Research article

Changes in cytosolic Mg2+ levels can regulate the activity of the plasma membrane H+-ATPase in maize

Stefan Hanstein, Xiaozhi Wang, Xiaoqing Qian, Peter Friedhoff, Ammara Fatima, Yuhua Shan, Ke Feng, Sven Schubert


Plant PM (plasma membrane) H+-ATPase, a major consumer of cellular ATP, is driven by the MgATP complex which may dissociate at low cytosolic Mg2+ activity. We investigated whether hydrolytic activity of PM H+-ATPase is inhibited at ATP concentrations exceeding the Mg2+ concentration. Activity in isolated maize PMs was measured at pH 6.5 in the presence of 5 mM Mg2+ (high) or 2 mM Mg2+ (low), whereas K+ was applied at concentrations of 155 mM (high) or 55 mM (low). In all experiments, with membrane vesicles either from roots or leaves, the enzyme activity decreased in the presence of Mg2+-free ATP. At inhibitory ATP concentrations, the activity was not influenced by the K+ concentration. The activity was restored after increasing the Mg2+ concentration. ATP inhibition also occurred at pH 7.5. Kinetic modelling shows that Mg2+-free ATP acted as a competitive inhibitor with a Ki in the range of the Km. Ki decreased by 75% at low K+ concentration. Ki was one order of magnitude lower at pH 7.5 compared with pH 6.5. The observed inhibition is consistent with a concept in which down-regulation of the cytosolic Mg2+ activity is involved in (phyto)hormonal stress responses.

  • free magnesium
  • Mg2+ activity
  • nucleotide binding
  • nucleotide inhibition
  • P-type ATPase


Plant PM (plasma membrane) H+-ATPase is the ‘powerhouse’ for nutrient uptake, turgor formation and cellular growth [1] pumping protons from the cytosol into the extracellular space (apoplast). Under resting conditions, enzyme activity is regulated by an autoinhibitory domain [1,2] and probably by reduced coupling efficiency [3]. The crucial physiological role of down-regulation is evident from fusicoccin effects. This phytotoxin binds to 14-3-3 proteins, inactivates the autoinhibitory domain and causes a persistent activation of PM H+-ATPase, leading to abnormal apoplastic acidification and stomatal opening [4].

The enzyme is driven by the MgATP complex. A simple type of down-regulation or inhibition which does not involve protein–protein interaction could arise from an imbalance between the molar concentrations of Mg2+ and ATP: ATP in excess of Mg2+ inhibits PM H+-ATPase of yeast [5] and of Acetabularia [6]. ATP is considered to be the most abundant intracellular chelating component of Mg2+ in mammalian and plant cells [7,8]. The Kd for MgATP2−, the prevailing MgATP species at normal physiological pH, is 28 μM at 25 °C [9]. Taking into account that ATP species with lower Mg2+ affinity [mainly (H-ATP)3−] and univalent cations competing with Mg2+ for free ATP are also present, an apparent Kd of 71 μM was determined at pH 7.4 [10]. If the latter value is correct, then the Mg2+ activity that could regulate H+-ATPase activity would be higher than 30 μM. Direct measurements of cytosolic Mg2+ activity in the plant cytosol are difficult. However, for wheat protoplasts, the cytosolic Mg2+ activity was estimated from measured adenylate concentrations to be 400 μM in the dark and 200 μM in the light [8]. Poor Mg2+ supply from the soil and intracellular processes such as Mg2+ sequestration, e.g. with phytic acid (myo-inositol hexakisphosphate), a common constituent in eukaryotic cells and probably a major fraction of the cellular phosphate pool [11,12], or Mg2+ complexation in chloroplasts [13], may keep or force the cytosolic Mg2+ activity to lower levels which then leave a major fraction of ATP uncomplexed. Under such conditions, an inhibition mechanism of PM H+-ATPase by Mg2+-free ATP with a sufficiently low inhibition constant may play a physiological role. The aim of the present study was to prove ATP inhibition of PM H+-ATPase in higher plants and to determine the kinetic parameters of ATP inhibition.

Previous kinetic studies of PM H+-ATPase from oat roots have demonstrated that ATP hydrolysis falls to 7% when Mg2+ is omitted from the assay [14]. With Mg2+ in excess of ATP, the kinetics were hyperbolic and a Km for MgATP of 0.3 mM was derived from double-reciprocal plots. Our investigation follows the consequences of a moderate decrease in Mg2+ concentration down to a level of 30% of the ATP concentration, but still severalfold higher than the Km in roots of oat and maize [15]. We show the non-hyperbolic kinetics of maize H+-ATPase in PM vesicles [1517] under various ionic conditions. Because of non-hyperbolic kinetics, non-linear regression analysis based on mechanistic models [18,19] was employed to determine Km values and constants of ATP inhibition. The discussion of physiological relevance includes a specification of important regulatory processes which may be accompanied by a decrease in cytosolic Mg2+ activity.


Plant cultivation

Maize seeds (Zea mays L. cv. Pioneer 3906) were germinated in the dark at 20 °C for 10 days. Seedlings were transferred to nutrient solution [20] which was half-strength for 2 days, then full-strength and replaced every 3 days. Cultivar Pioneer was cultivated in the growth chamber for 16 days at a light intensity of 130 W·m−2 (Philips Master HPI-T Plus 400W) at 50% relative humidity, with a 16 h light/8 h dark cycle at 21 °C/16 °C. For the experiment on discrimination between inhibition mechanisms, Zea mays L. cv. Amadeo (KWS) was grown hydroponically for 14 days with a 16 h light/8 h dark cycle at 26 °C/18 °C.

PM isolation

PM vesicles from the lower 10 cm of roots and from young leaves were prepared by aqueous two-phase partitioning [15,16,20]. Shoots were cut above the fourth leaf and immediately placed in a beaker with the cut end immersed in cold water. The lower segment of 10 cm length was cut and the mid-rib was peeled top-down and discarded. The remaining blade tissue was homogenized in the presence of 250 mM potassium iodide in order to remove peripheral membrane proteins [2123]. Potassium iodide was not used for preparation of root vesicles [15]. Phase partitioning was performed with a polymer concentration of 6.2% for roots and 6.1% for leaves [20]. Vesicle aliquots were stored in liquid N2. Protein was quantified using the method of Bradford [24]. The purity of vesicle preparation was confirmed using specific ATPase inhibitors [15,25].

Kinetic assays

In order to suppress hydrolytic activity originating from traces of non-specific phosphatases, all assays were performed in the presence of 1 mM Na2MoO4. Interference from ATP hydrolysis originating from other membranes was eliminated with two complementary methods. (i) In the vanadate method, the decrease in ATPase activity upon addition of 300 μM Na3VO4 was taken as PM ATPase activity [15,23,25]. This method has the advantage that non-physiologically high KNO3 concentrations and azide are avoided. However, since experimental errors of two hydrolytic assays contribute to final experimental error, experimental variation is inherently larger. Moreover, a higher amount of vesicle sample is required. (ii) In the cocktail method, ATPase activity in the presence of 1 mM Na2MoO4, 1 mM NaN3 and 100 mM KNO3 was considered as PM ATPase activity. Since there was no NO3-sensitive activity in leaf vesicles, KNO3 was not required in kinetic assays of leaves.

ATPase hydrolytic activity was determined using the method of Yan et al. [15] at 30 °C with 3 μg of vesicle protein in 0.5 ml of 30 mM BTP {1,3-bis[tris(hydroxymethyl)methylamino]propane}/Mes buffer with 0.02% Brij 58 (Sigma), 2 or 5 mM MgSO4 and potassium salts. Experiments at 155 mM K+ were performed with 50 mM KCl, 100 mM KNO3 and 5 mM potassium PEP (phosphoenolpyruvate) (in root and some leaf experiments) or with 150 mM KCl and 5 mM potassium PEP in a leaf experiment for discriminating between inhibitory ATP species. ATP was supplied as Na2ATP or di-Tris-ATP (both from Sigma), adjusted with BTP to the assay pH. At maximum activity, the increase in the concentration of Pi throughout the incubation time did not exceed 3% of the ATP concentration. Unless otherwise stated, formation of ADP and decrease in ATP was prevented by an ATP-regenerating system [26], which included 5 units of pyruvate kinase (Fluka) and 5 mM potassium PEP (Fluka) or 5 mM trisodium PEP (Serva). With the proton/cation exchanger gramicidin for preventing H+ gradient across the PM, the ATPase hydrolytic activity of root vesicles was 0.284 μmol·mg−1·min−1 compared with 0.271 without gramicidin (n=3). Unless otherwise indicated, ATPase assays were performed without gramicidin.

All enzyme assays were performed with two chemical parallels. Three or four vesicle preparations from parallel plant cultivations (biological replicates) were used to determine mean±S.E.M. values of ATPase activity.

Normalization of ATPase activities

The normalization procedure applied for some datasets as indicated eliminates differences between replicates, if for all ATP concentrations activities are higher (or lower) by a constant factor. For example, such differences occur when enzyme concentrations slightly differ between stored vesicle ‘aliquots’ for replicates. For each replicate, normalized activities were determined as follows: (i) calculating the sum of activities over the whole range of ATP concentrations for each replicate (individual sum), (ii) determining the mean sum of all replicates, and (iii) multiplying the activity values of a replicate with the ratio between the mean sum and the individual sum of that replicate.

Non-linear regression analysis with DynaFit

The principle of the analysis is to define a mechanistic model from which DynaFit derives a mathematical model, i.e. a set of simultaneous rate equations including the unknown rate constants [27] (see Supplementary Figure S1 at The mathematical model is then solved with numerical methods using iterative procedures. Mean values of non-normalized activity data were subjected to DynaFit regression analysis from which rate constants and their S.D. values were obtained. Regression curves are shown in the Figures. If the legend does not specify the inhibition mechanism which was applied for curve fitting, curves did not differ for competitive, uncompetitive and non-competitive inhibition. The fit of different mechanistic models was compared (model discrimination) using the Akaike information criterion [28]. The most plausible model is one with the highest Akaike weight (Max=1). The probability of one model in comparison with another model was calculated using the method of Mannervik [29].

DynaFit analysis was based on concentration values in μM and rate constants per s (see Supplementary Figure S1). Enzyme concentration (E) was set to 0.1 nM, assuming that 1 mg of PM protein contains 1% H+-ATPases with a relative molecular mass of 100000. The start value for fitting the catalytic rate constant (turnover number, kr) was 3500 min−1≈58 s−1. Koland and Hammes [5] determined a kr of 59 μmol·min−1 mg−1 for reconstituted yeast PM H+-ATPase. A 1 mg amount of a protein with a relative molecular mass of 100000 Da corresponds to 10 nmol. Thus this rate constant can be expressed as 5900 min−1≈100 s−1. Using a higher enzyme concentration and correspondingly lower kr start value did not change the values of Km and Ki calculated. Activity data were processed with the dynamic method based on the simulation of pre-steady-state dynamics of the biochemical system [27]. In contrast with the rapid-equilibrium method, this method is also valid when equilibration between enzyme and enzyme–substrate complex is slow compared with catalytic turnover. Equilibration time was set to 3 s (see Supplementary Figure S1, section [progress]).

Discrimination between ATP species

In order to investigate whether a single ATP species, e.g. (K-ATP)3−, is more inhibitory compared with others, inhibition in the presence of 155 mM K+ was compared with inhibition occurring at 55 mM K+ {Kd for (K-ATP)3−=68 mM [9]}. Non-linear regression analysis was performed considering three cases of inhibition which differed with respect to the assumed inhibitory ATP species: (K-ATP)3−, (H-ATP)3− or ATP4− (see Supplementary Figure S2 at For discriminating between these ATP species, the set of rate equations was extended by two equilibrium reactions for ATP4− association with H+ and K+ [9].

Unless otherwise indicated, inhibition constants refer to the sum of all Mg2+-free ATP species. If inhibition is attributed to a single ATP species, Ki(1) of this species will be according to the relationship: Embedded Image where Ki(A) is the Ki assuming that all Mg2+-free ATP acts inhibitorily, c(1) is the concentration of the ATP species and c(A) is the concentration sum of all Mg2+-free ATP (tested in Supplementary Table S1 at


H+-ATPase kinetics in root membranes

ATP was applied up to a concentration of 5 mM in the presence of 5 mM Mg2+. The kinetics showed a broad maximum between 1 and 3 mM with a slight decline at the highest ATP concentrations (Figure 1A). After normalization of data, S.E.M. values at 1–2 mM ATP were reduced (Figure 1A, lower error bars compared with upper), but the decrease in activity at 5 mM ATP compared with at 2 mM ATP remained insignificant (Student's t test). Four kinetic models were analysed with DynaFit with respect to their fit to the experimental data: three models of inhibition by Mg2+-free ATP (competitive, uncompetitive and non-competitive) and the Michaelis–Menten model (see Supplementary Figure S1). The probability of any of the inhibition models compared with the Michaelis–Menten model was 99.9%. Regression curves were identical for all models of inhibition (Figure 1A). Akaike weights were 0.332 for each of the inhibition models (see Supplementary Table S2 at In a second experiment, the range of ATP concentrations was extended to 13 mM. At this concentration, activity decreased to 57% of maximum activity (Figure 1B, and see non-normalized data in Supplementary Table S3 at Again, competitive, uncompetitive and non-competitive mechanisms all yielded the same fits of regression curves and identical Akaike weights of 0.333. Thus it was not possible to derive the mechanism of inhibition. For this purpose, a third experiment was conducted in which ATPase kinetics were measured at fixed concentration of the inhibitor, i.e. of Mg2+-free ATP (Figure 1C, thick line). In the case of competitive inhibition, the relative decrease in activity caused by the inhibitor should become smaller with increasing concentration of the substrate, i.e. of MgATP. Activity without inhibition was measured in the presence of 3 mM Mg2+ in excess of total ATP (Figure 1C, thin line). With increasing concentration of MgATP, relative inhibition decreased continuously from 85% at 0.1 mM MgATP to 23% at 7 mM MgATP. Indeed, model discrimination with DynaFit yielded the highest possible Akaike weight of 1 for competitive inhibition and Akaike weights of 0 for uncompetitive and non-competitive inhibition (DynaFit script and output for model discrimination in Supplementary Figure S3 at The probability of competitive inhibition in comparison with each of the other mechanisms was 100%.

Figure 1 Effect of Mg2+-free ATP on PM ATPase activity from roots

Assays were performed with vesicles (cytosolic side outside) at pH optimum (6.5) using Na2ATP adjusted to pH 6.5. The ADP released was converted into ATP with pyruvate kinase and PEP. (A) Experiment at ATP concentrations up to 5 mM at 5 mM Mg2+ and 55 mM K+. Upper error bar denotes S.E.M. of original data; lower error bar denotes S.E.M. of normalized data (n=3). Regression curves by DynaFit assuming Michaelis–Menten kinetics (broken line) or inhibition by Mg2+-free ATP (continuous line, script in Supplementary Figure S1 at Curves were identical for competitive, uncompetitive and non-competitive inhibition. (B) Experiment at ATP concentrations up to 13 mM at 5 mM Mg2+ and 155 mM K+. Gramicidin was used to prevent a pH gradient at the vesicle membrane. Results are means+S.E.M. of normalized data (n=3). Regression curves calculated by DynaFit were identical for all inhibition types tested. (C) Mg2+-free ATP is a competitive inhibitor. Thick line: activity in the presence of 2 mM Mg2+-free ATP. Thin line: activity with 3 mM Mg2+ in excess of total ATP at all ATP concentrations (no inhibition). K+ concentration was 155 mM. Results are means±S.E.M. of original data (n=3). Regression curves by DynaFit assuming competitive inhibition.

The shapes of the activity curves differed slightly at low ATP concentrations as indicated by the arrows in Figures 1(A) and 1(B). In the first experiment (Figure 1A), maximum activity was virtually attained at 1 mM ATP, whereas, in the second experiment (Figure 1B), activity at 1 mM ATP clearly remained below maximum activity. The first experiment was performed at a K+ concentration of 55 mM (vanadate method, see the Experimental section), whereas the second experiment was conducted at 155 mM K+ (cocktail method with additional 100 mM KNO3).

H+-ATPase kinetics in leaf membranes

PM ATPase activity was measured using the cocktail method. Compared with roots, a higher ATP concentration of 4.5 mM was required to reach maximum activity of PM ATPase (Figure 2A, and see non-normalized data in Supplementary Table S3). ATPase activity significantly decreased by 33% at 8.9 mM ATP. At 13.4 mM ATP, activity was only 47% of the observed maximum activity. Thus activity was the same at 13.4 mM ATP and 0.6 mM ATP.

Figure 2 Effect of Mg2+-free ATP on PM ATPase activity from leaves

Assays were performed with ATP regeneration at pH 6.5 and at 5 mM Mg2+. (A) Inhibition with two different ATP reagents: Na2ATP (●) and di-Tris-ATP (□). K+ concentration was 155 mM. Results are means+S.E.M. of normalized data (n=3). For di-Tris-ATP S.E.M. was omitted for clarity and is shown in (B). Regression curves calculated by DynaFit were identical for all inhibition types tested. (B) Inhibition at different K+ concentrations: 155 mM (□) and 55 mM (▲). ATP was provided as di-Tris-ATP. Data from the experiment at 155 mM K+ is shown in (A). Results are means±S.E.M. (n=3).

Influence of cations on ATP inhibition

ATP in excess of Mg2+ is partly protonated and partly bound to K+ with a Kd of 68 mM [9]. In the search for the inhibitory ATP species, it was investigated whether inhibition was modified when the K+ concentration was reduced from 155 mM to 55 mM. Figure 2(B) shows that the activities at high ATP concentrations were not changed by K+. However, the decrease in KCl concentration caused an increase in activity at low ATP concentrations (thick line compared with thin line). In this experiment, ATP was supplied as metal-free ATP (di-Tris-ATP), whereas Na2ATP was used in previous experiments. Therefore the experiment also showed that sodium was not responsible for the inhibition at high ATP concentration. Furthermore, the experiment demonstrated that inhibition was not related to high ionic strength imposed by increasing the ATP concentration.

Upon increasing Mg2+ concentration from 5 mM to 15 mM, the decrease in ATPase activity at high ATP concentration was prevented (Figure 3A). On the other hand, when the Mg2+ concentration was reduced to 2 mM, inhibition of PM ATPase began at lower ATP concentrations (Figure 3B).

Figure 3 Influence of [Mg2+]

(A) Recovery of ATPase activity after addition of Mg2+. The assay was performed with vesicles from leaves at Mg2+ concentrations of 5 mM (□) and 15 mM (●) with 150 mM K+ at pH 6.5. ATP was provided as di-Tris-ATP. The two Mg2+ concentrations were tested in the same assay. No ATP-regenerating system was used. (B) Inhibition of PM ATPase activity at low Mg2+ concentration. Assays were performed with vesicles from leaves at 2 mM Mg2+ and 155 mM K+ at pH 6.5. ATP was provided as di-Tris-ATP. Results are means±S.E.M. (n=4).

In the absence of free Mg2+, the surplus of ATP may bind to other complexing agents present in the assay solution. Since complex formation between the bivalent buffer cation BTP and ATP and inhibitory action of ATP–BTP could not be ruled out, we used the univalent Tris instead of BTP. Figure 4(A) shows that inhibition also occurred in the absence of BTP. Inhibition was also observed at pH 7.5 (Figure 4B).

Figure 4 Inhibitory effect of Mg2+-free ATP with different buffers

(A) Inhibition in the presence of bivalent (BTP, ●) or univalent (Tris, □) amino buffer cations. The concentration of buffer cations was 30 mM. In both cases, Mes was used as acidic buffer component. The assay was performed with vesicles from leaves at 5 mM Mg2+, 55 mM K+ with di-Tris-ATP at pH 6.5. The two buffer cations were tested in the same assay procedure. Regression curve refers to BTP. (B) Inhibition at pH 7.5 (○) in comparison with pH 6.5 (▲, from Figure 2B). Assays were performed with vesicles from leaves in BTP buffer at 5 mM Mg2+ and 55 mM K+ with di-Tris-ATP. Results are means+S.E.M. (n=2).

Kinetic constants

Kinetic constants were derived from non-linear regression analysis (Table 1). For kr and Km, only values based on the model of competitive inhibition are shown since these constants were very similar for all models (deviation less than 3%). The rate of ATP hydrolysis of PM H+-ATPase is known to increase with K+ concentration. Usually, a test range of up to 50 mM K+ is employed [3]. For our leaf preparations kr was 33% higher at 155 mM K+ compared with 55 mM K+. However, this increase in K+ concentration did not change the kr of root preparations (Table 1). The Km for leaf PM ATPase was similar with Na2ATP and with metal-free ATP (977±70 μM and 907±155 μM, n=3). When the concentration of Mg2+ was decreased from 5 mM to 2 mM in leaf preparations at 155 mM K+, the Km decreased from 907±155 μM (n=3) to 340±77 μM (n=4). When the concentration of K+ was decreased from 155 mM to 55 mM, at 5 mM Mg2+, the Km decreased from 907±155 μM to 128±30 μM in leaf preparations and from 410±46 μM to 203±28 μM in root preparations (n=3). For leaves, the low Km at 55 mM K+ was confirmed by measuring the activity at ATP concentrations of 50, 100, 200, 600, 1000 and 3000 μM using Na2ATP. Km was 92±20 μM (n=3).

View this table:
Table 1 Kinetic constants of PM ATPase from maize roots and leaves

Assays were performed with different ATP reagents under the specified ionic conditions. Catalytic rate constant kr (s−1), Michaelis–Menten constant Km (μM) and inhibition constant Ki (μM) were calculated assuming competitive (c), uncompetitive (u) and non-competitive (n) inhibition without discriminating between ATP species (means±S.D.). kr and Km refer to (c), but deviation with (u) and (n) was less than 3% (see data in Supplementary Table S2 at In lines ‘3sep’ and ‘4sep’, the mean±S.D. value was calculated from kinetic constants determined separately for each replicate. n.d., not determined because of small range of inhibitory ATP concentrations (see the Results).

Inhibition constants shown in Table 1 refer to the sum of Mg2+-free ATP species. As illustrated in Figure 1(C), Ki values assuming competitive inhibition are most probable. It is evident that, in leaves, the high K+ concentration strongly increased Ki (Table 1, compare experiment 4 with experiment 5). At high K+ concentration, leaf Ki exceeded root Ki (Table 1, compare experiment 3 with experiment 2). However, leaf Km also exceeded root Km. It is clear that this change in Km is also relevant to the degree of inhibition in a mechanism in which substrate and inhibitor compete for the same site at the enzyme: at given concentrations of substrate and inhibitor, an increase in Km without a change in Ki decreases the activity. Thus, in comparing leaves and roots, the ratio between Ki and Km is an additional instructive parameter: at saturating substrate concentration, the competitive mechanism of inhibition allows us to relate the ratio between Ki and Km to the Mg2+ activity at which 50% inhibition by Mg2+-free ATP would occur, as explained below. For competitive inhibition, the relationship between the inhibitor concentration for 50% inhibition ([ATP]50) and the substrate concentration [MgATP] is given by eqn (9) of Cortes et al. [30]: Embedded Image

Dividing by [MgATP] yields: Embedded Image

At saturating substrate concentrations, 1/[MgATP] becomes negligible on the right side of the equation and: Embedded Image Since the dissociation equilibrium of MgATP is given by: Embedded Image where all concentrations refer to the free species, it follows that: Embedded Image Thus Ki/Km is inversely related to the free Mg2+ concentration at which 50% inhibition by Mg2+-free ATP occurs. A ratio of 1.92 was obtained for root preparations, whereas ratios were between 0.32 and 1.23 for leaf preparations. The ratio decreased by 50% when ATP was supplied as metal-free ATP instead of as Na2ATP. If inhibition were uncompetitive or non-competitive, Ki values would be in the order of 5–10 mM at pH 6.5 with 5 mM Mg2+ (Table 1).


All experiments in the present study showed that ATP in excess of Mg2+ causes a down-regulation of PM ATPase. A first explanation could be that the enzyme is inhibited by kinases. Kinase action is, without doubt, a crucial mechanism to regulate the activity of PM H+-ATPase in vivo under a variety of stress conditions such as osmotic stress or salt stress [13,31].

Consideration of kinase action

Ca2+-dependent kinases [32] can be ruled out, since our assays were performed without Ca2+. The involvement of a PKS5-like kinase [33], which may be present in the vesicle membrane in close vicinity to H+-ATPase [34], is very unlikely because of three arguments.

(i) Usually kinase activity also requires Mg2+. Our data demonstrate that inhibition increased while the MgATP concentration was constant. This means that the kinase, e.g. PKS5 [33], should utilize a Mg2+-free ATP species as substrate. It is generally accepted that the conserved catalytic domain of protein kinases binds MgATP or MnATP [35,36]. Binding of the bivalent cation neutralizes the negative charges of the nucleotide, in order to facilitate phosphoryl transfer [35]. Members of the SNF1-related kinase family, to which PKS5 belongs, require Mn2+ or Mg2+ for phosphorylation. PKS5 is not active in kinase assays without Mn2+ or Mg2+ [33]. PKS5 activity was stopped by chelating Mg2+ or Mn2+ in the presence of 12.5 mM EDTA [33]. When attributing H+-ATPase inhibition to PKS5-like kinases in the absence of the MgATP or MnATP complexes, one has to assume a high-affinity binding site of the kinase for the bivalent cation. The Kd of the Mg–kinase complex should be less than 30 μM, the Kd of MgATP [9], in order to maintain Mg2+ binding in the presence of free ATP. This would be unexpected since neither free ATP nor free Mg2+ bind well to the conserved catalytic domain of protein kinases [37].

(ii) Inhibition by a kinase of the PKS family in our measurements is also unlikely because ATPase activity decreased at high ATP concentrations above 5 mM. Therefore the kinase should not be saturated with ATP at 5 mM ATP. A Km value in the millimolar range would be required. Gong et al. [36] determined a Km for ATP of 0.83 μM for another member of the PKS family (PKS11). Accordingly for PKS5, kinase assays were conducted at 10 μM ATP [33]. Fuglsang et al. [33] performed H+-pumping assays with PM vesicles (Figure 8A in their publication) which for PKS5 support a Km for ATP below 0.5 mM. The initial rate of H+ pumping was determined at various ATP concentrations up to 3 mM in vesicles from PKS5-free plant material and after adding PKS5 to the vesicle preparation. Inhibition was already detectable at ATP concentrations of 0.5 mM. Importantly, relative inhibition did not increase at higher ATP concentrations which is consistent with the low Km reported for PKS11.

(iii) A third argument against a kinase-based interpretation of our results arises from the use of a chaotropic salt in our leaf vesicle preparation. PKS5 should associate with the PM in vivo and should remain associated with PM vesicles during vesicle preparation. Fuglsang et al. [33] clearly showed that vesicles prepared from Arabidopsis expressing the PKS5 gene had lower H+-pumping activity compared with vesicles from loss-of-function pks5 mutants. These experiments prove the presence of inhibitory phosphorylated sites in vesicle preparations. However, they do not prove that phosphorylation occurred during the pumping assay. This may already have occurred before vesicle preparation. Up to now, PKS5 was not identified in PM vesicles of Arabidopsis, although other members of this kinase family have been found in lipid rafts of vesicle preparations [34]. The SNF1-like kinase family to which PKS5 belongs does not contain integral membrane proteins [31]. Therefore PKS5 should have been anchored in the membrane by binding to another molecule with hydrophobic character before vesicle preparation, or PKS5 itself should have bound to the membrane surface as a peripheral membrane protein. Anchoring could have been achieved by interaction with a CBL protein, the N-terminal myristoyl group of which may immobilize PKS kinases at the PM. Indeed, interaction of PKS5 with a CBL protein was demonstrated in vivo. However, the physiological relevance of this interaction is not yet clear, since PKS5 was active in vitro without CBL association [33]. After association of PKS5 with the PM, this association should be stable enough to withstand our membrane isolation procedure. In contrast with the vesicle preparations from Shahollari et al. [34], in which CBL proteins were proven, we used the chaotropic salt potassium iodide, in order to remove peripheral membrane proteins as outlined in the Experimental section. This treatment removes 30–40% of total membrane protein in maize root vesicles [22]. Therefore the present study probably reveals direct effects of Mg2+-free ATP on PM H+-ATPase activity.

Interaction of Mg2+-free ATP with H+-ATPase

If the Mg2+ activity falls below the range of the Kd of the MgATP complex, then Mg2+-free ATP will accumulate. One possible interaction of Mg2+-free ATP with H+-ATPase could be that Mg2+-free ATP competes with the enzyme for Mg2+ which is required as a cofactor in P-type ATPases. In fact, for Na+/K+-ATPase, the Kd of Mg2+ interaction with the E1ATP(Na3) state was identical with the Kd of MgATP [38] which would cause Mg2+ withdrawal from the enzyme upon addition of Mg2+-free ATP. This would also occur in the normal functional cycle, in which ATP binds to the E2 or E2P state (see the next section), if the Kd of the Mg2+ interaction with respect to these enzyme states were similar or even larger compared with the E1ATP(Na3) state. Another mode of inhibition would be competition between MgATP and Mg2+-free ATP for the same enzyme site. Inhibition by Mg2+-free ATP has been observed for many Mg2+-dependent ATPases. For yeast mitochondrial F1-ATPase, Jenkins [39] suggested that excess ATP inhibits by complexing free Mg2+ assuming that it is free Mg2+ which binds to the enzyme–ATP complex. For crayfish mitochondrial F1-ATPase, Li and Neufield [40] explained the kinetic data with a classical model of competitive inhibition by Mg2+-free ATP.

Binding site for Mg2+-free ATP

Our data on inhibition by Mg2+-free ATP are consistent with a model of competitive inhibition (Figure 1C). Competition between MgATP and Mg2+-free ATP for the same binding site may occur in different states of the functional cycle of P-type ATPases. According to the most widely used model of the mechanism of P-type ATPases (Albers–Post model or E1-E2 model) ATP binds to the dephosphorylated E2 state [41]. Indeed, ATP binding to this state dramatically accelerates the E2→E1 transition of other structurally related P-type ATPases, i.e. PM Na+/K+-ATPase and sarcoplasmic/endoplasmic reticulum Ca2+-ATPase. However, ATP binding to the E2P phosphoenzyme of Na+/K+-ATPase was demonstrated recently utilizing a fluorescent dye which revealed changes in the E2P state at MgATP concentrations above 100 μM [42]. Moreover, for sarcoplasmic/endoplasmic reticulum Ca2+-ATPase, X-ray crystallography shows binding of an ATP analogue to the transition state of dephosphorylation (E2-P*). The authors suggest that ATP binds in exchange for ADP already in the E2P state [43]. Khalid et al. [42] showed that, in the absence of K+, MgATP bound to the E2P state of PM Na+/K+-ATPase plays a modulatory role by decreasing the rate of dephosphorylation. For PM Na+/K+-ATPase, it is now clear “that the enzyme only has a single ATP-binding site which fulfils both allosteric and catalytic functions in different parts of the reaction cycle” [42].

Competitive inhibition of PM Na+/K+-ATPase by Mg2+-free ATP has been first described by Hexum et al. in 1970 [44]. Robinson [45] confirmed this finding. His experiments were performed at pH 7.8, at which the concentration of (H-ATP)3− is negligible. He did not discriminate between (K-ATP)3− and ATP4− and determined an apparent Ki of 4.8 mM. For PM H+-ATPase in maize, we obtained considerably lower values between 0.01 and 1.1 mM depending on the ionic conditions of the assay. Competitive inhibition requires that Mg2+-free ATP binds at or in the ATP-binding pocket of the nucleotide-binding domain (N-domain). For PM Na+/K+-ATPase, site-directed mutagenesis was exploited to identify specific amino acid residues in the N-domain which contribute to high-affinity binding of Mg2+-free ATP [46]. Indeed, an arginine residue at the mouth of the ATP-binding pocket was shown to be important for binding of Mg2+-free ATP as well as MgATP.

Inhibitory ATP species

At inhibitory ATP concentrations, changes in K+ concentration did not influence PM ATPase activity (Figure 2B), although the concentration of all major Mg2+-free ATP species changed (Table 2). Does this observation rule out a model of inhibition by a single ATP species? The answer has to account for the important observation that the Km for MgATP decreased at low K+ concentrations (Table 1, compare experiment 5 with experiment 4). This increase in substrate affinity should increase hydrolytic activity in the presence of the inhibitory ATP species. Because this did not occur, obviously affinities increased for MgATP as well as for the inhibitor. Indeed, in any mechanism of competitive inhibition, a change in Km is likely to affect Ki as well. When regression analysis is not forced to use the same Km and Ki for both K+ concentrations, the K+-dependent concentration change in a given ATP species is completely compensated for by a change in Ki. As a consequence, good fits are obtained for any ATP species and Akaike weights are similar for all of them (Table 2). Therefore inhibition by a single ATP species cannot be ruled out.

View this table:
Table 2 Discrimination between different Mg2+-free ATP species

The enzyme activity data from the experiment at different K+ concentrations (Figure 2B) were analysed on the basis of three models of competitive inhibition by a single ATP species. For each ATP species, Ki (μM) was calculated (see Supplementary Figure S2 at Percentage values denote the ratio between Ki of the species and the sum of Ki values at the given K+ concentration. This ratio reflects the molar composition of the Mg2+-free ATP pool. The fit of each of the models to the experimental data is expressed as the Akaike weight. The model (ATP species) with highest Akaike weight is most plausible, but here differences are negligible.

Physiological relevance of inhibition by Mg2+-free ATP

Apparent Ki values for competitive inhibition by Mg2+-free ATP are in the range of the Km values for MgATP. This means that, at a given MgATP concentration, activity declines by 50% upon withdrawal of half of the Mg2+ from MgATP (in addition to the decrease due to smaller substrate concentration). It is evident that this mechanism is particularly useful for down-regulating PM H+-ATPase activity when MgATP concentrations are in the saturating range in which changes in substrate concentration have little influence on activity. Moreover, this mechanism does not require disposal or removal of Mg2+-free ATP, which remains available as a regulator for other processes or for fast reactivation by Mg2+.

Our investigation revealed differences between leaves and roots with respect to Ki and Km (Table 1, see experiments 2 and 3). The two parameters changed in the same direction, but the Ki/Km ratio was smaller for leaves. As explained in the Results section and utilizing a Kd for MgATP of 71 μM, a decrease in Mg2+ activity to a level of 63 μM ([Mg2+]50) would cause 50% inhibition of PM H+-ATPase in leaves, whereas in roots, a lower Mg2+ activity of 37 μM would be required to achieve 50% inhibition. For the leaves, the control potential of ATP inhibition is substantially higher at high K+ concentration with a [Mg2+]50 of 127 μM (experiment 4). It is particularly high at pH 7.5 with a [Mg2+]50 of 222 μM. This is well within the range of estimated Mg2+ activities in leaf mesophyll cells [8]. Control by cellular Mg2+ could play a physiological role in leaf cells once cellular turgor has been established. Inhibition by Mg2+-free ATP of E1 phosphorylation could act in concert with K+-stimulated dephosphorylation of the E1P state [3], both preventing E1P accumulation under conditions in which the membrane potential is close to the reversal potential of the pump. Since the E1P state can phosphorylate ADP (is ADP-sensitive), E1P accumulation could cause the pump to run backwards [3]. We did not observe the stimulatory effect of K+ on ATP hydrolysis of root vesicles. An explanation could be that hydrolytic stimulation of root isoforms is already saturated at 55 mM K+.

Mechanisms which decrease cytosolic Mg2+ activity

ATP inhibition, in vivo, of PM ATPases by Mg2+-free ATP depends on the existence of mechanisms, which force the cytosolic Mg2+ activity down to the range of 30 μM, the Kd of (MgATP)2−. Two types of mechanisms have to be considered [7]: (i) Mg2+ transport into organelles or to the extracellular space through membrane transporters or channels, and (ii) Mg2+ binding by other strong chelators or macromolecules.

For the small thin cytosolic compartment of a leaf mesophyll cell, the voluminous chloroplasts with their ability to take up Mg2+ appear to represent a competitive Mg2+ sink. Ion translocation within the chloroplast is known to have a profound influence on ionic conditions in the cytosol and on PM H+-ATPase activity. After switching off the light or when the substomatal CO2 concentration suddenly increases, massive proton release from the thylakoid lumen to the chloroplast stroma occurs. Also within the cytosol, proton activity increases transiently [47]. At the PM, both stimuli immediately invoke a transient hyperpolarization [47,48]. At the same time, apoplastic pH decreases [49] consistent with a transient stimulation of PM H+-ATPase. While protons pass the thylakoid membrane, Mg2+ ions move in the opposite direction for charge compensation. When the light is turned on, Mg2+ is transported to the stroma and is bound to the N-carboxylated Rubisco (ribulose-1,5-bisphosphate carboxylase/oxygenase), thereby activating it. It is clear that upon light-off, Mg2+ must be released to be bound again upon the next exposure to light. Release probably causes a transient increase in stromal Mg2+ activity [8]. Evidence indicates that, after switching off the light, the stromal Mg2+ concentration eventually decreases by several millimolar. Taking into consideration that Mg2+ can freely move through the outer chloroplast membrane, not only protons, but also Mg2+, may leave the chloroplast upon light-off and may contribute to the transient stimulation of PM H+-ATPase by complexing Mg2+-free ATP.

For plant cells, information on Mg2+ transport, its affinity for Mg2+ and its regulation is very limited [50]. There is one report on a candidate Mg2+ transporter of Arabidopsis thaliana which restores intramitochondrial Mg2+ concentration in a yeast mutant [50]. Membrane transporters for Mg2+ are better investigated in mammalian cells. The channel kinase TRPM7 (transient receptor potential melastatin 7) mediates Mg2+ uptake from the extracellular space into the cytosol and is essential for cellular viability [51]. This channel is strongly down-regulated by intracellular Mg2+ activity, MgATP and Mg–nucleotides, thereby restricting cytosolic access of Mg2+. Massive cAMP-mediated Mg2+ release from mitochondria into the cytosol occurs upon catecholamine administration [7]. Mg2+ release from isolated mitochondria upon administration of cAMP is accompanied by an increase in the respiration rate. The underlying activity changes in mitochondrial enzymes could be caused by a decrease in matrix Mg2+ activity. In the intact cell, relief of ATP inhibition of the Na+/K+-ATPase, which may consume 20–30% of the cellular ATP [46], by additional Mg2+ could boost ATP hydrolysis and thus contribute to enhanced respiration.

Besides ATP, a number of Mg2+ chelators are related to the regulation of metabolic activity. In chloroplasts, Mg–chelatase activity and Mg–protoporphyrin concentration increase upon ABA (abscisic acid) application [13] suggesting a mechanism for hormone-stimulated Mg2+ withdrawal from the cytosol. In non-phytosynthetic tissue, the cytosolic Mg2+ activity may be decreased by synthesis of phytic acid. Phytic acid probably exists mostly as a penta-Mg2+ salt in vivo [52]. ABA induces the enzyme inositol-3-phosphate synthase, leading to the accumulation of phytic acid in duckweed [53]. For roots of maize seedlings, it has been shown that exogenous phosphate is readily incorporated into phytic acid [54]. The level of the phytic acid precursor myo-inositol increased to 75% of soluble carbohydrates in roots of salt-adapted tomato plants [55]. Another physiological process involving Mg2+ binding to phosphate groups is polyamine catabolism. Upon degradation of these abundant polycations, their binding sites for DNA, RNA and proteins will adsorb free Mg2+ [56].

There is evidence that ABA responsiveness of guard cells requires attenuation of H+ pumping in addition to activation of anion channels [13]. For activation of the guard cell anion channel GCAC1 ATP is required, but for this PM transporter, the Mg2+-free nucleotide is sufficient for activation [57]. Because a large number of ABA-responsive genes are triggered by Mg2+ deficiency [58], it is clear that ABA responses are not hampered by Mg2+ deficiency. It is also conceivable that ABA responses may impose a physiological Mg2+ deficiency on the cytosolic compartment.

For PM Na+/K+-ATPase, inhibition was found not only by Mg2+-free ATP, but also by MgADP in a competitive manner [45]. In our assays, ADP was also inhibitory (A. Fatima, unpublished work). This is evident from assays which were performed with root vesicles at pH 6.5 in the presence of 5 mM ATP, 5 mM Mg2+ and 155 mM K+, but without an ATP-regenerating system. Derived from the increase in phosphate concentration, ADP accumulated to a concentration of 0.15 mM. Omission of the ATP-regenerating system decreased PM H+-ATPase activity by 15±2% (n=4). Possibly, the inhibition mechanisms of Mg2+-free ATP and of MgADP are closely related. Certainly, these mechanisms contribute to the complex metabolic and mineral requirements of PM H+-ATPase for maximum hydrolytic activity. With respect to mineral requirements, kinetic properties of PM H+-ATPase and emerging patterns of regulation of several intracellular Mg2+ chelators indicate that ATP inhibition of PM H+-ATPase could be involved in regulatory mechanisms for reducing the activity of this ‘powerhouse’ enzyme during stages of growth reduction, growth arrest or during the transition from active growth to dormancy.


Peter Friedhoff, Stefan Hanstein, Sven Schubert and Ke Feng were involved in the conception and design of the experiments. Xiaozhi Wang, Xiaoqing Qian, Yuhua Shan and Ammara Fatima prepared PM vesicles from hydroponically grown maize plants. Results were acquired by Xiaoqing Qian, Xiaozhi Wang and Stefan Hanstein. Xiaoqing Qian normalized the activity data. Peter Friedhoff, Stefan Hanstein and Xiaozhi Wang performed non-linear regression analysis based on recommended complexation constants. Stefan Hanstein interpreted inhibition constants and Ki/Km in the context of the cytosolic concentration of free Mg2+. Stefan Hanstein, in collaboration with the others, drafted the paper. Sven Schubert and Peter Friedhoff revised the paper critically for important intellectual content.


This work was supported by the National Basic Research Program of China [grant number 2007CB109303], by the Natural Science Foundation of China [grant number 30871588] and by the China Scholarship Council [grant number 2007832114].

Abbreviations: ABA, abscisic acid; BTP, 1,3-bis[tris(hydroxymethyl)methylamino]propane; PEP, phosphoenolpyruvate; PM, plasma membrane


View Abstract