Is ATP a Signaling Agent in Plants?

Physiological processes in plant cells are regulated by intrinsic and extrinsic signals. Numerous signaling molecules have been identified, including hormones, elicitors, and secondary metabolites. Cognate receptors and receptor genes have been reported in some cases (Hua and Meyerowitz, 1998; Inoue et al., 2001; Nam and Li, 2002). Several second messengers (responsible for signal amplification and encoding specificity) have been demonstrated in plants; for example, cytosolic Ca2+, cyclic nucleotides, and lipid derivatives. Calcium is a common second messenger in plant cell signaling; the amplitude and duration of transient elevations of [Ca2+]cyt may play an encoding role (Allen et al., 2001). Plants have evolved many plant-specific signaling molecules. Nevertheless they also "share" some signaling moieties with animals: nitric oxide, reactive oxygen species, and other regulators function in both Kingdoms. Glu (previously considered an exclusively animal signaling agent, a neurotransmitter) is now regarded as a likely plant signaling compound (Dennison and Spalding, 2000), and genes encoding putative Glu receptor subunits have been identified in the Arabidopsis genome (Lacombe et al., 2001). The possible conservation of Glu signaling between plants and animals suggests that other neurotransmitters might function in plants as signaling agents. One such candidate is extracellular ATP (eATP). Although ATP is an ubiquitous energy source, it also acts extracellularly as a neurotransmitter. The concept of ATP and purine derivatives as extracellular signaling molecules was born in the late 1920s and early 1930s when physiological effects of adenine derivatives were discovered (for review, see Ralevic and Burnstock, 1998). Further studies culminated in the proposal in the 1970s of purinergic neurotransmission mediated by specific plasma membrane receptors: "purinoceptors" (Burnstock, 1978). Two main families of purinoceptors (P1 and P2) have now been confirmed in animal cells (for review, see Ralevic and Burnstock, 1998); members exhibit varying sensitivity to eATP, purines, and pyrimidines. P1 receptors are activated by adenosine (P2 are not) and couple with heterotrimeric G proteins. P2 receptors divide into two classes; ligand-gated nonselective cation channels (P2X receptors) and G-protein-coupled receptors (P2Y receptors). P2X receptors activate at higher eATP concentrations (micromoles) than P2Y (nanomoles; Khakh, 2001). Purinoceptors may contribute to Ca2+ signal transduction cascades directly (through Ca2+ permeability of P2X channels) or indirectly (via downstream effects of the P2Y-coupled G-protein), and they are now implicated in a range of physiological functions ranging from neurotransmission to cell death (for review, see Ralevic and Burnstock, 1998; Burnstock and Williams, 2000). In contrast, eATP has received little attention from plant biologists. In the late 1960s and early 1970s, eATP effects on plants were largely being interpreted in terms of supplementation of cellular energy or chelation of divalent cations rather than signaling activity. Plant mechanical movements (e.g. Jaffe, 1973), K+ uptake (e.g. Lüttge et al., 1974), and endonuclease activity (Udvardy and Farkas, 1973) were found to be stimulated by eATP. Most studies tested eATP in the millimolar range, but it is notable that Venus fly trap closure was accelerated by 100 μm eATP (Jaffe, 1973), a concentration consistent with receptor activity (animal P2X receptors may exhibit one-half maximal activation by over 300 μm ligand; Humphrey et al., 1998). Since then, eATP effects have been rarely studied and mostly in the context of an exogenous energy supply. For example, stomatal aperture of Commelina communis was enhanced by eATP supplied at 5 to 20 mm, and tracer experiments indicated that eATP at such high levels could be taken up by epidermal tissue, possibly to act as an energy source for stomatal movement (Nejidat et al., 1983; Shaish et al., 1989). Only recently has experimentation begun to reveal possible purine signaling activity. Lew and Dearnaley (2000) found that eATP and extracellular ADP (eADP) depolarized the plasma membrane of Arabidopsis root hairs (one-half maximal effect produced by 0.4 mm eATP and 10 μm eADP). This would be consistent with ion channel opening. The effect of eATP on auxin distribution, gene expression, and gravitropism in Arabidopsis roots has been recently reported (Tang et al., 2003) and the authors considered one possible explanation to be eATP acting as a regulatory signal. Therefore, previous data suggest the existence of plant eATP signaling. The objective of the present work was to test for possible eATP-induced signaling events. As Ca2+ is a key, quantifiable second messenger in plant cells and is downstream of eATP in animal cells, we tested the null hypothesis that eATP has no effect on [Ca2+]cyt. To test for changes in [Ca2+]cyt, Arabidopsis plants constitutively expressing aequorin in the cytosol (35S promoter-driven expression; Knight et al., 1996) were used. Roots (excised) were tested because roots are naturally directly exposed to the soil solution where eATP occurs due to release from microorganisms (Thomas et al., 1999) and perhaps damaged or sloughed off root cells, and root hairs have been shown to be purine responsive (Lew and Dearnaley, 2000). An eATP range from 100 nm to 1 mm was tested. Application of assay solution alone caused a "touch response" (about 5 s long) and such transients, caused by injection of solution, were also observed in the case of eATP addition (Fig. 1A). After the touch response, eATP caused a dose-dependent transient elevation in [Ca2+]cyt (Fig. 1A), with even 300 nm causing a 2-fold increase in [Ca2+]cyt. At [eATP] < 3 μm, the [Ca2+]cyt elevation comprised two distinguishable peaks. The first peak reached its maximum in 10 to 20s, whereas the second maximum was attained 1 to 2 min after application. Above 3 μm [eATP], the first peak was not so clearly distinguishable (merging with the second peak). As a general trend, both peaks increased as [eATP] increased, but the ratio between the amplitudes of first and second peaks tended to decrease. [Ca2+]cyt transients lasted 5 to 10 min in total, and basal [Ca2+]cyt was fully recovered. eATP was added as the disodium salt, therefore, control measurements with equimolar NaCl were carried out. At 0.1 mm, NaCl caused no [Ca2+]cyt elevation (n = 3; data not shown). At 2 mm, NaCl caused a transient [Ca2+]cyt elevation that lasted approximately 1 min with a mean ± sem peak value of 0.32 ± 0.06 μm, n = 4 (compare with basal value before addition; 0.11 ± 0.01 μm). In contrast, 1 mm Na2ATP caused a mean peak value of 6.97 ± 0.82 μm, n = 5 (compare with basal value before addition; 0.10 ± 0.01 μm). Thus, [Ca2+]cyt transients were due to eATP rather than Na+. Extracellular Ca2+ depletion (10-0.1 mm) abolished 85% of the [Ca2+]cyt increase induced by 3 μm eATP, demonstrating eATP-activated Ca2+ entry from the extracellular space (Fig. 1A). Maximal peak [Ca2+]cyt values were plotted against [eATP] (Fig. 1B). The dose-dependence curve reached saturation at 0.3 mm eATP with EC50 at 2.6 μm (analyzed by SigmaPlot 4.01 software). Overall, the null hypothesis was disproved; in root cells at least, eATP can elevate [Ca2+]cyt. The specificity of this response was then tested. Effect of externally applied purines and UTP on Arabidopsis root [Ca2+]. A, Typical [Ca2+ cyt]cyt transients induced by different eATP concentrations. External Ca2+ was 10 mm unless indicated. B, Mean dose-response curves for effects of different agonists on [Ca2+]cyt. sem did not exceed 15% of mean (not shown). Experimental procedures are described in detail by Demidchik et al. (2003). Arabidopsis constitutively expressing cytosolic aequorin driven by the cauliflower mosaic 35S promoter (Knight et al., 1996), were grown vertically at 22°C (16-h daylength; 100 μmol m–2 s–1 irradiance) on full-strength Murashige and Skoog medium (Duchefa, Harlem) with 1% (w/v) Suc and 0.35% (w/v) Phytagel (Sigma, St. Louis). Recording solution contained 10 mm CaCl2 (or 0.1 mm), 2 mm MES, pH 5.7 (Tris), and 4 μg mL–1 coelenterazine (free base; NanoLight Technology, Prolume, Pittsburgh). Luminescence was recorded at 1 Hz. Duration and amplitude of purine-induced Ca2+ transients in animal cells are determined by the structure of the purine derivative (White et al., 2001). The effects of ATP derivatives (see Ralevic and Burnstock, 1998), ADP, and the pyrimidine UTP are shown in Figure 1B. All compounds tested elicited a double peak [Ca2+]cyt transient response and maximal peak values are reported. ADP was a potent activator; with up to 10 μm ADP, the magnitude of response was lower but still comparable with ATP (e.g. mean response to 3 μm addition: eATP, 480 nm Ca2+; eADP, 310 nm Ca2+). 2-Methylthio ATP (an ATP analog with a modified purine ring) was, at lower concentrations, as effective as ATP and ADP. BzATP (ATP analog with modified Rib moiety) produced a weaker effect over the concentration range tested. Critically, the nonhydrolyzable ATP analogs αβmeATP (methylene insert between first and second phosphates) and βγmeATP (methylene insert between second and third phosphates) were effective (Figs. 1B and 2A). This clearly demonstrates that hydrolysis is not required for the activation of [Ca2+]cyt transients by eATP. αβmeATP was the more potent of the two. Even though ADP was an effective agonist, AMP and Pi (as other ATP break-down products; tested at 0.1-1 mm) did not elicit statistically significant elevations of [Ca2+]cyt (measured 1 min after addition). For AMP experiments, mean ± sem [Ca2+]cyt: basal, 0.096 ± 0.004 μm, n = 10; 1 mm AMP, 0.138 ± 0.017 μm, n = 4. For Pi (as H3PO4, buffered with Tris), mean ± sem [Ca2+]cyt: basal, 0.112 ± 0.006 μm, n = 4; 1 mm Pi 0.119 ± 0.007 μm, n = 4. The pyrimidine UTP (which can also elicit a [Ca2+]cyt response in animal cells) was only effective at higher concentrations (>100 μm), thus demonstrating not only the significance of purine structure, but also that the ATP effect was unlikely to have been due to a surface charge effect mediated by the phosphate groups of the agonist. Adenosine did not cause [Ca2+]cyt transients (mean ± sem [Ca2+]cyt: basal, 0.107 ± 0.007 μm; 1 mm adenosine 1 min postaddition, 0.106 ± 0.006 μm, n = 5); the simplest structure, adenine, was also ineffective (mean ± sem [Ca2+]cyt: basal, 0.119 ± 0.008 μm; 1 mm adenine 1 min postaddition, 0.122 ± 0.009 μm, n = 5). Effect of antagonists on eATP-induced [Ca2+]cyt elevation. A, Typical [Ca2+]cyt response to 10–4 m ATP or αβmeATP and the effect of 0.3 mm PPADS. B, Reduction of αβmeATP-induced [Ca2+]cyt response (mean ± sem). Antagonists were added 10 min before addition of αβmeATP. Here, the [Ca2+]cyt transients manifested from a multicellular system comprising various cell types in different developmental and physiological states, all with potentially different capacities to respond to purine derivatives. Therefore, quantitative comparison of the data shown in Figure 1B would have little meaning. Having demonstrated that changes in ligand structure evoked measurably different responses, the mechanism of [Ca2+]cyt elevation was explored. Reducing external Ca2+ diminished the eATP-induced transient (see above), suggesting that Ca2+ influx contributed to the response. To confirm the operation of a plasma membrane influx pathway, the effect of cation channel blockers was investigated. In this pharmacological analysis, the nonhydrolyzable ATP analog αβmeATP (0.1 mm) was used to prevent possible effects caused by phosphorylation or ATP breakdown. Of the conventional cation channel blockers (Fig. 2B), application of the K+ channel blocker tetraethylammonium+ (20 mm) or Ca2+ channel blocker verapamil (0.1 mm) resulted in mean maximum [Ca2+]cyt increases that were 65% and 68% of the control value, respectively (n = 5; Fig. 2B). Whatever the mechanism of initiating the [Ca2+]cyt increase, it was resistant to these channel antagonists. The 35% and 32% decreases in peak [Ca2+]cyt response by tetraethylammonium and verapamil, respectively, may reflect the sensitivity of contributing transport systems downstream of the initiating mechanism in this multicellular test system. The [Ca2+]cyt increase was most strongly affected by Gd3+ (6% of mean control response at 0.1 mm, Fig. 2B; n = 5). Gd3+ is a nonspecific blocker of plant cation channels (including Arabidopsis root nonselective cation channels; Demidchik et al., 2002a,b) and this result clearly demonstrates that plasma membrane channel-mediated Ca2+ influx is an initial step in eATP-induced [Ca2+]cyt elevation. The role of Ca2+ intracellular stores (if any) in generating the transients requires greater resolution than aequorin affords. The mechanistic basis of eATP-induced [Ca2+]cyt elevation in plants could be absolutely different from that found in animals or could show some degree of conservation. If the mechanism of activating the Ca2+-permeable channels that underlie eATP-induced Ca2+ elevation were similar in plants and animals, it is reasonable to predict a conserved pharmacological profile. Animal purinoceptor-mediated signaling is highly sensitive to pyridoxalphosphate-6-azophenyl-2′,4′-disulfonic acid (PPADS) and suramin (see Ralevic and Burnstock, 1998). In animal preparations, PPADS has been shown to specifically block P2 receptors (see Ralevic and Burnstock, 1998). Suramin, apart from blocking animal P2 receptors, may also affect other receptors and enzymes in animal and plant cells (Ralevic and Burnstock, 1998; Stratmann et al., 2000), but it remains a key diagnostic tool for P2 delineation. We have tested these antagonists (0.3 mm) on excised roots. The eATP and αβmeATP effects on [Ca2+]cyt were strongly attenuated to a degree comparable with application of Gd3+ (Fig. 2; n = 5). If, as the data suggest, eATP were to activate Ca2+-permeable plasma membrane channels, what is the most likely mechanism of activation? If eATP were to permeate, cation channel phosphorylation at the cytosolic face could activate Ca2+ influx. Hyperpolarization-activated Ca2+ channels (which can contribute to Ca2+ influx across the plant cell plasma membrane; Véry and Davies, 2000; Demidchik et al., 2002a) perhaps activate at cytosolic [ATP] > 0.5 mm (Köhler and Blatt, 2002). Cytosolic ADP or nonhydrolyzable ATP do not increase the activity of these channels, therefore, the ATP effect is due to phosphorylation (Köhler and Blatt, 2002). Additionally, if eATP were to permeate, it could conceivably activate ABC transporters by binding to their cytoplasmic nucleotide-binding domains. ABC transporters are increasingly recognized as being capable of regulating ion channel activity and have a role in plant cell signaling (Schmid-Antomarchi et al., 1987; Bryan and Aguilar-Bryan, 1999; Klein et al., 2003). In the present study, it is difficult to make a case for the ATP influx necessary for cytosolic phosphorylation of channels or ABC activation. The negative membrane voltage of plant root cells (about –160 mV in the conditions used here; Demidchik et al., 2002a) and the estimated species concentrations of cytosolic ATP (Davies et al., 1993) strongly support anion channel-mediated ATP efflux over the range of eATP tested. Even if ATP influx were feasible, phosphorylation at the cytosolic or extracellular face is clearly not essential for generation of the [Ca2+]cyt transients, as ADP and αβmeATP (nonhydrolyzable ATP analog) were potent agonists acting at micromolar concentrations and exhibiting saturation. Thus, it is also unlikely that Ca2+ influx was due simply to a stimulation of the plasma membrane H+-ATPase and hence membrane hyperpolarization. However, it is important to note that the αβmeATP response does not rule out the possibility of the involvement of a regulatory ABC transporter. If αβmeATP were to permeate (although this appears unlikely), it could feasibly change the conformational state of an ABC transporter, as this can be achieved by ATP binding alone (Rosenberg et al., 2001). The antagonistic effect of PPADS and suramin raises the possibility of P2 equivalents in plant cells. It is feasible that eATP-activated P2Y equivalent(s), sensitive to PPADS and suramin, couple to Gd3+-sensitive Ca2+-permeable channels. It is feasible that there is an eATP-binding and PPADS/suramin-sensitive periplasmic or membrane moiety that regulates Gd3+-sensitive Ca2+-permeable channels. It is equally feasible that the eATP-induced [Ca2+]cyt elevation is mediated by a class of ionotropic receptors, the eATP-gated Ca2+-permeable cation channel. In all cases, other plasma membrane Ca2+-permeable channels could lie downstream of the initial eATP response and could contribute to the observed [Ca2+]cyt elevation. These possibilities must now be addressed for individual root cell types using electrophysiological techniques. This would enable direct testing of at least ionotropic versus metabotropic activity and permit comparison of the pharmacological response of ion channels with that of the excised roots reported here. The Arabidopsis genome has been reported to contain no sequences with obvious similarities to animal P2X genes (Tang et al., 2003), but extensive in silico analysis has revealed sequences predicted to encode polypeptides with two membrane-spanning domains linked by a Cys-rich extracellular loop bearing a putative ATP-binding motif, reminiscent of a P2X subunit (V. Demidchik, M. Oliynyk, B. J. Glover, and J. M. Davies unpublished data). The ability of eATP (and to some extent eADP) to elevate [Ca2+]cyt in plants implies further possible conservation with animal purine signaling. The mechanism of eATP/eADP signaling in plants (by analogy with animals) should include systems for rapid release from cells and rapid removal from the extracellular space. In animal cells, ATP is anion channels, ABC or and Burnstock, 2001; et al., ATP and ADP are by extracellular the and et al., To in the ABC is to ATP release from Arabidopsis (Thomas et al., 2000), and extracellular (which are found in are to ATP more than other extracellular and et al., Therefore, plant cells do have the to eATP as a signaling It is notable that if eATP (or perhaps eADP) were a signal by then (as shown in this the hydrolysis AMP and Pi would be effect on the signal would be This has demonstrated that eATP could be a signaling in The physiological downstream of an eATP-induced [Ca2+]cyt transient to be but in of the present a of reported eATP effects on plant cells is now For example, of the Arabidopsis root response by eATP millimolar levels into the has been reported recently (Tang et al., 2003). Here, a lower ATP concentration that can elicit a root [Ca2+]cyt response also mm 0.5 mm and 1 mm MES, pH were applied to Arabidopsis roots before a in the After 3 the mean ± sem response of the control was ± (n = plants in and the mean 100 μm ATP response was ± (n = this is a statistically significant at the experiments that 100 μm ADP is also but AMP and UTP are to eATP to gravitropism [Ca2+]cyt now requires a of the and cells. of ATP could be from damaged cells or as a result of cell In this a role in signaling could be In to the animal plant cells could also be exposed to eATP by other can release ATP and other purines, at the may perhaps eATP as a signaling ATP release by plant cells may be mediated by ABC transporters (Thomas et al., and it is to note the possible between this of and the eATP-activated Ca2+ influx reported here. eATP is to ABC transporters in Arabidopsis (Thomas et al., 2000), and of Ca2+ (by Ca2+ has been found to increase of an ABC in root et al., 2002). ATP release by ABC for [Ca2+]cyt eATP activation of Ca2+ influx. In animals, neurotransmitters and a between eATP could play a similar role in perhaps to developmental or to Further studies are now required to the plant and physiological We Knight for and for on animal