Plant immune responses, including the production of reactive oxygen species (ROS), are triggered when pattern recognition receptors (PRRs) become activated upon detection of microbe-associated molecular patterns (MAMPs). Receptor-like cytoplasmic kinases are key components of PRR-dependent signaling pathways. In tomato, two such kinases, Pti1a and Pti1b, are important positive regulators of the plant immune response. However, it is unknown how these kinases control plant immunity at the molecular level and how their activity is regulated. To investigate these issues, we used mass spectrometry to search for interactors of Pti1b in Nicotiana benthamiana leaves and identified a PP2C protein phosphatase, referred to as Pic1. An in vitro pull-down assay and in vivo split-luciferase complementation assay verified this interaction. Pti1b was found to autophosphorylate on threonine-233, and this phosphorylation was abolished in the presence of Pic1. An arginine-to-cysteine substitution at position 240 in the Arabidopsis MARIS kinase was previously reported to convert it into a constitutive-active form. The analogous substitution in Pti1b made it resistant to Pic1 phosphatase activity, although it still interacted with Pic1. Treatment of N. benthamiana leaves with the MAMP flg22 induced threonine phosphorylation of Pti1b. The expression of Pic1, but not a phosphatase-inactive variant of this protein, in N. benthamiana leaves greatly reduced ROS production in response to treatment with MAMPs flg22 or csp22. The results indicate that Pic1 acts as a negative regulator by dephosphorylating the Pti1b kinase, thereby interfering with its ability to activate plant immune responses.
Plants exist in an environment containing large numbers of diverse microorganisms, some of them detrimental to plant health. Consequently, their survival depends on their ability to specifically recognize and respond to pathogenic microbes. This recognition is enabled by the function of plasma membrane-localized pattern recognition receptors (PRRs), which recognize microbe-associated molecular patterns (MAMPs) [1–3]. The best understood PRR is FLS2, which perceives the peptide flg22, a part of the flagellin protein, which forms the bacterial flagellum [4,5]. Two PRRs that are present only in solanaceous species, including tomato, are FLS3, which recognizes a flagellin-derived peptide, flgII-28 and CORE, which detects an MAMP in the bacterial cold-shock protein, csp22 [6,7]. The binding of the extracellular part of the receptor to the appropriate ligand activates the cytoplasmic kinase domain of the receptor .
Activated PRRs induce a defense response referred to as pattern-triggered immunity (PTI). PTI is associated with the generation of reactive oxygen species (ROS), activation of mitogen-activated protein kinases (MAPKs), transcriptional reprograming, production of antimicrobial molecules and the reinforcement of the cell wall, among other processes [9–12]. Receptor-like cytoplasmic kinases (RLCK) play a key role in linking PRRs to downstream signaling components . RLCKs constitute a large family of proteins consisting primarily of a serine–threonine kinase domain, but some members also possess EGF, LRR, WD40 and transmembrane domains .
The Arabidopsis thaliana genome encodes 179 RLCKs divided into 17 subfamilies: RLCK-II and from RLCK-IV to RLCK-XIX. The majority of RLCKs which are known to have a role in plant immunity belong to sub-family VII. For example, AtBIK1, a member of sub-family VII, is activated upon phosphorylation by FLS2 after its perception of flg22 [13,14]. The activated BIK1 induces the generation of ROS by phosphorylating the NADPH oxidase, RBOHD [11,15]. Similarly, the activated PRR chitin elicitor receptor kinase 1 (CERK1) phosphorylates the RLCK protein PBL27, which then directly activates an immunity-associated MAPK kinase cascade . Some RLCKs play a negative regulatory role such as PBL13 whose activation reduces the generation of ROS in Arabidopsis and hinders resistance against the bacterial pathogen Pseudomonas syringae pv. tomato .
RLCKs also play important roles in regulating plant growth, development and reproduction. One such RLCK is MARIS, which belongs to sub-family VIII. MARIS was identified in a forward genetic screen as a suppressor of the anx1/anx2 mutations . This double mutation disrupts cell wall integrity (CWI) signaling and results in a burst of the pollen tube after pollen germination. The MARIS allele identified in the screen contains a mutation that introduces an arginine-to-cysteine substitution at position 240 (R240C) in the activation loop of the kinase which rescues the pollen tube bursting phenotype of the anx1/anx2 mutant . This and other observations suggested that MARIS acts downstream of the ANX1/ANX2 receptors in CWI signaling and that the R240C variant is a constitutively active form of MARIS , although the molecular basis for this constitutive activity is unknown.
Defense responses in plants need to be tightly regulated to prevent damage to plant tissue. Since immune responses are often activated by protein kinases, protein phosphatases are natural candidates as a negative regulator of plant responses to pathogen attack. Indeed, a few phosphatases are known to function as regulators of PRRs, immunity-associated RLCKs and MAPKs. The kinase-associated protein phosphatase (KAPP), for example, interacts with FLS2 and KAPP overexpression blocks flg22-dependent signaling . The rice PRR XA21 which confers resistance to the bacterial pathogen Xanthomonas oryzae pv. oryzae interacts with XB15, which belongs to the protein phosphatase 2C (PP2C) family . XB15 dephosphorylates the autophosphorylated kinase domain of XA21. Rice plants with a mutation in the Xb15 gene show symptoms of activated defense responses, whereas plants overexpressing Xb15 are more susceptible to X. o. pv. oryzae . In Nicotiana benthamiana, virus-induced gene silencing of a gene encoding a catalytic subunit of protein phosphatase 2A results in activation of plant defense responses, but the mechanism is not known . In Arabidopsis, the PRR co-receptor BAK1 interacts with a phosphatase PP2A holoenzyme. PP2A also negatively controls the phosphorylation status of BAK1, thereby interfering with subsequent PTI signaling . The RLCK BIK1 is known to interact with phosphatase PP2C38. The phosphatase dephosphorylates BIK1 and impedes its ability to phosphorylate the NADPH oxidase RBOHB causing impaired ROS production and decreased stomatal-mediated immune responses . Finally, some MAPK phosphatases are known to negatively regulate aspects of PTI including ROS production [24,25].
In tomato, two highly similar kinases referred to as Pti1a (Pto interactor 1a) and Pti1b play a role in PTI against P. s. pv. tomato . Like MARIS, Pti1a and Pti1b are members of RLCK family VIII and consist primarily of a serine–threonine protein kinase domain . We previously generated and characterized tomato plants carrying a hairpin (hp)-Pti1 construct, which consequently have reduced expression of both Pti1a and Pti1b. The hpPti1 transgenic plants are impaired in ROS production in response to treatment with MAMPs flg22 or flgII-28, whereas activation of MAPK kinases is unaffected . Importantly, the hpPti1 plants are significantly more susceptible to P. s. pv. tomato. Pti1a was previously shown to intramolecularly autophosphorylate on threonine-233 in the activation loop of the kinase; however, the possible role of this phosphorylation in Pti1a-mediated PTI signaling is unknown [27,28].
Homologs of the Pti1 kinases exist in many other plant species. In Arabidopsis, it was shown that Pti1–2 and Pti1–4 play a role in oxidative stress [29,30], and the kinase activity of Pti1–2 is induced by flg22 . Arabidopsis Pti1–1 and Pti1–2 proteins, but not Pti1–3 or Pti1–4, show strong autophosphorylation activity . In cucumber, a Pti1 homolog CsPti1-L is a positive regulator of plant immunity and salt tolerance and also exhibits autophosphorylation activity . In contrast, it was shown in rice that OsPti1a acts as a negative regulator of basal defense against X.o. pv. oryzae and Magnaporthe oryzae [32,33]. As with the tomato Pti1a kinase, threonine-233 of OsPti1a undergoes autophosphorylation in vitro and additionally is phosphorylated by the OsOxi1 kinase . The authors postulate that the negative regulatory function of OsPti1a might be abolished by OsOxi1-dependent phosphorylation of OsPti1a on threonine 233 . An OsPti1a(T233A) variant is unable to restore basal resistance against X. o. pv. oryzae in an ospti1a mutant background in comparison with OsPti1 wildtype .
Although much is known about the enzymatic activity and upstream activators of Pti1-like kinases, little is known about their downstream substrates or associated proteins that might regulate their autophosphorylation status. To find potential substrates and regulators of the Pti1b kinase, we used a combined FLAG co-immunoprecipitation and Strep pull-down approach coupled with mass spectrometry. We focused our research on Pti1b because previous RNA-Seq analyses indicated that the expression of the Pti1b gene and not the Pti1a gene is induced by MAMPs csp22 and flgII-28 and infection by P. syringae pv. tomato, suggesting it might play a more prominent role in PTI . We found that Pti1b expressed in N. benthamiana leaves co-purifies with a phosphatase of the PP2C family. We verified this interaction using additional assays and investigated the role of the phosphatase in regulating Pti1b. Our observations indicate that Pic1 dephosphorylates Pti1b and acts as a negative regulator of PTI signaling. Consequently, we refer to the phosphatase as pattern-triggered immunity inhibiting PP2C1, Pic1.
The AtPIP2a gene was transferred from plasmid pCambia-AtPIP2a-cLuc (a kind gift from G. Coaker; Univ. California-Davis) into the pJLSmart Gateway entry vector  and then moved into the pGWB-SF destination plasmid that enables protein expression in fusion with a 2xStrep-FLAG- at the C terminus of the protein. The pGWB-SF plasmid was obtained by cloning a 2xStrep sequence into pGWB411 vector . Pti1b(K96N)-pJLSmart and Pti1b(R234C)-pJLSmart were generated with PCR-based site-directed mutagenesis with construct Pti1b-pJLSmart  as the template. Pti1b, Pti1b(K96N) and Pti1b(R234C) genes were then transferred from pJLSmart into the pGWB-SF destination plasmid. For protein expression in Escherichia coli, the Pti1b gene was amplified from Pti1b-pJLSmart with primers that enabled cloning into the pET30a+ vector using restriction enzyme NdeI and NotI. Pti1b(K96N), Pti1b(S232A), Pti1b(T233A), Pti1b(S232A,T233A), Pti1b(T233D), Pti1b(R234C) and Pti1b(T233A,R234C) variants were generated with PCR-based site-directed mutagenesis. The appropriate Pti1b variant in the pET30a+ plasmid was used as the template for the PCR. The Pic1 gene was amplified from tomato cDNA and cloned into the pASK3 plasmid using restriction enzymes XbaI and AfeI. The Pic1(NN) variant was generated with PCR-based site-directed mutagenesis with Pic1-pASK3 as the template. For ROS assays in N. benthamiana Pic1 variants were amplified from pASK3 plasmids and cloned into pJLTRBO  with restriction enzymes NotI and AvrII. Similarly, the GFP gene was amplified from plasmid pGWB505 and cloned into the pJLTRBO plasmid using the same restriction enzymes. To carry out the split-luciferase complementation assay (SLCA) Pti1b and Pic1 genes were amplified with PCR and cloned into pCambia-nLuc and pCambia-cLuc, respectively, using restriction enzymes KpnI and SalI. All primers used in this study are listed in Supplementary Tables S1 and S2.
Bacteria growth and infiltration
Agrobacterium tumefaciens GV3101 was grown on LB medium with appropriate antibiotics for 48 h at 30°C. The bacterial cells were collected and suspended in buffer containing: 10 mM MES, 10 mM MgCl2, 200 µM acetosyringone (pH 5.7). For agroinfiltration into N. benthamiana leaves, bacteria were diluted to OD 1.0 and incubated for 3 h at room temperature. N. benthamiana leaves were infiltrated with a needleless syringe.
Protein expression and purification
For mass spectrometry analysis, proteins in fusion with 2xStrep-FLAG were purified with Anti-FLAG M2 Magnetic Beads (Sigma Co.). Purification was carried out according to the manufacturer's protocol with minor modifications. Specifically, 5 g of N. benthamiana leaf tissue was ground in liquid nitrogen and dissolved in 7.5 ml of extraction buffer containing: 150 nM Tris–Cl, 150 mM NaCl, 5 mM DTT, 10 mM MnCl2, 1% [v/v] Triton X-100, PhosSTOP™ — phosphatase inhibitor tablets (Sigma), cOmplete™, Mini, EDTA-free Protease Inhibitor Cocktail (Roche Co.) (pH 7.4). After 30 min incubation on a rotator at 4°C, the protein samples were centrifuged at 18 000×g for 30 min to remove plant debris and the protein extract was filtered through Miracloth filter paper (Calbiochem Co.). Next, 20 µl of previously washed anti-FLAG magnetic beads in an amount equivalent to the 20 µl of magnetic bead suspension was added to the protein extract. The sample was incubated on a rotator for 1 h at 4°C, and the magnetic beads were then collected using a magnetic rack. The beads were washed three times with 200 µl extraction buffer and eluted with 100 µl of an elution buffer containing: 100 mM Tris–Cl, 150 mM NaCl, 5 mM DTT, 10 mM MnCl2 (pH 8.0) and 450 ng/µl 3xFLAG peptide. After elution, MagStrep ‘type 3’ XT beads (https://www.iba-lifesciences.com) in an amount equivalent to the 10 µl of magnetic bead suspension were added to the sample. After 1.5 h of incubation on a rotator at 4°C, the beads were washed two times with 200 µl of washing buffer containing: 100 mM Tris–Cl, 150 mM NaCl, 5 mM DTT (pH 8.0). The proteins bound to the beads were then eluted by boiling in Laemmli buffer.
To purify 6xHis-tagged proteins, a frozen pellet of E. coli strain DE3(pLys)Rosetta was thawed in buffer containing: 50 mM Tris–Cl, 1 M NaCl2, 10 mM Imidazole, 5% glycerol, cOmplete™ Mini EDTA-free Protease Inhibitor Cocktail (Roche Co.), 1 mg/ml lysozyme (pH8.0). After sonication, the protein extract was centrifuged at 18 000×g for 30 min to remove bacterial debris. The supernatant was incubated on a rotator with Ni-NTA Agarose (Qiagen Co.) for 1.5 h at 4°C. Next, the sample was loaded onto a column, and after liquid removal, the resin was washed three times with 5 ml of buffer containing: 100 mM Tris–Cl, 1 M NaCl2, 10% glycerol, 10 mM imidazole (pH 8.0). After washing, the protein bound to the resin was eluted with buffer containing: 50 mM Tris–Cl, 1 M NaCl2, 10% glycerol, 250 mM imidazole (pH 8.0). For purification of Strep-tagged proteins, frozen E. coli strain DE3(pLys)Rosetta cells were thawed in buffer containing: 150 mM Tris–Cl, 150 mM NaCl2, 5 mM DTT, cOmplete™ Mini EDTA-free Protease Inhibitor Cocktail (Roche) (pH 8.0). After sonication, the protein extract was centrifuged 18 000×g for 30 min to remove bacterial debris. The supernatant was incubated on a rotator with Strep-Tactin Macroprep resin (IBA Co.) for 1.5 h at 4°C. Next, the resin was washed three times with 5 ml of buffer: 150 mM Tris–Cl, 150 mM NaCl2, 5 mM DTT (pH 8.0), next the proteins bound to the resin were eluted with buffer containing: 150 mM Tris–Cl, 150 mM NaCl2, 5 mM DTT, 2.5 mM biotin (pH 8.0).
Protein samples were separated in 10% bisacrylamide gels at 200 V. Detection of phosphorylated proteins was performed in blocking solution (5% BSA, 0.1% [v/v] Tween-20) with anti-phosphothreonine antibodies (9381, Cell Signaling) and with anti-rabbit-HRP (W4011, Promega). Detection of tagged proteins was carried out in blocking solution (5% fat-free milk, 0.5% [v/v] Tween-20). Other antibodies used include: StrepMAB-Classic, HRP conjugate (2-1509-001, IBA), anti-HA High Affinity (11867423001, Roche), anti-polyHistidine antibody (H1029, Sigma) ANTI-FLAG® M2 (F1804, Sigma), anti-first 258 amino acids of luciferase (NB600, Novus), anti-C-terminal part of luciferase (sc-74548, Santa Cruz Biotechnology), anti-GFP (11814460001, Roche), anti-rat-HRP (sc-2006, Santa Cruz Biotechnology) and anti-mouse-HRP (sc2005, Santa Cruz Biotechnology).
Mass spectrometry analysis
Peptide mixtures were analyzed by LC–MS /MS (liquid chromatography coupled to tandem mass spectrometry) using Nano-Acquity (Waters) LC system and LTQ-FT-Orbitrap mass spectrometer (Thermo Electron Corp., San Jose, CA, U.S.A.). Prior to the analysis, proteins were subjected to a standard ‘in-solution digestion’ procedure during which proteins were reduced with 50 mM TCEP (for 60 min at 60°C), alkylated with 200 mM MMTS (45 min at room temperature) and digested overnight with trypsin (sequencing grade modified Trypsin — Promega V5111). The peptide mixture was applied to RP-18 precolumn (nanoACQUITY Symmetry® C18 — Waters 186003514) using water containing 0.1% TFA as a mobile phase and then transferred to a nano-HPLC RP-18 column (nanoACQUITY BEH C18 — Waters 186003545) using an acetonitrile gradient (5–35% AcN) for 180 min in the presence of 0.05% formic acid with a flow rate of 250 nl/min. Column outlet was directly coupled to the electrospray ion source of the spectrometer working in the regime of data dependent MS to MS/MS switch. A blank run to ensure a lack of cross-contamination from previous samples preceded each analysis. The data acquired were processed by Mascot Distiller followed by Mascot Search (Matrix Science, London, U.K., on-site license) against the database ‘NBenthLuteoviridaecRAPBSAProtAmod’ is accessible at ftp://ftp.solgenomics.net/proteomics/Nicotiana_benthamiana/Cilia_lab/proteomics_db/ (Stacy DeBlasio and Michelle Cilia, 2014 personal communication). Search parameters for precursor and product ions mass tolerance were 30 ppm and 0.1 Da, respectively, enzyme specificity: trypsin, missed cleavage sites allowed: 1, fixed modification of cysteine by methylthio and variable modification of methionine oxidation. Peptides with Mascot Score exceeding the threshold value corresponding to <5% expectation value, calculated by Mascot procedure, were considered to be positively identified. Detected phosphorylated peptides were manually inspected.
For kinase assays, the proteins were incubated for 1 h at room temperature in buffer containing: 150 mM HEPES, 150 mM NaCl2 10 mM MgCl2, 10 mM MnCl2, 2 mM DTT, 20 µM ATP and 1 µCi of [γ-32P]ATP (PerkinElmer Co.) (pH 8.0). The total volume of the reaction mixture was 20 µl. The reaction was terminated by adding Laemmli buffer and boiling for 5 min. The proteins were separated on 10% polyacrylamide gels and visualized by autoradiography. To confirm protein loading, the gels were stained with Coomassie Blue. The amount of protein used in each reaction is stated in the figure legend.
Strep-tag pull-down assay
To decrease nonspecific interactions between MagStrep ‘type 3’ XT beads and proteins, the beads were incubated in 100 µl of buffer containing: 150 mM HEPES, 150 mM NaCl2 10 mM MgCl2, 10 mM MnCl2, 2 mM DTT, 20 µM ATP, 0.1 µg/µl BSA, 1% Tween (pH 8.0) on a rotator for 1 h at 4°C. Next 30 µg of 6xHis-tagged Pti1b variants and Pic1 in fusion with Strep-tag were added to the solution with the beads. Samples were incubated for 2 h on a rotator at 4°C and the beads were collected using a magnetic rack and washed two times with 300 µl of washing buffer containing: 150 mM HEPES, 150 mM NaCl2 10 mM MgCl2, 10 mM MnCl2, 2 mM DTT, 20 µM ATP, 1% Tween (pH 8.0). Next, the beads were resuspended again in the washing buffer and transferred to a new Eppendorf tube. The beads were collected with a magnetic rack, and the proteins were eluted by boiling in Laemmli buffer. The obtained samples were separated on 10% polyacrylamide gels and transferred to a PVDF membrane. The presence of 6xHis-tagged proteins and Strep-tagged proteins was tested with 6xHIS TagAntibody HIS.H8 (Invitrogen Co.) and StrepMAB-Classic, HRP conjugate (IBA Co.), respectively.
Split-luciferase complementation assay
N. benthamiana leaves were infiltrated with A. tumefaciens GV3101 carrying pCambia plasmids with the genes in fusion with the N-terminal or C-terminal part of luciferase . Three days after agroinfiltration, discs were removed from the leaves using a cork borer size 2 and placed in 100 µl water in a well of a 96-well plate. After a 1-h incubation, 100 μl of a solution of 2 mM luciferin in water was added to each well and the relative light units were determined with a luminometer (Biotek Co.) over a period of 45 min. The expression level of recombinant proteins was determined by Western blotting.
ROS assays were performed as described in [39,40]. Leaf discs were removed from N. benthamiana leaves and incubated overnight in water, in white, ﬂat-bottom, 96-well plates (Greiner Bio-One Co.). After 16 h, the water was replaced with a water solution containing: 34 µg/ml luminol (Sigma–Aldrich Co.) and 20 µg/ml horseradish peroxidase (type VI-A, Sigma–Aldrich) and the designated concentration of flg22 and csp22. The relative light units were determined with a luminometer (Biotek Co.) over a period of 45 min.
Plant growth conditions
N. benthamiana accession Nb-1  was used in all experiments. Nb-1 plants were grown for 4–6 weeks in a controlled environment chamber with 16 h light, 65% relative humidity and a temperature of 24°C (light period) and 22°C (dark period).
The normality of the data was assessed using normal quantile plots (q–q plots) and the Shapiro–Wilk test. The equality of the variance was as assessed using the Bartlett test and variance of the samples is indicated by the standard deviation represented by error bars. Welch's or Tukey–Kramer tests were used to test the differences between the means, P-values are indicated in the figure legends. Statistical analysis was performed with JMP Pro 14 software.
Pti1b interacts with PP2C phosphatase Pic1 in vitro and in plant cells
To identify plant proteins that interact with Pti1b, constructs encoding the kinase and an inactive variant Pti1b(K96N), both with a C-terminal 2xStrep-FLAG epitope tag, were individually introduced into leaves of N. benthamiana using Agrobacterium-mediated transient transformation (agroinfiltration) and Pti1b proteins were purified with FLAG-immunoprecipitation followed by Strep pull-down (Supplementary Figure S1). Pti1b has S-acylation sites in its N terminus, which play a role in its localization to the cell periphery . As a control, therefore, we applied the same purification method using another membrane-associated protein from Arabidopsis, PIP2A (AT3G53420) . The samples obtained were submitted for mass spectrometry analysis to identify potential Pti1b-interacting proteins. A protein was considered a candidate interactor if a derived peptide was detected in at least two samples with Pti1b or Pti1b(K96N) and was absent from all PIP2a samples (Figure 1A).
Pti1b interacts with protein phosphatase Pic1 from tomato in vitro and in plant cells.
One of the N. benthamiana proteins that met these criteria was a PP2C with close similarity to tomato phosphatase Solyc07g066260. Phylogenetic analysis indicated that the closest proteins in Arabidopsis to Solyc07g066260 are Arabidopsis PP2C phosphatases At1g16220 and At1g79630, both of which are currently unnamed [43,44]. Based on findings described below, we refer to the PP2C in N. benthamiana and tomato (Solanum lycopersicum) as NbPic1 and SlPic1, respectively, for pattern-triggered immunity inhibiting PP2C1 (Supplementary Figure S2). Supporting a role for this phosphatase in PTI, the expression of the SlPic1 gene is induced in tomato in response to treatment with MAMPs flgII-28 and csp22 (Supplementary Table S3).
To confirm the interaction between Pti1b and tomato Pic1, we first used a Strep pull-down assay in combination with wild type Pti1b, its kinase-inactive form and two variants with substitutions at threonine-233, the major autophosphorylation site in the Pti1a kinase . Each Pti1b protein was purified as a 6xHIS-tagged fusion and incubated in kinase buffer with Pic1-Strep. After incubation, Pic1-Strep was purified with Strep-affinity chromatography and the possible presence of Pti1b-6xHIS was tested with an anti-6xHIS antibody. As shown in Figure 1B, each of the Pti1b variants was detected only in samples containing Pic1.
As further confirmation, we examined the interaction of Pti1b and Pic1 in plant cells using a split-luciferase complementation assay. Pti1b-nLuc and Pic1-cLuc or Pti1b-nLuc and cLuc-AtPIP2a (as a negative control) were transiently expressed in N. benthamiana leaves using agroinfiltration and the leaf discs were treated with 1 mM luciferin. We detected statistically significantly more luminescence from the leaf discs expressing Ptib-nLuc and Pic1-cLuc than from leaf discs expressing the negative control (Figure 1C). Western blotting confirmed the expression of each of the nLuc and cLuc proteins (Figure 1D). From these experiments, we conclude that Pti1b interacts with protein phosphatase Pic1 and, at least in vitro, this interaction occurs independently of Pti1b autophosphorylation.
Pti1b autophosphorylates in vitro on threonine-233
In previous work, we showed that the Pti1a kinase, a homolog of Pti1b, has protein kinase activity and threonine-233 in its activation loop is its major autophosphorylation site . The present study focused on Pti1b because a subsequent RNA-Seq experiment found that expression of the Pti1b gene is induced by MAMPs, whereas Pti1a is not (Supplementary Table S1). We therefore tested Pti1b and its presumed kinase-inactive variant for kinase activity in an in vitro assay. Wild-type Pti1b strongly autophosphorylated, whereas no phosphorylation was detected with the Pti1b(K69N) variant (Figure 2A). Next, using LC–MS/MS, we investigated whether threonine-233 is also the major phosphorylation site in Pti1b as it is in Pti1a. However, we were unable to detect the 230LHSTR234 peptide generated after trypsin or pepsin digestion in either phosphorylated or unphosphorylated form.
Threonine-233 is the major autophosphorylation site in Pti1b in vitro.
To overcome this obstacle, we generated a Pti1b variant where the arginine-234 at the trypsin digestion site was substituted to cysteine. This variant was based on a recent report that the MARIS kinase from Arabidopsis, which belongs to the same sub-family VIII of RLCKs as Pti1b, gains constitutive function when it carries the analogous arginine-to-cysteine substitution (R240C; Figure 2B) . Such a modified Pti1b, when digested with trypsin, would generate a longer peptide containing threonine-233 (LHSTCVLGTFGYHAPEYAMTGQLSSK) that might be more easily detected by LC–MS/MS. Indeed, with LC–MS/MS analysis of the autophosphorylated Pti1(R234C), we observed the presence of both phosphorylated and unphosphorylated forms of the longer peptide. Both threonine-233 and, unexpectedly, serine-232 were found to be phosphorylated in this peptide (Figure 2C, Supplementary Figure S3; Supplemental Table S4).
To investigate this observation further, we created a series of Pti1b variants with substitutions at serine-232 and threonine-233: Pti1b(S232A), Pti1b(T233A), Pti1b(S232A,T233A) and a potential phosphomimic substitution, Pti1b(T233D). These Pti1b variants were expressed in E. coli, purified and incubated in kinase buffer. Pti1b wild type and Pti1b(K96N) were used as positive and negative control, respectively. The results obtained with autoradiography showed that substitution of threonine-233 to either alanine or aspartic acid greatly decreased the autophosphorylation of Pti1b (Figure 2D). In contrast, the S232A substitution had no impact on the autophosphorylation of Pti1b. These data indicate that threonine-233 is likely the major site of autophosphorylation in Pti1b and suggest that the R234C substitution may cause enhanced phosphorylation of S232.
Pic1 dephosphorylates Pti1b in vitro and in plant cells
The interaction of Pic1 with Pti1b suggests the phosphatase might act to dephosphorylate the kinase possibly as a mechanism of desensitizing its activated state. To test this possibility, we incubated Pti1b-6xHIS in kinase buffer, either alone or in the presence of Pic1-Strep or a variant of Pic1-Strep that has the aspartate residues at positions 110 and 234 substituted to asparagine [Pic1(NN)-Strep]. Such asparagine substitutions are known to abolish phosphatase activity of PP2C proteins . We observed that in the presence of Pic1-Strep, the phosphorylated form of Pti1b was greatly reduced as compared with Pti1b incubated alone or in the presence of the inactive variant Pic1(NN)-Strep (Figure 3A).
Pic1 reduces the amount of autophosphorylated Pti1b in comparison with the catalytically inactive variant Pic1(NN) in vitro and in plant cells.
We further examined whether Pic1 affects Pti1b phosphorylation in vivo. Using agroinfiltration, we co-expressed Pti1b-2xStrep-FLAG with either Pic1-3xHA or Pic1(NN)-3xHA in leaves of N. benthamiana. Six days after agroinfiltration, the plant tissue was treated with 1 μM flg22 for 10 min. Pti1b was then purified with Strep-affinity chromatography and the phosphorylation level of Pti1b was tested with antibodies that specifically recognize phosphorylated threonine residues. The specificity of the antibodies for phosphorylated Pti1b was tested with different in vitro phosphorylated Pti1b variants (Supplementary Figure S4). We observed that the phosphorylation level of Pti1b-2xStrep-FLAG co-expressed with Pic1-3xHA is reduced in comparison with Pti1b co-expressed with Pic1(NN)-3xHA (Figure 3B). Collectively, the data from Figures 2 and 3 indicate that Pic1 dephosphorylates threonine-233, the major in vitro autophosphorylation site in Pti1b.
Pti1b(R234C) is recalcitrant to Pic1 phosphatase activity in vitro
The arginine-to-cysteine substitution at position 240 in MARIS causes the protein to gain a constitutive-active function which restores pollen tube CWI in an anx1/anx2 Arabidopsis mutant . The underlying reason for this gain-of-function activity is unknown [23,45]. The corresponding arginine in Pti1b, at position 234, is adjacent to threonine-233 and we speculated that an R234C substitution might affect the kinase activity of Pti1b. To test this possibility, we compared the autophosphorylation activity of Pti1b and Pti1b(R234C) in an in vitro kinase assay and found no difference between them (Figure 4A). We next examined whether the R234 substitution might affect the activity of Pic1 towards Pti1b. The kinase activity of Pti1b and Pti1b(R234C) were assayed in the presence of Pic1, Pic1(NN), calf intestinal phosphatase (CIP) or shrimp alkaline phosphatase (rSAP). Remarkably, Pic1 dephosphorylated Pt1b but was incapable of dephosphorylating Pti1b(R234C) (Figure 4B). Both Pti1b proteins were effectively dephosphorylated by CIP and rSAP (Figure 4B); for unknown reasons, CIP was phosphorylated independently of Pti1b (Supplementary Figure S5). We next investigated whether the inability of Pic1 to dephosphorylate Pti1b(R234C) was due to the loss of an interaction between these two proteins. Using the Strep pull-down assay described above, we found that Pic1-Strep interacted just as well with Pti1b as it did with Pti1b(R234C) (Figure 4C). Thus, the Pti1b(R234C) variant has enzymatic activity comparable to Pti1b and still interacts with Pic1 although it resists Pic1 phosphatase activity.
Pti1b(R234C) is recalcitrant to Pic1 but is sensitive to other phosphatases in vitro.
Peptide flg22 induces the phosphorylation of Pti1b on threonine residue(s) in plant cells
Perception of MAMPs like peptide flg22 by PRRs induces phosphorylation of downstream signaling components [13,15]. To test possible MAMP induction of Pti1b phosphorylation, we chose to use Pti1b(R234)-Strep to avoid dephosphorylation by Pic1. Pti1b(R234)-Strep was expressed in N. benthamiana leaves using agroinfiltration and the leaf tissue was infiltrated 3 days later with 1 µM flg22 or with water as a control. Ten minutes after the addition of flg22, the tissue was collected and Pti1b(R234C) protein was purified by Strep-affinity chromatography. One half of the purified protein was then incubated with CIP to remove phosphate from Pti1b(R234C) and the samples were analyzed by Western blotting using the antibody that specifically detects phosphorylated threonine residues. This analysis revealed a higher level of phosphothreonine on Pti1b(R234C) from flg22-treated leaf tissue than from the control treated with water alone (Figure 5). In turn, Pti1b(R234C) protein purified from the water-treated leaf tissue was phosphorylated to a higher level than the CIP-treated samples indicating that Pti1b is phosphorylated at a basal level even without flg22 treatment.
Peptide flg22 induces Pti1b(R234C) phosphorylation on threonine residue(s) in plant cells.
Pic1 decreases the production of ROS generated in leaves in response to flg22 and csp22
Tomato plants with decreased levels of Pti1a and Pti1b produce less ROS after treatment with flg22 compared with wild-type plants . We hypothesized that the dephosphorylation of Pti1b by Pic1 might negatively regulate the role of Pti1 in generating ROS in response to flg22. To test this hypothesis, we used agroinfiltration to overexpress Pic1-3xHA in leaves of N. benthamiana, which has four Pti1 homologs  and measured the production of ROS in response to flg22 and another MAMP, csp22. As controls, we overexpressed the inactive Pic1(NN)-3xHA variant. Leaf discs expressing Pic1 produced statistically significantly lower levels of ROS after treatment with flg22 and csp22 compared with leaf discs expressing Pic1(NN)-3xHA (Figure 6A,C). Each of the proteins was shown to be expressed well by Western blotting (Figure 6B,D). Pic1 therefore acts as a negative regulator of ROS production in plant leaves, likely by dephosphorylating Pti1b.
Pic1 reduces the amount of ROS produced in N. benthamiana leaves in response to MAMPs.
We reported previously that the Pti1 kinases are activators of ROS burst in tomato and positive regulators of disease resistance against P. syringae pv. tomato . Here, we identified Pic1 as an interactor of Pti1 with the ability to dephosphorylate the major Pti1b autophosphorylation site, threonine-233. Phosphorylation of Pti1b is induced upon treatment of leaves with MAMPs and overexpression of Pic1 in leaves greatly reduces the production of immunity-associated ROS. Below, we elaborate on our key findings about Pti1b phosphorylation status, the role of Pic1 in regulating Pti1b, and on the ability of the arginine-234-cysteine substitution to make Pti1b recalcitrant to Pic1 activity.
Our mass spectrometry experiments suggested that Pti1b is autophosphorylated in vitro on threonine-233 and/or on serine-232. However, with autoradiography, we were only able to find evidence for Pti1b phosphorylation on threonine-233 and our current data therefore support this residue as the major phosphorylation site. Phosphorylation of serine-232 was also not observed in a previous study that reported threonine-233 as the major site of phosphorylation in the Pti1a kinase . Interestingly, we do observe weak phosphorylation associated with both Pti1b(T233A) and Pti1b(T233D). The source of this signal could be phosphorylated serine-232 which may be a minor site of autophosphorylation in Pti1b. It is possible that the arginine-234-cysteine substitution causes more phosphorylation to occur on serine-232 than occurs in wild-type Pti1b. This possibility is supported by the greater number of peptides observed by mass spectrometry with phosphorylated S232 (Figure 2C) and by the experiment in which we compared the autophosphorylation of Pti1b(T233A) and Pti1b(T233A/R234C) and found the latter protein to be more strongly phosphorylated (Supplementary Figure S4A). It also appears that the serine-232-alanine substitution might strengthen the autophosphorylation on threonine-233 as shown in Figure 2B.
We found that flg22 induces the phosphorylation of Pti1b on threonine residue(s) in N. benthamiana. Although we have not investigated this further yet, it is likely this phosphorylation occurs on threonine-233 because that residue is the major phosphorylation site in vitro. It has been reported previously that OsPti1a is phosphorylated in rice cells on its threonine-233 (corresponding to threonine-233 in Pti1b) . Additionally, it was shown that threonine-233 of OsPti1a plays a role in immunity against X. o. pv. oryzae . Further insights into which residues of Pti1b are phosphorylated in plant cells will come from the future development and biochemical characterization of transgenic tomato plants with phosphonull and phosphomimic variants of threonine-233.
The role of phosphatase Pic1
There are relatively few protein phosphatases which are known to regulate RLCKs that are activators of plant immune responses. One of these, PP2C38 , which regulates the immunity-associated RLCK BIK1, has some similarities to Pic1. Like Pic1 and Pti1b, PP2C38 negatively controls the phosphorylation status of BIK1 . Treatment of Arabidopsis leaves with flg22 induces BIK1 activity causing it to phosphorylate PP2C38, which leads to the dissociation of the phosphatase from BIK1 . The dissociated BIK1 is then able to activate downstream targets. We did not observe Pti1b phosphorylation of either Pic1 or Pic1(NN) in vitro. Therefore, we have no evidence that the release of Pti1b from interaction with Pic1 relieves negative regulation by Pic1 indicating Pti1b-Pic1 may use a different mechanism than BIK1-PP2C38. A hypothesis that would be interesting to explore is that Pti1b might be phosphorylated by an upstream kinase such as Oxi1 which allows Pti1b to mitigate the negative regulation imposed by Pic1.
The influence of the arginine-234-cysteine substitution on Pic1 activity
The MARIS protein, which is a member of the same RLCK sub-family as Pti1b, acts as a regulator of CWI in Arabidopsis . The variant of this kinase with an arginine-to-cysteine substitution was identified as a constitutive-active version of MARIS, but the underlying reason for this activity is unknown . Here, we found that the analogous substitution in Pti1b (R234C) makes the kinase insensitive to dephosphorylation by Pic1 although its kinase activity is comparable to Pti1b wild type. Based on this observation, it is possible that MARIS(R240C) is also able to resist dephosphorylation by an associated phosphatase that normally negatively regulates MARIS. This could explain how the constitutively active variant of MARIS is able to restore pollen tube CWI in the absence of Anx1/Anx2.
A model for Pti1b and Pic1
Based on the results presented here, we propose a model for Pti1b and Pic1 in PTI (Supplementary Figure S6). Various MAMPs, including flg22 and csp22, activate PRRs by binding to their extracellular leucine-rich repeat domains. These PRRs act in concert with the co-receptor BAK1 (SERK3B) in signaling pathway(s) which lead to multiple defense responses including production of ROS. Pti1b is proposed to act downstream or with PRR signaling complexes and plays a role in ROS production. Phosphorylation of threonine-233 in Pti1b is proposed to contribute to this mechanism and Pic1 acts to dephosphorylate this residue, thereby inactivating Pti1b and desensitizing the associated signaling pathway. In the future, we will investigate the possibility that Pti1b is present in PRR complexes, how it might be affected by activated PRRs, and search for its potential substrates which are expected to include proteins involved in ROS production.
anxur receptor-like kinase
brassinosteroid insensitive 1-associated kinase 1
chitin elicitor receptor kinase 1
cold-shock protein receptor
cell wall integrity
epidermal growth factor
flagellin sensing 2
kinase-associated protein phosphatase
microbe-associated molecular patterns
mitogen-activated protein kinases
nicotinamide adenine dinucleotide phosphate
protein phosphatase 2C
pattern recognition receptors
Pto interactor 1b
respiratory burst oxidase homolog protein D
receptor-like cytoplasmic kinases
reactive oxygen species
XA21 binding protein 15
F.G. and G.B.M. conceived and designed the experiments, F.G. performed experiments, F.G. and G.B.M. analyzed the data and wrote the paper
This research was supported by National Science Foundation [IOS-1451754 and IOS-1546625] and the USDA-Binational Agriculture Research Fund (BARD) [IS-4931-16C].
We thank Fabio Rinaldi and Robyn Roberts helpful comments on the manuscript, Konrad Thorner, Sam Wolfe and Lydia Zamidar for technical assistance, and Gitta Coaker for the cLuc-PIP2A plasmid.
The Authors declare that there are no competing interests associated with the manuscript.