Enzymatic Catalysis of Proton Transfer at Carbon

Abstract

More than 80% of the rate acceleration for enzymatic catalysis of the aldose-ketose isomerization of (R)-glyceraldehyde 3-phosphate (GAP) by triosephosphate isomerase (TIM) can be attributed to the 3′-phosphodianion group of GAP. We examine here the necessity of the covalent connection between the phosphodianion and triose sugar portions of the substrate by “carving up” GAP into the minimal neutral two-carbon sugar glycolaldehyde and phosphite dianion pieces. This “two-part substrate” preserves both the α-hydroxycarbonyl and oxydianion portions of GAP. TIM catalyzes proton transfer from glycolaldehyde in D₂O, resulting in deuterium incorporation that can be monitored by ¹H NMR spectroscopy, with k_cat/K_m = 0.26 M^-1 s^-1. Exogenous phosphite dianion results in a very large increase in the observed second-order rate constant (k_cat/K_m)_obsd for turnover of glycolaldehyde, and the dependence of (k_cat/K_m)_obsd on [HPO₃^2-] exhibits saturation. The data give k_cat/K_m = 185 M^-1s^-1 for turnover of glycolaldehyde by TIM that is saturated with phosphite dianion, so that the separate binding of phosphite dianion to TIM results in a 700-fold acceleration of proton transfer from carbon. The binding of phosphite dianion to the free enzyme (K_d = 38 mM) is 700-fold weaker than its binding to the fleeting complex of TIM with the altered substrate in the transition state (K_d^‡ = 53 μM); the total intrinsic binding energy of phosphite dianion in the transition state is 5.8 kcal/mol. We propose a physical model for catalysis by TIM in which the intrinsic binding energy of the substrate phosphodianion group is utilized to drive closing of the “mobile loop” and a protein conformational change that leads to formation of an active site environment that is optimally organized for stabilization of the transition state for proton transfer from α-carbonyl carbon.

Despite the wealth of mechanistic and structural data available for enzyme catalysts of proton transfer at α-carbonyl carbon (1-4), the origins of the rate accelerations effected by these enzymes remain elusive. The principal burden for such enzymes is the very large thermodynamic barrier to the formation of simple enolates in aqueous solution (5-7), but the physical mechanism(s) by which they lower this barrier and the nature of the corresponding transition state stabilization remain topics of current interest.

Triosephosphate isomerase (TIM)¹ is the prototypical protein catalyst of proton transfer at α-carbonyl carbon and catalyzes the reversible stereospecific 1,2-shift of the pro-R proton at dihydroxyacetone phosphate (DHAP) to give (R)-glyceraldehyde 3-phosphate (GAP) by a single-base (Glu-165) mechanism through a cis-enediol(ate) intermediate (Scheme 1) (8). The early extensive studies of TIM provided a clear description of the chemical events that occur at the enzyme active site (9, 10), and they show that the enzyme approaches “perfection” in its catalysis of the isomerization of triose phosphates (11).

Scheme 1.

TIM presents a unique opportunity for detailed study of the origin of the enzymatic rate acceleration for proton transfer at carbon for several reasons: (1) TIM catalyzes the aldose-ketose isomerization of small and chemically simple phosphodianion substrates, and it requires no co-substrates, metal ions or other cofactors (8-10). (2) The mechanism (9, 10) and free energy profile (11) have been well-defined in extensive chemical (12-14), spectroscopic (15-18), kinetic (19-23) and isotope exchange (8, 11, 24-27) studies. (3) The overall canonical (β,α)₈-barrel structure of the protein in which the active site lies at the C-terminal end of the barrel (28-30), the identity of the catalytic residues (principally the base Glu-165 and the electrophiles His-95 and Lys-12 (9, 10)),² and the architecture of the active site have been characterized in extensive mutagenesis (18, 31-33) and crystallographic (34-44) studies. (4) The importance of the 11-residue “mobile loop” (loop 6) has been explored via crystallographic (36-38, 40-42, 44, 45), mutagenesis (46-51), computational (52-56) and dynamical approaches (52, 57-59).

The binding of phosphodianion ligands to TIM results in a large ca. 7 Å motion of the mobile loop which acts as a “lid” to cover the bound ligand in the active site (34, 36-38, 40-42, 44, 56, 57, 59). The closing of this loop leads to formation of an important hydrogen bond between the backbone amide NH group of Gly-171 at the tip of the mobile loop and a peripheral oxygen of the bound phosphodianion group (36, 38-40, 42, 44, 45). The anionic oxygens of the phosphodianion group also interact with the protein via direct hydrogen bonds with backbone amide NH groups in the phosphate binding motif located in loops 7 and 8, and via a water molecule with the essential (31) positively charged Lys-12 in the active site (37, 40, 42, 44, 60). Loop closure results in the exclusion of bulk solvent from the active site (36-38, 40, 42) and serves to protect the reactive enediolate intermediate from bulk solution where it undergoes rapid nonenzymatic elimination to give methylglyoxal (46, 61). Despite these well-documented conserved interactions of the protein with the phosphodianion group of bound ligands, there is no consensus about the physical mechanism by which they may be harnessed to stabilize the transition state for proton transfer.

We showed previously (62) that deprotonation of the neutral triose sugar (R)-glyceraldehyde by TIM is ca. 10⁸-fold slower than the partly diffusion-controlled (11) turnover of the natural phosphorylated substrate GAP, and that more than 80% of the 4 × 10¹⁰-fold enzymatic rate acceleration of proton transfer (61, 62) can be attributed to the presence of the small nonreacting phosphodianion group of the substrate.

Utilization of the binding energy of the substrate phosphodianion group for stabilization of the ground state Michaelis complex with bound substrate is important, but uninteresting with respect to a rationalization of the enzymatic rate acceleration for TIM. Our earlier data demand that the intrinsic binding energy of the phosphodianion group of GAP be utilized specifically for stabilization of the transition state for proton transfer from carbon. However, this conclusion has not gained wide acceptance, perhaps as a result of the complexity of the experimental evidence that supports it. The goal of the current work was to provide simple and more compelling evidence that a portion of the intrinsic binding energy of the substrate phosphodianion group is expressed only on moving from the Michaelis complex to the transition state for deprotonation of GAP by TIM. The demonstration that the interactions of TIM with the phosphodianion group are “catalytic” in the absence of a covalent connection between the phosphodianion and carbon acid portions of GAP would be unequivocal evidence that the oxydianion is not simply an “anchor” to fix the carbon acid at the active site. This would correspond to an “allosteric” activation of TIM by the separately bound phosphodianion moiety of a truncated substrate.

We report here that the slow deprotonation of the minimal two-carbon neutral substrate glycolaldehyde (h-GLY) by TIM in D₂O, which results in formation of monodeuteriated glycolaldehyde (d-GLY), can be monitored by ¹H NMR spectroscopy (Scheme 2). The separate binding of exogenous phosphite dianion to TIM results in a very large 700-fold increase in the second-order rate constant for turnover of h-GLY. The results strongly support a physical model for catalysis by TIM in which the intrinsic binding energy of the substrate phosphodianion group is utilized to drive closing of the mobile loop and a protein conformational change that leads to formation of an active site environment that is optimally organized for stabilization of the transition state. They suggest that a substantial portion of the rate acceleration for TIM results directly from utilization of the intrinsic binding energy of the remote nonreacting substrate phosphodianion group to lower the kinetic barrier to proton transfer from α-carbonyl carbon.

Scheme 2.

MATERIALS AND METHODS

Triosephosphate isomerase from rabbit muscle (lyophilized powder) was from Sigma and had a specific activity towards GAP of 5500 - 6000 units/mg at pH 7.5 and 25 °C. Glycerol 3-phosphate dehydrogenase (GPDH) from rabbit muscle was from Roche or MP Biomedicals. Bovine serum albumin (Fraction V, BSA) was from Life Technologies. Glycolaldehyde dimer, triethanolamine hydrochloride, imidazole, sodium phosphate (tribasic) and tetramethylammonium hydrogensulfate were from Aldrich. NADH (disodium salt), D,L-glyceraldehyde 3-phosphate diethyl acetal (barium salt), phosphoglycolic acid (tri(monocyclohexylammonium) salt) and Dowex 50WX4-200R were from Sigma. Sodium phosphite (dibasic, pentahydrate) was from Riedel-de Haën (Fluka). Deuterium oxide (99.9% D) and deuterium chloride (35% w/w, 99.9% D) were from Cambridge Isotope Laboratories. Imidazole was recrystallized from benzene. All other chemicals were reagent grade or better and were used without further purification.

Preparation of Solutions

Stock solutions of D,L-glyceraldehyde 3-phosphate (D,L-GAP) at pH 7.4 or 7.5 were prepared by hydrolysis of the diethyl acetal using Dowex 50WX4-200R (H⁺ form) followed by neutralization with 1 M NaOH. The concentration of GAP in these solutions was determined by coupled enzymatic assay, as described previously (25).

Triethanolamine buffers (pH 7.5) were prepared by neutralization of the hydrochloride with 1 M NaOH. Buffered solutions of imidazole (50 mM, 70% free base, pH 7.4), phosphite (70 mM, 90% free base) and phosphate (60 mM, 85% free base) for use in competitive inhibition studies were prepared by addition of measured amounts of 1 M HCl to the basic form or the sodium salt.

Buffered solutions of imidazole (30 mM, 20% free base) at pD 7.0 were prepared by dissolving the basic form, and where appropriate NaCl, in D₂O followed by addition of 0.8 eqs of DCl to give the required acid/base ratio at I = 0.024 or 0.10.

Before preparation of solutions in D₂O, Na₂HPO₃·5H₂O was dried under vacuum at 25 °C for 16 h. Gravimetric analysis and titration showed that this resulted in the loss of 4.6 equivalents of H₂O, which corresponds to ca. 40% of the initial weight. The resulting salt of formula Na₂HPO₃·0.4H₂O was stored in a dessicator. Stock solutions of phosphite (40 or 80 mM, I = 0.10 or 0.20, respectively) at pD 7.0 were prepared by dissolving Na₂HPO₃·0.4H₂O in D₂O followed by addition of 0.5 eqs of DCl to give 1:1 dianion:monoanion.

A stock solution of phosphate (60 mM, I = 0.21) at pD 7.1 was prepared by dissolving sodium phosphate trianion in D₂O followed by addition of 1.5 eqs of DCl to give 1:1 dianion:monoanion.

Sodium phosphoglycolate tri(monocyclohexyl)ammonium salt in D₂O at pD 6.9 (100 mM, 1:1 trianion:dianion, prepared by addition of 0.5 eqs of DCl) was freed of monocyclohexylamine by passage down a short column of Dowex 50WX4-200R (D⁺ form) that had been equilibrated with D₂O. The acidic fractions were combined and the concentration of 2-phosphoglycolic acid in the resulting solution (41 mM) was determined by ¹H NMR by quantitative integration of the signal due to the methylene group of 2-phosphoglycolate relative to that for the C(4,5)-protons of imidazole buffer that was added as an internal standard. The acidic solution of 2-phosphoglycolic acid (pD ≈ 1.4) was neutralized by the addition of 3 eqs of concentrated NaOD to give a stock solution of 39 mM 2-phosphoglycolate (PGA) at pD 6.9 (ca. 1:1 trianion:dianion, I ≈ 0.2, NaCl).

Stock solutions of glycolaldehyde (100 mM) in D₂O at I = 0.10 or zero were prepared by dissolving glycolaldehyde dimer, and where appropriate NaCl, in D₂O and were stored at room temperature in order to minimize the dimer content. ¹H NMR analysis showed that there was negligible dimeric glycolaldehyde present after a five-fold dilution of this stock solution into 0.1 M NaCl in D₂O (63, 64).

Commercial rabbit muscle TIM (10 - 16 mg/mL protein) was dialyzed at 4 °C against 30 mM imidazole buffer (20% free base) in D₂O at pD 7.0 and I = 0.024 for reactions in the presence of added phosphite or phosphate dianions, or I = 0.10 (NaCl) for reactions in their absence.

Determination of pL and the pK_a of Phosphite and Phosphate Dianions in D₂O

Solution pH or pD was determined at 25 °C using an Orion Model 720A pH meter equipped with a Radiometer pHC4006-9 combination electrode that was standardized at pH 4.00 and 7.00 at 25 °C. Values of pD were obtained by adding 0.40 to the observed reading of the pH meter (65). An apparent value of pK_BD = 6.99 for ionization of phosphite monoanion in D₂O at 25 °C and I = 0.10 (NaCl) was determined as the pD of a solution of 40 mM sodium phosphite 1:1 monoanion:dianion. An apparent value of pK_BD = 7.22 for ionization of phosphate monoanion in D₂O at 25 °C and I = 0.10 (NaCl) was determined as the pD of solution of 30 mM sodium phosphate 1:1 monoanion:dianion.

Enzyme Assays

Enzyme assays were carried out at 25 °C. Changes in the concentration of NADH were calculated from the changes in absorbance at 340 nm using an extinction coefficient of 6220 M^-1 cm^-1. GPDH was dialyzed at 4 °C against 20 mM triethanolamine buffer (pH 7.5) and was assayed by monitoring the oxidation of NADH by DHAP, as described previously (25).

The activity of TIM in the reaction mixtures in D₂O that were monitored by ¹H NMR (10 - 80 units/μL) was determined in standard assays in which the isomerization of GAP was coupled to the oxidation of NADH using GPDH (20). Assay mixtures (1.0 mL) contained 100 mM triethanolamine (pH 7.5, I = 0.1), 0.2 mM NADH, ca. 6 mM D,L-GAP (3 mM GAP, ∼ 7K_m), 0.3 - 0.5 units of GPDH (3 - 5 μg) and 5 μL of the reaction mixture after its 10,000- or 50,000-fold dilution into 20 mM triethanolamine (pH 7.5). Before addition of the reaction mixture containing TIM, the background velocity due mainly to the isomerization of GAP catalyzed by TIM present as a contaminant in the GPDH, v_o, was determined over 2 - 4 min. This background velocity generally represented ≤ 3% of the total initial velocity, v_i, that was determined over 5 - 10 min after the addition of the diluted reaction mixture containing TIM. Values of V_max (M s^-1) were calculated from the values of (v_i - v_o) using K_m = 0.45 mM for GAP (20, 62). The concentration of TIM in the assay mixture was calculated from the values of V_max (M s^-1) using the relationship [TIM]_assay = V_max/k_cat with k_cat = 4300 s^-1(19, 66).

Values of K_i for competitive inhibition of TIM by phosphite and phosphate dianions in H₂O at pH 7.4 and I = 0.12 (NaCl) were determined from the effect of added 35 mM phosphite buffer (90% free base, 31.5 mM dianion) or 27 mM phosphate buffer (85% free base, 23 mM dianion) on the apparent value of K_m for GAP determined in 10 mM imidazole buffer (70% free base) at pH 7.4, 25 °C and I = 0.12 (NaCl). For these studies the stock solution of TIM (ca. 0.2 μg/mL) was stabilized by the inclusion of BSA (100 μg/mL) and the amount of GPDH in the coupled assays was increased to 1 - 2 units (10 - 20 μg).

Background Reactions of Glycolaldehyde in D₂O Buffered by Imidazole, Phosphite and Phosphate Followed by ¹H NMR

The reactions of glycolaldehyde (19 mM) in D₂O buffered by 24 mM imidazole (20% free base, pD 7.0), 32 mM phosphite (50% free base, pD 7.0) or 30 mM phosphate (50% free base, pD 7.2) at 25 °C and I = 0.10 (NaCl) were followed by ¹H NMR spectroscopy. Two identical reaction mixtures (1 mL) were prepared by addition of an aliquot of 100 mM glycolaldehyde in D₂O (I = 0.10, NaCl) to the buffer. For the reaction in phosphite buffer the reaction mixture also contained ca. 1 mM tetramethylammonium hydrogensulfate as an internal standard. For the reaction in phosphate buffer the reaction mixture also contained 6 mM imidazole buffer (pD 7.0). 750 μL of the first reaction mixture was transferred to an NMR tube and the NMR spectrum was recorded within 1 hour, after which the NMR tube was incubated at 25 °C. The second reaction mixture was incubated at 25 °C and was used to monitor any change in pD during these slow reactions. The reaction in the presence of imidazole was followed for 35 days (19% disappearance of glycolaldehyde, ca. 0.2 unit decrease in pD). The reaction in the presence of phosphite was followed for 39 days (48% reaction, ca. 0.4 unit decrease in pD). The reaction in the presence of phosphate was followed for 13 days (40% reaction, 0.1 unit decrease in pD).

Turnover of Glycolaldehyde by TIM in D₂O in the Absence and Presence of 2-Phosphoglycolate Followed by ¹H NMR

The turnover of glycolaldehyde (19 mM) by TIM in D₂O in the presence of 24 mM imidazole buffer (20% free base, pD 7.0) at 25 °C and I = 0.10 (NaCl) was followed by ¹H NMR spectroscopy. The reaction in a volume of 850 μL was initiated by the addition of an aliquot (170 μL) of 100 mM glycolaldehyde in D₂O (I = 0.10, NaCl) to a solution of TIM (ca. 100 units/μL, 10 mg/mL) in 30 mM imidazole buffer at pD 7.0 in D₂O (I = 0.10, NaCl). 780 μL of the reaction mixture was transferred to an NMR tube and the NMR spectrum was recorded immediately, after which the NMR tube was incubated at 25 °C. The remainder was incubated at 25 °C and the activity of TIM in the reaction mixture was determined by periodic standard assay (see above). The reaction was followed for 38 hours during which time there was 68% reaction of glycolaldehyde and no significant decrease in enzyme activity (± 10%). The solution was then removed from the NMR tube and 163 μL of a solution of PGA at pD 7.0 in D₂O (39 mM, I = 0.20, prepared as described above) was added to 650 μL of the recovered reaction mixture. This gave a new reaction mixture containing ca. 5 mM unreacted glycolaldehyde, 7.8 mM PGA and 19 mM imidazole in D₂O (I = 0.12). 750 μL of this solution was transferred to an NMR tube and the NMR spectrum was recorded immediately, after which the NMR tube was incubated at 25 °C. The remainder was incubated at 25 °C and the activity of TIM was determined by periodic standard assay (see above). The reaction in the presence of PGA was followed for 5 days during which time there was 68% reaction of the remaining glycolaldehyde and no significant decrease in enzyme activity (± 10%). After 7 days the solution was removed from the NMR tube, the protein was removed by ultrafiltration using an Amicon Microcon device (10,000 MWCO), and the pD of the filtrate was determined. This showed that there was only a small 0.15 unit decrease in the pD of the reaction mixture during the total reaction time of 9 days.

Turnover of Glycolaldehyde by TIM in D₂O in the Presence of Phosphite and Phosphate Dianions Followed by ¹H NMR

The turnover of glycolaldehyde by TIM in D₂O in the presence of 4.9 - 40 mM phosphite buffer (50% dianion, pD 7.0) or 30 mM phosphate buffer (50% dianion, pD 7.2) at 25 °C and I = 0.10 (NaCl) was followed by ¹H NMR spectroscopy. Reactions in a volume of 750 μL were initiated by addition of 150 μL of a solution of TIM (ca. 60 units/μL, 6 mg/mL) in 30 mM imidazole buffer at pD 7.0 in D₂O (I = 0.024) to the reaction mixture containing glycolaldehyde, phosphite or phosphate buffer, and NaCl in D₂O to give final concentrations of 19 or 9.5 mM glycolaldehyde, 6 mM imidazole and 2.5 - 20 mM inorganic oxydianion at I = 0.10 (NaCl). 700 μL of the reaction mixture was transferred to an NMR tube and the NMR spectrum was recorded immediately. The remainder was incubated at 25 °C and the activity of TIM was determined by periodic standard assay (see above). For the reactions in the presence of phosphite, the NMR tube was left in the probe at 25 °C and spectra (8 -16 transients) were recorded at 12 - 24 min intervals for up to 13 hours. The reaction in the presence of phosphate was followed for 6 days. In several cases the solution was removed from the NMR tube at the end of the reaction, the protein was removed by ultrafiltration using an Amicon Microcon device (10,000 MWCO), and the pD of the filtrate was determined. This showed that there was no significant change in pD during the turnover of glycolaldehyde by TIM in these experiments.

¹H NMR Analyses

¹H NMR spectra at 500 MHz were recorded in D₂O at 25 °C using a Varian Unity Inova 500 spectrometer that was shimmed to give a line width of ≤ 0.4 Hz for the downfield peak of the triplet due to the C-1 proton of glycolaldehyde hydrate. The probe temperature was determined from the difference in the chemical shifts of the hydroxyl and methyl protons of methanol containing a trace of HCl (67). Spectra (8 - 32 transients) were obtained using a sweep width of 6000 Hz, a pulse angle of 90° and acquisition times of 6 - 9 s, with zero-filling of the data to 128 K. In order to ensure accurate integrals for the protons of interest, a relaxation delay between pulses of 90 - 120 s (> 8T₁) was used (see below). Baselines were subjected to a first-order drift correction before determination of integrated peak areas. Chemical shifts are reported relative to HOD at 4.67 ppm.

The following relaxation times T₁ were determined in D₂O at 25 °C and I = 0.10 (NaCl): (a) The protons of glycolaldehyde hydrate (19 mM) in 24 mM imidazole buffer at pD 7.0: T₁ = 11 s (C1-proton), 3 s (C2-protons). (b) The protons of the free carbonyl form of glycolaldehyde (19 mM) in 32 mM phosphite buffer at pD 7.0: T₁ = 14 s (C1-proton), 6 s (C2-protons). (c) The C(4,5)-protons of imidazole (24 mM) at pD 7.0: T₁ = 15 s. (d) The P-H proton of phosphite (32 mM) at pD 7.0: T₁ = 5 s.

Integrated peak areas were normalized to that of an appropriate internal standard. For the background reactions in imidazole buffer and in phosphate buffer containing imidazole the internal standard was the signal at 7.3 ppm due to the C(4,5)-protons of imidazole. For the background reaction in phosphite buffer the internal standard was the triplet at 3.1 ppm due to the methyl groups of added tetramethylammonium hydrogensulfate. For the TIM-catalyzed reactions in the presence of phosphite the internal standard was the upfield peak of the doublet at 6.7 ppm (J = 600 Hz) due to the P-H proton of phosphite. For the TIM-catalyzed reactions in the absence of phosphite the internal standard was the signal at 7.3 ppm due to the C(4,5)-protons of imidazole.

The integrated area of the signal due to the C1-proton of h-GLY hydrate was evaluated using eq 1, where a is the area of the downfield peak of the triplet due to the C1-proton of h-GLY hydrate. The integrated area of the signal due to the C1-proton of d-GLY hydrate was evaluated using eq 2, where b is the sum of the areas of the downfield peak of the doublet due to the C1-proton of d-GLY hydrate and the center peak of the triplet due to the C1-proton of h-GLY hydrate (see Figure 1).

Figure 1.

Partial ¹H NMR spectra at 500 MHz obtained at various times during the reaction of 19 mM h-GLY in the presence of 0.31 mM (8.3 mg/mL) rabbit muscle TIM and 24 mM imidazole buffer (pD 7.0) in D₂O at 25 °C (I = 0.10). The extent of reaction of h-GLY is given above each spectrum. A. The turnover of h-GLY results in disappearance of the triplet at 4.945 ppm (J = 5 Hz) due to the C1-proton of h-GLY hydrate and the appearance of a broad upfield-shifted doublet at 4.942 ppm (Δδ = 0.003 ppm, J = 5 Hz) due to the C1-proton of the hydrate of glycolaldehyde labeled with deuterium at C2 (d-GLY). At late reaction times the upfield peak of this doublet appears larger than the downfield peak as a result of the formation of the hydrate of glycolaldehyde labeled with two deuteriums at C2, which exhibits a broad singlet at 4.936 ppm. B. The turnover of h-GLY results in disappearance of the doublet at 3.409 ppm (J = 5 Hz) due to the two C2-protons of h-GLY hydrate and the appearance of an upfield-shifted double triplet at 3.389 ppm (Δδ = 0.02 ppm) due to the single C2-proton of d-GLY hydrate. This proton is coupled to both the C1-proton (J = 5 Hz) and the geminal deuterium (J = 1.6 Hz).

$$A_{h-\mathrm{GLY}}=4a$$ (1)

$$A_{d-\mathrm{GLY}}=2b-4a$$ (2)

The fractions of h-GLY and d-GLY present at time t were calculated according to eqs 3 and 4 where (A_h-GLY)_o and (A_d-GLY)_o are the integrated areas of the signals due to the C1-protons of the hydrates of h-GLY and d-GLY, respectively, obtained from the first NMR spectrum, for which t was defined to be zero. This analysis neglects the small amounts (≤ 3%) of dideuteriated glycolaldehyde that is already formed at zero time in the relatively fast TIM-catalyzed reactions in the presence of phosphite dianion. The total concentration of glycolaldehyde present in the reaction mixture was obtained by comparison of the value of {(A_h-GLY)_o + (A_d-GLY)_o} with the integrated area of the signal due to the C(4,5)-protons of imidazole, or that of the upfield peak of the doublet due to the P-H proton of phosphite, with a correction for the presence of 6.1% of the free carbonyl form of glycolaldehyde. For the slow background reactions in the absence of TIM there was no detectable formation of d-GLY at zero time so that (A_d-GLY)_o = 0.

$$f_{h-\mathrm{GLY}}=\frac{A_{h-\mathrm{GLY}}}{{\left(A_{h-\mathrm{GLY}}\right)}_{\mathrm{o}}+{\left(A_{d-\mathrm{GLY}}\right)}_{\mathrm{o}}}$$ (3

$$f_{d-\mathrm{GLY}}=\frac{A_{d-\mathrm{GLY}}}{{\left(A_{h-\mathrm{GLY}}\right)}_{\mathrm{o}}+{\left(A_{d-\mathrm{GLY}}\right)}_{\mathrm{o}}}$$ (4

Observed first-order rate constants k_obsd (s^-1) for the disappearance of h-GLY were determined from the slopes of linear semi-logarithmic plots of reaction progress against time according to eq 5, where (f_h-GLY)_o is the fraction of h-GLY present in the reaction mixture at time zero. For the slow background reactions in buffered D₂O, (f_h-GLY)_o = 1.0 and these plots covered up to ca. 50% disappearance of the starting h-GLY. For the TIM-catalyzed reactions, (f_h-GLY)_o ≥ 0.8 and these plots covered disappearance of 60 - 80% of the starting h-GLY.

$$\ln \{f_{h-\mathrm{GLY}}∕{\left(f_{h-\mathrm{GLY}}\right)}_{\mathrm{o}}\}=-k_{\mathrm{obsd}}t$$ (5)

RESULTS

Quantitative integration of the triplet at 4.95 ppm due to the C1-proton of glycolaldehyde hydrate and the singlet at 4.34 ppm due to the C2-protons of the free carbonyl form of glycolaldehyde in the ¹H NMR spectrum of 19 mM glycolaldehyde in 32 mM phosphite buffer (50% free base, pD 7.0) showed that glycolaldehyde exists as 93.9% hydrate and 6.1% free carbonyl form in D₂O at 25 °C and I = 0.10 (NaCl). This gives K_hyd = 15.4 as the equilibrium constant for hydration of the carbonyl group of glycolaldehyde under these conditions (Scheme 3). This is similar to the values of K_hyd ≈ 17.5 calculated from NMR data for 0.1 M glycolaldehyde in D₂O at 35 °C (63), and K_hyd = 18.0 ± 3.3 in D₂O at 25 °C determined by NMR (64). In the presence of imidazole buffer (≥ 6 mM) at pD 7.0 and 25 °C the ¹H NMR signals due to the free carbonyl form of glycolaldehyde are severely exchange-broadened or not observed, as a result of formation of the adduct of imidazole to the carbonyl group. The ¹H NMR spectrum of 19 mM glycolaldehyde in 0.8 M imidazole buffer (70% free base, pD 7.9) at 25 °C and I = 0.3 (NaCl) exhibits signals due to glycolaldehyde hydrate at 4.86 ppm (triplet, C1H) and 3.32 ppm (doublet, C2H), along with signals assigned to an approximately equal amount of the corresponding imidazole adduct at ca. 5.7 ppm (very broad, C1H) and 3.71 ppm (broad, C2H) (data not shown).

Scheme 3.

Figure 1 shows partial ¹H NMR spectra at 500 MHz obtained at various times during the reaction of 19 mM glycolaldehyde (h-GLY) in the presence of 0.31 mM (8.3 mg/mL) rabbit muscle triosephosphate isomerase (TIM) and 24 mM imidazole buffer (pD 7.0) in D₂O at 25 °C and I = 0.10 (NaCl). The slow turnover of h-GLY by TIM results in disappearance of the triplet at 4.945 ppm (J = 5 Hz) due to the C1-proton of h-GLY hydrate and the appearance of a broad upfield-shifted doublet at 4.942 ppm (Δδ = 0.003 ppm, J = 5 Hz) due to the C1-proton of the hydrate of glycolaldehyde labeled with deuterium at C2 (d-GLY, Scheme 3). At late reaction times the upfield peak of this doublet appears larger than the downfield peak as a result of the formation of glycolaldehyde hydrate labeled with two deuteriums at C2, which exhibits a broad singlet at 4.936 ppm. Similarly, disappearance of the doublet at 3.409 ppm (J = 5 Hz) due to the two C2-protons of h-GLY hydrate is accompanied by the appearance of an upfield-shifted double triplet at 3.389 ppm (Δδ = 0.02 ppm) due to the single C2-proton of the first-formed d-GLY hydrate product, which is coupled to both the C1-proton (J = 5 Hz) and the geminal deuterium (J = 1.6 Hz) (Scheme 3). The magnitude of these deuterium isotope effects on ¹H chemical shifts and the geminal H-D coupling constant are similar to those observed in our earlier work for deuterium incorporation into GAP (25), DHAP (25), dihydroxyacetone (68, 69) and hydroxyacetone (70).

Figure 2 (●) shows the semi-logarithmic timecourse for the disappearance of 68% of 19 mM h-GLY in the presence of 0.31 mM (8.3 mg/mL) rabbit muscle TIM and 24 mM imidazole buffer (pD 7.0) in D₂O at 25 °C and I = 0.10 (NaCl), followed by ¹H NMR spectroscopy by monitoring the decrease in the area of the downfield peak of the triplet due to the C1-proton of h-GLY hydrate (Figure 1A). The data give k_obsd = 8.2 × 10^-6 s^-1 as the first-order rate constant for disappearance of h-GLY, which is 120-fold larger than k_o = 7.0 × 10^-8 s^-1 determined for disappearance of h-GLY in the absence of TIM under the same conditions (see below). Therefore, essentially all of the reaction of h-GLY can be attributed to the presence of the protein catalyst with k_obsd = 8.1 × 10^-6 s^-1 for turnover of h-GLY by TIM in this experiment.

Figure 2.

Semi-logarithmic timecourses for the reaction of h-GLY in D₂O at pD 7.0 and 25 °C catalyzed by rabbit muscle TIM, followed by ¹H NMR spectroscopy by monitoring the decrease in the area of the downfield peak of the triplet due to the C1-proton of h-GLY hydrate. Key: (●)19 mM h-GLY in the presence of 0.31 mM TIM and 24 mM imidazole buffer (I = 0.10). (■) 5 mM h-GLY in the presence of 0.24 mM TIM, 19 mM imidazole buffer and 8 mM PGA (I = 0.12).

Figure 2 (■) shows the semi-logarithmic timecourse for the disappearance of 68% of 5 mM h-GLY in the presence of 0.24 mM (6.4 mg/mL) rabbit muscle TIM, 19 mM imidazole buffer (pD 7.0) and 8 mM of the potent competitive inhibitor 2-phosphoglycolate (PGA, K_i ≈ 5 μM at pH 7.0 and I = 0.1 (22)) in D₂O at 25 °C and I = 0.12 (NaCl), followed by ¹H NMR spectroscopy. The data give k_obsd = 2.7 × 10^-6 s^-1 as the first-order rate constant for disappearance of h-GLY, which is 22-fold larger than k_obsd ≈ 1.2 × 10^-7 s^-1 estimated³ for disappearance of h-GLY in the absence of TIM under the same conditions. This gives k_obsd = 2.6 × 10^-6 s^-1 for the nonspecific protein-catalyzed disappearance of h-GLY in the presence of saturating PGA in this experiment (62).

Observed first-order rate constants for the slow nonenzymatic background reactions of 19 mM h-GLY in buffered D₂O at 25 °C and I = 0.10 (NaCl) were determined by ¹H NMR spectroscopy by monitoring the decrease in the area of the downfield peak of the triplet due to the C-1 proton of h-GLY hydrate (Figure 1A). The data give k_o = 7.0 × 10^-8 s^-1 for reaction in the presence of 24 mM imidazole buffer (20% free base, pD 7.0) determined during disappearance of 19% h-GLY, k_o = 2.4 × 10^-7 s^-1 for reaction in the presence of 32 mM phosphite buffer (50% free base, pD 7.0) determined during disappearance of 48% of h-GLY, and k_o = 4.3 × 10^-7 s^-1 for reaction in the presence of 30 mM phosphate buffer (50% free base, pD 7.2) containing 6 mM imidazole determined during disappearance of 40% of h-GLY.

Figure 3 shows representative semi-logarithmic timecourses for the disappearance of 19 mM h-GLY catalyzed by 42 μM rabbit muscle TIM in the presence of various concentrations of phosphite dianion and 6 mM imidazole in D₂O at pD 7.0 and 25 °C (I = 0.10, NaCl). Table 1 summarizes the observed first-order rate constants k_obsd (s^-1) determined for the reaction of 19 or 9.5 mM h-GLY in experiments conducted in the presence of 2.5 - 20 mM phosphite dianion.

Figure 3.

Representative semi-logarithmic timecourses for the reaction of 19 mM h-GLY catalyzed by 42 μM rabbit muscle TIM in the presence of phosphite dianion and 6 mM imidazole in D₂O at pD 7.0 and 25 °C (I = 0.10), followed by ¹H NMR spectroscopy by monitoring the decrease in the area of the downfield peak of the triplet due to the C1-proton of h-GLY hydrate. Key: (●) [HPO₃^2-] = 5 mM; (■) [HPO₃^2-] = 10 mM; (▲) [HPO₃^2-] = 20 mM.

Table 1.

Rate constants for the turnover of glycolaldehyde (h-GLY) to give monodeuteriated glycolaldehyde (d-GLY) by 42 μM triosephosphate isomerase in the presence of various concentrations of phosphite dianion and 6 mM imidazole in D₂O at pD 7.0 and 25 °C (I = 0.10, NaCl)

[HPO32-] a	[GLY]totb	kobsdc	kTIMd	(kcat /Km)obsde
mM	mM	s-1	M-1 s-1	M-1 s-1
0	19		0.0158 f	0.26
2.5	19	3.60 × 10-5	0.853	14.0
5.0	19	6.08 × 10-5	1.44	23.6
10.0	19	9.52 × 10-5	2.26	37.0
10.0	9.5	9.98 × 10-5	2.36	38.7
15.0	19	1.32 × 10-4	3.13	51.3
20.0	19	1.64 × 10-4	3.89	63.8
20.0	9.5	1.70 × 10-4	4.03	66.1

^aConcentration of phosphite dianion added as phosphite buffer (50% free base, pD 7.0).

^bTotal concentration of glycolaldehyde.

^cObserved first-order rate constant for the disappearance of h-GLY, determined from the slope of a semi-logarithmic plot of reaction progress against time according to eq 5 (see Figure 3).

^dSecond-order rate constant for turnover of h-GLY by TIM to give d-GLY, calculated using the relationship k_TIM = k_obsd/[TIM] with [TIM]= 42 μM, unless noted otherwise. No corrections are made for the background reaction catalyzed by phosphite dianion (k_o = 2.4 × 10^-7 s^-1 in the presence of 32 mM phosphite buffer at pD 7.0) and the nonspecific TIM-catalyzed reaction that occurs outside the active site (k_non [TIM] = 4.5 × 10^-7 s^-1) because they contribute ≤ 1% to the overall reaction (see text).

^eSecond-order rate constant for turnover of the free carbonyl form of h-GLY by TIM, calculated using the relationship (k_cat/K_m)_obsd = k_TIM(1 + K_hyd) with K_hyd = 15.4.

^fSecond-order rate constant for turnover of glycolaldehyde at the active site of TIM determined in the presence of 24 mM imidazole buffer (pD 7.0).

A value of k_obsd = 3.3 × 10^-6 s^-1 was determined for the reaction of 19 mM h-GLY in the presence of 42 μM rabbit muscle TIM, 6 mM imidazole and 30 mM phosphate buffer (50% free base, pD 7.2). This is only 7.7-fold faster than k_o = 4.3 × 10^-7 s^-1 determined for reaction in the absence of TIM under the same conditions (see above). The net rate constant for turnover of h-GLY by 42 μM TIM in the presence of 15 mM phosphate dianion at pD 7.2 is therefore k_obsd = 2.9 × 10^-6 s^-1.

The Michaelis constant for turnover of GAP (0.07 - 3.6 mM) by rabbit muscle TIM in 10 mM imidazole buffer in H₂O at pH 7.4, 25 °C and I = 0.12 (NaCl) was determined as K_m = 0.50 mM. This is comparable with K_m = 0.45 mM determined in 0.1 M triethanolamine buffer at pH 7.5, 25 °C and I = 0.1 (62). The addition of 35 mM phosphite buffer (31.5 mM dianion) resulted in competitive inhibition and a value of (K_m)_app = 1.04, which can be combined with K_m = 0.50 mM to give K_i = 29 mM for binding of phosphite dianion to TIM in H₂O at pH 7.4 and I = 0.12. There was no detectable change in the value of (K_m)_app for GAP under these conditions upon the addition of 20 mM glycolaldehyde. Similarly, the addition of 27 mM phosphate buffer (23.0 mM dianion) resulted in competitive inhibition and a value of (K_m)_app = 1.40, which gives K_i = 13mM for binding of phosphate dianion to TIM in H₂O at pH 7.4 and I = 0.12. This is consistent with K_i ≈ 5 mM for phosphate determined at pH 7.7 and low ionic strength (71) and the marked effect of ionic strength on the affinity of TIM for phosphodianion ligands (22).

DISCUSSION

We showed previously (62) that the TIM-catalyzed reaction of the neutral triose sugar (R)-glyceraldehyde in D₂O gives a complex mixture of products resulting from isomerization to give dihydroxyacetone, isomerization with deuterium exchange to give [1-²H]-dihydroxyacetone, “simple” deuterium exchange to give [2-²H]-glyceraldehyde, and a substantial amount of methylglyoxal from the formal elimination of water (Scheme 4). Moreover, all of these exist as mixtures of free carbonyl and hydrated forms in aqueous solution (25, 62, 64, 72, 73).

Scheme 4.

By contrast, isomerization of the minimal two-carbon substrate glycolaldehyde (h-GLY) is a degenerate reaction, its isomerization with deuterium exchange and simple deuterium exchange reactions both give glycolaldehyde labeled with one deuterium at C2 (d-GLY), and no elimination reaction is possible (Scheme 5). This greatly simplifies study of the turnover of this neutral substrate by TIM in D₂O because the sole initial product d-GLY corresponds to that for simple deuterium incorporation, glycolaldehyde exists largely as its hydrate in solution (K_hyd = 15.4, see Results), and h-GLY and d-GLY will be hydrated to the same extent.⁴

Scheme 5.

Figure 1 shows that the slow reaction of 19 mM h-GLY in the presence of a large amount of TIM (0.31 mM) in D₂O buffered at pD 7.0 can be conveniently monitored by ¹H NMR spectroscopy because there are distinct signals for the hydrates of the substrate h-GLY and the first-formed product d-GLY in both the C1 and the C2 regions of the spectrum.⁵ Good first-order kinetics were observed for this reaction (Figure 2,●) and the data give the second-order rate constant for the TIM-catalyzed disappearance of h-GLY as k_enz = 8.1 × 10^-6 s^-1/[TIM] = 0.0264 M^-1 s^-1. However, the observation that the presence of a saturating concentration (8 mM) of the potent competitive inhibitor 2-phosphoglycolate (K_i ≈ 5 μM at pH 7.0 and I = 0.1 (22)) results in less than a three-fold decrease in the observed rate constant for disappearance of h-GLY (Figure 2,■) shows that there is a substantial protein-catalyzed reaction of h-GLY that occurs outside the active site. The second-order rate constant for this nonspecific protein-catalyzed deuterium exchange reaction of h-GLY can be calculated as k_non = 2.6 × 10^-6 s^-1/[TIM] = 0.0106 M^-1 s^-1. Therefore the turnover of h-GLY by deprotonation at the active site of TIM with an observed second-order rate constant of k_TIM = (0.0264 - 0.0106) = 0.0158 M^-1 s^-1 is so slow that it is only 1.5-fold faster than its nonspecific protein-catalyzed deuterium exchange reaction.

Similarly, we reported previously that the TIM-catalyzed turnover of (R)-glyceraldehyde is only 2-fold faster than its nonspecific protein-catalyzed degradation reaction that occurs outside the active site (62). For (R)-glyceraldehyde the nonspecific reaction results mainly in the formation of methylglyoxal, the product of the formal elimination of water. However, for glycolaldehyde, no elimination reaction is possible and the nonspecific reaction results in simple deuterium exchange. The second-order rate constants for the nonspecific protein-catalyzed reactions of glycolaldehyde (0.01 M^-1 s^-1, this work) and (R)-glyceraldehyde (0.009 M^-1 s^-1, calculated from our earlier data (62)) are essentially identical. This suggests a common mechanism for these nonspecific reactions, likely involving formation of a Schiff’s base adduct between the substrate aldehyde group and lysine residues (74, 75), followed by proton abstraction to give an enzyme-bound iminium ion-stabilized carbanion that undergoes either deuterium incorporation from solvent to give d-GLY or β-elimination to give methylglyoxal.

TIM is specific for the binding and turnover of the free carbonyl forms of its natural substrates DHAP and GAP (11) and the true second-order rate constant for turnover of the free carbonyl form of glycolaldehyde (which constitutes 6.1% of the total) at the active site of TIM can be calculated as (k_cat/K_m)_E = 0.0158/0.061 = 0.26 M^-1 s^-1. This is similar to k_cat/K_m = 0.34 M^-1 s^-1 for turnover of the free carbonyl form of (R)-glyceraldehyde by TIM under the same conditions (62). Therefore, the truncation of this neutral triose to give the minimal neutral substrate glycolaldehyde has a negligible effect on the ability of TIM to catalyze proton transfer from α-carbonyl carbon. This shows that there is little or no overall contribution of the terminal hydroxymethylene group of (R)-glyceraldehyde to the catalytic rate acceleration for TIM, and that glycolaldehyde is an appropriate model for the reacting triose sugar portion of the natural substrate GAP.

Activation of TIM by Binding of Exogenous Phosphite Dianion

Figure 3 shows that there is a large dependence of the observed first-order rate constant for turnover of 19 mM h-GLY by 42 μM TIM in D₂O on the concentration of exogenous phosphite dianion at a constant ionic strength of 0.10 (NaCl). The data are summarized in Table 1. The observation of identical values of k_obsd for the disappearance of 19 or 9.5 mM glycolaldehyde, corresponding to 1.2 or 0.6 mM of the free carbonyl form, shows that there is no significant saturation of TIM by glycolaldehyde at these concentrations. This is further supported by the absence of any detectable effect of 20 mM glycolaldehyde on the value of (K_m)_app for turnover of GAP by TIM in the presence of 32 mM phosphite dianion at pH 7.4 (see Results).

The values of k_obsd (s^-1) in Table 1 are at least 100-fold larger than k_o = 2.4 × 10^-7 s^-1 determined for the background reaction of glycolaldehyde in the presence of 32 mM phosphite buffer (50% dianion, pD 7.0) and k_non[TIM] = (0.0106 M^-1 s^-1 × 42 μM) = 4.5 × 10^-7 s^-1 for the nonspecific protein-catalyzed reaction of h-GLY catalyzed by 42 μM TIM (see above). Therefore, observed second-order rate constants (k_cat/K_m)_obsd (M^-1 s^-1) for turnover of the free carbonyl form of h-GLY by TIM in the presence of phosphite dianion (Table 1) were calculated directly from the values of k_TIM = k_obsd/[TIM] using the relationship (k_cat/K_m)_obsd = k_TIM(1 + K_hyd) with K_hyd = 15.4 determined for hydration of the carbonyl group of glycolaldehyde under the experimental conditions.

Figure 4 shows the dependence of the values of (k_cat/K_m)_obsd (M^-1 s^-1) for turnover of the free carbonyl form of h-GLY by TIM in D₂O on the concentration of exogenous phosphite dianion at pD 7.0 and 25 °C (I = 0.10, NaCl). There is a large 250-fold increase in the observed second-order rate constant for turnover of h-GLY by TIM upon addition of 20 mM phosphite dianion to the reaction mixture (Table 1), and the dependence of (k_cat/K_m)_obsd on [HPO₃^2-] exhibits saturation (Figure 4). This behavior is consistent with a model in which the binding of phosphite dianion to the free enzyme E gives a noncovalent E·HPO₃^2- species that is much more reactive towards proton transfer from glycolaldehyde (S) than is the free enzyme E (Scheme 6).

Figure 4.

Dependence of the observed second-order rate constant (k_cat/K_m)_obsd (M^-1 s^-1) for turnover of the free carbonyl form of h-GLY by TIM in D₂O on the concentration of added phosphite dianion at pD 7.0 and 25 °C (I = 0.10). The data were fit to eq 6 to give K_d = 38 ± 9 mM for binding of phosphite dianion to the free enzyme, and (k_cat/K_m)_E·HPi = 185 ± 30 M^-1 s^-1 for turnover of h-GLY by TIM that is saturated with phosphite dianion. Key: (●) [GLY]_tot = 19 mM; (○) [GLY]_tot = 9.5 mM.

Scheme 6.

$${(k_{\mathrm{cat}}∕K_{\mathrm{m}})}_{\mathrm{obsd}}=\left(\frac{K_d}{K_d+\left[{{\mathrm{HPO}}_3}^{2-}\right]}\right){(k_{\mathrm{cat}}∕K_{\mathrm{m}})}_{\mathrm{E}}+\left(\frac{\left[{{\mathrm{HPO}}_3}^{2-}\right]}{K_{\mathrm{d}}+\left[{{\mathrm{HPO}}_3}^{2-}\right]}\right){(k_{\mathrm{cat}}∕K_{\mathrm{m}})}_{E\cdot \mathrm{HPi}}$$ (6)

The data in Table 1 and Figure 4 were fit to eq 6 derived for the model shown in Scheme 6 in which the free enzyme catalyzes proton transfer from h-GLY with (k_cat/K_m)_E = 0.26 M^-1 s^-1, while the binding of phosphite dianion to the free enzyme to give the noncovalent E·HPO₃^2- complex results in an enzyme that is much more reactive, with a second-order rate constant (k_cat/K_m)_E·HPi·. This procedure gave K_d = 38 ± 9 mM for binding of phosphite dianion to the free enzyme and (k_cat/K_m)_E·HPi = 185 ± 30 M^-1 s^-1 as the second-order rate constant for turnover of h-GLY by TIM that is saturated with phosphite dianion. Therefore, the separate binding of phosphite dianion to TIM results in a dramatic 700-fold increase in the second-order rate constant for turnover of the minimal truncated substrate glycolaldehyde. The value of K_d = 38 mM at in D₂O at pD 7.0 and I = 0.10 is similar to K_i = 29 mM determined for competitive inhibition of the turnover of GAP by phosphite dianion in H₂O at pH 7.4 and I = 0.12. This provides further support for the conclusion that the saturation observed in Figure 4 results from the binding of phosphite dianion to the free enzyme.

The thermodynamic cycle shown in Scheme 6 shows that the values of (k_cat/K_m)_E = 0.26 M^-1 s^-1 and (k_cat/K_m)_E·HPi = 185 M^-1 s^-1 can be combined with K_d = 38 mM for binding of phosphite dianion to the free enzyme in the ground state to give K_d^‡ = 53 μM as the dissociation constant for phosphite dianion from the fleeting complex of the enzyme with the transition state for deprotonation of h-GLY to form the enediolate intermediate, E·S^‡. Therefore, phosphite dianion binds 700-fold more tightly to the enzyme in the transition state for proton transfer from carbon than it does to the free enzyme in the ground state.

These facts are represented graphically in Figure 5A which relates the free energy profiles for turnover of glycolaldehyde (S) by the free enzyme (E) and by TIM saturated with phosphite dianion (E·HPO₃^2-). The observed tighter binding of phosphite dianion to the transition state complex E·S^‡ to give the ternary complex E·HPO₃^2-·S^‡ (K_d^‡ = 53 KM, ΔG_o = -5.8 kcal/mol) than to the free enzyme E (K_d = 38 mM, ΔG_o = -1.9 kcal/mol) represents a large “interaction energy” of 3.9 kcal/mol (76, 77). Conversely, it also corresponds to the tighter binding of the altered substrate in the transition state to the binary complex E·HPO ^2- ₃to give the same ternary complex E·HPO₃^2-·S^‡ than to the free enzyme to give E·S^‡ (Figure 5A). In other words, the free enzyme has LOW affinities for both phosphite dianion (K_d = 38 mM) and the altered substrate in the transition state (represented by (k_cat/K_m)_E = 0.26 M^-1 s^-1)), but its binary complexes have HIGH affinities for phosphite dianion (K_d^‡ = 53 μM) and the altered substrate in the transition state (represented by (k_cat/K_m)_HPi = 185 M^-1 s^-1).

Figure 5.

A. Free energy profiles for turnover of glycolaldehyde (S) by free TIM (E) and by TIM that is saturated with phosphite dianion (E·HPO₃^2-). Activation free energy changes were calculated using the Eyring equation at 298 K. The observed tighter binding of phosphite dianion to the transition state complex E·S^‡ (ΔG_o = -5.8 kcal/mol) than to the free enzyme E (ΔG_o = -1.9 kcal/mol) represents the “interaction energy” of 3.9 kcal/mol (76, 77). B. The difference between the total intrinsic binding energy of -5.8 kcal/mol and the observed free energy change of -1.9 kcal/mol for binding of phosphite dianion to the inactive open ground state enzyme E_O to give the active closed liganded enzyme E_C·HPO₃^2- can be attributed to ΔG_conf = +3.9 kcal/mol for an unfavorable conformational change that converts E_O to the active closed enzyme E_C. The observed value of ΔG^‡ = +18.8 kcal/mol for turnover of glycolaldehyde (Figure 5A) may be partitioned conceptually into ΔG^‡ = +14.4 kcal/mol for proton transfer from glycolaldehyde to E_C and ΔG_conf = +3.9 kcal/mol for the obligate unfavorable conformational change that converts E_O to E_C.

The free energy change ΔG_o = -5.8 kcal/mol for binding of phosphite dianion to the binary complex E·S^‡ represents the total intrinsic binding energy of exogenous phosphite dianion available from its interactions with the protein at the transition state complex (Figure 5A). There should be no direct stabilizing interactions between bound phosphite dianion and the altered substrate in the transition state at the E·HPO₃^2-·S^‡ complex: any simple interaction of the negatively charged phosphite dianion with the negative charge at the developing enediolate intermediate would be expected to be destabilizing. Therefore, the smaller value of ΔG_o = -1.9 kcal/mol observed for the binding of phosphite dianion in the ground state strongly suggests that the binding of phosphite dianion to the free enzyme is accompanied by a second unfavorable process with a free energy change of +3.9 kcal/mol.

Origin and Economics of Catalysis of Proton Transfer by TIM

Crystallographic studies show that TIM exists predominantly in a “loop open” conformation in the absence of bound ligands but that the binding of phosphodianion ligands such as the transition state/intermediate analogs PGA (36, 44) and phosphoglycolohydroxamate (37, 42, 45), the substrate analog glycerol 3-phosphate (38) and the substrate DHAP (34, 43) is accompanied by a large conformational change (71). At a minimum this includes a ca. 7 Å movement of the 11-residue mobile loop (loop 6) to close over the mouth of the active site, and results in formation of a hydrogen bond between the backbone amide NH group of Gly-171 at the tip of the loop and an anionic oxygen of the bound phosphodianion group (36-38, 40-42, 44, 45). Importantly, ligand binding and loop closing is also accompanied by a 2 - 3 Å movement of the side-chain of the catalytic base Glu-165 towards the bound ligand, from a “swung-out” to a “swung-in” position (36, 37, 40, 42, 45, 78). This movement of Glu-165 places an oxygen of its carboxylate group between C1 and C2 of bound dihydroxyacetone phosphate where it is optimally aligned for proton transfer (43).

There is crystallographic evidence that the binding of phosphate dianion to trypanosomal TIM results in a “loop closed” conformation in which this simple inorganic oxydianion occupies the same binding site with the same protein-ligand contacts that are observed for the phosphodianion group of bound glycerol 3-phosphate (79). Moreover, the binding of phosphate dianion is also accompanied by the movement of Glu-165 into the active “swung-in” position that is observed for TIM complexed with phosphodianion substrate and intermediate analogs (79). By analogy, the binding of phosphite dianion to TIM should also result in a substantial conformational change, involving closing of the mobile loop over the active site, the formation of a hydrogen bond between the backbone NH of Gly-171 and the phosphite ligand, along with the movement of Glu-165 into the active “swung-in” position where it would be positioned for proton abstraction from co-bound glycolaldehyde.

We propose that a large portion of the intrinsic binding energy of phosphite dianion is used to drive a conformational change from an inactive open form of TIM that is not competent for proton transfer, to an active closed form of the enzyme at which the active site is organized for optimal stabilization of the transition state for proton transfer from bound substrate. This organization would involve the repositioning of Glu-165 into a conformation that is optimal for proton transfer from bound substrate, along with the exclusion of solvent from the active site cavity which in turn would lower its effective dielectric constant. Crystallographic studies show that loop closing and the repositioning of Glu-165 results in a 3 - 4 Å movement of its carboxylate group towards Leu-230 in loop 8 (44, 78), along with a 3 Å movement of Ile-170 in loop 6 deeper into the active site towards Glu-165 (39, 40). Thus loop closing and reorganization of the active site places the anionic carboxylate group of Glu-165 into close proximity with the bulky hydrophobic side-chains of the conserved residues Ile-170 and Leu-230 (38-40, 43, 44, 78). We propose that this makes a major contribution to the observed ca. 3 unit higher pK_a of Glu-165 in the complex with 2-phosphoglycolate than at the free enzyme (22, 80), which in turn serves to increase the driving force for proton transfer (81). The placement of Glu-165 in a hydrophobic environment provides a strong driving force for proton transfer from bound substrate because this moves essentially localized negative charge from a highly hydrophobic environment onto the enolic oxygen of the enediolate intermediate. Stabilization of negative charge at the enediolate by electrostatic and/or hydrogen bonding interactions with His-95 and Lys-12 is strongly favored by the low effective dielectric constant of the active site that arises from the exclusion of water and the proximity of Ile-170 and Leu-230 (6, 82).

The “economics” of our model for catalysis by TIM are illustrated in Figure 5B (77). In this model both phosphite dianion and the altered substrate in the transition stabilize the same active closed form of the enzyme, E_C. The difference between the total intrinsic binding energy of -5.8 kcal/mol and the observed free energy change of -1.9 kcal/mol for binding of phosphite dianion to the inactive open ground state enzyme E_O to give the active closed liganded enzyme E_c·HPO₃^2- can be attributed to ΔG_conf = +3.9 kcal/mol for an obligate unfavorable conformational change that converts the open enzyme E_O to the closed enzyme E_C and results in reorganization of the active site as described above (77, 83). We propose that the reactivity of the active closed enzyme E_C towards bound substrate is essentially independent of the presence of bound phosphite dianion. Therefore, the observed value of ΔG^‡ = +18.8 kcal/mol for turnover of glycolaldehyde (Figure 5A) may be conceptually partitioned into ΔG^‡ = +14.4 kcal/mol for proton transfer from glycolaldehyde to the active closed enzyme, and ΔG_conf = +3.9 kcal/mol for the conformational change from the inactive open to the active closed enzyme that must occur in order to facilitate proton transfer (Figure 5B). Furthermore, this model accounts for the observation that the binary complexes E_C·S^‡ and E_C·HPO₃^2- exhibit much higher affinities for HPO₃^2- and the altered substrate in the transition state, respectively, because the binding of S^‡ or of HPO₃^2- results in a conformational change to the closed enzyme at which full intrinsic binding energies may be expressed upon subsequent ligand binding to give a ternary complex.

The values of K_m for turnover of glycolaldehyde by the free enzyme and the E·HPO₃^2- complex should be very similar because the binding interactions of this simple two carbon sugar with the enzyme should be largely confined to those with His-95 and Lys-12, whose positions do not change significantly upon closure of the mobile loop (36, 40). Therefore, the 700-fold larger value of k_cat/K_m for turnover of glycolaldehyde by the E·HPO₃^2- binary complex than by free TIM likely also translates to a 700-fold larger value of k_cat for E·HPO₃^2-. In this case, the intrinsic binding energy of phosphite dianion is used directly to increase the rate constant for proton transfer from bound substrate to the enzyme.

Comparison of Phosphite and Phosphate Dianions

Scheme 7 shows that the α-hydroxycarbonyl and inorganic oxydianion “pieces” of GAP are modeled by the minimal two-carbon sugar glycolaldehyde and phosphite dianion. Truncation of both the sugar and the formal phosphate dianion portions of the substrate is necessary in order to generate a “two-part substrate” that maintains the same approximate steric bulk of the “holistic substrate” (84).

Scheme 7.

The value of k_obsd = 2.9 × 10^-6 s^-1 for turnover of h-GLY by 42 μM TIM in the presence of 15 mM phosphate dianion can be corrected for the nonspecific reaction that occurs outside the active site (k_non = 0.0106 M^-1 s^-1) and the presence of 6.1% of the free carbonyl form of glycolaldehyde to give (k_cat/K_m)_obsd = 0.96 M^-1 s^-1. This is only 3.7-fold larger than (k_cat/K_m)_E = 0.26 M^-1 s^-1 for turnover of h-GLY by TIM in the absence of phosphate dianion (Table 1). By contrast, the addition of 15 mM phosphite dianion results in a 200-fold increase in (k_cat/K_m)_obsd for the turnover of h-GLY (Table 1). The value of K_i =13 mM for competitive inhibition of TIM by phosphate dianion in H₂O at pH 7.4 (I = 0.12) is 2.2-fold smaller than K_i = 29 mM for phosphite dianion under the same conditions (this work). These data, together with K_d = 38 mM for phosphite dianion in D₂O, can be used to estimate a value of K_d = 17 mM for binding of phosphate dianion to TIM in D₂O at pD 7.0 (I = 0.10). Therefore the value of (k_cat/K_m)_obsd = 0.96 M^-1 s^-1 can be substituted into eq 6 rewritten for phosphate dianion with K_d = 17 mM to give (k_cat/K_m)_E·Pi ≈ 1.7 M^-1 s^-1 as the second-order rate constant for turnover of h-GLY by TIM that is saturated with phosphate dianion. This is 100-fold smaller than (k_cat/K_m)_E·HPi = 185 M^-1 s^-1 for TIM that is saturated with phosphite dianion. Therefore, there is a large sensitivity of the reactivity of TIM towards proton transfer from glycolaldehyde to the steric and electronic properties of the co-bound oxydianion activator (85). Replacement of the H group at phosphite dianion with the larger OH group to give phosphate dianion may prevent the correct productive co-binding of both glycolaldehyde and the oxydianion in the active site. This could result in incomplete or incorrect closing of the mobile loop and an active site environment that is less than optimal for catalysis of proton transfer. There may also be unfavorable electrostatic consequences, for example an unfavorable interaction of the positive dipole of the OH group at bound phosphate dianion with the positive charge of Lys-12 in the active site.

Utilization of the Intrinsic Binding Energy of the Remote Substrate Phosphodianion Group: A General Strategy for Enzymatic Catalysis?

Our earlier comparison of the TIM-catalyzed reactions of the neutral triose (R)-glyceraldehyde and GAP showed that the total intrinsic binding energy of the substrate phosphodianion group in the transition state for proton transfer, calculated from the ratio of the second-order rate constants for turnover of GAP and (R)-glyceraldehyde, is 14.1 kcal/mol (62). The large intrinsic binding energy for exogenous phosphite dianion in the transition state for deprotonation of glycolaldehyde, 5.8 kcal/mol, shows that a significant fraction of the transition state stabilization due to the phosphodianion group at GAP can be “reconstituted” by binding a small oxydianion ligand to TIM. The remaining ca. 8 kcal/mol may be attributed to several factors:

1. The entropic price for binding phosphite dianion to the binary complex E·S^‡ to give the ternary complex E·S^‡·HPO₃^2-. This price does not have to be “paid” by the intramolecular phosphodianion group which is already present at the binary transition state complex for the reaction of GAP. The moderate value of 8 kcal/mol is considerably less than the estimated maximal value of 11 - 12 kcal/mol (76, 86).

2. The separate binding of glycolaldehyde and phosphite dianion may result in incomplete or incorrect closure of the mobile loop such that the active site environment is not completely optimized for catalysis of proton transfer.

3. The combination of the enediolate intermediate generated from deprotonation of glycolaldehyde and phosphite dianion is unlikely to be a perfect steric and electronic match for the enzyme-bound enediolate intermediate generated from deprotonation of GAP. Small differences in the steric and/or electrostatic and hydrogen bonding interactions of the two fragments with the enzyme in the transition state, compared with those for the reaction of GAP, could result in less than optimal stabilization of the intermediate and a corresponding destabilization of the transition state for its formation.

4. A potentially important interaction between Lys-12 and the bridging oxygen of the substrate phosphodianion group has been identified in the crystal structure of the Michaelis complex of yeast TIM with DHAP (43). It has also been observed that the TIM-catalyzed isomerization of the phosphonate analog of GAP, (R)-2-hydroxy-4-phosphonobutyraldehyde, is 800-fold slower than that of GAP (21). By analogy, the absence of this bridging oxygen may reduce (k_cat/K_m)_E·HPi for turnover of glycolaldehyde by the TIM·HPO₃^2- complex.

In parallel with the work described here we have shown that the binding of phosphite dianion to orotidine 5′-monophosphate decarboxylase (OMPDC) stabilizes the transition state for decarboxylation of a neutral truncated substrate that lacks the 5′-phosphodianion group of the natural substrate by 7.8 kcal/mol (87). TIM and OMPDC have come to represent paradigms for enzyme-catalyzed proton transfer and decarboxylation reactions, respectively. OMPDC is the titular member of the OMPDC “suprafamily” (28, 88, 89) and shares the ubiquitous (β,α)₈-barrel fold that was first observed for TIM (90, 91), along with a common canonical phosphate binding site at loops 7 and 8 (28, 29, 88). Moreover, both TIM and OMPDC are characterized by interactions of flexible loop(s) with the phosphodianion group of bound ligands (loop 6 for TIM and loop 7 for OMPDC) (90, 91), the closure of which sequester the bound substrate from bulk solvent. It is possible that the utilization by these enzymes of the intrinsic binding energy of the remote phosphodianion group of the substrate for transition state stabilization is a coincidence, and of limited general significance. However, we propose that the interactions of the closed flexible loops of TIM and OMPDC with the substrate phosphodianion group represent a common strategy that enables efficient recruitment of the intrinsic binding energy of small remote binding determinant for transition state stabilization. By contrast with TIM, the mechanism of action of the extraordinarily efficient OMPDC remains largely undefined. Further studies to elucidate the structural and dynamic strategies used to elicit its enormous 10¹⁷-fold rate acceleration (92) are clearly warranted.