The mRNA decay factor PAT1 functions in a pathway including MAP kinase 4 and immune receptor SUMM2

AtPAT1 is an ScPAT1 orthologue and interacts with MPK4 in planta

We previously conducted a yeast two-hybrid screen to identify Arabidopsis proteins that interact with MPK4 (Andreasson et al, 2005). In addition to the MPK4 substrate MKS1, we identified two clones encoding PAT1 (At1g79090), an mRNA decapping stimulator involved in post-transcriptional gene regulation (Coller & Parker, 2005). Two PAT1 homologs are encoded in the Arabidopsis genome (AtPAT1H1 At3g22270; AtPAT1H2 At4g14990, Supplementary Fig S1A). The steady-state expression level of PAT1 and the homologs compared to the housekeeping gene ACTIN8 (At1g49240) was analyzed by RT-PCR (Supplementary Fig S1B). We focused on PAT1 as the other two homologs were not identified by yeast two-hybrids. To confirm the PAT1-MPK4 interaction in planta, doubly transgenic Arabidopsis lines were generated in the Ler mpk4-1 background that expressed PAT1 with a C-terminal Myc tag and HA-tagged MPK4 under the control of their own promoters. Anti-Myc immunoprecipitation from either MPK4-HA or double transgenic MPK4-HA/Pat1-Myc tissue detected a 50 kDa band corresponding to MPK4-HA only in double transgenic lines (Fig1A). Thus, MPK4 and PAT1 can be found in complex in Arabidopsis.

Yeast PAT1 engages with translating mRNPs and is involved in translational repression and decapping activation (Marnef & Standart, 2010). Since the function of PAT1 in Arabidopsis was unknown, we examined whether it functions similarly to yeast PAT1. To this end, a full-length Arabidopsis PAT1 cDNA was cloned from Col-0 (Supplementary Fig S1C and D) and transformed into wild-type yeast (B4742) and a yeast mutant (Y15797) in which yeast PAT1 was replaced with a G418 resistance cassette (BY4742 (YCR077c) pat1Δ::KanMX). In contrast to the wild-type, yeast lacking PAT1 (pat1Δ) display a temperature-sensitive phenotype and are impaired at 37°C but grow normally at 30°C (Tharun et al, 2005). This phenotype is reverted to wild-type in yeast containing Arabidopsis PAT1, as growth at 37°C was restored (Fig1B). As an additional control, we transformed Arabidopsis PAT1 into wild-type yeast (B4742/AtPAT1), and this grew similarly to wild-type at 30°C and almost as well at the wild-type at 37°C (Fig1B). The expression of Arabidopsis PAT1 in yeast was confirmed by anti-PAT1 immunoblotting of yeast protein extracts (Supplementary Fig S1E). This provides compelling evidence for the orthologous functions of these yeast and Arabidopsis PAT1 proteins. As PAT1 is found in complex with MPK4, these results provide a link between MPK4 and post-transcriptional regulation of mRNA stability.

We next analyzed the interaction between PAT1 and conserved components of mRNA decapping. PAT1-LSM1-7 complexes function in mRNA decapping and deadenylation (Bouveret et al, 2000; Tharun, 2009; Haas et al, 2010; Ozgur et al, 2010; Totaro et al, 2011). We therefore transiently expressed in Nicotiana benthamiana LSM1-GFP and PAT1-HA and then immunoprecipitated LSM1 with GFP Trap beads. PAT1-HA could be detected in LSM1 immunoprecipitates but did not adhere to GFP Trap beads in the absence of LSM1-GFP (Fig1C). This is consistent with the detection of peptides corresponding to PAT1 and its homologues in LSM1 immunoprecipitates (Golisz et al, 2013) and supports a role of PAT1 in mRNA decapping. In other organisms, interactions between PAT1 and LSM1 are robust, while those between PAT1 and other mRNA decapping proteins, including the DCP1-DCP2 complex and XRN1, are more transient (Bouveret et al, 2000; Nissan et al, 2010; Ozgur et al, 2010). This is consistent with our difficulty in detecting DCP1 in complex with PAT1 in Arabidopsis (Supplementary Fig S1F).

PAT1 is required for decapping of selected mRNAs

In order to determine whether PAT1 behaves as an activator of mRNA decapping, we used 5′ RACE to compare the levels of capped mRNAs in Col-0 and pat1 mutants. To this end, we identified an allele, pat1-1 (Salk_040660), with a T-DNA insertion in the last exon of PAT1 (Supplementary Fig S1C). We also generated an anti-PAT1 antibody against a C-terminal peptide (Supplementary Fig S1D). Immunoblotting of Col-0 protein extracts with this antibody detected a clear band around 90 kDa. In contrast, no protein could be detected in pat1-1 mutant extracts (Supplementary Fig S1G). This indicates that pat1-1 harbors either a truncated version of PAT1, no PAT1 protein, or levels of the protein that are below detection.

5′ RACE was performed on transcripts known to be degraded by the decapping complex (EXPL1; UGT87A2) (Perea-Resa et al, 2012), as well as a housekeeping transcript EIF4A1. We found that capped EXPL1 and UGT87A2 accumulated in pat1-1 mutants, while capped EIF4A1 mRNA was present in equal amounts in Col-0 and pat1-1 (Fig1D). This indicates that PAT1 plays a role in mRNA decay via decapping.

PAT1 is an MPK4 substrate

Since MPK4 and PAT1 are found in complexes in planta, we asked whether PAT1 is an MPK substrate. PAT1 contains 5 Ser-Pro (SP) motifs which are commonly phosphorylated by MPKs (Pearson et al, 2001; Ubersax & Ferrell, 2007) (Supplementary Fig S1D). To characterize PAT1 phosphorylation in vivo, we identified PAT1 phosphopeptides by mass spectrometry. Since MPK3/4/6 are activated by flagellin and by virulent strains of Pseudomonas syringae pv. tomato (Pto) DC3000 (Asai et al, 2002; Brader et al, 2007; Suarez-Rodriguez et al, 2007; Bethke et al, 2009, 2012; Rasmussen et al, 2012), PAT1 was immunoprecipitated from extracts of untreated control and flg22-treated wild-type Col-0 or PAT1-GFP transgenic lines. Bands corresponding to PAT1-GFP (130 kDa) were excised from the gel (Supplementary Fig S2), subjected to in-gel tryptic digestion, and peptides were extracted. Phosphopeptides were enriched by TiO₂ chromatography and analyzed by liquid chromatography (RP-HPLC) coupled to electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS). This identified several phosphopeptides from PAT1-GFP IPs that were not detectable in the negative control (Col-0) (Table1; Supplementary Fig S2). The most abundant phosphopeptide, based on the extracted ion chromatogram from the LC-MS/MS analysis, revealed phosphorylation of Ser208 in an SP motif (Fig2A). Another peptide was identified with phosphorylation of Ser343. However, this site is not within an SP motif, making it a less likely site for phosphorylation by MPKs. Importantly, both these peptides have previously been detected in Arabidopsis by mass spectrometry (Phosphat Database, http://phosphat.mpimp-golm.mpg.de/). It should be noted that the PAT1 phosphopeptides were detected both whether or not the sample had been treated with flg22. Thus, under the conditions used here, PAT1 was phosphorylated and remained so after exposure to flg22.

To determine whether PAT1 is a substrate of a specific MPK, we carried out in vitro kinase assays using immunoprecipitated, PAMP-activated MPKs with purified His₆-PAT1 protein. Phosphorylated His₆-PAT1 was detectable as a radioactive band around 95 kDa after incubation with flg22-activated MPK4 (Fig2B), while MPK6 caused only low levels of PAT1 phosphorylation and MPK3 did not significantly alter PAT1 phosphorylation (Supplementary Fig S3). Each MPK was also incubated with the generic MPK substrate myelin basic protein (MBP) to verify their activation (Fig2B; Supplementary Fig S3). This confirms that activated MPK4, and to a lesser extent, MPK6, is responsible for the phosphorylation of PAT1. A version of His₆-PAT1 with Ser208 mutated to alanine (S208A) was also incubated with MPK4 and MPK6 IPs, and this mutant had significantly lower levels of phosphorylation (Fig2B). This supports the identification of S208 as a key phosphorylation site in PAT1.

pat1 mutants exhibit autoimmunity similar to mpk4

Given that MPK4 is an important immune regulator in Arabidopsis and that mpk4 mutants have de-repressed defense responses, we examined whether PAT1 may also be involved in immunity. The pat1-1 mutant has a distinct leaf serration phenotype and a slightly smaller rosette than Col-0 (Fig3). A similar rosette phenotype was seen in another T-DNA insertion line (pat1-2, WiscDsLox_734_D04 in Supplementary Figs S1C and S4), indicating that this phenotype is due to loss-of-function of PAT1. Importantly, however, the phenotype of pat1-1 is not as extreme as mpk4-2, which is much smaller and has more pronounced leaf curling and reduced fertility as well as constitutive defense gene expression (Petersen et al, 2000); Fig3).

In order to determine whether pat1-1 is a constitutive defense mutant similar to mpk4, quantitative reverse-transcription PCR (qRT-PCR) was used to measure the steady-state level of the pathogenesis-related PR1 and PR2 genes in adult plants (Fig4A). Previous work showed that the enhanced defense response of mpk4 mutants is suppressed by mutations in EDS1 (Brodersen et al, 2006). Thus, eds1-2 was crossed with pat1-1 to explore the pat1 phenotype in the absence of this regulator (Fig3). Compared to Col-0, pat1-1 mutants accumulated 1,000-fold more PR1 and 150-fold more PR2 transcripts (Fig4A). In contrast, the levels of PR mRNAs in pat1-1 eds1-2 double mutants were similar to those in wild-type (Fig4A). Under the same conditions, mpk4-2 mutants accumulated 4,000-fold more PR1 and 800-fold more PR2 transcripts (Fig4A).

We found that the elevated PR gene expression in pat1-1 correlated to an enhanced resistance to infection by syringe-infiltrated Pto DC3000 when compared to Col-0 (Fig4B). While Pto DC3000 growth reached 5.5 cfu/cm² in Col-0, pat1-1 mutants supported tenfold lower accumulation of Pto DC3000. In this experiment, eds1-2 mutants showed enhanced bacterial growth as expected for this mutant with compromised defense responses (Fig4B). pat1 disease resistance is EDS1 dependent, as bacterial growth pat1-1 eds1-2 double mutants was similar to that in eds1-2 (Fig4B). These findings indicate that, similar to mpk4, pat1 mutants express EDS1-dependent autoimmunity in the absence of microbes.

PAT1 detection in P-bodies is induced by PAMPs

Processing bodies (PBs) are cytoplasmic granules involved in both mRNA decay and translational repression pathways (Kulkarni et al, 2010). PAT1 is a conserved PB component in yeast (Rodriguez-Cousino et al, 1995), C. elegans (Boag et al, 2008), Drosophila (Marnef et al, 2010) and mammals (Scheller et al, 2007) and PBs can be significantly induced in Arabidopsis by hypoxia and heat stress (Weber et al, 2008). To investigate whether PAT1 is found in PBs, we produced transgenic Arabidopsis lines that express PAT1-GFP from its native promoter complementing the pat1 phenotype (Supplementary Fig S5A and B). We detected GFP signal in small numbers of distinct foci by confocal microscopy in roots of young seedlings (Fig5A). Within 20 min of flg22 treatment, a significant increase in the number of GFP-positive foci could be seen in the root tips (Fig5A). To test whether these foci correspond to PBs, we treated the roots with cycloheximide in DMSO which is known to abrogate PB formation in plants (Goeres et al, 2007). This revealed that cycloheximide inhibited flg22-induced foci (Fig5A). Importantly, the control DMSO treatment did not reduce the number of foci (Supplementary Fig S6A and B). As a control, we also tracked the localization of the known decapping component VCS (Xu et al, 2006). Similar to PAT1-GFP, VCS-GFP was seen in flg22-induced PBs and absent when treated with cycloheximide (Fig5B). Interestingly, PAT1 protein, which is hardly detectable under steady-state conditions, also accumulated in response to flg22 treatment, with a peak by 60 min, and a return to normal levels after 2 h (Fig5C). PAT1-GFP mirrors this effect and was similarly up-regulated in response to flg22 treatment, as detected by anti-GFP Western blotting of seedling protein (Fig5D). Importantly, PAT1 transcript levels were not highly induced by flg22 treatment in Col-0 seedlings (Supplementary Fig S6C), suggesting that PAT1 induction occurs post-transcriptionally. These data indicate that activation of PTI leads to up-regulation of PAT1 protein levels by an unknown post-transcriptional mechanism, and this facilitates the detection of PAT1 in PBs. In addition, PTI could induce PB formation as part of cellular reprogramming, and PAT1 may localize to PBs to engage in this process.

The MPK4 suppressor summ2 also suppresses the pat1 resistance phenotype

Autoimmunity caused by loss of different components of the MPK4 kinase cascade can be suppressed by mutations in the resistance protein SUMM2 (Zhang et al, 2012). An explanation for this is that SUMM2 keeps this PAMP responsive pathway under surveillance and that mutations in its components mimic the effects of microbial effectors that prevent phosphorylation within or below the cascade (Zhang et al, 2007). To further probe the connection between MPK4 and PAT1, we generated pat1-1 summ2-8 double mutants. These mutants retained the leaf serration phenotype of pat1-1 single mutants (Supplementary Fig S7A). However, double mutants no longer accumulated excessive PR1 and PR2 transcripts (Fig6A) and displayed summ2 levels of susceptibility to syringe-infiltrated Pto DC3000 (Fig6B). Importantly, the pat1 growth phenotype was not caused by overexpression of SUMM2 transcripts, as the level of SUMM2 in pat1-1 was similar to wild-type (Supplementary Fig S7B). To test whether the accumulation of 5′ capped transcripts in pat1-1 (Fig1D) was merely an effect of inappropriate activation of SUMM2, we tested the accumulation of 5′ capped UGT87A2 in an XRN1 sensitivity assay. XRN1 degrades all uncapped RNA leaving 5′ capped RNA intact (Blewett & Goldstrohm, 2012). The accumulation of 5′ capped versus uncapped UGT87A2 transcripts was at similar levels in pat1-1 and pat1-1 summ2-8 (Fig6C). Col-0 and summ2-8 plants had similarly reduced 5′ capped versus uncapped ratios but, importantly, these were much lower than the levels of pat1-1 and pat1-1 summ2-8 (Fig6C). Therefore, the accumulation of capped transcripts in pat1-1 mutants is not an artifact of defense activation but reflects a role for PAT1 in mRNA decay. In order to determine whether the SUMM2-mediated resistance in pat1 mutants is specific to pathogens of a specific class, we compared susceptibility to pathogens with different lifestyles. Pto DC300 is a hemi-biotrophic pathogen, which grows in living tissue during early infection. In contrast, the necrotrophic fungal pathogen Botrytis cinerea relies on dying tissue for its propagation in plants (Glazebrook, 2005). Infection of pat1-1 summ2 double mutants with B. cinerea resulted in similar growth of the fungus as detected in summ2-8 single mutants (Supplementary Fig S8).

To more accurately address to what extent PAT1 phosphorylation is MPK4 dependent and to measure PAT1 phosphorylation profiles before and after flg22 treatment, we generated PAT1-HA transgenic lines in the mkk1/2 summ2-8 background. The suppression of mpk4 by summ2-8 is only partial, but the autoimmunity phenotype is fully suppressed in mkk1/2 summ2 triple mutants. We next applied quantitative mass spectrometry (iTRAQ) to PAT1-HA IPs from Col-0 and mkk1/2-summ2 and this revealed increased phosphorylation stoichiometry of PAT1 at Ser208 in untreated Col-0 compared to mkk1/2 summ2 plants (Fig6D). Flg22 treatment leads to increased levels of PAT1 Ser208 in both genotypes, but the levels in Col-0 were higher than what we found for PAT1 in mkk1/2 summ2 (Fig6D). Since flg22 treatment augments PAT1 Ser208 phosphorylation in the absence of MKK1/2, MPK4 might be activated by other upstream kinases or PAT1 may be phosphorylated by MPK6. Nevertheless, these data indicate that flg22 treatment leads to increased phosphorylation of PAT1 at Ser208 and the MKK1/2 MPK4 pathway contributes to this phosphorylation.

Our data indicate that PAT1 is part of the pathway including MPK4 and the SUMM2 R protein. SUMM2 and PAT1 might thus interact in planta. To test this, we transiently expressed and immunoprecipitated PAT1-GFP in N. benthamiana also transiently expressing MPK4-HA or SUMM2-HA. When PAT1-GFP was pulled down using GFP Trap beads, we could easily detect MPK4-HA and SUMM2-HA in the immunoprecipitates (Fig7A). To verify the association of MPK4, SUMM2 and LSM1, we also pulled down PAT1-HA and detected MPK4-GFP, SUMM2-GFP and LSM1-GFP in the immunoprecipitates but not Myc-GFP (Supplementary Fig S9). To further confirm the association of PAT1 and SUMM2, we used bimolecular fluorescence complementation (BiFC), which produces a fluorescent readout upon reconstruction of YFP. The BiFC assay showed detectable YFP signal in the cytoplasm when PAT1 and SUMM2 were co-expressed (Fig7B). No signals were observed when PAT1 was expressed with another R protein, At4g12010, implying that the association/reconstruction is specific (Fig7B, bottom panel). This indicates that PAT1 is in complex with SUMM2 in planta. Thus, MPK4 and PAT1 are both physically and genetically linked to SUMM2.

Plant materials and growth conditions

Arabidopsis thaliana ecotype Columbia (Col-0) was used as a control. Seeds for T-DNA insertion lines were from NASC (Nottingham, UK). The T-DNA lines for At1g79090 (PAT1), both of which have insertions in the last exon, were Salk_040660 (here named pat1-1) and WiscDsLox437D04 (pat1-2). The T-DNA insertion in At1g12280 (SUMM2) summ2-8 (SAIL_1152A06), mkk1/2 (Zhang et al, 2012) and eds1-2 (Parker et al, 1996) have been described. Genotyping primers for newly described T-DNA lines are provided in Supplementary Table S1. Arabidopsis plants were grown in 9 × 9 cm pots at 22°C with a 8-h photoperiod, or on plates containing Murashige–Skoog (MS) salts medium (Duchefa), 1% sucrose and 1% agar with a 16-h photoperiod.

Cloning and transgenic lines

The genomic PAT1 (At1g79090) DNA sequence (without stop codon), plus 2 kb upstream from the start codon, was amplified from Col-0 genomic DNA and cloned into pENTR-D-TOPO (Invitrogen). The entry clone was recombined into pGWB513, pGWB517 and pGWB504 (Nakagawa et al, 2007) to obtain a C-terminal HA, Myc and GFP tags, respectively. The expression clones were transformed into Agrobacterium tumefaciens strain GV3101 and transformed into Col-0 plants by floral dipping. Transformants were selected on hygromycin (30 μg/ml) MS agar and survivors tested for protein expression by Western blotting.

The VCS promoter region was amplified and inserted into pGEM-T easy and next into the HindIII SalI in pBN:GFP to generate pBNpVCS:GFP. The VCS coding region was amplified and cloned into pGEM-T easy next inserted into SalI and KpnI in pBNpVCS:GFP and transformed into Agrobacterium LBA4404 and transformed into Col-0 plants by floral dipping. The cDNA of LSM1 (At3g14080) without stop codon was cloned into pENTR-D-TOPO (Invitrogen) and recombined into 35S promoter-containing pGWB505 (Nakagawa et al, 2007) to obtain C-terminal GFP-tagged LSM1. Genomic DNA of MPK4 and cDNA of SUMM2 were cloned into pENTR-D-TOPO and recombined into pGWB514 (Nakagawa et al, 2007) to obtain C-terminal HA-tagged constructs. The genomic DNA and promoter of DCP1 was cloned into pENTR-D-TOPO and recombined into pGWB513 to obtain C-terminal HA-tagged constructs under the control of the native promoter. Clones were transformed into A. tumefaciens strain GV3101 and used for transient expression in N. benthamiana or floral dipping in Arabidopsis. Genomic PAT1-GFP was transformed into the Col-0 background. DCP1-HA was dipped into the Col-0 background and T3 progeny were crossed with Col-0/PAT1-GFP T4 lines to obtain double transgenic lines. Genomic PAT1-Myc was transformed into Ler mpk4-1/MPK4-HA transgenic lines (described in Petersen et al, 2000) to obtain double transgenic lines.

The entry clones of PAT1 and SUMM2 were further recombined into pcYGW and pnYGW (Hino et al, 2011), respectively, obtaining the N- and C-terminal split YFP tags used for BiFC. At4g12010 clones were obtained by the same procedure. Clones were transformed into A. tumefaciens strain GV3101 and used for transient expression in N. benthamiana.

Mutagenesis of His₆-PAT1

Site-directed mutagenesis of Ser208 to Ala was accomplished using pET15b-PAT1 as a template for PCR mutagenesis. The primers used were PAT1 S208A F&R (see Supplementary Table S1).

PAT1 protein purification and in vitro kinase assays

For in vitro experiments, PAT1 protein was purified from E. coli. The PAT1 cDNA was cloned into pET15b (for an N-terminal His fusion) and transformed into E. coli BL21 (pLysS). Protein expression was induced by overnight treatment with 0.5 mM IPTG at 18°C, added to cells at OD₆₀₀ = 0.6. PAT1 protein was insoluble and thus purified from inclusion bodies using Bugbuster (Novagen). Proteins were solubilized in 6 M urea, 0.7% N-lauroylsarcosine, 100 mM Tris-HCl pH 8 and refolded overnight at 4°C in 0.88 M L-arginine, 55 mM Tris-HCl, 2 mM NaCl, 0.88 mM KCl, protease inhibitors. Protein was then dialyzed against 20 mM Tris-HCl pH 8, 100 mM NaCl using 3,500 MWCO dialysis tubing. Purified protein was concentrated using Centriprep 30K spin columns (Millipore) and then diluted with glycerol to a final concentration of 0.1 mg/ml.

For kinase assays, Col-0 plants were immersed in 200 nM flg22 for 10 min of to activate MAP kinases. MPK3, 4 and 6 were immunoprecipitated from 3 mg total protein extracted from flg22-treated tissue using 2 μg of each of their specific antibodies (Sigma) and 30 μl EZview protein A agarose beads (Sigma). Four microgram of purified myelin basic protein (MBP, Sigma) or 20 μg His₆-PAT1 protein was incubated with washed MPK immunoprecipitates for 60 min at 37°C with 3 μCi γ-ATP in kinase buffer (62.5 μM ATP, 100 mM Tris pH 7.5, 150 mM NaCl, 150 mM MgCl₂, 10 mM EGTA, 5 mM DTT, Phosstop inhibitor (Roche)). Kinase reactions were diluted with 4× SDS buffer and boiled for 5 min before loading on 12% SDS-PAGE gels. Following electrophoresis, gels were stained with Coomassie Brilliant Blue and incubated with gel drying buffer, followed by drying on a Bio-Rad gel dryer. Dried gels were exposed to a phosphor screen overnight.

Yeast transformation

Yeast strains pat1Δ (BY4742 (YCR077c) pat1Δ::KanMX) and B4742 were obtained from Euroscarf. PAT1 was cloned from A. thaliana Col-0 cDNA into pENTR-D-TOPO and recombined into yeast expression vector pVV215 (C-terminal HA tag, -URA selection; (Van Mullem et al, 2003) by Gateway recombination. Yeast pat1Δ and B4742 cells were transformed using lithium acetate/polyethylene glycol according to the Clontech Yeast Protocols handbook. Transformed yeast was selected on SD-URA agar plates and re-streaked after 3–4 days onto fresh selection plates. Pat1Δ and wild-type-transformed yeast was grown in liquid culture overnight, and protein was extracted from pelleted cells by vortexing with glass beads in 1× SDS extraction buffer. Boiled proteins were subjected to SDS-PAGE and anti-PAT1 immunoblotting. For temperature sensitivity assays, overnight cultures of wild-type, pat1Δ and PAT1-expressing transformants were plated on YPAD and grown at 30 and 37°C for 3 days.

Flg22 kinetics

Seedlings were grown on MS agar (Col-0) or MS agar containing 30 μg/ml hygromycin (Col-0/PAT1-GFP) for 5 days before being transferred to MS liquid medium in 24-well plates. After 10 days, seedlings were treated by the addition of flg22 to a final concentration of 100 nM (2 seedlings per well × 3 wells per treatment time point). Seedlings were harvested at the indicated times and immediately frozen in liquid nitrogen for later RNA or protein extraction.

Semi-quantitative and qRT-PCR

Total RNA was extracted from seedlings with Tri Reagent (Sigma). RNA samples were treated with DNase Turbo DNA-free (Ambion), quantified with a NanoDrop spectrophotometer (Thermo Scientific) and reverse-transcribed into cDNA with SuperScript III reverse transcriptase (Invitrogen). For semi-quantitative reverse-transcription PCR (RT-PCR), Col-0 seedling cDNA was used as a template for PCR with primers specific for ACTIN8, PAT1, PAT1H1 or PAT1H2 using Sigma Jumpstart REDTaq Readymix. PCR products (20 μl) were separated on 2% (w/v) agarose gels and visualized with ethidium bromide. Brilliant II SybrGreen master mix (Agilent) was used for qPCRs. The UBQ10 (At4g05320) gene was used for normalization. Gene expression of PR1, PR2, PAT1 and SUMM2 was measured by qPCR analysis, normalized to UBQ10 expression and plotted relative to Col-0 expression level. These experiments were repeated in three independent biological replicates, each with three technical replicates, and representative data are shown. Standard error is represented by error bars on figures, and statistical significance is indicated by letters above error bars. These are derived from one-way ANOVA with Tukey's multiple comparison test (GraphPad Prism).

RNA extraction and RACE PCR

RNA extraction used TRI reagent and 10 μg RNA was used for 5′ RACE according to instructions (First Choice RACE, Ambion). PCR was carried out on 1 μl of products from reverse transcription of capped RNA using DreamTaq polymerase (Fermentas) with 25 cycles. RACE PCR products (10 μl) were separated on 2% (w/v) agarose gels and visualized with ethidium bromide.

Quantification of capped versus uncapped transcripts

Total RNA was extracted from seedlings with NucleoSpin RNA columns (Machery-Nagel). To remove RNA with no 5′ cap structure, 1 μg of total RNA was incubated with 1 unit XRN1 (New England Biolabs) or no enzyme at 37°C for 1 h. Next RNA was reverse-transcribed into cDNA with SuperScript III reverse transcriptase (Invitrogen). UGT87A2 transcript accumulation was measured by qPCR using SybrGreen master mix (Agilent) and normalized to ACT2. Calculating 5′ capped versus uncapped transcripts was done by comparing transcript levels from XRN1 and mock-treated samples for the individual genotypes.

Confocal microscopy

Col-0/PAT1-GFP plants were grown on MS agar containing 30 μg/ml hygromycin for 5 days. Seedlings were placed on glass microscope slides with water, 1% DMSO or 100 nM flg22 for 20 min. For following cycloheximide and DMSO treatment, seedlings were removed from flg22 and placed on new glass microscope slides. Cycloheximide was used at 100 μg/ml in a 1% DMSO dilution, and the control was done with 1% DMSO. Imaging was done using a Leica SP5 inverted microscope.

Infection assays

Pseudomonas syringae pv. tomato DC3000 (Pto DC3000) strains were grown in overnight culture in Kings B medium supplemented with appropriate antibiotics. Cells were harvested by centrifugation and pellets resuspended in sterile water to OD₆₀₀ = 0.002. Bacteria were infiltrated with a needleless 1-mm syringe into four leaves on four plants per genotype and maintained in growth chambers for 3 days. Samples were taken using a cork-borer (6.5 mm) to cut leaf disks from four leaves per plant and four plants per genotype. Leaf disks were ground in water, serially diluted and plated on Kings B with appropriate selection. Plates were incubated at 28°C and colonies counted 2–3 days later. These experiments were repeated in three independent biological replicates, and representative data are shown.

For drop inoculation of B. cinerea (strain B05.10), 10 μl of 2.5 × 10⁵ spores/ml in Gamborg B5/2% sucrose (pH 5.6) was placed on the adaxial surface of fully expanded leaves of 4-week-old plants and sampled by harvesting 10–15 leaf disks per genotype were at each time point, namely 0, 2 and 3 days. Data are the average of 3 biological replicates. Disease severity was measured as accumulation of the B. cinerea CUTINASE A transcript by qPCR relative to A. thaliana α-SHAGGY KINASE (At5g26751) (primers are according to Gachon & Saindrenan, 2004).

Transient expression in Nicotiana benthamiana

Agrobacterium tumefaciens strains were grown in LB medium supplemented with appropriate antibiotics overnight. Cultures were spun down and resuspended in 10 mM MgCl₂ to OD₆₀₀ = 0.8. A. tumefaciens strains carrying PAT1-GFP and MPK4-HA or SUMM2-HA were mixed 1:1 and syringe-infiltrated into 3-week-old N. benthamiana leaves. Samples for protein extraction were harvested 2 days post-infiltration (dpi). Agrobacterium tumefaciens strains carrying either PAT1-HA, LSM1-GFP or PAT1-HA + LSM1-GFP were mixed 1:1 and syringe-infiltrated into 3-week-old N. benthamiana leaves. Samples for protein extraction were harvested 3 dpi. For BiFC, A. tumefaciens strains carrying PAT1 fused to the N-terminal part of YFP and SUMM2 or At4g12010 fused to the C-terminal part were mixed 1:1 and syringe-infiltrated into N. benthamiana leaves at an final OD₆₀₀ = 0.8. Confocal microscopy on leaf disks was conducted 2 days post-infiltration under a Leica SP5 inverted microscope.

Protein extraction and immunoprecipitation in Nicotiana benthamiana

Leaves were ground in liquid nitrogen and extraction buffer [50 mM Tris-HCl pH 7.5; 150 mM NaCl; 10% (v/v) glycerol; 10 mM DTT; 10 mM EDTA; 1% (w/v) PVP; protease inhibitor cocktail (Roche); 0.1% (v/v) IGEPAL CA-630 (Sigma); Phosstop (Roche) added at 2 ml/g tissue powder. Samples were clarified by 20 min centrifugation at 4°C and 13,000 rpm. Supernatants (1.5 ml) were adjusted to 2 mg/ml protein and incubated 2 h at 4°C with 20 μl GFPTrap-A beads (Chromotek) or anti-HA antibodies (2 μg, Santa Cruz) and 30 μl EZview protein A agarose beads (Sigma). Following incubation, beads were washed four times with IP buffer, before adding 30 μl 2× SDS and heating at 95°C for 5 min.

Arabidopsis protein extraction and immunoprecipitation for mass spectrometry analysis

Leaves from adult plants were ground in liquid nitrogen and extraction buffer [50 mM Tris-HCl pH 7.5; 150 mM NaCl; 10% (v/v) glycerol; 5 mM DTT; 2 mM EDTA; protease inhibitor cocktail (Roche); 0,1% (v/v) IGEPAL CA-630 (Sigma) and Phosstop (Roche) added at 2 ml/g tissue powder. Samples were clarified by 20 min centrifugation at 4°C 13,000 rpm. Supernatants (45 ml) were adjusted to 3 mg/ml protein and incubated 4 h at 4°C with 50 μl GFPTrap-A beads (Chromotek) or anti-HA antibodies (4 μg, Santa Cruz) and 100 μl EZview protein A agarose beads (Sigma). Following incubation, beads were washed four times with IP buffer before adding 2× SDS and heating to 95°C for 5 min.

SDS-PAGE and immunoblotting

SDS-PAGE gels were prepared with either 8, 10 or 12% cross-linking. Proteins were loaded and gels run at 100–150 V for 1.5 h before electroblotting onto PVDF membrane (GE Healthcare). Membranes were rinsed in TBS and blocked for 1 h in 5% (w/v) non-fat milk in TBS-Tween (0.1% (v/v)). Antibodies were diluted in blocking solution to the following dilutions: anti-GFP (AMS Biotechnology (rabbit) 1:5,000 or Santa Cruz (mouse) 1:1,000), anti-HA 1:1,000 (Santa Cruz). Membranes were incubated with primary antibodies for 1 h to overnight. Membranes were washed 3 × 10 min in TBS-T (0.1%) before 1-h incubation in secondary antibodies, anti-rabbit or anti-mouse-HRP or anti-rabbit or anti-mouse AP conjugate (Promega; 1: 5,000). Chemiluminescent substrate (ECL Plus, Pierce) was applied before exposure to film (AGFA CP-BU). For AP-conjugated primary antibodies, membranes were incubated in NBT/BCIP (Roche) until bands were visible. For probing immunoblots with multiple antibodies, stripping was carried out using Restore Western Blot Stripping Buffer (Pierce) for 15–30 min, following by three washes with TBS-Tween.

Antibodies

Polyclonal anti-PAT1 antibodies were generated by immunizing rabbits with synthetic peptides derived from the N-terminus [EQRIPDRTKLYPEPQ] and C-terminus [KRSMLGSQKTEPVLS] of PAT1. Antibodies (final bleed) were affinity-purified against the C-terminal peptide (Eurogentec). Antibody specificity was verified by immunoblotting with plant extracts derived from Col-0 and pat1-1 tissue. Mouse anti-HA and anti-GFP antibodies were obtained from Santa Cruz. Rabbit anti-GFP antibodies were obtained from AMS Biotechnology. Anti-MPK3, anti-MPK-4 and anti-MPK-6 antibodies were obtained from Sigma. Secondary antibodies were obtained from Promega.

In-gel digestion, TiO₂ chromatography and mass spectrometry

Bands excised from SDS-PAGE were chopped into small pieces and incubated for 1 h in 50 mM TEAB, acetonitrile (50:50) on a shaker at room temperature. After incubation, the sample was centrifuged shortly and the supernatant was discarded. Gel pieces were dried in a vacuum centrifuge for 15 min and subsequently 30 μl of trypsin (10 ng/μl) in 20 mM TEAB pH 7.5 was added to cover the dried gel pieces. The solution was incubated in 4°C for 30 min. After incubation, the trypsin solution was replaced with 30 μl 20 mM TEAB pH 7.5 and the tube was incubated at 37°C overnight. Peptides from the digestion solution after incubation were recovered in a low binding Eppendorf tube (Sorensen Bioscience), and the gel pieces washed with 50% acetonitrile for 15 min to extract more peptides. The washing solution was mixed with the recovered peptides, and the peptide solution was lyophilized. Phosphopeptides were purified by titanium dioxide (TiO₂) chromatography (Larsen et al, 2005). Lyophilized peptides were redissolved in loading buffer for TiO₂ chromatography (80% acetonitrile, 5% TFA, 1 M glycolic acid), and 0.3 mg TiO₂ beads (GL Science, Japan) were added to the solution and incubated for 10 min. After incubation, the beads were pelleted by centrifugation and the supernatant was removed. The beads were washed once with 80% acetonitrile, 1% TFA and once with 10% acetonitrile, 0.1% TFA. Phosphopeptides were eluted using 1% ammonium hydroxide and desalted and concentrated prior to LC-MS/MS using a Poros Oligo R3 micro-column (Engholm-Keller et al, 2012).

For quantitative phosphopeptide analysis, the tryptic peptide mixtures after in-gel digestion were labeled with iTRAQ 4 plex (according to the manufacturer's protocol) and the samples were mixed prior to TiO₂ enrichment as described above. The non-phosphorylated peptides were concentrated and desalted using a Poros Oligo R3 micro-column (Engholm-Keller et al, 2013). The non-phosphorylated peptides were used to normalize the PAT1 protein level.

LC-MS/MS analysis was performed using an EASY-LC system (Proxeon, Thermo Fischer Scientific) coupled to an LTQ-Orbitrap XL mass spectrometer (Thermo Fisher Scientific) as described previously (Engholm-Keller et al, 2012). Peptides were separated using a 20 min gradient from 0–34% B-buffer (A-buffer: 0.1% formic acid; B-buffer: 90% acetonitrile, 0.1% formic acid). The data-dependent analysis was performed using one MS full scan in the area 300–1,800 Da performed in the Orbitrap with 60,000 in resolution, followed by the five most intense ions selected for MSMS (collision induced dissociation) performed in the linear ion trap.

Raw data from the LTQ-Orbitrap-XL were processed using the Proteome Discoverer (PD) program (Thermo Fisher Scientific, Bremen, Germany). The data were searched against the Arabidopsis database (33,596 sequences; 13,487,687 residues) using the Mascot 2.2 version database. The parameters for the database search were as follows: Enzyme—trypsin; missed cleavages—2; MS mass accuracy—10 ppm; MSMS mass accuracy—0.8 Da; Variable modifications—Phosphorylation (S, T, Y), Oxidation (met), Propionamide (C). The identified peptides were filtered for 1% false discovery rate using the ‘Fixed Value PSM Validator’ in PD. All identified phosphopeptides were manually validated. For iTRAQ-labeled phosphopeptides, the PD quantitation node was used and the data were normalized based on the non-phosphorylated peptides.