Plant extracellular ATP signalling by plasma membrane NADPH oxidase and Ca2+ channels


  • Vadim Demidchik,

    1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK
    Search for more papers by this author
    • Present addresses: VD; Department of Biology, University of Essex, Colchester, CO43SQ, United Kingdom. ZS; College of Life Science, Hebei Normal University, Yuhua East Road, Shijiazhang 050016, Hebei, China. RS; Riken Plant Science Center, Yokohama City, Kanagawa 230-0045, Japan. LR; Departamento de Biología Vegetal, Facultad de Ciencias, Universidad de Málaga, Campus de Teatinos, 29071 Málaga, Spain.

    • These authors contributed equally to this work.

  • Zhonglin Shang,

    1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK
    Search for more papers by this author
    • Present addresses: VD; Department of Biology, University of Essex, Colchester, CO43SQ, United Kingdom. ZS; College of Life Science, Hebei Normal University, Yuhua East Road, Shijiazhang 050016, Hebei, China. RS; Riken Plant Science Center, Yokohama City, Kanagawa 230-0045, Japan. LR; Departamento de Biología Vegetal, Facultad de Ciencias, Universidad de Málaga, Campus de Teatinos, 29071 Málaga, Spain.

    • These authors contributed equally to this work.

  • Ryoung Shin,

    1. Donald Danforth Plant Science Center, St Louis, MO 63132, USA
    Search for more papers by this author
    • Present addresses: VD; Department of Biology, University of Essex, Colchester, CO43SQ, United Kingdom. ZS; College of Life Science, Hebei Normal University, Yuhua East Road, Shijiazhang 050016, Hebei, China. RS; Riken Plant Science Center, Yokohama City, Kanagawa 230-0045, Japan. LR; Departamento de Biología Vegetal, Facultad de Ciencias, Universidad de Málaga, Campus de Teatinos, 29071 Málaga, Spain.

    • These authors contributed equally to this work.

  • Elinor Thompson,

    1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK
    Search for more papers by this author
  • Lourdes Rubio,

    1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK
    Search for more papers by this author
    • Present addresses: VD; Department of Biology, University of Essex, Colchester, CO43SQ, United Kingdom. ZS; College of Life Science, Hebei Normal University, Yuhua East Road, Shijiazhang 050016, Hebei, China. RS; Riken Plant Science Center, Yokohama City, Kanagawa 230-0045, Japan. LR; Departamento de Biología Vegetal, Facultad de Ciencias, Universidad de Málaga, Campus de Teatinos, 29071 Málaga, Spain.

  • Anuphon Laohavisit,

    1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK
    Search for more papers by this author
  • Jennifer C. Mortimer,

    1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK
    Search for more papers by this author
  • Stephen Chivasa,

    1. School of Biological and Biomedical Sciences, University of Durham, Durham DH1 3LE, UK
    Search for more papers by this author
  • Antoni R. Slabas,

    1. School of Biological and Biomedical Sciences, University of Durham, Durham DH1 3LE, UK
    Search for more papers by this author
  • Beverley J. Glover,

    1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK
    Search for more papers by this author
  • Daniel P. Schachtman,

    1. Donald Danforth Plant Science Center, St Louis, MO 63132, USA
    Search for more papers by this author
  • Sergey N. Shabala,

    1. School of Agricultural Sciences, University of Tasmania, Hobart, Tas. 7001, Australia
    Search for more papers by this author
  • Julia M. Davies

    Corresponding author
    1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK
      *(fax +44 (0) 1223 333 953; e-mail
    Search for more papers by this author

*(fax +44 (0) 1223 333 953; e-mail


Extracellular ATP regulates higher plant growth and adaptation. The signalling events may be unique to higher plants, as they lack animal purinoceptor homologues. Although it is known that plant cytosolic free Ca2+ can be elevated by extracellular ATP, the mechanism is unknown. Here, we have studied roots of Arabidopsis thaliana to determine the events that lead to the transcriptional stress response evoked by extracellular ATP. Root cell protoplasts were used to demonstrate that signalling to elevate cytosolic free Ca2+ is determined by ATP perception at the plasma membrane, and not at the cell wall. Imaging revealed that extracellular ATP causes the production of reactive oxygen species in intact roots, with the plasma membrane NADPH oxidase AtRBOHC being the major contributor. This resulted in the stimulation of plasma membrane Ca2+-permeable channels (determined using patch-clamp electrophysiology), which contribute to the elevation of cytosolic free Ca2+. Disruption of this pathway in the AtrbohC mutant impaired the extracellular ATP-induced increase in reactive oxygen species (ROS), the activation of Ca2+ channels, and the transcription of the MAP kinase3 gene that is known to be involved in stress responses. This study shows that higher plants, although bereft of purinoceptor homologues, could have evolved a distinct mechanism to transduce the ATP signal at the plasma membrane.


ATP is a ubiquitous intracellular energy source, but, in animals, ATP also acts at the extracellular face of the plasma membrane as a paracrine or autocrine signalling agent (Khakh and North, 2006). There, extracellular ATP is involved (together with extracellular ADP and adenosine) in the regulation of a wide range of cellular processes, including neurotransmission, immune response, and cell growth and death (Khakh and North, 2006). Cellular responses to extracellular purines are mediated by specific ‘purinoceptors’ that comprise the P1 family (heterotrimeric G-protein-coupled adenosine receptors) and P2 family (P2X, ligand-gated ion channels; P2Y, G-protein-coupled receptors). The elevation of intracellular cytosolic free Ca2+, [Ca2+]cyt, as a second messenger is a common outcome of purinoceptor activation (Khakh and North, 2006).

It is now over 30 years since the first observations on the effects of extracellular ATP in plants, namely Venus flytrap closure (Jaffe, 1973), induction of endonucleases in oat leaves (Udvardy and Farkas, 1973) and stimulation of K+ uptake (Lüttge et al., 1974), were made. During this period, the molecular mechanisms underpinning purine signalling in animal cells have been elucidated, whereas scant attention has been paid to plant systems. Recently, it has become clear that extracellular ATP regulates plant viability (Chivasa et al., 2005), growth (Kim et al., 2006; Roux and Steinebrunner, 2007; Wu et al., 2007a), gravitropism (Tang et al., 2003) and stress responses (Thomas et al., 2000; Jeter et al., 2004; Song et al., 2006). That ATP operates in such processes places it alongside better established regulators such as ABA, in terms of fundamental importance. The way in which extracellular ATP functions as a plant cell regulator must now be established.

ATP is released from plant cells by plasma membrane ABC transporters (Thomas et al., 2000; Roux and Steinebrunner, 2007), vesicular efflux (Kim et al., 2006) and wounding (Jeter et al., 2004). The levels of extracellular ATP are regulated by apyrases (extracellular nucleotide phosphohydrolases) that catalyse its breakdown (Thomas et al., 2000; Wu et al., 2007a). Physiological studies point to roles for extracellular ATP in cell viability, defence and growth. Experimental depletion of extracellular ATP triggers plant cell death in a range of species, and hence it has been proposed to act as a suppressor of a default cell death pathway (Chivasa et al., 2005). Commitment to cell death in the model plant Arabidopsis thaliana caused by the mycotoxin fumonisin B1 (an elicitor of death and defence responses) correlates with a loss of extracellular ATP (probably through hydrolysis), and can be reversed by its application (Chivasa et al., 2005).

In roots, extracellular ATP levels are highest in the elongation zone and at the root hair apex (Kim et al., 2006), implicating ATP in growth regulation. The depletion of extracellular ATP (by application of apyrase) severely inhibits Medicago root hair elongation (Kim et al., 2006), whereas the application of extracellular ADP results in increased elongation of Arabidopsis root hairs (Lew and Dearnaley, 2000). The current model assumes that extracellular ATP must be tightly regulated, as too little ATP results in cessation of growth and death, but too much ATP also arrests growth, possibly by impairing auxin transport (Tang et al., 2003; Kim et al., 2006; Roux and Steinebrunner, 2007; Wu et al., 2007a). In Arabidopsis seedlings and suspension cells, high extracellular ATP concentrations cause the transcription of genes involved in stress responses (Jeter et al., 2004; Song et al., 2006), suggesting the involvement of extracellular ATP in the switch from growth to stress adaptation (Roux and Steinebrunner, 2007).

How plants perceive extracellular ATP remains unknown. There are no equivalents to animal purinoceptors evident in higher plant genomes (Kim et al., 2006; Fountain et al., 2008). Although a P2X receptor has recently been identified in the unicellular green alga Ostreococcus tauri, it does not function at the plasma membrane (Fountain et al., 2008). In Arabidopsis, extracellular ATP transiently elevates [Ca2+]cyt in whole organs, possibly by increasing Ca2+ influx at the plasma membrane (Demidchik et al., 2003a; Jeter et al., 2004; Kim et al., 2006), but the underlying mechanisms of extracellular ATP sensing and Ca2+ flux have not yet been identified. Plasma membrane Ca2+-permeable channels would be capable of mediating the influx of Ca2+. In contrast to animals, the plasma membrane may not even be the site of ATP recognition and initiation of signalling. The plant extracellular matrix contains a suite of secreted proteins capable of being phosphorylated by ATP (Chivasa et al., 2002; Ndimba et al., 2003), and it has been suggested that these could relay a signal to the cell (Chivasa et al., 2005).

At present there is no direct evidence for channel activation in response to extracellular ATP, or even for the ATP ‘target’ residing at the plasma membrane, rather than in the extracellular matrix. Here, we have aimed to resolve the site of extracellular ATP perception in Arabidopsis and to delineate an ATP-regulated stress signalling pathway. To do so, we have refined the study of ATP effects in plants by moving away from whole-organ studies to a defined cell type: the root epidermis. Using this cell type, we have determined that extracellular ATP is sensed at the plasma membrane rather than in the wall. The perception of extracellular ATP causes the production of reactive oxygen species (ROS) through the activation of a specific plasma membrane NADPH oxidase. ROS, in turn, activate plasma membrane hyperpolarization-activated Ca2+ influx channels, thereby mediating [Ca2+]cyt elevation. Extracellular ATP-induced ROS accumulation, Ca2+ channel activation and transcription of the stress-signalling mitogen-activated protein kinase MPK3 (Kovtun et al., 2000; Rentel et al., 2004) are impaired in a loss-of-function NADPH oxidase mutant. This indicates that extracellular ATP controls [Ca2+]cyt-dependent signalling in plants, acting via redox-activated Ca2+ channels, but with receptors that are structurally distinct from those of animals.


Extracellular ATP increases Ca2+ influx at the root epidermis

Extracellular ATP effects on plant Ca2+ have previously been determined only at the whole-organ level of roots and leaves (Demidchik et al., 2003a; Jeter et al., 2004). Here, to elucidate the mechanistic basis of ATP responses, we have examined a single cell type. The mature, fully expanded root epidermis of Arabidopsis is ideal for determining ATP stress signalling pathways, as it should be capable of sensing environmental change, and ATP responses here should be unrelated to the growth mechanism. To discriminate between ATP perception at the cell wall or plasma membrane, we first checked for the ATP activation of Ca2+ influx in the mature epidermis, and then removed cell walls enzymatically and tested the protoplasts for responsiveness.

The influx of Ca2+ at the mature epidermis of excised roots was measured using a vibrating Ca2+-selective microelectrode, and, under control conditions, the mean (± SE) net Ca2+ influx was 0.53 ± 1.6 nmol m−2 sec−1 (= 97). The extracellular Ca2+ concentration was maintained at 0.1 mm to aid electrode function. A modification of the chamber design allowed solution exchange, and a resumption of recording after 20–30 sec rather than after minutes, as was possible previously (Shabala et al., 1997). Thus early responses to test agents could be recorded. In preliminary trials an H+-selective microelectrode was used to confirm that the addition of purines caused no change in extracellular pH, and, as was found previously (Demidchik et al., 2007), the addition of control buffer alone (no purines and acting as a control for mechanical disturbance) did not cause a change in the net Ca2+ flux (n = 4). Application of 30 μm ATP (a concentration generated by wounding; Song et al., 2006) or its stable analogue adenosine 5′-(α,β-methylene)triphosphate (αβme-ATP) caused transient net Ca2+ influx at the mature and fully-expanded epidermis (Figure 1a), approximately 1 min after the addition of agonist. The peak net influx in response to 30 μm ATP was 13 ± 9 nmol m−2 sec−1, = 5. In response to 30 μmαβme-ATP, the peak net influx was 7 ± 2 nmol m−2 sec−1, = 4. AMP was far less potent, requiring an application of 1 mm to evoke a transient increase in net Ca2+ influx of 3 ± 3 nmol m−2 sec−1 (= 4). As free external Ca2+ activity was constant during the application of purine, the increased influx was unlikely to be a homeostatic response to chelation of extracellular Ca2+ by ATP.

Figure 1.

 Extracellular ATP increases the net Ca2+ influx and elevates the intracellular cytosolic free Ca2+, [Ca2+]cyt, in the root epidermis.
(a) Extracellular adenosine 5′-(α,β-methylene)triphosphate (30 μm; αβme-ATP) caused a transient increase in the net Ca2+ influx to fully-expanded Arabidopsis root epidermis. Extracellular Ca2+ was maintained at 0.1 mm.
(b) The transient increase in [Ca2+]cyt in protoplasts of fully-expanded Arabidopsis root epidermal cells, caused by treatment with 100 μm ATP, and its inhibition by 0.5 mm suramin.
(c) Mean ± SE peak transient [Ca2+]cyt increase in protoplasts evoked by ATP, αβme-ATP, ADP or AMP (= 9–12).
(d) Effect of 300 μm Gd3+ or 0.5 mm suramin on the mean ± SE peak transient protoplast [Ca2+]cyt increase evoked by 100 μm ATP, αβme-ATP or ADP (= 4–12).
(b–d) Extracellular 1 mm CaCl2.

Extracellular ATP is sensed at the plasma membrane

Enzymatic wall removal from the mature epidermis still permitted ATP responses. Extracellular ATP caused the transient and dose-dependent elevation of [Ca2+]cyt in mature epidermal protoplasts (detected using cells that constitutively express aequorin; Figure 1b,c). The resting level of [Ca2+]cyt under control conditions (1 mm CaCl2, pH 6) was 107 ± 2 nm (= 521). In these experiments the level of extracellular Ca2+ was higher than that used previously, in order to increase the driving force for Ca2+ and to maximize the resultant signal. The peak [Ca2+]cyt response occurred approximately 1 min after the application of the agonist. Agonists were tested over a concentration range from 1 μm to 1 mm. At the lowest concentration tested (1 μm), ATP caused a peak increase in [Ca2+]cyt of 20 ± 14 nm (= 9), representing a 19% increase above the basal level. The duration of the transient [Ca2+]cyt elevation was dependent on agonist concentration; e.g. the duration of the transient induced by 100 μm and 1 mm ATP was 5.5 ± 0.5 min (n = 9) and 11 ± 0.4 min (n = 3), respectively. Basal levels of [Ca2+]cyt were always fully recovered after agonist challenge. Extracellular αβme-ATP and ADP also caused the transient dose-dependent elevation of protoplast [Ca2+]cyt (Figure 1c). As the Li+ salt of αβme-ATP was used, the effect of Li+ on [Ca2+]cyt was also tested. The addition of up to 2 mm LiCl did not induce significant [Ca2+]cyt transients (n = 9). ADP was added as its Na+ salt: NaCl as a control was not effective as an elicitor, and even 1 mm only raised [Ca2+]cyt by 0.01 ± 0.01 μm (= 4). That ADP was an effective agonist indicates that hydrolysis of the terminal phosphate of ATP was not necessary for the elevation of [Ca2+]cyt. Protoplasts exhibited a low sensitivity to AMP in terms of response threshold and magnitude (Figure 1c).

Lowering extracellular Ca2+ from 1 mm to 5 μm caused an 85% reduction in the peak [Ca2+]cyt response to 100 μmαβme-ATP. The mean ± SE was 200 ± 40 nm in 1 mm Ca2+, and 30 ± 10 nm in 5 μm Ca2+ (= 4), which is indicative of the plasma membrane Ca2+ influx generating the majority of the [Ca2+]cyt response. Gd3+ (a blocker of root cell Ca2+-permeable channels; Demidchik et al., 2007; Foreman et al., 2003) is known to inhibit the elevation of extracellular ATP-induced [Ca2+]cyt in Arabidopsis roots and seedlings (Demidchik et al., 2003a; Jeter et al., 2004). Here, 300 μm Gd3+ inhibited the elevation of protoplast [Ca2+]cyt in response to 100 μm ATP, αβme-ATP or ADP by 88, 75 and 92%, respectively (= 9–12; Figure 1(d)). Responses were also inhibited by 0.5 mm suramin (100 μm ATP, 82%; 100 μmαβme-ATP, 97%; 100 μm ADP, 95%; = 3–9;Figure 1d), an antagonist of animal purinoceptors known to inhibit the ATP-induced elevation of root [Ca2+]cyt (Demidchik et al., 2003a). Suramin also significantly inhibited root elongation (control, 5.6 ± 0.02 mm day−1; 50 μm suramin, 1.5 ± 0.002 mm day−1; = 30; 6–8-day growth interval; P < 0.001, Student’s t-test), supporting the premise that normal growth requires the perception of extracellular ATP (Kim et al., 2006; Roux and Steinebrunner, 2007; Wu et al., 2007a). Overall, the results demonstrate that extracellular ATP is sensed at the plasma membrane.

ATP evokes a plasma membrane Ca2+ conductance

Having resolved the extracellular ATP regulation of plant [Ca2+]cyt at the single cell-type level, we then addressed the underlying transport reactions. The inhibition of ATP-induced elevation of [Ca2+]cyt by channel blockers implied that plasma membrane Ca2+-permeable channels mediate Ca2+ influx. To investigate such channels, protoplasts from mature root epidermal cells were patch clamped. This technique relies on complete wall removal, and therefore its successful use verifies the protoplasting protocol used in the aequorin studies. Application of 20 μm extracellular ATP to protoplasts (in ‘whole cell’ recording mode) transiently activated a plasma membrane hyperpolarization-activated Ca2+ channel conductance (Figure 2a), similar to those characterized previously in this and other cell types, including guard cells (Pei et al., 2000; Foreman et al., 2003; Demidchik et al., 2007; Wu et al., 2007b). An increase in inwardly-directed current was observed 1–3 min after the addition of ATP, and was evident over the physiological voltage range for this cell type (Maathuis and Sanders, 1993). At −200 mV, the current increased approximately twofold from −164 ± 42 pA to a peak of −362 ± 65 pA after 5–20 min exposure (= 6), after which the current declined. The ATP analogue 2′-O-(4-benzoylbenzoyl) adenosine 5′-triphosphate (BzBzATP) caused comparable activation of channel conductance, which was fully inhibited by 100 μm Gd3+ (Figure 2b; 30 μm, = 7).

Figure 2.

 Extracellular ATP evokes a plasma membrane Ca2+ conductance in root epidermal protoplasts.
(a) Effect of 20 μm extracellular ATP. Whole-cell currents are shown above mean ± SE current–voltage relationships (= 6).
(b) 2′-O-(4-Benzoylbenzoyl) adenosine 5′-triphosphate (30 μm; BzBzATP) transiently increased the whole-cell conductance (G, ○), and 100 μm Gd3+ inhibited the response (•; = 7).
(c) The ATP-activated conductance was variously permeable to Ca2+, Ba2+, Mg2+ and Zn2+, and was inhibited by 100 μm Gd3+ (= 3 for all trials; 20 mm external divalent cation). The control current–voltage relationship was recorded in external Ca2+ with no ATP; Gd3+ was applied in the Ca2+ + ATP test.
(d) Suramin (0.5 mm) prevented the activation of the Ca2+ conductance by 20 μm ATP (= 6).

The ATP-activated conductance was highly permeable to Ba2+ and Mg2+, but was only weakly permeable to Zn2+ (100 μm ATP; Figure 2c), consistent with the operation of Ca2+ channels, rather than K+ channels (Hirsch et al., 1998; Foreman et al., 2003). The permeability sequence (determined from conductance values; Demidchik et al., 2003b) was Ba2+ (1.21 ± 0.14)>Ca2+ (1.00) > Zn2+ (0.19 ± 0.05); = 3. The ATP-induced Ca2+ conductance was blocked by 100 μm Gd3+ (Figure 2c; = 3), and activation was prevented by 0.5 mm suramin (control I at −200 mV, −217 ± 38 pA, n = 6; suramin, −198 ± 43 pA, = 6; Figure 2d). These data indicate that the hyperpolarization-activated Ca2+ conductance is required for the ATP-activated elevation of [Ca2+]cyt observed in epidermal protoplasts and whole roots (Demidchik et al., 2003a).

A 19-pS Ca2+ channel is activated in response to extracellular ATP

We next determined the identities of the channels contributing to this transport reaction. At the single-channel level, a 4–5 pS channel (typical of the non-selective cation channels found previously in this cell type (Demidchik et al., 2002), was sometimes observed prior to the application of ATP (or BzBzATP) to the extracellular face of excised outside-out membrane patches. Its activity was insensitive to agonists. The application of 30 μm ATP (or BzBzATP) stimulated a transient burst of inwardly-directed positive current, which occurred up to 1 min after the estimated arrival of agonist at the membrane (ATP, three out of six patches, Figure 3(a); BzBzATP, seven out of 11 patches; Figure 3b). The burst duration was 17 ± 2 min with ATP (= 3), and 21 ± 1 min with BzBzATP (= 7), consistent with the time course of current activation measured from the entire plasma membrane. Up to four, single, seemingly identical, open channel states were observed early in the burst. The channel amplitude increased on the hyperpolarization of the membrane voltage (Figure 3c). The single channel conductances derived from the first open state were 19 ± 0.8 and 20 ± 1.9 pS, for ATP and BzBzATP, respectively (Figure 4a). The extrapolated reversal potentials were 19 ± 11 and 33 ± 4 mV, for ATP and BzBzATP, respectively. The displacement of the reversal potential away from the equilibrium potential for Cl (−36 mV), and towards that for Ca2+ (+154 mV), is indicative of Ca2+ permeability, in addition to possible Cl permeability. The open probability (Popen, estimated for the first open state at the height of the purine-activated burst) decreased over time (Figure 4b). Hyperpolarization increased Popen (Figure 4c), whereas 100 μm Gd3+ decreased it (Figure 4d). The Ca2+ permeability, voltage dependence, transient activation and pharmacological profile are fully consistent with this channel's contributing to the ATP-activated Ca2+ conductance.

Figure 3.

 Extracellular ATP activates single channels in excised outside-out plasma membrane patches from root epidermal protoplasts.
(a) Burst of channel opening in response to 30 μm ATP, with 20 mm external CaCl2; HP, holding potential.
(b) Burst of channel opening in response to 30 μm 2′-O-(4-benzoylbenzoyl) adenosine 5′-triphosphate (BzBzATP).
(c) The channel amplitude increased as the voltage was hyperpolarized (30 μm BzBzATP).

Figure 4.

 A 19-pS Ca2+-permeable channel is activated in response to extracellular ATP.
(a) Single-channel conductance of the first open state in response to 30 μm ATP and 30 μm 2′-O-(4-benzoylbenzoyl) adenosine 5′-triphosphate (BzBzATP).
(b) The open probability (Popen) declines in the continued presence of 30 μm ATP.
(c) The mean ± SE increases in Popen upon hyperpolarization (30 μm BzBzATP).
(d) The mean ± SE Popen at −100 mV is decreased by 100 μm extracellular Gd3+ (with the addition of 30 μm BzBzATP for activation). All trials were performed in the presence of 20 mm external CaCl2.

Extracellular ATP increases AtrbohC NADPH oxidase activity

Extracellular ATP causes the accumulation of ROS in roots (Kim et al., 2006) and leaves (Song et al., 2006). ROS increase plasma membrane hyperpolarization-activated Ca2+-permeable channel (HACC) activity in root and guard cells (Pei et al., 2000; Demidchik et al., 2003b, 2007; Foreman et al., 2003). Therefore, it was reasoned that the Ca2+ conductance was activated by ATP via the production of ROS. In a preliminary study to test whether ROS production in the root could respond to purine nucleotides, extracellular superoxide anion production was quantified through the reduction of the soluble tetrazolium dye, sodium 3′-[1-(phenylamino-carbonyl)-3,4-tetrazolium]-bis(4-methoxy-6-nitro) benzene-sulfonic acid hydrate (TXX; Sutherland and Learmouth, 1997; Able et al., 1998). This probe has been used previously to examine Ca2+-stimulated NADPH oxidase activity in Arabidopsis roots (Mortimer et al., 2008). Here, superoxide anion production by excised Arabidopsis roots was found to nearly double in response to 100 μm ADP (control, 28 ± 8 nmol superoxide μg−1 protein, = 6; ADP, 45 ± 6 nmol superoxide μg−1 protein, = 6). However, this approach could not be applied to single roots, and did not permit the spatial resolution of ROS production. Therefore, ROS imaging was used instead.

The application of ATP to individual Arabidopsis roots resulted in the rapid (response observed by 15 sec), dose-dependent (from 10 μm to 1 mm) accumulation of intracellular ROS (Figure 5a), detected by ester-loaded 5- and 6-chloromethyl-2′,7′-dichlorodihydrofluorescein (CM-H2DCF; Foreman et al., 2003; Shin and Schachtman, 2004). The accumulation of ROS occurred primarily in the mature root (Figure 5a). The fluorescence intensity over the length of single primary roots was 32 ± 3 fluorescence per pixel under control conditions, with a peak response of 117 ± 4 fluorescence per pixel evoked by 100 μm ATP (= 27; < 0.05, Tukey’s multiple comparison test). The response to 100 μm ATP was significantly inhibited by the NADPH oxidase inhibitor diphenylene iodonium (DPI; Bolwell and Wojtaszek, 1997; Foreman et al., 2003) (20 μm DPI; 15 ± 1 fluorescence per pixel, = 27; < 0.05; Figure 5a). Of the ten A. thaliana respiratory burst oxidase homologs (AtRBOH) encoding NADPH oxidases, AtRBOHC is expressed in the epidermis, and its loss-of-function mutant (root hair defective 2, rhd2) fails to produce ROS to activate channel-mediated Ca2+ influx in this tissue (Foreman et al., 2003). Basal levels of ROS in rhd2/AtrbohC roots were lower than that in the wild type (WT; 25 ± 1 fluorescence per pixel, = 27), and the mutant failed to accumulate ROS to WT levels in response to 100 μm ATP (45 ± 2 fluorescence per pixel, = 27; < 0.05; Figure 5b). The increase observed in the mutant is compatible with the known presence of AtRBOHD or AtRBOHF in the roots (Kwak et al., 2003), which in leaves can support the induction of ROS by ATP (Song et al., 2006), or the activation of alternative generators of ROS. AtRBOHC is also activated by increased [Ca2+]cyt (Takeda et al., 2008). Blocking the influx of Ca2+ with 300 μm Gd3+ caused a 52% inhibition of the WT ROS accumulation evoked by 100 μm ATP (56 ± 3 fluorescence per pixel, n = 27; < 0.05). Although Gd3+ was a very effective blocker of the ATP-induced Ca2+ conductance (Figures 2c and 4d), it still allowed some ATP-induced increase in [Ca2+]cyt in protoplasts (Figure 1c). This suggests that ATP causes an increase in [Ca2+]cyt through intracellular Ca2+ release, which could activate AtRBOHC to generate ROS, which in turn activate Ca2+ influx through the HACC conductance. This channel-mediated Ca2+ influx could then potentiate AtRBOHC activity.

Figure 5.

 The rhd2/AtrbohC NADPH oxidase mutant shows an impaired accumulation of reactive oxygen species (ROS) in response to extracellular ATP.
(a) Accumulation of ROS in a representative wild-type (WT) root in response to 100 μm ATP. Upper panels, bright-field and false-colour mapped images of the WT control. The mean ± SE level was 32 ± 3 fluorescence per pixel (= 27). Middle panels, WT response to 100 μm ATP, recorded 30 sec after application. The mean ± SE level was 117 ± 4 fluorescence per pixel (= 27). Lower panels, WT response to 100 μm ATP with 20 μm diphenylene iodonium (DPI), which is an NADPH oxidase inhibitor (Bolwell and Wojtaszek, 1997; Foreman et al., 2003). The mean ± SE level was 15 ± 1 fluorescence per pixel (= 27).
(b) The accumulation of ROS evoked by 100 μm ATP was impaired in rhd2/AtrbohC. The mean ± SE level was 45 ± 2 fluorescence per pixel (= 27). Scale bars for all panels: 2 mm.

The Ca2+ conductance lies downstream of ATP-activated AtRBOHC

Patch-clamp electrophysiology was used to determine whether the HACC conductance lies downstream of AtRBOHC. Under control conditions, the Ca2+ conductance in rhd2/AtrbohC root mature epidermal protoplasts was not significantly different to that in WT (whole-cell I at −200 mV: rhd2/AtrbohC, −203 ± 56 pA, = 6; WT, −217 ± 38 pA, n = 6). However, unlike in the WT, the Ca2+ conductance in rhd2/AtrbohC did not increase in response to extracellular ATP (Figure 6a; I at −200 mV; −195 ± 48 pA, = 6). This suggests that the WT channel activation was caused by increasingly oxidative conditions, which is consistent with ATP-induced ROS production by RHD2/AtRBOHC. Reducing conditions, imposed by 1 mm DTT at both membrane faces, caused WT mature epidermal protoplasts to phenocopy rhd2/AtrbohC, thereby preventing the increase in Ca2+ conductance by extracellular ATP (Figure 6b; n = 6). The results are consistent with a signalling pathway in the mature epidermis in which the stress levels of ATP activate AtRBOHC, and hence the hyperpolarization-activated Ca2+ channel conductance.

Figure 6.

 Ca2+ channel activity and transcription are impaired in rhd2/AtrbohC.
(a) ATP (20 μm) did not evoke Ca2+ channel activity in epidermal protoplasts from rhd2/AtrbohC roots (with the same recording conditions as described in Figure 2a). The whole-cell currents are shown above mean ± SE current–voltage relationships (= 6).
(b) Reducing conditions (1 mm DTT in bath and pipette solutions) prevented Ca2+ channel activation by 20 μm ATP in wild-type (WT) root epidermal protoplasts (= 6).
(c) Root MPK3 transcription did not increase in rhd2 upon 15 min of exposure to 1 mm ATP. Primers produced a 959-bp transcript.

ATP-induced transcription requires AtRBOHC

No difference in root cell death kinetics (Chivasa et al., 2005) was observed between the WT and rhd2/AtrbohC (data not shown), suggesting that this response is not governed by the ATP/AtRBOHC pathway. However, extracellular ATP in leaves stimulates the transcription of genes involved in stress signalling, such as AtMPK3 (Song et al., 2006). This ATP effect is dependent on Ca2+ influx (Song et al., 2006). Root AtMPK3 transcription is also upregulated by stress-related ROS (Kovtun et al., 2000; Rentel et al., 2004). As stress conditions such as wounding, osmotic stress and touch cause ATP release (Song et al., 2006; Roux and Steinebrunner, 2007), and as extracellular ATP causes NADPH oxidase-dependent ROS production, we reasoned that root AtMPK3 transcription lies downstream of the ATP activation of RHD2/AtRBOHC and the HACC. Application of ATP to rhd2/AtrbohC roots failed to cause increased AtMPK3 transcription (Figure 6c), confirming this premise.


The possibility that extracellular ATP has a regulatory role to play in plants has received relatively little attention since the initial studies were published. It is now clear that ATP is involved in plant growth and stress responses, but mechanistically our understanding of ‘how’ has remained largely at the black-box level. Whereas there are parallels with animal cells for mechanisms of ATP release and regulation of its extracellular concentration (Kim et al., 2006; Roux and Steinebrunner, 2007), a fundamental question of whether the wall is involved in plant ATP signalling has only now been answered. In the mature root epidermis, the wall is redundant in the elevation of [Ca2+]cyt, and ATP is sensed at the plasma membrane. This does not preclude the activity of cell wall proteins in other cell types, and the possibility of their involvement would certainly mark a distinctive split between plant and animal ATP signalling systems.

This study has shown that plant extracellular ATP results in the activation of plasma membrane NADPH oxidases. These enzymes catalyse the formation of the extracellular superoxide anion, and are implicated in plant development, response to abiotic stress and defence against pathogens (Kwak et al., 2003; Apel and Hirt, 2004; Carrol et al., 2005). The control of rboh NADPH oxidases in Arabidopisis is also exerted by ABA (Kwak et al., 2003); the relationship of ATP and ABA in rboh-mediated growth and stress responses must now be examined. In animal cells, plasma membrane P2 ionotropic receptors bind extracellular ATP, causing an influx of Ca2+, which acts as a second messenger to activate NADPH oxidase, resulting in a respiratory oxidative burst (Parvathenani et al., 2003). The application of La3+ (as a cation channel blocker) or the P2 receptor antagonist pyridoxalphosphate-6-azophenyl-2′,4′-disulfonic acid (PPADS) prevented superoxide production by extracellular ATP in Arabidopsis leaves (Song et al., 2006). As both lanthanides and PPADS were found previously to inhibit ATP-induced [Ca2+]cyt elevation in Arabidopsis roots (Demidchik et al., 2003a) and leaves (Jeter et al., 2004), it has been proposed that, as in animals, extracellular ATP activates a receptor in the plasma membrane that causes channel-mediated Ca2+ influx, and subsequently the elevated [Ca2+]cyt activates NADPH oxidases (Song et al., 2006). However, the present study on roots shows that plasma membrane Ca2+ channels operate downstream of an NADPH oxidase, and that it is most likely that an ATP-induced release of Ca2+ from internal stores lies at the beginning of the [Ca2+]cyt elevation. Thus, PPADS is more likely to affect an ATP-binding protein equivalent to a metabotropic, rather than an ionotropic, receptor. The combination of the plasma membrane AtRBOHC NADPH oxidase and its partner ROS-activated plasma membrane Ca2+ channels would increase the magnitude of the [Ca2+]cyt response, to trigger transcription.

NADPH oxidase and plasma membrane Ca2+ channel activity are involved in a range of plant developmental and environmental response pathways (Pei et al., 2000; Foreman et al., 2003; Apel and Hirt, 2004). The plasma membrane Ca2+ channels lying downstream of extracellular ATP are transiently activated, voltage dependent and appear to be redox sensitive. As they would open over the plasma membrane voltage range reported for the mature epidermis (Maathuis and Sanders, 1993), they would facilitate transient Ca2+ influx, and contribute to [Ca2+]cyt elevation. With a single-channel conductance of 19–20 pS, these channels appear to be distinct from the 15-pS Ca2+-permeable channel reported previously in the plasma membrane of the mature root epidermis, under an identical transmembrane Ca2+ gradient (Demidchik et al., 2007). The 15-pS channel was also hyperpolarization activated, and its Popen increased transiently in response to intracellular H2O2 (Demidchik et al., 2007). The identity of the ROS activating the 19–20-pS channel is unknown at present, and it is also feasible that ROS activation could be indirect.

The reliance of ATP-induced AtMPK3 transcription on the AtRBOHC NADPH oxidase identifies a signalling pathway that operates in stress adaptation (Figure 7). AtMPK3 is part of a stress-responsive phosphorylation cascade initiated in some cases by the activity of ANP1, a mitogen-activated protein kinase kinase kinase (Kovtun et al., 2000). Touch, low temperature and salinity responses involve AtMPK3 (Mizoguchi et al., 1996), and it functions downstream of the oxidative signal-inducible 1 (OXI1) kinase involved in root growth and basal resistance to pathogens (Rentel et al., 2004). ATP is released by Arabidopsis in response to hyper- and hypo-osmotic stress (Jeter et al., 2004). AtRBOHC is implicated in the response to hypo-osmotic shock (Macpherson et al., 2008), and AtMPK3 is known to be involved in hypo-osmotic and moderately hyperosmotic stress responses, where its transcription is dependent on Ca2+ influx (Droillard et al., 2002). The results obtained here for roots, and previously for the Ca2+ influx-dependent ATP induction of AtMPK3 transcription in leaves (Song et al., 2006), identify extracellular ATP, released during stress, as a regulator of the MPK cascade. In the present study, roots were grown on medium with an osmolarity of 115 mOsM, and were then immersed in assay medium with an osmolarity of approximately 11 mOsM for non-invasive slowly vibrating microelectrode (MIFE; microelectrode ion flux estimate) determinations of net Ca2+ influx, ROS imaging and transcript analysis. This represents a hypo-osmotic stress that could have ‘primed’ the system, but at present, the implications for the magnitude of the subsequent response to applied ATP are unknown.

Figure 7.

 Schematic model of a possible pathway for the induction of MPK3 transcription by extracellular ATP. The putative ATP-binding protein is depicted as residing in the plasma membrane (PM). The perception of ATP triggers (by an unknown mechanism) a release of intracellular Ca2+ from stores such as the vacuole, endoplasmic reticulum or mitochondria. Elevated intracellular cytosolic free Ca2+, [Ca2+]cyt, stimulates AtRBOHC NADPH oxidase activity via its EF hands (Takeda et al., 2008). The superoxide anion would then be produced extracellularly, and converted to other ROS, such as H2O2 and hydroxyl radicals (Foreman et al., 2003). Peroxide could enter the cytosol through PM aquaporins (Bienert et al., 2007; Dynowski et al., 2008) and contribute to the accumulation of intracellular ROS shown in Figure 5. The identity of the ROS activating the PM Ca2+-permeable channel are unknown at present, but activation could be at the extra- or intracellular membrane face (Demidchik et al., 2003b, 2007). Channel opening would further increase [Ca2+]cyt, leading to further NADPH oxidase stimulation and the induction of MPK3 transcription (Droillard et al., 2002; Song et al., 2006). Intracellular ROS could also contribute to MPK3 transcription (Kovtun et al., 2000; Rentel et al., 2004).

The identity of the plasma membrane receptor for ATP remains enigmatic. Despite the recent breakthrough findings of P2X equivalents in the amoeba Dictyostelium and the unicellular green alga Ostreococcus tauri, neither of these function at the plasma membrane (Fountain et al., 2007, 2008), and they offer no clues to the identity of higher plant receptors. Higher plant genomes appear bereft of P2 homologues. This study has shown that higher plants could have evolved a distinct mechanism to transduce the ATP signal at the plasma membrane. Isolation of the ATP receptor clearly cannot rely on an animal or even algal paradigm, and represents a challenging task. However, the strategies used successfully to identify ligand–receptor interactions in plant membranes (Yamaguchi et al., 2006) may yet prove fruitful.

Experimental procedures

Growth conditions and measurements

Arabidopsis thaliana (Col-0 and rhd2/AtrbohC from laboratory stock) was grown aseptically and vertically at 22°C for 5–15 days (16-h day, with 100 μmol m−2 sec−1 irradiance) on medium comprising 0.3% (w/v) Phytagel (Sigma-Aldrich,, full-strength MS (Duchefa, medium and 1% (w/v) sucrose. The rhd2/AtrbohC mutant is defective in the At5g1060 gene (GenBank accession number AF055355; TIGR,, and four mutant alleles have been defined (Foreman et al., 2003). The rhd2-4 mutant is restored by complementation (Foreman et al., 2003), and was used here. For growth determination, plants were grown on 5 mm KNO3, 5 mm MgSO4, 5 mm 2-(N-morpholino)-ethanesulphonic acid (MES), 2 mm NaCl, 1 mm CaCl2, 1 mm (NH4)2HPO4 and 1% (w/v) sucrose, supplemented with vitamins and micronutrients, and adjusted to pH 5.7 with bis-tris propane. Free ion contents were estimated with GEOCHEM (Parker et al., 1995) and Maxchelator (Patton et al., 2004). The root lengths were measured using ImageJ software ( at the 6–8-day interval. Test additions did not affect germination.

Net Ca2+ fluxes

Net Ca2+ fluxes were measured using the MIFE technique (Shabala et al., 1997). Excised roots (5–7 mm from the apex) were mounted in a perspex chamber by an agar drop, and immersed in 0.5 ml of assay solution comprising 0.1 mm KCl, 0.1 mm CaCl2, 4 mm MES and 2 mm Tris base (pH 6.0). The root was viewed under an Olympus I×50 microscope (Olympus, using 200× magnification. Ion-selective microelectrodes with an external tip diameter of approximately 3 μm were manufactured and calibrated as described previously (Shabala et al., 1997). Electrodes were calibrated before and after use. Purines (tested up to 1 mm; Sigma-Aldrich) did not affect calibration values or electrode response times. The microelectrode was placed 20 μm above the root surface. During measurements, the electrode was moved between two positions, 20 and 50 μm above the root surface, in a square-wave manner with a 5-sec half cycle. Measurements of mature epidermal flux were taken 1–1.5 mm from the root apex. The net Ca2+ flux was measured for 5 min prior to the addition of 0.5 ml of assay solution; Ca2+ was maintained at 0.1 mm, estimated using Maxchelator (Patton et al., 2004), and confirmed using a Ca2+-selective microelectrode in root-free assay. Measurements were resumed after 20–30 sec, when unstirred layer conditions were reached.

[Ca2+]cyt measurement

Protoplasts were isolated from the mature root epidermis of Arabidopsis expressing cytosolic (apo)aequorin (Cauliflower mosaic virus, CaMV, 35S promoter) using the protocol described by Demidchik et al. (2007), with the modification that incubation with wall-degrading enzymes was increased from 50 to 60 min. Protoplasts were held for 3–4 h (in the dark, at 28°C) in recording medium (1 mm CaCl2, 1 mm KCl, 1 mm MgCl2, 10 mm glucose, pH 6.0, 2 mm Tris base/MES; 290–300 mOsM, with d-sorbitol) with 10 μm coelenterazine (NanoLight Technology, After washing, recording medium (without coelentrazine) and standard luminometry procedures were used (Demidchik et al., 2007). The discharging solution contained 2 m CaCl2 and 20% (w/v) ethanol, and [Ca2+]cyt values were estimated according to the method described by Fricker et al. (1999).


Protoplasts used for patch clamping were 20 ± 1.5 μm in diameter. For ATP studies, the pipette solution contained 40 mm K-gluconate, 10 mm KCl and 1 mm 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid (free Ca2+ to 100 nm with 0.4 mm CaCl2), pH 7.2 (2 mm Tris/MES). For BzBzATP, Na-gluconate and NaCl replaced K-gluconate and KCl. The control bathing solution comprised 20 mm CaCl2, 0.1 mm KCl and 0.3 mm NaCl (varied to supply equimolar Na+ to that in the purine-containing test), pH 6 (2 mm Tris/4 mm MES). Solutions were adjusted to 290–300 mOsM with d-sorbitol. The patch-clamp procedures were carried out as described previously (Demidchik et al., 2003b, 2007). The current–voltage relationships were recorded within 1 min.

Imaging and transcription

The ROS imaging with 50 μm 5- and 6-chloromethyl-2′,7′-dichlorohydrofluorscein diacetate, acetyl ester (Molecular Probes, now part of Invitrogen, was performed as described previously by Shin and Schachtman (2004). For imaging and transcription studies, WT and rhd2/AtrohC roots were treated with control and ATP-containing solutions, as used for determining net Ca2+ influx. DPI was purchased from Sigma-Aldrich. The images of roots were acquired with an SMZ 1500 microscope (Nikon, and a Retiga cooled 12-bit camera (Qimaging, The preliminary study of superoxide anion production was performed according to the method described by Mortimer et al. (2008). Roots were excised from approximately 30 seedlings, and submerged in 10 mm phosphate buffer with 0.5 mm CaCl2 and 500 μm XTT (Sigma-Aldrich) (Sutherland and Learmouth, 1997; Able et al., 1998), pH 6.0, with or without 100 μm ADP. Roots were incubated at 21°C in the dark for 30 min and the sample absorbance at 470 nm was measured (1-cm path length; Helios γ spectrophotometer; Unicam, Superoxide production was estimated using the XTT extinction co-efficient of 2.16 × 104 (mol L−1)−1 cm−1. Roots were snap-frozen in liquid nitrogen and homogenized with buffer comprising 50 mm Tris acetate, 100 mm potassium acetate, 1 mm EGTA, 1 mm DTT and 20% (v/v) glycerol. The sample was clarified by centrifugation for 10 min at 4°C, and the supernatant was assayed for protein content (Bio-Rad, The xanthine oxidase/hypoxanthine (XO/HX) superoxide anion-generating system (Suzuki, 2000) was used as a cell-free check on the ability of ADP to perturb the XTT assay. XO (Sigma-Aldrich) was freshly prepared in experimental buffer at 0.5–5 U ml−1. HX (Sigma-Aldrich) was added to a final concentration of 500 μm with or without ADP.

RNA was extracted from excised roots (RNEasy; Qiagen, cDNA was prepared using reverse transcriptase (BioScript; Bioline,, and PCR was carried out with the Arabidopsis actin 8 gene (At1g49240; forward primer, 5′-AGAAAGATGCGTATGTTGGTGA-3′; reverse primer, 5′-CTGCTGGAAAGTGCTGAGGGAA-3′). The number of cycles and levels of cDNA required to amplify the product during the exponential phase, and at an equal level between all wild-type, mutant and ATP-treated samples, were determined. RT-PCR was then carried out for MPK3 (At3g45640; forward primer, 5′-TGTTGGATACGGAGACGA-3′; reverse primer, 5′-GCTGCACTTCTAACCGTA-3′) for comparison of MPK3 transcript levels between treatments. These primers produce a 959-bp transcript.


The study was funded by the Leverhulme Trust, the BBSRC, the Royal Society, the Australian Research Council, the Fulbright Foundation and the Monsanto Company.