ABA-Regulated G Protein Signaling in Arabidopsis Guard Cells: A Proteomic Perspective Zhixin Zhao,† Bruce A. Stanley,‡ Wei Zhang,†,§ and Sarah M. Assmann*,† Biology Department, 208 Mueller Laboratory, Penn State University, University Park, Pennsylvania 16802, and Section of Research Resources, The Pennsylvania State University College of Medicine, 500 University Drive, Hershey, Pennsylvania 17033 Received April 22, 2009
Signaling cascades mediated by heterotrimeric G proteins are ubiquitous and important signal transduction mechanisms in both metazoans and plants. In the model plant Arabidopsis thaliana, the sole canonical G protein R subunit, GPA1, has been implicated in multiple signaling events, including guard cell movement regulated by the plant stress hormone abscisic acid (ABA). However, only a handful of proteins have been demonstrated to be involved in GPA1 signaling to date. Here, we compared the proteome composition of guard cells from wild type Col vs gpa1-4 null mutants with and without ABA treatment using iTRAQ technology to identify guard cell proteins whose abundance was affected by ABA and/or GPA1. After imposition of strict selection criteria, the abundance of two proteins in Col and six proteins in gpa1-4 was found to be affected by ABA in guard cells, and 18 guard cell proteins were quantitatively affected by the mutation of GPA1. On the basis of known functions of the differentially expressed proteins, our data suggest that GPA1 inhibits guard cell photosynthesis and promotes the availability of reactive oxygen species (ROS) in guard cells. These results exemplify how iTRAQ can be used to quantitatively study single cell signaling pathways in Arabidopsis. Keywords: Arabidopsis guard cells • heterotrimeric G protein • abscisic acid (ABA) • iTRAQ
Introduction Abscisic acid (ABA), one of the major plant hormones, regulates many aspects of plant growth and development, including seed dormancy and germination, early seedling development and stomatal movements.1-3 Under stress conditions, especially under drought stress, accumulation of ABA in plants facilitates acclimation responses of plants to environmental changes.4 The global effect of ABA on the plant transcriptome has been studied extensively;5-7 however, it is widely recognized that changes in the transcriptome may only account for a portion of the changes occurring at the protein level, with either significant increases in protein amounts or activity occurring with no changes in cognate transcripts or, conversely, no changes in protein amounts or activity occurring in the face of significant changes in cognate transcript levels.8,9 Here we studied global changes in protein abundance induced by ABA in a single cell type, guard cells, in both wild type Col and a mutant lacking the GR subunit GPA1. Heterotrimeric G proteins, composed of the three different subunits R, β, and γ, are major components of signal transduction pathways in both mammalian and plant systems. In the canonical model in mammalian cells, GR couples with * To whom correspondence should be addressed. Prof. Sarah M. Assmann. Email:
[email protected]. Phone: 814-863-9579. FAX: 814-865-9131. † Penn State University. ‡ The Pennsylvania State University College of Medicine. § Current address: School of Life Science, Shandong University, Jinan, 250100, Shandong Province, P.R. China. 10.1021/pr901011h
2010 American Chemical Society
guanosine diphosphate (GDP) and associates with the Gβγ dimer when inactive. When activated by G protein coupled receptors (GPCRs), GR dissociates from the Gβγ dimer and GDP is replaced by guanosine triphosphate (GTP). Upon dissociation, either GR or the Gβγ dimer can act as a functional unit and induce downstream signaling. In contrast to mammalian cells, where multiple R, β, and γ genes exist, there is only one prototypical GR (GPA1), one Gβ (AGB1), and two known Gγ (AGG1 and AGG2) genes in Arabidopsis.10,11 Despite the comparative simplicity of players, G proteins have been shown to participate in multiple signaling pathways in Arabidopsis, including many that are regulated by ABA, notably seed germination and early seedling development,12,13 guard cell signaling,14,15 and oxidative stress responses.16,17 Mutants lacking the GR subunit show phenotypes in multiple aspects of guard cell function, including hyposensitivity to ABAinhibition of inward K+ channels and stomatal opening,14,15 hyposensitivity to ABA-activation of anion channels through a pH-independent pathway,15 hyposensitivity to sphingosine-1phosphate (S1P)-regulation of stomatal movements and ion channel activities18 and hyposensitivity to phyto-S1P-regulation of stomatal movements.19 Since we are particularly interested in G-protein signaling in guard cells, in this study, we used a null mutant lacking the GR subunit in the Col background, gpa1-4, to study proteins whose abundance is affected by loss of GPA1 in Arabidopsis guard cells. To date, only one proteomic study of heterotrimeric G protein R subunit signaling in plants has been published.20 By Journal of Proteome Research 2010, 9, 1637–1647 1637 Published on Web 02/19/2010
research articles image analysis of Coomassie-stained gels followed by classic protein sequencing to identify relevant proteins, protein amounts of six rice embryo globulin-2 proteins and a receptor for activated C-kinase (RACK) were shown to be decreased in rice embryos from the rice GR mutant d1 as compared to wild type. Conversely, RACK protein abundance, relative to wild-type, increased in rice embryos of d1 mutants complemented with a constitutively active QL form of GR transgenic rice seed embryos.20 Upon seed imbibition, the protein level of RACK was increased in wild type if ABA was present, but ABA did not have this effect in d1 embryos or in QL/d1 embryos. These results led the authors to propose that RACK might be upregulated by ABA signaling through GR activation during rice seed germination.20 This paper provides support for the use of proteomics methods to study GR function in plants. The strength of comparative proteomics lies in its ability to reveal quantitative protein changes between samples on a large scale. Isotope tags for relative and absolute quantification (iTRAQ) is a gel-free method for quantitative comparisons of protein samples. The iTRAQ reagent (Applied Biosystems) labels lysine residues and the N termini of peptides with specific reporter moieties for each of four samples, which can then be combined for simultaneous determination of protein amounts from each of the multiple samples. Upon fragmentation of the combined labeled peptides by collision-induced dissociation to produce MS/MS spectra, the relative proportion of each peptide arising from each sample is correlated to the relative intensity of the reporter ion peaks, and thus to the amount of the protein from which it was derived in each sample.21 To date only a few papers have reported using iTRAQ to study protein abundance in Arabidopsis.22-25 The main objective of this study was to discover candidate proteins involved in ABA and/or GPA1 signaling in guard cells, using iTRAQ technology. We first showed that iTRAQ is efficient in identifying Arabidopsis leaf proteins. Since it was shown previously that GPA1 is required for proper execution of ABAsignaling pathways in guard cells, we then compared protein abundances from wild type Col and gpa1-4 guard cell protoplasts (GCPs) with and without 50 µM ABA treatments. ABA treatment significantly affected the amounts of two proteins in Col guard cells while affecting six proteins in gpa1-4. Eighteen proteins were significantly affected in abundance in gpa1-4 guard cells compared to Col. ABA and GPA1 signaling models are proposed on the basis of our iTRAQ study. The reproducibility of the iTRAQ technology and the fold changes identified by the iTRAQ technology in plant iTRAQ studies to date are also discussed.
Materials and Methods Guard Cell Protoplast Preparation and Protein Extraction. Plant growth conditions and GCP isolation procedures were the same as described by Zhao et al.26 For control or ABA treated GCPs, 50 µM (final concentration) ABA or equivalent volume of ethanol was added for four hours during the enzyme steps in cell wall digestion for GCP preparation.27 Given that ABA-induced changes in transcript abundance have been observed within 15 min after ABA application to guard cells,28 and that ABA stimulates subnuclear organization within guard cells by 20 min,29,30 we surmised that four hours of ABA treatment would be sufficient to elicit changes in protein abundance. The concentration of ABA used was chosen to match that employed in other cell signaling studies on guard 1638
Journal of Proteome Research • Vol. 9, No. 4, 2010
Zhao et al. 15,31
cells and is within the range of concentrations commonly employed for guard cell studies.5,32 About four million GCPs from ∼400 leaves were used per genotype per treatment for protein extraction in each replicate. In total, ∼48 million GCPs from ∼4800 leaves were used. Proteins were extracted from GCPs according to Zhao et al.26 Trypsin Digestion and iTRAQ Labeling. One hundred micrograms of proteins from each sample were dissolved, reduced, alkylated and trypsin digested according to the protocol provided with the iTRAQ kit (Applied Biosystems, part #: 4374321), with the exceptions that iodoacetamide rather than MMTS was used as an alkylating agent, and the trypsin used for the digestion steps was Promega Sequencing grade Trypsin. Peptides in each tube were labeled by one iTRAQ reagent as follows: for the first and second replicates, peptides from Col GCPs without ABA were labeled by reagent 114; peptides from Col GCPs with 50 µM ABA treatment were labeled by reagent 115; peptides from gpa1-4 GCPs without ABA were labeled by reagent 116 and peptides from gpa1-4 GCPs with 50 µM ABA treatment were labeled with reagent 117; for the third replicate, peptides from gpa1-4 GCPs without ABA were labeled by reagent 114 and peptides from gpa1-4 GCPs with 50 µM ABA treatment were labeled with reagent 115; peptides from Col GCPs without ABA were labeled by reagent 116; peptides from Col GCPs with 50 µM ABA treatment were labeled by reagent 117. All labeling steps were performed according to the protocol provided by Applied Biosystems. Before peptides from the four samples were pooled together, the labeling reaction was quenched by adding 100 µL distilled water, and then incubated at room temperature for 30 min. Three biological replicates were performed and compared. MS Identification and Data Analysis. All peptides were separated through two sequential columns: strong cation exchange (SCX) and C18 nanoflow chromatography, and spotted onto MALDI plates. As soon as each stainless steel MALDI target plate was placed in the 4800 MALDI TOF-TOF mass spectrometer, MS spectra were acquired by accumulating 500 laser shots per spot at a laser power of 3800, and then MS/MS spectra were taken in an immediately subsequent pass over each MALDI target plate using a plate-wide interpretation method, essentially a plate-wide dynamic exclusion method whereby only the spot containing the best representative of each observed unique m/z species across the entire target plate was chosen for subsequent MS/MS analysis of a particular peak m/z. These data-dependent MS/MS spectra were acquired by accumulation of 2000 spectra with a laser power setting of 4600, using air as a CID gas at approximately 4 × 10-6 Torr and Metastable Suppression ON. Raw MS and MS/MS spectra were analyzed by 4000 Series Explorer Software version 3.6, build 999, from Applied Biosystems/MDS Sciex. No smoothing or background subtraction was used on the raw data, but a two-pass signal-to-noise filter was applied to select peaks. The first pass defined any peaks as real if they had a signal-to-noise ratio of 3 or higher, using a local noise window width of 250 m/z, in order to see as many isotope peaks as possible and not miss possible monoisotopic peaks. An ion cluster S/N optimization filter was then applied, where only the monoisoptopic peak of ion clusters having a signalto-noise ratio of at least 10 were accepted as peaks. The peaks thus generated were searched against a constructed FASTA Arabidopsis-specific database consisting of all the Arabidopsis entries contained in the NCBInr database on July 23, 2008 (138 418 protein sequences) concatenated to a FASTA list of
ABA-Regulated G Protein Signaling in Arabidopsis Guard Cells common potential contaminants such as trypsin, keratins, and BSA. One missed cut of trypsin cleavage was allowed, and all standard post-translational modifications (PTMs), as recognized by ProteinPilot, were allowed. The standard set of “Variable” PTMs that ProteinPilot allows includes 51 PTMs that may occur in the course of sample preparation, for example, oxidation of methionine, plus any specific anticipated PTMlike side reactions resulting from the Cys-alkylation procedures used. We also searched using the “Biological Modifications” classification of ProteinPilot, which adds another 126 possible PTMs, although they are not all given the same weight or likelihood, that is, if some “likely” PTMs are predicted by the sequence tag obtained and the mass of the parent ion, they are weighted more than “less likely” PTMs by the ProteinPilot algorithm. PSPEP software33 integrated with ProteinPilot software (Applied Biosystems) was used to concatenate a decoy database consisting of the reverse sequences of everything in the “normal arabidopsis + contaminants” database, and the MS/MS data were searched against this normal + decoy database using the Paragon algorithm34 contained in the ProteinPilot software package, using settings of trypsin digestion, Urea-modification, Thorough Search, and Biological Modifications. Possible multiple identifications of homologous proteins were collapsed into the minimum set of proteins accounting for all the ms/ms spectral data using the ProGroup algorithm in ProteinPilot. Finally, the local False Discovery Rate (FDR; the probability that any identified protein is a false positive) was calculated by the PSPEP algorithm, and only protein identifications that had a ProteinPilot confidence score of 95% or better (Unused Score above 1.3) AND had an estimated Local FDR of 5% or lower were accepted. (Note that this criterion of only accepting protein IDs with an estimated local FDR of 5% or lower is much more stringent than an estimated Global FDR of 5% or even in many cases 1% or 2%ssince an estimated Global FDR of 5% applies to a whole set of IDs, many of the lower-ranked proteins on that list may actually have much greater probabilities of being false positives than the overall rate of 5%; however, the estimated Local FDR of 5% means that the lowest ranked protein on that list itself has an estimated probability of being a false positive of only 5%). Relative quantitative analysis of proteins between samples based on ratios of iTRAQ reporter ions from all unique peptides belonging to that protein was also analyzed by ProteinPilot, using only ratios from peptides which could be uniquely assigned to that protein without being a sequence shared with other identified homologous proteins. The relative quantification of proteins is the ratio of peak areas at 114, 115, 116, and 117, and proteins were only accepted that had relative quantification p-values less than 0.05. Identified proteins with quantification p-values 1.17 or 1.17 or
