The kinetic properties of a human PPIP5K reveal that its kinase activities are protected against the consequences of a deteriorating cellular bioenergetic environment

Abstract

We obtained detailed kinetic characteristics–stoichiometry, reaction rates, substrate affinities and equilibrium conditions–of human PPIP5K2 (diphosphoinositol pentakisphosphate kinase 2). This enzyme synthesizes ‘high-energy’ PP-InsPs (diphosphoinositol polyphosphates) by metabolizing InsP₆ (inositol hexakisphosphate) and 5-InsP₇ (5-diphosphoinositol 1,2,3,4,6-pentakisphosphate) to 1-InsP₇ (1-diphosphoinositol 2,3,4,5,6-pentakisphosphate) and InsP₈ (1,5-bis-diphosphoinositol 2,3,4,6-tetrakisphosphate), respectively. These data increase our insight into the PPIP5K2 reaction mechanism and clarify the interface between PPIP5K catalytic activities and cellular bioenergetic status. For example, stochiometric analysis uncovered non-productive, substrate-stimulated ATPase activity (thus, approximately 2 and 1.2 ATP molecules are utilized to synthesize each molecule of 1-InsP₇ and InsP₈, respectively). Impaired ATPase activity of a PPIP5K2-K248A mutant increased atomic-level insight into the enzyme's reaction mechanism. We found PPIP5K2 to be fully reversible as an ATP-synthase in vitro, but our new data contradict previous perceptions that significant ‘reversibility’ occurs in vivo. PPIP5K2 was insensitive to physiological changes in either [AMP] or [ATP]/[ADP] ratios. Those data, together with adenine nucleotide kinetics (ATP K_m=20–40 μM), reveal how insulated PPIP5K2 is from cellular bioenergetic challenges. Finally, the specificity constants for PPIP5K2 revise upwards by one-to-two orders of magnitude the inherent catalytic activities of this enzyme, and we show its equilibrium point favours 80–90% depletion of InsP₆/5-InsP₇.

INTRODUCTION

PP-InsPs [diphosphoinositol polyphosphates, such as 1-InsP₇ (1-diphosphoinositol 2,3,4,5,6-pentakisphosphate), 5-InsP₇ (5diphosphoinositol 1,2,3,4,6-pentakisphosphate) and InsP₈ (1,5-bis-diphosphoinositol 2,3,4,6-tetrakisphosphate); see Figure 1] have been shown to regulate diverse cellular processes such as vesicle trafficking, exocytosis, cytoskeletal dynamics, apoptosis, environmental stress response and insulin secretion and signalling (reviewed in [1–6]). Research into the mechanisms of action of the PP-InsPs has identified a cyclin/cyclin kinase ‘receptor’ in yeast that is allosterically regulated by 1-InsP₇ [7]. There is also evidence that 5-InsP₇ can regulate cell-signalling by competing with PtdIns(3,4,5)P₃ for binding to pleckstrin homology domains [8]. Finally, PP-InsPs have been reported to promote the diphosphorylation of certain proteins thereby regulating inherent protein activities and protein/protein interactions [9–11]. Each of these proposed mechanisms of action of the PP-InsPs are based upon a fundamental principle in signal transduction: their cellular levels are altered in a predictable manner in response to a specific intracellular or extracellular stimulus. For example, levels of 1-InsP₇ are elevated during nutrient stress in yeast [7]. Total InsP₇ levels also increase substantially following the addition of growth factors to cells that have been serum-starved overnight [8]. As for InsP₈, its levels are elevated when cells are subjected to either hyperosmotic stress or a thermal challenge [12,13]. There are other circumstances that cause InsP₇ and InsP₈ levels to decrease, such as following oxidative stress [14]. Finally, InsP₈ concentration decreases during bioenergetic stress [15].

Figure 1. Pathways of enzymatic synthesis of PP-InsPs.

Figure modified from Shears 2009 [5]. The IP6Ks (myoinositol hexakisphosphate kinase; Kcs1 in yeast [48]), of which there are three isoforms in mammals, IP6K1, IP6K2 and IP6K3 [26,48,52], phosphorylate InsP₆ and 1-InsP₇ at the 5 position [41,53]. The PPIP5Ks (Vip1 in yeast [21]), of which there are two isoforms in mammals, PPIP5K1 and PPIP5K2 [18,19], phosphorylate InsP₆ and 5-InsP₇ at the 1 position [20,41]. Thus, the concerted actions of the IP6Ks and the PPIP5Ks leads to two routes of synthesis to InsP₈, which are designated pathway I and pathway II after Padmanabhan et al. [17]. The diphosphate groups that are added by these kinases are hydrolysed by a family of PP-InsP phosphohydrolases (DIPPs) [44,45].

It is because of all of these signalling activities that there is a need to characterize how the turnover of PP-InsPs is regulated in intact cells. As with any signal transduction cascade, in order to understand its dynamic behaviour in vivo, it is critical to determine the kinetic parameters of its constituent signalling enzymes [16,17]. Such information–enzyme stoichiometry, reaction rates, substrate affinities and equilibrium conditions–helps us to understand the regulation of product generation as well as substrate depletion. In the current study, we provide new information concerning these properties of a PPIP5K (diphosphoinositol pentakisphosphate kinase), an enzyme that phosphorylates InsP₆ (inositol hexakisphosphate) and 5-InsP₇ to 1-InsP₇ and InsP₈, respectively (Figure 1).

Mammalian cells express two PPIP5Ks: types 1 and 2 that are 160 and 138 kDa, respectively [18,19]. The expression and purification of recombinant enzymes of such a large size are problematic. Fortunately, these are modular proteins; the N-terminal one-third of each of these proteins contains their kinase domain [18,20,21], following which is a centrally located PtdIns(3,4,5)P₃-binding module [22]. The active sites of these two kinase domains appear to be structurally and catalytically indistinguishable [18,20]; the relatively small number of residues elsewhere in these domains that are dissimilar are confined to the protein surface that is distal to the active site [20]. Thus, using the kinase domain of human PPIP5K2^KD (PPIP5K2 kinase domain) as a model, we have now characterized the kinetic properties of this class of enzyme.

For an ATP-dependent kinase such as IP6K (inositol hexakisphosphate kinase) or PPIP5K, its activity may not solely depend upon the concentrations of inositol phosphate substrates. There is growing interest in the possibility that changes in the levels of adenine nucleotides may also be of importance in regulating IP6K and PPIP5K activity in vivo [1,2,5,6]. This idea has taken support from the evidence that bioenergetic homoeostasis interfaces with PP-InsP turnover and cell signalling [15,23–26]. For example, some experiments indicate that bioenergetic health sets the cellular levels of PP-InsPs [15,24–26]. Other work [23] indicates that it is bioenergetic status that is regulated by the PP-InsPs. One of the goals of the current study was to increase our understanding of the role of PPIP5K in this complex inter-relationship between signalling and metabolism.

Interest in the possibility of a link between cellular bioenergetics and PP-InsP turnover began with the recognition that these molecules contain high-energy phosphoanhydride bonds that undergo rapid turnover in vivo [27,28]. Subsequently, in what was a startling exception for kinases in general [29], and inositol phosphate kinases in particular [30], a high K_m (>1 mM) of the IP6K for ATP was determined [26]. IP6Ks were further shown to have a similarly high K_m for ADP, facilitating an ability to act ‘in reverse’ as ATP-synthases [26]. This ATP-synthase activity was proposed to occur in vivo [26]. Thus, akin to InsP₅ (inositol 1,3,4,5,6-pentakisphosphate) 2-kinase in plants [31], the IP6K reaction direction and rate of activity appear to be controlled by changes in ATP/ADP ratios [24]. These data have consolidated the idea that PP-InsP turnover interfaces with cellular bioenergetic homoeostasis.

Experiments with partially purified native preparations of PPIP5K [25] have led to the conclusion that, just like IP6K, PPIP5K is reversible under in vivo conditions, such that cellular InsP₈ levels can be controlled by physiological changes in ATP/ADP ratios [25]. This hypothesis has been well received [32–34]. However, it is a natural concern that the properties of partially purified native enzyme preparations may be influenced by contaminating proteins and other small-molecule modulators. The intrinsic lability of native PPIP5K [19] is also an issue. Thus, one of the goals of the current study was to use recombinant kinase to investigate through a kinetic study the manner in which PPIP5K activity–in both the ‘forward’ and ‘reverse’ directions–may be influenced by physiological changes in levels of adenine nucleotides as well as by the concentrations of inositol phosphate substrates.

Additionally, we should note that reliable data on reaction rates, reversibility and steady-state conditions are not the only parameters that systems biologists utilize when transferring kinetic data with isolated enzymes into mathematical models of metabolic pathways in vivo; information on reaction stoichiometry is also important [16]. In the current study we determined that PPIP5K's substrate-independent ATPase activity [20] is considerably more significant than was previously appreciated, and is stimulated up to 4-fold upon the binding of InsP (inositol phosphate) substrate to the active site. We discuss the significance of this bioenergetic cost to the cell in terms of the roles of PPIPKs in vivo. We also discuss how this discovery of substrate-dependent ATPase activity of PPIP5K provided an opportunity to gain new insight into the reaction mechanism of this particular kinase.

One of the reasons that such extensive kinetic studies have not previously been performed with recombinant PPIP5Ks is the difficulties in obtaining samples of the PP-InsPs with the appropriate degree of purity. PP-InsPs are not commercially available and, at the time of writing, we are unaware of any laboratories that are currently chemically synthesizing these materials. PP-InsPs can be prepared enzymatically but these highly charged molecules are inherently difficult to isolate in a salt-free form to a high degree of purity. Recently, a new method was introduced, which accomplishes that goal [35]. We have used these procedures in our study.

MATERIALS AND METHODS

Materials

InsP₆ was purchased from Calbiochem and [³H]InsP₆ (~13.0 Ci/mmol) from PerkinElmer. Non-radiolabelled and [³H]-radiolabelled PP-InsPs (1-InsP₇, 5-InsP₇ and InsP₈) were all synthesized enzymatically with the appropriate purified enzymes (see below). Deionized water was used throughout.

Protein expression and purification

The human PPIP5K2 kinase domain (residues 1–366) [hPPIP5K2^KD, NCBI (National Center for Biotechnology Information) accession number NP_056031.2] and its single-site mutants were expressed in Escherichia coli and purified as previously described [20]. The protein was stored at −80°C.

The human IP6K1 (NCBI accession number NP_695005.1) was expressed and purified as previously described for human PPIP5K1 and PPIP5K2 with the following modifications [18]: the protein was expressed with an N-terminal GST (glutathione transferase) tag in BL21 (DE3) E. coli (Stratagene) by induction overnight. Cells were sonicated in lysis buffer containing an EDTA-free complete protease inhibitor (Roche). The clarified lysate was loaded on to a 5 ml GSTrap HP (GE Healthcare) column, washed and eluted with a linear gradient from 0 to 30 mM of reduced glutathione in 25 mM Tris pH 8.0, 50 mM NaCl and 1 mM DTT (dithiothreitol). The protein was stored with 0.5 mg/ml BSA and 50% glycerol at −80°C. The purity of our enzyme preparations were >95% (see Supplementary Figure S1 at http://www.bioscirep.org/bsr/033/bsr033e022add.htm).

Preparation of PP-InsPs

Enzymatic reactions were carried out in siliconized tubes (PGC Scientific) at 37°C in 20 mM Hepes, pH 6.8, 50 mM NaCl, 6 mM MgSO₄, 1 mM DTT, 6 mM phosphocreatine, 24 unit/ml creatine kinase and 5 mM ATP disodium salt. Reactions (15 min) for the synthesis of [³H]-radiolabelled 1-InsP₇, 5-InsP₇ and InsP₈ contained 9.7 μCi/ml [³H]InsP₆ (~13.0 Ci/mmol) and either, respectively, 0.076 mg/ml PPIP5K2^KD, 0.21 mg/ml GST–IP6K1 or 0.35 mg/ml GST–IP6K1 and 0.076 mg/ml PPIP5K2^KD. Reactions for the synthesis of non-radiolabelled 1-InsP₇, 5-InsP₇ and InsP₈ contained 1.2 mM InsP₆ and were conducted as follows, respectively: 0.12 mg/ml PPIP5K2^KD for 22.5 h, 0.17 mg/ml GST–IP6K1 for 3 h or 0.17 mg/ml GST–IP6K1 and 0.11 mg/ml PPIP5K2^KD for 3 h. Reactions were quenched and neutralized with, respectively, 0.2 volumes 2 M HClO₄ and 0.34 volumes 1 M K₂CO₃, 40 mM EDTA or placed at 100°C for 3–5 min, then concentrated in a SpeedVac (Savant) at 43°C to 300 μl.

The PP-InsPs were then purified via a PAGE-based method [35] with the following modifications in order to scale up the procedure. Gels (31 cm×38.5 cm×1.5 mm) were run in a model S2 sequencing gel electrophoresis system (Life Technologies Gibco BRL) at 4°C. Gels were prerun at 1000 V for 1 h prior to sample loading, then run at 1000 V for 48–60 h. The elution positions of the PP-InsPs were determined in a 1 cm wide column at the edge of the gel. The non-radiolabelled PP-InsPs were detected by staining with Toluidine Blue [35]. For the detection of PP-[³H]InsPs, the excised region of the gel was cut into 0.5 cm bands which were each mixed with 1 ml of Solusol (National Diagnostics) on a rocker (25 rev./min) for at least 5 h at room temperature, after which 4 ml of Soluscint XR (National Diagnostics) was added and samples were counted by liquid scintillation spectroscopy. To correct for any unevenness in the chromatography, the Orange-G dye front was used as a guide for determining the elution positions of the products (see Supplementary Figure S2 at http://www.bioscirep.org/bsr/033/bsr033e022add.htm). The appropriate sections of the gels were excised and washed with 5–10 ml 20% (v/v) methanol, followed by a brief rinse with 5 ml water. The gel fragments were homogenized (Tekmar Tissuemizer Mark II) in 5 ml water at 20500 rev./min at 0–4°C prior to extraction of the PP-InsP. Samples were concentrated to 1 ml (non-radiolabelled) or 4000 DPM/μl ([³H]-radiolabelled) in a SpeedVac, aliquoted and stored at −80°C. No sample degradation was detected through a dozen freeze–thaw cycles over a period of several months.

Purity of PP-[³H]InsPs was assessed by HPLC (see the Results section). Non-radiolabelled PP-InsPs were quantified by mass assay of the released orthophosphate upon complete hydrolysis by wet-ashing at 120°C for 48 h [36,37].

Enzyme assays

Kinetic parameters (K_m, V_max, k_cat and k_cat/K_m) for adenine nucleotides and inositol phosphates were determined for PPIP5K2^KD under pseudo first-order conditions [38]. Reactions (200 μl, 37°C) simulating intracellular conditions contained 20 mM Hepes-NaOH pH 7.2, 50 mM KCl, 1 mg/ml bovine serum albumen, 1 mM EDTA disodium salt, 1 mM DTT, ATP or ADP disodium salt (1.1 μM to 10 mM), inositol phosphate (10 nM–10 μM), appropriate [³H]-radiolabelled inositol phosphate tracer (4000 DPM) and PPIP5K2. Total MgSO₄ varied from 2 to 12 mM depending on ATP and ADP concentrations. Free Mg²⁺ was 1.2±0.1 mM (estimates performed with WebMaxc Extended, http://www.stanford.edu/~cpatton/webmaxc/webmaxcE.htm, last updated 7/3/09, accessed 8/24/11 [39]). Reactions were quenched and neutralized with, respectively, 0.2 volumes 2 M HClO₄ and 0.34 volumes 1 M K₂CO₃, 40 mM EDTA, then analysed by HPLC.

The mobile phase was initially run through a 4.6×250 mm silica presaturation column (Grace), before samples were loaded onto a 4.6×125 mm Partisphere 5 SAX column (Whatman). Buffer A was 1 mM EDTA disodium salt, Buffer B was 1.3 M (NH₄)₂HPO₄ (Sigma) plus 1 mM EDTA disodium salt, pH 3.85 with phosphoric acid (Aldrich). The gradient (1 ml/min) was as follows: 0–5 min, 0% B; 5–10 min, B increased linearly from 0 to 45%; 10–60 min, B increased linearly from 45 to 100%; 60–75 min, B was 100%. Each fraction (~30/assay, 1 ml each) was mixed with 4 ml MonoFlow 4 scintillant (National Diagnostics) and counted on a liquid scintillation counter for 10 min. For the kinetic determinations, assay linearity was confirmed under all conditions. The Michaelis–Menten equation was fit to plots of initial velocity against substrate concentration using non-linear regression to determine K_m and V_max values (GraphPad Prism v5.03). Results are presented as the means±S.E.M. (n≥2). Similar assay conditions were used to determine equilibrium points.

ATPase activity

ATPase activity (non-productive hydrolysis of ATP to ADP that did not drive phosphorylation of an InsP) was determined by measurement of free orthophosphate in the reaction solution by the Malachite Green detection method used for quantification of PP-InsP preparations (see above) 27–270 μg/ml PPIP5K2^KD was incubated at 37°C for 120 min in 50 μl reaction mixtures containing 20 mM Tris/HCl, pH 7.5, 10 mM ATP (unless otherwise indicated), 100 mM KCl, 0.1 mM EDTA and 5 mM MgCl₂. Results are presented as the means±S.D. (n=3).

ATP synthesis by PPIP5K2^KD

The ATP synthase activity of PPIP5K2^KD was determined in a 96-well microplate format by incubating 0.3 μg/ml enzyme at 37°C for 15 min in 20 μl assays containing 1.4× reaction buffer (ATP Determination Kit, Molecular Probes), 650 μM ADP (ATP contamination<1 ppm, Apollo Scientific Limited), 250 nM InsP₈ and the indicated concentrations of either InsP₆ or InsS₆ (inositol hexasulphate). Then 16 μl of either an ATP standard solution or reaction solution was added into 100 μl Standard Reaction Solution (ATP Determination Kit, Molecular Probes) and luminescence was recorded after 15 s using a BioTek Synergy 2 microplate reader. The IC₅₀ values were determined using GraphPad Prism v5.03.

Crystal structure determination

Crystals of PPIP5K2^KD in complex with ADP were soaked with InsP₈ using previously published procedures [20].

RESULTS AND DISCUSSION

Enzymatic synthesis and purity of PP-InsPs

PP-InsPs possess energetic phosphoanhydride bonds and undergo rapid turnover [27,28], so their cellular synthesis requires a considerable bioenergetic investment. Indeed, variations in cellular bioenergetic status have been proposed to influence PP-InsP synthesis (see the Introduction section). We have investigated the role of PPIP5Ks in this process by determining the kinetic properties of the kinase domain of recombinant human PPIP5K2 (PPIP5K2^KD).

One of our goals was to analyse the ‘forward’ (ATP-consuming) and ‘reverse’ (ATP-producing) reactions; that information has not previously been derived from recombinant enzyme. We therefore required three PP-InsPs for this study, both radiolabelled and non-radiolabelled (1-InsP₇, 5-InsP₇ and InsP₈; Figure 1). These PP-InsPs were not commercially available, and to our knowledge there was not a laboratory that was chemically synthesizing these molecules. We therefore prepared them enzymatically, and purified them by a PAGE-based technique [35]. We determined that the purities of 5-[³H]InsP₇, 1-[³H]InsP₇ and [³H]InsP₈ were 91, 98 and 92%, respectively (Figures 2A–2C). The purities of 5-InsP₇, 1-InsP₇ and InsP₈ were estimated to be 96, 98 and 97%, respectively (Figure 2D). This is an improvement over previous studies in which PP-InsPs were either prepared enzymatically or chemically synthesized, whereupon the final products were 75–90% pure [18,19,40–42].

Figure 2. Purity of PP-InsP preparations.

HPLC (see the Materials and methods section) of 5-[³H]InsP₇ (A), 1-[³H]InsP₇ (B) and [³H]InsP₈ (C). The purities of 5-[³H]InsP₇, 1-[³H]InsP₇ and [³H]InsP₈ were 91, 98 and 92%, respectively. (D) Toluidine Blue-stained PAGE analysis (16 cm gel) of purity of non-radiolabelled PP-InsP preparations (20 nmol/lane). The image was enhanced in ImageJ [54] by increasing contrast and smoothing. Purity was quantified by densitometric analysis (right-hand panels) of the central portion of each lane as indicated by the rectangles in each of the left-hand panels using ImageJ [54]. Commercial InsP₆ (not further purified) served as a control. Purity was as follows: 5-InsP₇, 96%; 1-InsP₇, 98%; and InsP₈, 97%.

Kinetic analysis of PPIP5K2^KD

We first used HPLC to analyse the PPIP5K-mediated phosphorylation of [³H]-radiolabelled inositol phosphate substrates using saturating concentrations of ATP (see below) (Figures 3 and 4). Our data (Table 1) indicate that 5-InsP₇ is the higher-affinity substrate; we found the K_m value of PPIP5K2^KD for InsP₆ (0.4 μM; Table 1) to be 6.5-fold higher than that for 5-InsP₇ (0.06 μM; Table 1). Our results are similar to the previously determined K_m values for recombinant PPIP5Ks, 0.1 μM for InsP₆ and 0.1–0.3 μM for 5-InsP₇ [1,18,19]. These kinetic data demonstrate that the active site will be near-saturated by the levels of inositol phosphate substrates that occur in vivo [1,43]. However, one significant difference with earlier data is that we found that the V_max values for these reactions (Table 1) were about 20–130-fold higher than those previously reported [18,19] for recombinant PPIP5K. This in turn is largely responsible for our values for k_cat/K_m–the specificity constants–being up to 100-fold greater (Table 1) than that of previous determinations [18,19]. This brings the specificity constant for the PPIP5Ks much closer to that of the DIPPs (diphosphoinositol-polyphosphate phosphohydrolases) [44,45], and as such is more consistent with the long-standing observation that there is high ongoing turnover of PP-InsPs in vivo [2,28]. Our much higher estimates for the V_max, which are likely due our obtaining correctly folded protein to a greater degree of purity, counter the argument that PPIP5K activity is too low to participate in ‘phasic’ signalling events [1].

Figure 3. Phosphorylation reactions of PPIP5K2^KD and substrate saturation plots for ATP and ADP.

(A) Either no enzyme (broken line) or 27 μg/ml PPIP5K2^KD (solid line) was incubated with 10 μM [³H]InsP₆, and 11 μM ATP for 20 min as indicated in the Materials and methods section. The reactions were quenched and neutralized and analysed by Partisphere SAX HPLC as described in the Materials and methods section. Representative HPLC data are shown. (B) The initial velocity of InsP₆ phosphorylation was determined for forward reactions under pseudo first-order rate conditions [38] in which the concentration of the designated inositol phosphate was fixed at saturating levels (10 μM; Table 1), and the concentration of nucleotide was varied as indicated. Each individual data point was analysed by HPLC as described under the Materials and methods section. The Michaelis–Menten equation was fitted to the data using non-linear regression (GraphPad Prism v5.03). Results are presented as the means±S.E.M. (C) Either no enzyme (broken line) or 0.011 μg/ml PPIP5K2^KD (solid line) was incubated with 100 nM 5-[³H]InsP₇ and 10 mM ATP for 7 min as indicated in the Materials and methods section. The reactions were quenched and neutralized and analysed by Partisphere SAX HPLC as described in the Materials and methods section. Representative HPLC data are shown. (D) The initial velocity of 5-InsP₇ phosphorylation was determined as described in the legend to panel (B) for InsP₆.

Figure 4. Dephosphorylation reactions of PPIP5K2^KD and substrate saturation plots for ATP and ADP.

(A) Either no enzyme (broken line) or 0.56 μg/ml PPIP5K2^KD (solid line) was incubated with 10 μM [³H]InsP₈ and 5 mM ADP for 10 min as indicated in the Materials and methods section. The reactions were quenched and neutralized and analysed by Partisphere SAX HPLC as described in the Materials and methods section. Representative HPLC data are shown. (B) The initial velocity of InsP₈ dephosphorylation was determined for reverse reactions under pseudo first-order rate conditions [38] in which the concentration of the designated inositol phosphate was fixed at saturating levels (10 μM; Table 1), and the concentration of nucleotide was varied as indicated. Each individual data point was analysed by HPLC as described under the Materials and methods section. The Michaelis–Menten equation was fitted to the data using non-linear regression (GraphPad Prism v5.03). Results are presented as the means±S.E.M. (C) Either no enzyme (broken line) or 27 μg/ml PPIP5K2^KD (solid line) was incubated with 10 μM 1-[³H]InsP₇ and 28 μM ADP for 45 min as indicated in the Materials and methods section. The reactions were quenched and neutralized and analysed by Partisphere SAX HPLC as described in the Materials and methods section. Representative HPLC data are shown. (D) The initial velocity of 1-InsP₇ dephosphorylation was determined as described in the legend to panel (B) for InsP₈.

Table 1. Kinetic parameters of PPIP5K2^KD:PP-InsP synthesis.

Kinetic parameters were determined under pseudo first-order conditions [38] for phosphorylation of InsP₆ or 5-InsP₇ for both ATP and InsP substrates (see the Materials and methods section). When fixed, ATP was 10 mM and InsP 10 μM. Initial velocities were measured by conversion of an appropriate [³H]-radiolabelled inositol phosphate tracer (4000 DPM) as analysed by Partisphere 5 SAX HPLC (e.g., see Figures 3A and 3C) for each data point. The Michaelis–Menten equation was fit to plots of initial velocity against substrate concentration using non-linear regression (GraphPad Prism v5.03). No substrate inhibition was observed (results not shown). Results are presented as the means±S.E.M. (n≥2 independent experiments).

		Km (μM)				kcat/Km (M−1·s−1)
Substrates		Nucleotide	InsP	Vmax (nmol·min−1·mg−1)	kcat (s−1)	Nucleotide	InsP
ATP	InsP6	37±5	0.39±0.09	43±2	0.030±0.001	7.9×102	7.6×104
ATP	5-InsP7	22±7	0.060±0.009	190±10	0.13±0.01	5.9×103	2.2×106

Using saturating concentrations of either InsP₆ or 5-InsP₇, we next derived a substrate saturation plot for ATP; this showed Michaelis–Menten kinetics (Figures 3B and 3D). The value of the K_m for ATP (22–37 μM; Table 1) was similar for both InsP₆ and 5-InsP₇. At an [ATP] of 1 mM, the V_max values were already attained (Figures 3B and 3D); there was also no significant increase in reaction velocity when [ATP] was further increased to 10 mM (results not shown). These data lead us to conclude that PP-InsP synthesis by PPIP5Ks is protected from physiologically relevant changes in [ATP] (1–5 mM [46]), contrary to the conclusion of a previous report [25]. As well as providing this new information about the characteristics of PPIP5Ks, our kinetic data (Table 1) can also be viewed as a successful quality control test of the recently developed PAGE-based procedures for purification of enzymatically prepared PP-InsPs [35].

Reversibility of PPIP5K and the consequences of substrate competition

Several inositol phosphate kinases have been shown to be physiologically reversible, that is, their relative rates of phosphorylation and dephosphorylation can be influenced by changes in cellular [ATP]/[ADP] ratios: IP6K [26], InsP₅ 2-kinase [31] and ITPK1 (inositol 1,3,4-trisphosphate 5/6-kinase) [47]. A previous report [25] has included the PPIP5Ks in this category. The degree of the ATP-synthase activities of IP6Ks and PPIP5Ks were further proposed [25,26] to be regulated by physiological changes in adenine nucleotide levels. This idea, which dovetails with the concept that PP-InsPs are highly energetic phosphate donors [26,27], has become well accepted in the field [32–34,48].

We found that PPIP5K2^KD was reversible in vitro when it was incubated with ADP plus either 1-InsP₇ or InsP₈ (Figure 4). Substrate saturation curves showed the reactions to follow Michaelis–Menten kinetics (Figures 4B and 4D). The dephosphorylation of InsP₈ proceeded at a slightly greater V_max (Tables 1 and 2) than did the corresponding forward reaction (5-InsP₇ phosphorylation). For the dephosphorylation of InsP₈ the K_m for ADP was 0.9 mM (Table 2), which is within the range of physiological total ADP concentrations (0.7–1.7 mM; [46]). Interestingly, for the dephosphorylation of 1-InsP₇, the value for the K_m of ADP was 175-fold lower, and thus much closer to the K_m values for ATP in the phosphorylation reactions (Table 2).

Table 2. Kinetic parameters of PPIP5K2^KD:ATP synthesis.

Kinetic parameters were determined under pseudo first-order conditions [38] for dephosphorylation (ATP synthesis) of InsP₈ or 1-InsP₇ for both ADP and InsP substrates (see the Materials and methods section). When fixed, ADP was 10 mM and InsP 10 μM. Initial velocities were measured by conversion of an appropriate [³H]-radiolabelled inositol phosphate tracer (4000 DPM) as analysed via Partisphere 5 SAX HPLC (e.g., see, Figures 4A and 4C) for each data point. The Michaelis–Menten equation was fit to plots of initial velocity against substrate concentration using non-linear regression (GraphPad Prism v5.03). No substrate inhibition was observed (results not shown). Results are presented as the means±S.E.M. (n≥2 independent experiments).

		Km (μM)				kcat/Km (M−1·s−1)
Substrates		Nucleotide	InsP	Vmax (nmol·min−1·mg−1)	kcat (s−1)	Nucleotide	InsP
ADP	InsP8	900±280	0.022±0.011	270±30	0.19±0.02	2.1×102	8.5×106
ADP	1-InsP7	5.2±1.5	0.11±0.01	2.8±0.1	0.002±0.000	3.7×102	1.7×104

The high rate of dephosphorylation of InsP₈ by PPIP5K2^KD was somewhat surprising in view of our earlier structural data [20] which indicated that, in the crystal complex of PPIP5K2^KD, the InsP₈ product was under considerable steric strain. It had seemed reasonable to conclude that the increase in entropy upon product release would discourage re-binding of InsP₈ for dephosphorylation. Thus, we soaked InsP₈ into a crystal complex of PPIP5K2^KD and ADP. We found the conformation and catalytic environments of this complex to be very similar (results not shown) to those that we described previously when InsP₈ was synthesized inside the crystals [20]. We therefore have no evidence that the mechanism of InsP₈ dephosphorylation proceeds any differently than by a reversal of 5-InsP₇ phosphorylation.

One of the other inositol phosphate kinases that is reversible, ITPK1, is PPIP5K's closest structural homologue [20]. Both PPIP5K2 and ITPK1 envelop ATP tightly within two sets of antiparallel β-sheets that comprise an ATP-grasp fold; in each case, less than 10% of the nucleotide is solvent exposed [20,47]. In the case of ITPK1, the bound nucleotide promotes a phosphotransferase activity by which the dephosphorylation of Ins(1,3,4,5,6)P₅ to Ins(3,4,5,6)P₄ is stimulated by the phosphorylation of Ins(1,3,4)P₃ [47]. The analogous reaction for PPIP5K would be the stimulation of InsP₈ dephosphorylation by the presence of InsP₆ (Figure 5A). To investigate if that were possible, we assayed the ‘reverse’ (i.e., InsP₈ dephosphorylation) activity of PPIP5K2^KD in the presence of physiological [InsP₆] (Figure 5B). Unlike ITPK1, the phosphorylation of one substrate (InsP₆) by PPIP5K2^KD merely inhibited (by >70%) the dephosphorylation of another (InsP₈) (Figure 5B). We conclude that even though both ITPK1 and PPIP5K are ATP-grasp kinases, only ITPK1 can act as a phosphotransferase.

Figure 5. PPIP5K2^KD is not a phosphotransferase: InsP₆ inhibits InsP₈ dephosphorylation by PPIP5K2^KD.

(A, top) This graphic depicts how Ins(1,3,4)P₃ stimulates Ins(1,3,4,5,6)P₅ dephosphorylation by ITPK1 [47]. The 1-phosphate group (shown in red) that is removed from Ins(1,3,4,5,6)P₅ is not released into the bulk phase but is retained by the nucleotide and donated to Ins(1,3,4)P₃; this returns the enzyme to its ADP state, ready for another cycle of Ins(1,3,4,5,6)P₅ dephosphorylation. (A, bottom) The analogous potential reaction sequence for PPIP5K. (B) Either no enzyme (broken line) or 0.63 μg/ml PPIP5K2^KD (solid line; dotted line) was incubated with 10 μM [³H]InsP₈ and 10 mM ADP for 10 min as indicated in the Materials and methods section. Where indicated (dotted line) incubations also contained 10 μM InsP₆. The reactions were quenched and neutralized and analysed by Partisphere SAX HPLC as described in the Materials and methods section. Representative HPLC data are shown.

The inhibition of InsP₈ dephosphorylation by InsP₆ was also demonstrated by using a luciferase-based ATP-detection method (IC₅₀=0.32 μM; Figure 6). The latter result indicates that the luciferase assay could be employed as a high-throughput tool to screen for inhibitors of this enzyme. For example, by using this method we found that InsS₆ also inhibited the enzyme (IC₅₀=0.48 μM; Figure 6).

Figure 6. InsP₆ and InsS₆ inhibit InsP₈ dephosphorylation by PPIP5K2^KD; a luciferase-based ATP assay.

A total of 0.3 μg/ml PPIP5K2^KD was incubated at 37°C with 1.4× reaction buffer (ATP Determination Kit, Molecular Probes), 650 μM ADP and 250 nM InsP₈ for 15 min as described in the Materials and methods section. InsS₆ and InsP₆ were included in the assays as indicated. Then 16 μl of either an ATP standard solution or reaction solution was added into 100 μl Standard Reaction Solution (ATP Determination Kit, Molecular Probes) and luminescence was recorded after 15 s using a BioTek Synergy 2. Data are presented as the means±S.E.M. (n≥3). A dose–response curve was fit using GraphPad Prism v5.03. The IC₅₀ values for InsS₆ and InsP₆ were 0.48 and 0.32 μM, respectively.

Equilibrium and rate studies of PPIP5K2^KD under varied [ATP] and [ADP]

Tables 1 and 2 show that PPIP5K2^KD is reversible in vitro and also InsP₈ dephosphorylation was associated with a K_m value for ADP that has potential physiological relevance. We therefore next investigated to what extent changes in the initial ATP/ADP ratios would influence either the initial reaction rate or equilibrium point of the kinase reaction.

In planning these experiments, we noted that total cellular [ATP] generally varies between 1 and 5 mM, with total cellular ATP/ADP ratios generally in the range of 2–12 [46,49,50]. At physiologically relevant [ATP] and [ADP] concentrations of 5 and 0.5 mM, respectively, the PPIP5K2^KD-catalysed equilibrium point was attained at near complete (80–90%) phosphorylation of either InsP₆ or 5-InsP₇ (Figures 7A–7C). Furthermore, the PPIP5K equilibrium point for the phosphorylation of 5-InsP₇ to InsP₈ was neither affected by varying total ATP (1 or 5 mM) nor by changing the ATP/ADP ratio from 2 to 20 (Figure 7C). We also assayed PPIP5K2^KD activity under physiological extremes of ATP/ADP ratio (2 against 20), using both 1 and 5 mM total [ATP] (results not shown). The changes that we made to the ATP/ADP ratios did not alter reaction rates. These data indicate that the reversibility of PPIP5Ks is not biologically significant in vivo, in contrast to a previous proposal [25]. Although the V_max for InsP₈ dephosphorylation is relatively high, the enzyme's relatively low affinity for ADP may in part explain why that reaction is not favoured, as long as physiological levels of ATP are also present. Our data also demonstrate that the synthesis of both 1-InsP₇ and InsP₈ is likely well-protected from the consequences of the deteriorating [ATP]/ [ADP] ratio that cells experience when undergoing bioenergetic stress.

Figure 7. Effect of ATP/ADP and AMP upon the equilibrium condition for PPIP5K2^KD.

(A) and (B) show a time course of InsP₆ and 5-InsP₇ phosphorylation, respectively. In (A), 20 μg/ml enzyme was incubated with reaction buffer (see the Materials and methods section) containing 15 μM InsP₆, 5 mM ATP and 0.5 mM ADP. In (B), 0.33 μg/ml enzyme was incubated with reaction buffer (see the Materials and methods section) containing 1 μM 5-InsP₇ and 5 mM ATP. Assays were quenched and neutralized and analysed by Partisphere SAX HPLC as described in the Materials and methods section. (C) and (D) show the equilibrium point for the phosphorylation of 5-InsP₇ to InsP₈ in which 0.33 μg/ml enzyme was incubated as described in the Materials and methods section under initial conditions of 1 μM 5-InsP₇ and either 1 (squares) or 5 (diamonds) mM ATP; either the ATP/ADP ratio (C) or the ATP/AMP ratio (D) was varied as indicated. Assays were conducted for 60 min. The reactions were then quenched and neutralized and analysed by Partisphere SAX HPLC as described in the Materials and methods section.

Equilibrium and rate studies of PPIP5K2^KD under varied [AMP]

InsP₈ levels in vivo have been found to decrease in response to mild bioenergetic stress that is reflected more in an increase in [AMP] rather than a decrease in [ATP]/[ADP] [15]. However, we did not find any effect of AMP upon the equilibrium condition catalysed by PPIP5K2^KD (Figure 7D). We further found that these same changes in [ATP]/[AMP] ratios did not affect PPIP5K2^KD reaction rate (results not shown). Thus the decrease in InsP₈ in relation to elevated [AMP] in vivo [15] likely proceeds through a signalling pathway, rather than through direct effects on InsP₈ synthesis by PPIP5K2. We [15] have previously excluded AMPK (AMP-activated protein kinase) from being involved, so we will need to consider other mechanisms.

The stoichiometry of PPIP5K2^KD

Reaction stoichiometry is pertinent to the regulation of metabolic pathways [16]. Moreover, since PPIP5K activity consumes ATP, stoichiometry is also highly relevant to the potential interface of PP-InsPs with cellular bioenergetic status [5,15]. We [20] previously demonstrated that PPIP5K2 showed a slow ATPase activity (0.5 nmol min⁻¹ mg⁻¹) when it was incubated with 0.1 mM ATP in the absence of inositol phosphate substrate. That did not seem significant in relation to the V_max values for InsP₆ and 5-InsP₇ phosphorylation (43 and 190 nmol min⁻¹ mg⁻¹; Table 1). However, in the current study we found that at physiological [ATP] (5 mM), the inositol phosphate-independent ATPase activity increased to approximately 12 nmol min⁻¹ mg⁻¹ (Figures 8A and 8B). Moreover, the rate of ATP hydrolysis was further elevated up to 4-fold as increasing concentrations of either InsP₆ or 5-InsP₇ were added, with a maximally effective concentration of 25 μM (Figure 8A).

Figure 8. ATPase activity of PPIP5K2^KD.

(A) and (B) Orthophosphate released from the hydrolysis of ATP was measured using a Malachite Green detection method [55]. The 27–270 μg/ml PPIP5K2^KD was incubated at 37°C for 120 min in 50 μl reactions containing 20 mM Tris/HCl, pH 7.5, 10 mM ATP (unless otherwise indicated), 100 mM KCl, 0.1 mM EDTA and 5 mM MgCl₂. (A) Stimulation of PPIP5K2^KD ATPase activity by InsP₃, InsP₄, InsP₅, InsP₆ and 5-InsP₇ (concentrations as indicated, [ATP]=5 mM). (B) Effect of ATP concentration (0.2, 1 or 5 mM) on basal (absence of InsP) and InsP-stimulated (25 μM) ATPase activity. Results are presented as the means±S.D. (n=3). (C) and (D) provide representative HPLC data indicating that 15 μM InsP₅ (dotted lines) does not inhibit phosphorylation of either InsP₆ or 5- InsP₇, respectively. In (C), 20 μg/ml PPIP5K2^KD was incubated for 15 min with reaction buffer (see the Materials and methods section) containing 15 μM [³H]InsP₆, 5 mM ATP and 0.5 mM ADP. In (D), 0.41 μg/ml PPIP5K2^KD was incubated for 5 min with reaction buffer (see the Materials and methods section) containing 1 μM 5-[³H]InsP₇, 5 mM ATP and 0.5 mM ADP. The reactions were quenched and neutralized and analysed by Partisphere SAX HPLC as described in the Materials and methods section.

InsP₃ (inositol 1,3,4-trisphosphate), InsP₄ (inositol 1,3,4,5-tetrakisphosphate) and InsP₅ also stimulated ATPase activity (Figures 8A and 8B). These InsPs elevated ATPase activity even at low (0.2 μM) [ATP] (Figure 8B). It is surprising that this ATPase activity can be stimulated by an inositol phosphate that is neither a substrate [20] nor an inhibitor (Figures 8C and 8D). Nevertheless, such data (Figures 8C and 8D) lead us to conclude that even the relatively high levels of InsP₅ that occur in cells cannot access the enzyme's active site to stimulate ATPase activity when InsP₆ and 5-InsP₇ are available.

A comparison of the rates of the ATPase activity (Figures 8A and 8B) in relation to the kinase activity (Table 1) indicates that approximately two molecules of ATP are consumed for every molecule of InsP₆ that is phosphorylated. As the V_max is higher for 5-InsP₇ phosphorylation compared with InsP₆ phosphorylation (Table 1), there is a much smaller impact of the ATPase activity upon that reaction stoichiometry (1.2 molecules of ATP are consumed for every molecule of 5-InsP₇ that is phosphorylated). In other words, the bioenergetic cost to the cell is much smaller for 5-InsP₇ phosphorylation (17% ATP ‘wastage’) than it is for InsP₆ phosphorylation (50% ATP ‘wastage’). Thus, the synthesis of InsP₈ by pathway I is more energetically efficient than is pathway II (Figure 1).

An analogous ATPase reaction has been observed in ATP-grasp ligases [51], but to our knowledge this is the first demonstration that this is a property of an ATP-grasp kinase. We propose that the dynamic nature of the active site and the partly associative reaction mechanism make it inevitable that there will be some non-productive ATPase activity. We [20] have also previously determined that InsP₆ is more mobile within the active site than is 5-InsP₇. That may explain why the phosphorylation of InsP₆ is less efficient.

We studied the mechanism of this ATPase activity through structural and mutagenesis studies. Previous analysis [20] of the crystal structure of the PPIP5K2^KD indicated that there are two water molecules (W1 and W2; Figure 9A) that are candidates to carry out nucleophilic attack on the γ-phosphate of ATP (Figure 9A). The two amino acid residues that are candidates for activating these putative nucleophiles are Arg²¹³ and Lys²⁴⁸, respectively (Figure 9A). We therefore assayed ATPase activities of R213A and K248A mutants of PPIP5K2^KD. Both mutants showed a reduction in InsP₆-stimulated ATPase activity, as expected since Arg²¹³ and Lys²⁴⁸ are required for binding of InsP₆ [20]. In contrast, only the K248A mutant showed a significant reduction in the rate of inositol phosphate-independent ATPase activity (Figure 9B). We therefore conclude that W2 is the water molecule that is activated for nucleophilic attack upon the γ-phosphate of ATP.

Figure 9. The effects of R213A and K248A mutations upon ATPase activity of PPIP5K2^KD.

(A) Structure of a portion of the wild-type PPIP5K2^KD catalytic centre [20] showing water molecules bridging the interaction between the ATP analogue AMPPNP (adenosine 5′-[β,γ-imido]triphosphate) and R213 and K248. (B) The effect of single-site mutants of PPIP5K2^KD on basal and InsP₆-stimulated (25 μM) ATPase activity. ATPase activity was measured as in Figure 8.

CONCLUDING COMMENTS

Our new kinetic information concerning PPIP5K stoichiometry, rates, substrate affinities and equilibrium conditions assists our understanding of the dynamic behaviour of this signalling enzyme in vivo and the role it has in PP-InsP generation. Since the catalytic and structural features of the active sites of PPIP5K1 and PPIP5K2 are so similar [18,20], we believe the conclusions that we have made are equally applicable to both enzymes. Furthermore, our studies with recombinant protein have led us to conclude that contrary to an earlier report [25], the kinetic characteristics of PPIP5Ks do not match the synthesis of InsP₈ to fluctuations in adenine nucleotide levels. Instead, the opposite is true; the kinetic properties of PPIP5Ks ensure that InsP₈ synthesis is protected against cellular bioenergetic stress. The features of this enzyme stand in contradiction to the reduction in InsP₈ levels that accompanies experimental manipulations in cultured cells that are generally thought to invoke mild bioenergetic stress [15]. It will be important to uncover the signalling pathways that are involved and their biological relevance.

AUTHOR CONTRIBUTION

Jeremy Weaver, Huanchen Wang and Stephen Shears designed the experiments, analysed the results and wrote the paper. Jeremy Weaver and Huanchen Wang conducted the experiments.

FUNDING

This work was supported by the Intramural Research Program of the National Institutes of Health/National Institute of Environmental Health Sciences.