al., 2004; Vernoud et al., 2003). Rops also regulate the production of reactive oxygen species (ROS; Park et al., 2000; Potikha et al., 1999), which is generally required during plant growth (Foreman et al., 2003; Liszkay et al., 2004) and pathogen defence responses (Agrawal et al., 2003; Kawasaki et al., 1999; Ono et al., 2001; Park et al., 2004). Several Rop genes are upregulated during xylogenesis (Brembu et al., 2005; Nakanomyo et al., 2002) and function in lignification (Kawasaki et al., 2006). Recently, Berken et al. (2005) and Gu et al. (2006) identified a plant-specific family of Rop GDP/GTP exchange factors (RopGEFs) called PRONE/KPP, which may link specific RLKs via Rop activation to regulation of cytoskeletal dynamics (Berken, 2006; Kaothien et al., 2005; Nibau et al., 2006; Uhrig and Hulskamp, 2006). Functional specificity of GTP-bound Rop-containing complexes is determined by the associated binding partners, which need to be functionally characterized in order to understand the downstream effects of Rop activation. Known molecular targets of Rop GTPases belong to different functional categories, namely regulation of cytoskeletal dynamics, exocytosis, cell wall biosynthesis and ROS production (reviewed by Basu et al., 2004; Kawasaki et al., 2006; Lavy et al., 2007; Molendijk et al., 2004). In addition, Rop has also been identified as a component of the active CLV1 signalling complex and found to be associated with mitogen-activated protein kinase (MAPK) complexes, but the precise molecular interactions in these complexes are unknown (Lieberherr et al., 2005; Trotochaud et al., 1999). Animal and yeast Rac/Cdc42 GTPases target protein kinases of the Ste20/Pak group by binding to specific GTPase-binding domains (GBDs) containing the Cdc42/Rac interactive binding (CRIB) motif (reviewed by Zhao and Manser, 2005). Although these GBDs are present in plant RopGAPs and the small RIC proteins which regulate cytoskeletal dynamics (Fu et al., 2005; Wu et al., 2000, 2001), they are not contained in any Arabidopsis protein kinases, and protein kinases have not yet been identified as direct molecular interactors of Rops in plants.

To determine whether Rop signalling might rely on protein kinase coupling, we used yeast two-hybrid screening to obtain Rop protein kinase interactors and proceeded with their functional characterization. We identified a novel cysteine-rich receptor kinase (NCRK) and two group VI receptor-like cytosolic kinases RBK1 and RBK2 (Rop Binding protein Kinases) as direct molecular interactors of Rop GTPases. Rops interacted directly with the protein kinase domains, and binding did not require a full size catalytically active protein kinase. The NCRK protein was hyperphosphorylated and localized to endosomes in plant protoplasts, indicating efficient activation and internalization from the plasma membrane. The occurrence of a transient interaction with Rop GTPase at the plasma membrane, could be demonstrated for NCRK and RBK1 using BiFC assays. Gene expression data suggested roles for NCRK during tracheary element differentiation and for RBK1 during pathogen defence.


Molecular interaction of Rop GTPases with RLK kinase domains

To identify molecular interactors of Rop GTPases, yeast two-hybrid screening was performed on an Arabidopsis cell suspension culture cDNA library according to Soellick and Uhrig (2001), using constitutively active and prenylation-deficient AtRop4 G15V C193S as a bait. Several receptor-like protein kinase cDNAs were obtained belonging to the RLCK-VI family according to the classification by Shiu and Bleecker (2001). One partial clone, named RBK1-322 corresponded to At5g10520 and two partial clones, named RBK2-311a and b to At3g05140, respectively. Another partial clone corresponded to an unclassified receptor kinase gene NCRK (At2g28250), named NCRK-313. Prey kinase clones did not interact with a nucleotide-free and prenylation-deficient bait AtRop4 T20N C193S, used as a negative control (Figure S1). Sequencing showed that RBK1-322 and RBK2-311 contained complete kinase domains and C-termini, whereas NCRK-313 contained a partial kinase domain from subdomain V to XI and ended within the variable C-terminus outside of the catalytic domain (Figure 1a). Proper expression of bait AtRop4 and prey kinase clones was confirmed by western analysis of yeast whole cell extracts using Rop4 monoclonal antibodies and anti-HA antibodies, respectively (Figure 1b). C-terminal deletion of kinase subdomains IX–XI from NCRK-313 or subdomain XI from RBK1-322 abolished interaction with AtRop4 in the yeast two-hybrid system, while the C-termini outside the catalytic domains of RBK1-322 and RBK2-311 were not sufficient for interaction (data not shown). Using the NCRK-313 clone as a bait in a second screening, three full-length independent AtRop11 clones (At5g62880) were among several prey clones pulled out from the cell suspension culture cDNA library. The AtRop11 isoform was found also to interact with RBK1-322 and RBK2-311 in yeast two-hybrid assays (data not shown). Prey AtRop11 in yeast does not contain the G15V mutation and is therefore probably mainly in the GDP-bound form, indicating that the observed difference in interaction of the prey kinases to G15V and T20N mutants of AtRop4 baits might primarily be caused by the predicted lack of guanine nucleotide binding caused by the T20N mutation (Gibson and Wilson-Delfosse, 2001). Using an in vitro binding assay, the interaction of Rops and RLKs was examined for the RBK1 clone 322 protein. His(6×)-tagged RBK1-322 protein bound equally well to either GDP or GTPγS-loaded GST-AtRop4 (Figure 1c). In contrast, a maltose-binding protein (MBP) fusion of a GBD derived from AtGAP2 gene At4g03100 bound more efficiently to GTPγS loaded GST-Rop4. Binding of RBK1-322 protein to either GDP or GTPγS-bound AtRop4 could also be demonstrated using a protein autophosphorylation assay (Figure 1c). This indicated that binding of the RBK1-322 protein to AtRop4 was independent of the activation state of the GTPase.

Figure 1.

 Rop GTPase-interacting receptor-like kinases (RLKs).
(a) Schematic representation of RLK clones obtained from yeast two-hybrid screening: S, secretory signal sequence; T, transmembrane sequence; KD, kinase domain. Arrows indicate the 5′ and 3′ ends of prey clones.
(b) Western analysis of yeast strains co-expressing combinations of Rop bait and RLK prey proteins.
(c) In vitro binding assay of AtRop4 and RBK1 clone 322. GST-AtRop4 was affinity-purified and loaded with either GDP or GTP-γS. Pull-down assays were performed using affinity-purified maltose-binding protein (MBP)-tagged GTPase binding domain (GBD) or His(6×)-tagged RBK1 322 protein. Binding was detected by western blot analysis using MBP and His(6×) antibodies or by phosphorylation assay using 32P γATP.

NCRK contains a novel cysteine-rich receptor domain

The extracellular receptor domain encoded by NCRK genes contains two repeats of about 40 amino acids sharing a novel conserved cysteine-rich motif, WXCXCX13–18CX3CXC (Figure S2a). The C-terminus encoded by NCRK-like sequences from different plant species such as the magnoliids Persea americana and Liriodendron tulipifera and the conifers Cryptomeria japonica and Picea glauca contains an absolutely conserved DLTEPR core motif (Figure S2b), within an acidic PEST-like sequence (Rechsteiner and Rogers, 1996). The RLCK-VI gene subfamily in Arabidopsis consists of two separate clades of seven genes each (Shiu and Bleecker, 2001), with RBK1 and RBK2 belonging to the same clade, here named RLCK-VIb. RLCK-VIb homologous genes were represented by expressed sequence tags (ESTs) from the gymnosperms Cycas rumphii and different conifers (not shown) indicating that this group of RLCKs is likely to be conserved in all seed plants.

Preferential binding of GTP-bound AtRop4 to NCRK in insect cells

We decided to study the molecular interaction between Rop GTPase and NCRK by flow-cytometric fluorescence resonance energy transfer (FRET) assay in insect cells. This method has the advantage over in vitro binding assays of determining the interaction of full-length trans-membrane NCRK with prenylated AtRop4 in a more natural membrane environment. The combination of a YFPv-tagged GBD derived from AtGAP2 gene At4g03100 and enhanced cyan fluorescent protein (ECFP)-tagged AtRop4 G15V was used as a positive control FRET pair (Figure 2a). Comparable high FRET efficiency was also measured with a control CFP–yellow fluorescent protein (YFP) fusion protein. Low but significant FRET was observed between NCRK-YFPv and either wildtype or G15V ECFP-AtRop4. The interaction with the G15V mutant version was consistently higher than with wildtype AtRop4, indicating a preferential binding of NCRK to GTP-bound AtRop4. Fluorescence resonance energy transfer between AtRop4 T20N and NCRK did not exceed background levels, in agreement with a lack of interaction observed in yeast two-hybrid assays. Histogram analysis of the cells in the FRET channel showed a shift of the distribution maximum to higher signal strength when using ECFP-AtRop4 G165V (Figure S5). C-terminal deletion of NCRK protein kinase subdomains IX–XI, also abolished FRET with AtRop4 as expected from yeast two-hybrid results (Figure 2a). Proper expression of YFPv- and ECFP-tagged proteins was confirmed by the levels of fluorescence observed in FRET analysis and by western analysis of total transfected insect cell lysates using GFP monoclonal antibodies, indicating comparable levels of expression for ECFP-Rops and NCRK-YFPv, and higher levels of expression for cytosolic YFPv-GBD (Figure 2b). Yellow fluorescent protein serves as an acceptor fluorophor and thus lower expression levels of the transmembrane protein NCRK-YFPv correlated well with lower FRET efficiency. When transformed into plant protoplasts, NCRK-YFPv displayed a large range of mobility on low-percentage SDS-PAGE due to hyperphosphorylation (see below), but not so in insect cells, possibly due to the absence of a plant-specific ligand required for kinase activation. NCRK remained unphosphorylated in the presence of interacting constitutively active AtRop4 G15V, indicating that GTP-bound Rop was not sufficient to activate NCRK in insect cells (Figure 2c).

Figure 2.

 Fluorescence resonance energy transfer (FRET) analysis of molecular interactions between enhanced cyan fluorescent protein (ECFP) and yellow fluorescent protein (YFP)v-tagged Arabidopsis proteins expressed in insect cells.
(a) Percentage of transfected insect cells detected in the FRET channel for different combinations as indicated: CFP-YFP, intramolecular FRET control; 1, ECFP-Rop4 T20N; 2, ECFP-Rop4 wt; 3, ECFP-Rop4 G15V; 5, YFPv-GBD wt; 6, NCRK-YFPv with C-terminal deletion starting from kinase subdomain IX; 7, NCRK-YFPv. Values are means of duplicate experiments with SE.
(b) A GFP western blot of a 15% SDS-PAGE gel of total lysates of co-transfected insect cells used in the FRET experiment. Lane 1, YFPv-GBD wildtype (wt) and ECFP-Rop4 G15V; lane 2, NCRK-YFPv and ECFP-Rop G15V. Protein size markers as indicated.
(c) A GFP western blot of a 6% SDS-PAGE showing NCRK-YFPv proteins expressed in co-transfected insect cells (lanes 1, 2, 3) and transformed Arabidopsis protoplasts (lane 4). Lane 1, NCRK-YFPv and ECFP-Rop4 T20N; lane 2, NCRK-YFPv and ECFP-Rop4 wt; lane 3, NCRK-YFPv and ECFP-Rop4 G15V; lane 4, NCRK-YFPv.

To determine the possibility of the Rop GTPases being RLK substrates, we used immunoprecipitated NCRK-YFPv and RBK1-YFPv expressed in insect cells in protein kinase assays with added recombinant Rop4 protein. Autophosphorylation activity was detected for RBK1, but not for NCRK (not shown). Cross-phosphorylation of either GDP- or GTPγS-loaded Rop4 by RBK1 did not exceed background levels (Figure S3), suggesting that Rop4 is not a RBK1 substrate.

Rop4 GTPase interacts with NCRK and RBK1 kinases in plant protoplasts

To demonstrate in vivo interaction between Rop4 and NCRK and RBK1 kinases, we used the bimolecular fluorescence complementation (BiFC) method, which is well suited to monitoring transient molecular interactions in plants (Bhat et al., 2006). In co-transformed Arabidopsis seedling protoplasts, split YFP-tagged GTP-bound Rop4 G15V interacted with both split YFP-tagged NCRK and RBK1 kinases (Figure 3a), as evident from the presence of reconstituted YFP fluorescence at the plasma membrane. In contrast, YFP-tagged NCRK and RBK1 localized predominantly to endomembrane compartments (Figure 3b). Nucleotide-free Rop4 T20N was relatively ineffective in BiFC, whereas comparable amounts of the split YFP proteins could be detected by immunoblot analysis (not shown). The observation that the T20N mutation in Rop4 led to a decreased BiFC signal, in agreement with the yeast two-hybrid and FRET data, indicated that specific molecular interactions between Rop4 and the kinases caused the BiFC effect in protoplasts.

Figure 3.

 Bimolecular fluorescence complementation (BiFC) analysis of molecular interactions between Rop4 GTPase mutants and NCRK and RBK1 kinases expressed in Arabidopsis seedling protoplasts.
(a) Percentage of protoplasts with higher BiFC signals for eight different combinations as indicated. Combinations A are yellow fluorescent protein (YFP)n-Rop4 with Kinase-YFPc, whereas combinations B areYFPc-Rop4 with Kinase-YFPn. Each percentage represents the mean value of four transformation experiments. Standard errors are shown, except where too low.
(b) Epifluorescence microscopy of BiFC-positive combinations, each field including one positive cell with plasma membrane localized BiFC. Insets, transformed protoplasts showing localization of YFP-tagged NCRK (upper panels) and RBK1 (lower panels) proteins.

NCRK partially localizes to RabF2a positive endosomes in plant protoplasts

The intracellular localization of YFPv-tagged kinases was examined in transformed plant protoplasts using a plant expression plasmid that allowed relatively high expression levels (Reichel et al., 1996). In tobacco mature leaf protoplasts, NCRK localized to a few (one to ten) large globular endomembrane compartments of 2–4 μm diameter with visible lumen. In protoplasts derived from Arabidopsis cell suspension, NCRK-YFPv was mainly associated with numerous smaller vesicles, although a few larger compartments were present as well (Figure 4). A few cells with low amounts of plasma membrane-localized NCRK-YFPv could also be observed (not shown). In Arabidopsis protoplasts, the larger NCRK positive endomembrane compartments and a subpopulation of the NCRK-positive small vesicles were also positive for ECFP-RabF2a used commonly as a marker for early endosomes and prevacuolar components (PVCs) (Ueda et al., 2004). ECFP-RabF2a was present in endomembranes and also showed a diffuse fluorescence consistent with the cycling of Rab GTPases between membranes and cytosol (Figure 4h,k), whereas NCRK-YFPv was confined to membranes. The GFP-AtRop4 localized to both plasma membrane and cytosol, while ECFP-AtRop11 was predominantly associated with the plasma membrane (Figure 4), probably related to the different lipid modifications predicted for these Rop isoforms. The presence of NCRK in early endosomes and PVCs indicated that the protein was internalized from the plasma membrane, suggesting that molecular association of Rop GTPases and NCRK might occur transiently at the plasma membrane. Dominant negative AtRop4 T20N did not change the endomembrane localization of NCRK, indicating that active GTP-bound Rops are not required for internalization of NCRK (Figure 5). Co-expression of NCRK with wildtype or constitutively active Rops resulted in retention of NCRK in a perinuclear membrane compartment (Figure 5), which partially co-localized with the endoplasmic reticulum (ER) and Golgi marker ER-CFP (not shown). In transgenic Arabidopsis expressing NCRK-YFPv under the control of the 35S promoter, the receptor kinase was also exclusively present in endomembranes, including the globular compartments of diameter up to 5 μm with visible lumen, showing that NCRK internalization is not specific for protoplasts (Figure S4).

Figure 4.

 Localization of NCRK and RBK1 in transformed plant protoplasts.
Localization of NCRK, RBK1, Rop4, Rop11 and RabF2a. (a–c) Tobacco leaf protoplasts and (d–l) Arabidopsis cell suspension protoplasts.
(a–c) Fluorescent protein images merged with chloroplast autofluorescence (red) images. (a) GFP-Rop4; (b) NCRK-YFPv; (c) RBK1-YFPv.
(d) GFP-Rop4; (e) ECFP-Rop11; (f) RBK1-YFPv.
(g–i) partial co-localization in cell co-transformed with NCRK-YFPv (g) and ECFP-RabF2a (h); (i) merged images (g) and (h).
(j–l) More complete co-localization in cell co-transformed with NCRK-YFPv (j) and ECFP-RabF2a (k); (l) merged images (j) and (k).

Figure 5.

 Localization of NCRK and RBK1 in Arabidopsis cell suspension protoplasts co-transformed with Rop GTPases.
(a) RBK1-YFPv (1), ECFP-Rop4 T20N (2), merged (3).
(b) RBK1-YFPv (1), ECFP-Rop4 G15V (2), merged (3).
(c) NCRK-YFPv (1), ECFP-Rop4 T20N (2), merged (3).
(d) NCRK-YFPv (1), ECFP-Rop4 G15V (2), merged (3).
(e) NCRK-YFPv (1), Rop11-ECFP (2), merged (3).

RBK1 localizes to cytosol and endomembranes

RBK1-YFPv was mainly cytosolic in tobacco leaf protoplasts, as indicated by a diffuse fluorescence (Figure 4), but was also associated with endomembrane compartments in Arabidopsis protoplasts, with occasional nuclear localization as shown (Figure 5). Since RBK1 is predicted from its sequence to be a cytosolic protein this indicated the presence of endogenous endomembrane-localized RBK1-interacting proteins in Arabidopsis protoplasts. Co-expression of RBK1 with either dominant negative or constitutively active Rop mutants had no effect on localization of RBK1 (Figure 5). RBK1-YFPv and ECFP-Rops did not co-localize at the plasma membrane, which was indicated as the site of interaction in the BiFC experiments, suggesting the molecular interaction at the plasma membrane was transient.

The ER localized NCRK is unphosphorylated, while endosomal NCRK is hyperphosphorylated

Tagged proteins were correctly expressed in plant protoplasts (Figure 6a). In contrast to NCRK-YFPv expressed in insect cells, plant NCRK-YFPv showed a large range of electrophoretic mobility on low-percentage SDS-PAGE gels (Figure 6b). This mobility shift could be eliminated by lambda protein phosphatase treatment, indicating the protein was mostly hyperphosphorylated when expressed in plant protoplasts (Figure 6c, lanes 1 and 2). To determine a possible effect of Rops on regulation of kinase activity, we examined the phosphorylation status of NCRK in protoplasts co-transformed with RopGAP and different Rop constructs (Figure 6b). Co-transformation experiments with either dominant negative ECFP-Rop4 T20N or RopGAP2, which are predicted to deplete the cell of endogenous active Rops, indicated that Rop-GTP is not necessary for hyperphosphorylation of NCRK (Figure 6b, lanes 3 and 4). When NCRK-YFPv was co-expressed with either wildtype or constitutively active ECFP-Rops, the mobility of NCRK-YFPv in SDS PAGE was similar to protein phosphatase-treated NCRK-YFPv from single transformations (Figure 6c, lanes 3 and 4), and was resistant to protein phosphatase treatment (Figure 6c, lanes 5 and 6), indicating that NCRK remained inactive under these conditions, where it was retained in the ER. The mobility range for RBK1 was smaller and more difficult to resolve on 6% acrylamide gels, but lower-mobility forms were visible as a ladder, suggesting that RBK1 was also partly phosphorylated in plant protoplasts irrespective of the presence of co-expressed Rop mutants (Figure 6d).

Figure 6.

 Phosphorylation status of NCRK and RBK1 in transformed Arabidopsis protoplasts.
(a) GFP immunoblot of transformed protoplast extracts separated on 10% SDS-PAGE. Lane 1, RBK1-YFPv; lane 2, RBK1-YFPv + ECFP-Rop4 T20N; lane 3, RBK1-YFPv + ECFP-Rop4 G15V; lane 4, NCRK-YFPv + ECFP-Rop4 T20N; lane 5, NCRK-YFPv + ECFP-Rop4 G15V. Molecular weight markers as indicated.
(b) GFP immunoblot of transformed protoplast extracts separated on 6% SDS-PAGE showing NCRK-YFPv; lane 1, NCRK-YFPv control expressed in insect cells; lane 2, NCRK-YFPv; lane 3, NCRK-YFPv + RopGAP2; lane 4, NCRK-YFPv + ECFP-Rop T20N; lane 5, NCRK-YFPv + ECFP-Rop G15V. Lower-mobility phosphorylated kinases are indicated with asterisk.
(c) GFP immunoblot of λ PPase-treated transformed protoplast extracts separated on 6% SDS-PAGE showing NCRK-YFPv. Lane 1, NCRK-YFPv –λ PPase; lane 2 NCRK-YFPv + λ PPase; lane 3, NCRK-YFPv + λ PPase; lane 4, NCRK-YFPv + ECFP-Rop4 G15V –λ PPase; lane 5, NCRK-YFPv + ECFP-Rop4 G15V –λ PPase; lane 6, NCRK-YFPv + ECFP-Rop4 G15V + λ PPase. Lower-mobility phosphorylated kinases are indicated with asterisk.
(d) GFP immunoblot of transformed protoplast extracts separated on 6% SDS-PAGE showing RBK1-YFPv and NCRK-YFPv. Lane 1, RBK1-YFPv; lane 2, RBK1-YFPv + ECFP-Rop4 T20N; lane 3, RBK1-YFPv + ECFP-Rop4 G15V; lane 4, NCRK-YFPv. Lower-mobility phosphorylated kinases are indicated with asterisk.

Gene expression analysis of Rop-interacting protein kinases

Examination of public Arabidopsis micro-array databases indicated that RLCK-VIb genes are ubiquitously expressed, with several members of the gene family especially highly upregulated during differentiation of tracheary elements and in pollen. The single-copy NCRK gene was also highly expressed during differentiation of tracheary elements but not in pollen (Table S1). Transgenic Arabidopsis lines containing an NCRK promoter GUS fusion construct were used to analyse gene expression in 1–3-week-old seedlings. The NCRK gene appeared ubiquitously expressed, with the highest levels of GUS activity in leaf primordia, in the root apical meristem, in lateral root primordia, and in the stele of older roots and hypocotyls (Figure 7a,d–f). In cotyledons and leaves, expression was highest in vasculature and hydathode endothem (Figure 7a–c). Trichomes expressed relatively high levels of GUS activity (Figure 7a). In situ mRNA detection on 6-day-old Arabidopsis seedling sections showed that NCRK gene expression was upregulated in the apical shoot meristem and young leaf primordia (Figure 7g,h), and confirmed higher levels of gene expression associated with vascular strands as shown for cotyledon and hypocotyl sections (Figure 7g–i).

Figure 7.

 Expression analysis using NCRK promoter GUS and in situ mRNA detection in Arabidopsis seedlings.
(a–f) GUS detection in Arabidopsis seedlings: (a) primary leaves and shoot apical region; (b) leaf; (c) leaf vascular strands; (d) and (e) lateral root primordia; (f) primary root tip.
(g–i) In situ NCRK mRNA detection on sections of 6-day old Arabidopsis seedlings: (g) cotyledons and shoot apical region; (h) magnified showing leaf primordia and shoot apical meristem; (i) cotyledon with cross-cut vascular strand.
Bar = 100 μm in (d)–(f); 200 μm in (a), (g–i); 400 μm in (b), 50 μm in (c).

Arabidopsis RBK1 promoter GUS seedlings indicated relatively high expression levels in vasculature, hydathode endothem, leaf mesophyll cells and trichomes. In contrast to NCRK, the RBK1 gene was not expressed at high levels in the root apical meristem and lateral root primordia (Figure 8e,f). In situ mRNA detection experiments confirmed the relatively high levels of RBK1 gene expression associated with vasculature, as shown for cotyledon and hypocotyl sections (Figure 8g–i).

Figure 8.

 Expression analysis using RBK1 promoter GUS and in situ mRNA detection in Arabidopsis seedlings.
(a)–(f) GUS detection in Arabidopsis seedlings: (a) cotyledons, and primary leaves with GUS-positive trichomes; (b) cotyledon and primary leaves with GUS-positive vasculature; (c) leaf with GUS-positive vasculature, trichomes, and hydathodes; (d) leaf vascular strand; (e) lateral root primordium; (f) primary root tip.
(g–i) In situ RBK1 mRNA detection on sections of 6-day old Arabidopsis seedlings: (g) cross-section of cotyledon; (h) cotyledon, cross-cut vascular strand; (i) longitudinally cut hypocotyl with vascular strand.
Bar = 200 μm in (a–c), (g–i); 100 μm in (d–f).

Locally induced expression of the RBK1 gene upon pathogen exposure

Public microarray data indicated strong upregulation of RBK1 gene expression in leaves exposed to oomycete and fungal pathogens (Table S1). Since several RLK genes have been functionally characterized in relation to pathogen defence, we decided to study this phenomenon in more detail. When Phytophthora infestanszoospores were applied to Arabidopsis rosette leaves, RBK1 mRNA accumulated peaking at 6 h after infection as demonstrated by northern analysis (Figure 9a). Independent RBK1 promoter GUS lines showed local induction of gene expression at the sites of P. infestans application (Figure 9b). Local application of Botrytis cinerea spores resulted in RBK1 mRNA upregulation at 24 h after inoculation as measured by real-time RT-PCR (Figure 9c), and likewise resulted in local induction of GUS activity in RBK1 promoter GUS lines (Figure 9d). A general ethylene responsiveness of the RBK1 promoter has been shown by Millenaar et al. (2005) and Zimmermann et al. (2004), and could be confirmed in RBK1 promoter GUS lines upon ethylene treatment as an induction of transcription of the gene in the vascular cylinder of primary roots and in the root cortex in the root/hypocotyl junction zone (Figure 9e). Upregulation of RBK1 transcription during B. cinerea exposure could be reduced to levels comparable to those detected in the mock-treated samples by the inhibitor of ethylene perception 1-MCP (Figure 9c), suggesting that ethylene production during B. cinerea attack was a major RBK1-inducing signal.

Figure 9.

 Expression analysis of RBK1 in Arabidopsis leaves challenged with Phytophthora infestans or Botrytis cinerea pathogens.
(a) RNA isolated from leaves of 5-week-old Arabidopsis plants, which had been either mock-treated (‘H2O’) or inoculated with a zoospore suspension of P. infestans (5 × 105 spores ml−1, ‘P. inf ’), was subjected to northern analysis. Hybridization was carried out with a radioactively labelled fragment of RBK1. The ethidium bromide stained gel is shown as a loading control.
(b) Transgenic Arabidopsis plants carrying the RBK1 promoter GUS construct were inoculated with a zoospore suspension of P. infestans. Duplicate leaves were stained for GUS activity 6 or 24 h after infection.
(c) Real-time PCR on cDNA obtained from Arabidopsis thaliana Columbia (WT) exposed to B. cinerea spores or mock (PDB) controls. Treatment with 1-methylcyclopropene (1-MCP; 10 p.p.m.) was given 8 h before inoculation and throughout the experiment; the y-axis indicates arbitrary expression units; error bars indicate standard deviation.
(d) Transgenic Arabidopsis plants carrying the RBK1 promoter GUS construct were mock-treated or inoculated with a spore suspension of B. cinerea (5 × 105 spores ml−1) strain SAS 56 for 24 h and assayed for GUS activity.
(e) Root tips and hypocotyl/root junctions of transgenic Arabidopsis RBK1 promoter GUS seedlings treated with 10 p.p.m. ethylene or air for 24 h.


In this study we identified a cysteine-rich receptor kinase NCRK and two RLCK-VIb cytosolic kinases RBK1 and RBK2 as novel molecular interactors of Rop GTPases by means of yeast two-hybrid screening. Plant Rops are unique among the Rho GTPases in interacting directly with receptor protein kinase domains. A molecular interaction of full-length NCRK and Rop GTPase was confirmed by flow cytometric FRET analysis of YFPv and ECFP-tagged proteins expressed in insect cells. The occurrence of the Rop/NCRK and Rop/RBK1 interactions in plant cells was demonstrated by BiFC assay, which established the plasma membrane as the site of interaction. The NCRK gene is highly expressed during tracheary differentiation, while RBK1 gene expression is upregulated by pathogen exposure and ethylene. Our data indicate that different groups of Rop GTPases have a general potential to interact with specific RLKs, and mark these previously uncharacterized protein kinases as participating in Rop-regulated aspects of plant development and pathogen defence.

Consequences of the Rop/RLK molecular interactions

Rho-associated kinases are an important class of Rho effectors in yeast and animal cells, whereas target kinases for Rops in plants have not yet been found. Rho GTPases commonly bind to conserved binding domains outside the catalytic domain of their target kinases (Zhao and Manser, 2005), which is not a feature of the Rop/RLK interaction. An important question is whether these RLKs might phosphorylate Rops or whether Rops might modulate RLK signalling. The presence of conserved putative phosphorylation sites in Rop GTPases has been noted previously, and the possibility of Rops being substrates for RLK kinase activity has been suggested by Agrawal et al. (2003). However, the lack of efficient phosphorylation of Rop-GDP and Rop-GTP by RBK1 as observed in in vitro kinase assays argued against Rop being a natural substrate for phosphorylation by RLCK-VIb kinases, even though this will have to be clarified by in vivo experiments. Although yeast two-hybrid results and pull-down experiments showed that both GDP- and GTP-bound Rops are able to interact with NCRK and RLCK-VIb proteins, FRET experiments with full-size NCRK and prenylated Rop in a membranous environment indicated a preferential interaction with the GTP-bound conformation, suggesting that GTP-bound Rop might be the preferred interactor in vivo. NCRK expressed in insect cells, with or without co-expressed constitutively active Rop GTPase, was not hyperphosphorylated, indicating that GTP-Rop binding is not sufficient for NCRK activation and also suggesting the absence of the relevant activating plant ligand.

NCRK localization and activation in plant protoplast

Partial co-localization of NCRK with the endosomal and PVC marker RabF2a (Ueda et al., 2004) in co-transformed plant protoplasts indicated that the NCRK protein is internalized from the plasma membrane. NCRK was hyperphosphorylated in transformed plant protoplasts but not in insect cells, which indicated the presence of a functional activating stimulus in protoplasts, presumably a ubiquitous higher plant ligand. Since Rops and NCRK expressed in protoplasts were not co-localized, it is likely that a transient molecular interaction of Rop and NCRK preceded internalization of NCRK from the plasma membrane, as was also indicated by the occurrence of BiFC at the plasma membrane. Depletion of active endogenous Rop in protoplasts co-transformed with NCRK and either dominant negative Rop4 T20N or RopGAP2 did not inhibit NCRK hyperphosphorylation, indicating that receptor kinase activation did not require Rop GTPase. When NCRK-YFPv and wildtype or constitutively active CFP-Rops were co-expressed from plasmids that allowed high expression levels, NCRK-YFPv was retained in the ER. We think this protein trafficking blockage might be due to ectopic molecular interaction of Rops with ER-localized NCRK, and was not caused by a general inhibition of membrane cycling in the presence of high intracellular Rop protein levels as described by Bloch et al. (2005), since the plasma membrane localization of a control CFP-tagged RLK BRL3 (Cano-Delgado et al., 2004) in the presence of over-expressed Rops was unchanged (not shown). Taken together, the data are consistent with a general characteristic of receptor kinase signalling, i.e. ligand-induced activation at the plasma membrane and internalization of receptor kinase oligomers.

RLCK-VIb gene expression is suggestive of a role in plant development and pathogen defence

Specific Rop genes in Arabidopsis are highly upregulated in elongating pollen tubes and support tip growth (Gu et al., 2003, 2005; Vernoud et al., 2003). Other Rops are associated with morphogenesis of epidermal leaf cells (Fu et al., 2002, 2005) or are highly expressed in vasculature (Brembu et al., 2005). Apart from their role in different aspects of plant development, Rops are also involved in plant defence responses (see reviews by Agrawal et al., 2003; Gu et al., 2004). Therefore, gene expression patterns of Rop interactors may indicate their importance in different Rop-mediated processes. Arabidopsis RLCK-VIb genes are differentially expressed during plant development with specific genes upregulated during xylogenesis or in pollen, as in case of the RBK2 gene (Pina et al., 2005), and during pathogen attack as for the RBK1 gene. The RBK1 promoter was ethylene responsive and maximal RBK1 mRNA levels during B. cinerea exposure required ethylene perception. Higher gene expression levels of the seven-member RLCK-VIb gene clade generally correlated with higher levels of cell wall biosynthesis and/or ROS production, whether during tracheary differentiation, in pollen, trichomes or during the hypersensitive response caused by P. infestans and B. cinerea pathogens. This expression profile might indicate that RLCK-VIb protein kinases function in cell wall sensing, similar to WAK and THE1 kinases (Decreux and Messiaen, 2005; Hematy et al., 2007).The presence of a universal stress protein (USP) domain in a number of RLCK-VIa and -VIb protein kinases (Kerk et al., 2003) might also suggest a role for this family of protein kinases in oxidative stress signalling, although the activating stimulus for RLCK-VIb kinases remains unknown.

NCRK, an orphan receptor kinase gene highly upregulated during tracheary differentiation

Homozygous Arabidopsis NCRK gene knock-out lines or overexpressing lines did not show obvious abnormalities (results not shown), offering no preliminary indications for gene function. High NCRK gene expression levels in trichomes were consistent with a relatively large number of plant NCRK ESTs originating from cotton fibre cDNA libraries in public databases (not shown), and seem to argue against a role for NCRK in cell type specification or intercellular communication, but instead for a role in cell wall sensing and detection of an autosecretory ligand. Relatively high levels of NCRK gene expression in vasculature were consistent with data by Kubo et al. (2005), which show that the NCRK gene is highly upregulated in developing tracheary cells in Arabidopsis cell suspensions treated with brassinolide to undergo trans-differentiation to xylem elements. The changes in cell morphology during differentiation of tracheary elements are accompanied by extensive rearrangements in the microtubular and actin cytoskeletons and altered cell wall synthesis (Fukuda, 2004), which will require Rop action and possibly a number of cell wall sensing RLKs.

We currently hypothesize that ligand-dependent NCRK signalling and RLCK-VIb kinase signalling are modulated by Rop GTPases. It will be interesting to identify RLCK-VIb- and NCRK-interacting proteins in future and elucidate the related signalling pathways.

Experimental procedures

Yeast two-hybrid screening

AtRop4 constructs were cloned into the bait vector pAS2 (Clontech, http://www.clontech.com/), and yeast two-hybrid screening of an Arabidopsis cell suspension culture cDNA library in pACT2 (Nemeth et al., 1998) was carried out according to Soellick and Uhrig (2001). Protein extracts of yeast double transformants were prepared according to the Clontech yeast protocol handbook and assayed by western blotting. Prey kinase proteins were detected using an anti-HA monoclonal (Covance; http://www.covance.com), while AtRop4 bait proteins were detected using either affinity-purified AtRop4 polyclonals (Eurogentec, http://www.eurogentec.com/) or monoclonal AtRop4 antibodies (Nanotools, http://www.nanotools.de/).

Insect cell transfection and FRET measurements

Full-length NCRK cDNA was obtained from the RIKEN Bioresource Center (RBC) and tagged at the C-terminus with YFPv (Venus version of EYFP; Nagai et al., 2002). GTPase-binding domain cDNA was tagged at the N-terminus with YFPv, while Rop4 wildtype and mutant Rop4 cDNAs G15V and T20N (Molendijk et al., 2001) were tagged at the N-terminus with CFP (ECFP, Clontech). Tagged plant cDNAs were subcloned into pRmHa-3 for transfection of Drosophila melanogasterS2 cells as described (Rolli et al., 2002). Flow cytometric FRET measurements were performed on transfected cells expressing fluorescent proteins essentially according to Chan et al. (2001). For western analysis, frozen insect cell lysates were dissolved in SDS-PAGE loading buffer and samples were run on 15% acrylamide gels. Expressed ECFP/YFPv-tagged proteins were detected using GFP monoclonal antibodies (Roche, http://www.roche.com/) and an ECL detection kit (Amersham, http://www5.amershambiosciences.com/).

Immunoprecipitation and protein kinase assays

Full-length RBK1 cDNA was obtained from RBC, tagged at the C-terminus with YFPv and subcloned into pRMHa-3. NCRK-YFPv and RBK1-YFPv were expressed in Drosophila S2 cells. Immunoprecipitation was performed on cell lysates using monoclonal GFP antibodies (Roche). AtRop4 cDNA lacking the polybasic tail (amino acids 1–179) was cloned in pET45b (Novagen; http://www.merckbiosciences.com). Bacterially expressed AtRop4 was affinity purified using nickel nitrilotriacetic acid (NiNTA), followed by HPLC gel filtration. Kinase assays were performed according to standard procedures using 32P γATP.

Plant protoplast transformation and confocal microscopy

The ECFP-tagged wildtype AtRop4, ECFP-tagged G15V and T20N mutant versions of AtRop4, YFPv-tagged NCRK and YFPv-tagged RBK1 inserts were transferred from pRMHa-3 constructs to plant expression vector pCKGFP (Reichel et al., 1996) from which the GFP encoding DNA had been removed (pCK). AtRabF2a (Rha1; At5g45130) and BRL3 (At3g13380) cDNAs were obtained from the BRC. AtRabF2a and BRL3 were tagged with ECFP at the N-terminus and C-terminus, respectively, and subcloned into pCK. Full-length RopGAP2 (At4g03100) cDNA was also subcloned into pCK. The ER-CFP marker was used as in Boevink et al. (1998) but substituting GFP for ECFP and cloning the marker cDNA into pCK. Protoplasts were transformed according to the polyethylene glycol-mediated transformation protocol (Koop et al., 1996) using 0.5 × 106 cells and 25–50 μg of each plasmid DNA per transformation. As visualized by fluorescence microscopy, transformation efficiencies of approximately 50–80% were routinely obtained, with co-transformation efficiencies at almost 100% (our observations). For microscopy, a Zeiss LSM 510 confocal microscope equipped with a 32-channel spectral detector (Zeiss LSM 510 META) was used with LSM software (Zeiss, http://www.zeiss.com/).

BiFC measurements

NCRK and RBK1 cDNAs were cloned in BiFC vectors pUC-SPYCE and pUC-SPYNE for C-terminal split YFP tagging (Walter et al., 2004). Rop4 T20N and Rop4 G15V cDNAs were cloned in two home-made vectors for N-terminal split YFP tagging. Twenty-five micrograms of plasmid DNA for each BIFC construct was used to co-transform Arabidopsis seedling protoplasts. The percentage of BiFC-positive protoplasts was determined by cell counting using an Axiovert 200M microscope (Zeiss) with Sensi Cam camera (PCO Computer Optics, http://www.pco.de/) and Metamorph software (MDS; http://www.moleculardevices.com). A minimum of 600 cells were counted for each transformation sample.

Lambda protein phosphatase treatments and SDS-PAGE mobility shift assay

Transformed protoplast pellets were resuspended in phosphatase buffer containing protease inhibitor cocktail, 2 mm MnCl2 and 1% Triton X-100. Protein dephosphorylation reactions using lambda protein phosphatase (λ PPase, 400 U, 30 min, 30°C) were performed according to the manufacturer’s instructions (New England Biolabs, http://www.neb.com/) with phosphatase inhibitor cocktail (Sigma, http://www.sigmaaldrich.com/) added to untreated controls. Samples corresponding to 0.5 × 106 cells were analysed using 20 cm-long 6% acrylamide SDS-PAGE gels, followed by western blotting according to standard protocols.

In situ RNA detection

The DNAs corresponding to the transcriptional trailer regions of NCRK (207 nt) and RBK1 (157 nucleotides) genes were obtained using PCR (Platinum Taq; Invitrogen, http://www.invitrogen.com/) on whole seedling cDNA, cloned in transcription vector pGEM-T easy (Promega, http://www.promega.com/) and sequenced. Biotin-labelled probe RNA was transcribed using the Ribomax kit as recommended by the manufacturer (Promega). Six-day-old Arabidopsis seedlings were embedded in paraffin and processed for microtome sectioning. Tissue sections on slides were hybridized with probe RNAs, using a slide processing robot (Cnops et al., 2006). Immunological detection of hybridized probe was performed using alkaline phosphatase conjugated anti-biotin monoclonal antibody (Sigma).

Arabidopsis transformation and binary promoter GUS constructs

The complete intergenic region and transcriptional leader sequences including ATG start codons of the NCRK (1.94 kb) and RBK1 (2.07 kb) genes were obtained using PCR (AccuPrime pfx DNA polymerase; Invitrogen) on whole Arabidopsis genomic DNA. Promoter DNAs were cloned into Gateway® donor plasmid pDONR207 (Invitrogen), and from there into promoter GUS reporter destination vector pMDC163 (Curtis and Grossniklaus, 2003). Binary constructs were sequenced, and electroporated into Agrobacterium tumefaciensGV3101 pMP90. Agrobacterium-mediated Arabidopsis transformation was carried out using the floral dip method (Clough, 2005).

Pathogen experiments, plant treatments and expression analyses

Phytophthora infestans isolate 208 M2 (Si-Ammour et al., 2003) was grown on oat-bean medium at 18°C in the dark. Ten to twelve-day-old mycelium was incubated with 10 ml of sterile distilled water at 4°C for 3 h. After filtration, zoospores were counted and the appropriate concentration of the zoospore suspension was adjusted. Arabidopsis plants were inoculated with 10 μl drops of the zoospore suspension. After inoculation, plants were incubated in a phytochamber with 16 h of light (200 μE) at 20°C, 8 h dark at 18°C and 100% humidity. Botrytis cinerea (strain SAS 56) inoculation was carried out on mature rosette leaves by spotting four drops of 5μl each (100 spores μl−1) per leaf and collecting infected and mock-treated leaf material after 24 h of incubation at the above described conditions. Botrytis cinerea spores were resuspended in potato dextrose broth (PDB) medium and with mock treatment consisting of 5 μl drops of PDB medium only. 1-Methylcyclopropene (1-MCP) treatments were carried out by incubating plants in closed containers with a saturating concentration (10 p.p.m.) of the chemical. Extraction of RNA was performed by using a commercial kit (NucleoSpin® RNA Plant; Macherey Nagel, http://www.macherey-nagel.com/) following the manufacturer’s instructions. Ethylene treatment was carried out by flushing 7-day-old seedlings, grown on Arabidopsis medium agar plates, in gas-tight jars with 10 p.p.m. ethylene or air (control) for 24 h. Quantitative real-time PCR was carried out on a DNA Engine Opticon2 (Biorad; http://www.bio-rad.com) using SYBR Green RealMasterMix (Eppendorf, http://www.eppendorf.com/) as described by Nonis et al. (2007).


The work was supported by grants from the Deutsche Forschungsgemeinschaft to KP (SFB, 746), the Alexander von Humbold Foundation, the FCI, a Marie Curie Fellowship of the European Community programme ‘Improving Human Research Potential’ under contract number HPMF-CT-2002-01880 to BR, and a DFG grant under contract number Hu794/3-2 to MH. Additional support was provided by the Freiburg Initiative in Systems Biology (# 03139219), a program of the German Federal Ministry in Research and Education. We are grateful to Dr Bingshan Wang for providing plant BiFC vectors, to Drs Olaf Tietz and Taras P. Pasternak for advice on histochemical analysis, to Dr C. S. V. Rajendrakumar for providing RopGAP2 cDNA, to Claudia Gilles for plant transformation and propagation, to Dr Roland Nitschke at the Freiburg University Life Imaging Centre for advice on confocal microscopy, and to Dr William Teale for critical reading of the manuscript. We are also grateful to Professor Giuseppe Firrao and Dr Patrick De Marta, University of Udine, Dipartimento di Biologia Applicata alla Difesa delle Piante, for their kind help with the Botrytis experiments and to dr. Alberto Nonis, Dipartimento di Scienze Agrarie e Ambientali, for his help with real-time RT-PCR.