Significant Loop Motions in the SsoPTP Protein Tyrosine Phosphatase Allow for Dual General Acid Functionality

Abstract

Conformational dynamics are important factors in the function of enzymes including protein tyrosine phosphatases (PTPs). Crystal structures of PTPs first revealed the motion of a protein loop bearing a conserved catalytic aspartic acid, and subsequent NMR and computational analyses have shown the presence of motions, involved in catalysis and allostery, within and beyond the active site. The tyrosine phosphatase from the thermophilic and acidophilic Sulfolobus solfataricus (SsoPTP) displays motions of its acid loop together with dynamics of its phosphoryl-binding P-loop and the Q-loop, the first instance of such motions in a PTP. All three loops share the same exchange rate, implying their motions are coupled. Further evidence of conformational flexibility comes from mutagenesis, kinetics, and isotope effect data showing that E40 can function as an alternate general acid to protonate the leaving group when the conserved acid, D69, is mutated to asparagine. SsoPTP is not the first PTP to exhibit an alternate general acid (after VHZ and TkPTP), but E40 does not correspond to the sequence or structural location of the alternate general acids in those precedents. A high-resolution X-ray structure with the transition state analog vanadate clarifies the role of the active site arginine R102, which varied in structures of substrates bound to a catalytically inactive mutant. The coordinated motions of all three functional loops in SsoPTP, together with the function of an alternate general acid, suggest that catalytically competent conformations are present in solution that have not yet been observed in crystal structures.

Graphical Abstract

INTRODUCTION

Protein phosphorylation of side-chain hydroxyl groups is a major mechanism for regulation of protein function via the cooperative function of kinases and phosphatases.^1–2 The phosphatases include the classic protein tyrosine phosphatases (PTPs), specific for phosphotyrosine, and the closely related dual-specificity phosphatases (DSPs) that dephosphorylate phosphothreonine and phosphoserine as well as phosphotyrosine. PTPs and DSPs utilize the same catalytic machinery and mechanism portrayed in Figure 1. The conserved sequence C(X)₅R(S/T), termed the P-loop, contains the nucleophilic cysteine, appropriately oriented backbone amide bonds, and an arginine that provide for substrate binding and transition state stabilization. The PTP mechanism utilizes a conserved aspartic acid located on a mobile general acid loop of about a dozen residues, known as the WPD-loop in the classical PTPs, where the WPD sequence within this mobile loop is highly conserved.^3–5 In the catalytic reaction, the cysteine thiolate attacks the phosphorus as the conserved aspartic acid on the general acid loop protonates the scissile P-O ester oxygen atom. Kinetic isotope effects have shown that protonation occurs simultaneously with P-O bond fission, keeping the leaving group neutral in the transition state. The phosphocysteine intermediate is hydrolyzed in the second step of the reaction, in which the carboxylate of the same aspartic acid acts as a general base, deprotonating the nucleophilic water molecule. The positioning of the nucleophilic water in the second step is assisted by a glutamine residue on the Q-loop, another common feature of the PTP family.

Figure 1.

Chemical steps of the PTP-catalyzed reaction. The WPD-loop assumes a catalytically active closed conformation with the general acid (Asp) in position to protonate the leaving group during formation of the phosphoenzyme intermediate. In the second step this intermediate is hydrolyzed. After the phosphate product is released, the open WPD-loop conformation becomes favored.³

Protein motions are a key part of PTP function. Protein motions in PTPs were first noted in crystal structures that showed both open and closed positions of the WPD-loops of classical PTPs.⁶ NMR dynamics studies of two of the best characterized classic PTPs, the bacterial YopH and human PTP1B, found that the first chemical step is correlated with the rate of loop closure, which brings the general acid into the position needed for catalysis.⁷ Recent studies have shown that in PTP1B, and possibly other members of the family, other motions are correlated with WPD-loop motions and may also play a part in regulation of activity.⁸ Protein dynamics involving the acid loop and neighboring protein regions have also been revealed in VHR, the first DSP shown to exhibit such motions.⁹ Evidence is growing that structural fluctuations in the PTP family are more widespread than the WPD-loops of classical PTPs, and include protein dynamics tied to the catalytic reaction and to allostery.

Archaeal PTPs are less well characterized than eukaryotic and bacterial PTPs. Evolutionary studies indicate that the first recognizable member of the protein kinase (PK) and phosphatase superfamilies appeared after the divergence of the archaeal/eukaryal line from bacteria, but before the separation of Archaea from Eukarya. This suggests that the sophisticated phosphorylation/dephosphorylation known in Eukarya and bacteria today evolved from Archaea.^10–11 It was recently reported that the thermophilic phosphatase Tk-PTP undergoes a temperature-dependent conformational change of the substrate-binding P-loop.¹² Another such enzyme is the PTP from Sulfolobus solfataricus (S. solfataricus), a hyperthermoacidophilic organism discovered in the Solfatara volcano, belonging to the phylum Crenarchaeota that grows optimally from 60°- 90° C in acidic environments. The genome of S. solfataricus encodes for a single PTP, named SsoPTP, which has been characterized structurally and kinetically (in vitro) as a classical PTP.¹³ SsoPTP has a molecular weight of approximately 18 kDa and contains 161 amino acids, making it one of the smallest PTPs known. Nonetheless, the catalytic region of SsoPTP is structurally analogous to other PTPs (Figure 2). SsoPTP was classified as a PTP based on its active site depth, which is a distinguishing factor between PTPs and DSPs.¹³ SsoPTP contains the conserved P-loop HCX₅R(S/T) motif, but differs by having an isoleucine in place of the normally conserved tryptophan in the loop bearing the general acid (IPD rather than WPD), a difference it shares with some other PTPs, such as VHZ and Tk-PTP.^{12, 14–15} This loop will be referred to as the acid loop in this manuscript.

Figure 2.

Alignment of the catalytic regions of PTP1B (red)¹⁶, YopH (blue)⁶, and SsoPTP (green) each with bound vanadate, showing the common geometries of their active sites. The Q-loop is characterized by a conserved glutamine, shown in sticks. This residue positions a nucleophilic water molecule in the second catalytic step as shown in Figure 1. The acid-loop in PTPs contains a conserved general acid, shown in sticks, (D69 in SsoPTP) which protonates the leaving group in the first catalytic step. The binding-loop, also referred to as the P-loop, contains the nucleophilic cysteine, shown in sticks at the bottom of the figure which attacks the phosphorus in the first step of the mechanism.¹⁷

Crystal structures have been reported of several substrates bound to the inactive C96S mutant. In the structure with bound pNPP, the conserved arginine in the P-loop (R102) shows an unusual conformation, different from that in complexes with two pY-containing peptides.¹³ The latter two structures show the guanidinium group in bidentate hydrogen bonds with two nonbridging oxygen atoms of the substrate phosphoryl (-PO₃) group, consistent with this conserved residue’s positioning in other PTP crystal structures. In contrast, in the pNPP-bound enzyme structure, R102 makes one hydrogen bond with a nonbridging oxygen and one with the scissile, ester oxygen atom (Figure S1).¹³ This raises the possibility that the role of R102 may differ in SsoPTP from other PTPs.

This report presents a high-resolution crystal structure of SsoPTP bound to 2-chloroethylsulfonate and a complex with the transition state analog vanadate. Solution NMR evidence is presented showing that SsoPTP undergoes significant and distributed structural fluctuations of all three active site motifs: the acid loop, P-loop, and the Q-loop. One mechanistic consequence of these structural fluctuations manifests itself in the kinetic results, which indicate the ability to access two potential general acids in the first step of the mechanism. This is a rare, but not unprecedented, feature in the PTP family that has previously been found in the human enzyme VHZ¹⁴ and in Tk-PTP.¹² The alternate general acid implicated by the kinetic results is found in the crystal structures in a position such that structural reorganization, as implied by the NMR results, must occur to bring it into position for this role. During kinetics studies it was observed that SsoPTP activity is inhibited by some commonly used buffers, which is also summarized.

MATERIALS AND METHODS

Dithiothreitol (DTT) and ampicillin (AMP) were purchased from GoldBio. Restriction enzymes were purchased from Integrated DNA Technologies. Protease-inhibitor tablets were purchased from Sigma-Aldrich. All other buffers and reagents were purchased from Sigma-Aldrich or Fisher. The substrate p-nitrophenyl phosphate (pNPP) was synthesized using published methods.¹⁸ Crystallography screens, trays, and coverslips were purchased from Hampton Research. For labeled protein expression, deuterium oxide (D₂O), ¹⁵N-ammonium chloride, and ¹³C₆-D-glucose were purchased from Cambridge Isotope Laboratories (Tewksbury, MA).

An E. coli codon-optimized SsoPTP gene was obtained from General Biosystems. The gene was cloned into a pET-45b (+) vector using NcoI and SacI restriction sites. These restriction sites do not include the histidine tag and allow for tagless SsoPTP purification.

Mutagenesis

Template DNA of SsoPTP WT will be mutated using the Q5 Site-Directed Mutagenesis Kit from NEB. The primers were used to introduce the mutation aspartate (D)69 to alanine (A). and sequencing was performed by ACGT to confirm the presence of the mutation.

Protein expression and purification for kinetics, kinetic isotope effects, and crystallization.

The E. coli codon-optimized expression plasmid of SsoPTP was inserted into vector pET45B(+) and transformed in E. coli BL21(DE3) competent cells. Cells were grown in 1L of LB media containing ampicillin at 37°C until OD_600nm reached 0.6 −0.8, then induced with 100 mg of IPTG (final concentration 100 mg/L). The flask was transferred to a room temperature shaker and incubated for 14–18 hours. The cells were harvested by centrifugation at 4°C and 18,700 g. The cells were then re-suspended in buffer containing 50mM sodium succinate, 1mM EDTA, 5 mM DTT, 10% glycerol, pH 5.5 (4°C). An EDTA-free protease inhibitor cocktail tablet was added for every 50 mL of buffer. Cells were disrupted by sonication on ice in a metal beaker at 60% power and 60% pulse for two minutes. This was repeated four times. The cell lysate was centrifuged at approximately 12,800 g for 30 minutes at a temperature of 4°C. The supernatant was treated with a 10 % w/v solution of polyethyleneimine (PEI, Mw 50000, pH 8.0) at 4°C added dropwise to reach a final PEI concentration of 0.5% (w/v) and allowed to mix on ice for 15 minutes. This mixture was then centrifuged in the same fashion stated above. The supernatant was loaded at 1.5 mL/min on two Q-HiTrap (GE Healthcare) columns pre-washed with 2M NaCl and then pre-equilibrated with buffer containing 50 mM sodium succinate, 1 mM EDTA, 3 mM DTT, 10% glycerol, pH 5.5. The protein was loaded and washed with the same buffer used to equilibrate the column. The protein was then eluted with a 150 mL linear-gradient (2 mL/min) buffer containing 50 mM sodium succinate, 1 mM EDTA, 600 mM NaCl, 3 mM DTT, 10% glycerol, pH 5.5. The fractions showing UV absorbance at 280nm were tested for phosphatase activity by mixing 100 μL of protein from designated fractions with pNPP. The fractions showing phosphatase activity toward pNPP were assayed on a 15% SDS page gel to confirm the presence of SsoPTP. Confirmed fractions containing SsoPTP were concentrated to 5–8 mL and loaded on a Superdex 75 26/60 gel filtration column (GE-Healthcare) pre-equilibrated with buffer (640 mL corresponding to two-column volumes) containing 100mM succinate, 150 mM NaCl, 1 mM EDTA, 3 mM DTT, pH 6.0 (at 4°C). Purified protein was collected, concentrated, and flash-frozen in liquid nitrogen in the same buffer with 25% glycerol added. The concentration of each enzyme was determined by measuring absorbance at 280 nm using calculated extinction coefficients.

Protein expression and purification for NMR

E. coli BL21(DE3) cells were grown in M9 minimal medium supplemented with ¹⁵N-ammonium chloride (1.0 g/L), ¹³C₆-D-glucose (2.0 g/L) as the sole nitrogen and carbon sources, Cacl₂, MgSo₄, MEM vitamins, and either protonated or perdeuterated by substitution of D₂O for H₂O. Alternatively, for NMR relaxation experiments SsoPTP was expressed using ¹⁵N-NH₄Cl and perdeuterated with natural abundance (¹²C) glucose as the carbon source. Growths contained 100 mg/mL carbenicillin at 37 °C to an OD₆₀₀ of 0.7 – 1.0, induced with 0.5 mM IPTG, and SsoPTP was expressed at 37 °C for 5 hours. The cells were harvested by centrifugation, and the cell pellet was resuspended in the lysis buffer containing 20 mM CAPS (pH 10), 300 mM NaCl, 1 mM EDTA, 1 mM β-mercaptoethanol. The cells were disrupted by sonication on ice and incubated at 70 °C for ~ 30 min to remove unwanted proteins and subsequently separated by centrifugation. A 10% (w/v) solution Polyethyleneimine (PEI, Mw = 50000, pH 8.0 and 4 °C) was added to the supernatant to 0.5% (w/v) and incubated for 1~2 hours¹⁴. The white precipitate formed and was removed by centrifugation. Ammonium sulfate was added to the supernatant to 65% (w/v) incubated for 1~2 hours, and precipitated protein was collected by centrifugation and resuspended in Lysis buffer. The resuspended solution was dialyzed into a Q-column loading buffer containing 20 mM CAPS (pH 10.0), 50 mM NaCl, 1 mM β-mercaptoethanol. The supernatant was loaded onto a gravity column packed with a Q-sepharose high-performance resin (GE Healthcare) and washed extensively with the Q-loading buffer. The SsoPTP was eluted using a buffer containing 20 mM CAPS (pH 10.0), 500 mM NaCl, and 1 mM β-mercaptoethanol. The elute was dialyzed into NMR buffer 20 mM MES (pH 6.4), 50 mM KCl, 0.5 mM TCEP, 8% D₂O, and concentrated to 0.3–0.7 mM.

X-ray crystallography

Purified SsoPTP was concentrated to 14 mg/mL for crystallization trials at room temperature using the hanging drop diffusion method. Crystallization conditions were identified from the PACT Premier (Molecular Dimensions) screen condition C12 and further optimized to final crystallization condition of 0.1 M HEPES pH 7.0, 10.0 mM zinc chloride, and 18–20% PEG 6000 at a 1:1 ratio of protein:precipitant solution. For the vanadate bound crystal, vanadate was prepped beforehand by dissolving metavanadate in 0.1 M NaOH, boiled for 20 minutes, and frozen until use. This process ensures the monomeric form of vanadate.¹⁹ Vanadate was thawed immediately before use and added to the protein solution to reach a 1 μM concentration of vanadate, and kept on ice for at least 30 minutes until it was added to the crystallization drop. Crystals were harvested by transferring them into a stabilization solution that contained 50% sucrose. The crystals were flash-frozen in liquid nitrogen and stored under liquid nitrogen.

Diffraction data on the apo and vanadate bound forms of SsoPTP were collected at the Stanford Synchrotron Radiation Lightsource (SSRL). As described in the results and discussion section, a HEPES precursor molecule (2-chloroethylsulfonate) was identified at the active site of the apo structure. In the vanadate bound structure, 2-chloroethylsulfonate is not observed. Data was processed and indexed using HKL3000.²⁰ Molecular replacement was performed with Phaser²¹ as implemented in Phenix,²² using the previously published ligand-free structure of SsoPTP (PDB ID 2I6I) as a search model.¹³ Model building was performed using Coot.²³ Structure refinement was performed using phenix.refine (Table 1).²⁴ PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC. was used to create figures in this manuscript.

Table 1.

Data collection and refinement statistics for SsoPTP-2-chloroethylsulfonate and Sso-vanadate complexes.

	SsoPTP-2-chloroethylsulfonate	SsoPTP-vanadate
Data collection
PDB ID	7MPD	7MPC
Beamline	SSRL 9–2	SSRL 9–2
Wavelength (Å)	0.979	0.979
Unit cell dimensions (a, b, c (Å); α, β, γ (˚))	83.9, 83.9, 42.4, 90.0, 90.0, 120	84.6, 84.6, 42.3, 90.0, 90.0, 120
Resolution range (outer shell) (Å)	36.3 – 1.05 (1.09 – 1.05)	36.7 – 1.75 (1.81– 1.75)
No. of reflections
Unique	78644 (7241)	17577 (1704)
Total	1339872	307727
Average redundancy	17.0 (9.40)	17.5 (8.60)
Mean I/σ(I)	12.0
Completeness (%)	99.1 (91.6)	99.8 (98.2)
Rmeas	0.081 (0.700)	0.191 (0.595)
CC1/2	0.995 (0.894)	0.985 (0.759)
Space group	P63	P63
Refinement
Rwork	0.129 (0.160)	0.166 (0.277)
Rfree	0.141 (0.1780)	0.201 (0.335)
Atoms in structure
Protein	1382	1383
Water	184	74
Ligands/ions	42	5
Average B-factors (Å 2 )
Protein	16.8	38.0
Water	33.4	42.9
RMSD bond (Å)/angle (°)	0.013/1.234	0.013/ 1.47
Protein Geometry
Ramachandran favored (%)	96.2	96.2
Ramachandran outliers (%)	0.63	0.63
Rotamer outliers (%)	0.00	0.00

Kinetics

Steady-state kinetics at room temperature.

Michaelis-Menten kinetics were conducted as previously reported.^{3, 25–26} All kinetics experiments were carried out using from 0.5 – 15 mM p-nitrophenyl phosphate (pNPP) in 200 – 500 mM succinate buffer using enzyme concentrations in the range of .05 – 3.3 μM. Reactions were carried out in 96-well plates and initiated by addition of enzyme to a final reaction volume of 330 μL. After times ranging from 5–20 minutes, reactions were quenched by addition of 50 μL 10 M NaOH. The formation of p-nitrophenolate was measured from the absorbance at 400 nm using the molar extinction coefficient of 18,300 M⁻¹ cm⁻¹. Final absorbances were corrected for background hydrolysis of the substrate obtained by replicating each condition in the absence of enzyme. Kinetic parameters were determined by fits of the initial rate versus concentration of pNPP to the Michaelis-Menten equation.

Determination of inhibition constants.

The effect of buffers and other inhibitors on the activity of SsoPTP was tested with pNPP in 100 mM sodium acetate buffer (pH 5.0), which showed no inhibitory effect, with inhibitor concentrations in the range 5–400 mM. The kinetic data were fit to competitive, noncompetitive, and mixed inhibition models using the Visual Enzymics software to determine the modes of inhibition reported in Table 2.

Table 2.

Buffers, modes of inhibition, and K_i values

Compound	Inhibition type	Ki, mM
Acetic acid	No inhibition observed	-
Citric acid	Competitive	8 ± 1
MOPS	No inhibition observed	-
Bis-Tris	Competitive	123 ± 1
Tris	Competitive	109 ± 2
Succinic acid	No inhibition observed	-
CAPS	Mixed	54 ± 11
HEPES	Competitive	44 ± 8

Kinetic isotope effects (KIEs).

The kinetic isotope effects (KIEs) were measured by the same methods used in prior studies of PTPs^27–28 using the internal competition method, and thus are isotope effects on k_cat/K_m, commonly referred to as V/K. This means that the kinetic isotope effects report on the part of the overall mechanism up to and including the first irreversible step, which is phosphoenzyme intermediate formation. Natural abundance pNPP, synthesized with commercial p-nitrophenol containing the natural abundance of ¹⁴N and ¹⁵N, was used for the measurements of ¹⁵(V/K). The ¹⁸O KIEs were measured by the remote label method using the nitrogen atom in pNPP as a reporter for the isotope fractionation in the labeled oxygen positions. The methodology and the rationale for using nitrogen isotope ratios in the reactant close to natural abundance have been described.²⁹ Figure S2 shows the isotopic isomers of the substrate used for KIE determinations. Figure 3 shows the positions where KIEs were measured, and the designations used.

Figure 3:

The substrate p-nitrophenyl phosphate (pNPP) showing the positions at which kinetic isotope effects were measured: the bridge oxygen atom, the position of bond cleavage, ¹⁸(V/K)_bridge; the nitrogen atom in leaving group, ¹⁵(V/K).¹⁴

Isotope effect determinations were carried out in triplicate at 25 °C in either 50 mM succinate at pH 4.75 (wild type SsoPTP) or 50 mM sodium acetate at pH 4.5. In each KIE trial, 40 mg of pNPP was dissolved in 5 mL of buffer. 200 μL of enzyme, ranging from 0.3–5.9 mg/mL, were added to start the reaction. Production of p-nitrophenol product was followed by following the absorbance at 400 nm of an aliquot of the reaction solution added to 0.1N NaOH. After reactions reached 40%−60% of completion, HCl was used to titrate the solution to a pH of ~3.0 to stop the enzymatic reaction. The solution was extracted with three 50 mL portions of diethyl ether to isolate the p-nitrophenol product. Ether fractions were collected, dried with anhydrous magnesium sulfate, filtered, and ether was removed by rotary evaporation. 3 mL of 1.0 M Tris buffer was added to the aqueous layer, and the pH was adjusted to 9.0. Alkaline phosphatase was added to hydrolyze the remaining unhydrolyzed pNPP and allowed to react overnight. The pNP product was then extracted and collected as described for the first part of the reaction. Samples were then purified by sublimation at 95°C under vacuum. After sublimation, 1 mg of p-nitrophenol was collected from each trial and analyzed by isotope ratio mass spectrometry. The isotope effect was calculated from the nitrogen isotopic ratios in the p-nitrophenol product at partial reaction (R_p), in the residual substrate (R_s), and the starting material (R_o). Each experiment gives two independent determinations of the isotope effect: R_s and R_o (Eq. 1), the other from R_p and R_o (Eq. 2).³⁰ For each isotope effect the values calculated from R_o and R_s and R_o and R_p were averaged to give the results reported in Table 3.

$$\text{isotope}\text{effect}\text{=}\text{log}\left(\text{1}-\text{f}\right)\text{/}\text{log}\left[\left(\text{1}-\text{f}\right)\left({\text{R}}_{\text{s}}{\text{/}\text{R}}_{\text{o}}\right)\right]$$ [1]

$$\text{isotope}\text{effect}\text{=}\text{log}\left(\text{1}-\text{f}\right)\text{/}\text{log}\left(\text{1}-\text{f}\left({\text{R}}_{\text{p}}{\text{/}\text{R}}_{\text{o}}\right)\right)$$ [2]

Table 3.

Kinetic isotope effects for reactions of members of the PTP superfamily with the substrate pNPP. Standard errors are in the range 0.0001–0.0008 if not shown.

Enzyme	15(V/K)	18(V/K)bridge
YopH, PTP1, VHR, PTP1B28	0.9999 – 1.0004	1.0118 – 1.0152
D to N mutants of YopH, PTP1, VHR, PTP1B28	1.0024 – 1.0030	1.0275 – 1.0297
Stp128	1.0007	1.0171
VHZ28	1.0013	1.0164
SsoPTP WT	0.9994 ± 0.0008	1.0198 ± 0.0080
SsoPTP D69N	0.9999 ± 0.0006	1.0219 ± 0.0088
SsoPTP D69N E138Q	1.0008 ± 0.0004	1.0011 ± 0.0071

The ¹⁵(V/K) KIE is given directly from these equations. In the ¹⁸O isotope effect experiments, the observed KIEs obtained from the above equations were corrected as previously described for the ¹⁵N isotope effect and for incomplete levels of isotopic incorporation.²⁷ The levels of isotopic incorporation for the labeled substrates and the mixtures used for the ¹⁸O experiments were determined by mass spectrometry.³¹

NMR Experiments

All NMR data were collected on Varian spectrometers (600 MHz, 700 MHz, and 800 MHz). The NMR data were processed and analyzed using NMRPipe³² and NMRFAM-SPARKY³³, respectively. The temperature was calibrated using methanol³⁴, and the chemical shift was referenced internally using sodium trimethylsilylpropanesulfonate (DSS).³⁵

Two-dimensional ¹H-¹⁵N HSQC spectra were collected on protonated or perdeuterated ¹⁵N SsoPTP at 35 °C in the NMR buffer (20 mM MES (pH 6.4), 50 mM KCl, 0.5 mM TCEP, 8% D₂O) performed at 600, 700, and 800 MHz. The ¹H frequency was centered at the water resonance with a spectral width of 12,000, 14,000, and 16,000 Hz at 600, 700, and 800 MHz, respectively. The ¹⁵N carrier frequency was centered at 120 ppm with a ¹⁵N spectral width of 2,500, 2,916, and 3,333 Hz at 600, 700, and 800 MHz, respectively. Forty-eight transients were collected for each of 128 t₁ points.

To assign the NMR resonances to sequence-specific residues in SsoPTP, a suite of backbone assignment experiments was performed: HNCA, HN(CO)CA, HNCACB, HN(CO)CACB, HNCO, and HN(CA)CO³⁶ on ligand-free SsoPTP. In addition, the HNCACB experiment was acquired on SsoPTP in the presence of saturating amounts of tungstate. Saturation was achieved through titration with sodium tungstate while collecting 2D-HSQC spectra at 35 °C in the NMR buffer. NMR spectra were acquired for titration points of 0.10, 0.59, 3.30, 12.5, 20.0 mM ligand.

NMR Relaxation experiments

R_1ρ experiments

Off-resonance R_1ρ experiments were acquired at a static magnetic field of 14.4 T (600 MHz). The spin-lock field strength (ω₁) was calibrated by measuring residual one-bond scalar coupling (¹J_NH) in the absence and in the presence of ¹⁵N decoupling during t₂ in a ¹H-¹⁵N HSQC as described.³⁷ R^β_1ρ values were measured using the TROSY-selected R_1ρ experiment.³⁸ Data were collected on 0.5 mM perdeuterated ¹⁵N SsoPTP at 35 °C in the NMR buffer. The ¹H frequency was centered at the water resonance at 4.68 ppm with a spectral width of 12,000 Hz. The ¹⁵N carrier frequency was centered at 120.0 ppm with a ¹⁵N spectral width of 2,500 Hz. Sixteen transients were collected for each of 80 t₁ points.

The R^β_1ρ is given by equation 3.

$$R_{1\rho }^{\beta }=R_1^{\beta }{\cos }^2{\theta }^{\beta }+R_2^{\beta }{\sin \theta }^{\beta }$$ [3]

$$R_2^{\beta }=R_2^{\beta ,0}+R_{ex}$$ [4]

In equation [3], $R_1^{\beta }\mathrm{a}\mathrm{n}\mathrm{d}{R}_2^{\beta }$ are the longitudinal and transverse relaxation rates of the slowly relaxing (β) component of the N-H doublet and θ^β = arctan(ω₁/Ω^β) is the tilt angle for the effective spin-lock field and Ω^β is the offset frequency of the applied field to the resonance shift for the TROSY component (Ω^β = Ω+πJ). R^β,0₂ is the transverse relaxation rate constant due to ¹⁵N CSA and ¹H-¹⁵N dipole-dipole relaxation mechanisms, and R_ex is the contribution of conformational exchange to the measured R^β_1ρ value in which $R_{ex}=\frac{{\varphi }_{ex}{k}_{ex}}{k_{ex}^2+{\omega }_e^2}$ k_ex is the sum of the forward and reverse conformational exchange rate constants and ${\varphi }_{ex}=p_a{p}_b{\mathrm{\Delta }\omega }_N^2$ in which p_a(b) are the equilibrium populations of the assumed, two conformations and Δω_N is the ¹⁵N chemical shift difference of the resonance of interest between the two conformations.

The off-resonance (Ω) field was varied by changing the values of the ¹⁵N frequency offset that occurs during the relaxation period: offsets used had values of −626, −1200, −1600, −2000, −2400, −2800, −3200, −3600, and −4000 Hz. At each of the different offset values, intensity decay of individual resonances was measured in each spectrum collected at seven different relaxation delay time points with two duplicates for error analysis. Peak heights were measured and fit to a single exponential curve to obtain the R^β_1ρ; I(t) = I(0)exp[−R^β_1ρt]. The relaxation delay times used were: 0.002 (x2), 0.02, 0.05, 0.08 (x2), 0.1, 0.13, 0.16 s.

Measurement of TROSY-selected R^β₁

R^β₁ values were measured to reduce the number of variables needed for fitting relaxation dispersion curves. In each R^β₁ experiment, intensity decays of resonances were measured in spectra collected at ten different relaxation delay time points with two duplicates for error analysis. Peak heights were measured and fit to a single exponential curve to obtain the R^β₁ constant; I(t) = I(0)exp[-R^β₁t]. The relaxation delay times used were: 0.002 (x2), 0.1, 0.2, 0.3, 0.4 (x2), 0.5, 0.6, 0.7, 0.8, and 0.9 s. R^β₂ and R_ex were determined from the measured values of R_1ρ^β and R₁^β and the effective spin-lock field (${\omega }_e={({\omega }_1^2+{\mathrm{\Omega }}^2)}^{1/2}$ using in-house written Perl scripts.

Carr-Purcell Meiboom-Gill (CPMG) NMR relaxation dispersion experiments

CPMG data were collected on 0.5 mM perdeuterated tungstate-saturated ¹⁵N SsoPTP at 35 °C in the NMR buffer (20 mM MES (pH 6.4), 50 mM KCl, 0.5 mM TCEP, 8% D₂O) performed at 600 MHz and 700 MHz. The ¹H frequency was centered at the water resonance at 4.725 ppm with a spectral width of 8,400 Hz. The ¹⁵N carrier frequency was centered at 120 ppm with a ¹⁵N spectral width of 2,500 Hz. Sixteen transients were collected for each of 100 t₁ points.

The transverse relaxation constant, R₂, was measured at different 𝜏_cp values during the CPMG cycles. Heat compensated CPMG pulses were employed during the recycle delay to ensure a consistent number of ¹⁵N 180º pulses during each experiment. 𝜏_cp values used were: 0.5, 0.75, 1, 1.25, 2, 3.5, and 5 ms giving overall relaxation times: 200, 180, 176, 180, 176, 168, and 200 ms.

The intensity decay of each resonance was measured in spectra collected at different relaxation delay points with two duplicates for error analysis. Peak heights were measured and fit to a single exponential curve to obtain the R₂ constant; I(t) = I(0)exp[-R₂t]. The resulting dispersion curves were best fit to the expression for conformational exchange under the fast-limit according to the Luz-Meiboom³⁹ equation.

RESULTS AND DISCUSSION

Buffer inhibition.

In light of our previous finding that certain buffers inhibit VHZ.¹⁴ inhibition studies were conducted with several biological buffers, including Tris, Bis-Tris, succinate, acetate, and CAPS. Several buffers were found to be weakly inhibitory (Table 2). CAPS and HEPES presumably inhibit due to their sulfonate group, which sterically and electrostatically resembles phosphate, although MOPS did not show measurable inhibition. The origins of the relatively strong inhibition by citrate, or the weaker effect of Tris and Bis-tris, are uncertain. The inhibition constant for citrate is comparable to the K_M for pNPP (3.4 mM), ruling it out as a useful buffer. Succinate or acetate was used in all subsequent kinetics.

X-ray structural results.

The SsoPTP complex with vanadate.

Crystal structures of WT PTPs complexed with nonhydrolyzable substrate analogs, such as phosphonates, show the conserved active site arginine in bidentate hydrogen bonds with the phosphoryl group. The analogous orientation is seen in PTPs complexed with the transition state analog vanadate and vanadate esters.^{15, 25–26, 40–46} Mutation of the active site cysteine to serine disables catalysis and allows complexation of phosphate ester substrates. Three such complexes of C96S SsoPTP have been reported: one with pNPP and two with pY-containing peptides.¹² The two peptide complexes show the typical interactions of Arg 102 with the phosphoryl group. In contrast, the pNPP complex shows one of the nitrogen atoms donating a hydrogen bond to the scissile ester bond (Figure S1).

The SsoPTP crystal structure complexed with vanadate (Figure 4) shows a strong resemblance to other PTP-vanadate structures.^{15, 25–26, 40–46} The vanadate complex shows the trigonal bipyramidal geometry typical in PTP-vanadate complexes, and the guanidinium side chain of R102 forms bidentate hydrogen bonds. A carboxylate oxygen atom of D69 is 2.7 Å from the apical oxygen corresponding to the scissile oxygen atom of substrate in the catalytic reaction. This is consistent with its assignment as a general acid based on its structural analogy with other PTPs. The side-chain carbonyl oxygen of Q135 in the Q-loop is 3.3 Å away from this apical oxygen, consistent with a role positioning the nucleophilic water in the second step, also a common PTP active site characteristic.

Figure 4.

SsoPTP complexed with the transition state analog vanadate. The R102 interactions in this complex are typical of those observed in other PTPs indicative of bridging interactions with two oxygen atoms of the -PO₃ (phosphoryl) group. Positions of the conserved general acid D69 and the potential general acid E40 are shown. The pH where this structure and previous ones have been obtained is well above optimal activity, where E40 is ionized and found in a salt bridge with R102 and not in a location amenable to protonate the leaving group. Structural reorganization, potentially facilitated at lower pH where E40 should be largely protonated, would be required to bring it into proximity to fill the role implied by the kinetic data. Alternatively, given the structure with pNPP bound to the inactive mutant (Figure S1), E40 may act in concert with R102 as the proton donor.

The transition state analog vanadate has the trigonal bipyramidal geometry in the active site, and interactions with active site residues, that are typical of PTP-vanadate complexes. However, the vanadium-C96 sulfur distance of 2.9 Å is longer than usually observed (for comparison, this distance is 2.4 Å in the vanadate complexes of PTP1B and VHZ) and the electron density indicates that no covalent bond has formed (Figure S3). A similar finding was reported in the vanadate complex with VHZ.¹⁵ A computational analysis showed that even in the more conventional PTP1B complexes with vanadate and a vanadate ester, the apical interactions are not classic 2c-2e bonds, but are highly delocalized bonds.⁴⁷ The chemistry of vanadate is more complex than phosphorus, but the propensity for vanadate to form stable trigonal bipyramidal complexes, structurally mimicking the transition state for phosphoryl transfer, make it a potent inhibitor of PTPs and other phosphatases. The absence of a covalent interaction is consistent with data showing the transition state for phosphoryl transfer is loose, with low bond orders to the nucleophile and leaving group.

While the SsoPTP-vanadate complex shows some differences in V-O and V-S distances compared with the analogous complex in PTP1B, the overall geometry is consistent with the trigonal bipyramidal transition state typical of PTPs, and together with the kinetic isotope effects (discussed below), supports the conservation of the typical PTP mechanism and transition state.

SsoPTP-2-chloroethylsulfonate structure.

Because the published ligand-free structure was crystallized using citrate buffer (PDB ID: 2I6I), which we found to be inhibitory, we attempted to obtain a ligand-free structure in the absence of citrate. When crystallized using the buffer HEPES, we observed electron density at the SsoPTP active site that is consistent with a precursor in the commercial HEPES synthesis, 2-chloroethylsulfonate (Figure S4). The presence of 2-chloroethylsulfonate in the HEPES buffer was confirmed by mass spectrometry (Figure S5). The sulfonate portion of the molecule binds at the same position as the vanadate in the SsoPTP-vanadate structure described above. Notably, a complete HEPES molecule bound in the same position would result in significant clashes with the SsoPTP backbone. HEPES has been reported to inhibit the bovine low molecular weight PTP about 10-fold more strongly than we observe for SsoPTP, together with a crystal structure of that enzyme’s complex with HEPES.⁴⁶ These observations suggest that the observed inhibition of SsoPTP by HEPES may be due to the presence of low levels of the precursor.

Solution NMR experiments.

Solution NMR Backbone Resonance Assignments of ligand-free SsoPTP.

SsoPTP consists of 161 amino acid residues, 7 of which are proline and do not yield detectable signals in the two-dimensional ¹H-¹⁵N TROSY spectrum. Initially, for ligand-free SsoPTP, 130 residues out of 154 were assigned (84%). The remainder of the unassigned resonances were deemed to result from two factors. First, there were fewer observable peaks in the HSQC compared to the expected number. Second, many of the unassigned peaks were located in loop regions (Figure 5 and S4). Both observations suggested that conformational dynamics in these regions of the enzyme, resulting in exchange broadening and low signal-to-noise (S/N), prevented their observation.^48–49 The assigned residues and unassigned residues are mapped onto the crystal structure of the ligand-free SsoPTP as solid and dashed ribbons, respectively, in Figure S6. Specifically, the GG motif in the P-loop, C96-V97-G98-G99, were unassigned, which is included in the conserved PTP motif, HCVGGIGRT.

Figure 5.

2D ¹H-¹⁵N-TROSY spectra of SsoPTP at 35 °C. (Top) ligand-free SsoPTP. Unassigned residues are shown absent a label. (Bottom) SsoPTP in the presence of 200-fold molar excess of tungstate at 35 °C. Residues corresponding to each backbone amide H-N correlation are labeled. Resonances that appear in the presence of tungstate are G98, G101, R102, T103, and G104.

We hypothesized that the loop regions, including the P-loop, might be flexible in the absence of the substrate or product. Flexibility in the P-loop, while uncommon is not without precedent as crystallographic evidence in Tk-PTP suggest a P-loop that is able to assume two conformations {Yun, 2018 #17187}. The P-loop, in particular, binds to substrate/product via several hydrogen bonds and charge interactions, and we envisioned this binding might restrict these motions and limit the conformational exchange broadening. To test the hypothesis, we added the product-state mimic ligand, tungstate to free SsoPTP. The affinity of tungstate, which is a chemical analog of phosphate and a competitive inhibitor versus the substrate, is tighter than that of phosphate. A TROSY spectrum of SsoPTP was collected in the presence of excess tungstate (1:200 molar ratio). Interestingly, we observed three new resonances in the spectrum of tungstate-saturated SsoPTP (Figure 5). Moreover, several other residues became more intense in the HSQC upon tungstate binding. These remaining resonances, together with additional resonances, were assigned by HNCA, HNCACB experiments. As a result, the P-loop residue of G98, G99, and the conserved motif G101 and R102 were assigned and aided further assignments near the P-loop; thus, 138 residues out of 154 were assigned (90%) in total (Figure 5). The assigned residues and unassigned residues are mapped onto the crystal structure of the tungstate bound SsoPTP as solid and dashed ribbons, respectively (Figure 6).

Figure 6.

The assigned residues are mapped onto the crystal structure of the tungstate (shown in CPK rendering) bound SsoPTP as shown as a solid ribbon, and the unassigned residues are shown in dashed ribbon. Proline residues are colored in black. Secondary structures are colored as follows: α-helices are in red, β-strands are in yellow, and loop regions are represented in green. PDB ID: 2I6M

The amide backbone assignment results suggest that the P-loop is less flexible when coordinated to tungstate. Unexpectedly, however, we observed resonance broadening (or disappearance, in some cases, from the spectrum) of G70 (acid loop) and V97 (P-loop) in the tungstate bound SsoPTP spectrum, suggesting some residual or altered conformational exchange motions specifically in an intermediate exchange regime. This was the first indication of multiple dynamic events experienced by the acid and P-loop in SsoPTP. Consequently, the structure and dynamic properties of the conformational exchange process upon tungstate-binding were further characterized by monitoring amide chemical shift changes for ligand-free and tungstate-bound SsoPTP.

Tungstate Titration.

NMR has served as an exquisite technique to identify enzyme-ligand interaction sites^50–52 and provide unique information in the identification of allosteric sites in enzymes.^{50, 53–54} The chemical shift of amide proton (¹H) and nitrogen (¹⁵N) resonances reveal the structure and dynamic changes in response to the ligand binding. Tungstate was titrated in the ligand-free SsoPTP solution, and 2D TROSY-HSQC spectra were collected at different enzyme-ligand ratios at 35 °C (Figure 7). The enzyme-ligand ratio was varied from the free (ligand-free) to 1:0.5, 1:3.6, 1:21, 1:100, and 1:200. During the tungstate titration, the movement of the NMR resonances followed four patterns: significant chemical shift changes, resonances broadening, appearance of new resonances, and no response.

Figure 7.

The overlay of 2D ¹H-¹⁵N-TROSY spectrum of the SsoPTP during the tungstate titration at 35 °C. The ligand-free (free) spectrum is shown in red, and the tungstate bound spectra (enzyme-ligand ratio) are shown in orange (1:0.5), yellow (1:3.6), green (1.21), blue (1:100), and purple (1:200). Residues that appear in the presence of tungstate are located at the active site, and identified by the blue circle. Residues that disappear in the presence of tungstate are also near the active site and are highlighted with the red circle.

To quantify the effects of tungstate binding, we analyzed the chemical shift perturbations (CSP) and fit equation 5 to the titration data of quantifying the dissociation constant (K_d) both globally and for the active site residues only (5).

$$∆\delta =0.5{∆\delta }_{max}\left[1+\mathrm{X}+\frac{{\mathrm{K}}_{\mathrm{d}}}{\left[\mathrm{S}\mathrm{s}\mathrm{o}\mathrm{P}\mathrm{T}\mathrm{P}\right]}\right]-\sqrt{{(1+\mathrm{X}+\frac{{\mathrm{K}}_{\mathrm{d}}}{\left[\mathrm{S}\mathrm{s}\mathrm{o}\mathrm{P}\mathrm{T}\mathrm{P}\right]})}^2-4\mathrm{X}}$$ [5]

In equation 5, X is the SsoPTP:WO₄^2- ratio at each titration point, and ${∆\delta }_{max}$ is the maximal chemical shift measured for each residue. The active site residues were globally fit to yield the K_d of 0.64 ± 0.059 mM (Figure S7), and entire residues were globally fit to yield the K_d of 0.71 ± 0.033 mM. The similarity of the K_d values indicates an identical global response of SsoPTP to tungstate binding and supports a single tungstate binding site.

To summarize, two resonances from the acid residue G70 and the P-loop residue V97 were only present in the ligand-free spectrum, however, as the titration proceeds, the peak heights were reduced, and the peak widths were increased, leading to the resonance broadening in the tungstate-saturated HSQC spectrum shown in (Figure 5). This resonance broadening is the hallmark of motions on the intermediate exchange timescale. In addition, G98, G101, R102, which are located in the conserved P-loop motif, were not present in the ligand-free HSQC spectrum, but these resonances appeared in HSQC spectra in the presence of high concentrations of tungstate (Figure 5), suggesting that G98, G101, and R102 residues experience intermediate conformational exchange (i.e. k_ex ~ Δω) in the ligand-free state, and restriction in their conformational motions in the tungstate bound state.

Overall, different dynamic processes were observed in the ligand-free and the tungstate-bound forms of SsoPTP in which the active site has a selective mode of conformation to coordinate the product state. To probe the conformational timescale of the active-site residues (G70, V97, G98, G101, and R102) in the tungstate saturated form compared to the ligand-free enzyme, we performed NMR relaxation dispersion experiments to assess molecular motions from the μs – ms timescale.

Characterizing microsecond to millisecond motions.

The backbone dynamics of the ligand-free and tungstate-bound enzyme were probed utilizing the TROSY-selected (TS) off-resonance R_1ρ experiment, and the relaxation compensated CPMG (rcCPMG) experiment at 35 °C, the same temperature at which the backbone assignments experiments and the ligand titrations were performed. These two NMR methods are sensitive to enzymatic motions occurring from microsecond to millisecond, in which the ¹⁵N-R_1ρ experiment is capable of detecting faster motions than the ¹⁵N-CPMG experiment.

Motions of two of acid loop residues, D69 and V72, were identified as dynamic in the sub-millisecond timescale (Figure 8). Dispersion curves of these residues were fit to the eq 3, yielding a shared conformational exchange rate constant (k_ex) value of 3140 ± 470 s⁻¹. The identical k_ex suggests that the motions of these correlated residues undergo the same dynamic process. Individual fits of D69 and V72 are 4000 ± 2170 s⁻¹ and 3600 ± 1940 s⁻¹, indicate that this global fit is a better representation of the experimental data, not surprising given their spatial proximity in the SsoPTP structure.

Figure 8.

Conformational motions of ligand-free SsoPTP as probed by R_1ρ experiments. The dynamic residues, D69 and V72, in the acid loop are shown onto the crystal structure of the ligand-free SsoPTP as red spheres. Relaxation data was collected at 600 MHz. Global fit to the two-site exchange equation (eq 3) is represented with the solid line. PDB code: 2I6I

CPMG experiment to characterize tungstate bound SsoPTP.

To measure motions of tungstate-bound SsoPTP, the CPMG relaxation dispersion experiment was performed at two static magnetic field strengths (600 MHz and 700 MHz), and the transverse relaxation rate, R₂, as a function of the applied spin-echo pulsing rate, 𝜏_cp, was measured. Initially, all relaxation dispersion curves for residues in secondary structure units were fit by grouping residues in the same secondary structure. For the acid loop residues, I67, D69, and V72 were identified as dynamic and yielded dispersion curves that could be fit with good confidence (Figure S8). Relaxation dispersion curves were fit with an exchange rate value of k_ex = 860 ± 200 s⁻¹. Thus, in the presence of bound tungstate, the motions of the acid loop were reduced by four-fold in comparison to the ligand-free acid loop (3140 s⁻¹). This result suggests that tungstate binding reduces the timescale of acid loop motions.

Residues distinct from the acid loop also show evidence for conformational exchange motions in the tungstate bound SsoPTP enzyme. Residues in the conserved P-loop motif, G99, R102, and T103, undergo conformational exchange in the presence of tungstate, with a shared k_ex = 660 ± 140 s⁻¹ (Figure S9). Also, Y2 is located outside of the active site but shows a relaxation dispersion profile, with a similar k_ex values, (k_ex = 610 ± 260 s⁻¹), to those the P-loop, forming a small dynamic cluster (Figure S9). A weak van Der Waals interaction exists between the side chain of Y2 and T103 that likely facilitates these coupled motions. In addition, Yun and coworkers¹² have shown that a stabilizing interaction between the hydroxyl group of Y2 and the carbonyl group of G99 in the P-loop via a hydrogen bond. Thus, the motions in the conserved P-loop motif (G99, R102, and T103) are propagated to the nearby N-terminus of SsoPTP.

Lastly, the Q-loop residues A133 and Q135 display dispersion curves in the CPMG experiments, the globally fit k_ex value for these two residue is k_ex = 900 ± 370 s¹ (Figure S10). Importantly, Q135 is the conserved glutamine residue that serves to help orient the water at the active site during the hydrolysis step of catalysis.

Importantly, the conformational exchange motions of the P-loop and Q-loop have not previously been reported in class I PTP enzyme. Other classical PTPs display a rigid P-loop and Q-loop. The similarity of the k_ex values from the individual fits of the dispersion curves for the separate loop regions suggested that they might all be part of the same conformational exchange process. To test this, we fit all the relaxation dispersion data globally using a single k_ex value (Figure 9). This global fit yielded a k_ex value = 810 ± 220 s⁻¹. In addition, the fitted uncertainty for the global fit is smaller than those for the individual fits. An F-test comparison (p = 0.66) indicates the simpler (global) model should not be rejected. This suggests and provides the first evidence for coupled motions of all the active site loops in the product bound state of SsoPTP. Further evidence of active site flexibility comes from kinetic data indicating that SsoPTP can access a second residue to function as a general acid, a phenomenon previously characterized in VHZ and in another thermophilic enzyme, TkPTP.

Figure 9.

Conformational motions of tungstate bound SsoPTP as probed by CPMG relaxation dispersion experiments. (left) ¹⁵N-CPMG relaxation dispersion curves of the dynamic residues in tungstate bound SsoPTP are plotted. Global fit to the fast two-site exchange equation⁵⁵ represented with solid line. (right) The dynamic residues in the acid, P-, and Q-loop are mapped onto the crystal structure of the tungstate bound SsoPTP as shown as spheres. PDB ID: 2I6M

In conclusion, conformational motions in the ligand-free and tungstate bound of SsoPTP that occur on a time scale of sub-milliseconds to milliseconds are summarized in Figure 10. Motions are reduced by four-fold, and functional loops containing conserved catalytic residues undergo conformational exchange upon tungstate binding, implying that loop conformation and motions could have a functional role in the catalytic process. The NMR relaxation dispersion experiments uncovered that acid loop is mobile in ligand-free, and tungstate binding allows the propagation of motions from the acid loop to the P- and Q-loop.

Figure 10.

Conformational motions of ligand-free and tungstate bound SsoPTP are probed by R_1ρ and CPMG experiments. In ligand-free, D69, V72 of the acid loop mobile with a k_ex = 3140 s⁻¹. In the tungstate bound form, I67, D69, V72 of the acid loop, G99, R102, and T103 of the conserved PTP motif, Y2, and A133, Q135 of the Q-loop are also flexible with k_ex reduced and = 760 s⁻¹. The dynamic residues are mapped onto the crystal structure of the ligand-free and tungstate bound SsoPTP, as shown in spheres. PDB code: 2I6I (ligand-free), 2I6M (tungstate bound).

Kinetic results

Kinetic evidence for dual general acids.

The native SsoPTP displays the bell-shaped pH-rate profile typical of the PTP family (Figure 11). The acidic limb reflects the nucleophilic cysteine, and the basic limb the general acid. The residue D69 occupies the typical position of the general acid in PTP active sites. The mutation of the general acid in PTPs from Asp to Asn typically results in observations consistent with full loss of general acid catalysis: a rate reduction of 2–3 orders of magnitude with the abolishment of the basic limb of the pH-rate profile^56–57 and leaving group kinetic isotope effects with the substrate pNPP that show the leaving group departs as the anion.²⁸ In contrast, the SsoPTP D69N variant showed only a modest reduction in rate, and retention of a bell-shaped pH-rate profile.

Figure 11.

pH profiles of SsoPTP WT (blue squares), D69N (red circles), E40Q (green diamonds), and D69N E40Q (black triangles). Data were obtained at 25°C in 200mM succinate. D69N, WT, and E40Q all display bell-shaped trends consistent with operational general acid catalysis. The rate reduction seen in the double mutant D69N/E40Q (black), along with the elimination of the basic limb, indicate abolished general acid catalysis.

These kinetic observations suggest that a second, or alternate, general acid functions in the D69N variant. This was confirmed through kinetic isotope effects (KIEs), where both the ¹⁵(V/K) and the ¹⁸(V/K)_bridge KIEs were not significantly different from WT SsoPTP (Table 3). The ¹⁵(V/K) KIE directly reflects the proficiency of general acid catalysis.²⁸ This isotope effect results from the delocalization of negative charge into the aryl ring in the transition state. Catalysis in native PTPs shows near-unity ¹⁵(V/K) KIEs because general acid catalysis maintains a neutral leaving group as the P-O bond breaks. Loss of general acid catalysis by mutation results in ¹⁵(V/K) KIEs in the range of 1.0024–1.0030.²⁸ The ¹⁵(V/K) KIE for native SsoPTP and the D69N mutant are the same within experimental error, indicating general acid catalysis remains. The ¹⁸(V/K)_bridge in the scissile oxygen position has two contributions in the transition state: a normal effect from P-O bond fission, and an inverse effect from O-H bond formation; in native PTPs this KIE ranges from 1.0118–1.0152. This KIE becomes larger in PTP variants with disabled general acid catalysis due to the loss of the inverse effect from protonation, and ranges from 1.0275 to 1.0297.²⁸

Both KIEs for D69N SsoPTP are similar to those of the native enzyme. This, and the retention of a basic limb on the pH-rate profile, suggest that SsoPTP has a second residue that can act as an alternate general acid in the mutant. While unusual, this phenomenon has been previously observed in Tk-PTP and VHZ. In those enzymes, a glutamic acid (E134 in VHZ ¹⁴ and E132 in TkPTP¹²) performs this role, and both are in a similar structural location (Figure 12). However, SsoPTP has a glutamine residue, Q135, in the corresponding structural position. Residue E138 in the crystal structure of SsoPTP was identified as potentially residing close enough to the active site that, given the conformational mobility revealed in the NMR data, might act as the alternate general acid. With this in mind, the E138Q D69N double mutant of was tested for complete abolishment of general acid catalysis. However, the double mutant exhibited similar kinetics as the single D69N mutant (data not shown).

Figure 12.

(A). VHZ (yellow)¹⁵, Tk-PTP (purple)¹², and SsoPTP (green)¹³ in complex with metavanadate, vanadate, and vanadate, respectively. (B) Sequence alignment of VHZ, Tk-PTP, and SsoPTP. Structures were obtained from PDB IDs 4ERC, 5Z5A, and 2I6I, respectively. Highlighting in the first column shows the glutamate residue found to function as an alternative general acid in SsoPTP and the corresponding residues in the other two enzymes. The aspartic acids highlighted in the second column are the conserved general acid residues common to all PTPs and found in the same active site locations. In the third column, the nucleophilic cysteine is highlighted in all three proteins. Residues highlighted in yellow in the 4th column show the glutamates that can act as the alternate general acids in VHZ and Tk-PTP and the glutamine in the corresponding location in SsoPTP.

The next nearest acidic residue found near the active site is E40; however, E40 is found in a salt bridge with Arg 102 (Figure 4). This same interaction is seen in the new structures reported here, all of which indicate the deprotonated state of E40, one in which it would not be feasible to act as a general acid. Our crystals and all published SsoPTP structures were obtained at pH >7. In contrast, in the acidic pH range where SsoPTP is most active, E40 will be mostly in the protonated state. Attempts to grow crystals at lower pH, or to buffer exchange crystals to more acidic conditions, were unsuccessful. A glutamate residue is commonly found in this position in the PTP crystal structures, sometimes in multiple conformations or disordered.⁵⁸

A notable exception of the usual orientation of the conserved Arg residue in the P-loop is seen in the X-ray structure of pNPP bound to the C96S mutant of SsoPTP (Figure S1). The orientation of R102 in this structure raises the possibility that it could function as the secondary general acid, despite the high solution pK_a of 13.8 for arginine.⁵⁹ Although there are enzymes in which arginine functions in such a role,^60–64 in those reactions the leaving group being protonated is an alkoxide, much more basic than the tyrosine phenoxide leaving group in the PTP-catalyzed reaction. Additionally, those arginines are found in interactions with other positively charged residues or in a hydrophobic environment, both of which would increase the acidity of the guanidinium group. The interactions of R102 in the SsoPTP-substrate complex are with anions (the substrate and the carboxylate ion of E40) and would have the opposite effect.

To investigate the kinetic importance of E40, the D69N E40Q double mutant was prepared and found to exhibit the substantially reduced activity and loss of pH dependency expected from complete loss of general acid catalysis. For E40 to carry out this role, the salt bridge observed in the SsoPTP crystal structures must be disrupted and E40 be protonated. Notably, all SsoPTP crystal structures to date have been obtained at pH values well above the range where the enzyme is optimally active. The pH optimum of the D69N mutant is slightly lower than native SsoPTP (Figure 11). Under these conditions, E40 will be primarily in the protonated state, which may weaken its interaction with R102. Furthermore, the structural alterations documented by NMR could both disrupt this interaction and change the positioning of E40 relative to the scissile P-O bond of the substrate. The possibility that E40 acts in a cooperative role with R102 for it to act as the proton donor cannot be ruled out, but for reasons summarized above, we feel that R102 acting alone as the alternate general acid is unlikely. The kinetic data show that E40 is necessary for alternate general acid catalysis to occur in the D69N variant.

CONCLUSIONS

The SsoPTP enzyme has the active site architecture common to the PTP family, including the positioning of a general acid (D69) on a mobile loop. Kinetic and isotope effect data show that some other residue protonates the leaving group in the first step of the reaction in the D69N mutant, an unusual but not unprecedented finding noted previously in two other PTPs. Kinetic data from double mutants with the D69N mutation combined with conservative E to Q mutations are most consistent with E40 in this role. The position of E40 in crystal structures suggest it either acts directly as a general acid in a conformation not yet captured in a crystal structure, or, in cooperation with R102.

The protein dynamics in SsoPTP revealed by NMR expand our view of protein motions in the PTP family. Our data provides the first description Q-loop motions in PTPs. Additionally, motions are observed in the P-loop and the more well-known acid loop. Furthermore, the coordinated motions of all three functional loops in SsoPTP is the first such finding in the PTP family. These results, together with the kinetic data indicating the function of an alternate general acid taking on the role when D69 is mutated, suggest that SsoPTP adopts catalytically competent conformations that have not been observed in crystal structures, all of which have been crystallized at a pH that is significantly higher than the optimal pH. SsoPTP is the second PTP from a thermophile demonstrating conformational flexibility, after TkPTP, which shows a nonfunctional and a functional orientation of the P-loop that were identified crystallographically. The NMR results with SsoPTP demonstrate the presence of significantly more extensive, and coordinated, flexibility among key protein loops involved in catalysis.

The finding of an alternate general acid in SsoPTP is the third occurrence of this phenomenon in the PTP family, with two of the three cases (SsoPTP and TkPTP) in archaeal hyperthermophiles. The evolutionary origin and possible role of such a feature remains to be determined. The thermal energy available at high temperatures enables enhanced sampling of conformational space, which may facilitate the functioning of two alternate general acids in different active site geometries. This could be useful in a hyperthermophile in which conformational flexibility of the active site results in some conformations in which the traditional general acid is not always in a catalytically productive position. The presence of two potential general acids would increase the number of conformations in which the catalytically important contribution from general acid catalysis can be attained. A glutamic acid interacting with the P-loop arginine side chain, analogous to E40 in SsoPTP, is found in other PTPs although this is the first instance where data support a catalytic role. It may be that during the evolution of PTPs in thermophiles this glutamic acid residue had a catalytic role that was later lost, and only a structural role remained in positioning the guanidinium side chain of the conserved arginine to better aid transition state stabilization.