Substrate Binding in Protein-tyrosine Phosphatase-like Inositol Polyphosphatases

Background: PTP-like inositol polyphosphatases hydrolyze myo-inositol hexakisphosphate via an ordered pathway.

Results: X-ray structures in complex with substrates and fluorescence binding reveal novel features of the kinetic mechanism.

Conclusion: PTP-like inositol polyphosphatases have a two-step binding mechanism that facilitates specificity and catalysis.

Significance: Structural and binding studies are essential for understanding complex kinetic mechanisms.

Abstract

Protein-tyrosine phosphatase-like inositol polyphosphatases are microbial enzymes that catalyze the stepwise removal of one or more phosphates from highly phosphorylated myo-inositols via a relatively ordered pathway. To understand the substrate specificity and kinetic mechanism of these enzymes we have determined high resolution, single crystal, x-ray crystallographic structures of inactive Selenomonas ruminantium PhyA in complex with myo-inositol hexa- and pentakisphosphate. These structures provide the first glimpse of a myo-inositol polyphosphatase-ligand complex consistent with its known specificity and reveal novel features of the kinetic mechanism. To complement the structural studies, fluorescent binding assays have been developed and demonstrate that the K_d for this enzyme is several orders of magnitude lower than the K_m. Together with rapid kinetics data, these results suggest that the protein tyrosine phosphatase-like inositol polyphosphatases have a two-step, substrate-binding mechanism that facilitates catalysis.

Introduction

myo-Inositol polyphosphates (IPPs)² are ubiquitous in nature and are involved in a number of important cellular signaling events (1). myo-Inositol 1,2,3,4,5,6-hexakisphosphate (phytic acid, InsP₆) is the most abundant cellular IPP and plays a central role in numerous essential cellular processes, including DNA repair, RNA processing, mRNA export, plant development, apoptosis, and bacterial pathogenicity (2–7). Inositol polyphosphatases (IPPases) that degrade IPPs are found throughout nature and have been identified in prokaryotes, protists, fungi, animals, and plants (8–10). IPPases that are able to hydrolyze InsP₆ are generically referred to as phytases. To date, four classes of IPPases have been described that are structurally and mechanistically diverse, including histidine acid phosphatases, β-propeller phytases, purple acid phosphatases, and protein-tyrosine phosphatase-like phytases (PTPLPs) (11–13). Each of the IPPase classes catalyze the stepwise removal of one or more phosphates from IPPs. The products generated during the breakdown of a specific IPP are released from the enzyme and serve as substrate for subsequent cycles of hydrolysis. Although the biological function of some of these enzymes is unclear, they have been found in a wide range of bacteria, including plant and human pathogens (14, 15).

PTPLPs are particularly interesting because the stepwise removal of phosphate from IPP substrates occurs in a relatively specific order. They have a protein-tyrosine phosphatase (PTP) fold, including the P-loop or PTP active-site signature sequence (CX₅R(S/T)) and utilize a classical PTP reaction mechanism (12–13). Interestingly, these enzymes display no catalytic activity against classical PTP substrates due to several unique structural features that confer specificity for IPPs (12–13, 15–17). The structural basis of IPP binding to a representative PTPLP (PhyAsr; Selenomonas ruminantium PhyA) has been investigated by x-ray crystallography using the competitive inhibitor, myo-inositol hexasulfate (MIHS) (13). As expected, the inhibitor binds in a deep electropositive cleft where it makes extensive contacts with both the active-site signature sequence (P-loop) and structural determinants that are not present in PTPs. Notably, MIHS binds with the C5 sulfate adjacent to the nucleophilic thiol in contrast to the known specificity of the enzyme for the C3 phosphate of InsP₆, suggesting that the binding modes for substrate and inhibitor are distinct.

In this work, we describe high resolution x-ray crystallographic structures of inactive PhyAsr mutants in complex with InsP₆ and Ins(1,2,3,5,6)P₅. The PhyAsrC252S·InsP₆ complex structure is the first IPPase·InsP₆ structure that is consistent with the known substrate specificity of the enzyme. The PhyAsrC252A·Ins(1,2,3,5,6)P₅ complex structure and previous inhibitor structures (13) suggest that IPP substrates have multiple, overlapping binding sites within the binding pocket of the enzyme. To complement the structural studies, fluorescent binding assays were developed and demonstrate that the K_d for InsP₆ is several orders of magnitude lower than the apparent K_m determined from steady-state kinetic studies. These data suggest that the earliest steps of the PhyAsr reaction mechanism involve an initial substrate binding event followed by a slower catalytic step that may require substrate reorientation. This interpretation is supported by preliminary results from a rapid kinetic study of substrate binding which suggest a multiphasic binding involving, at minimum, a rapid initial phase and another slower phase.

EXPERIMENTAL PROCEDURES

Cloning and Mutagenesis

The pET28b expression construct (13) containing the full-length S. ruminantium ORF (minus the putative signal peptide) was used as the template for all cloning and mutagenesis procedures. The signal peptide sequence was determined using SignalP 3.0 (19). PhyAsr numbering begins with 1 at the N terminus of the protein sequence found in GenBank (AAQ13669) including the predicted 27-residue signal peptide. All mutant proteins (C252S, C252A, H188C, and H188C/C252S) were prepared by site-directed mutagenesis using counter-PCR amplification of the expression plasmid as described previously (20). To verify the identity of the construct and the presence of the desired mutations, all PCR products were sequenced at the University of Calgary Core DNA and Protein services facilities. Sequence data were analyzed using the programs SEQUENCHER 4.0 (Gene Codes Corp.) and MacDNAsis 3.2 (Hitachi Software Engineering Co.). All of the primers used in this study were purchased from Integrated DNA Technologies.

Protein Production and Purification

Protein expression was carried out in Escherichia coli BL21(DE3) cells (Novagen) for 18 h at 310 K. Cells were grown to an A_{600 nm} of 0.6–0.8, and protein expression was induced by adding isopropyl-β-d-thiogalactopyranoside to a final concentration of 1 mm. Induced cells were harvested and resuspended in lysis buffer (20 mm KH₂PO₄ (pH 7.0), 300 mm NaCl, 1 mm β-mercaptoethanol), 15 mm imidazole (pH 8.0)). Cells were lysed by sonication, and cell debris was removed by centrifugation at 20 000 × g. PhyAsr was purified to homogeneity by metal chelating affinity chromatography (nickel-nitrilotriacetic acid-agarose; Qiagen). Bound protein was washed with lysis buffer and eluted with lysis buffer containing 400 mm imidazole (pH 8.0). PhyAsr was further purified by cation exchange (Macro-Prep High S) and size exclusion chromatography (GE Healthcare, S200). The homogeneity of the purified protein was confirmed by SDS-PAGE (21) and Coomassie Brilliant Blue R-250 staining. Protein (containing an N-terminal His tag) was either used immediately or dialyzed into 20 mm ammonium bicarbonate (pH 8.0) and lyophilized for long term storage.

Crystallization and Ligand Soaking

Crystallization experiments were carried out at room temperature using sitting-drop vapor diffusion with a drop ratio of 2 μl of 20 mg/ml protein solution (in water) to 2 μl of reservoir solution. Crystals were grown in 8–10% PEG 8000, 200–300 mm NaCl, 50 mm sodium acetate (pH 4.8). Ligand solutions were prepared by dissolving InsP₆ (Sigma-Aldrich) in mother liquor to a final concentration of 2 mm. Ligand solutions were added to drops containing crystals and incubated for approximately 1 h prior to flash freezing in a solution containing the crystallization reagents, ligand, and 25% glycerol.

Data Collection, Image Processing, and Structure Refinement

Diffraction data (λ = 1.1159 Å) for both complexes were collected at 100 K on beamline 8.3.1 at the Advanced Light Source. Diffraction images were processed with MOSFLM and scaled with SCALA within the CCP4 program suite. Phases derived from the isomorphous D223N structure (Protein Data Bank (PDB) code 2B4P) (13) were used to solve the structures of the PhyAsr ligand complexes. CNS 1.2 (22) was used to refine the models iteratively using simulated annealing and positional and B-factor refinement followed by manual model building in XFIT (23). The N-terminal His tag and residues 27–32 lack corresponding electron density and were not included in the final model of either structure. Unless indicated otherwise, figures were prepared with PyMOL (24). Initial topology and parameter files for the ligands were generated using the structure of InsP₆ bound to the E. coli phytase (PDB code 1DKQ) as a model. For the refinement of the Ins(1,2,3,5,6)P₅ complex, these files were edited to reflect the structures of the ligands. PROCHECK was used throughout refinement to assess the stereochemistry of the model (25). Statistics for the data collection and refinement are presented in Table 1.

TABLE 1.

Data collection and refinement statistics for the PhyAsrC252S·InsP₆ and PhyAsrC252A·Ins(1,2,3,5,6)P₅ complex structures

Values in parentheses are for the highest resolution shell.

Parameters	PhyAsrC252S·InsP6	PhyAsrC252A·InsP5
Data collection
Space group	P21	P21
a, b, c (Å)	46.1, 136.9, 79.9	46.0, 137.3, 80.0
β (°)	102.9	103.0
Wavelength (Å)	1.1159	1.1159
Resolution (Å)	51-1.60	37.5-1.55
Observed reflections	560,803	456,682
Unique reflections	120,565	126,093
Completeness (%)	92.9 (77.0)	89.9 (56.8)
Redundancy	4.6 (4.2)	3.6 (2.8)
Rmerge	0.083 (0.76)	0.074 (0.38)
I/σI	10.7 (1.5)	15.5 (2.8)
Refinement statistics
Resolution (Å)	50-1.60	50-1.55
No. reflections work set	113,134	119,687
No. reflections test set	6,015	6,346
Rwork/Rfree (%)	17.0/18.5	16.4/16.9
Asymmetric unit	Dimer	Dimer
Protein atoms	5,103	5,304
Solvent atoms	932	886
Ligand atoms	80	72
Small molecule atomsa	47	47
Wilson B (Å2)	17.9	15.1
Average B protein (Å2)	16.0	14.6
Average B solvent (Å2)	30.6	29.3
Average B ligand (Å2)	39.1	23.2
R.m.s.d.b bonds (Å)	0.013	0.011
R.m.s.d. angle (°)	1.66	1.65
Ramachandran distribution
Most favored (%)	93.1	92.7
Additionally allowed (%)	6.6	6.9
Generously allowed (%)	0.4	0.4

^a Each structure contains 1 chloride, 4 acetate, and 5 glycerol molecules.

^b R.m.s.d., root mean square deviation.

Fluorescent Labeling

150 nmol of PhyAsrH188C or PhyAsrH188C/C252S was batch-bound to nickel-nitrilotriacetic acid resin (Qiagen) in binding buffer (20 mm Tris-HCl (pH 8.0), 100 mm NaCl, 10 mm β-mercaptoethanol), and buffer was exchanged with 4 column volumes of labeling buffer (20 mm Tris-HCl (pH 8.0), 100 mm NaCl, 10% glycerol). A 20-fold molar excess of 5-iodoacetamidofluorescein (5-IAF) was added to 2 ml of labeling buffer and incubated with the bound protein for 48 h at 310 K with gentle agitation. The bound protein was washed with labeling buffer until the flow-through was colorless, eluted with elution buffer (20 mm Tris-Cl (pH 8.0), 100 mm NaCl, 2.5 m imidazole (pH 8.0)), and subsequently buffer-exchanged into storage buffer (20 mm sodium acetate (pH 5.0), 300 mm NaCl, 20% glycerol). Labeling efficiencies were ∼70%.

Binding Studies

Dissociation constants (K_d) were determined by titrating a fixed amount of 5-IAF labeled PhyAsrH188C/C252S with InsP₆. Fluorescence measurements were made using a Cary Eclipse fluorometer (Varian). The 5-IAF fluorophore was excited at 480 nm, and the resulting emission between 490 nm and 550 nm was measured in 0.5-nm increments at a rate of 300 nm/min after a 2-min equilibration. Runs were performed in quadruplicate at 293 K after a 2-min equilibration period following injection of titrant. Blank titrations were performed by titrating 5-IAF-labeled PhyAsrH188C/C252S with buffer. Fluorescent data points were obtained by averaging data from fluorescent emissions across 20 nm, centered on the fluorescence emission maxima (517 nm). The change in fluorescence was obtained by subtracting the blank titrations from the InsP₆ titrations. The resulting difference data were evaluated using the following equation,

where B_max is amplitude, [P]_t is the total concentration of PhyAsrH188C/C252S, [L]_t is the total concentration of InsP₆/MIHS, and K_d is the dissociation constant (26).

Phosphate Inhibition

PhyAsr (1 μm) was incubated with 2–50 mm InsP₆ in the presence and absence of 1.0 and 10.0 mm inorganic phosphate for 5 min, and the products were separated using a CarboPax-100 column (Dionex) following standard protocols (27).

RESULTS

Structure of PhyAsrC252S·InsP₆ Complex

To examine the interactions between PhyAsr and its substrate, InsP₆, we have determined the high resolution structure (1.6 Å) of the inactive C252S mutant (active-site nucleophile) in complex with InsP₆ (PDB code 3MMJ). Analysis of 2F_o − F_c omit and difference electron density maps clearly identify density for InsP₆ bound in the active site. The InsP₆ molecule displays excellent electron density (Fig. 1A); however, refinement of the ligand with full occupancy results in negative difference density for all of the atoms in the ligand with the exception of the C3 phosphate. To account for this density, we modeled PhyAsrC252S as binding either InsP₆ or inorganic phosphate (an impurity in commercial InsP₆) with occupancies of 0.75 and 0.25, respectively. Subsequent hydrolysis assays in the presence and absence of inorganic phosphate confirm product inhibition occurs (supplemental Fig. S1) under conditions similar to those used for crystallization and likely accounts for the observed phosphate binding.

FIGURE 1.

InsP₆ ring conformation and electron density. A, 2F_o − F_c electron density contoured at 1.5 σ for the InsP₆ bound in the active site. B, superposition of the PhyAsrC252S·InsP₆ complex (gray) and the PhyAsr·MIHS complex (red; PDB code 1U26). InsP₆ binds in a chair conformation with the C2 phosphate in an axial position and the remaining phosphates in equatorial positions. The C3 phosphate is bound in the active site consistent with known specificity of this enzyme. MIHS binds with the C2 phosphate in an equatorial position and with the C5 sulfate in the active site. C, divergent stereo view of the interactions between InsP₆ and PhyAsrC252S. Residues involved in InsP₆ binding are shown as sticks, and dashed lines represent specific hydrogen bonding or electrostatic interactions. The P-loop (yellow) accounts for all backbone contacts with InsP₆, whereas the GA loop (cyan) and six additional residues (green) make direct contacts through their side chains.

In the PhyAsrC252S·InsP₆ complex structure, the ligand conformation and orientation are significantly different from previous complex structures with the competitive inhibitor MIHS (13). InsP₆ binds to PhyAsrC252S in a chair conformation with the C1 and C3–C6 phosphates in equatorial positions, whereas the MIHS inhibitor was modeled in a chair conformation with only C2 adopting an equatorial position (Fig. 1B). As a consequence of this conformational difference, the InsP₆ leaving group oxygen of the scissile phosphate is 0.60 Å closer to the general acid (Asp-223) than in the inhibitor complex (12). This was previously predicted from substrate docking studies that suggested an InsP₆ conformation similar to that observed in our complex has the lowest conformational energy and the lowest binding energy and positions the leaving group oxygen for protonation by the general acid (13). Next, InsP₆ binds to PhyAsrC252S with the C3 phosphate in the active site (Fig. 1C) in agreement with the known PhyAsr degradation pathway that shows InsP₆ dephosphorylation is initiated at and highly specific for the C3 phosphate (13). Finally, we note that InsP₆ binding induces strain in the ligand with the inositol ring adopting a partially twisted chair conformation and increasing the C-O-P angle for the active-site phosphate by 27 degrees relative to the other phosphates. The twisted conformation of the ligand maximizes the agreement with the electron density for the ligand and relieves steric interactions between the general acid loop (Asp-223, His-224) and the C2 phosphate (Fig. 1C). Additionally, the increase in the C-O-P angle of the phosphate in the active site relieves steric clashes between the inositol ring and P-loop, specifically Ala-254. This conformation is stabilized by the extensive contacts made to both the C2 phosphate and the scissile phosphate.

InsP₆ Contacts with PhyAsrC252S

The binding pocket of PhyAsr is an electropositive cleft that is large enough to accommodate the highly negative InsP₆ molecule (Fig. 2A). The C3 phosphate is bound adjacent to C252S and positioned for nucleophilic attack. There are extensive contacts between the C2, C3, and C4 phosphates and the protein at the base of the binding pocket and relatively few contacts involving the solvent-exposed C1, C5, and C6 phosphates (Fig. 2). Main-chain contacts between PhyAsrC252S and InsP₆ are limited to the scissile C3 phosphate (five contacts) and the adjacent C4 phosphate (one contact). To facilitate the discussion of ligand binding to various IPPs, we will refer to the PhyAsr phosphate binding sites as P_s (scissile phosphate) P_a/P_a′ (adjacent to scissile phosphate), P_b/P_b′, and P_c (Fig. 2c). Residues that form hydrogen bonds and ion pairs with the substrate are listed in Table 2 and shown schematically in Fig. 2b. Ordered solvent molecules are observed coordinating the phosphates in the P_a, P_a′, and P_b sites. Interestingly, most of the residues involved in binding InsP₆ are also involved in binding MIHS (12) despite the significant conformational differences between ligands. Because there are no significant changes in the main-chain conformation of PhyAsr residues contacting the ligands in the two structures (0.2 Å root mean square deviation for 15 residues), it appears that there are several modes or ways in which the binding pocket can accommodate ligands.

FIGURE 2.

A, binding of InsP₆ to PhyAsrC252S InsP₆ binds deep in the highly basic binding pocket. The C5, C6, and C1 phosphates furthest from the scissile phosphoester bond are exposed to bulk solvent. The electrostatic surface potential was calculated using the APBS plugin in PyMOL (24). Positively and negatively charged regions are colored blue and red, respectively. B, schematic is two-dimensional representation of the contacts (dashed lines) between PhyAsrC252S and each of the InsP₆ phosphates. C, divergent stereo view of the interactions between InsP₆ and PhyAsrC252S highlights the different phosphoryl group binding sites. The binding sites are labeled P_s, P_a, P_a′, P_b, P_b′, and P_c to facilitate discussion of the different ligand binding modes identified in this work.

TABLE 2.

Contacts (<3.5 Å) in the PhyAsrC252S·InsP₆ and PhyAsrC252A·Ins(1,2,3,5,6)P₅ complex structures

InsP6			InsP5
Residue	Phosphate	Contact	Residue	Phosphate	Contact (Å)
		Å			Å
Arg-57	Pa′	2.73, 2.88, 2.48	Arg-57	Pa′	2.91, 3.06, 3.35
Arg-68	Pb′	3.12
Asp-153	Pa′	3.09
Lys-189	Pc/Pa′	3.47/3.37	Lys-189	Pc/Pa′	2.66/3.16
Asp-223	Pa′	3.10, 2.57, 2.74	Asp-223	Pa/Pa′	2.63/2.87
His-224	Pa	2.97	His-224	Pa	2.73
Ser-252	Ps	2.50
Glu-253	Ps	3.00	Glu-253	Ps	2.82
Ala-254	Ps	3.14	Ala-254	Ps	2.96
Gly-255	Ps	2.83	Gly-255	Ps	3.02
Val-256	Ps	2.67	Val-256	Ps	3.17, 2.97
Gly-257	Pa	3.24	Gly-257	Ps/Pa	3.11/3.42
Arg-258	Ps	2.89, 2.87, 2.79	Arg-258	Ps	2.78, 2.83, 2.94
			Thr-259	Ps	2.78
			Lys-305	Pb′/Pa	2.86/2.71
Tyr-309	Pb	3.26	Tyr-309	Pb	2.60
Lys-312	Pa	2.83	Lys-312	Pa	3.07

Two notable differences in the binding of InsP₆ and MIHS involve residues Arg-68 and Gly-257. In the absence of substrate and in the MIHS complex structure, the Arg-68 side chain is directed away from the binding pocket and into the solvent. In the InsP₆ complex structure, the Arg-68 side chain undergoes a large, induced-fit movement that brings it into contact with the C1 phosphate. This movement is a result of a large rotation about χ1 and χ2 that shifts the Nη1 and Nη2 atoms of the guanidinium group by to 8.3 and 9.8 Å, respectively. This movement results in the formation of a 3.1 Å contact with the C1 phosphate and may be important for productive substrate binding. Unlike Arg-68, Gly-257 does not change its conformation in the absence of substrate or in the presence of either ligand. Instead, differences in the ligand binding modes give rise to an additional hydrogen bond between the amine of Gly-257 and the bridging oxygen of the C4 phosphate (P_a site) in the InsP₆ complex structure. This hydrogen bond is absent in the inhibitor complex structure and may have a role fixing the orientation of InsP₆ relative to the nucleophile.

Structure of PhyAsrC252A Bound to Ins(1,2,3,5,6)P₅

In an attempt to obtain a structure of the PhyAsrC252A mutant in complex with InsP₆ we unexpectedly observed the binding of Ins(1,2,3,5,6)P₅ and inorganic phosphate (Fig. 3A, PDB code 3MOZ). This inositol polyphosphate is a contaminant in commercial InsP₆ purified from plant sources and has previously been observed in the structure of a pleckstrin homology domain soaked with InsP₆ (28). The smaller alanine side chain enables the inorganic phosphate to bind 0.8 Å deeper within the active site than the scissile phosphate of the InsP₆ complex. The binding of inorganic phosphate sterically prevents InsP₆ binding while allowing the binding of Ins(1,2,3,5,6)P₅. The inorganic phosphate in the active site makes a 2.6-Å hydrogen bond with the C4-hydroxyl of Ins(1,2,3,5,6)P₅. The Ins(1,2,3,5,6)P₅ ring adopts a chair conformation similar to the InsP₆ ring without the slight twist. However, the ring of Ins(1,2,3,5,6)P₅ is shifted away from the base of the active site by roughly 1 Å and rotated relative to InsP₆ (Fig. 3B).

FIGURE 3.

Binding of Ins(1,2,3,5,6)P₅ and inorganic phosphate to PhyAsrC252A. A, divergent stereo view of the interactions between Ins(1,2,3,5,6)P₅, inorganic phosphate, and PhyAsrC252A. Residues contacting the substrate are shown as sticks and colored as in Fig. 1C. Electrostatic interactions and H-bonds are shown as dashed lines. B, divergent stereo view of the least squares superposition of the PhyAsrC252S·InsP₆ (gray) and PhyAsrC252A·Ins(1,2,3,5,6)P₅ (red) including the inorganic phosphate bound within the active site. The InsP₅ is rotated and translated away from the P-loop to accommodate the inorganic phosphate. C, schematic, two-dimensional representation of the contacts (dashed lines) between PhyAsrC252A and each of the Ins(1,2,3,5,6)P₅ phosphates.

PhyAsrC252A contacts with Ins(1,2,3,5,6)P₅ are closely similar to those observed in both the InsP₆ and MIHS complex structures despite the differences in ligand orientation within the active site (Table 2). The C3, C5, C6, and C2 phosphates of Ins(1,2,3,5,6)P₅ contact residues that form the P_a′, P_a, P_b, and P_c sites, respectively, in the PhyAsrC252S·InsP₆ complex (Fig. 3, B and C). The solvent exposed C1 phosphate fails to make direct contacts with PhyAsrC252A whereas the anion makes 10 contacts at the base of the active site, involving main-chain amines of the P-loop and the side chains of Arg-258 and Thr-259.

Based on the shift in the inorganic phosphate position and the smaller size of the alanine side chain, we investigated the possibility that the inorganic phosphate was positioned similarly to the cysteinyl-phosphate intermediate in PTP1B (29). Superposition of the PhyAsrC252A·Ins(1,2,3,5,6)P₅ complex with the PTP1B phosphoenzyme intermediate (PDB code 1A5Y) reveals only small differences in the position of the phosphates that are likely due to the noncovalent nature of the bound inorganic phosphate in our structure (Fig. 4). The similarities suggest the hypothesis is reasonable and the observed contacts between the phosphate anion and PhyAsrC252A will be analogous to those present in a PhyAsr cysteinyl-phosphate intermediate.

FIGURE 4.

PhyAsrC252A·Ins(1,2,3,5,6)P₅ complex resembles the PTP1B phosphoenzyme intermediate. Least squares superposition of PhyAsrC252A mutant (red) bound to Ins(1,2,3,5,6)P₅ and inorganic phosphate and the phosphocysteine intermediate of PTP1B (cyan; PDB code 1A5Y) is shown. The general acid and phosphate binding loops are labeled GA loop and P-loop, respectively.

InsP₆ Binding Studies

To date, the reported binding affinities of phytases for InsP₆ or any other inositol phosphates have been based on the Michaelis-Menten kinetic parameter K_m. To examine the K_d of PhyAsr for InsP₆ we developed a fluorescent based assay to measure substrate binding. The binding curves obtained for InsP₆ are characteristic of a tight binding interaction and indicate that the K_d for InsP₆ binding is comparable with, or lower than, the concentration of protein (1 μm) used in the assay (Fig. 5A) (30). The sensitivity of our current fluorescent based assay precludes us from significantly decreasing the protein concentration. Although we cannot currently obtain an accurate K_d value, the curve is consistent with PhyAsrC252S having a submicromolar binding affinity for InsP₆. In contrast, K_m values for InsP₆ are dependent on ionic strength and vary from 1290 ± 240 μm (100 mm ionic strength) to 150 ± 10 μm (500 mm ionic strength) (31). At all ionic strengths, the K_d for InsP₆ determined from the fluorescence binding assay is 2 or more orders of magnitude less than the K_m. These finding suggest that PhyAsr does not utilize a classical Michaelis-Menten reaction mechanism as assumed in previous works.

FIGURE 5.

InsP₆ binding to PhyAsrH188C/C252S detected by changes in relative fluorescent intensity. The 5-IAF fluorophore was introduced at the periphery of the active site (H188C) near the P_c site. A, titration of PhyAsrH188C/C252S with InsP₆. The binding curve suggests a maximal dissociation constant in the low micromolar range. Preliminary rapid kinetics measurements suggest a two-phase binding over 10 s (B) and 100 s (C).

DISCUSSION

Substrate Conformation

This first structure of a PTPLP in complex with InsP₆ has allowed us to identify novel features not observed in the MIHS inhibitor complex. In particular, our PhyAsrC252S·InsP₆ structure is consistent with the known InsP₆ degradation pathway, positions the leaving group oxygen for protonation by Asp-223 (12), and the InsP₆ adopts the energetically preferred ring conformation present under physiological conditions (Fig. 1C). In addition, the binding of InsP₆ induces strain in the ligand similar to that observed in high resolution structures of several glycosidase and glycosyltransferases (32–34). As a consequence, the substrate adopts a conformation closer to the presumed transition state which minimizes unfavorable steric and electrostatic clashes with the protein while maximizing the stabilizing contacts with the ligand.

In the PhyAsrC252A·Ins(1,2,3,5,6)P₅ complex structure, the observed phosphate anion contacts resemble those of a cysteinyl-phosphate intermediate, whereas the ligand adopts the same energetically preferred ring conformation observed in the PhyAsrC252S·InsP₆ complex. However, the Ins(1,2,3,5,6)P₅ is shifted away from the active site. This accommodates the bound phosphate anion and relieves the steric and electrostatic interactions that induce strain in InsP₆ (Fig. 2). These observations are consistent with the PhyAsrC252S·InsP₆ complex structure representing interactions prior to catalysis and the PhyAsrC252A·Ins(1,2,3,5,6)P₅ structure representing a postcatalytic complex (Fig. 4). Notably, the bound phosphate anion cannot be released from the active site in the observed structure without prior dissociation of the Ins(1,2,3,5,6)P₅ ligand. This agrees with previous findings that clearly indicate that PhyAsr is nonprocessive and releases all products prior to subsequent catalytic cycles (12).

Phosphoryl Group Binding Sites

We have identified several conserved phosphoryl group binding sites (P_a, P_s, and P_a′) near the base of the active site, and several variable, solvent-accessible binding sites (P_b, P_c, and P_b′). The P_s site is the catalytic center and as such is directly involved in catalysis. Notably, this site is primarily composed of P-loop residues and accounts for virtually all the main-chain interactions with the Ins(1,2,3,5,6)P₅, InsP₆, and MIHS ligands. The P_a and P_a′ sites (Arg-57, Lys-189, Asp-223, His-224, and Lys-312) are located adjacent to the catalytic site and appear to function as anchor points that ensure the scissile phosphate is correctly oriented for nucleophilic attack. The functions of the variable sites are less clear, but they may be important for binding and discriminating between the lower phosphorylated IPPs that are generated during the degradation of InsP₆.

The catalytic properties of several PTPLPs have recently been described (12,16–18), and these enzymes can be separated into high activity and low activity groups. Interestingly, all of the binding residues identified in PhyAsr are conserved in all other high activity PTPLPs whereas the low activity PTPLPs have conservative and nonconservative mutations or deletions at several of these sites. The conservation of these binding residues also appears to play a role in defining substrate specificity because two of the enzymes with mutations to binding residue have significantly altered substrate specificity (16–17, 35).

In each complex structure, residues that form the conserved P_s, P_a, and P_a′ subsites (Arg-57, Lys-189, Asp-223, His-224, Lys-312, and the P-loop) account for the majority of the contacts to the ligand. In contrast, residues that compose the variable subsites (P_b, P_c, and P_b′) make relatively few contacts with solvent-exposed phosphoryl groups. Further, there are differences in both the number and nature of the contacts. With the exception of the remote standby site identified by Chu et al. (13) the same residues are involved in binding the distinctive Ins(1,2,3,5,6)P₅, InsP₆, and MIHS inhibitor conformations. The P_a, P_a′, and variable subsites are almost exclusively composed of side chains atoms. Consequently, differences of several angstroms in phosphoryl group position can be accommodated by the movement of side-chain atoms without significant main-chain movements.

Arg-68 adopts multiple conformations in the InsP₆ complex structure and apparently undergoes a large induced fit movement upon substrate binding and makes a specific contact with the C1 phosphate of InsP₆. While this residue is not interacting with Ins(1,2,3,5,6)P₅ substrate it is observed in a closed conformation. In contrast, this residue adopts an open conformation in all PhyAsr structures lacking substrate. Although Arg-68 is not strictly conserved in known PTPLPs, it may participate in substrate binding prior to catalysis, either guiding or orienting the ligand within the PhyAsr active site.

Mechanistic Insight into PhyAsr Catalysis

The K_d values for InsP₆ binding (this study) are at least 2 orders of magnitude lower than previously reported K_m values under similar conditions (12, 31). There are several potential explanations for the large, reported K_m values. Based on our structural analysis, binding studies, and the findings of Puhl et al. (12) we suggest that a two-step binding mechanism accounts for the discrepancy. Preliminary rapid kinetics data (Fig. 5, B and C) suggest that binding involves a rapid initial binding of the substrate followed by a slower step. Structural studies clearly indicate that PhyAsr binds substrate or inhibitor in at least three distinct conformations within the electropositive binding pocket. Further, only one conformation (InsP₆ complex) is consistent with catalytic competence. We speculate that the rapid initial binding phase represents substrate association with the electropositive binding pocket in one of several conformations followed by a slower phase that results in the substrate adopting a catalytically competent conformation.

To our knowledge, there have been no studies examining the binding of InsP₆ to phytases; however, the K_d for the binding of the competitive inhibitor MIHS (K_i) has been determined for both PhyA and PhyB from Aspergillus ficuum (36) and PhyA from Aspergillus fumigates (37). Interestingly, for each of these enzymes their K_i is at least 1 order of magnitude smaller than their K_m for InsP₆. These phytases belong to the histidine acid phosphatase family of phytase and are structurally similar to the E. coli phytase. It is not known whether the K_d for the substrate, InsP₆, is significantly lower than the K_m for these enzymes. Interestingly, the only other phytate-degrading enzyme that has been structurally characterized in complex with InsP₆ is the E. coli phytase (PDB code 1DKQ) (14). This enzyme hydrolyzes the C6 phosphate preferentially but was observed to bind the C3 phosphate within the catalytic site. This result suggests that E. coli phytase can bind InsP₆ in multiple conformations as observed for the PTPLP, PhyAsr. Similar to PhyAsr, the binding pocket of this enzyme is a highly basic cleft that makes extensive contacts with the phosphoryl groups near the catalytic site at the base of pocket.

The discrepancy between the binding affinity of MIHS and the K_m of InsP₆ in histidine acid phosphatases, the similarities in substrate binding by PhyAsr and the E. coli phytase, and the finding that the E. coli phytase can bind InsP₆ in multiple conformations, all suggest that histidine acid phosphatases and PTPLPs may both utilize a similar binding mechanism. This mechanism likely involves a rapid initial substrate binding followed by a slower reorientation and may be a general feature of all IPPases.