Structural and catalytic analyses of the InsP₆ kinase activities of higher plant ITPKs

Abstract

Inositol phosphate signaling in plants is of substantial agricultural interest, with a considerable focus on the inositol tris/tetrakisphosphate kinase (ITPK) family of inositol phosphate kinases. Historically, the 4–6 isoforms of ITPKs that higher plants each express have been studied for their multiplexing a metabolic pathway to synthesis of inositol hexakisphosphate (i.e., InsP₆ or phytate), through the phosphorylation and dephosphorylation of multiple inositol phosphates, including Ins(1,3,4,5,6)P₅ (inositol-1,3,4,5,6-pentakisphosphate). A more recent discovery is ITPK-catalyzed phosphorylation of InsP₆ to inositol pyrophosphates, which regulate plant immunity and phosphate homeostasis. However, a molecular-based explanation for these alternate catalytic activities has been missing, because no plant ITPK structure has previously been solved. Herein, we provide biochemical and structural analyses of ITPKs from Zea mays and Glycine max. For this work we introduce a simple, enzyme-coupled microplate-based assay of InsP₆ kinase activity that should promote more general access to this important field. Furthermore, a ZmITPK1/InsP₆ crystal complex is described at a resolution of 2.6 Å, which identifies a number of catalytically-important residues; their functionality is confirmed by mutagenesis. We further demonstrate ZmITPK1 adds a β-phosphate to the 3-position of Ins(1,2,3,4,5)P₅, yielding a candidate signal for regulating phosphate homeostasis. An impactful discovery is our description of a 29-residue catalytic specificity element; by interchanging this element between GmITPK1 and GmITPK2, we demonstrate how its isoform-specific sequence specifically determines whether the host protein phosphorylates InsP₆, without substantially affecting Ins(1,3,4,5,6)P₅ metabolism. Our structural rationalization of key catalytic differences between alternate ITPK isoforms will complement future research into their functional diversity.

Introduction

Inositol phosphates are a large multifunctional family of signaling molecules that are enzymatically synthesized from the combinatorial positioning of phosphate groups around the inositol ring through a network of specialized kinases and phosphatases (1–3). All eukaryotes utilize these signaling pathways to regulate a number of fundamental functions such as energy metabolism, phosphate homeostasis and immune responses (2).

Among wide-ranging research into inositol phosphate metabolic enzymes in plants, the ITPK (inositol tris/tetrakisphosphate kinase (4)) gene family receives particular attention (2). The founding member of this family was originally discovered in mammalian cells through its ability to phosphorylate Ins(1,3,4)P₃ (5, 6). Higher plants each express either four or six isomers of ITPKs isomers (4, 7, 8). These enzymes multiplex a metabolic pathway to Ins(1,3,4,5,6)P₅ synthesis through the phosphorylation and dephosphorylation of multiple inositol phosphate isomers (4, 7). The Ins(1,3,4,5,6)P₅ end-product is then phosphorylated by a separate and selective 2-kinase to produce InsP₆, which forms deposits that are of particular importance as storages depot for inositol and phosphate (3, 9).

An extension to the purely metabolic function for ITPKs is their generally robust catalysis of a reversible Ins(3,4,5,6)P₄ 1-kinase / Ins(1,3,4,5,6)P₅ 1-phosphatase activity (4, 7, 10), although the low Ins(3,4,5,6)P₄ 1-kinase activity of AtITPK4 suggests this is not a universal activity (4). In mammals, Ins(3,4,5,6)P₄ is a well-characterized inhibitor of a Ca²⁺-activated Cl⁻ conductance in the plasma membrane that influences salt and fluid secretion and cell migration (11, 12). Ins(3,4,5,6)P₄ appears to have a similar inhibitory action upon a Cl⁻ conductance that regulates pollen tube growth (13).

More recently, it has been discovered that some higher plant ITPK orthologs harbor InsP₆ kinase activity, thereby generating 5-InsP₇, an inositol pyrophosphate (PP-InsP) (14–16). This particular catalytic activity is important for several reasons: 5-InsP₇ regulates a functionally diverse set of proteins through a unique form of covalent modification known as ‘pyrophosphorylation’ (17, 18). Additionally, 5-InsP₇ is itself a precursor for 1,5-InsP₈, which enhances plant immunity (19), promotes tolerance to drought stress (20), and maintains inorganic phosphate (Pi) homeostasis in both plants and animals (15, 21, 22). In plants, the binding of InsP₈ to the SPX1 receptor promotes the latter’s inhibitory interaction with the phosphate starvation response 1 (PHR1) transcriptional regulator. When Pi supply becomes limiting, cellular 1,5-InsP₈ levels decrease, uncoupling SPX1 from PHR1, leading to transcription of Pi starvation-induced genes (22). Interestingly, changes in Pi-homeostasis have recently been linked to adjustments in plant defenses; plant immunity is suppressed upon Pi-depletion, to allow successful colonization of phosphate-remobilizing microbiota to enhance Pi availability and uptake (23). Pi-depleted conditions also inhibit 1,5-InsP₈ synthesis, likely contributing to the mechanism by which plant immunity is suppressed. ITPKs have been implicated in contributing to this regulatory relationship between immunity and Pi balance, as 5-InsP₇ synthesis by these kinases is thought to be regulated by changes in cellular [ATP], which go hand-in-hand with fluctuations in Pi levels (15, 24).

There is some emerging evidence that the various ITPK isoforms may differ in their ability to phosphorylate InsP₆ (16) but the chromatographic methods in general use that assay this particular kinase activity (such as thin layer chromatography, gel-electrophoresis or HPLC) are laborious, low throughput, and in the case of HPLC, rather expensive to operate. In the current study we describe a new enzyme-coupled InsP₆ kinase assay that can be performed in a 96-well microplate format. We believe this simple assay will make this field of research more accessible, and thereby improve our understanding of the various physiological roles of alternate ITPKs.

Another item we have addressed in the current study is the absence of structural data for any plant ITPK, which hinders a formulation of a molecular understanding of the catalytic diversity of this family of enzymes in plants. A certain amount of modeling (e.g., (24)), has been attempted based on previously published structures of the human (25) and Entamoeba (26) orthologs, but there are considerable uncertainties as to the accuracy of such predictions, particularly with regards to the side chain conformations of catalytic residues.

In the current study, we have studied ITPK1 from both Zea mays and Glycine max; we present X-ray analysis of crystal complexes, together with data arising from molecular engineering of these enzymes. We also present ITPK1 kinetic parameters. These experiments have led us to identify an important structural feature that has not previously been attributed to this enzyme family: a catalytic specificity element that, through isoform-dependent variations in its primary sequence, can reciprocally specify the relative rates of InsP₆ kinase and Ins(1,3,4,5,6)P₅ phosphatase activities of individual members of this versatile enzyme family.

Materials and Methods

Protein Expression and Purification

The codon-optimized cDNA of ZmITPK1 (Zea mays ITPK1, Uniprot: Q84Y01) was purchased from Genscript Inc. The Gateway expression system (Invitrogen) was used to subclone ZmITPK1 into the pDest-566 vector. This vector encodes tandem His₆ and maltose-binding protein tags that are linked to the N-terminus through a tobacco etch virus protease cleavage site. The recombinant plasmid was transformed into competent E. coli BL21(DE3). An overnight culture of the transformed E. coli cells was inoculated into nutrient-rich 2×YT medium (16g/L Tryptone, 10g/L yeast extract and 5g/L NaCl at pH 7.5) which was cultured at 37 °C to an optical density of 0.7 at 600 nm. Isopropyl β-d-thiogalactopyranoside (0.1 mM) was then added and cultures were continued at 15 °C for 22 hours. For preparation of selenomethionine-labeled ZmITPK1, the E. coli cells were inoculated into a synthetic M9 minimal medium that was supplemented with the following amino acids: 100 mg/L Lys, 100 mg/L Thr, 100 mg/L Phe, 50 mg/L Leu, 50 mg/L Ile, 50 mg/L Val and 50 mg/L selenomethionine.

Codon-optimized cDNAs of His₆MBP-GmITPK1 (Glycine max ITPK1, Uniprot: A7X657) and His₆MBP-GmITPK2 (Uniprot: A7X665) were purchased from Genscript Inc. They also were subcloned into pDest-566 vector and separately used to transform competent E. coli BL21(DE3); proteins were expressed as described above.

The cDNA of full-length HsITPK1 (25) was subcloned into pDest-566 vector and transformed into competent E. coli BL21(DE3) with chaperone plasmid pGro7 (Takara Bio Inc). An overnight culture of the transformed E. coli cells was inoculated into nutrient-rich 2×YT medium supplemented with 0.07% (w/v) L-arabinose and 30 mg/L chloramphenicol and grown at 37 °C to an optical density of 0.7 at 600 nm. Isopropyl β-d-thiogalactopyranoside (0.1 mM) was then added and cultures were continued at 15 °C for 22 hours.

All proteins were purified in the following manner: E. coli cells were harvested by centrifugation at 5000 × g for 10 min and disrupted using a Constant Cell Disruption System (Constant System Ltd) at 20 KPsi. Recombinant protein was then purified at 4 °C as follows: cell supernatant was first mixed with Ni-NTA agarose (Qiagen; 5 ml agarose per liter of cells), then washed with 20 column volumes of buffer containing 300 mM NaCl, 20 mM Tris-HCl, pH 7.2, 20 mM imidazole, and the target protein was eluted with the same buffer containing 250 mM imidazole. The eluate was loaded to 5ml HiTrap^™ Heparin HP column (GE Healthcare) and eluted with 10 column volumes of a 50–2000 mM NaCl gradient in 20 mM Tris-HCl, pH 7.2. Finally, protein was applied to a Superdex^™ 200 size exclusion column (GE Healthcare) and eluted with 150 mM NaCl, 20 mM Tris-HCl pH 7.2. These preparations (designated His₆MBP-ZmITPK1) were concentrated and stored at −80 °C; purity was validated by 4–20% SDS-PAGE.

For some experiments, including all structural studies of ZmITPK1, the enzyme fractions in the HiTrap^™ Heparin HP column eluate described above were treated with tobacco etch virus protease to remove both N-terminal tags. Samples were then re-run through the HiTrap^™ Heparin HP column and then further purified on a Superdex^™ 200 size exclusion column as described above.

Mutants were prepared using a site-directed mutagenesis kit (Strata gene) and Q5 site-directed mutagenesis kit (Biolabs), mutants were verified by sequencing.

The His₆-HsDIPP1 was prepared as previously described (27).

Enzyme assays.

Enzyme activity was performed at 25°C. Unless otherwise stated, the enzyme-coupled Pi-release assays were performed for 30–150 minutes in 100 μl reaction mixtures containing 0.007–1.2 mg/ml wild-type or mutant ITPK, 0.05 mg/ml His₆-HsDIPP1, 20 mM HEPES pH 7.2, 100 mM KCl, 3.5 mM MgCl₂, 20 μM EDTA, 1 mM ATP (Gold Biotechnology, Catalog number A-081–1), and 25 μM of substrate (either InsP₆ (Sigma Aldrich, P8810), Ins(2,3,4,5,6)P₅, Ins(1,2,3,4,5)P₅, Ins(1,3,4,5,6)P₅, Ins(1,2,4,5,6)P₅, Ins(1,2,3,5,6)P₅, or Ins(1,2,3,4,6)P₅ (thus, less than 2.5% of the ATP can be consumed in these assays). All InsP₅ isomers were purchased from Sichem. Reactions were quenched by addition of 100 μl of phosphate detection reagent (36:1 v/v of 2.6% sodium molybdate in 2.5 M HCl : 0.126% malachite green chloride). Pi release was quantified from the absorbance at 620 nm (28). The Z’-factor was calculated as described previously (29).

For assays of ITPK1 that included [³H]-InsP₆ (20,000–40,000 d.p.m.; 25 μM) (American Radiolabeled Chemicals, Inc., ART 1915) or [¹⁴C]Ins(1,3,4,5,6)P₅ (80,000 d.p.m; (30); 10 μM) the reaction medium was as described above except that DIPP1 was not included in the reaction buffer, only 1–2 μg ZmITPK1 was added, and for the Ins(1,3,4,5,6)P₅ phosphatase reactions, 1 mM ADP replaced the ATP. The InsP₆ kinase activity of 77 μg HsITPK1 was also studied using [³H]-InsP₆ in the same DIPP1-free medium. In other assays with ZmITPK1, 0.1 mg of protein was incubated for 120 min in 100 μl buffer containing 1 μCi [³³P]-γ-ATP, 1mM InsP₆ or Ins(1,2,3,4,5,)P₅, and 20 mM HEPES pH 7.2 100mM KCl, 20 μM EDTA, 8mM MgCl₂, 2mM ATP, 3 units of creatine kinase, and 5mM creatine phosphate.

All assays containing radiolabels were quenched with 0.2 volumes of 2 M perchloric acid + 1 mg/ml of InsP₆, neutralized, and analyzed by ion-exchange HPLC, using a 4.6 × 125 mm, 5 μm PartiSphere SAX column. The elution gradient (1 ml/min) was generated by mixing Buffer A (1 mM Na₂EDTA) with Buffer B (Buffer A plus 2.5 M NH₄H₂PO₄, pH 3.9); the elute was mixed with 2.5 ml/min Monoflow4 scintillation liquid (National Diagnostics) and radioactivity was monitored in-line with a Beta-RAM 6 radio flow detector (LabLogic).

Crystallization

The full length wild-type apo-ZmITPK1 (7TN5; Supplementary Table 1) or the Se-Met derivative (Supplementary Table 1) were crystallized by hanging drop vapour diffusion against a well buffer of 1.0 M potassium sodium tartrate, 100mM imidazole pH 8.0 and 200mM NaCl. The construct of ZmITPK1 (residues ¹⁸⁻²¹⁸Gly-Ser-Gly-Ser-Gly^248−328) was crystallized by hanging drop vapour diffusion against a well buffer of 1.4 M ammonium phosphate dibasic, 20% ethylene glycol, 100mM Tris pH 8.0 (7TN7; Supplementary Table 1). Crystals of full length ZmITPK1^H192A or ZmITPK1^F189A/H192A were formed in 12% PEG3350, 100mM HEPES pH 7.0, 200mM calcium acetate and 10% Glycerol. For some experiments, the latter two crystals were subsequently soaked in buffer containing 30% PEG3350, 100mM sodium acetate pH 5.0, 5mM AMP-PNP and 15% glycerol (7TN6 and 7TN3; Supplementary Table 1); note that no electron density for the nucleotide was detected. In other experiments, 10 mM InsP₆ was soaked into ZmITPK1^H192A crystals in 4 μl buffer containing 30% PEG3350, 100mM sodium acetate pH 5.0, and 15% glycerol (7TN8; Supplementary Table 1).

To identity reaction products of ZmITPK1 kinase activity, 1 mM of either InsP₆ or Ins(1,2,3,4,5)P₅ was incubated with 2 mg/ml ZmITPK1 enzyme at 37°C for 12 h in 100 μl reaction buffer containing 20 mM Tris-HCl, pH 7.2, 150 mM NaCl, 2 mM MgCl₂, 50 μM Na₂EDTA and 3 mM ATP. The reactions were then dried under a vacuum at room temperature, and then 10 μl of soaking buffer was added: 200 mM LiCl, 20% (w/v) PEG 8000, 30% (v/v) isopropanol, 20 mM MgCl₂, 100 mM NaF, and 100 mM NaAc, pH 5.0. A precipitate was discarded by centrifugation and the supernatant was added to crystals of HsDIPP1 (residues 1–148) for 12h at 25°C (27).

Data collection, structure determination, and refinement.

Diffraction data were collected using APS beam lines 22-ID and 22-BM. All data were processed with the program HKL2000. The initial structures of ZmITPK1 were determined by single-wavelength anomalous dispersion with Autosol in PHENIX follow by the default autobuild procedure. The initial structure was manually rebuilt with COOT and refined with PHENIX. The molecular graphics representations were prepared with the program PyMol (Schrödinger, LLC). Atomic coordinates and structural factors have been deposited with the Protein Data Bank with accession codes 7TN3, 7TN4, 7TN5, 7TN6, 7TN7, 7TN8 (and see Supplementary Table 1).

Results

Characteristics of PP-InsP synthesis by ZmITPK1.

Many higher plants each express multiple isoforms of ITPKs, some of which have been shown to add a β-phosphate to InsP₆ to yield 5-InsP₇ (14–16). These particular enzymatic activities have classically been demonstrated by HPLC (31), and more recently by gel electrophoresis (32); the latter is a more accessible method, but both techniques are relatively laborious and low throughput. For this study, we have developed a higher-throughput method to assay the InsP₆-kinase activity of ZmITPK1: we describe an enzyme-coupled assay in which the newly added β-phosphate is cleaved by recombinant human DIPP1 (33), and the resulting inorganic phosphate (Pi) is assayed in a microplate-format colorimetric assay (Figure 1A). This new method yields consistent data over a range of assay conditions (Figure S1A,B,C). The performance of high-throughput assays is typically ascertained from its Z’-factor; values between 0.5 and 1 are considered high quality (29). The Z’-factor for our coupled assay was calculated to be 0.66. Moreover, the InsP₇ product is dephosphorylated back to InsP₆ by DIPP1 during the kinase assay, thereby preventing the ITPK1 acting “in reverse” as an InsP₇ phosphatase. The kinetic data obtained from the enzyme-coupled assay are similar to those we obtained by HPLC analysis of radiolabeled [³H]InsP₆ substrate (Figure 1B,C; Figure S1D,E). We further verified the phosphorylation of InsP₆ by ZmITPK1 using [³³P-γ]ATP; the addition of HsDIPP1 hydrolyzed >98% of the 5-InsP₇ that was formed (Figure 1D,E,F).

Figure 1. InsP₆ kinase activity of recombinant ZmITPK1.

A, Graphical representation of the coupled enzyme assay showing chair conformations of InsP₆ and 5-InsP₇. When non-radiolabeled ATP was present, free Pi was recorded with a malachite green colorimetric assay. B, Michaelis-Menten profile using the coupled assay with 25 μM InsP₆ and the indicated concentrations of ATP. K_M values were computed by SigmaPlot. Data are means and standard errors from 3 independent experiments. C, Michaelis-Menten plot obtained using the coupled assay with 1 mM ATP and the indicated concentrations of InsP₆. Data are means and standard errors from 6 independent experiments. D, E, F show HPLC analyses of the kinase activity of ZmITPK1 towards 1 mM InsP₆ at either zero time (panel D) or after 120 min either without (panel E) or with the addition of HsDIPP1 (panel F). Representative traces are shown from 3 independent experiments.

The coupled assay yielded a K_M value for ATP of 364 μM (Figure 1B), compared to the value of 211 μM similar obtained upon HPLC analysis of [³H]-InsP₆ phosphorylation (Figure S1D). These values for ZmITPK1 are below those K_M values for ATP that have previously been determined for AtITPK1 (523 μM (15); 1200 μM; (24)). Assuming cytoplasmic ATP levels of around 1 mM, we conclude that the ITPK1 ortholog from Zea mays is less responsive to fluctuations in cellular ATP levels than is AtITPK1 (15, 24).

Previous studies with AtITPK1 have examined its reactivity towards the full range of six InsP₅ isomers, and Ins(1,2,3,4,5)P₅ was found to be the only significant substrate (15, 24). We confirmed those earlier findings using our enzyme-coupled assay by incubating ZmITPK1 with 25 μM of each of these InsP₅ isomers; the following are the reaction rates (pmol / nmol enzyme / min) from 4 independent experiments (means ± standard errors): Ins(2,3,4,5,6)P₅, 5.1 ± 1; Ins(1,3,4,5,6)P₅, 9.1 ± 0.8; Ins(1,2,4,5,6)P₅, 6.1 ± 0.7; Ins(1,2,3,5,6)P₅, 2.2 ± 1; Ins(1,2,3,4,6)P₅, 3.7 ± 0.6; Ins(1,2,3,4,5)P₅, 134 ± 5.9. A Michaelis Menten plot of Ins(1,2,3,4,5)P₅ phosphorylation yielded a K_M value of 10 μM and an approximately 4-fold lower V_max compared to that for InsP₆ phosphorylation (Figure 1C, 2C). In separate assays that contained Ins(1,2,3,4,5)P₅ and [³³P-γ]-ATP, which were analyzed by HPLC, we found that a [³³P]-radiolabeled reaction product eluted 1 min prior to 5-InsP₇ (Figure 1E;2A). This product of Ins(1,2,3,4,5)P₅ phosphorylation was sensitive to HsDIPP1-catalyzed dephosphorylation (Figure 2A,B), consistent with it being a PP-InsP₄.

Figure 2. Kinase activity of ZmITPK1 towards Ins(1,2,3,4,5)P₅.

A, B, show HPLC analyses of the kinase activity of ZmITPK1 towards 1 mM Ins(1,2,3,4,5)P₅ after 120 min either without (panel A) or with the addition of HsDIPP1 (panel B). Representative traces are shown from 3 independent experiments. C, Michaelis-Menten plot obtained using the coupled assay with 1 mM ATP and the indicated concentrations of Ins(1,2,3,4,5)P₅. Data are means and standard errors from 5 independent experiments. The K_M value was computed by SigmaPlot. Panels D and E show the identities of the products of ZmITPK1-dependent phosphorylation of Ins(1,2,3,4,5)P₅ and InsP₆ respectively, from their electron densities after being soaked into crystals of HsDIPP1. The 2Fo-Fc electron density maps (blue mesh) are contoured at 1.5 σ. Phosphate groups are numbered according to standard nomenclature for myo-inositol derivatives.

The isomeric nature of the Ins(1,2,3,4,5)P₅ reaction product has not previously been determined (24). We soaked the product into crystals of HsDIPP1; X-ray diffraction analysis revealed a single PP-InsP₄, which was identified as 3-PP-Ins(1,2,4,5)P₄ (Figure 2D). In control experiments we soaked the product of InsP₆ phosphorylation into HsDIPP1 and confirmed the reaction product to be 5-InsP₇ (Figure 2E). Our data provide a new example of how the axis of symmetry of the myo-inositol ring can license members of the ITPK family to bind two distinct inositol phosphates in alternate yet productive orientations (34, 35), such that in this case either the 5- or 3-phosphate groups can be further phosphorylated

The 3-PP-Ins(1,2,4,5)P₄ described above was previously not known to occur in nature; therefore, it is different from the 5-PP-Ins(1,3,4,6)P₄ isomer identified as an IP6K product of Ins(1,3,4,5,6)P₅ phosphorylation in mammalian cells (36, 37). Interestingly, a recent study of PP-InsP turnover in Arabidopsis described root-specific, ITPK1-dependent synthesis of an unknown PP-InsP₄ isomer that is regulated by Pi availability (15). The possibility that this apparently biologically active PP-InsP₄ isomer might be 3-PP-Ins(1,2,4,5)P₄ is an opening for synthetic chemists to prepare this material for future biological studies.

Overall structure of ZmITPK1

There are no previous descriptions of the structure of any plant ITPK. We solved the crystal structure of full-length ZmITPK1, using selenium single wavelength anomalous dispersion phasing, and among a number of crystal complexes, the best resolution we obtained was 2.25 Å (Supplementary Table 1). Each asymmetric unit comprises one molecule of protein.

We were unable to obtain crystals of wild-type ZmITPK1 containing InsP₆ substrate, but in the process of studying the significance of a “hinge mutant” (H192A; see below) we established that InsP₆ could be soaked into crystals of ZmITPK1^H192A (Supplementary Table 1; Figure 3A). It is notable that the His192Ala mutation itself has no significant impact upon the overall conformation of the apoenzyme but the additional presence of InsP₆ stabilizes residues 216–230 (Figure S2A,B,C,D,E), thereby revealing a short, four-residue β-strand (β11) that is not visible in either the wild-type or H192A apoenzymes (Figure 3A, S2E, S3). Nevertheless, four other short sections of ZmITPK1 remain recalcitrant to our structural analysis: N-terminal residues 1–15, C-terminal residues 326–342, residues 231–256 between β11 and α8, plus Ala119/Asp120 (Figure S3). We were unable to obtain a crystal complex that contained nucleotide.

Figure 3. Overall structure of ZmITPK1^H192A.

A, Ribbon plot of the crystal structure of the ZmITPK1^H192A/InsP₆ crystal complex (PDB code 7TN8). The terminology for the three separate lobes follows that for EhITPK1 (26). Here, the N-terminal lobe is colored gold, the central lobe is colored blue, and the C-terminal lobe is colored green. The hinge (i.e., the β8-β9 loop; see text for details) is colored magenta. The individual α-helices and β-strands are numbered. A chloride ion is shown as a cyan sphere. The InsP₆ ligand is depicted as a stick model (green for carbon, orange for phosphorus, red for oxygen); the inset shows the 2Fo-Fc electron density map contoured at 1.5 σ. (blue mesh). B, A superimposition in ribbon format of the central lobes of ZmITPK1^H192A (blue) and HsITPK1 (pink; PDB code 2Q7D). The hinge regions are also shown (magenta for ZmITPK1^H192A 1, lime for HsITPK1). Several substantial displacements of corresponding structural elements are highlighted with broken lines (Zm vs Hs): C, Partial, structure-guided sequence alignment of wild-type ZmITPK1 and HsITPK1. The secondary structural elements are color coded to match panels A and B. Also highlighted are the particular residues used as reference points (at each Cα) to determine the displacements described in panel B, i.e., ZmITPK1 and HsITPK1: Ala136 in α6 vs Asp140 in α7 (yellow); Glu175 in α7 vs Glu176 in 3₁₀3 (green); Ala147 in 3₁₀2 vs Asn148 in α7 (dark purple); hinge residues Val190 vs Ile191 (cyan).

The overall crystal structures of ZmITPK1 and the ZmITPK1^H192A mutant (Figure 3A; S2A,B,C) exhibit an α-β-α fold distributed between three separate lobes, which are named N-terminal, central and C-terminal, by using the nomenclature previously applied to the structures of the human (25) and Entamoeba histolytica (26) orthologs (Figure S4,A,B). Superimpositions of the latter two structures upon the InsP₆-bound ZmITPK1^H192A structure (Figure S4C,D) indicates greater overall structural similarity of the plant enzyme with the human ortholog (RMSD = 1.9 Å; 1231 comparable atoms) rather than EhITPK1 (RMSD = 3.4 Å; 1141 comparable atoms). Even so, there are key structural elements of our ZmITPK1 crystal complexes that take on a distinctly unique conformation, particularly with regards to key aspects of the nucleotide-binding ATP-grasp region. The latter is classically formed from two sets of anti-parallel β-sheets, which in HsITPK1 and EhITPK1 are constructed from four β-strands in the C-terminal lobe and four in the central lobe (Figure S4A,B). In ZmITPK1, the equivalent pattern is four plus three. Thus, one β-sheet in the C-terminal lobe of HsITPK1 comprises β11, β10, β13 and β14 (25) (Figure S3, S4A); these are spatially equivalent to β10, β9, β12, and β13 in ZmITPK1 (Figure 3A, S3). The opposing β-sheet in HsITPK1 comprises β12, β8, β7 and β9 (25); the latter three strands correspond to β7, β6 and β8 in ZmITPK1 (Figure 3A, S3, S4A), but the plant enzyme does not have a structural element that corresponds spatially to β12 in HsITPK1 and EhITPK1 (Figure 3A, S3, S4A,B).

In a sequence alignment, β−11 in ZmITPK1 corresponds to β12 in HsITPK1 (Figure S3), but a structural comparison shows a relative spatial displacement of approximately 20 Å between the two β-strands (compare Figure 3A with Figure S4A). The β−11 strand in ZmITPK1 is proximal to a 26-residue sequence that is disordered in our crystal structures (see above), suggesting this entire structural element could be mobile. This is intriguing because the corresponding sequence in HsITPK1 is also flexible (its orientation varies between each asymmetric unit of the published crystal structure); its mobility may provide the protein with conformational control over entry and exit of nucleotide, as well as contributing to the productive orientation of the bound nucleotide (25).

A separate structural feature that might also contribute to catalytically-important conformational changes in ZmITPK1/ZmITPK1^H192A: a 6-residue β8-β9 loop that connects the central and C-terminal lobes (Figure 3A). Both HsITPK1 and EhITPK1 have structurally equivalent regions (Figure S4A,B), and in the Entamoeba ortholog, the Ile171 backbone NH in the equivalent loop interacts with the nucleotide’s adenine ring; this residue is conserved as Ile191 in HsITPK1 (26). This raises the possibility that this β8-β9 loop in ITPKs is orthologous to the flexible “hinge” that links two opposing lobes in other inositol phosphate kinases and protein kinases (38). In addition to directly contributing to nucleotide binding and conformational flexibility, hinge residues also dictate what is a productive orientation of the nucleotide, through their impact on the internal topology of the nucleotide pocket. Interestingly, a structural superimposition shows that hinge residue Val190 in ZmITPK1^H192A is displaced by 4.8 Å relative to Ile191 in HsITPK1 (Figure 3B) and Ile171 in EhITPK1, which interacts with nucleotide (26). This comparison suggests the nucleotide binding site of the plant enzyme has a specialized topology.

To pursue our proposal that the β8-β9 loop in ZmITPK1 may be functional, we noted that Phe189 and His192 in ZmITPK1 are both conserved in HsITPK1 (Figure 3C). Therefore, each of these residues was separately mutated to Ala, and both mutants of His₆MBP-ZmITPK1 hosted substantially reduced InsP₆ kinase activities (Figure 4A). In control experiments we found that the overall conformation of the ZmITPK1^F188A/H191A double mutant was not significantly different from that of wild-type enzyme (Figure S5B), the residues in the mutated hinge were fully traceable (Figure 4B), and the main structural impact of the mutations is to increase the volume of the nucleotide-binding pocket (Figure 4C). We therefore propose that the functions of Phe189 and His192 are to contribute to licensing a productive orientation of nucleotide in ZmITPK1.

Figure 4. Biochemical analysis of F189A and H192A mutations in the hinge region of His₆MBP-ZmITPK1.

A, InsP₆ kinase activities (25 μM InsP₆; 1 mM ATP) of wild-type His₆MBP-ZmITPK1 and the corresponding F189A and H192A mutants. Data are means and standard errors from 10 independent experiments. B, Stick model of the hinge region (¹⁸⁹AVNAGG¹⁹⁴) from ZmITPK1^F189A/H192A (PDB code 7TN3); carbon is blue, oxygen is red, nitrogen is blue. The 2Fo-Fc electron density map is shown in blue mesh (contoured at 1.0 σ). C, Superimposition of the hinge regions of wide-type ZmITPK1 (magenta) with ZmITPK1^F189A/H192A (cyan). F189, H192 and the Ala mutants are shown as stick models.

Other significant differences between the two nucleotide-binding central lobes of HsITPK1 and ZmITPK1 include a 10.4 Å displacement of their respective 3₁₀ helix and α7 helices, while the α7 helix in HsITPK1 is replaced in ZmITPK1 with two displaced structural elements: a smaller α-helix that is displaced by 15.1 Å plus an adjacent and more tightly wound 3₁₀ helix that is displaced by 14.8Å (Figure 3B,C).

Analysis of InsP₆ binding to ZmITPK1.

In the ZmITPK1^H192A crystal complex with InsP₆, in which the ligand lies in a shallow and open binding pocket, the 3-, 4-, 5- and 6-phosphates make direct polar contacts with the sidechains of several positive charged C-terminal lobe residues: Lys198, Arg211, Asn280, and Lys306 (Figure 5A, B, C). In addition, the 5-phosphate has hydrogen bonds with Tyr200 and the backbone α-amino group of Gly303 (Figure 5B,C). Furthermore, Lys29 and Lys70 from the N-terminal lobe participate in polar contacts with the 6-phosphate through two water molecules (Figure 5B,C). Lys164 and Lys167 contribute through Van Der Waal interactions with the 1- and 2-phosphates, respectively (Figure 5C). Interestingly, this ZmITPK1^H192A crystal complex also utilizes a chloride atom to stabilize the ligand-enzyme structure through a network of polar interactions with the sidechains of Lys29, Ser32, and also the backbone α-amino group of Tyr304 and Ala305 (Figure 5D); previous structural analyses of HsITPK1 and EhITPK1 did not detect any anion with a structural role. Finally, the Lys29 side chain makes polar contact with the backbone carboxyl of Pro302. The importance of each of these putative polar interactions was confirmed by mutagenesis, which reduced kinase activity 8–19 fold (Figure 5E). Three of these, Ser32, Asn280 and Lys306 are not conserved in HsITPK1 (Figure S3). Thus, we conclude that anion-mediated structural stabilization may be a unique property of the plant enzyme.

Figure 5. InsP₆ binding pocket of ZmITPK1^H192A.

A, An electrostatic surface plot of the occupied InsP₆ binding pocket. Electrostatic surface potentials were calculated using Pymol (blue is electropositive, red is electronegative). The ligand is depicted as a stick model (green for carbon, orange for phosphorus, red for oxygen); the 4- and 5-phosphates are labeled. B, Descriptions of amino acid residues (stick format) that have polar interactions (within 3.5 Å, black dashed lines) with the bound InsP₆. Two water molecules (W₁ and W₂) are depicted as purple spheres. C, A rendering of the ligand-protein interactions created by Ligplot+. D, Polar contact network formed around the bound chloride ion (cyan sphere; electron density is shown in blue mesh, contoured at 1.0 σ). E, InsP₆ kinase activities (25 μM InsP₆; 1 mM ATP) of wild-type His₆MBP-ZmITPK1 and mutants targeting key residues identified by structural analysis. Data are means and standard errors from at least 10 independent experiments.

Characterization of a structural element in plant ITPKs that imparts InsP₆ kinase catalytic activity.

We next further explored the significance of residues 219–247 which we have identified (see above) as a potentially mobile element at the entry point into the nucleotide binding pocket. Only two of these residues are fully conserved in HsITPK1 (Figure 6A), suggestive of functional diversity. Indeed, in our hands HsITPK1 is an extremely weak InsP₆ kinase: no activity was found using our enzyme-coupled assays (Figure 6B), although the more sensitive HPLC analysis of phosphorylation of 25 μM [³H]InsP₆ (see Methods) detected a minor kinase activity (0.87 ± 0.09 pmol / nmol enzyme / min (n=3)) i.e., approximately 55-fold lower than its Ins(1,3,4,5,6)P₅ phosphatase activity (Figure 6B). In view of this substantial difference in catalytic preferences, and the lack of conservation in HsITPK1 of ZmITPK1 residues 219–247, we hypothesized this sequence in the plant enzyme might contribute to its distinctive InsP₆ kinase activity. To test this idea, we replaced these residues with a Gly-Ser-Gly-Ser-Gly linker. This construct (ZmITPK1^218GSGSG248) exhibited >90% less InsP₆ kinase activity than wild-type enzyme (Figure 6A,B).

Figure 6. Molecular engineering of the catalytic specificity domains in ITPK1/2.

A, Multiple sequence alignments of HsITPK1 (pink), His₆MBP-ZmITPK1 (blue), His₆MBP-ZmITPK1^218GSGSG248 (blue), His₆MBP-GmITPK1 (dark green) and His₆MBP-GmITPK2 (orange), centered upon selected residues from within the regions of the catalytic specificity elements (CSE) selected for molecular engineering: brown for ZmITPK1, light green for GmITPK1 and yellow for GmITPK2. Red font highlights residues in ZmITPK1 that are identical to either GmITPK1 or GmITPK2 within the regions selected for a domain swap experiment. Gray font is used for the GSGSG linker in ZmITPK1^218GSGSG248, to emphasize it is not part of the wild-type sequence. B, InsP₆ kinase activity (upper panel; 1 mM ATP) and ADP-dependent Ins(1,3,4,5,6)P₅ phosphatase activity (lower panel; 1 mM ADP) of HsITPK1, ZmITPK1, and ZmITPK1^218GSGSG248 (i.e., ΔCSE). C, InsP₆ kinase activity (upper panel) and ATP-dependent Ins(1,3,4,5,6)P₅ phosphatase activity (lower panel) of (counting bars left-to-right) wild type His₆MBP-GmITPK1 (color-coded K1/CSE-1/K1), His₆MBP-GmITPK1^ITPK2-CSE (color-coded K1/CSE-2/K1), wild-type His₆MBP-GmITPK2 (color-coded K2/CSE-2/K2) and His₆MBP-GmITPK2^ITPK1-CSE (color-coded K2/CSE-1/K2). All graphs show means and standard errors from at least 4 independent experiments.

Could this deletion have unexpected conformational impact upon the protein structure? To investigate this possibility, we derived a crystal structure for the version of ZmITPK1 that lacks the catalytic specificity element, provided we also deleted the non-traceable residues 1–17 and 329–342 (PDB code 7TN7; Supplementary Table 1). This truncation of the β10-α8 loop had minimal impact upon the overall conformation of the rest of the protein (RMSD 0.258 Å compared to wild-type ZmITPK1; 1762 atoms; Figure S5A). We conclude that residues 219–247 in ZmITPK1, which includes the short β11 strand, are a catalytic requirement for robust InsP₆ kinase activity.

The ITPK1 family of enzymes in plants and animals has also been studied for its reversible, ATP-dependent Ins(3,4,5,6)P₄ 1-kinase / ADP-dependent Ins(1,3,4,5,6)P₅ 1-phosphatase activities (4, 7, 10, 39). These reactions are key not only to regulating the cell signaling role of Ins(3,4,5,6)P₄ (13), but also are a central aspect of the metabolic contributions that ITPKs make towards Ins(1,3,4,5,6)P₅ and InsP₆ synthesis (4, 7). Remarkably, in assays that contained 1 mM ADP, we found that the Ins(1,3,4,5,6)P₅ 1-phosphatase activity of ZmITPK1^218GSGSG248 was little different from that of wild-type ZmITPK1 (Figure 6B). We therefore name the region of ZmITPK1 that contains residues 219–247 as a catalytic specificity element (CSE) because it selectively sanctions InsP₆ kinase signaling activity.

Higher plant genomes encode multiple ITPK genes; in Arabidopsis and rice, the degree of InsP₆ kinase activity varies considerably between the various corresponding proteins (15). Thus, our next goal was to compare the functionality of the catalytic specificity elements in some alternate ITPKs. Problems with protein expression prevented us from studying other ZmITPKs; ITPKs from soybean (10) proved to be more amenable experimental models. Furthermore, a multiple sequence alignment (Figure 6A) identified a candidate catalytic specificity element in GmITPK1 that is 9 residues longer than that in GmITPK2. Moreover, this particular sequence in GmITPK1 shares 13 identical residues with ZmITPK1, whereas only 7 residues are found in the equivalent sequence of GmITPK2. We therefore hypothesized that GmITPK1 would harbor greater InsP₆ kinase activity than GmITPK2.

Indeed, in assays that contained 1 mM ATP, we found that recombinant His₆MBP-GmITPK1 is an active InsP₆ kinase (Figure 6C, S6A). Furthermore, GmITPK2 exhibited negligible InsP₆ kinase activity (Figure 6C; S6B). We next compared the Ins(1,3,4,5,6)P₅ phosphatase activities of these two enzymes; GmITPK2 exhibited the higher activity (Figure 6C). Therefore, we next exchanged the putative catalytic specificity elements between the two soybean proteins. The GmITPK2^ITPK1-CSE hybrid then expressed a substantial gain-of-function InsP₆ kinase activity (Figure 6C, S6C); the Ins(1,3,4,5,6)P₅ phosphatase activity supported by 1 mM ADP was reduced about 40%, although this is a less dramatic effect than the many fold increase in InsP₆ kinase activity (Figure 6C). On the contrary, the GmITPK1^ITPK2-CSE hybrid lost much of its InsP₆ kinase activity without impact upon its Ins(1,3,4,5,6)P₅ phosphatase activity (Figure 6C, S6D). These data reinforce the biological significance of the catalytic specificity element through its licensing of InsP₆ kinase activity.

Discussion.

Among the most important points to emerge from study is the identification of a 29-residue catalytic specificity element in plant ITPKs that, through isoform-dependent variations in its primary sequence, can reciprocally specify the relative rates of ATP-dependent InsP₆ kinase versus ADP-dependent Ins(1,3,4,5,6)P₅ phosphatase activities of this versatile enzyme family (Figure 3,6). Thus, for the first time it is possible to attach biochemical-based significance to the differential patterns of expression of GmITPK1 and GmITPK2 (10, 40): tissue-dependent variability in PP-InsP production. There may also be a practical advantage to our identification of a catalytic specific element that is required for InsP₆ kinase activity in certain plant ITPKs: genomic engineering of this element offers a path to specifically target inositol pyrophosphate synthesis without impeding the additional role of ITPK1 in generating precursor material for InsP₆ synthesis.

Prior to our solving the structure of a plant ITPK, it had been unknown which features in ITPKs might impart isoform-specific catalytic specificity (e.g., (41)). What is especially notable about our characterization of the catalytic specificity element is that it lies in a region with considerable evolutionary diversity, and its residues have almost entirely escaped attention in previous mutagenic efforts to identify catalytic requirements (7). Currently, many of the residues within the catalytic specificity element are not traceable in our crystal structures. However, that information in itself is instructive, as it speaks to the likely flexibility of this region. The corresponding region of HsITPK1 is also flexible, and this may control nucleotide entry into the protein, and also contribute to establishing a productive orientation of the bound nucleotide (25). These considerations raise the possibility that the nucleotide binding pocket of our apo-enzyme has been captured in a relatively open configuration; perhaps the catalytic specificity element closes around bound nucleotide, directing its appropriate orientation for optimizing InsP₆ kinase activity.

Our previous studies (42, 43) have shown that, compared to the inositol phosphate kinases that phosphorylate hydroxyl groups, the enzymatic addition of a β-phosphate to an inositol phosphate may be facilitated by the nucleotide’s β- and γ-phosphates being presented much closer to perpendicular to the inositol ring, to minimize steric and electrostatic clashing with non-reacting phosphates. Therefore, it can be anticipated that during ADP-dependent removal of a monophosphate (as in Ins(1,3,4,5,6)P₅ dephosphorylation), the relative orientations of the nucleotide and substrate will be different from those of ATP and substrate during InsP₆ phosphorylation. This may help explain why the ratios of activities of Ins(1,3,4,5,6)P₅ phosphatase : InsP₆ kinase differs between GmITPK isoforms (approximately 9:1 for GmITPK1 and 0.02:1 for GmITPK2; Figure 6C). Similarly, previous work with all four GmITPK isoforms demonstrated up to 22-fold differences in the ratios of activities of Ins(1,3,4,5,6)P₅ phosphatase : Ins(1,3,4)P₃ kinase (10).

Our current structural data suggest there are some spatial differences between the nucleotide-binding hinge regions of the plant and human enzymes (Figure 3B). These adaptations could conceivably also contribute to modifying the presentation of nucleotide to substrate in ZmITPK1 compared to the human ortholog, which exhibits only weak InsP₆ kinase activity. Such conclusions serve a cautionary note to previous work (24) that has attempted to model plant enzyme/ATP complexes, based on the relative orientations of HsITPK1 protein and nucleotide.

To what extent is the InsP₆ kinase activity of plant ITPKs shared by the mammalian orthologs? Laha et al (16) have reported that HsITPK1 harbors measurable InsP₆ kinase activity. However, they did not state how much enzyme was added to their reactions, nor do they calculate specific activity data and compare that with the plant enzymes. To our knowledge, ours is the first study to produce such a comparison; the low level of InsP₆ kinase activity that we recorded for the human enzyme is approximately 50-fold less than its Ins(1,3,4,5,6)P₅ phosphatase activity (Figure 6B) that generates the Ins(3,4,5,6)P₄ cellular signal. Such in vitro data indicate that the synthesis of 5-InsP₇ by HsITPK1 is not sufficiently adequate to be the major signaling activity of this enzyme. Moreover, compared to HsITPK1, both ZmITPK1 and GmITPK1 2–3 orders of magnitude higher rates of InsP₆ kinase, respectively, indicating this is a specialized activity of the plant enzymes.

Previous work has shown that some AtITPK1 has 523–1200 μM affinity for ATP, making it likely that its InsP₆ kinase activity is responsive to physiologically-relevant fluctuations in cellular [ATP], which are anticipated to be in the low mM range (15, 24). Our data suggest this scenario is less likely to apply to ZmITPK1 (K_M for ATP = 211–364 μM; Figure 1B; S1D), raising the possibility of further functional differences among the various plant ITPK family members.

We have augmented previous work (15) that identified a root-specific, ITPK1-dependent PP-InsP₄ isomer, the synthesis of which is regulated by Pi availability; the latter study did not determine the configuration of this isomer, other than it was shown to be distinct from the 5-PP-Ins(1,3,4,6)P₄ isomer. In the current study we have shown that ZmITPK1 phosphorylates Ins(1,2,3,4,5)P₅ to a previously unknown PP-InsP₄ isomer that we determined is 3-PP-Ins(1,2,4,5)P₄ (Figure 2D). Thus, we conclude that Ins(1,2,3,4,5)P₅ can be accommodated into the catalytic pocket in a different orientation to that of InsP₆, such that the former can be phosphorylated at the 3-position. Such positional flexibility is typical for inositol phosphate kinases (44). Future work can now focus on the chemical synthesis of the 3-PP-Ins(1,2,4,5)P₄, to examine the proposal in (15) that this molecule should be investigated as a potential SPX domain ligand and/or a potential regulator of organ-specific Pi-dependent accumulation. Such properties would add considerably to the repertoire of cell-signaling functions for plant ITPKs.

Finally, we wish to note how we anticipate our study might be of particular significance to the agricultural field: our simple, enzyme-coupled microplate-based assay of PP-IP synthesis should promote more general access to biochemical analysis of ITPKs, by supplanting current reliance on expensive and/or laborious chromatographic analyses. In this way, there can be improved understanding of the contributions made by various ITPK orthologs in different plant species, not just to PP-InsP synthesis, but also their multiplexing a metabolic pathway to the synthesis of InsP₆; this is work that has direct relevance to immunity, defense again environmental stress, intracellular phosphate homeostasis, and nucleotide balance (4, 7, 15, 23, 24, 45). Together with the structure/function data that we have provided, new synthetic biology approaches to improving crop yield and food security may also now be envisaged.

DATA AVAILABILITY STATEMENT

Coordinates and structure factors have been deposited in the Protein Data Bank. All other data in this study can be provided by the corresponding author if needed.