CAST AWAY, a membrane-associated receptor-like kinase, inhibits organ abscission in Arabidopsis

Receptor-like kinase-mediated cell signaling pathways play fundamental roles in many aspects of plant growth and development. A pair of Arabidopsis thaliana leucine-rich repeat receptor-like kinases (LRR-RLKs), HAESA (HAE) and HAESA-LIKE2 (HSL2), have been shown to activate the cell separation process that leads to organ abscission. Another pair of LRR-RLKs, EVERSHED (EVR) and SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASE1 (SERK1), act as inhibitors of abscission, potentially by modulating HAE/HSL2 activity. Cycling of these RLKs to and from the cell surface may be regulated by NEVERSHED (NEV), a membrane trafficking regulator that is essential for organ abscission. We report here the characterization of CAST AWAY (CST), a receptor-like cytoplasmic kinase that acts as a spatial inhibitor of cell separation. Disruption of CST suppresses the abscission defects of nev mutant flowers, and restores the discrete identity of the trans -Golgi network in nev abscission zones. After organ shedding, enlarged abscission zones with obscured boundaries are found in nev cst flowers. We show that CST is a dual-specificity kinase in vitro, and that myristoylation at its N-terminus promotes association with the plasma membrane. Using the bimolecular fluorescence complementation assay, we have detected interactions of CST with HAE and EVR at the plasma membrane of Arabidopsis protoplasts and hypothesize that CST negatively regulates cell separation signaling directly and indirectly. A model integrating the potential roles of receptor-like kinase signaling and membrane trafficking during organ separation is presented. promote organ separation, respectively. Our studies of CST suggest a distinct mode of RLCK action in which an RLCK and RLK partner may act in a step-wise fashion to inhibit the activity or alter the location of a ligand-binding receptor-like kinase. wild-type, and between nev cst and nev tissues are indicated by single (Fisher’s exact test: P <0.012) and double ( P <0.02) asterisks, respectively. A statistical difference was not detected between cst and wild-type tissues. n ≥ 26 cells per genotype. G2A -GFP) or the predicted palmitoylation site (CST C4S -GFP) results in localization of CST mutant proteins throughout the cytoplasm. Localization of the mutant proteins is still observed at or near the plasma membrane.


Introduction
Signaling by transmembrane receptor-like kinases (RLKs) underlies diverse aspects of plant growth and development. Surprisingly, a substantial number of plant RLKs do not contain either extracellular or transmembrane domains. Although receptor-like cytoplasmic kinases (RLCKs) account for at least 125 of the 610 annotated RLKs in Arabidopsis, much remains to be learned about their functions within cell signaling complexes (Shiu and Bleecker, 2001;Shiu et al., 2004;Goring and Walker, 2004;Jurca et al., 2008). Several of the 46 RLCKs assigned to the class VII subfamily have been found to function in pathogen response and developmental signaling pathways (Swiderski and Innes, 2001;Shao et al., 2003;Murase et al., 2004;Muto et al., 2004;Veronese et al., 2006;Ade et al., 2007;Lu et al., 2010;Zhang et al., 2010).
Functional studies of class VII RLCKs have identified four different modes of action; RLCKs can act as co-receptors of RLKs, in signal relays, as repressors, and as activators of signaling. The M-Locus Protein Kinase (MLPK) RLCK functions as a coreceptor of a ligand-binding RLK to transduce signaling. During the self-incompatibility response in Brassica rapa flowers, MLPK has been found to interact with the ligandactivated S-Locus Receptor Kinase and is essential for the cell signaling leading to rejection of self-pollen (Murase et al., 2004;Kakita et al., 2007). The BOTRYTIS-INDUCED KINASE 1 (BIK1) RLCK functions in a signaling relay with an activated ligand-binding RLK and its co-receptor. BIK1 was shown to interact with two leucine-rich repeat receptor-like kinases (LRR-RLKs), the ligand-binding FLAGELLIN-SENSITIVE 2 (FLS2) and its co-receptor BRI1-ASSOCIATED KINASE 1 (BAK1) (Veronese et al., 2006;Lu et al., 2010). FLS2 binding of the bacterial flagellin-derived peptide, flg22, triggers interaction of FLS2 and BAK1, and downstream signaling for pathogenassociated molecular patterns (PAMP)-triggered immunity (Chinchilla et al., 2007;Heese et al., 2007). BIK1, which independently associates with FLS2 and BAK1 in the absence of ligand, is rapidly phosphorylated by BAK1 upon flg22 treatment (Lu et al., 6 The AvrPphB Susceptible 1 (PBS1) RLCK acts as a repressor. PBS1 is cleaved by an effector of the pathogen Pseudomonas syringae, activating the nucleotide binding site-leucine rich repeat (NB-LRR) protein, RPS5, which triggers programmed cell death (Warren et al., 1999;Swiderski and Innes, 2001;Shao et al., 2003;Ade et al., 2007). A recent study has revealed that like PBS1, BIK1 and several PBS1-like (PBL) RLCKs are also substrates of the bacterial AvrPphB protease effector (Zhang et al., 2010). This discovery and other work suggests that a bacterial effector can suppress PAMPtriggered immunity in plants by cleaving RLCKs known (BIK1) or proposed (PBL1, PBL2) to positively interact with RLKs that bind PAMPs, including FLS2, CHITIN ELICITOR RECEPTOR KINASE1, and the EF-Tu RECEPTOR (Zipfel et a., 2006;Miya et al., 2007;Zhang et al., 2010). During effector-triggered immunity, another RLCK,

Organ separation is restored in nev cst flowers
To identify novel regulators of organ abscission, a genetic screen was conducted for mutations that restored abscission in nev-3 mutant flowers (Lewis et al., 2010). A recessive mutation identified in this screen, cast away (cst-1), was found to rescue organ separation in nev flowers ( Fig. 1A-C). A second mutant allele of CST from the SAIL T-DNA collection (cst-2;SAIL_296_A06;Sessions et al., 2002) dominantly rescues organ abscission in nev-3 flowers (Fig. 1D). Flowers with mutations in CST alone have a wild-type appearance and organ shedding occurs normally (Fig. 1E,F).
Since NEV, IDA, HAE and HSL2 each regulate the cell separation stage of organ separation (Butenko et al., 2003;Cho et al., 2008;Stenvik et al., 2008;Liljegren et al., 2009), we tested whether disruption of CST activity might also suppress the ida and hae hsl2 mutant phenotypes. We found that mutations in CST do not rescue the shedding defects of ida or hae hsl2 flowers ( Fig. 1G-J). These results suggest that CST acts upstream of IDA and HAE/HSL2, or in a parallel pathway that converges at HAE/HSL2 activity or further downstream.

The organ AZs of nev cst flowers are enlarged and disorganized.
Although organ separation occurs in nev-3 cst-1, nev-3 cst-2/+ and nev-3 cst-2 flowers, the AZ regions have a visibly rough appearance compared to the smooth surfaces of the organ detachment sites in wild-type flowers (Fig. 1A,C,D). To further characterize this phenotype, we examined longitudinal sections and scanning electron micrographs of nev cst flowers at the time of shedding compared to wild-type and cst single mutant flowers ( Fig. 2A,C-G). While the remaining AZ cells of wild-type expand to form discrete scars ( Fig. 2A,E), cells in the AZ regions of nev cst flowers have a disordered appearance and show increased, uneven cell expansion (Fig. 2C,F). After organ shedding, nev cst AZ regions were found to be significantly enlarged compared to wildtype ( Fig. 2H) and the boundaries between individual organ detachment sites and with the floral stem are notably obscured (Fig. 2E-G). These results suggest that CST acts as a spatial inhibitor of signaling that modulates AZ cell adhesion and expansion.

Disruption of CST activity suppresses the subcellular defects of nev flowers
Our studies have previously shown that mutations in NEV are associated with a unique set of trafficking defects in flowers undergoing organ separation (Liljegren et al., 2009).
To determine whether disruption of CST activity suppresses these subcellular changes as well as restoring organ separation in nev cst mutant flowers, we carried out transmission electron microscopy of wild-type and mutant flowers shortly after organ abscission (stage 17) ( Fig. 2A-D). Whereas the structure and organization of the Golgi cisternae and trans-Golgi network are altered in nev-3 mutant cells (Fig 3B,E) compared to wild-type (Fig. 3A,E), we found that these organelles are unaffected in nev-3 cst-2 ( Fig. 3C,E) and cst-2 cells (Fig. 3D,E). We also discovered that the hyperaccumulation of extracellular vesicles in nev-3 cells (

CST encodes a receptor-like cytoplasmic kinase with dual specificity
The cst-1 mutation was mapped to chromosome 4 and found to affect At4g35600, a gene encoding a predicted receptor-like cytoplasmic kinase (RLCK) of the class VII subfamily (Fig. 4A). Although an early study suggested that the At4g35600 gene product shared sequence homology with animal connexins (Meiners et al., 1991), subsequent analysis has shown that the first open reading frame predicted was incorrect (Mushegian and Koonin, 1993;Arabidopsis Genome Inititative, 2000;Yamada et al., 2003). No evidence of connexin homology in the accepted sequence of the At4g35600 gene product has been found.
The cst-1 mutation introduces an amino acid substitution immediately after subdomain IV of the kinase domain, which is involved in binding ATP (Fig. 4B) (Hanks, 2003). Although residues in this region are not highly conserved among eukaryotic protein kinases (Hanks and Hunter, 1995), the affected glycine is invariant in the kinase domains of all 46 predicted class VII RLCKs in Arabidopsis (Fig. 4B, data not shown).
The cst-2 mutant allele contains a T-DNA insertion upstream of the kinase domain, and is predicted to cause production of a truncated protein (Fig. 4A To test CST kinase activity, full-length proteins of wild-type (WT), a traditional kinase-dead mutant (K124E) (Horn and Walker, 1994), and the cst-1 mutant (G157R) were expressed as N-terminal 6XHis-tagged fusion proteins in E. coli. Whereas a His antibody recognizes purified, presumably phosphorylated CST WT protein migrating as a single band of ~55 kDa, it recognizes purified CST K124E and CST G157R proteins migrating as single bands of ~49 kDa in agreement with the predicted size of the tagged protein (49 kDa) (Fig. 4D). Phosphoserine, phosphothreonine, and phosphotyrosine antisera were used to detect phosphorylated residues on the recombinant proteins. Each of these antisera recognized the CST WT protein and neither of the mutant proteins (Fig.   4D). These results suggest that CST is a dual-specificity kinase that autophosphorylates serine, threonine and tyrosine residues in vitro, and that its kinase activity is abolished by the cst-1 mutation.

Analysis of allele-specific interactions between NEV and CST
We have observed allele-specific differences in the number of copies of cst-1 and cst-2 required to restore organ shedding in nev-3 flowers. While a single cst-2 allele is sufficient to dominantly rescue abscission in nev-3 flowers (Fig. 1B,D), both copies of the cst-1 mutant allele must be present to restore organ shedding in this background To determine whether we could uncover additional allele-specific interactions between CST and NEV, the cst-1 and cst-2 alleles were crossed to the nev-2 and nev-6 mutant alleles. While we have not detected significant phenotypic differences between the nev-2, nev-3 and nev-6 mutants in a Landsberg erecta background, the molecular nature of the respective mutations is quite different. The nev-3 mutation introduces an amino acid substitution in the ARF GAP domain at an invariant arginine (R59K), which is known to be essential for ARF GAP enzymatic activity (Luo et al., 2007;Liljegren et al., 2009). The nev-2 mutation introduces a stop codon downstream of the ARF GAP domain (Q198*) and is predicted to cause production of a truncated protein with an ARF GAP domain. The nev-6 allele contains a T-DNA insertion in the first intron, and is expected to produce a truncated protein without an ARF GAP domain.

1
While the cst-1 mutation recessively rescues organ shedding in nev-2 and nev-6 flowers ( Fig. 1M,N; data not shown), and cst-2 dominantly rescues shedding in nev-6 flowers (data not shown), the cst-2 allele was unable to restore abscission in nev-2 flowers even if both mutant copies of CST were present (Fig. 1O). These results are partially consistent with an allele-specific compensatory mutation, in which the suppressor mutation restores a physical interaction between the affected components (Michels, 2002). However, if the truncated cst-2 mutant protein were to interact with the nev-3 mutant protein in such a way that its ARF GAP activity was restored, one would expect that the cst-2 mutant allele should also not be able to rescue abscission in a nev mutant protein missing the ARF GAP domain.
To address whether the dominant rescue of organ shedding in nev flowers by the cst-2 allele is due to either a dominant-negative interaction or haploinsufficiency, a wildtype copy of the CST cDNA driven by its predicted 1.4 kb promoter was introduced into nev-3 cst-2 homozygous mutant plants. We predicted that one copy of CST would not suppress organ shedding in either a dominant-negative or haploinsufficient situation, whereas a block of organ shedding by two copies of CST would be consistent with haploinsufficiency but not a dominant-negative interaction. We observed that for two independent T1 lines with T2 kanamycin-resistance segregation ratios characteristic of a single insertional locus, presence of the CST::CST transgene was sufficient to block organ abscission in nev cst flowers, restoring the nev mutant phenotype in plants hemizygous and homozygous for the transgene (Fig. 1K,L; Supplemental Table SI).
Since multiple T-DNA insertions can be present at a single locus (Jorgensen et al., 1987), these results do not rule out a dominant-negative interaction involving truncated CST protein produced from the cst-2 allele. Multiple copies of wild-type CST could efficiently dilute the dominant-negative effect of a single locus producing truncated CST protein.

Localization of CST to the plasma membrane is supported by N-terminal myristoylation
CST is predicted to associate with membranes in part via myristoylation of its Nterminus ( Fig. 4C) Warren et al., 1999;Boisson et al., 2003;Murase et al., 2004;Veronese et al., 2006;Kakita et al., 2007). Palmitoylation of N-terminal cysteine residues is also predicted to allow reversible membrane association for CST and many other class VII RLCKs (Fig. 4C;Sorek et al., 2009;Zhang et al., 2010). To visualize CST protein within Arabidopsis cells, we generated a CST-GFP fusion construct driven by the constitutive viral 35S promoter that could be transfected into mesophyll protoplasts. Attempts to visualize CST-YFP under the control of its native promoter in vivo were unsuccessful, likely due to the limited expression of CST in roots and leaves ( The distinct localization profile of CST compared to those of EVR, SERK1 and HAE may indicate that the mechanism of CST inhibition of abscission differs from that of EVR and SERK1.

CST is expressed in organ AZs, lateral roots, and developing guard cells
To determine the expression pattern directed by CST regulatory regions, a construct with a translational fusion of the predicted 1.4 kb promoter region to the β-glucuronidase  6G). These results suggest that expression of CST is restricted to specific cell types and tissues, and that CST may function during other phases of plant development.

CST interacts with the HAE and EVR receptor-like kinases at the plasma membrane
To test for interactions between CST and other RLKs that modulate abscission, we used the bimolecular fluorescence complementation (BiFC) assay in Arabidopsis mesophyll protoplasts (Walter et al., 2004;Yoo et al., 2007). This approach has been successfully used to detect interactions between membrane-bound LRR-RLKs, such as the SERK family members (SERK1, SERK2, BAK1 and BAK1-LIKE1) and BAK1-  7D). As depicted, no fluorescence was observed in the majority of protoplasts that were successfully transfected with the CFP-tagged mitochondrial expression control, indicating that CST and the C-terminus of YFP do not interact.
As a control for non-specific interactions between RLCKs and LRR-RLKs, we tested for interactions between the PBS1 class VII RLCK and the HAE and EVR LRR- and either HAE or EVR at the plasma membrane.
Taken together, these results suggest that CST inhibits signaling that promotes abscission both directly and indirectly by physically interacting with HAE and EVR at the cell surface.

Discussion
We report here the identification of CST, a membrane-associated RLCK that functions as an inhibitor of organ abscission. Like the EVR and SERK1 LRR-RLKs (Leslie et al., due to ectopic, prolonged signaling of HAE and HSL2, the putative receptors of the IDA peptide. We have found that the cst-2 mutation restores the structure of the Golgi and location of the TGN in nev flowers, and significantly reduces the hyperaccumulation of extracellular vesicles (Fig. 3) Functional studies of class VII RLCKs have revealed that several act in plant defense cell signaling pathways; CST is one of the first RLCKs found to regulate a developmental process (Table I) MLPK and the ligand-activated S-Locus receptor kinase are proposed to form a heteromeric complex in stigmatic cells that mediates signaling leading to the rejection of self pollen (Murase et al., 2004;Kakita et al., 2007). As with the cst mutant, redundancy appears to mask the phenotype of the Arabidopsis constitutive differential growth1 Although the genetic nature of the cst-2 allele has not been resolved, we speculate that it is likely acting as a dominant negative mutation. If the truncated mutant protein is acting as a non-functional kinase and is capable of homodimerizing with wildtype CST protein and/or heterodimerizing with the HAE and EVR LRR-RLKs, it has the potential to interfere with the activity and regulation of its partners. Dominant negative LRR-RLK mutants are associated with particular missense mutations in the extracellular LRR and kinase domains of CLAVATA1 (Diévart et al., 2003;Diévart and Clark, 2003) and with truncation of the entire kinase domain in some LRR-RLKs such as ERECTA (ER) (Shpak et al., 2003). While most class VII RLCK mutations either occur within the kinase domain or are expected to truncate part of the kinase domain, the cst-2 mutation affects a codon positioned upstream of the kinase domain ( Fig. 4A; Table I intriguing that cst-2 dominantly suppresses the nev-3 and nev-6 mutants, but not the nev-2 mutant (Fig. 1D,O; data not shown). Allele-specific interactions for genes with opposing functions in a biological process suggest that the CST and NEV proteins may function in a single complex, and that the ARF GAP domain may regulate this interaction. We plan to explore these possibilities in future genetic and biochemical studies.

Integrated model for RLK regulation of organ separation
Our studies suggest a model in which CST and EVR inhibit cell separation signaling by acting in a stepwise manner to mediate HAE/HSL2 receptor complex formation and internalization ( within the early endosomal system and ultimately recycle them back to the plasma membrane (Fig. 8). Loss of NEV could lead to the hyperaccumulation of inactivated receptors within endosomal compartments, while a secondary loss of CST, EVR or SERK1 may stabilize the HAE/HSL2 RLKs at the plasma membrane. At the proper timing for abscission in wild-type flowers, our model predicts that IDA ligand-binding stabilizes HAE/HSL2, leading to activation of downstream signaling events required for the loss of cell adhesion (Fig. 8) internally in endosomes or externally in the observed extracellular vesicles. Secondary loss of either the CST or EVR negative regulators may restore localization of the HAE/HSL2 RLKs to the plasma membrane. Since the primary function of CST may be to sequester RLKs at the plasma membrane and facilitate interactions between their intracellular kinase domains, future experiments will also investigate whether CST interacts with SERK1, and whether EVR and/or SERK1 interact with HAE.
There is a growing body of evidence that class VII RLCKs can mediate cell signaling by forming heteromeric complexes with RLKs. MLPK acts as a co-receptor for the S-Locus receptor kinase during the self-incompatibility reponse (Murase et al., 2004;Kakita et al., 2007). BIK1 has been found to interact with the FLS2 and BAK1 LRR-

Protein interaction and localization assays
The coding regions for CST, HAE and PBS1 were PCR amplified from the          Interactions between a set of receptor-like kinases may modulate the timing and Information to identify the cst-1 and cst-2 mutant alleles is provided.

Supplemental Figure 1. CST localization is disrupted by mutations that affect N-terminal myristoylation or palmitoylation.
Transfected Arabidopsis leaf protoplasts were imaged using confocal microscopy. GFP fluorescence (green), chlorophyll autofluorescence (magenta), brightfield and merged images are shown for each protoplast. Protoplast focal planes are equivalent to that shown in Fig. 5A. Receptor-like kinase interactions and non-interactions (Fig. 7) were confirmed by performing BiFC assays in protoplasts transfected with the inverse YFP fusions.