Abstract
Protein phosphorylation regulates a wide range of cellular processes. Here, we report the proteome‐wide mapping of in vivo phosphorylation sites in Arabidopsis by using complementary phosphopeptide enrichment techniques coupled with high‐accuracy mass spectrometry. Using unfractionated whole cell lysates of Arabidopsis, we identified 2597 phosphopeptides with 2172 high‐confidence, unique phosphorylation sites from 1346 proteins. The distribution of phosphoserine, phosphothreonine, and phosphotyrosine sites was 85.0, 10.7, and 4.3%. Although typical tyrosine‐specific protein kinases are absent in Arabidopsis, the proportion of phosphotyrosines among the phospho‐residues in Arabidopsis is similar to that in humans, where over 90 tyrosine‐specific protein kinases have been identified. In addition, the tyrosine phosphoproteome shows features distinct from those of the serine and threonine phosphoproteomes. Taken together, we highlight the extent and contribution of tyrosine phosphorylation in plants.
Visual Overview
Identification of more than two thousand phosphorylation sites in Arabidopsis.
Tyrosine phosphoproteome does exist in plants in spite of the absence of typical human‐type tyrosine‐specific protein kinases.
The tyrosine phosphorylation profile has the different features from that of the serine/threonine phosphorylation in terms of the site location and the site conservation in plant homologs.
Most of the identified pY motifs are novel and distinct from the human pS, pT and pY motifs.
Introduction
Protein phosphorylation is a critical regulatory step in signaling networks and is arguably the most widespread protein modification affecting almost all basic cellular processes in various organisms (Hunter, 2000; Manning et al, 2002).
Advances in mass spectrometry (MS)‐based technologies accompanied with phosphopeptide enrichment methods paved the way for high‐throughput, large‐scale in vivo phosphorylation site mapping, and indeed, several pioneering plant phosphoproteome studies have been reported in the past 5 years (Nuhse et al, 2003, 2004, 2007; de la Fuente van Bentem et al, 2006; Benschop et al, 2007). Although these studies provided new insights into phosphorylation events in plants, the analyses were restricted to subfractionated samples, such as plasma membrane proteins, containing a few hundred phosphoproteins. No plant study has yet been reported using unfractionated whole cells to provide a wide‐ranging view of cellular phosphorylation events.
More than 1000 phosphorylation sites have recently been identified in animal and yeast cells, using a combination of two or more methods for phosphopeptide enrichment coupled with mass spectrometric phosphopeptide‐oriented techniques, such as neutral loss‐triggered MS3 to generate fragment ions after elimination of labile phosphate groups, multistage activation, and electron transfer dissociation (Olsen et al, 2006; Bodenmiller et al, 2007a; Chi et al, 2007; Molina et al, 2007; Villen et al, 2007). We also reported the identification of more than 2000 in vivo phosphorylated sites in unstimulated HeLa cells employing an aliphatic hydroxy acid‐modified metal oxide chromatography (HAMMOC) as a phosphopeptide enrichment method (Sugiyama et al, 2007). Since different phosphopeptide enrichment methods are likely to have distinct preferences for particular properties of phosphopeptides (Bodenmiller et al, 2007b), it is reasonable to use two or more phosphopeptide enrichment methods for evaluation of proteome‐wide phosphorylation.
Comparative genome analyses revealed substantial differences in the ensembles of kinases (kinomes) in eukaryotes (Diks et al, 2007). The Arabidopsis genome encodes at least two times more protein kinases than the human genome (Manning et al, 2002; Champion et al, 2004). Importantly, the Arabidopsis genome (Initiative, 2000) does not contain any predicted human‐type TKs (Rudrabhatla et al, 2006). However, plants are likely to utilize tyrosine phosphorylation signaling, as bona fide tyrosine‐specific protein phosphatases do exist in Arabidopsis (Xu et al, 1998; Luan, 2003), and a few early studies detected tyrosine phosphorylation by using pY antibodies (Barizza et al, 1999; Kameyama et al, 2000; Luan, 2002). In addition, a previous Arabidopsis phosphoproteome study identified a small number of phosphorylated tyrosine residues, although the actual data sets were missing in the report (Benschop et al, 2007). Thus, evidence for tyrosine phosphorylation in plants is limited so far.
Here, we present a large‐scale phosphoproteome analysis of Arabidopsis, providing an overview of in vivo phosphorylation events in Arabidopsis at the cellular level. Importantly, we show the extent of tyrosine phosphorylation in plants, which has been largely underestimated to date.
Results and discussion
Large‐scale in vivo phosphorylation site mapping in Arabidopsis
To collect a comprehensive data set of Arabidopsis phosphorylation sites, we employed six distinct methods for phosphopeptide enrichment (Supplementary information). Our approach identified 2172 unique phosphorylation sites with very high confidence on 1346 proteins from unfractionated Arabidopsis cell lysates; this is one of the largest data sets available for a plant to date (Supplementary Table I and Supplementary information). A large majority (1155; 85.8%) of the identified phosphoproteins are novel, while 191 (14.2%) were reported in the previous phosphoproteome studies that focused on plasma membrane proteins (Nuhse et al, 2004; Benschop et al, 2007) (Supplementary Figure 2).
Arabidopsis phosphoproteome
To obtain an overview of phosphorylation events in Arabidopsis, protein abundance distribution, cellular localization, molecular function, and biological processes of identified phosphoproteins were analyzed and compared with those of all proteins encoded by the Arabidopsis genome (Supplementary Figures 3 and 4). Phosphoproteins were generally less abundant, as expected, even when we did not take account of the degree of phosphorylation (Supplementary information). Proteins from all subcellular compartments were found to be targets for phosphorylation. However, approximately 40% of phosphorylation occurred on predicted nuclear proteins. Since nuclear proteins account for only approximately 20% of all genome‐encoded proteins and 15% of the experimentally identified proteins in this study, phosphorylation is likely to target nuclear proteins preferentially (Supplementary Figures 4A and 5). The distributions of the molecular function and biological processes of phosphoproteins and that of all genome‐encoded proteins were relatively similar (Supplementary Figures 4B and C). This indicates that most cellular processes in Arabidopsis are likely to be regulated at least in part by various phosphorylation events.
To our surprise, of the 2172 identified phosphorylation sites, we found 94 sites to be tyrosine residues (Table I). The kinome of Arabidopsis does not contain any of the typical TKs found in humans, suggesting that plants and humans do not share mechanistic features of tyrosine phosphorylation. Nevertheless, the relative abundances of pS, pT, and pY in Arabidopsis were estimated to be 85.0, 10.7, and 4.3%, which are strikingly close to the human phosphoproteome profile recently reported. The proportion of pY among phospho‐residues in human cells is estimated to be between 1.8 and 6.0%, depending on the analyzed samples (Olsen et al, 2006; Molina et al, 2007; Sugiyama et al, 2007). These data clearly indicate that the importance of tyrosine phosphorylation in plants has been greatly underestimated.
Arabidopsis tyrosine phosphoproteome
The 94 identified pY residues were mapped on 95 proteins (Supplementary Table II). The difference in the number of pY residues and corresponding proteins is due to matching of single phosphopeptides to several different proteins. Since the sequences surrounding tyrosine phosphorylation sites on the listed protein kinases are often well conserved, the number of protein kinases is over‐represented. To investigate whether tyrosine phosphorylation is targeted to a specific subset of proteins, gene ontology analyses of serine‐, threonine‐, or tyrosine‐phosphorylated proteins were performed as described (Figure 1). Tyrosine phosphorylation preferentially occurs on proteins that possess kinase activity or transferase activity (Figure 1B). Otherwise, no outstanding differences were found in the distributions.
Gene ontology analysis of the serine‐, threonine‐, or tyrosine‐phosphorylated proteins. (A) Cellular localization, (B) molecular function, and (C) biological process.
Location of phosphorylation sites on characterized protein domains
To assess whether trends or patterns exist in the position of tyrosine phosphorylation sites, we investigated whether these sites are located in conserved domains. Pfam search (Bateman et al, 2004) was used to extract domain information of the identified phosphoproteins. Of the 1346 proteins, we obtained domain information for the 1118 proteins. In these proteins, 77.95% of phosphorylation sites (1548 sites) were located outside of conserved domains (Table II). The tendency that the majority of phosphorylation occurs outside of conserved domains is consistent with the observations from the phosphoproteome study of plasma membrane proteins (Nuhse et al, 2004). Interestingly, however, nearly half (48.5%) of pYs were found to be located on conserved domains (Table II). These data indicate that tyrosine phosphorylation may have more impact on domain‐associated function compared to serine and threonine phosphorylation.
Conservation of tyrosine phosphorylation sites in plant homologs
Conservation of the tyrosine phosphorylation sites between homologous proteins in Arabidopsis, rice (Oryza sativa), and poplar (Populus trichocarpa) was investigated to get an overview of tyrosine phosphorylation events in other plant species. Of the 95 tyrosine‐phosphorylated proteins (109 tyrosine phosphorylation sites), 84 proteins (97 sites) were validated to possess homologs in Arabidopsis (paralogs), while 89 (103 sites) and 92 (106 sites) proteins had corresponding homologs (orthologs) in rice and poplar, respectively (Supplementary Table II). Multiple sequence alignments of the homologous proteins were created using ClustalW (Thompson et al, 1994), and the conservation of the phosphorylatable tyrosine residues was verified manually (Figure 2). In total, 72 sites are conserved within Arabidopsis paralogs, while 72 and 79 sites are conserved in rice and poplar orthologs, respectively. Most of these sites (61 sites) are conserved in all three plant species, indicating that the tyrosine phosphorylation sites are nearly equally conserved in paralogs and orthologs. This observation is in clear contrast to the case of serine phosphorylation sites, which are less conserved in paralogs compared to orthologs (Nuhse et al, 2004).
Venn diagram showing the number of conserved tyrosine phosphorylation sites in homologs. Blue, yellow, and red circles indicate the conserved sites in Arabidopsis paralogs, rice orthologs, and poplar orthologs, respectively.
Distribution of the phosphorylation sites
We found that most (76.3%) of the pY‐containing phosphopeptides are multiply phosphorylated, while the majority (80.9%) of phosphopeptides are singly phosphorylated (Table III). In other words, tyrosine phosphorylation seems to occur near other phospho‐residues (Supplementary Table III). Since the amino acids surrounding the phosphorylation sites often contribute substantially to recognition by protein kinases, the phosphorylation status of neighboring residues is an essential factor in determining whether the phosphorylation site is targeted by a particular protein kinase. It would be very interesting to investigate whether and how the phosphorylation state of the neighboring residues affects tyrosine phosphorylation events in Arabidopsis.
Tyrosine phosphorylation motifs
An obvious question arising from our finding is, which kinases carry out tyrosine phosphorylation in plants? To address this question, we attempted to extract significant patterns surrounding the pY residues from our data set, assuming that conserved phosphorylation sites within functionally related proteins tend to be well targeted by structurally similar protein kinases. We have extracted 20 pY motifs through the substrate‐driven approach (Schwartz and Gygi, 2005) (Supplementary information). Most of the identified pY motifs are novel and distinct from the human pS, and pT and pY motifs in the human protein reference database (Amanchy et al, 2007). These results indicate that tyrosine phosphorylation in plants is carried out by a novel class(es) of plant kinases. One candidate might be dual‐specific serine/threonine/tyrosine protein kinases (Rudrabhatla et al, 2006). Other possible candidates would be tyrosine‐specific protein kinase‐like kinases (TKLs), which are especially abundant in plants: 776 in Arabidopsis and nearly 1000 in rice, compared to 55 in humans (Miranda‐Saavedra and Barton, 2007). Tyrosine phosphorylation by human TKLs has not been reported. Functions of plant TKLs remain also unknown, but the large number of TKLs in plants may suggest that they carry out important and diverse plant‐specific functions. In this sense, it is of particular interest to investigate if any of TKL possesses tyrosine phosphorylation activity.
Materials and methods
Plant material
Arabidopsis cell suspension line (ecotype Landsberg erecta) (Maor et al, 2007) was grown in Murashige and Skoog medium (pH 5.7) containing 3% sucrose, 0.59 g/l MES, 100 mg/l myo‐inositol, 10 mg/l thiamine‐HCl, 1 mg/l pyridoxine‐HCl, 1 mg/l nicotinic acid, 0.5 mg/l 1‐naphthaleneacetic acid, and 0.05 mg/l 6‐benzylaminopurine under a 16‐h light/8‐h dark cycle at 22°C. Seven‐day‐old Arabidopsis suspension cultures were harvested by vacuum filtration, frozen immediately in liquid nitrogen, and kept at −80°C until the analysis.
Digestion of Arabidopsis cell cytoplasmic fraction
Arabidopsis cells (0.2 g, wet) were frozen in liquid nitrogen and then disrupted with a Multi‐beads shocker (MB400U; Yasui Kikai). The disrupted cells were suspended in 0.1 M Tris–HCl (pH 8.0), containing protein phosphatase inhibitor cocktails 1 and 2 (Sigma) and protease inhibitors (Sigma). The homogenate was centrifuged at 1500 g for 10 min and the supernatant was reduced with dithiothreitol, alkylated with iodoacetamide, and digested with Lys‐C, followed by dilution and trypsin digestion as described (Saito et al, 2006). These digested samples were desalted using StageTips with C18 Empore disk membranes (3 M) (Rappsilber et al, 2003). The peptide concentration of the eluates was adjusted to 1.0 mg/ml with 0.1% TFA and 80% acetonitrile.
Enrichment of phosphopeptides
HAMMOC using titania and zirconia was performed as described previously (Sugiyama et al, 2007) with some modifications. Custom‐made MOC tips were prepared using C8‐StageTips and metal oxide bulk beads (0.5 mg beads per 10 μl pipette tip), as described for SCX(beads)‐C18 tips (Ishihama et al, 2006). Prior to loading samples, the MOC tips were equilibrated with 0.1% TFA, 80% acetonitrile, containing a hydroxy acid as a selectivity enhancer (solution A). As an enhancer, lactic acid was used at a concentration of 300 mg/ml for titania MOC tips and β‐hydroxypropanoic acid at 100 mg/ml for zirconia MOC tips. The digested sample from 100 μg of Arabidopsis total proteins was diluted with 100 μl of solution and loaded onto the MOC tips. After successive washing with solution A and solution B (0.1% TFA and 80% acetonitrile), 0.5% ammonium hydroxide or 1.0% disodium hydrogen phosphate was used for elution. The eluted fraction was acidified with TFA, desalted using C18‐StageTips as described above, and concentrated in a Tony CC‐105 vacuum evaporator (Tokyo, Japan), followed by the addition of solution A for subsequent nanoLC‐MS/MS analysis.
Fe‐IMAC was conducted using Phos‐Select (Sigma) as described previously (Kokubu et al, 2005; Ishihama et al, 2007), except for the use of C8‐StageTips instead of C18‐StageTips for packing Phos‐Select beads. Briefly, after loading the sample solutions, the tips were rinsed with 0.5 ml of 50% ACN in 0.3% TFA. Then, 0.5% ammonium hydroxide or 1.0% disodium hydrogen phosphate was used for elution. The eluted fraction was acidified with TFA and desalted using C18‐StageTips using HAMMOC methods. The eluted phosphopeptide fraction was concentrated in the vacuum evaporator and resuspended in solution A for nanoLC‐MS/MS analysis.
NanoLC‐MS system
A Finnigan LTQ‐Orbitrap (Thermo Fisher Scientific, Bremen, Germany) coupled with a Dionex Ultimate3000 pump (Germering, Germany) and an HTC‐PAL autosampler (CTC Analytics AG, Zwingen, Switzerland) was used for nanoLC‐MS/MS analyses throughout this study. ReproSil C18 materials (3 μm; Dr Maisch, Ammerbuch, Germany) were packed into a self‐pulled needle (150 mm length × 100 μm i.d., 6 μm opening) to prepare an analytical column needle with ‘stone‐arch’ frit (Ishihama et al, 2002). An x–y–z nanospray interface (Nikkyo Technos, Tokyo, Japan) was used to hold the column needle and to set the appropriate spray position. A spray voltage of 2400 V was applied. The injection volume was 5 μl and the flow rate was 500 nl/min. The mobile phases consisted of (A) 0.5% acetic acid and (B) 0.5% acetic acid and 80% acetonitrile. A three‐step linear gradient of 5–10% B in 5 min, 10–40% B in 60 min, 40–100% B in 5 min and 100% B for 10 min was employed throughout this study. The MS scan range was m/z 300–1400, and the top 10 precursor ions were selected for subsequent MS/MS scans. A lock mass function was used for the LTQ‐Orbitrap to obtain constant mass accuracy during gradient analysis (Olsen et al, 2005).
Database searching
Mass Navigator v1.2 (Mitsui Knowledge Industry, Tokyo, Japan) was used to create peak lists on the basis of the recorded fragmentation spectra. Peptides and proteins were identified by means of automated database search using Mascot v2.1 (Matrix Science, London) against TAIR7_pep_20070425 (ftp://ftp.arabidopsis.org/home/tair/Sequences/blast_datasets/TAIR7_blastsets/) with a precursor mass tolerance of 3 p.p.m., a fragment ion mass tolerance of 0.8 Da and strict trypsin specificity (Olsen et al, 2004), allowing for up to two missed cleavages. Carbamidomethylation of cysteine was set as a fixed modification, and oxidation of methionines and phosphorylation of serine, threonine, and tyrosine were allowed as variable modifications. Peptides were considered identified if the Mascot score was over the 95% confidence limit based on the ‘identity’ score of each peptide and at least three successive y‐ or b‐ions with a further two and more y‐, b‐, and/or precursor‐origin neutral loss ions were observed, based on the error‐tolerant peptide sequence tag concept (Mann and Wilm, 1994). A randomized decoy database created by a Mascot Perl program estimated a 2.1% false‐positive rate for identified peptides within these criteria. Note that most sulfated peptides can be discriminated from phosphopeptides because of the ultrahigh accuracy of the Orbitrap instrument that we used.
Phosphorylated sites were unambiguously determined when y‐ or b‐ions between which the phosphorylated residue exists were observed in the peak lists of the fragment ions.
Bioinformatics
We used the KAGIANA tool (http://pmnedo.kazusa.or.jp/kagiana/index.html) to extract cellular localization information of Arabidopsis proteins predicted by the WoLF PSORT program (Horton et al, 2007). For molecular function and biological process annotations extraction, the TAIR gene ontology annotation search tool (Berardini et al, 2004) was used.
For the homologs search, BlastP searches (Altschul et al, 1997) were performed against the protein databases, TAIR7_pep_2007425, rap1_all_orf_amino, and proteins.Poptr1_1.JamboreeModels for Arabidopsis, rice, and poplar, respectively (Ohyanagi et al, 2006; Tuskan et al, 2006). The E‐value cutoff of 10−3 was used for the initial search and if there were no protein hits, the cutoff value was lowered stepwise to 10−2 and 10−1. In some cases, E‐value cutoff of 10−6 was used for AT1G70520.1, E‐value cutoff of 10−5 was used for AT3G05140.1, and E‐value cutoff of 10−4 was used for AT2G30940.1. For multiple sequence alignment, ClustalW (Thompson et al, 1994) was performed with default parameter settings. The aligned sequences were further manually analyzed using the MEGA4 program (Tamura et al, 2007).
Pfam domain information was extracted from the database, TAIR7_all.domains (ftp://ftp.arabidopsis.org/home/tair/Proteins/Domains/).
Acknowledgements
We thank Sumiko Ohnuma (Keio University) for technical assistance, Alex Jones (The Sainsbury Laboratory, UK) for initial introduction of mass spectrometry at RIKEN, and Tetsuro Toyoda and Yoshiki Mochizuki (RIKEN) for kindly uploading the phosphorylation site data to a RIKEN OmicBrowse database. This study was supported by research funds from Yamagata Prefecture and Tsuruoka City to Keio University, MEXT Grants‐in‐Aid for Scientific Research (nos. 19678001 and 19039034 to KS and no. 19710169 to HN) and a grant from the Japan Society for the Promotion of Science to AD.
Supplementary Information
Supplementary Information [msb200832-sup-0001.doc]
Supplementary Figures [msb200832-sup-0002.pdf]
Supplementary Table 1 [msb200832-sup-0003.xls]
Supplementary Table 2 [msb200832-sup-0004.xls]
Supplementary data 1 [msb200832-sup-0005.pdf]
Supplementary data 2 [msb200832-sup-0006.doc]
Supplementary data 3 [msb200832-sup-0007.zip]
References
This is an open‐access article distributed under the terms of the Creative Commons Attribution License, which permits distribution, and reproduction in any medium, provided the original author and source are credited. This license does not permit commercial exploitation without specific permission.
- Copyright © 2008 EMBO and Nature Publishing Group
