A receptor-like cytoplasmic kinase, BIK1, associates with a flagellin receptor complex to initiate plant innate immunity

Flg22 Rapidly Induces BIK1 Phosphorylation.

Protein phosphorylation plays essential roles in diverse MAMP responses from receptor activation to downstream immune response gene expression (18). Although some phosphorylated proteins have been identified in response to MAMP treatment, their exact mode of action in MAMP signaling has remained elusive (19, 20). To identify kinases with a particular interest in RLCKs involved in MAMP signaling in Arabidopsis, we analyzed kinase-encoding transcripts that are induced upon flg22 and other MAMP treatment by searching through microarray data sets (21, 22). One gene identified in this way was BIK1, which was rapidly induced by flg22 and EF-Tu treatments (21, 22). The induction of BIK1 by flg22 was further confirmed by RT–PCR analysis in which BIK1 expression was significantly enhanced 30 min after flg22 treatment in 8-day-old Arabidopsis seedlings (Fig. 1A). Intriguingly, when we expressed HA-epitope-tagged BIK1 in Arabidopsis protoplasts, the flg22 treatment induced a mobility shift of BIK1 as detected by Western blot (Fig. 1B). We examined several other RLCKs that have been shown previously to play roles in plant defense or other signaling: oxidative signal-inducible 1 (OX1) is involved in oxidative-burst-mediated signaling in Arabidopsis (23), brassinosteroid-signaling kinase 1 (BSK1) mediates Arabidopsis BR signaling (24), and Pto induces effector-triggered immunity in tomato (25). The flg22 treatment did not induce the mobility shift of OX1, BSK1, and Pto as detected by Western blot (Fig. 1B).

The mobility of modified BIK1 could be restored to that of the unmodified form by the treatment of calf alkaline intestinal phosphatase (CIP) or a general kinase inhibitor K252a (Fig. 1C). The data indicate that the modification of BIK1 as detected by Western blot was caused by flg22-induced phosphorylation. We further performed a BIK1 phosphorylation time-course assay upon flg22 treatment. BIK1 phosphorylation was evident as early as 1 min upon flg22 treatment and gradually increased within 10 min (Fig. 1D). The rapid phosphorylation of BIK1 upon flg22 treatment demonstrates that BIK1 phosphorylation is one of the earliest events in flagellin signaling. BIK1 phosphorylation in response to flg22 was further confirmed in transgenic plants expressing BIK1 fused with an HA epitope tag (Fig. 1E). Plants could respond to multiple MAMPs and activate the convergent signaling (21). We therefore examined BIK1 phosphorylation upon different MAMP treatments. Significantly, elf18 (the N-terminal 18-amino-acid peptide of EF-Tu), but not chitin, also induced BIK1 phosphorylation in Arabidopsis transgenic plants (Fig. 1E). The results suggest that BIK1 is a convergent component involved in multiple MAMP signaling.

Flg22-Induced BIK1 Phosphorylation Depends on FLS2 and BAK1.

The flg22-induced BIK1 phosphorylation was dependent on the flg22 receptor FLS2 and its signaling partner BAK1 as phosphorylation was absent in the fls2 and bak1 mutant seedlings (Fig. 2A). Importantly, expression of FLS2 in the fls2 mutant protoplasts or of BAK1 in the bak1 mutant protoplasts could restore the BIK1 phosphorylation induced by flg22 (Fig. 2B). However, expression of the corresponding kinase-inactive mutants, FLS2Km or BAK1Km, failed to complement BIK1 phosphorylation, indicating that the kinase activity of FLS2 and BAK1 is required for BIK1 phosphorylation (Fig. 2B). Interestingly, the BAK1 closest homolog, SERK4, is not required for BIK1 phosphorylation as the flg22-induced BIK1 phosphorylation occurred normally in the serk4 mutant (Fig. 2B). The flg22-induced BIK1 phosphorylation could be suppressed by expression of the bacterial effector AvrPto (Fig. 2C), which is consistent with the finding that AvrPto suppresses MAMP signaling by targeting receptor complexes (13, 26). Therefore, the results demonstrate that BIK1 could be quickly phosphorylated by different MAMPs downstream of MAMP receptor complexes.

BIK1 Associates with FLS2 and BAK1.

Our data suggest that BIK1 is a component in the early MAMP signaling. Consistent with this notion, expression of BIK1 in Arabidopsis protoplasts moderately, but significantly, activated FRK1-LUC and WRKY29-LUC, two MAMP-responsive genes fused to a luciferase reporter (Fig. S1A). MAPK activation is an immediately early event upon MAMP perception (27). To determine the relationship of BIK1 phosphorylation and MAPK activation, we transiently coexpressed BIK1 with constitutively active MKK5, MEKK1, or full-length MEKK1 that has been shown to be activated in flg22 signaling (27). The MAPK components did not induce the phosphorylation of BIK1 (Fig. S1B), suggesting that BIK1 functions either upstream of the MAPK cascade or independently of MAPK signaling.

BIK1 is a plasma membrane localized protein with a putative myristoylation motif (17). To test whether BIK1 could associate with FLS2 and BAK1, we performed a coimmunoprecipitation (Co-IP) assay with coexpressing HA-epitope-tagged BIK1 and FLAG-epitope-tagged FLS2 or BAK1 in protoplasts. Clearly, BIK1 immunoprecipitated both FLS2 and BAK1 in vivo (Fig. 3 A and B). Interestingly, the association of BIK1 with FLS2 or BAK1 appears to be reduced upon flg22 treatment (Fig. 3 A and B). This result suggests that BIK1 might be released from the receptor complex upon phosphorylation by flagellin. The kinase activity of BIK1, BAK1, and FLS2 seems not to be required for the association because their kinase-inactive mutants still associate normally by Co-IP assay (Fig. S2 A and B). The BIK1-FLS2 association is independent of BAK1, and BIK1–BAK1 association is independent of FLS2 because bak1 and fls2 mutants did not affect the association of BIK1 with FLS2 or BAK1 (Fig. S3 A and B). To confirm this interaction, we further performed a pull-down assay with an N-terminal GST–BIK1 fusion protein immobilized on glutathione beads as bait against lysates of Arabidopsis protoplasts expressing HA-epitope-tagged FLS2 or BAK1. Both FLS2 and BAK1 could be pulled down strongly by GST-BIK1, but not by GST itself (Fig. 3C). The treatment of flg22 seems not to significantly affect FLS2–BIK1 and BAK1–BIK1 interaction with this assay in which BIK1 was not exposed to flg22 treatment. BIK1 is a membrane-associated intracellular cytoplasmic kinase. We next tested whether FLS2 and BAK1 associate with BIK1 through their kinase domains. Using a GST pull-down assay, BIK1 could be clearly pulled down by GST-BAK1K and GST-FLS2K, the kinase domain of BAK1 (BAK1K) or FLS2 (FLS2K) fused to an N-terminal GST (Fig. 3D). This was further confirmed by a Co-IP assay, in which BIK1 coimmunoprecipitated FLS2K (Fig. 3E) and BAK1K (Fig. 3F). Taken together, the data demonstrate that BIK1 functions in MAMP signaling by interacting with the MAMP receptor complex.

BIK1 Transphosphorylates FLS2 and BAK1.

BIK1 is predicted to encode a Ser/Thr protein kinase (17). To determine the amino acid residues that are essential for flg22-induced phosphorylation and gain insight into how BIK1 functions, we searched the potential BIK1 phosphorylation sites in The Arabidopsis Protein Phosphorylation Site Database (http://phosphat.mpimp-golm.mpg.de/app.html) and aligned the activation domain of BIK1 with several related kinases (Fig. 4A). We individually substituted several putative Ser and Thr phosphorylation sites in the BIK1 activation domain with Ala and tested their effect on flg22-induced BIK1 phosphorylation. Significantly, a Thr²³⁷ → Ala²³⁷ (T237A) mutation completely eliminated the BIK1 phosphorylation by flg22 whereas Ser²³³, Ser²³⁶, and Thr²⁴² seem to be dispensable for flg22-induced BIK1 phosphorylation (Fig. 4B). The data suggest that Thr²³⁷ is a major phosphorylation site of BIK1 in response to flg22. However, additional phosphorylation sites that are not detected by Western blot might be present in BIK1 (19). BIK1^T237A did not reduce its interaction with FLS2 and BAK1 as detected by an in vivo Co-IP assay (Fig. S2).

Our findings on BIK1 interaction with FLS2/BAK1 and rapid flg22-induced BIK1 phosphorylation suggest that BIK1 might be a substrate of BAK1 and/or FLS2. BIK1 exhibited strong autophosphorylation activity with an in vitro kinase assay (Fig. S4A). This autophosphorylation was eliminated in BIK1Km, which carries a mutation in the ATP-binding site (Fig. S4A). Thus, GST-BIK1Km was used as a substrate for the in vitro kinase assay. We purified the cytosolic domains of BAK1 (BAK1CD) and FLS2 (FLS2CD) as MBP or GST fusion proteins in Escherichia coli. The cytosolic domain includes both the intracellular kinase domain and the juxtamembrane domain. Significantly, BAK1CD directly phosphorylated BIK1Km in vitro in the presence of [³²P]-γ-ATP (Fig. 4C). The phosphorylation depends on the kinase activity of BAK1 because BAK1CDKm completely eliminated its phosphorylation on BIK1 (Fig. 4C). A similar result was obtained with an immunocomplex kinase assay in which FLAG-epitope-tagged BAK1 was expressed in protoplasts with flg22 treatment for an additional 10 min. BAK1 was pulled down with FLAG antibody and subjected to an in vitro kinase assay using GST-BIK1Km as a substrate. The immunoprecipitated BAK1, not kinase-inactive BAK1Km, phosphorylated BIK1Km (Fig. S5). The BIK1 phosphorylation seems to be enhanced on flg22 treatment (Fig. S5). We did not observe the in vitro phosphorylation of BIK1Km by FLS2CD. However, the autophosphorylation activity of FLS2CD was weaker than that of BAK1CD under our in vitro assay condition (Fig. S4B). The data indicate that BAK1 could directly phosphorylate BIK1 and that BIK1 is a substrate of BAK1.

Surprisingly, BIK1 could also phosphorylate FLS2 and BAK1. In an immunocomplex kinase assay, the immunoprecipitated BIK1 could phosphorylate GST-BAK1K and GST-FLS2K, the kinase domains of BAK1 and FLS2 fused with GST (Fig. 4D). GST-BAK1K and GST-FLS2K did not exhibit detectable autophosphorylation activity (Fig. S4A). Apparently, the phosphorylation was significantly enhanced with the BIK1 that was activated upon flg22 treatment (Fig. 4D). This unexpected finding was substantiated by an in vitro kinase assay in which GST-BIK1 directly phosphorylated GST-FLS2K and GST-BAK1K (Fig. 4E and Fig. S6A). The specificity of this phosphorylation was demonstrated by the fact that BIK1Km did not phosphorylate FLS2K and BAK1K (Fig. 4E and Fig. S6A). To rule out the possibility that phosphorylation of GST-BAK1K and GST-FLS2K might result from the enhanced autophosphorylation of BAK1K or FLS2K in the presence of BIK1, we generated the catalytically inactive kinase mutants BAK1KKm and FLS2KKm. GST-BIK1 also phosphorylated GST-BAK1KKm and GST-FLS2KKm in vitro (Fig. 4E). Similarly, BIK1, but not BIK1Km, could phosphorylate BAK1CDKm (Fig. S6B). Our data suggest the transphosphorylation of BIK1 with FLS2/BAK1 in the flagellin receptor complex. This phosphorylation was significantly reduced by BIK1^T237A, although not completely eliminated (Fig. 4E). BIK1^T237A also significantly reduced its autophosphorylation activity (Fig. 4E). The results suggest that BIK1 Thr²³⁷, the flg22-mediated phosphorylation site, is required for its phosphorylation on FLS2 and BAK1. Importantly, expression of BIK1^T237A significantly reduced the activation of the MAMP-responsive genes FRK1-LUC and WRKY29-LUC compared to wild-type BIK1 (Fig. S7), suggesting that this residue is essential for BIK1 to function in MAMP signaling. The functional importance of Thr²³⁷ in plant defense gains further support from the recent study concluding that the corresponding residue in the tomato BIK1 ortholog TPK1b is required for mediating Botrytis cinerea resistance in Arabidopsis (28).

BIK1 Is Required in Flagellin-Triggered Immunity.

To further verify the functional significance of flg22-induced BIK1 phosphorylation, we isolated an insertional mutant of the BIK1 gene from the Salk T-DNA insertion collection (Salk_005291) (29). RT–PCR analysis confirmed that bik1 is a null allele with undetectable transcript [Fig. S8 A and B (17)]. The bik1 mutant did not affect flg22-induced FLS2 and BAK1 association (Fig. S3C), suggesting that BIK1 is not required for flg22 binding to FLS2. We tested the bik1 sensitivity to flg22 with seedling growth assay. The bik1 mutant showed a clear reduction of sensitivity to flg22 compared to wild-type plants although to a lesser extent than the bak1 mutant (Fig. 5A). We next performed a pathogen infection assay with wild-type and bik1 mutant seedlings and adult plants. Significantly, bik1 mutant seedlings lost flg22-induced resistance to Pseudomonas syringae pv. tomato (Pst) DC3000 infection. Flg22-pretreated wild-type seedlings were more resistant than H₂O-pretreated seedlings as detected by bacterial multiplication 3 and 5 days after infection (Fig. 5B). However, the bacterial growth did not differ significantly in flg22- and H₂O-pretreated bik1 seedlings (Fig. 5B). Although bik1 mutants are more resistant to virulent Pst DC3000 infection than wild-type plants because of the elevated salicylic acid (SA) level (17), the enhanced resistance of plate-grown seedlings was not as pronounced as soil-grown plants probably because high humidity in plates could reduce the high SA level in the bik1 mutant (Fig. 5B and Fig. S8C). The flg22-induced resistance was also not observed in 4-week-old bik1 mutant plants inoculated with Pst DC3000 (Fig. S8C). Furthermore, when we infected wild-type and bik1 mutant seedlings with the nonpathogenic Pst DC3000 type III secretion mutant hrcC, bik1 mutant seedlings were significantly immunocompromised (Fig. 5C and Fig. S8D). Bacterial growth assay indicated that the bacterial number was about five- to-eightfold higher in the bik1 mutant than in wild-type seedlings 3 days after infection (Fig. 5C). Six days after infection with hrcC, bik1 mutant seedlings exhibited severe necrotic symptoms, whereas the wild-type seedlings were still green at this time point (Fig. S8D). The bik1 mutant complemented with the BIK1 gene completely restored plant immunity to hrcC to the wild-type level (Fig. 5C and Fig. S8D). Plant immunity to nonpathogenic Pst DC3000 hrcC infection is likely attributed by the action of multiple MAMPs. The data strongly support that BIK1 is an important signaling component in flagellin and likely in other MAMP signaling.

Plant Material and Growth Conditions.

Wild-type (Col-0), bik1, fls2, bak1-4, and serk4 mutant Arabidopsis plants were grown in a growth room at 23°C, 60% relative humidity, 70 μE light with a 12-h photoperiod for 30 days before protoplast isolation or bacterial inoculation. The fls2, bak1-4, and serk4 mutants and bik1 complementation lines were reported previously (13, 17). The bik1 mutant (Salk_005291) was obtained from the Arabidopsis Biological Resource Center and confirmed by PCR and RT–PCR analyses. Seedlings were grown on a 1/2 Murashige and Skoog medium (MS) plate with 1% sucrose and 0.9% agar at 23°C and 70 μE light with a 12-h photoperiod for 12 days. Seedlings were transferred to 2 mL H₂O in the six-well tissue culture plates 1 day before flg22 treatment for 10 min. RT–PCR analysis and flg22-mediated inhibition of seedling growth were carried out as described (13, 32). All experiments were repeated three to four times with reproducible results.

Plasmid Constructs, Protoplast Transient Assay, and Generation of Transgenic Plants.

Arabidopsis BAK1 and FLS2 constructs were reported previously (13). BIK1, OX1, and BSK1 genes were amplified by PCR from Col-0 cDNA and introduced into a plant expression vector with an HA or FLAG epitope tag at the C terminus. BIK1 point mutations were generated by a site-specific mutagenesis kit (Stratagene). The primer sequences for different kinases and BIK1 point mutations are listed in the SI Experimental Procedures. Full-length BIK1, a cytosolic domain, or the kinase domain of BAK1 and FLS2 were subcloned into the modified GST fusion protein expression vector pGEX4T-1 (Pharmacia) or pMAL-c2 (New England Biolabs) with BamHI and StuI digestion. Protoplast transient assay was carried out as described (13, 27). For BIK1 phosphorylation assays, 0.1-mL protoplasts at a density of 2 × 10⁵/mL were transfected with 20 μg of plasmid DNA. For Co-IP assays, 1-mL protoplasts were transfected with 200 μg of DNA. For GST pull-down assays, 0.5-mL protoplasts were transfected with 100 μg of DNA. The BIK1 transgenic plants in Col-0, fls2, and bak1-4 were generated by Agrobacterium-mediated transformation with the BIK1 construct under the control of a constitutive cauliflower mosaic virus 35S promoter with an HA epitope tag.

Elicitor and Chemical Inhibitor Treatments.

The flagellin peptide flg22 (32) and the EF-Tu peptide elf18 (21) were used in a concentration of 1 μM if not otherwise stated. Chitin fragments (33) were used at a final concentration of 50 μg/mL. Chemical inhibitors K-252a, diphenylene iodonium (DPI), and U0126 were purchased from A. G. Scientific, prepared as a stock solution of 1 mM, and used at a final concentration of 1 μM for K-252a, 5 μM for DPI, and 1 μM for U0126. Different chemical inhibitors were added 1 h before the 1-μM flg22 treatment. CIP was purchased from New England BioLabs, and the treatment was carried out following the instruction.

Coimmunoprecipitation and GST Pull-Down Assays.

Protoplasts were lysed with 0.5 mL of extraction buffer (10 mM Hepes, pH 7.5, 100 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% Triton X-100, and a protease inhibitor mixture from Roche). After vortexing vigorously for 30 s, the samples were centrifuged at 12,470 ×g for 10 min at 4°C. For the Co-IP assay, the supernatant was incubated with anti-HA or anti-FLAG antibody for 2 h and then with protein-G-agarose beads for another 2 h at 4°C with gentle shaking. For the GST pull-down assay, the supernatant was incubated with prewashed GST, GST-BIK1, GST-BAK1K, or GST-FLS2K beads for 2 h at 4°C with gentle shaking. The beads were collected and washed three times with washing buffer (10 mM Hepes, pH7.5, 100 mM NaCl, 1 mM EDTA, 10% glycerol, and 0.1% Triton X-100) and once with 50 mM Tris·HCl, pH7.5. The immunoprecipitated proteins were analyzed by Western blot with an anti-HA or anti-FLAG antibody. The protein bands with appropriate molecular weight are shown.

In Vitro Phosphorylation and Immunocomplex Kinase Assays.

Expression of the GST and MBP fusion proteins and affinity purification were performed as standard protocol. The protein concentration was determined with the BIO-RAD Quick Start Bradford Dye Reagent and confirmed by the NanoDrop ND-1000 Spectrophotometer. For in vitro kinase assay, kinase reactions were performed in 30 μl of kinase buffer (20 mM Tris·HCl, pH 7.5, 10 mM MgCl₂, 5 mM EGTA, 100 mM NaCl, and 1 mM DTT) containing 10 μg of fusion proteins with 0.1 mM cold ATP and 5 μCi of [³²P]-γ-ATP at room temperature for 3 h with gentle shaking. The reactions were stopped by adding 4× SDS loading buffer. The phosphorylation of fusion proteins was analyzed by autoradiography after separation with SDS/PAGE. For immunocomplex kinase assays, protoplasts were lysed with 0.5 mL of IP buffer (50 mM Tris·HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1 mM DTT, 2 mM NaF, 2 mM Na₃VO₃, 1% Triton, and a protease inhibitor mixture from Roche). After centrifugation at 12,470 ×g for 10 min at 4°C, the supernatant was incubated with anti-HA or anti-FLAG antibody for 2 h and then with protein-G–agarose beads for another 2 h at 4°C with gentle shaking. The beads were collected and washed once with IP buffer and once with kinase buffer (20 mM Tris·HCl, pH 7.5, 20 mM MgCl_2, 5 mM EDTA, and 1 mM DTT). The kinase reactions were performed in 20 μl of kinase buffer with 2 μg of GST fusion proteins, 0.1 mM cold ATP, and 5 μCi of [³²P]-γ-ATP at room temperature for 1 h with gentle shaking. The phosphorylation of GST fusion proteins was analyzed by 10% SDS/PAGE.

Pathogen Infection Assays.

P. syringae tomato DC3000 and hrcC strains were grown overnight at 28°C in King’s B medium with 50 μg/mL rifampicin. Bacteria were collected, washed, and diluted to the desired density with H₂O. For the 4-week-old plant flg22 protection assay, leaves were preinoculated with 200 nM flg22 or an H₂O control for 24 h and then infiltrated with Pst DC3000 at the concentration of 5 × 10⁵ cfu/mL using a needleless syringe (22). For seedling infection assay, seedlings were grown in 1/2 MS medium for 10 days in a 12-well tissue culture plate. Bacteria were added at the concentration of 2 × 10⁸ cfu/mL in each well. Bacterial counting was performed from six leaves as three repeats by surface sterilization with 70% ethanol (34, 35).