The phosphoproteome of Arabidopsis plants lacking the oxidative signal-inducible1 (OXI1) protein kinase


Author for correspondence:
Lee J. Sweetlove
Tel: + 44 (0) 1856 275000


  • The AGC protein kinase OXI1 is a key protein in plant responses to oxidative signals, and is important for two oxidative burst-mediated processes: basal resistance to microbial pathogens and root hair growth. To identify possible components of the OXI1 signalling pathway, phosphoproteomic techniques were used to detect alterations in the abundance of phosphorylated proteins and peptides in an oxi1 null mutant of Arabidopsis thaliana.
  • The relative abundance of phosphorylated proteins was assessed either using two-dimensional gel electrophoresis and staining with the phosphoprotein stain Pro-Q Diamond or by the identification and quantification, by mass spectrometry, of stable-isotope labelled phosphopeptides.
  • A number of proteins show altered phosphorylation in the oxi1 mutant. Five proteins, including a putative F-box and 3-phosphoinositide-dependent kinase 1, show reduced phosphorylation in the oxi1 mutant, and may be direct or indirect targets of OXI1. Four proteins, including ethylene insensitive 2 and phospholipase d-gamma, show increased phosphorylation in the oxi1 mutant.
  • This study has identified a range of candidate proteins from the OXI1 signalling pathway. The diverse activities of these proteins, including protein degradation and hormone signalling, may suggest crosstalk between OXI1 and other signal transduction cascades.


Reactive oxygen species (ROS) are important signalling molecules in plants, being produced by the oxidative burst and as by-products of numerous metabolic pathways. A range of biotic and abiotic stimuli have been shown to induce ROS production, including pathogen attack, wounding and temperature stress (Apel & Hirt, 2004; Fujita et al., 2006). While there is much evidence supporting a role for ROS in plant signal transduction, relatively little is known about the protein components that link ROS to downstream signalling. Previously, an AGC protein kinase, termed OXI1, has been shown to be induced directly by ROS and is also induced by stimuli that result in ROS production, such as wounding and pathogen infection (Anthony et al., 2004; Rentel et al., 2004). OXI1 has been shown to be important in two plant responses linked to ROS production: pathogen resistance and root hair growth. Arabidopsis plants with an oxi1-null mutation have enhanced susceptibility to infection by virulent strains of the oomycete pathogen Peronospora parasitica (Rentel et al., 2004) and also the biotrophic bacterial pathogen Pseudomonas syringae pv. tomato DC3000 (Petersen et al., 2009). In addition, oxi1-null mutant plants exhibit reduced root hair elongation (Anthony et al., 2004; Rentel et al., 2004).

OXI1 may also be activated in a ROS-independent manner by 3-phosphoinositide-dependent kinase 1 (AtPDK1). Upstream of OXI1, PDK1 activity is regulated by binding to phosphatidic acid. Upon activation, PDK1 can then bind to and phosphorylate OXI1, resulting in OXI1 activation (Anthony et al., 2004). Downstream, OXI1 has been shown to be required for the activation of the mitogen- activated protein kinases MPK3 and MPK6 (Rentel et al., 2004) and also the serine/threonine protein kinase PTI1-2 (Anthony et al., 2006). However, additional components of the OXI1 signalling pathway have remained elusive. The identification of such components is essential if we wish to elucidate the precise mechanisms by which OXI1 regulates plant growth and helps plants respond to biotic stress.

In the current investigation, a proteomics strategy was used for the identification of proteins that may be directly or indirectly phosphorylated by OXI1 kinase. By comparing quantitative changes in protein phosphorylation in an oxi1-null mutant of Arabidopsis with wild-type, it was possible to identify proteins whose phosphorylation was directly or indirectly regulated by OXI1. Altered abundance of phosphorylated proteins was assessed using either two-dimensional gel electrophoresis and staining with Pro-Q Diamond (Invitrogen, Paisley, UK) or by direct analysis using mass spectrometry of purified phosphorylated peptides labelled with heavy isotopes.

Materials and methods

Plant growth conditions

Wild-type Arabidopsis thaliana (L.) Heynh ecotype Ws-2 and the OXI1 T-DNA insertion mutant (Rentel et al., 2004) were grown in a sterile hydroponic system. Sterilized seeds were sown on stainless steel wire mesh platforms (with a mesh diameter of 0.45 mm). Mesh platforms were placed within Magenta pots and 130 ml of Murashige and Skoog (MS) media (4.4 g l−1 MS, 10 g l−1 sucrose, 0.4 g l−1 MES (2-(N-morpholino)ethanesulfonic acid), pH 5.8) was added to the pots so that the final volume was just below the surface of the wire mesh. Plants were grown for 14 d under a photoperiod of 16 h and at a constant temperature of 21°C. After 14 d plant roots were treated for 30 min with 0.1% w/v cellulase (Melford Laboratories Ltd, Ipswich, UK) to induce OXI1 expression and roots were harvested, blotted on filter paper and frozen immediately in liquid nitrogen.

Protein extraction

Proteins were extracted from Arabidopsis root tissue according to the method described by Isaacson et al. (2006). Briefly, root tissue was homogenized in liquid nitrogen and suspended in ice-cold extraction buffer (0.7 M sucrose, 0.1 M KCl, 0.5 M Tris-HCl, pH 7.5, 50 mM EDTA, 2% v/v β-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 50 mM sodium pyrophosphate, 25 mM sodium fluoride and 1 mM sodium molybdate) and an equal volume of ice-cold phenol saturated with Tris-HCl (pH 7.5). Samples were shaken for 30 min at 4°C and then centrifuged at 5000 g for 30 min at 4°C. The upper phenolic phase was collected and an equal volume of extraction buffer was added for re-extraction. Samples were shaken and centrifuged as before and the upper phenolic phase was re-extracted once more. Proteins were then precipitated from this phase with five volumes of ice-cold 0.1 M ammonium acetate in methanol at −20°C overnight. Proteins were pelleted by centrifugation at 5000 g and the protein pellet washed three times with ice-cold methanol followed by three washes with ice-cold acetone.

Two-dimensional SDS-PAGE and phosphoprotein staining

Protein samples were suspended in sample buffer (10 ml of buffer contains 3.6 g urea, 1.52 g thiourea, 0.2 g 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate, 0.03 g dithiothreitol (DTT)) and total protein quantified by Bradford assay. Once quantified, IPG buffer (GE Healthcare, Chalfont St Giles, UK) (200 μl per 10 ml) and trace bromophenol blue were added to the samples and 280 μg of protein was loaded on to 180 mm, pH 4–7, nonlinear Immobiline DryStrips (GE Life Sciences, Buckinghamshire, UK). Isoelectric focusing (IEF) was performed on a multiphor II electrophoresis unit (GE Life Sciences) for 22 h at 20°C at the following voltage: 500 V for 5 h, gradient to 3500 V over 5 h, 3500 V for 12 h (current and power were were maintained at 1 mA and 5 W, respectively). Once IEF was complete, the Immobiline Strips were equilibrated and placed on 12% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) gels and run at 10 mA overnight. Gels were stained for protein phosphorylation using Pro-Q Diamond stain following the manufacturer’s instructions. Gels were subsequently stained for total protein using colloidal Coomassie stain (10% w/v ammonium sulphate, 10% v/v phosphoric acid, 20% v/v methanol and 0.12% w/v G250). Protein spots were quantified using PD-Quest software (Bio-Rad). Data were acquired from three independent biological replicates.

Protein spots were cut and identified by digestion with trypsin followed by peptide mass fingerprinting as described by Millar et al. (2001) but with digestion carried out at room temperature overnight. Mass spectra were obtained using a Shimadzu Axima CFR MALDI-TOF (Shimadzu Biotech, Milton Keynes, UK) in positive ion reflectron mode, and peptide mass lists were searched against the National Centre for Biotechnology Information gene database using an in-house Mascot server (Matrix Science Ltd, London, UK). The following settings were used: peptide mass tolerance ± 50 ppm, maximum missed cleavages = 3.

Stable isotope labelling

Wild-type A. thaliana ecotype Ws-2 and the OXI1 T-DNA insertion mutant were grown for 14 d in a sterile hydroponic system as described earlier but with wild-type plants grown in MS media in which nitrogen compounds (potassium nitrate and ammonium nitrate) were replaced with 15N-labelled stable isotopes (labelled to ≥ 98%; CK Gas Products Ltd, Hook, UK). Plants were treated with cellulase, roots harvested and protein extracts generated as described earlier. Wild-type and OXI1 mutant protein samples were suspended in 10 mM Hepes, 6 M urea, 2 M thiourea and quantified by Bradford assay before being combined 1 : 1 (1 mg of wild-type protein extract and 1 mg of OXI1 mutant protein extract). Samples were collected from three independent biological replicates.

Protein digestion and phosphopeptide enrichment

Protein samples were reduced for 45 min at room temperature with the addition of DTT to a final concentration of 1 mM and alkylated for 30 min with the addition of iodoacetamide to a final concentration of 5 mM. Protein samples were diluted with 10 mM Tris, pH 8.0, to achieve a final urea concentration of 2 M before digestion overnight with 20 μg trypsin. Digested protein samples were fractionated by strong-cation exchange chromatography (SCX) as described by Olsen et al. (2006) but using a gradient of NH4HCO2 of 5–200 mM. Seven 4 ml peptide fractions were collected and enriched for phosphopeptides using titanium dioxide (TiO2) chromatography columns. One millilitre of 30 g l−1 2,5-dihydroxybenzoic acid (DHB) in 80% acetonitrile (ACN)/0.1% trifluoroacetic acid (TFA) was added to each pooled peptide fraction before the addition of TiO2 beads (5 mg of beads in 3g l−1 DHB with 80% ACN/0.1% TFA). Peptide fractions were incubated with TiO2 beads, shaking for 1 h at room temperature, and the beads were harvested by centrifugation. Beads were washed with a solution of 3 g l−1 DHB, 80% ACN/0.1% TFA before being washed three times with 80% ACN/0.1% TFA. The beads were then loaded on to a homemade C8 column within a gel-loader tip and washed with 80% ACN. Phosphopeptides were eluted with two successive washes of 100 μl elution buffer (2% NH4OH, 40% ACN). Eluted samples were dried in a Speed-Vac and suspended in 10 μl of 0.1% TFA before being subjected to LC-MS/MS analysis.

Mass spectrometry data were acquired on an Orbitrap mass spectrometer (Thermo Scientific, Hemel Hempstead, UK) fitted with a nanospray source (Proxeon, Thermo Scientific) coupled to a U3000 nano HPLC system (Dionex, Leeds, UK). Samples were loaded on to a 15-cm-long, 100 μm internal diameter, home-packed column made by packing a Picotip emitter (New Objective, Woburn, MA, USA ) with ProntoSIL C18 phase; 120 angstrom pore, 3 μm bead C18 (ReproSil-Pur C18-AQ 3 μm; Dr Maisch GmbH, Germany). The nano HPLC was run in a direct injection configuration. Buffer A consisted of 94.5% H2O, 5% ACN and 0.5% acetic acid, and buffer B consisted of 99.5% ACN and 0.5% acetic acid. Samples were loaded on to the column at a flow rate of 700 nl min−1, and resolved using a 120 min gradient at a flow rate of 300 nl min−1. The Orbitrap was run in a data-dependent acquisition mode in which the Orbitrap resolution was set at 60 000 and the top five multiply charged species were selected for MS/MS.

MS data analysis

Raw data files were searched against the NCBInr database ( using MASCOT with the following settings: peptide mass tolerance ± 20 ppm, fragment mass tolerance ± 0.5 Da, maximum missed cleavages = 3, 15N metabolic quantitation and the variable modifications carbamidomethylation of cysteine, oxidation of methionine and phosphorylation of serine, threonine and tyrosine. Peptides were quantified using MSQuant software ( with the default settings for quantification of 15N-labelled peptides (Mortensen et al., 2010). 15N peptide ratios were corrected for inaccurate peptide ratios introduced by incomplete 15N labelling using the method described by Gouw et al. (2008). Additional 15N-labelled peptides were also detected following the method described by Gouw et al. (2008). The degree of 15N incorporation was determined using the isotope distribution calculator ICR-2LS (Pacific Northwest National Laboratory, Richland, WA, USA;


Conditions for inducing OXI1 expression

Previously OXI1 kinase has been shown to be induced in roots and leaves of A. thaliana in response to oxidative stress. OXI1 expression is also up-regulated in response to cellulase and hydrogen peroxide treatment (Rentel et al., 2004). Before commencing detailed phosphorylation studies we wished to confirm OXI1 expression in our experimental system. The roots of 14-d-old Arabidopsis seedlings were treated with 0.1% (w/v) cellulase for 30 min. This time point and cellulase concentration were selected based on the results of Rentel et al. (2004) in which cellulase was shown to be a strong activator of OXI1 with expression increasing over time. Using an OXI1:: luciferase Arabidopsis transgenic line we confirmed OXI1 expression in the roots and leaves of Arabidopsis seedlings treated with 0.1% cellulase (data not shown).

Two-dimensional PAGE reveals differences in protein phosphorylation

A preliminary screen using two-dimensional PAGE and staining with Pro-Q Diamond, identified four candidate proteins that showed reduced phosphorylation in the oxi1 null mutant background when compared with wild-type plants (Supporting Information, Fig. S1). Three of these candidate proteins were identified by mass spectrometry: citrate synthase 4 (AT2G44350), a ubiquitin-activating enzyme (AT2G30110) and COP9 signalosome 5A (AT2G30110) all showed reduced phosphorylation in the oxi1 null mutant (Fig. S1, Table S1). All three proteins showed at least a 1.5-fold difference in phosphorylation and this difference was observed in three independent biological replicates (Fig. S2). None of these proteins were altered in overall protein abundance to a degree sufficient to account for altered Pro-Q staining (Table S2), indicating that the difference in protein phosphorylation is not the result of altered protein expression. It is important to note that a single gel spot may comprise multiple proteins, any one of which may be phosphorylated and contribute to a signal on a Pro-Q stained gel. The candidate proteins identified in this study would therefore require their phosphorylation status to be confirmed by further analysis.

Stable isotope labelling of whole plants and phosphopepetide enrichment enables a direct comparison of the abundance of wild-type and oxi1 mutant phosphopeptides

Wild-type and oxi1 null mutant plants were grown in a sterile hydroponic system in which the nitrogen source within the growth medium was either unlabelled (in the case of oxi1 null mutant plants) or 15N-labelled (in the case of wild-type plants). Such a strategy allows protein samples from the two lines to be combined and processed together, eliminating any variation in sample preparation and any variations in peptide identification and quantitation during mass spectrometry. An overview of the complete experimental strategy is shown in Fig. 1. The degree of 15N incorporation was calculated to be 98.3%, and this value was used to adjust the quantified peptide data for improved accuracy according to the method described by Gouw et al. (2008). Phosphopeptide enrichment and LC-MS/MS analysis enabled the identification of over 1600 phosphopeptides (Table S3), representing a good coverage of the phosphoproteome when compared with the results of other studies (Nuhse et al., 2004; Benschop et al., 2007; Sugiyama et al., 2008).

Figure 1.

 Experimental strategy for isotope labelling, protein extraction, phosphopeptide enrichment and LC-MS/MS based proteomics. Wild-type and oxi1 null mutant plants were grown in a sterile hydroponic system supplemented with light (14N) or heavy (15N) nitrogen. After 14 d, roots were treated with 0.1% cellulase for 30 min to induce OXI1 expression. Roots were harvested and proteins extracted before combining wild-type and OXI1 protein samples 1 : 1. The protein sample was digested with trypsin, and peptides fractionated with strong cation exchange chromatography (SCX) followed by phosphopeptide enrichment using titanium dioxide columns. Peptides were analysed by LC-MS/MS, and the 14N and 15N peptides, corresponding to OXI1 mutant and wild-type proteins, respectively, were quantified using MSQuant software. Data were collected from three independent biological replicates.

Quantifiable differences in phosphopeptide abundance are observed between wild-type and oxi1 mutant plants

Stable isotope labelling of whole plants combined with phosphopeptide enrichment and peptide identification by LC-MS/MS identified a number of proteins that show altered abundance of phosphorylated peptides in the oxi1 null mutant. Peptides were only considered as differentially abundant if they showed at least a twofold difference in abundance in all three biological replicates. Peptides showing conflicting alterations in abundance between fractions and between the three replicates were discarded from the data set. Two categories were identified: proteins that show decreased phosphopeptide abundance in oxi1 mutant plants compared with wild-type, and proteins that show increased phosphopeptide abundance in oxi1 mutant plants compared with wild-type. A putative F-box protein (AT1G27540) and phosphoinositide-dependent kinase 1 (AtPDK1, AT5G04510; putative PDK1, AT3G10540) both showed decreased phosphopeptide abundance in the oxi1 mutant compared with wild-type plants (Fig. 2, Table 1). Four proteins showed increased phosphopeptide abundance in oxi1 mutant plants compared with wild-type: ethylene insensitive 2 (EIN2, AT5G03280), seed imbibition 1-like (SIP1, AT5G40390), phospholipase d-gamma (AT4G11850) and an unknown protein (AT1G31870) (Fig. 2, Table 1). Annotated LC-MS/MS spectra showing peptide fragmentation and predicted phosphorylation sites are shown in Fig. S3. The putative F-box protein (AT1G27540) was only sequenced by LC-MS/MS in one replicate. Additional data supporting the identity of this protein and its quantitation are provided in Fig. S4.

Figure 2.

 Representative LC-MS spectra showing quantifiable differences in phosphopeptide abundance when comparing oxi1 null mutant plants to wild-type. Protein identity is shown, as is the predicted position of the phosphorylated residue (underlined). The mass of the monoisotopic peak (where present) is shown for each peptide. The horizontal double-headed arrow above each spectrum indicates the pairs of 14N- and 15N-labelled peptides. Peptide ratios and identification data are presented in Table 1 for all peptides and for the three biological replicates. ATPDK1, 3-phosphoinositide-dependent kinase 1; EIN2, ethylene insensitive 2; SIP1, seed imbibition 1-like.

Table 1.   Identity of proteins with altered phosphorylation in the oxi1 mutant as revealed by stable isotope labelling and phosphopeptide enrichment
 ProteinPeptide sequenceMascot ion scorePeptide ratio (oxi1 : WT)
  1. PDK1, phosphoinositide-dependent kinase 1; EIN2, ethylene insensitive 2; SIP1, seed imbibition 1-like.

  2. *The putative F-box protein was only sequenced by LC-MS/MS in one replicate.

  3. Phosphopeptides with at least a twofold change in abundance in the oxi1 mutant compared with wild-type are shown. Predicted phosphorylation sites are underlined. Where unambiguous phosphorylation site assignment cannot be achieved, multiple sites are underlined. The table shows the mascot score for each peptide and the peptide ratio (average of three biological replicates ± SD). Additional data supporting the identity of this protein and its quantitation are provided in Fig. S4.

AT5G04510/AT3G10540PDK1LASIDSFDSR56, 65, 530.42 ± 0.16
AT5G03280EIN2AAPTSNFTVGSDGPPSFR46, 52, 722.43 ± 0.36
AT5G40390SIP1SDSGINGVDFTEK67, 66, 724.20 ± 1.56
AT4G11850Phospholipase d-gammaEVPVGTVSVYNSPR85, 52, 713.45 ± 1.16
AT1G31870Unknown proteinSFGSNADLSPPGR67, 70, 822.41 ± 0.54


This study aimed to identify components of the OXI1 signalling pathway in A. thaliana by examining the phosphorylation status of the proteome of wild-type and oxi1 null mutant plants. Two-dimensional gel electrophoresis along with stable isotope labelling and phosphopeptide enrichment have enabled us to identify a number of candidates that show altered phosphoprotein/peptide abundance in a oxi1 null mutant of Arabidopsis.

In the current investigation, six proteins were identified that showed altered phosphopeptide abundance in the oxi1 null mutant, using isotope labelling, phosphopeptide enrichment and LC-MS/MS. All of these proteins were also shown to be phosphorylated in a recent study by Reiland et al. (2009). In fact, all but PDK1 were shown to be phosphorylated on the same peptide as those identified by Reiland et al. (2009), raising confidence in the identification of the phosphopeptides in the current investigation.

Within our list of phosphorylated candidates, two proteins, phosphoinositide-dependent kinase 1 (PDK1) and phopholipase d-gamma (PLD), have previously been shown to be upstream in the OXI1 signal transduction pathway. The PDK1 phosphopeptide was found in lower abundance in the oxi1 mutant, while the PLD phosphopeptide was found in greater abundance in the oxi1 mutant relative to wild-type. OXI1 is activated by PDK1-mediated phosphorylation and PDK1 activity is regulated by binding the lipid signalling molecule phosphatidic acid (PA). PA-induced activation of PDK1 is likely to occur through the activity of PLD (Anthony et al., 2006). A range of stresses, including oxidative stress, have been shown to activate PLD and subsequent PA biosynthesis (Bargmann & Munnik, 2006). PDK1 has previously been shown to be phosphorylated in mammals and in plants, and phosphorylation is important for PDK1 activity (Casamayor et al., 1999; Otterhag et al., 2006). Our observation that PDK1 is phosphorylated to a reduced extent in the oxi1 mutant may suggest feedback from OXI1 to PDK1 for the control of PDK1 activity. The phosphorylation of PDK1 may be dependent on the presence of its substrate, OXI1. The phosphorylation of PDK1 may occur directly via OXI1 kinase activity or indirectly by the activity of another kinase in the OXI1 signalling pathway. The activity of plant PLD enzymes may also be regulated by phosphorylation (Novotna et al., 2003). PLD phosphorylation was increased in the oxi1 mutant, indicating that PLD is not likely to be a direct downstream target of OXI1 but rather an indirect target during the regulation of the OXI1 signalling cascade.

A putative F-box protein was also found to have reduced phosphopeptide abundance in the oxi1 null mutant. F-box proteins are part of the SCF (SKP1-CUL1-F-box) protein complex, which is a class of E3 ubiquitin ligases. The SCF complex catalyses the ubiquitination of target proteins, which are subsequently degraded by the 26S proteosome. The role of F-box proteins within the SCF complex is to recognize and bind a substrate protein for degradation (Lechner et al., 2006). Arabidopsis is estimated to have almost 700 F-box proteins and these proteins are involved in diverse activities, including hormone signalling, root and shoot development and pathogen resistance (Xu et al., 2009).

A phosphorylated peptide of the ethylene response regulator EIN2 was found to be more abundant in the oxi1 mutant background. This increase in phosphorylation suggests that EIN2 is not a direct target of OXI1, but is affected as a secondary consequence in the oxi1 mutant, possibly as a result of cross-talk between the OXI1 and ethylene signalling pathways. Corroborating the link between OXI1 and ethylene signalling, both oxi1 and ein2 mutants share a common phenotype of reduced root hair elongation (Pitts et al., 1998; Rentel et al., 2004). While EIN2 phosphorylation has been detected in previous large-scale phosphorylation studies (Sugiyama et al., 2008, Reiland et al., 2009),, the importance of this modification for the plant ethylene response is yet to be determined.

The candidate proteins identified in this investigation may be direct or indirect targets of OXI1 kinase and may be positioned upstream or downstream of OXI1 within the signal transduction pathway. Further work will be required to establish whether any of the five proteins with decreased phosphorylation in the oxi1 mutant are direct phosphorylation targets of OXI1.

In summary, we have identified a range of alterations in the phosphoproteome of oxi1 mutant plants and have discovered a number of new candidate proteins for the OXI1 signalling cascade. The phosphoproteomic approaches employed in this study have proved extremely sensitive for the quantification of changes in protein phosphorylation. These results may suggest that the plants’ response to oxidative stress through OXI1 is closely linked to other signalling pathways, including plant hormone signalling and protein degradation.


The authors would like to thank Edward Huttlin (University of Wisconsin) and Joost Gouw (Utrecht University) for advice with 15N peptide quantitation, and Peter Mortensen (University of Southern Denmark) for advice on using MSQuant. We thank David Trudgian (University of Oxford) for useful discussions regarding data analysis. This work was supported by a BBSRC grant awarded to L.J.S. and M.R.K. M.S. was supported by Welcome Trust Grant GR076558MA.