The EVERSHED receptor-like kinase modulates floral organ shedding in Arabidopsis

Plants have evolved expansive groups of transmembrane receptor-like kinases (RLKs) that allow complex sampling of environmental conditions and developmental status. With over 600 RLKs annotated in the Arabidopsis genome, signaling complexity is enhanced by genetic redundancy (Shiu and Bleecker, 2001). Allowing for even further complexity, RLKs containing extracellular leucine-rich repeats (LRRs) can oligomerize and participate in multiple receptor complexes (Dievárt and Clark, 2003). For example, by interacting with the ligand-binding LRR-RLKs BRASSINOSTEROID INSENSITIVE1 (BRI1) and FLAGELLIN-SENSITIVE2 (FLS2), BRI1-ASSOCIATED RECEPTOR KINASE1 (BAK1) functions in two discrete signaling complexes that regulate cell growth and pathogen defense, respectively (Li et al., 2002; Chinchilla et al., 2007).

Although RLKs are expected to play crucial roles in plant development, deduction of these roles has often required the phenotypical analysis of multiple loss-of-function mutants (Shpak et al., 2004; DeYoung et al., 2006; Morillo and Tax, 2006; Xu et al., 2008). A recent breakthrough is the discovery that a pair of LRR-RLKs, HAESA (HAE) and HAESA-LIKE2 (HSL2), are redundantly required for floral organ abscission (Jinn et al., 2000; Cho et al., 2008; Stenvik et al., 2008). Whereas the outer whorls of the Arabidopsis flower, containing the sepals, petals and stamens, are shed following fertilization in wild type, floral organs remain attached indefinitely in hae hsl2 flowers. As shown for other LRR-RLKs (Li et al., 2002; Kinoshita et al., 2005; Karlova et al., 2006; Chinchilla et al., 2007), HAE and HSL2 probably function as members of larger receptor complexes that are activated through interactions with an extracellular ligand. A small, secreted peptide, INFLORESCENCE DEFICIENT IN ABSCISSION (IDA), is required for abscission and is likely to be the ligand for HAE/HSL2 (Butenko et al., 2003; Cho et al., 2008; Stenvik et al., 2008; Butenko et al., 2009).

With the identification of these and other mutations that block or delay floral organ shedding (reviewed by Leslie et al., 2007; González-Carranza et al., 2007; McKim et al., 2008; Ogawa et al., 2009; Liljegren et al., 2009), the Arabidopsis flower has become a model for studying abscission. In plants, abscission events allow the shedding of leaves, flowers, fruit and seeds, and can facilitate growth, reproduction, and defense against pathogens. As with most developmental events, proper timing and spacing are crucial during organ separation (Roberts and González-Carranza, 2007). In order for shedding to occur at the correct time, cell wall modifying enzymes must be secreted from cells within abscission zones (AZs) at the base of each organ to be shed (Roberts et al., 2002). If these enzymes are released too soon, premature organ loss could have an irreparable effect on reproduction (e.g. shedding of stamens before pollination). Conversely, a delay or block in abscission can also have an adverse effect on plant growth and reproduction (e.g. lack of seed dispersal) (Pinyopich et al., 2003). Of equal importance is the spatial regulation of abscission; the release of cell wall modifying enzymes must be restricted to discrete AZs to prevent general loss of tissue integrity or the shedding of neighboring organs. For example, misexpression of IDA leads to the deregulated expansion of floral AZ cells and can cause premature organ shedding, as well as ectopic loss of developing fruit (Stenvik et al., 2006; Cho et al., 2008).

Membrane trafficking is emerging as an additional means of regulating plant transmembrane RLKs and is likely to modulate signaling during complex developmental events (Geldner and Robatzek, 2008). We recently reported that NEVERSHED (NEV), an ADP-ribosylation factor GTPase-activating protein, is required for proper membrane trafficking and abscission (Liljegren et al., 2009). Mutations in NEV lead to altered organization of the Golgi and the associated trans-Golgi network, the accumulation of vesicles in paramural bodies, impaired fruit growth, and a block in organ separation. We hypothesized that NEV, which is localized to the trans-Golgi network and other distinct endosomal compartments, is required for the proper trafficking of key abscission factors.

Here, we report that mutations in EVERSHED (EVR), which encodes an LRR-RLK, rescue organ shedding in nev flowers. The sepals, petals and stamens of nev evr flowers are shed prematurely, and ectopic AZ cell expansion occurs. Loss of EVR kinase activity also restores normal Golgi morphology in nev AZ cells. Our results suggest that the EVR RLK functions as an inhibitor of abscission.

The evr-1 and evr-2 alleles were obtained through ethyl methanesulfonate screens of nev-3 (Landsberg erecta, Ler) plants as described (Liljegren et al., 2000). The evr-3 (sobir1-12; SALK_050715) and evr-4 (srrlk-2; SALK_009453) alleles (Col) contain T-DNA insertions within codons 516 and 600, respectively (Alonso et al., 2003; Gao et al., 2009; Katiyar-Agarwal et al., 2007). The At1g17750 (SALK_098161) allele contains an insertion within codon 137. Genotyping primers and restriction enzymes are listed in Table S1 in the supplementary material. Mutant alleles described previously include nev-2, nev-3, nev-6, ida-2, hae-1, hsl2-1 and pepr1-1 (SALK_014538) (Yamaguchi et al., 2006; Cho et al., 2008; Stenvik et al., 2008; Liljegren et al., 2009).

A 759-bp region of EVR, from 750 nucleotides upstream of the predicted translational start site into the third codon, was PCR amplified from Col DNA with 5′-GATCAAGCTTTCACTATGGAACCCACAG-3′ and 5′-GATCTCTAGAACAGCCATTTTAATTAGAG-3. This fragment was cloned into pCR2.1 (Life Technologies, Carlsbad, CA, USA), excised with HindIII/XbaI, and cloned into pDW137 (Blázquez et al., 1997) to create a translational fusion of the EVR promoter to β-glucuronidase (GUS). A 1689-bp region of HAE (Jinn et al., 2000) was amplified from Col DNA with 5′-GTTTCTGTTGCATGTCAGGATTAGC-3′ and 5′-GGATCCAGCATTTTTTTGGAAAAGGAATCG-3′, cloned into pCR2.1, excised with XbaI/BamHI, and cloned into pDW137. Eleven out of 12 HAE::GUS T1 lines showed AZ expression. GUS expression was analyzed as described (Blázquez et al., 1997).

A 2670-bp region of EVR, including the predicted promoter and open reading frame, was amplified from Col DNA with 5′-CACCTCACTATGGAACCCACAGCG-3′ and 5′-GTGCTTGATCTGGGACAACATGGTC-3′, cloned into pENTR/D-TOPO (Life Technologies) and recombined into pGWB40 to add a YFP C-terminal tag (Nakagawa et al., 2007). Sixty-four EVR::EVR-YFP T1 lines were generated.

To generate 35S::IDA-GFP/GUS plants, a tandem repeat of the 35S promoter was excised as an EcoRI/HindIII fragment from pBIN-JIT (Ferrandiz et al., 2000) and cloned in pBluescriptIISK(+) (Stratagene, La Jolla, CA, USA). The IDA cDNA was amplified with 5′-AAGCTTGACCCTTCATTCATTTACTC-3′ and 5′-GTCGACTGAGGAAGAGAGTTAACAAAAGAG-3′ from Col DNA, cloned into pCR2.1, excised with HindIII/SalI, and cloned into the 35S/pBluescript construct. The 35S::IDA fragment was excised with SacI/SalI, and replaced the 35S::CLV3 fragment in the pBGF-0 vector (Chytilova et al., 1999; Rojo et al., 2002) to create a translational fusion of IDA to GFP/GUS. Twelve out of 68 T1 lines showed the described phenotype.

To determine the identity of EVR, the nev-3 evr-1 mutant was crossed to nev-6 (Col). Using 631 F2 plants and PCR-based markers, including several designed from Ler polymorphisms (http://www.Arabidopsis.org/Cereon/), the evr-1 mutation was mapped to a 122-kb interval on chromosome 2, between the CER100665 and CER103194 markers.

EVR kinase domains were amplified with 5′-CACCAGAGGATCAGAAAAACCACCAGG-3′ and 5′-CTAGTGCTTGATCTGGGACAACATG-3′ from Col and evr-2 DNA, to create KD^WT and KD^E407K, respectively. Fragments were cloned into pENTR/D-TOPO. Site-directed mutagenesis (Stratagene) with 5′-GGAAGATCATAGCTGTGGAGAAAGTGATCCAACCG-3′ and 5′-CGGTTGGATCACTTTCTCCACAGCTATGATCTTCC-3′ was used to generate EVR-KD^K377E. Recombination with pDEST17 (Life Technologies) generated N-terminal 6×His-tagged EVR kinase domains for expression in Escherichia coli. Recombinant proteins were purified by Co²⁺ affinity chromatography (Clontech Laboratories, Mountain View, CA, USA).

To generate EVR-specific antiserum, a C-terminal peptide, CTLDDPKQRPNSKDVRTMLSQIKH, was synthesized and used to immunize chickens (Aves Labs, Tigard, Oregon, USA). Antisera were used at the following dilutions: anti-EVR (1:10,000), anti-phosphoserine (Sigma-Aldrich; 1:2000), anti-phosphotyrosine (Sigma-Aldrich; 1:2000), anti-phosphothreonine (Zymed/Invitrogen; 1:800). HRP-conjugated goat anti-chicken (Abcam, Cambridge, MA, USA) and chicken anti-mouse (Santa Cruz Biotechnology, Santa Cruz, CA, USA) secondary antibodies were used at a 1:10,000 dilution.

For scanning electron microscopy, flowers were fixed in 2% glutaraldehyde in 0.05 M sodium phosphate buffer, treated with 2% osmium tetroxide and dehydrated through an ethanol series. Samples were dried using a Samdri-795 critical point dryer (Tousimis Research Corporation, Rockville, MD, USA) and coated in gold-palladium using a Hummer X sputtering system (Anatech, Alexandria, VA, USA). Images were captured using a Zeiss Supra 25 scanning electron microscope with SmartSEM acquisition and imaging software. For transmission electron microscopy, flowers were analyzed as described (Liljegren et al., 2009). Confocal laser scanning microscopy of leaf petiole (stem), root, and cotyledon epidermal cells was performed with an LSM-510 (Carl Zeiss, Thornwood, NY, USA). Image brightness and contrast were adjusted with Adobe Photoshop CS4.

Wild-type floral organs are shed after fertilization, whereas floral organs senesce and remain attached indefinitely in nevershed (nev) flowers (Fig. 1A,B). To identify novel regulators of abscission, we carried out a suppressor screen for mutations that restore floral organ shedding in nev-3 plants (M.W.L., E. Fulcher and S.J.L., unpublished). Two recessive mutations identified in this screen form a complementation group that we have named evershed (evr) (Fig. 1C; see also Fig. S1A in the supplementary material). The interaction between NEV and EVR was subsequently observed in multiple allele combinations (see Fig. S1C-F in the supplementary material). Because the phenotypes of nev evr-1 and nev evr-2 flowers are indistinguishable, analysis was primarily performed using the evr-2 allele.

Mutations in NEV block abscission during the separation phase; the early patterning and differentiation of nev abscission zone (AZ) cells are unaffected (Liljegren et al., 2009). The secondary loss of EVR restored cell separation within nev AZs, such that all of the outer floral organs were shed from nev evr flowers (Fig. 1C; see also Fig. S1A in the supplementary material). As the densely cytoplasmic cells of nev sepal and petal AZs fail to become vacuolated and expand as they do in wild type (Fig. 1E,F) (Liljegren et al., 2009), we examined longitudinal sections of nev evr flowers at the time of shedding to see whether these defects were rescued by mutations in EVR. We found that cell vacuolation and expansion are not only rescued, but that nev evr AZ cells expand to a greater extent than do those of wild type (Fig. 1E-G).

To determine whether mutations in EVR affect organ separation in otherwise wild-type flowers, we isolated evr-1 and evr-2 single mutants. Floral organ shedding and AZ expansion appeared to be unaffected by loss of EVR alone compared with wild type (Fig. 1A,D,E,H; see also Fig. S1B in the supplementary material).

By examining the progression through flower development, we discovered that nev evr floral organs are shed prematurely (Fig. 2A,C,E,G). In wild-type flowers, buds open and pollen grains are released by the anthers (stage 13); fertilization of the ovules occurs shortly thereafter (stage 14) (Müller, 1961; Smyth et al., 1990). After the fruit begins to elongate (stage 15), the outer floral organs start to senesce and undergo abscission (stage 16) (Fang and Fernandez, 2002). To compare these developmental stages in wild-type and mutant flowers, the youngest open bud (stage 13) at the inflorescence apex was defined as the first floral position, the next older flower produced as the second position, and so on (Patterson, 2001). The percentage of flowers at each floral stage (13-17) was assessed for positions one to ten. Whereas all floral organs were shed (stage 17) from wild-type inflorescences by position 8, shedding was complete in nev evr inflorescences by position 6 (Fig. 2A,C). In nev evr flowers, abscission occurred before the petals wither, at a developmental timepoint resembling stage 15 (Fig. 2C, stage 15*) rather than stage 16. Similar results were obtained with petal breakstrength assays (Lease et al., 2006) of wild-type, nev and nev evr flowers. Whereas a similar force was required to remove petals from nev flowers at each position, the force required to remove nev evr petals dropped to zero by position 4, one position earlier than for wild type (see Fig. S2 in the supplementary material). These results suggest that EVR prevents premature organ abscission during the transition between fertilization and early fruit development.

nev and nev evr mutants show other growth defects, including reduced stature (see Fig. S3A in the supplementary material) and decreased fruit growth (Fig. 2F,G) (Liljegren et al., 2009). To confirm that premature shedding of nev evr floral organs was not due to a general defect in the timing of floral meristem formation, we also tracked individual wild-type and mutant flowers over the course of 5 days, from bud opening (day 0; stage 13) to the completion of abscission (stage 17). Whereas wild-type floral organs began shedding 3 days post-bud opening, nev evr floral organs began shedding one day earlier (see Fig. S3B in the supplementary material). These results suggest that nev evr floral organs abscise prematurely with respect to both floral age and stage of development.

Mutations in EVR alone did not substantially affect the timing of abscission (Fig. 2D,H; see also Fig. S3B in the supplementary material). However, evr fruit were significantly shorter (79%) than wild type (Fig. 2E,H), and nev evr fruit were shorter (88%) than nev fruit (Fig. 2F,G).

We found that expansion of nev evr AZ cells is increased in comparison to wild type at the time of shedding (Fig. 1E,G). To further characterize this phenotype, we took scanning electron micrographs (SEMs) of wild-type, evr and nev evr flowers after organ separation was complete (youngest stage 17). In wild-type flowers, discrete AZs were found at the base of the developing fruit (Fig. 3A). Loss of EVR alone did not appear to affect AZ development in comparison to wild type (Fig. 3B). By contrast, nev evr AZs were less defined and appeared to expand into the neighboring fruit and stem tissues (Fig. 3C). In a comparison of sepal AZ height in stage 17 SEMs, we found that nev evr sepal AZs were 2-fold taller than either wild-type or evr sepal AZs (Fig. 3E). In addition, the stem-like base of the fruit known as the gynophore was hidden owing to expansion of the stamen AZs in mature nev evr flowers (Fig. 3C, arrow). Cells within nev evr sepal AZs were also significantly larger in comparison to wild-type and evr cells (cells/10³ μm² measurement, Fig. 3F), indicating that cell expansion contributes in part to the increase in AZ size. In older flowers, the cell expansion was even more striking — discrete AZs were less recognizable and the floral nectaries became enveloped in the expanding tissue (Fig. 3G). Later on, the expanded nev evr AZ cells burst, leaving behind a visible collar of rough tissue around the base of the fruit (Fig. 1C).

A similar phenotype was reported for flowers misexpressing the putative signaling ligand INFLORESCENCE DEFICIENT IN ABSCISSION (IDA) (Stenvik et al., 2006; Cho et al., 2008). We generated plants that constitutively express an IDA-GFP/GUS translational fusion under the control of the viral 35S promoter. We also observed enlargement of the floral AZs and overexpansion of individual AZ cells (Fig. 3D-F,H). Similar to what we observed in nev evr flowers, premature abscission of the sepals, petals and stamens has been reported for 35S::IDA flowers (Stenvik et al., 2006). These results suggest that the regulation of AZ size and cell expansion by NEV, EVR and IDA is important for the proper timing of organ shedding in Arabidopsis.

The nev evr phenotype suggests an important role for EVR in abscission. To clarify its mechanism of action, we identified single nucleotide changes for the evr-1 and evr-2 alleles within At2g31880, one of the 34 predicted genes in the mapping interval. This intronless gene encodes a 641 amino acid protein of the leucine-rich repeat receptor-like kinase (LRR-RLK) family (Fig. 4A). Eleven additional alleles of EVR were identified in a screen for mutations that rescue seedling lethality in bak1-interacting receptor-like kinase1 (bir1) mutants (Gao et al., 2009). Loss of BIR1 is predicted to cause constitutive activation of a SUPPRESSOR OF BIR1 (SOBIR1)/EVR-mediated signaling pathway for disease resistance (Gao et al., 2009). Also encoded within the EVR open-reading frame is a heterogeneous cluster of small RNAs that are expressed upon pathogen infection, and for which EVR has also been named small RNA-generating RLK (Katiyar-Agarwal et al., 2007) (see Fig. S4A in the supplementary material).

Based on the sequence of its kinase domain, EVR has been assigned to the 28-member LRR-RLK subfamily XI (Shiu and Bleecker, 2001). This subfamily includes the HAESA (HAE) and HAESA-LIKE2 (HSL2) RLKs known to redundantly promote organ shedding (Jinn et al., 2000; Cho et al., 2008; Stenvik et al., 2008), the CLATAVA1 and BARELY ANY MERISTEM RLKs that control stem cell differentiation in shoot and flower meristems (Clark et al., 1997; DeYoung et al., 2006; DeYoung and Clark, 2008), the HAIKU RLK that regulates seed size (Luo et al., 2005) and the PEP1 RECEPTOR1 (PEPR1) that is involved in pathogen response (Yamaguchi et al., 2006). Although most members of subfamily XI have extensive sets of 19-32 LRRs in their extracellular domains (Shiu and Bleecker, 2001), EVR has only five predicted LRRs (Fig. 4A). However, in the context of the entire LRR-RLK family, EVR is not unique; the majority of Arabidopsis LRR-RLKs have fewer than ten LRRs (Shiu and Bleecker, 2001).

The evr-1 and evr-2 mutations are both predicted to be loss-of-function alleles that are unlikely to affect transcriptional regulation of EVR or the alternatively encoded small RNAs. The evr-1 allele introduces a stop codon upstream of the first LRR, suggesting that it is a null allele (Fig. 4A). The evr-2 allele changes a conserved glutamic acid to a lysine within subdomain III of the kinase domain (Fig. 4A,B). This subdomain is involved in ATP binding and catalysis; a loss-of-function mutation in the HAR1 LRR-RLK gene of Lotus japonicus introduces an identical missense mutation (Nishimura et al., 2002; Diévart and Clark, 2003). Two additional mutant alleles (Alonso et al., 2003), evr-3 and evr-4, contain T-DNA insertions within subdomains VII and X of the kinase domain, respectively (Fig. 4A; see also Fig. S1H,I in the supplementary material). These mutations also restore organ separation and cause premature abscission in nev flowers (see Fig. S1G,J-L, Fig. S2 in the supplementary material; data not shown).

Although up to 20% of Arabidopsis RLKs are predicted to be kinase dead (Castells and Casacuberta, 2007), isolation of the evr-2 mutation suggests that EVR kinase activity is required to regulate abscission (Fig. 4A,B). To facilitate the analysis of EVR kinase activity, we generated EVR-specific antiserum using a 24-amino acid peptide corresponding to its unique C terminus. The EVR antibody recognizes the EVR-KDs of wild type (WT), a kinase-dead mutant (K377E) (Horn and Walker, 1994) and the evr-2 mutant (E407K) expressed as N-terminal 6×His-tagged fusion proteins in E. coli (Fig. 4B,C). Whereas the purified KD^K377E and KD^E407K proteins migrate as single bands of ~40 kDa near the predicted size of 39 kDa, the purified KD^WT migrates as two distinct bands of ~40 and 46 kDa, suggesting that the wild-type protein is phosphorylated (Fig. 4C).

To test whether EVR is a functional serine/threonine kinase, we used phosphoserine and phosphothreonine antisera to detect phosphorylated residues on the recombinant KDs. Both antisera recognized the upper KD^WT band but not the lower, presumably unphosphorylated, KD^WT band. Neither KD^K377E nor KD^E407K appear to have kinase activity, as the antisera did not recognize these mutant proteins (Fig. 4C). As some LRR-RLKs have also been demonstrated to be dual-specificity kinases (Mu et al., 1994; Oh et al., 2009), we also tested the ability of the recombinant EVR-KDs to undergo autophosphorylation at tyrosine residues. We found that phosphotyrosine antiserum exhibited the same recognition pattern as that of the phosphoserine and phosphothreonine antisera (Fig. 4C), and that treatment with calf intestinal alkaline phosphatase was able to completely dephosphorylate the tyrosine residues of EVR-KD^WT (see Fig. S4B in the supplementary material). These results suggest that EVR is a dual-specificity kinase that autophophorylates serine, threonine and tyrosine residues in vitro.

EVR, like other transmembrane receptor-like kinases, is predicted to localize to the plasma membrane (Alexandersson et al., 2004). To visualize EVR localization in vivo, we generated transgenic plants expressing EVR-YFP under control of the EVR promoter (see Fig. S4A in the supplementary material). Using confocal laser scanning microscopy, six out of six independent transgenic lines showed localization of EVR-YFP chimeric protein to the periphery of epidermal cells (Fig. 4D), and early colocalization of EVR-YFP with the lipophilic dye FM4-64 was observed (Fig. S4C in the supplementary material). These results suggest that EVR is localized to the plasma membrane and/or internal membrane compartments (Vida and Emr, 1995).

RT-PCR and global expression experiments indicate that EVR transcripts are present in multiple tissues, with increasing expression in older flowers (see Fig. S5A,B in the supplementary material) (Schmid et al., 2005). To determine whether EVR is expressed at the right time and place to modulate organ separation, we generated transgenic plants that express β-glucuronidase (GUS) under the control of the predicted EVR promoter (see Fig. S4A in the supplementary material). Of 83 T1 plants examined, 58 showed GUS expression in floral organ AZs, typically just prior to organ shedding (stage 15) with strengthened expression in older flowers (stages 16 and early 17; Fig. 5A,C). This profile is similar to the reported expression patterns of HAE::GUS and HSL2::GUS (Fig. 5A-D) (Jinn et al., 2000; Cho et al., 2008), and is consistent with a role for EVR in modulating the timing of organ shedding.

Two independent EVR::GUS lines with single transgene insertions were characterized in the T2 generation. In addition to expression in floral organ AZs, each line showed GUS expression within the style of the developing fruit (stage 15; Fig. 5E), at the bases of the cauline leaves (Fig. 5G), and in the stems of the first rosette leaves (see Fig. S5C in the supplementary material). Because evr mutants showed defects in fruit growth (Fig. 2H), and global transcriptional analysis suggests that EVR has a broader expression profile than our EVR::GUS markers (see Fig. S5B in the supplementary material) (Schmid et al., 2005); these might not reflect the complete expression profile of EVR. The HAE promoter was also observed to direct GUS expression in the style (Fig. 5F), at the cauline leaf bases (Fig. 5H), and at the junction between the floral and inflorescence stems (Fig. 5B, arrow). These observations suggest that EVR and HAE may co-regulate the development of a set of plant tissues.

As mutations in EVR rescue abscission in nev flowers, we examined AZ cells at the time of organ shedding to determine whether the membrane trafficking defects observed in nev flowers were also rescued. Loss of NEV leads to a distinctive bending of the Golgi cisternae, disruption of the trans-Golgi network (TGN) and the accumulation of clusters of vesicles known as paramural bodies (PMBs) between the plasma membrane and cell wall of AZ cells (Liljegren et al., 2009). Unlike nev cells, nev evr cells contain flat stacks of Golgi cisternae with a wild-type appearance and closely associated TGN (Fig. 6A-C,E). The Golgi stacks and TGN of evr cells resemble those of wild type (Fig. 6D-E). As with nev cells, PMBs with more than 30 vesicles were observed in evr cells, but not in wild-type or nev evr cells (Fig. 6F-J; see also Fig. S6 in the supplementary material). As the defects in Golgi morphology and TGN location correlate with non-shedding floral organs, these might represent the primary cellular changes associated with and potentially responsible for the block of organ shedding in nev flowers. Furthermore, the restoration of Golgi/TGN structure in nev evr flowers and the defects in PMB formation in evr flowers suggest the possibility that, like NEV, EVR might regulate membrane trafficking at the transition to floral organ shedding.

To potentially identify additional regulators of abscission, we evaluated whether mutations in two EVR-related genes could also rescue nev-mediated defects. EVR shares 42 and 45% amino acid identity within the kinase domains of the PEPR1 and At1g17750 LRR-RLKs, respectively (Fig. 4B). PEPR1 was previously found to be a receptor for the PEP1 peptide involved in the innate immune response of plants to pathogen attack (Yamaguchi et al., 2006), and At1g17750 has been shown to be transcriptionally induced by the fungal protein, Nep1 (Necrosis and ethylene-inducing peptide1) (Qutob et al., 2006). EVR is also expressed in response to viral and bacterial infection (Whitham et al., 2003; Katiyar-Agarwal et al., 2007; Ascencio-Ibanez et al., 2008). Despite the similarities within the kinase domain and the overlapping expression patterns (see Fig. S5B in the supplementary material), predicted loss-of-function mutations in PEPR1 and At1g17750 did not rescue organ shedding in nev flowers (Fig. 7). These results suggest that these closely related kinases do not regulate organ separation, or that their roles are hidden by additional genetic redundancy.

As mutations in NEV, IDA, and the redundant genes HAE and HSL2 all appear to block floral organ shedding at a similar point during the cell separation stage (Butenko et al., 2003; Cho et al., 2008; Stenvik et al., 2008; Liljegren et al., 2009), we sought to use mutations in EVR to test the genetic relationship between these genes. We found that loss of EVR does not rescue shedding in either ida or hae hsl2 flowers (Fig. 8A-D). These results suggest that NEV and EVR could act upstream of IDA and HAE/HSL2, or in a parallel pathway that converges at the point of HAE/HSL2 function or further downstream (Fig. 8E).

Here, we report the characterization of EVR, an Arabidopsis LRR-RLK that modulates floral organ abscission and promotes fruit development. Our studies suggest that EVR functions to inhibit organ separation by regulating signaling that affects the timing of AZ activity, AZ size and cell expansion.

We were able to detect a role for EVR in organ separation through a sensitized genetic screen of nev flowers. NEV encodes an ADP-ribosylation factor GTPase-activating protein that is likely to regulate membrane trafficking during multiple aspects of plant development (Liljegren et al., 2009). At the time of abscission, loss of NEV activity could disrupt the movement of signaling molecules crucial for activating this process, thereby blocking release of the hydrolytic enzymes essential for the AZ cell wall modifications associated with cell separation. As described below, we propose that secondary loss of the EVR LRR-RLK might bypass the requirement for NEV-mediated trafficking during abscission by affecting the localization and/or activity of such a signaling complex.

We have discovered that EVR acts as both a temporal and spatial regulator of abscission. EVR regulatory regions first direct GUS expression in floral AZs during the transition from fertilization to fruit growth (stage 15; Fig. 5A,C). In wild-type flowers of the Ler ecotype, organ separation (stage 16) occurs about three days after the buds open and the stamens release their pollen (see Fig. S3B in the supplementary material). Loss of EVR causes premature abscission of floral organs in nev flowers (stage 15*) two positions earlier than in wild-type inflorescences (Fig. 2C), and about two days after bud opening (see Fig. S3B in the supplementary material). These results indicate that younger flowers are competent to respond to signals to shed their organs, as was previously observed for flowers treated with ethylene or misexpressing IDA (Butenko et al., 2003; Stenvik et al., 2006). More importantly, our results suggest that EVR is one of the factors that acts to inhibit this response. Although mutations in EVR alone do not affect the timing of abscission, multiple levels of genetic redundancy probably exist to ensure that organ separation does not occur prematurely.

Mutations in EVR appear to alter the spatial restriction of AZ signaling. Whereas smooth, scar-like surfaces form at the sites of organ detachment in wild-type flowers (Fig. 3A), increased expansion of individual AZ cells leads to visible collars of rough, broken tissue at the bases of nev evr fruit (Fig. 1C, Fig. 3F,G). AZ regions are also enlarged in nev evr flowers compared with wild type (Fig. 3E). These AZ phenotypes strongly resemble those of plants constitutively expressing IDA (Fig. 3E-H) (Stenvik et al., 2006; Cho et al., 2008), suggesting that excess levels of the IDA ligand or an activated HAE/HSL2 receptor complex might be present in nev evr flowers. An intriguing possibility is that ectopic AZ signaling causes cell wall loosening and expansion of cells both within and neighboring the original AZs of nev evr flowers. Although the relationship between AZ cell expansion and organ separation in wild-type Arabidopsis flowers is unclear (Patterson, 2001), cell expansion might physically enable the shedding of floral organs during the cell wall loosening process. In this capacity, the contribution of AZ cell expansion to organ abscission might be analogous to that of lignified fruit cells to Arabidopsis pod dehiscence (Liljegren et al., 2004).

With five LRRs in its extracellular domain, EVR may inhibit organ separation by regulating the behavior of ligand-binding LRR-RLKs. One of the best characterized interactions between non-ligand and ligand-binding LRR-RLKs is that of BAK1 (SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASE3) and FLAGELLIN-SENSITIVE2 (FLS2) during pathogen-triggered immunity (Geldner and Robatzek, 2008; Boller and Felix, 2009). Recognition of the flagellin-derived flg22 peptide by FLS2 triggers interaction with BAK1 (Chinchilla et al., 2007; Heese et al., 2007). As a co-receptor for FLS2, BAK1 promotes both flg22-induced signaling and internalization of FLS2 from the plasma membrane (Chinchilla et al., 2007). Although EVR may function similarly to BAK1 — by mediating the membrane trafficking of a ligand-binding receptor — we predict that it plays a unique role by acting as an inhibitor of its LRR-RLK partner prior to ligand binding, rather than as a co-receptor after ligand binding. Further studies will be necessary to pinpoint the location(s) of EVR activity; our initial results suggest that EVR functions at the plasma membrane and in closely associated vesicles (Fig. 4D; see also Fig. S4C in the supplementary material).

A model for EVR function during the transition from fertilization to floral organ shedding is shown in Fig. 9. In wild-type AZ cells, plasma membrane-localized EVR may interact with and inhibit the activity of ligand-binding LRR-RLK(s), such as HAE and HSL2. In contrast to BAK1, for which FLS2 binding is enhanced by the presence of ligand, EVR may promote the internalization of an inactive receptor complex, thereby limiting the pool of available receptors and delaying abscission signaling. Increased availability of the HAE/HSL2 receptors, or of the putative ligand IDA, may trigger cell separation in older flowers. Loss of EVR function alone would not alter the timing of floral organ abscission owing to the activity of a redundant LRR-RLK(s). This RLK would also be predicted to promote internalization of inactive receptor complexes in wild-type cells. Loss of NEV function in the TGN and other endosomal compartments could alter the trafficking of receptor complexes containing EVR and other inhibitory RLKs, such that the components required to activate cell separation are not recycled back to the plasma membrane. Secondary loss of EVR function could bypass the requirement for NEV in floral organ shedding, potentially resulting in a constitutive signal for cell separation that is no longer internalized or targeted for degradation. Increased signaling might be responsible for the premature organ shedding, enlarged AZs and deregulated cell expansion that we observe in nev evr flowers.

In addition to restoring abscission, we found that a mutation in EVR rescues the structural defects in the Golgi and TGN that are present in nev flowers. Defects in NEV-mediated membrane trafficking might leave the integrity of the Golgi/TGN vulnerable to organizational stress due to the high volume of traffic blocked during organ abscission. If mutations in EVR restore abscission signaling and thereby alleviate the backlog of traffic in nev AZ cells, they might be sufficient to indirectly restore Golgi/TGN structure. Alternatively, EVR/SOBIR1 might function in a pathway(s) that directly regulates Golgi dynamics during periods of cellular stress, such as abscission and pathogen attack. Perturbations in Golgi structure have been associated with cell death-associated kinase signaling and GTPase activity (Landry et al., 2009), and AZ tissues that undergo periodically high volumes of membrane trafficking might require specific signaling pathways to maintain the integrity of membrane-bound organelles. Future experiments investigating the role of the EVR RLK in regulating membrane trafficking during abscission are likely to advance our understanding of the signaling complexities that control plant growth and development.

We thank M. Duncan, M. Peifer, J. Reed, J. Kieber, B. Duronio, J. Dangl, C. Argueso, C. Burr and M. Simon for helpful discussions; V. Madden, S. Ray and T. Perdue for assistance with TEM, SEM and confocal microscopy; I. Chen, A. Larberg and R. Mulamba for technical assistance; J. Fletcher for the 35S::CLV3-GFP/GUS construct; T. Nakagawa for the pGWB destination vectors; ABRC for seed stocks; and Monsanto for access to Ler polymorphisms. This research was supported by an NSF grant (IOB-0517550) and an UNC URC grant to S.J.L. M.E.L. was supported by William R. Kenan, William C. Coker and Alma Holland Beers fellowships, and the UNC Curriculum in Genetics and Molecular Biology; M.W.L. was supported by the NIH-funded UNC Developmental Biology training program, and by William C. Coker and Alma Holland Beers fellowships. Deposited in PMC for release after 12 months.