Proteomic Analysis of Extracellular ATP-Regulated Proteins Identifies ATP Synthase β-Subunit as a Novel Plant Cell Death Regulator*

Extracellular ATP is an important signal molecule required to cue plant growth and developmental programs, interactions with other organisms, and responses to environmental stimuli. The molecular targets mediating the physiological effects of extracellular ATP in plants have not yet been identified. We developed a well characterized experimental system that depletes Arabidopsis cell suspension culture extracellular ATP via treatment with the cell death-inducing mycotoxin fumonisin B1. This provided a platform for protein profile comparison between extracellular ATP-depleted cells and fumonisin B1-treated cells replenished with exogenous ATP, thus enabling the identification of proteins regulated by extracellular ATP signaling. Using two-dimensional difference in-gel electrophoresis and matrix-assisted laser desorption-time of flight MS analysis of microsomal membrane and total soluble protein fractions, we identified 26 distinct proteins whose gene expression is controlled by the level of extracellular ATP. An additional 48 proteins that responded to fumonisin B1 were unaffected by extracellular ATP levels, confirming that this mycotoxin has physiological effects on Arabidopsis that are independent of its ability to trigger extracellular ATP depletion. Molecular chaperones, cellular redox control enzymes, glycolytic enzymes, and components of the cellular protein degradation machinery were among the extracellular ATP-responsive proteins. A major category of proteins highly regulated by extracellular ATP were components of ATP metabolism enzymes. We selected one of these, the mitochondrial ATP synthase β-subunit, for further analysis using reverse genetics. Plants in which the gene for this protein was knocked out by insertion of a transfer-DNA sequence became resistant to fumonisin B1-induced cell death. Therefore, in addition to its function in mitochondrial oxidative phosphorylation, our study defines a new role for ATP synthase β-subunit as a pro-cell death protein. More significantly, this protein is a novel target for extracellular ATP in its function as a key negative regulator of plant cell death.

ATP is a ubiquitous, energy-rich molecule of fundamental importance in living organisms. It is a key substrate and vital cofactor in many biochemical reactions and is thus conserved by all cells. However, in addition to its localization and functions inside cells, ATP is actively secreted to the extracellular matrix where it forms a halo around the external cell surface. The existence of this extracellular ATP (eATP) 1 has been reported in several organisms including bacteria (1), primitive eukaryotes (2), animals (3), and plants (4 -6). This eATP is not wasted, but harnessed at the cell surface as a potent signaling molecule enabling cells to communicate with their neighbors and regulate crucial growth and developmental processes.
In animals, eATP is a crucial signal molecule in several physiological processes such as neurotransmission (7,8), regulation of blood pressure (9), enhanced production of reactive oxygen species (ROS) (10), protein translocation (11), and apoptosis (12). Extracellular ATP signal perception at the animal cell surface is mediated by P2X and P2Y receptors, which bind ATP extracellularly and recruit intracellular second messengers (13,14). P2X receptors are ligand-gated ion channels that provide extracellular Ca 2ϩ a corridor for cell entry after binding eATP, facilitating a surge in cytosolic [Ca 2ϩ ] that is essential in activating down-stream signaling. P2Y receptors transduce the eATP signal by marshalling heteromeric G-proteins on the cytosolic face of the plasma membrane and activating appropriate downstream effectors.
Although eATP exists in plants, homologous P2X/P2Y receptors for eATP signal perception have not yet been identified, even in plant species with fully sequenced genomes. Notwithstanding the obscurity of plant eATP signal sensors, some of the key downstream messengers recruited by eATPmediated signaling are known. For example, eATP triggers a surge in cytosolic Ca 2ϩ concentration (15)(16)(17) and a heightened production of nitric oxide (18 -20) and reactive oxygen species (17,21,22). Altering eATP levels is attended by activation of plant gene expression (16,21) and changes in protein abundance (5,23), indicating that eATP-mediated signaling impacts on plant physiology. Indeed eATP has been demonstrated to regulate plant growth (20, 24 -26), gravitropic responses (27), xenobiotic resistance (4), plant-symbiont interactions (28), and plant-pathogen interactions (23,29). However, the mechanism by which eATP regulates these processes remains unclear, largely because the eATP signal sensors and downstream signal regulatory genes and proteins have not been identified.
We previously reported that eATP plays a central regulatory role in plant cell death processes (5). Therefore, an understanding of the signaling components galvanized by eATP in cell death regulation might serve a useful purpose in providing mechanistic detail of how eATP signals in plant physiological processes. We found that eATP-mediated signaling negatively regulates cell death as its removal by application of ATP-degrading enzymes to the apoplast activates plant cell death (5). Remarkably, fumonisin B1 (FB1), a pathogen-derived molecule that activates defense gene expression in Arabidopsis (30), commandeers this eATP-regulated signaling to trigger programmed cell death (5). FB1 is a mycotoxin secreted by fungi in the genus Fusarium and initiates programmed cell death in both animal and plant cells (31,32). In Arabidopsis, FB1 inaugurates cell death by inactivating eATPmediated signaling via triggering a drastic collapse in the levels of eATP (5). FB1-induced Arabidopsis programmed cell death is dependent on the plant signaling hormone salicylic acid (33), which is a key regulator of eATP levels (29). Because concurrent application of FB1 and exogenous ATP to remedy the FB1-induced eATP deficit blocks death, FB1 and exogenous ATP treatments can therefore be used as probes to identify the key signal regulators downstream of eATP in cell death control. This is vital for achieving the global objective of elucidating the mechanism of eATP signaling in plant physiology.
Gel-based proteomic analyses have been previously applied to successfully identify the novel role of eATP in the regulation of plant defense gene expression and disease resistance (23,29). We have now employed FB1 and ATP treatments together with two-dimensional difference in-gel electrophoresis (DIGE) and matrix-assisted laser desorptiontime of flight MS (MALDI-TOF MS) to identify the changes in Arabidopsis protein profiles associated with a shift from normal to cell death-inception metabolism. Additional reverse genetic analyses enabled us to definitively identify a putative ATP synthase ␤-subunit as a target for eATP-mediated signaling with an unexpected function in the regulation of plant programmed cell death.

Plant Material and Growth Conditions-Cell suspension cultures of
Arabidopsis thaliana var. Landsberg erecta (34) were grown at 22°C under a 16 h photoperiod (100 mol/m 2 /s) regimen. Cell cultures were used for experiments in mid-exponential growth phase (3-4 days postsubculturing). Soil grown-plants were incubated in a growth chamber with a 16 h photoperiod (100 -120 mol.mϪ 2 .s Ϫ1 ) maintained at 20°C during the light phase and 15°C during the dark phase. Plants were used for experiments 4 -5 weeks following sowing.
Cell Culture Treatment-Stock solutions of FB1 (Sigma, Haverhill, UK) were prepared in 70% methanol and filter-sterilized before using to treat cell cultures. Mock treatments were performed with an equivalent dilution of 70% methanol. Stocks of filter-sterilized 100 mM ATP, pH 6.5 (adjusted with KOH) were prepared fresh every time. Samples for proteomic analysis were prepared by treating 3 days old Arabidopsis cell cultures adjusted to a cell density of 5% (w/v) in a 100 ml culture volume. One set of cultures (FB1) was treated with 1 M FB1 at the beginning of the experiment, whereas the second set of cultures (FB1ϩATP) was similarly treated with FB1, but 1 mM ATP was added 40 h later. The controls were mock-treated with methanol, the carrier solution for FB1. Each treatment had 4 independent biological replicates. The cells were harvested 48 h following the start of the experiment and frozen in liquid nitrogen.
Protein Sample Preparation-Cells were pulverised at 4°C in a homogenization buffer (1 mM EDTA/10 mM Tris-HCl, pH 8.0) using a French Press (Constant systems Ltd., Warwick, UK) and the homogenate centrifuged (20,000 ϫ g, 30 min, 4°C). The supernatant, containing microsomal membranes and the total soluble protein fraction, was centrifuged (100,000 ϫ g, 1 h, 4°C) to separate these fractions. Total soluble protein (TSP) was recovered from the supernatant by precipitation (80% [v/v] acetone, Ϫ20°C, 12 h) and extracted from the precipitate with a protein solubilization solution (9 M urea, 2 M thiourea, 4%[w/v] CHAPS). The microsomal membrane pellets were washed three times with homogenization buffer and protein extracted using the protein solubilization solution.
Protein Labeling and Gel Electrophoresis-Protein samples from four biological replicates of each treatment were labeled with CyDyes (GE Healthcare, Buckinghamshire, UK) as described before (35). Each biological replicate sample was split into two and one labeled with Cy3 and the other with Cy5 in a dye swap experimental design to preclude dye-specific artifacts (36). The pooled standard consisting of equal amounts of all the samples was labeled with Cy2. Protein mixtures containing 12.5 g each of Cy3-and Cy5-labeled samples plus the Cy2-labeled pooled standard were resolved in 24 cm linear gradient pI 4 -7 IPG strips for 70kVh as previously described (35). Second dimension separation was performed on an Ettan DALT Twelve System (GE Healthcare, Buckinghamshire, UK) in custom made 10 -20% gradient polyacrylamide gels cast using the automated 2DE Optimizer gel caster (NextGen sciences, Cambridge, UK). Following resolution, the gels were immediately scanned using a Typhoon 9400 variable mode imager (GE Healthcare, Buckinghamshire, UK).
Image Analysis-Gel images were analyzed using DeCyder Differential Analysis Software version 6.5 (GE Healthcare, Buckinghamshire, UK) as described previously (35). Briefly, four biological and two technical replicates (from the dye-swap) comprised a total of 8 images per treatment. Spot detection, quantification and normalization of spot volume against the internal standard were performed automatically using the Differential In-Gel analysis module with estimated number of spots set to 5000. The Biological Variation Analysis module was used for spot matching and differential protein analysis. Matched spots were manually checked between gels to minimize false spot matching and exclude spot artifacts. The software automatically generated a ratio of sample spot volume to the pooled standard spot volume and normalized these ratios across all gels to generate standardized spot abundance values. The standardized spot abundance values from the four biological replicates and two technical replicates were then averaged and the means subjected to Student's t test to check for statistically significant differences. Only manually inspected spots present in all replicate gels and displaying a significant (p Յ 0.05) FB1-induced change in abundance of a minimum of 20% (TSP) or 50% (microsomal protein) were selected for further analyses. The response of these spots to ATP added after FB1 was analyzed by statistical comparison of FB1ϩATP average with FB1 only average. Only spots with an average standardized abundance that significantly (p Յ 0.05) shifted from the FB1 treatment average in response to the FB1ϩATP treatment were selected for further analyses as ATP-responsive proteins. Probability values associated with the comparison of means were calculated using Student's t test.
MALDI-TOf MS Protein Identification-Preparative gels for protein identification loaded with 200 g of unlabeled protein were stained with Sypro Ruby TM total protein stain (Genomic Solutions, Huntington, UK) according to the manufacturer's instructions. Differentially expressed spots were robotically excised using a ProPick Work station (Genomic Solutions, Huntington, UK) for identification by MALDI-TOF using a Voyager DE-STR Biospectrometry work station (Applied Biosystems, Warrington, Cheshire, UK) as described previously (37). Briefly, spot plugs of 2.0 mm diameter were digested with modified trypsin (Promega, Madison, WI) in a ProGest work station (Genomic Solutions, Huntington, UK) using the standard overnight digestion protocol supplied with the instrument. A total of 0.5 l of each digest was spotted together with 0.5 l of a saturated solution of ␣-cyano-4-hydroxycinnamic acid matrix directly onto a MALDI target plate using an Applied Biosystems Symbiot robot (Boston, MA). Instrument calibration was carried out for each sample using PE Sequazyme calibration mixture I (containing des-Arg-bradykinin, angiotensin I, Glu-fibrinopeptide B, and neurotensin). After each spot's spectra were acquired, automated peak detection, peak de-isotoping and noise reduction was carried out using Applied Biosystems Data Explorer 2.1.0 to generate peak mass tables. Trypsin peptide peaks (842.5 ϩ 2211.1) were used as internal calibrants and were excluded from database searches. The peptide mass fingerprint for each spot was used to search Viridiplantae (green plants) sequences in the nonredundant NCBInr database version 20070713 (5269953 sequences; 1825351362 residues) using the Mascot 4.0 search engine (Matrix Science, London, UK). The following search parameters were used: peptide mass tolerance: Ϯ 50 ppm, maximum number of missed cleavages: 1, fixed modifications: carbamidomethylation of cysteine residues, variable modifications: oxidation of methionine residues. The Mascot software probability-based MOWSE score cutoff for a significant (p Յ 0.05) positive protein identification of 71 was applied. Where more than one database entry was obtained from a single spot, the spots were excluded because it was impossible to know which of the proteins in the mixture were differentially regulated.
Cell Death Assays-Discs of 8 mm diameter were cored from leaves of 4-week-old plants and floated on 10 M FB1 solutions in triplicate Petri-dishes. Each replicate had 10 leaf discs each originating from one of 10 replicate plants. The discs were incubated in the dark for 48 h to allow uptake of FB1 prior to the onset of cell death. After the dark incubation, the discs were placed under a 16 h photoperiod regime and the conductivity of the underlying solution measured at 24 h intervals using a Jenway conductivity meter (Jenway Ltd., Felsted, UK). In addition, 10 M FB1 was also infiltrated into the apoplast of attached leaves from the abaxial surface using a syringe without a needle. Symptom development was visually monitored and photographs taken 4 days following infiltration.

RESULTS
Establishing the Experimental System-To identify eATPregulated proteins with a putative function in cell death regulation, we used Arabidopsis cell cultures treated with FB1 or FB1ϩATP. FB1 inactivates eATP-mediated signaling by triggering removal of the input signal (eATP), thereby initiating cell death (5). Exogenous ATP supplied back to the FB1-treated cultures rescues the cells from death, most probably by reestablishing the eATP-mediated signaling. Thus, exogenous ATP treatment can be used as a filter to identify the subset of FB1-induced genes/proteins whose expression and abundance is altered in response to the specific depletion of eATP. We used the cell death marker SEN1 (At4g35770), a gene activated during senescence-associated programmed cell death (38), to validate the utility of the exogenous ATP filter in this experimental system. FB1 treatment up-regulated SEN1 expression, but, in accordance with its ability to blockade FB1-induced cell death (5), exogenous ATP attenuated the response of this gene to FB1 treatment (Fig. 1). Because exogenous ATP effectively blocked FB1 effects at the transcript level, it is most likely that this is reflected at the protein level as well. Therefore, a comparative analysis of protein profiles of mock-treated cultures with FB1-and FB1ϩATPtreated cultures can reveal eATP-regulated proteins that mediate its physiological effects, including cell death control.
We have previously shown that exogenous ATP rescues cells from death if added concurrently with FB1 or any time up to ϳ40 h later, but fails to rescue the cells if added ϳ48 h or thereafter (5). Although 40 h coincides with the time when eATP levels have diminished below detection (5), these findings indicate that the cells' metabolism irreversibly commits to cell death at ϳ48 h, with cell death symptoms appearing ϳ24 h later (5). Therefore, we chose to perform proteomic analyses at the 48 h time point in order to detect the changes that switch normal metabolism to a death program, but that precede actual cell death. The timing of events triggered by FB1 addition to Arabidopsis cell cultures is illustrated schematically in Fig. 1.
Protein Gel Analysis and Protein Identification-Fractions enriched for total soluble protein (TSP) and microsomal membrane protein were prepared from Arabidopsis cell cultures exposed to FB1 or a combination of FB1ϩATP treatments. In the latter, ATP was added to the cell cultures 40 h after FB1. Treated cells were harvested for protein extraction at the 48 h time point. Images of the protein gels revealed big differences in the profiles of TSP and microsomal membrane proteins ( Fig. 2), indicating that analysis of the two separate fractions enabled the coverage of a wider range of proteins than would be achievable if only one of the protein fractions was targeted for analysis. As we used a gel-based approach, it is obvious that only a fraction of the membrane-associated proteins are represented in this study because highly hydrophobic proteins do not easily enter two-dimensional gels. However, this inadvertently helped to simplify the protein profile by reducing the number of proteins.
Quantitative analysis was performed using two-dimensional DIGE on four independent biological replicates of each treatment and two technical replicates of each sample. Average standardized spot volumes of the control and FB1 treatments were compared using Student's t test and the fold-change in protein abundance calculated by generating the ratio [FB1/ control] for up-regulated spots or [control/FB1] for downregulated spots (Tables I and II). To quantify the effects of ATP on FB1-induced changes in protein abundance, average spot volumes of the FB1 and FB1ϩATP treatments were compared using the Student's t test and the ratio [FB1ϩATP/FB1] or [FB1/FB1ϩATP] generated for up-regulated or down-regulated spots, respectively (Tables I and II). The ratio of downregulated spots is indicated by a minus sign in both Tables I  and II. Approximately 5000 features, including both authentic protein spots and some artifacts, were automatically detected on the pooled standard master gels. The majority of protein spots in both the TSP and microsomal protein fractions did not respond to FB1 treatment. The abundance of 145 protein spots in the TSP fraction was significantly (p Յ 0.05) altered in response to FB1 treatment by at least 20%. Of these 145 TSP spots, only 75 were positively identified ( Table I)  low abundance spots, from which inadequate sample was present in the preparative gel to enable identification, whereas a few were in protein mixtures. Spots with protein mixtures were excluded because it was not possible to determine which of the proteins was changing in abundance in response to the treatments. A total of 83 protein spots significantly (p Յ 0.05) responded to FB1 by at least a 50% change in abundance in the microsomal membrane fraction. Only 57 of these were positively identified (Table II) whereas identification of the rest was similarly hindered by low abundance or existence as multiple protein mixtures on the gels. The total number of positively identified protein spots in both the TSP and microsomal membrane fractions was 132 (Tables I and II), but this represented 74 unique proteins. The redundancy revealed by the inequality between the number of protein spots and unique proteins arises from the existence of post-translationally modified polypeptides and an overlap of 13 proteins that were identified in both protein fractions. Additional data relating to protein identification is presented in supplemental Table 1 and supplemental Protein Mass Spectra. The exogenous ATP filter revealed the subset of FB1-responsive proteins, which is regulated by the level of eATP. The response of 24 TSP spots and 16 microsomal membrane protein spots to FB1 was attenuated by exogenous ATP (Fig.  2C profile ii). These are the proteins most likely to mediate the physiological effects of eATP. A total of 26 unique proteins were in this category. However, the majority of the FB1responsive proteins remained unaffected by exogenous ATP (Fig. 2C profile i), indicating that FB1 has other targets and physiological effects that are independent of its ability to trigger eATP depletion. A minority of the spots (eight spots) had their response to FB1 enhanced by the addition of exogenous ATP (Fig. 2C profile iii). This result suggests that FB1 and exogenous ATP activate common, but as yet uncharacterized, plant signaling cascades where a combination of the two compounds has a synergistic effect.
Classification of Differentially Expressed Proteins-The identified proteins were classified into several functional categories (Tables I and II). Molecular chaperones and heat shock proteins were highly represented in the data sets and they were largely up-regulated in response to FB1. Remarkably, exogenous ATP attenuated this heightened increase in molecular chaperones, indicating that FB1-induced depletion of eATP is the central cue for the deployment of this protein response. Many subunits of the ATP synthesis machinery were down-regulated by FB1 treatment, but again this response was dependent on the level of eATP because exogenous ATP impeded this response. Although the majority of spots in this category were from the mitochondrial ATP synthase, vacuolar and chloroplastic ATP synthase subunits were also identified. Several cellular redox control proteins were suppressed in response to FB1. These included a glutathione S-transferase, dehydroascorbate reductase, and thioredoxin. Curiously, exogenous ATP enhanced the suppression of these proteins, indicating that eATP depletion per ser is not the cue for this response. Central metabolic pathways, such as glycolysis and amino acid synthesis were also affected by  FB1 treatment. Other functional categories of FB1-responsive proteins included members of the cell's protein degradation machinery, cytoskeleton-associated proteins with a structural role, and various enzymes involved in several other cellular and metabolic processes.
ATP Synthase ␤-subunit is a Cell Death Regulator-After identifying eATP-regulated proteins using FB1 treatments and the exogenous ATP filter, we initiated a systematic investigation of their possible role in plant cell death control. From the list of 28 proteins identified in this category, we selected candidates for which Arabidopsis transfer-DNA (T-DNA) gene knockout mutants from the SALK T-DNA collection (39) existed. We searched the Arabidopsis genome database for gene family members and relegated to the bottom of the list candidates from large gene families, whose function is unlikely to be obtained via reverse genetic studies owing to the high likelihood of gene redundancy. The response of the knockout mutants to FB1 treatment was examined using qualitative and quantitative cell death assays and any mutant with cell death kinetics and phenotype different from wildtype plants clearly defined novel cell death signal regulatory proteins under eATP control.
At the top of our candidate protein list was ATP synthase ␤-subunit (AT5G08690), a dominant microsomal membrane fraction eATP-regulated protein found in spots 30, 32, 33, and 35-39 ( Fig. 2B; Table II). Although seven of the spots appeared as a charge train of ϳ55 kDa, one spot (spot 39) had a lower molecular weight of ϳ49 kDa (Fig. 2B). The same ATP synthase ␤-subunit protein was identified in the TSP fraction as a ϳ38 kDa spot (spot 33) (Fig. 2A). Distribution of the identified peptides from the two low molecular weight species was predominantly in the central region of the primary sequence of ATP synthase ␤-subunit (Fig. 3), suggesting the possibility that they arise from N-and/or C-terminal cleavage of the ϳ55 kDa protein.
We obtained three independent homozygous gene knockout mutants of ATP synthase ␤-subunit (SALK_005252, SALK_024990, and SALK_135351) with a single T-DNA inserted into predicted exonal regions of the gene (Fig. 4A). Gene knockout status was confirmed by RT-PCR. Primers designed to straddle the insertion sites successfully amplified the expected 1795 bp gene-specific product in wild-type Columbia-0 plants whereas the absence of this product in all knockout lines confirmed the lack of a functional copy of the gene (Fig. 4B). Next we established the cell death kinetics of leaf tissue obtained from the knockout plant lines against wild-type plant tissues. The assay involves floating leaf discs on FB1 solutions and measuring the conductivity of the solution, which rises as dying cells release their contents. Measured over time, the conductivity profile reflects the relative rates and extent of FB1-induced cell death. The rate and extent of cell death were significantly diminished in all three knockout lines (Fig. 4C, 4E), indicating that they were resistant to FB1. Leaf discs from the ion leakage assay photographed 96 h post-treatment revealed advanced stages of chlorosis, that accompany cell death, in wild-type tissues (Fig. 4F). Corresponding tissues from SALK_135351 and SALK_005252 did not have these symptoms (Fig. 4F), displaying the resistance engendered by disruption ATP synthase ␤-chain subunit gene expression. Moreover, cell death symptom development was also suppressed in the SALK_024990 knockout line when FB1 was infiltrated into leaves left attached to growing plants (Fig. 4D). Overall, these results demonstrate that the ATP synthase ␤-subunit is a novel cell death regulator identified via proteomics.

DISCUSSION
Rationale of using FB1 and an ATP Filter to Identify eATPregulated Proteins-In order to identify eATP effectors and target proteins, we sought to use an experimental system in which endogenous eATP is depleted so as to trigger changes in the abundance of proteins whose gene expression is tightly regulated by the presence of eATP. This could easily be achieved using enzymatic eATP-sequestering systems such as exogenous apyrase or a glucose/hexokinase mixture. However, we chose to use FB1 treatments, which have been shown to activate progressive depletion of eATP prior to the onset of cell death (5). The advantage of this system is twofold. First, eATP diminishes gradually, which enables the reverse treatment of replenishing eATP in FB1-treated cultures by addition of exogenous ATP. Such a treatment would filter out proteins responding to FB1 only from those responding to FB1-induced eATP removal. Second, FB1 is linked to a physiologically relevant process because it is a pathogen-derived FIG. 3. ATP synthase ␤-subunit sequence and sequence coverage. A, sequence coverage by peptides identified from ATP synthase ␤-subunit in spot 33 (shown in Fig. 2A) of the TSP fraction. B, sequence coverage by peptides identified from ATP synthase ␤-subunit in spot 39 (shown in Fig. 2B) of the microsomal protein fraction. Residues in bold constitute the trypsin-generated peptides identified by mass spectrometry. Underlined residues form the cleavable predicted mitochondrial targeting sequence.
toxin whose use could result in a better understanding of events surrounding cell death processes triggered by certain plant pathogens. Presently it is not clear how endogenous eATP is depleted, but using FB1 can provide clues that may lead to discovery of the mechanism by which pathogens deplete eATP. Because cell cultures treated with FB1 are primed for eATP depletion, we used an excessive concentration of exogenous ATP, as high as 1 mM, to ensure that the FB1-induced eATP deficit was cancelled for the duration of the treatment.
Proteomic Changes Underlying the Switch to FB1-induced Cell Death-There was a clear up-surge in the abundance of molecular chaperones in response to FB1 treatment (Table I  and II). This group of proteins is important for maintaining cellular homeostasis by regulating protein folding, assembly, translocation, and degradation (40). Heightened expression of chaperone genes during stress is designed to re-establish normal protein conformation, thereby mitigating the adverse effects of stress (41). Because FB1 activates programmed cell death preceded by a shift in cellular metabolism underpinned by drastic changes in global gene expression, it is not surprising that chaperone proteins are deployed to accommodate the accompanying increased traffic through protein synthesis, translocation and degradation pathways. Not only does FB1 activate cell death processes, but also triggers pathogen defense systems that include synthesis of an array of pathogenesis-related proteins (30), most of which transit through the endoplasmic reticulum and increase the demand for chaperones. Similar chaperone increases occur in Arabidopsis responding to programmed cell death-eliciting patho- Inverted triangles indicate insertion sites and gray boxes represent exons. B, RT-PCR amplification of ATP ␤-synthase subunit (ATPase-␤) in RNA samples derived from Columbia-0 (Col-0) plants and the three independent T-DNA insertion mutants. Actin-2 (ACT2) was used as a constitutive reference control. C, cell death profiles of leaf disc tissues. Leaf discs were treated with FB1 and incubated in the dark for 48 h and then exposed to light. The conductivity of FB1 solutions on which discs were floating was measured at 48 h and every 24 h thereafter. Values and error bars represent means Ϯ se (n ϭ 3). An asterisk indicates data points at which the mutants are significantly (p Ͻ 0.05) lower that the wildtype (Col-0). D, appearance of Col-0 and SALK_024990 leaves 4 days after infiltration with FB1. E, leaf disc cell death assay with SALK_135351 and SALK_005252 as described in (C). F, leaf discs treated as described in (E) and photographed at 96 h. gens (42) or exogenous plant hormones (43,44), which activate enormous changes in the transcriptome. Remarkably, exogenous ATP blocked the rise in chaperone levels, indicating that this response was triggered by the cell death signal depletion of eATP.
Our study has revealed that a prime target for FB1 is cellular ATP synthesis. We identified subunits of the vacuolar and chloroplastic ATP synthase proteins as responsive to FB1, with the great majority belonging to the mitochondrial F 1 F 0 -ATP synthase machinery. The mitochondrial F 1 F 0 -ATP synthase has two main parts; the hydrophilic catalytic F 1 complex and the transmembrane proton-transporting F 0 subcomplex. We identified subunits ␣, ␤, and ␦ of the F 1 complex and subunit-D of the F 0 complex. Consistent with our findings, other cell death treatments, such as oxidative stress, have been reported to suppress expression of ␣-subunit and ␤-subunit genes (45). Though not determined experimentally in this study, the repression of ATP synthase proteins by FB1 is likely to disrupt oxidative phosphorylation and lead to a significant depression in cellular ATP levels. Harpin, a pathogen-derived cell death-activating elicitor, disrupts oxidative phosphorylation by triggering cytochrome c release (46), thereby inhibiting mitochondrial ATP production prior to onset of cell death (46,47). Reduction in cellular ATP could likely account for growth retardation imposed by FB1 on Arabidopsis (5) as general metabolism is slowed. However, the fact that exogenous ATP blocked the suppression of ATP synthase genes by FB1 provides a profoundly fascinating insight. Extracellular ATP appears to positively regulate mitochondrial ATP synthesis, directly linking the cell death trigger-eATP depletion-to the key organelle now established as a central hub for programmed cell death control (48). Thus eATP depletion, driven by FB1 treatment (5), negatively regulates intracellular ATP production, which may lead to decreased ATP secretion that further reduces eATP. This step could serve as a signal amplification loop to ensure the onset of cell death in response to FB1 treatment.
Among the proteins differentially expressed following FB1 treatment were several enzymes involved in protecting cells from oxidative damage. Glutathione transferase and 2-alkenal reductase play significant roles in plant detoxification of lipid peroxide-derived cytotoxic compounds (49,50). Monodehydroascorbate reductase and dehydroascorbate reductase (Table I) function in the anti-oxidant glutathione-ascorbate cycle (51) whereas thioredoxin (Table I) together with thioredoxin reductase constitute the thioredoxin anti-oxidant system (52). Except for glutathione transferase and 2-alkenal reductase that increased, all the other anti-oxidative enzymes were down-regulated by FB1 (Table I). An overall reduction of these enzymes may account for the observed accumulation of ROS in Arabidopsis plants exposed to FB1 (30). ROS can function as signaling molecules (53) or triggers of cell death by attacking membrane lipids to give rise to phytotoxic lipid peroxides (54). Shutdown of some anti-oxidative enzymes and up-regulation of others by FB1 (Table I) could be indicative of a finely tuned balance between harnessing the signaling capability of ROS and tightly controlling their propensity to trigger membrane damage and cell death. Attenuation of FB1 effects on these proteins by exogenous ATP (Table I) implicates them as regulatory or effector elements downstream of eATP signaling in cell death control.
Glycolysis proved to be a major target for FB1 suppression as reflected by seven different proteins belonging to the pathway that were identified as down-regulated proteins ( Table I). The reduction in these proteins inevitably constricts flux through the pathway, a strategy that could be useful to divert energy and metabolites toward essential processes required for the response to FB1. For example, up-regulation of components of the amino acid biosynthetic pathways (Table I) could provide building blocks for pathogenesis-related protein synthesis that is triggered by FB1 (30). In addition, slowing down central metabolic pathways could also starve the cells of energy in preparation for programmed cell death. In rescuing the cells from death, exogenous ATP did not attenuate the response of all the glycolytic proteins to FB1, but it enhanced the suppression of some (Table I). This possibly indicates the need for ATP to reset the global metabolic processes in a highly ordered fashion before the full switch back to normal metabolism can be achieved. Reasons for the targeting of glycolytic enzymes by FB1 and exogenous ATP in this study may not be simplistic. It is now known that hexokinase, which catalyzes the first commitment step of glycolysis, is a critical regulator of programmed cell death in plants (55) and animals (56,57). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is now known to translocate to the nucleus of neuronal cells (58) where it activates apoptosis (58,59). In view of this, a possible role for GAPDH, and the other glycolytic proteins identified in this study, in FB1-induced cell death warrants investigation.
Enzymes of the protein degradation machinery also responded to FB1 treatment (Table I). Changes in protein abundance in response to treatment may entail synthesis of new proteins, for up-regulation, and degradation of existing proteins, for down-regulation. Thus, it is not surprising that the cellular protein degradation machinery was invoked by FB1. A previous study reported that an E3 ubiquitin ligase is a critical regulator of FB1-induced cell death in Arabidopsis (60), revealing a crucial function of the ubiquitin-dependent protein degradation pathway in this response. However, an interesting observation is that exogenous ATP did not significantly attenuate the FB1 effects on 4 of the 5 protein spots in this category (Table I). This probably implies that in resetting metabolism from inception of cell death to normal growth metabolism, exogenous ATP requires the action of the protein degradation machinery to eradicate prodeath proteins that had accumulated prior to its addition. Thus, both FB1 and exogenous ATP may recruit the same processes, but for different purposes.
Finally, the rest of the unclassified proteins differentially expressed in response to FB1 reveal that the toxin effects are widespread. The significance of a majority of these proteome changes is not yet clear to us, but when more data on the physiological effects of FB1 on Arabidopsis become available in the future, our data sets could prove useful in understanding the basis for FB1 action.
ATP Synthase ␤-subunit is a Novel Cell Death Signal Regulator-Reverse genetics, using T-DNA insertional mutants from the SALK collection, in combination with FB1 treatments revealed a new role for ATP synthase ␤-subunit in the regulation of cell death. Although the use of sequence-indexed T-DNA mutants accelerates gene discovery, occasionally the observed phenotype is caused by a secondary mutation and not by the T-DNA-tagged gene indexed in the mutant collection database. Therefore, confirmation of the observed phenotype with either a second independent mutant line or complementation of mutant plants with a copy of the native gene is required (61)(62)(63). In this study, we used three independent T-DNA mutant lines and all gave the same phenotype, confirming that ATP synthase ␤-subunit has a pro-cell death function in Arabidopsis.
In two-dimensional gels of microsomal membrane fractions, ATP synthase ␤-subunit existed as a charge train of seven spots of ϳ55 kDa and an extra spot (spot 39) of ϳ49 kDa (Fig. 2B). Another ATP synthase ␤-subunit spot (spot 33) of ϳ38 kDa was identified in the TSP fraction ( Fig. 2A). ATP synthase ␤-subunit has 566 amino acid residues and, according to MitoProt II prediction algorithm (64), the first 38 residues on the N terminus constitute the cleavable mitochondrial targeting sequence. Therefore, the mature sequence with 528 residues has a predicted molecular weight of 54,586.03 Da, which is very close to the experimentally determined molecular weight of ϳ55 kDa we observed for the charge train (Fig.  2B). All the seven ϳ55 kDa spots in the charge train were down-regulated in response to FB1 whereas the ϳ38 kDa and ϳ49 kDa were up-regulated (Tables I and II). The basis for the appearance of these lower molecular weight species is still unclear, but they possibly could arise from proteolytic processing of the ϳ55 kDa protein spots. If all the amino acids upstream of the identified most extreme N-terminal peptide and downstream of the identified most extreme C-terminal peptide (Fig. 3A) were cleaved off from the protein sequence, the predicted size of the truncated sequence becomes ϳ38.7 kDa, which almost matches the experimentally determined size of the ϳ38 kDa TSP low molecular weight spot (spot 33, Fig. 2A). Doing the same for the ϳ49 kDa microsomal spot based on the identified peptides (Fig. 3B) yields a putative product whose predicted size of ϳ47.6 kDa is close to the low molecular weight observed on gels (Fig. 2B). Therefore, the increase in the abundance of the low molecular weight spots, which is reciprocated by a decrease in the ϳ55 kDa charge train, could be accounted for by post-translational cleavage of the mature protein. The protein could be proteolytically pro-cessed in response to FB1 and ATP treatment protects it from degradation. The respective contribution to cell death regulation of the intact and truncated protein forms and the responsible proteases await further investigation.
The finding that FB1-induced eATP depletion down-regulates the abundance of Arabidopsis ATP synthase ␤-chain subunit agrees with our previous findings in tobacco. Treatment of tobacco with ␤,␥-methylene adenosine 5Ј-triphosphate (AMP-PCP), a nonhydrolysable ATP analog that interferes with eATP signaling, causes a dramatic suppression of several subunits of mitochondrial and chloroplast ATP synthase proteins (23). Therefore, eATP depletion achieved by FB1 treatment or by competitive exclusion from its binding sites with excess amounts of AMP-PCP have the same effect on mitochondrial and chloroplast ATP synthase proteins. Given that the ␥ phosphate of AMP-PCP is recalcitrant to cleavage, we can conclude that eATP-mediated regulation of ATP synthase proteins in Arabidopsis may similarly require cleavage of the ␥ phosphate as in tobacco (23). The nature of this reaction will become clear once the primary eATP target or receptor protein(s) in the extracellular matrix are identified. Although ATP synthase proteins had been identified as eATPregulated proteins in tobacco (23), we still did not know which eATP controlled physiological processes these proteins mediated. Therefore, in addition to confirming the previous findings in tobacco (23), the current study has now revealed a novel function of the Arabidopsis ATP synthase ␤-subunit in cell death regulation.
The mechanism by which ATP synthase ␤-subunit promotes death is the focus of current research in our laboratory. Nevertheless, we envisage three possible ways by which ATP synthase ␤-subunit could perform this function. First, it could directly interact with other core cell death factors in a protein complex that is independent from its classical function in ATP production. This would be similar to cytochrome c, a mitochondrial oxidative phosphorylation protein, which translocates to the cytosol, following exposure to a cell death stimulus, and forms a complex with caspase-9 and Apaf-1 to initiate apoptosis (65). Second, ATP synthase ␤-subunit might be capable of influencing gene expression directly or indirectly, thereby activating cell death genes. Cytosolic proteins such as gluceraldehyde-3-phosphate dehydrogenase (58) and enolase (66) are now known to translocate to the nucleus to affect gene expression, though they are classical glycolytic enzymes. ATP synthase ␤-subunit could have a similar secondary function in regulation of gene expression as revealed by altered basal expression of several genes in the knockout mutant plants (data not shown). Finally, ATP synthase ␤-subunit within the F 1 complex could be targeted for direct binding by FB1 or another prodeath protein/signal, leading to an inhibition of mitochondrial ATP production. Likewise, the basis for phytotoxicity of tentoxin, another fungal-derived toxin, is cellular depletion of ATP caused by inhibition of chloroplastic photophosphorylation (67). In this scenario, FB1 resistance in the At5g08690 knockout mutants could result from the lack of a binding site in the mitochondrial F 1 complex as the target At5g08690 gene product would be absent and replaced by products from one of the two other family members, At5g08680 and At5g08670. The Arabidopsis ATP synthase ␤-subunit protein belongs to a multigene family consisting of three members having 98% sequence similarity at the amino acid level, with differences only in the first 61 amino acids, making this region an ideal focal point for future genetic analyses to determine the basis for its cell death promotional function. Certain residues of the chloroplastic ATP synthase ␤-subunit were found to be critical for binding of the cell death-inducing tentoxin to the chloroplast F 1 -ATP synthase (68). Mutagenesis of a specific chloroplastic ATP synthase ␤-subunit gene codon of tentoxin-resistant Chlamydomonas reinhardtii to match the corresponding codon of tentoxinsensitive Nicotiana species rendered the alga tentoxin-sensitive (68), demonstrating very specific structural requirements for chloroplastic ATP synthase ␤-subunit function in cell death promotion. Similarly, a single amino acid substitution in the chloroplastic ␤-subunit switched tentoxin-resistant F 1 -ATP synthase of thermophilic Bacillus PS3 to a tentoxin-sensitive enzyme (69). A similar situation could account for the nonredundant function of At5g08690 gene product in cell death as revealed by FB1 resistance in the gene knockout plants.