Transition State Structures and the Roles of Catalytic Residues in GAP-Facilitated GTPase of Ras As Elucidated by ¹⁸O Kinetic Isotope Effects

Abstract

Ras-catalyzed guanosine 5′ triphosphate (GTP) hydrolysis proceeds through a loose transition state as suggested in our previous study of ¹⁸O kinetic isotope effects (KIE) [Du, X. et al. ((2004)) Proc. Natl. Acad. Sci. U.S.A. 101, 8858–8863]. To probe the mechanisms of GTPase activation protein (GAP)-facilitated GTP hydrolysis reactions, we measured the ¹⁸O KIEs in GTP hydrolysis catalyzed by Ras in the presence of GAP³³⁴ or NF1³³³, the catalytic fragment of p120GAP or NF1. The KIEs in the leaving group oxygens (the β nonbridge and the β–γ bridge oxygens) reveal that chemistry is rate-limiting in GAP³³⁴-facilitated GTP hydrolysis but only partially rate-limiting in the NF1³³³-facilitated GTP hydrolysis reaction. The KIEs in the γ nonbridge oxygens and the leaving group oxygens reveal that the GAP³³⁴ or NF1³³³-facilitated GTP hydrolysis reaction proceeds through a loose transition state that is similar in nature to the transition state of the GTP hydrolysis catalyzed by Ras alone. However, the KIEs in the pro-S β, pro-R β, and β–γ oxygens suggest that charge increase on the β–γ bridge oxygen is more prominent in the transition states of GAP³³⁴-and NF1³³³-facilitated reactions than that catalyzed by the intrinsic GTPase activity of Ras. The charge distribution on the two β nonbridge oxygens is also very asymmetric. The catalytic roles of active site residues were inferred from the effect of mutations on the reaction rate and KIEs. Our results suggest that the arginine finger of GAP and amide protons in the P-loop of Ras stabilize the negative charge on the β–γ bridge oxygen and the pro-S β nonbridge oxygen of a loose transition state, whereas Lys-16 of Ras and Mg²⁺ are only involved in substrate binding.

Ras is the prototypical member of the small G protein super-family that mediates diverse signaling pathways in eukaryotes (1, 2). Like other G proteins, Ras alternates between the guanosine 5′ diphosphate (GDP¹)-bound inactive state and the GTP-bound active state. GTP-bound Ras is capable of binding to different effector proteins and thereby activating multiple signaling pathways. The intrinsic GTPase activity of Ras is very low, necessitating the down-regulation of Ras activity in vivo by GTPase activating proteins (GAPs) (3), including p120GAP (4) and neurofibromin 1(NF1) (5). Both GAPs increase the rate of GTP hydrolysis by more than 10⁵-fold, from 10⁻⁴ s⁻¹ to at least 8 s⁻¹ (6, 7). Mutations in Ras that impair either intrinsic or GAP-facilitated GTPase activity leave Ras in a prolonged state of activation and were found in 30% of cancers (8). Mutations in NF1 are responsible for the pathogenesis of neurofibromatosis type 1 (9).

The chemical mechanism of Ras-catalyzed GTP hydrolysis, in particular the nature of the transition state in the intrinsic or GAP-facilitated GTPase reaction, has been the focus of considerable debate. The traditional view holds that the hydrolysis of a phosphate monoester with a good leaving group, of which GTP is one example, proceeds through a metaphosphate-like, loose transition state that features extensive cleavage to the bond between the phosphorus and the leaving oxygen but little or no bond formation between the phosphorus and the nucleophile (10–14). Recent computational studies suggest that metaphosphate is generated as a discrete intermediate in the reaction pathway before it is attacked by the nucleophilic water to generate phosphate (15, 16). An alternative, substrate-assisted, mechanism involves a proton transfer from the attacking water to GTP as a pre-equilibrium step (17–19). The hydroxide thus formed then attacks the γ phosphorus to form a phosphorane intermediate, which decays in the rate-limiting step to generate GDP and phosphate. Resolution of the mechanism is the key to the elucidation of the catalytic roles of the enzyme residues in GTP hydrolysis.

Infrared and Raman spectroscopic measurements clearly revealed that, upon binding to Ras, the P_β–O_{β nonbridge} bonds of GTP are weakened, but P_γ–O_{γ nonbridge} bonds are strengthened (20–22). If the distortion of the electronic structure of GTP by Ras represents destabilization of the ground-state in the direction of the transition state, it can be inferred that the transition state is loose. Indeed, our earlier studies revealed substantial ¹⁸O KIE in the β–γ bridge oxygen and β nonbridge oxygens in Ras-catalyzed GTP hydrolysis, demonstrating substantial cleavage of P_γ–O_{βγ bridge} and P_β–O_{β nonbridge} bonds at the transition state (23). KIE in the two β nonbridge oxygens indicated that charge generated by the cleavage of P_γ–O_{βγ bridge} bond is delocalized into the two β nonbridge oxygen atoms through resonance structures. The KIE data, together with the crystal structures of Ras in complex with various GTP analogues (24, 25), provide insight into how Ras catalyzes GTP hydrolysis. Charge increase on the leaving group oxygens, namely, the β–γ bridge and the two β nonbridge oxygens, indicates that the interactions of Ras with the three leaving group oxygen atoms become stronger at the transition state and thereby afford preferential stabilization of the transition state. These interactions are mainly mediated by the P loop of Ras: the amide hydrogen of Gly-13 forms a short hydrogen bond with the β–γ bridge oxygen; the main chain NH groups of Val-14,Gly-15, and Lys-16, and the side chain of Lys-16 interact with the pro-S β nonbridge oxygen; the main chain NH of Ser-17 and Mg²⁺ interacts with the pro-R β oxygen (Figure 1). By contrast, interactions between the γ nonbridge oxygens and the Mg²⁺, the main chain NH groups of Thr-35 and Gly-60, and the side chain of Lys-16 are not expected to provide electrostatic stabilization of the transition state over the ground state.

Figure 1.

Active site of Ras in complex with GAP³³⁴ and GDP · AlF₃, a transition state analogue, in the crystal structure of Ras · GDP · AlF₃ 3 RasGAP (PDB ID 1WQ1). O1, O2, and O3 refer to the pro-S β nonbridge oxygen, the pro-R β nonbridge oxygen, and the β–γ bridge oxygen, respectively. The functional groups of Ras that contribute to interactions with the pro-S β nonbridge oxygen include the main chain NH groups of Val-14,Gly-15, andLys-16, as well as theNζ of Lys-16.The main chain NH of Ser-17 and Mg²⁺ interact with pro-Roxygen. The amide hydrogen of Gly-13 forms a strong hydrogen bond with the β–γ bridge oxygen. Interactions between Ras and γ nonbridge oxygens are realized by Mg²⁺, the main chain NH of Thr-35 andGly-60, the side chain of Gln-61, and the side chain of Lys-16.Arg-789 of GAP³³⁴ is positioned to interact with both the β–γ bridge oxygen and one γ nonbridge oxygen.

Structural, biochemical, and computational studies led to the notion that GAP³³⁴ facilitates the GTPase of Ras by stabilizing a catalytically competent conformation of Ras and by providing an arginine residue to electrostatically stabilize the transition state (13, 26–28). The crystal structure of Ras in complex with GAP³³⁴, a catalytic fragment of p120GAP, and GDP · AlF₃, a transition state analogue (Figure 1), revealed that GAP³³⁴ stabilizes Switch II of Ras in a catalytically competent conformation (26). In this conformation, the main chain carbonyl oxygen of Thr-35 of Ras and the side chain oxygen atom of Gln-61 of Ras, which points away from the γ phosphoryl group in the absence of GAP³³⁴, form hydrogen bonds with the nucleophilic water and orient it for inline attack on the γ phosphorus. In addition, Arg-789 of GAP³³⁴, often referred to as the arginine finger, extends into the active site of Ras to interact directly with the β–γ bridge oxygen and one of the γ nonbridge oxygen atoms. The critical role of the arginine finger in catalysis was confirmed by biochemical and computational studies (27, 28). NF1³³³, the catalytic fragment of NF1, is structurally similar to GAP³³⁴ and also possesses a catalytic arginine finger (27, 29). Therefore, NF1 was proposed to facilitate GTP hydrolysis by a similar mechanism (29).

Here, we address the mechanism of GAP-facilitated GTP hydrolysis using ¹⁸O KIE as a probe of the transition state structures in these reactions. We have synthesized a suite of ¹⁸O-labeled GTP molecules (Figure 2) that allow the determination of the KIEs in the oxygen atoms at or geminal to the scissile bond. The KIE in γ¹⁸O³-labeling is equivalent to the secondary KIE in the hydrolysis of phosphate monoesters. The charge on the β–γ bridge oxygen due to the cleavage of P_γ–O_{βγ bridge} bond is delocalized into the two β nonbridge oxygens. The electron delocalization effectively increases the bond order of the β–γ bridge oxygen but decreases the bond order of the two β nonbridge oxygen atoms. Therefore, the KIE in β¹⁸O₃, instead of the KIE in the β–γ bridge oxygen alone, better reflects the extent of cleavage to the P_γ–O_{βγ bridge} at the transition state (23). The KIE in the β–γ bridge oxygen is determined by the ratio of the KIE in β¹⁸O₃ to that in β¹⁸O₂ or by the ratio of KIE in γ¹⁸O₄ to that in γ₁₈O₃.KIEs in the β–γ bridge, the pro-S β, and the pro-R β oxygens afford stereospecific probes of changes in bond orders and thereby the changes in the charges on these atoms at the transition state. Analysis of the kinetic behavior and KIEs of certain Ras mutants has enabled us to address the roles of the highly conserved residues, Lys-16 andGln-61, in the catalytic site of Ras. In experiments with an arginine finger mutant ofNF1³³³, we have assessed the role of the arginine finger in GAP-facilitated GTP hydrolysis.

Figure 2.

Double-labeled nucleotides used in this study. Carbon, nitrogen, oxygen, and phosphorus atoms are colored in silver, blue, yellow, and red, respectively. ¹⁸O-labels are indicated by *.

MATERIALS AND METHODS

Materials

Guanine nucleotides uniformly labeled with ¹³C or ¹³C, ¹⁵N were purchased from Spectra Gases, Inc. The ¹³C enrichment is better than 99% as determined by electro-spray mass spectrometry. ¹³C-depleted GTP ([¹²C]GTP) was either purchased from Spectra Gases or synthesized as described earlier (30). The residual carbon isotope ratios in the [¹²C]GTP purchased or synthesized for this study were 0.23% and 0.09%, respectively. ${\mathrm{H}}_2^{18}\mathrm{O}$ (97% enriched) was from Isotech. Other reagents for chemical synthesis were obtained from Sigma.

Synthesis of ¹⁸O,¹³C-labeled nucleotides

¹⁸O, ¹³C-labeled nucleotides were synthesized by coupling ¹⁸O-labeled phosphate or phosphate to guanine nucleotides uniformly labeled with ¹³C. The schemes for the synthesis of the six double-labeled nucleotides, [¹³C₁₀, γ¹⁸O₃]GTP, [¹³C₁₀, γ¹⁸O₄]GTP, [¹³C₁₀, β¹⁸O₃]GTP, [¹³C₁₀,¹⁵N₅, β¹⁸O₂]GTP, [¹³C₁₀,¹⁵N₅, pro-S β¹⁸O]GTP, and [¹³C₁₀,¹⁵N₅, pro-R β¹⁸O] are provided as Supporting Information. The detailed procedures for the synthesis of [¹³C₁₀, γ¹⁸O₃] GTP, [¹³C₁₀, γ¹⁸O₄]GTP, [¹³C₁₀, β¹⁸O₃]GTP, and [¹³C₁₀,¹⁵N₅, β¹⁸O₂]GTP were described earlier (23, 30). Electrospray mass spectrometry was used to deduce the ¹⁸O isotopic purity of ¹⁸O-and ¹³C-labeled GTP. The ¹⁸O enrichment level for [¹³C₁₀, γ¹⁸O₃] GTP, [¹³C₁₀, γ¹⁸O₄]GTP, [¹³C₁₀, β¹⁸O₃]GTP, [¹³C₁₀,¹⁵N₅, β¹⁸O₂] GTP, [¹³C₁₀,¹⁵N₅, pro-Sβ¹⁸O]GTP, and [¹³C₁₀,¹⁵N₅, pro-R β¹⁸O] GTP were 96%, 94%, 96%, 94%, 95%, and 95%, respectively.

Synthesis of [¹³C₁₀,¹⁵N₅, pro-S β¹⁸O]GTP, and [¹³C₁₀,¹⁵N₅, pro-R β¹⁸O]GTP

The synthesis of [¹³C₁₀,¹⁵N₅, pro-S β¹⁸O]GTP and [¹³C₁₀,¹⁵N₅, pro-R β¹⁸O]GTP was carried out following the methods of Fritz Eckstein. R and S diastereomers of GTPβS were first synthesized (31). Sulfur in the R or S diastereomer was then substituted with ¹⁸O with the reversion of configuration, yielding [¹³C₁₀,¹⁵N₅, pro-S β¹⁸O]GTP, and [¹³C₁₀,¹⁵N₅, pro-R β¹⁸O]GTP, respectively (32). We used a modification of the protocol described by Connolly, et al. (32) to avoid the production of heavily bromated ¹⁸O GTP, as follows. GTPβS (1.5 mg) was mixed with 500 µL of GDP (80 mM) and 100 µL of dithiothreitol (300mM), and the solution was dried on a Speedvac. The residue was dissolved in 100 µL of ${\mathrm{H}}_2^{18}\mathrm{O}$ and re-evaporated, and the step was repeated. Finally, the residue was dissolved in 200 µL of ${\mathrm{H}}_2^{18}\mathrm{O}$ and 600 µL of N-bromosuccinimide (100 mM) in dry dioxane. After 5 min at room temperature, 100 µL of 1-mercaptoethanol was added to the solution. After 2min, 3mL of H₂O was added, and the pH was adjusted to approximately 7.0 with 150 µL of triethylamine. The solution was then loaded onto a 5 mL HiTrap Q (GE Healthcare) column. GDP and GTP were separated by a 20 column volume, 0 to 1M gradient of ammonium acetate. The fraction containing GTP was collected and further purified on a 1 mL Hi Trap Q column. Ammonium acetate was removed afterward by lyophilization.

Protein Expression and Purification

A synthetic gene encoding residues 1 to 166 of human H-Ras was subcloned into a pET28 vector (33). The catalytic fragments of SOS1 (residues 564–1049) (34), mouse Cdc25 (residue 976–1260, Cdc25^Mm285) (35), human NF1 (residue 1197–1528, NF1³³³) (36), and human p120GAP (residue 714–1097, GAP³³⁴) (6) were subcloned into pGEX vectors. Mutations in Ras and NF1³³³ were introduced using the QuickChange kit (Stratagene). His-tagged Ras and GST-fused Cdc25Mm285 were coexpressed in E. coli strain BL21 (DE3) and copurified as a complex through Ni-affinity, glutathione affinity, and ion-exchange column chromatography. Ras(K16A) in complex with Cdc25^Mm285 was prepared similarly. Unlike other Ras mutants, Ras(Q61H) interacts with Cdc25^Mm285 only weakly. When His-tagged Ras(Q61H) and GST-fused Cdc25^Mm285 were coexpressed bacterially, the stoichiometric complex was not obtained following Ni-affinity and glutathione affinity chromatography. Instead, the Ras(Q61H)–SOS1 complex was prepared and used in this study. First, GST-fused SOS1 was retained on Glutathione-Sepharose beads. A tight complex of Ras(Q61H)-SOS1 is formed when excess Ni-affinity purified, His-tagged Ras(Q61H) was loaded onto the column. The complex was eluted with 10 mM glutathione and further purified by ion-exchange column chromatography. GAP³³⁴, NF1³³³, or NF1³³³(R1276A) was expressed in BL21(DE3) and purified by glutathione affinity and ion-exchange chromatography.

Steady State Condition for GTP Hydrolysis Catalyzed by Ras

A typical GTP hydrolysis reaction for Ras was carried out in 40 µL volume that contained 136 µM (11.3 mg/mL) Ras · Cdc25^Mm285 complex, 1 mM GTP, 100 µM MgCl₂, and 50 mM HEPES at pH 7.5. Reactions typically reached 50% completion after 32 h at 22 °C. Aliquots of 13 µL were taken, placed in a 100 °C heating block for 20 s to denature the proteins, and then frozen at −20 °C before IRMS analysis. The turnover rate for Ras-catalyzed GTP hydrolysis, determined by measurement of the reaction period during which half of the reactant is consumed, was 0.115 h⁻¹ and was only one-sixth of the rate determined under single turnover conditions (20). Most of the reduction in rate can be ascribed to the inhibitory effect of Cdc25^Mm285, which is expected to compete effectively with GTP for Ras and secure Ras in the Ras · Cdc25^Mm285 complex (37). GDP also competes with GTP for Ras and product inhibition becomes significant even before the reaction is half-complete. On the basis of these considerations, it is estimated that about 20% to 25% of Ras existed in the Ras · GTP complex, while most of the Ras was in complex with Cdc25^Mm285 at the beginning of the reaction.

Steady State Condition for GAP-Facilitated GTP Hydrolysis Reactions

Reactions in the presence of GAP³³⁴ were carried out in 30 µL volumes that contained 48 µM (4.0mg/mL) Ras · Cdc25^Mm285 complex, 0.084 µM (0.0058 mg/mL) GAP³³⁴, 1 mM GTP, 25 µM MgCl₂, and 50 mM HEPES at pH 7.5. Typically, a reaction reached 50% completion in 40 min. For other hydrolysis reactions, the same concentration of Ras or Ras mutant in complex with Cdc25^Mm285 was used and the concentration of GAP was varied. The GAP concentration and the halftime of the reactions were as follows: 17 µM GAP³³⁴ and a t_1/2 of 40 min for Ras(K16A)-GAP³³⁴; 0.087 µM NF1³³³ and a t_1/2 of 60 min for Ras-NF1³³³; 4.3 µM NF1³³³ and a t_1/2 of 40 min for Ras (K16A)-NF1³³³; and 17 µM NF1³³³(R1276A) and a t_1/2 of 270 min for Ras-NF1³³³(R1276A). A typical experiment for Ras (Q61H)-NF1³³³ was carried out in 30 µL volume containing 51 µM (5.5 mg/mL) Ras(Q61H) · SOS1 complex and 67 µM (4.6 mg/mL) NF1³³³ and reached 30% completion in 240 min. The reactions for Ras-GAP³³⁴ and Ras-NF1³³³ were carried out in the presence of 25 µM MgCl₂.Other reactions were carried out in the presence of 50 µM MgCl₂.

Determination of KIEs

The method used to determine KIEs for GTP hydrolysis has been reported in detail (30) and is only briefly described here. An internal competition method was employed using ¹⁸O,¹³C-labeled GTP and ¹³C-depleted GTP. The double-labeled GTP is labeled with ¹⁸O at sites of mechanistic interest and with ¹³C at all carbon positions on the sugar and base of the nucleotide, while the ¹⁶O-nucleotide is depleted of ¹³C. The relative abundance of the labeled and unlabeled substrates or products is thereby reflected in the carbon isotope ratio (¹³C/¹²C) in GTP or GDP, which was measured using a LC-coupled IRMS. A KIE can be calculated from the change in the isotope ratio of the substrate GTP relative to that at the beginning of the reaction (R_s/R₀), by the changes in the isotope ratio of the product GDP relative to that when hydrolysis is complete (R_p/R₀), or by the difference between the isotope ratios in GDP and GTP (38). It was established that in our experimental system, the three methods give equivalent results, but the third method offers the best precision (30). Therefore, the changes in carbon isotope ratios of GDP and GTP (R_p/R_s) in the course of the reaction were used to calculate KIE according to eq 1. In a few cases, we were not able to accurately measure the isotope ratio of GDP, and the KIEs were calculated according to eq 2. Fractional reaction extent was calculated from mass 44 peak areas of GDP and GTP according to eq 3.

$$KIE=\frac{\text{ln}(1-f)}{\text{ln}\frac{1-f}{1-f+fRp/Rs}}$$ (1)

$$KIE=\frac{\text{ln}(1-f)}{\text{ln}[(1-f)R_s/R_0]}$$ (2)

$$f=\frac{\mathit{\text{area}}(\mathit{\text{GDP}})}{\mathit{\text{area}}(\mathit{\text{GDP}})+\mathit{\text{area}}(\mathit{\text{GTP}})}$$ (3)

RESULTS

KIEs are reported in Table 1 for GTP hydrolysis catalyzed by Ras and Ras mutants, in the presence or absence of GAP³³⁴, NF1³³³, or the arginine finger mutant of NF1³³³. In addition to wild type proteins, Ras(K16A), Ras(Q61H), and NF1³³³(R1276A) were employed to probe the kinetic mechanism and the perturbation of the transition state by the indicated mutations. Ras and Ras mutants were purified as complexes with guanine nucleotide exchange factors, Cdc25^Mm285 or SOS1, for several reasons: first, Ras proteins obtained in this way were free of GDP so that the isotope ratio of hydrolysis product (GDP) was not biased by the presence of residual, unlabeled GDP; second, Cdc25^Mm285 and SOS1 accelerate GTP release and thereby reduce the external forward commitment for the reaction; third, the exchange factors accelerate product (GDP) release so that the steady state rate is not limited byproduct release; last, active Ras(K16A) can only be obtained when copurified with Cdc25^Mm285. Six double-labeled nucleotides, [¹³C₁₀, γ¹⁸O₃]GTP, [¹³C₁₀, γ¹⁸O₄]GTP, [¹³C₁₀, β¹⁸O₃]GTP, [¹³C₁₀,¹⁵N₅, β¹⁸O₂]GTP, [¹³C₁₀,¹⁵N₅, pro-S β¹⁸O]GTP, and [¹³C₁₀,¹⁵N₅, pro-R β¹⁸O], were used in this study (Figure 2). KIEs in ¹³C-labeling or ¹³C,¹⁵N-labeling were measured directly using ¹³C- or ¹³C,¹⁵N-labeled nucleotides. Corrections due to the incomplete ¹³C depletion in ¹²C-GTP and incomplete ¹⁸O-labeling in double-labeled GTP for the computation of the KIE in ¹⁸O-labeling were carried out according to the formulation of Hermes et al. (39).

Table 1.

KIEs in GTP Hydrolysis Reactions Catalyzed by Ras, GAP (GAP³³⁴ or NF1³³³), and Mutants^a

labeling	Ras	Ras-GAP334	Ras(K16A)-GAP334	Ras-NF1333	Ras(K16A)-NF1333	Ras(Q61H)-NF1333	Ras-NF1333(R1276A)
13C	1.0014 (8, 0.03%)	1.0049 (11, 0.04%)	1.0051 (8, 0.06%)	1.0045 (11, 0.06%)	1.0039 (20, 0.06%)	1.0006 (6, 0.17%)	1.0035 (12, 0.04%)
13C, 15N	1.0016 (6, 0.06%)	1.0057 (9, 0.05%)	1.0060 (15, 0.05%)	1.0055 (9, 0.04%)	1.0049 (9, 0.08%)	1.0007 (6, 0.11%)	1.0036 (8, 0.03%)
γ18O3b	0.9994 (6, 0.09%)	1.0012 (15, 0.11%)	0.9989 (15, 0.17%)	0.9992 (9, 0.09%)	0.9983 (10, 0.08%)	0.9991 (6, 0.19%)	0.9994 (9, 0.05%)
γ18O4b	1.0106 (6, 0.04%)	1.0172 (15, 0.08%)	1.0140 (14, 0.14%)	1.0050 (9, 0.14%)	1.0030 (6, 0.11%)	1.0083 (6, 0.19%)	1.0132 (9, 0.10%)
β18O3b	1.0192 (5, 0.05%)	1.0220 (21, 0.14%)	1.0214 (17, 0.18%)	1.0077 (17, 0.09%)	1.0048 (11, 0.08%)	1.0179 (6, 0.18%)	1.0202 (9, 0.08%)
β18O2c	1.0078 (6, 0.08%)	1.0057 (9, 0.08%)	1.0051 (9, 0.14%)	1.0000 (9, 0.10%)	0.9995 (7, 0.09%)	1.0096 (6, 0.14%)	1.0069 (9, 0.05%)
pro-S β18Oc	1.0042 (6, 0.07%)	1.0044 (9, 0.11%)	1.0039 (9, 0.16%)	1.0012 (9, 0.11%)	1.0003 (7, 0.10%)	1.0046 (6, 0.12%)	1.0033 (9, 0.08%)
pro-R β18Oc	1.0028 (6, 0.07%)	1.0006 (8, 0.06%)	1.0010 (8, 0.11%)	0.9975 (9, 0.06%)	0.9972 (9, 0.16%)	1.0062 (6, 0.14%)	0.9994 (9, 0.06%)
γ18O4/γ18O3d	1.0112 (0.10%)	1.0160 (0.14%)	1.0151 (0.21%)	1.0058 (0.17%)	1.0047 (0.14%)	1.0092 (0.27%)	1.0138 (0.11%)
β18O3/β18O2e	1.0113 (0.09%)	1.0162 (0.15%)	1.0162 (0.24%)	1.0077 (0.13%)	1.0053 (0.12%)	1.0082 (0.22%)	1.0132 (0.09%)
β–γ bridgef	1.0113	1.0161	1.0157	1.0068	1.0050	1.0087	1.0135
rate g	0.115 h−1	2.5 s−1	0.012 s−1	1.6 s−1	0.048 s−1	0.00031 s−1	0.0018 s−1

^aEach KIE is an average with the number of determination and associated errors listed in parentheses. The following equation was utilized to correct for the incomplete ¹³C-depletion in [¹²C, ¹⁶O]GTP and incomplete ¹⁸O-labeling in double-labeled GTP ¹⁸(V/K) = WY/[A – (W – A)[(1 – B)Z/(BX)] – W(1 – Y)], where W = the observed isotope effect [KIE in ¹³,¹⁸(V/K) or ¹³,¹⁵,¹⁸(V/K)], A = the isotope effect in ¹³C- or ¹³C,¹⁵N-labeling, X = the fraction of ¹³C in the double-labeled GTP (0.99), Y = the fraction of ¹⁸O in double-labeled GTP (0.95), Z = the fraction of ¹³C in the [¹²C, ¹⁶O]GTP (0.23% or 0.09%), and B = the fraction of ¹³C- or ¹³C,¹⁵N-labeled GTP in the mixture of reactant (0.89% or 1.03% for Z of 0.23% or 0.09%, respectively).

^bThe associated errors for the KIEs in γ¹⁸O₃, γ¹⁸O₄, or β¹⁸O₃ are computed by Δ = [(Δ₁)²+(Δ₂)²]_1/2, where Δ₁ and Δ₂ are the standard deviations in the corresponding KIEs in ¹³C, ¹⁸O-labeling, and ¹³C-labeling, respectively.

^cThe associated errors for the KIEs in β¹⁸O², pro-S β¹⁸O, or pro-R β¹⁸O are computed by Δ = [(Δ₁)²+(Δ₂)²]_1/2, where Δ₁ and Δ₂ are the standard deviations in the corresponding KIEs in ¹³C, ¹³N, ¹⁸O-labeling, and ¹³C, ¹⁵N-labeling, respectively.

^dThe KIE in the β–γ bridge oxygen was computed as the ratio of the KIE in γ¹⁸O₄ to the KIE in γ¹⁸O₃.

^eThe KIE in the β–γ bridge oxygen was computed as the ratio of the KIE in β¹⁸O₃ to the KIE in β¹⁸O₂.

^fThe average of KIEs in the β–γ bridge oxygen computed in the two ways.

^gThe steady state rate for Ras, 0.5 [GTP]₀/([Ras]xt_1/2); steady state rates for GAP-facilitated reactions, 0.5 [GTP]₀/([GAP]xt_1/2).

DISCUSSION

Kinetic Mechanisms and Reaction Rates for GAP-Facilitated GTP Hydrolysis

The minimal kinetic mechanism for a GAP³³⁴- or NF1³³³-facilitated GTP hydrolysis reaction can be described by eq 4.

$$\text{Ras}+\text{GTP}\underset{k_{-1}}{\overset{k_1}{\rightleftharpoons }}\text{Ras}·\text{GTP}\underset{k_{-2}}{\overset{k_2}{\rightleftharpoons }}\text{Ras}·\text{GTP}·\text{GAP}\stackrel{k_3}{\to }\text{Ras}·\text{GDP}·\text{Pi}·\text{GAP}\stackrel{k_4}{\to }\text{Ras}·\text{GDP}+\text{Pi}+\text{GAP}$$ (4)

The reaction may be considered as a system with GAP as the enzyme and Ras · GTP as the substrate. The steady where state rate takes the form of Michaelis–Menten equation as in eq 5.

$$v=\frac{k_3[\text{GAP}][\text{Ras}·\text{GTP}]}{K_m+[\text{Ras}·\text{GTP}]}$$ (5)

Under saturating or limiting concentration of Ras · GTP, the reaction rate is given by

$$v=k_3[\text{GAP}]$$ (6)

$$v=(k_3/K_m)[\text{GAP}][\text{Ras}·\text{GTP}]$$ (7)

respectively. In a typical experiment with GAP, the concentration of Ras · GTP is estimated to be around 12 µM (~25% of total Ras; Materials and Methods). The values of K_m for GAP³³⁴ and NF1³³³ were reported to be 4.8 µM and 0.13 µM, respectively (7, 27). Therefore, the steady state rates for GAP³³⁴-facilitated or NF1³³³-facilitated GTP hydrolysis can be approximated by eq 6.

Practical Considerations in the Interpretation of KIEs

A KIE is defined as a ratio of the reaction rate of the substrate containing light isotopes of the atoms of interest to that of the substrate in which the corresponding atoms are substituted by a heavier isotope (40). Isotopic substitution lowers the vibrational energy levels of a molecule to an extent that depends on geometry and bond strength. If bond order decreases at the transition state, isotopic substitution lowers the zero point vibrational energy of the ground-state more than that of the transition state, which causes an increase in the activation energy for the heavier molecule and results in a normal isotope effect (> 1). In the opposite case where the bond order in the transition state increases, the activation energy is lower for the heavier isotopomer, and an inverse isotope effect (< 1) is observed. If the labeled atoms are involved in reaction coordinates, there is also a normal (> 1) contribution to the isotope effect from the reaction coordinate. The magnitudes of the KIEs often permit qualitative or quantitative description of the transition state (41, 42).

The KIE observed for an enzyme-catalyzed reaction is a function of both the intrinsic KIE in the chemical step and the rate constants for nonchemical steps such as substrate binding and release, formation of intermediates, and conformational changes (38, 43). For the GAP³³⁴- or NF1³³³-facilitated GTP hydrolysis reaction as described by eq 4, the observed KIE is an isotope effect in V/K and is given by the following:

$${}^{18}(V/K)=\frac{{}^{18}{k}_3+c_f}{1+c_f}$$ (8)

$$c_f=(k_3/k_{-2})(1+k_2/k_{-1})$$ (9)

Equation 8 predicts that the observed KIEs are reduced, relative to the intrinsic KIEs in the chemical step (¹⁸k₃), by forward commitment (c_f), which is the ratio of the forward rate of the Ras · GTP · GAP complex (k₃) to the net rate of GTP release from this complex to the solution. There has been no evidence in the literature for reversible cleavage the P_γ–O_βγ bond; therefore, a reverse commitment is not considered here. Two terms, k₃/k₋₂ and 1+ k₂/k₋₁, contribute to c_f. The first term is an internal commitment and relies only on the kinetic properties of the Ras · GTP · GAP complex, whereas the second term depends on the concentrations of Ras · GTP and GAP. In order to maximize the observed KIE, experimental conditions were designed to minimize 1+ k₂/k₋₁. First, a guanine nucleotide exchange factor, either Cdc25^Mm285 or SOS1, was included in the reaction mixture to accelerate GTP release from Ras · GTP (k₋₁). Second, with the exception of very slow reactions, such as the hydrolysis catalyzed by Ras(Q61H)-NF1³³³, only catalytic amounts of GAP were used so that k₂ is minimized. For all reactions, the observed KIEs did not change when the concentration of GAP was varied several fold in either direction. Since k₂ is expected to depend on the free concentration of GAP, the independence of KIEs on GAP concentration suggests that1+ k₂/k₋₁ is very close to unity and does not significantly contribute to c_f.

Transition State Structure for Ras-Catalyzed GTP Hydrolysis Is Loose

We previously reported the KIEs in the GTP hydrolysis reaction catalyzed by Ras (23). Having improved our analytical method (30) and adopted different reaction conditions, we redetermined these KIEs and report the values in Table 1. In contrast to the previous experiments, which were carried out using catalytic amounts of Cdc25^Mm285 at 37 °C, stoichiometric amounts of Cdc25^Mm285 were used in the present study and the reactions were carried out at room temperature. Despite these differences in reaction conditions, the KIEs for γ¹⁸O₃-, γ¹⁸O₄-, β¹⁸O₃-, and β¹⁸O₂-labeling are in agreement with previously reported values within experimental errors.

The nature of the transition state structure for a phosphoryl transfer reaction is revealed by the KIEs in the leaving oxygen and in the geminal nonbridge oxygens (44, 45). The comparable quantities for GTP hydrolysis are the KIEs in γ¹⁸O₃- and β¹⁸O₃- labeling, which reflect the changes in bond orders of the three γ nonbridge oxygens and the extent of cleavage of the P_γ–O_βγ bond, respectively. While interpretation of the KIE in the nonbridge oxygens can be complicated by the extent of proton transfer to the γ phosphoryl group at the transition state (46), the magnitude of the KIE in the leaving group oxygen is generally accepted as a reliable indicator of the nature of the transition state (44, 47). It is estimated by quantum-mechanical calculations that the KIE in β¹⁸O₃ for a dissociative pathway is around 1.03 (York, D., personal communication). The observed KIE of 0.9994 in γ¹⁸O₃-labeling and of 1.0192 in β¹⁸O₃-labeling reveal that the Ras-catalyzed GTP hydrolysis involves a loose transition state (23).

The KIE in β¹⁸O₂-labeling is 1.0078, indicating that the P_β–O_{β nonbridge} bonds are weakened in the transition state. This reduction in bond order presumably arises through resonance structures that allow distribution of the charge, which results from cleavage of the P_γ–O_β–γ bond, onto the two β nonbridge oxygens (23). The KIEs of 1.0042 in the pro-S oxygen and of 1.0028 in pro-R β oxygen suggest that charge increases slightly more on the pro-S β oxygen than the pro-R β oxygen. Crystal structures reveal that the pro-R β oxygen is coordinated to the Mg²⁺ cofactor and the amide proton of Ser-17, whereas the pro- S β atom accepts hydrogen bonds from three P-loop amide groups and forms an ion-pair with the P-loop Lys-16. It is quite possible that the electrostatic potential is different at two nonbridge oxygen atoms and that the asymmetry in charge increase may reflect the difference.

GAP³³⁴- and NF1³³³-Stimulated GTP Hydrolysis Reactions Also Proceed through Loose Transition States

The observed KIEs of 1.0012 in γ¹⁸O₃-labeling and 1.0220 for β¹⁸O₃-labeling in GAP³³⁴-stimulated GTP hydrolysis are similar to the respective KIEs of 0.9994 and 1.0192 in the intrinsic GTPase reaction, indicating that GAP³³⁴ does not change the structure of the transition state. In addition, the magnitude of KIE in β¹⁸O₃ indicates that the forward commitment is small and that transition-state formation is still rate-limiting in theGAP³³⁴-facilitated Ras GTPase reaction.

The KIE in β¹⁸O₃ for Ras-NF1³³³ (1.0077) is much smaller than that for Ras-GAP³³⁴ (1.0220), which suggests that there is significant forward commitment in the NF1³³³-facilitated GTP hydrolysis. This assessment is consistent with biochemical properties of NF1³³³ reported in the literature. The rate of dissociation of NF1³³³ from Ras · GTP · NF1³³³ (k₋₂) was estimated to be 5.5 s⁻¹ (7), and the rate of GTP cleavage (k₃) stimulated byNF1³³³ was reported to be 5.4 s⁻¹ (27) or 19.5 s⁻¹ (7) in two separate studies. These rates predict a c_f (k₃/k₋₂) in the range between 0.93 and 3.5, which, for an intrinsic isotope effect close to 1.02, could lead to an observed KIE in the range of 1.005 to 1.01, consistent with the measured value of 1.0077. If the commitment factor is very small, as is expected for the arginine finger mutant, NF1³³³(R1276A), a much larger KIE (1.0202) is observed. Therefore, the intrinsic KIE in β¹⁸O₃ for Ras-NF1³³³ is probably as large as that for Ras-GAP³³⁴, and it follows that NF1³³³-facilitated GTP hydrolysis reactions proceed through a loose transition state as well.

Charge Asymmetry at the pro-S and pro-R β Oxygens Increases in GAP-Facilitated Reactions

The asymmetry in the KIEs in the pro-S and pro-Rβ oxygens noted for the intrinsic GTPase activity of Ras is amplified in the GAP-facilitated reactions. The KIEs in pro-S β oxygen and pro-R β oxygen for Ras-GAP³³⁴ are 1.0044 and 1.0006, respectively, indicating that there is significant increase in the charge on the pro-S oxygen but not on the pro-R β oxygen at the transition state. KIEs for Ras-NF1³³³ are 1.0012 and 0.9975 in pro-S β and pro-R β oxygen, respectively, indicating that there is a decrease rather than an increase of charge on the pro-R β oxygen. The inverse nature of the KIE in pro-R β oxygen is confirmed by the observation that the KIE in β¹⁸O₂-labeling (1.0000) is less than that in pro-S β oxygen alone (1.0012).Note also that the intrinsic KIE in the pro-R β oxygen for the Ras-NF1³³³-catalyzed reaction is more inverse than the observed KIE of 0.9975 because of the significant commitment in the reaction. In summary, GAP³³⁴- and NF1³³³-facilitated GTP hydrolysis reactions share the loose nature of the transition states but not the pattern of charge increase on the pro-S and pro-R β oxygens.

Charge Increases Prominently on the β–γ Bridge Oxygen in GAP-Facilitated Reactions

Since the charge due to the cleavage of P_γ–O_βγ bond at the transition state is shared between the β–γ bridge oxygen and the two β nonbridge oxygens, it is of mechanistic interest to compare the pattern of charge increase on the three oxygens in different reactions. Computational studies on the hydrolysis of methyl triphosphate reveals that the KIE in the β–γ bridge oxygen likely has significant contribution from the reaction coordinate (York, D., personal communication) for a concerted mechanism. Therefore, the KIE in the β–γ bridge oxygen overestimates the charge increase on the β–γ bridge oxygen relative to the charge increase on the two β nonbridge oxygens at the transition state of a reaction. Here, we compare the KIEs in reactions catalyzed by Ras and Ras mutants in the presence of wild type and mutant GAPs and infer qualitative differences in charge distribution between these reactions. Contribution to the KIE in the β–γ bridge oxygen from the reaction coordinate presumably remains the same in different reactions, and the differences in the KIEs are due to changes in bond orders only, which is directly related to charge increase. Indeed, in all cases, an inverse relationship is observed between the magnitude of the KIE in the bridge oxygen and that in β¹⁸O₂, which is consistent with conservation of charge and validates the correlation between change in charge and KIE.

The KIE in the β–γ bridge oxygen for the GAP³³⁴-facilitated GTPase reaction is significantly larger (1.0161) than that for the intrinsic GTPase reaction (1.0113) even though the KIEs in β¹⁸O₃-labeling for the two reactions are similar (1.0220 vs 1.0192). Concomitantly, the KIE in β¹⁸O₂-labeling in GAP³³⁴-facilitated GTP hydrolysis is smaller (1.0057) than that for the intrinsic reaction (1.0078). These data suggest that in GAP³³⁴-facilitated GTP hydrolysis there is more charge on the β–γ bridge oxygen and less on the two β nonbridge oxygens than in the reaction catalyzed by the intrinsic activity of Ras. Because of the substantial, but imprecisely known, commitment in the NF1³³³-facilitated GTPase reaction, the KIEs for Ras-NF1³³³ cannot be compared directly with those in the reaction catalyzed by Ras alone. However, a concentration of charge on the β–γ bridge oxygen similar to that seen in the GAP³³⁴-facilitated GTPase reaction can be deduced by comparing the KIEs in the β–γ bridge and β nonbridge oxygens. A KIE of unity (1.0000) in β¹⁸O₂ for Ras-NF1³³³ indicates that there is no overall charge increase on the two β nonbridge oxygens at the transition state. The KIE in the β–γ bridge (1.0068), however, is very close to the KIE in β¹⁸O₃-labeling (1.0077), indicating that almost all charge increase occurs on the β–γ bridge oxygen. Thus, in both GAP³³⁴-and NF1³³³-assisted reactions, charge increases on the β–γ bridge oxygen at the expense of β nonbridge oxygens as compared with the reaction catalyzed by Ras alone. A computational study also showed that GAP³³⁴ increases electrostatic potential at the bridge oxygen (28). The increase was attributed not only to the presence of the positive charge on the arginine finger but also to the conformation of Ras in the crystal structure of Ras · GDP · AlF₃:GAP³³⁴, which was found to exert higher electrostatic potential at the β–γ bridge oxygen than the conformation of Ras in the crystal structure of Ras · GPPNP (28). The KIE data, together with the computational study, indicate that even though GAPs do not change the nature of the transition state, they can change the charge distribution of the transition state.

The GAP Arginine Finger Stabilizes Charge at the β–γ Bridge Oxygen at the Transition State

In the crystal structure of Ras · GDP · AlF₃:GAP³³⁴, Arg-789 of GAP³³⁴ interacts with both the would-be β–γ bridge oxygen and one of the γ nonbridge oxygens. It is generally accepted that the arginine finger of GAP participates in catalysis by stabilizing the transition state. However, without the knowledge of the structure of the transition state, it is not clear whether Arg-789 stabilizes the negative charge on the β–γ bridge oxygen of a loose transition state (13) or the charge on one of the γ nonbridge oxygens of a tight (associative-like) transition state.

In agreement with an earlier report, we found that GAP ³³⁴(R789A) stimulates the GTPase of Ras only marginally (27). Consequently, it was not possible to measure KIEs for Ras-GAP³³⁴(R789A) because the GTP hydrolysis due to intrinsic GTPase was significant relative to the overall level of hydrolysis even in the presence of a stoichiometric amount of GAP ³³⁴(R789A). NF1³³³(R1276A) is also a weak GAP but retains sufficient activity for measurement of KIEs (27). The KIE of 1.0202 in β¹⁸O₃ for Ras-NF1³³³(R1276A) is much larger than that of 1.0077 for wild type Ras-NF1³³³, which is consistent with a much reduced c_f as a result of the reduced turnover rate (k₃). In contrast, the mutation does not cause any change in the KIE in γ¹⁸O₃, which is 0.9994 for Ras-NF1³³³(R1276A) and 0.9992 for Ras-NF1³³³. This would be expected if there is no change in mechanism because the change in commitment alone is not expected to affect a KIE that is close to unity. Therefore, we can conclude that the mutation of the arginine finger does not change the nature of the transition state of NF1³³³-facilitated GTP hydrolysis. This is in keeping with our finding that wild type GAP³³⁴ orNF1³³³, with the arginine fingers, does not change the nature of the transition state of GTP hydrolysis catalyzed by Ras alone. A similar finding was made on studies of Yersinia protein tyrosine phosphatase, where mutation of a conserved Arg residue that interacts with the phosphoryl group does not alter the transition state (47). Together, these results would seem to argue against the notion that positive charges interacting with the phosphoryl group can change the nature of the transition state of the hydrolysis reaction of phosphate monoesters.

The mutation of Arg-1276 to Ala leads to increases in the KIEs in β nonbridge oxygens. Whereas the KIE in β¹⁸O₂ (1.0000) indicates that there is no overall change of charge at the transition state on the two β nonbridge oxygens forRas-NF1³³³, the KIE in β¹⁸O₂ (1.0069) is significant for Ras-NF1³³³(R1276A). Comparison of the KIEs in the β–γ bridge oxygen can be made only after scaled by the KIEs in β¹⁸O₃ to remove the differences in commitment factors in these two reactions. The comparison of the KIEs in β¹⁸O₂ for Ras-NF1³³³(0.68%/0.77%) and for NF1³³³(R1276A) (1.34%/2.02%) suggests the mutation results in smaller charge increase at the β–γ bridge oxygen atom at the transition state. These KIEs support the conclusion that Arg-1276 is a determinant of the electrostatic potential at the β–γ bridge oxygen and the mutation of Arg-1276 to Ala lowers the electrostatic potential at the β–γ bridge oxygen. The effect of the mutation on the charge distribution on the leaving group oxygens, together with early observation that there is no charge increase on the γ-nonbridge oxygens, strongly supports the proposal that the arginine finger of GAP facilitates GTPase of Ras by stabilizing the negative charge on the β–γ bridge oxygen of a loose transition state.

Lys-16 of Ras Is Involved in Substrate Binding but Not Transition State Stabilization

Lys-16 is highly conserved in P-loop nucleotide triphosphatases (48). In crystal structures of G proteins bound to GTP and GTP analogues, the conserved lysine residue is invariably found to engage in ionic interactions with both the pro-S β nonbridge oxygen and one of the γ nonbridge oxygens (24, 49, 50). It was postulated that Lys-16 is involved in stabilizing the negative charge on the pro-S oxygen at a loose transition state (13) or the negative charge on the γ nonbridge oxygen at a tight transition state (51–53).

Ras(K16A) has not been heretofore biochemically characterized, probably because it is difficult to express in E. coli as active protein. We obtained active Ras(K16A) by coexpression with CDC25^Mm285. The complex prepared in this way exhibited GTPase activity in the presence of either GAP³³⁴ or NF1³³³. As expected, the K16A mutation impairs the GAP-facilitated GTPase activity such that roughly 200-fold higher GAP³³⁴ or 35-fold higher NF1³³³ concentration was needed to achieve a steady state rate comparable to that for wild type Ras with GAP³³⁴ or NF1³³³, respectively. Rather surprisingly, the mutation of Lys-16 to Ala does not alter the KIEs in β¹⁸O₃, β¹⁸O₂, pro-S β¹⁸O, or pro-R β¹⁸O in either GAP³³⁴- or NF1³³³-facilitated GTP hydrolysis. Inspection of eq 5 reveals that a mutation of Ras can impair the steady state rate in three ways: by reducing the turnover rate (k₃); by increasing K_m of Ras · GTP for GAP; or by reducing the affinity of Ras for GTP and thereby lowering the concentration of Ras · GTP. The first two possibilities are unlikely. A decrease in k₃ or an increase in k₋₂, which is the only realistic way to increase K_m, is expected to lower the commitment (k₃/k₋₂) and lead to a significant increase in the observed KIEs. The paradoxical lack of any effect of the Lys-16 mutation on the KIEs despite its impairment of the steady state rate can be explained by the third possibility, that is, the mutation of Lys-16 reduces the affinity of Ras for GTP and therefore effectively reduces the steady state concentration of Ras · GTP. The mutation need not affect the rate of cleavage (k₃) or the dissociation of GAP from Ras · GTP (k₋₂), thereby leaving the commitment and KIEs unchanged. This proposal is consistent with the observation that the K16A mutation does not change the KIEs in the pro-S β, pro-R β, and the β–γ bridge oxygens, which suggests that the interaction of Lys-16 with the pro-R β oxygen atom is not strong enough to affect charge distribution on the three oxygen atoms at the transition state.

The proposed effect of the K16A mutation on the affinity of Ras for GTP offers a ready explanation for the observation that NF1³³³ is more efficient than GAP³³⁴ in stimulating the GTPase activity of Ras(K16A), the reverse of the case for wild type Ras (Table 1). As a consequence of weak binding of Ras(K16A) to GTP, the concentration of Ras(K16A) · GTP is likely much lower than K_m for Ras(K16A)-GAP³³⁴ or Ras(K16A)-NF1³³³. Consequently, the reaction rate depends on k₃/K_m as described by eq 7, instead of k₃, as is the case for wild type Ras and GAP. Although k₃ is slightly higher for GAP³³⁴, NF1³³³ is the more effective enzyme toward Ras(K16A) by virtue of its lower K_m (0.1 µM vs 5 µM) (7, 27).

Although the mutation of Lys-16 to Ala does not affect the KIEs in β oxygens, it does lower the KIEs in γ¹⁸O₃ from 1.0012 for Ras-GAP³³⁴ to 0.9989 for Ras(K16A)-GAP³³⁴. A concomitant decrease from 1.0172 to 1.0140 is also observed in the KIEs for [γ¹⁸O₄]GTP, which shares the ¹⁸O-labels at the three γ nonbridge oxygen positions with [γ¹⁸O₃]GTP. The decreases of 0.0023 and 0.0032 in these KIEs are statistically significant in light of the observation that differences in KIEs in β¹⁸O₃, β¹⁸O₂, pro-S β¹⁸O, and pro-R β¹⁸O for Ras-GAP³³⁴ and Ras(K16A)-GAP³³⁴ are all less than 0.0006. In an internal competition experiment, the isotope effects on V/K are measured. As discussed above, the K16A mutation does not affect V(k₃) of the reaction. Therefore, the reductions in the KIEs in γ¹⁸O₃ and γ¹⁸O₄ likely reflect a change in the isotope effect in binding of Ras to the γ nonbridge oxygens. If so, the KIE data suggest that Lys-16 contributes as much as 0.002 or 0.003 to the isotope effect in binding.

Mutation of Gln-61 Disrupts the Stabilization of the Leaving Group Oxygens

Gln-61 of Ras is a highly conserved residue in the G protein superfamily (54). Ras proteins mutated at Gln-61 are frequently found in cancers. Such mutations are oncogenic because they impair the intrinsic and GAP-facilitated GTPase activity of Ras, which results in an abnormally prolonged state of effector activation (8). The perceived roles of Gln-61 in catalysis have been evolving ever since the crystal structure of Ras was solved. An earlier proposal that Gln-61 is involved in the activation of the water molecule was disputed on the basis that Gln-61 is a weak base (24, 51, 55, 56). Mutational and computational studies argue effectively against the proposal that Gln-61 is involved in transition state stabilization (28, 56). The crystal structure of Ras · GDP · AlF₃ · GAP³³⁴ complex suggests that Gln-61 aligns the attacking water molecule and the γ-phosphoryl group with respect to the protein (26). In the metaphosphate-as-intermediate model, Gln-61 was proposed to mediate the proton transfer from the water to the metaphosphate intermediate (16, 57).

To probe its role in catalysis, we mutated Gln-61 to His, an amino acid that still confers weak activity upon Ras in the presence ofNF1³³³ (6). The virtually identical KIEs of 0.9992 and 0.9991 in γ¹⁸O₃ for Ras-NF1³³³ and Ras(Q61H)-NF1³³³, respectively, argues against the proposed role ofGln-61 as a mediator of proton transfer in the metaphosphate-as-intermediate model. The KIE of 1.0179 in β¹⁸O₃ is much larger than that for wild type Ras and NF1³³³ (1.0077), which is consistent with a much smaller commitment as a result of the smaller turn over rate (k₃). Surprisingly, a drastic effect of the mutation was observed on the KIEs in the leaving group oxygens. KIEs in the β–γ bridge oxygen (1.0087) and in β¹⁸O₂ (1.0096), when compared with those for the corresponding reactions with wild type Ras (1.0068 and 1.0000, respectively), indicate significantly higher charge increases in the two β nonbridge oxygens and concomitantly less charge increases on the β–γ bridge oxygen. The KIE of 1.0046 in pro-S β oxygen and 1.0062 in pro-R β oxygen suggest that there is greater charge increase on the pro-Rβ oxygen than on the pro-S β oxygen, which is a reversal of the pattern observed for Ras-NF1³³³. These results suggest that the Q61H mutation leads to substantial changes in the electrostatic potential at the three leaving oxygens and, in particular, reduces the electrostatic potential at the β–γ bridge oxygen. This is surprising in light of the observation that Gln-61 does not interact with the leaving group oxygens directly in the crystal structure of Ras · GDP · AlF₃:GAP³³⁴ (29). Thus, the KIE data suggest a previously unsuspected role of Gln-61 in organizing features of the active site that are involved in stabilizing the leaving group oxygens. The side chain NH₂ group of Gln-61 and the guanidinium moiety of GAP Arg-789 form hydrogen bonds with the same γ nonbridge oxygen. In addition, the NH₂ group of the Gln-61 side chain also forms a hydrogen bond with the main chain carbonyl group of Arg-789. It is conceivable that the removal of these interactions by the mutation of Gln-61 disrupts the correct positioning of the arginine finger in the active site of Ras and renders this residue unavailable for stabilizing the transition state.

Roles of Residues of Ras and GAP in Catalysis

A dominant theme revealed by the KIEs is that the transition state for Ras-catalyzed GTP hydrolysis is loose. This is the case for reactions catalyzed by the intrinsic GTPase activity of Ras as well as GAP³³⁴- or NF1³³³-facilitated Ras GTPase activity. Likewise, hydrolysis reactions catalyzed by Ras or GAP mutants proceed through similar transition states. The KIEs in γ¹⁸O₃ are close to unity for all of the reactions. The intrinsic KIEs in β¹⁸O₃ for GAP-facilitated GTP hydrolysis are around 1.02. These KIEs are in the range typically observed for the hydrolysis of phosphomonoesters, of which GTP is one example. In accordance with the existing body of evidence, the KIEs are consistent with loose transition states, in which charge is distributed on the β–γ bridge oxygen and (by resonance effects) the β-nonbridge oxygen atoms. This charge distribution imposes specific mechanistic requirements on an enzyme that accelerates the reaction by transition state stabilization. Specifically, we expect that interactions that stabilize charge at leaving group oxygens afford preferential stabilization of a loose transition state and are conducive to catalysis. Interactions with the γ phosphoryl group, while likely important for aligning the γ phosphoryl group, attacking water, and active site residues, do not provide electrostatic stabilization of the transition state. A second theme that arises from this work is that the distribution of charge among the three leaving group oxygens can differ significantly among reactions catalyzed by Ras, Ras mutants, and Ras–GAP complexes, presumably reflecting differences in electrostatic potential at those positions. Below, we discuss the catalytic roles of several active site residues in Ras, including the Mg²⁺ cofactor, and the catalytic arginine finger in Ras GAPs in light of the pattern of charge distribution on the pro-R β, pro-S β, and the β–γ bridge oxygens.

Mg²⁺ Cofactor

In general, Mg²⁺ provides electrostatic stabilization of substrate or the transition state in enzymatic reactions as a cofactor. In the structures of G proteins in complex with GTP, Mg²⁺ interacts with the pro-R β oxygen and one of the γ nonbridge oxygens (53, 54). In the two GAP-facilitated reactions, KIEs in γ¹⁸O₃ are close to unity, whereas KIEs in the pro-R β oxygen are either unity or inverse. Thus, KIE data reveals no charge increase on the γ nonbridge oxygens or the pro- R β oxygen, suggesting that Mg²⁺ does not preferentially stabilize the transition state in GAP-facilitated GTP hydrolysis. Its role in catalysis is likely limited to orientating GTP and neutralizing its negative charge in the ground state.

Ras Lys-16

The amine group of the conserved P loop lysine residue interacts with both the pro-S β nonbridge oxygen and one of the γ nonbridge oxygens (24). The mutation of Lys-16 to Ala changes neither forward commitment nor the pattern of charge distribution on the three leaving oxygens in either GAP³³⁴- or NF1³³³-facilitated GTP hydrolysis, strongly suggesting that Lys-16 is not involved in transition state stabilization. The effect of the mutation on KIEs in γ¹⁸O₃- or γ¹⁸O₄-labeling reveals its role in GTP binding through its interaction with one of the γ nonbridge oxygens.

P-Loop Amides

The amide proton of Ras Gly-13 forms a short hydrogen bond with the β–γ bridge oxygen, whereas the amide protons of residues 14, 15, and 16 interact with the pro-S β oxygen (24). In GTP hydrolysis facilitated by both GAP³³⁴ and NF1³³³, the KIEs in the β–γ bridge oxygen and the pro-S β nonbridge oxygen are normal and significant (the observed KIEs for Ras-NF1³³³ need to be adjusted for a commitment factor), revealing charge increase on these two oxygen atoms at the transition state. Therefore, the amide protons of residues 13, 14, 15, and 16 are expected to afford preferential stabilization of the β–γ bridge oxygen and the pro-S β nonbridge oxygen at the transition state. The catalytic role of the Gly-13 amide proton was first recognized by Maegley et al. (13).

Ras Gln-61

Essential to the intrinsic or GAP-facilitated GTPase activity of Ras, Gln-61 has been proposed to align the attacking water with respect to the γ phosphoryl group or to mediate proton transfer from the attacking water to the phosphoryl group. The effect of its mutation on the KIEs in the leaving oxygens suggests that Gln-61 is important for maintaining features of the active site that stabilize the charge on the leaving group oxygen, although the mechanism by which it might provide such stabilization is not obvious. It is also possible that histidine substitution in particular is responsible for perturbations in KIE. Perhaps more important, the mutation does not change the KIE in γ¹⁸O₃, which argues against its role as a mediator of proton transfer.

GAP Arginine Finger

It is generally accepted that the arginine finger of GAP accelerates GTP hydrolysis by providing an additional level of transition state stabilization (26). The KIEs for Ras-NF1³³³ and Ras-GAP³³⁴ reveal the loose nature of the transition state and the significant charge buildup on the β–γ bridge oxygen at the transition state. Moreover, the KIEs for Ras-NF1³³³(R1276A) reveal that the mutation of the arginine finger influences the transition state charge increase on the β–γ bridge oxygen but not on the γ nonbridge oxygen. Overall, these data strongly suggest that the arginine finger facilitates the GTPase activity of Ras by stabilizing the charge on the β–γ bridge oxygen at the transition state.

CONCLUSIONS

There is considerable evidence in the literature to support the notion that GAP proteins function by reorganizing the Switch II of Ras and by electrostatically stabilizing the transition state. The allosteric effect of GAP is critical for the correct positioning of the nucleophilic water and is potentially important for the proton transfer from the nucleophilic water to the phosphate product or bulk medium. The results of the experiments presented here define the loose nature of the transition state structures and emphasize the effect of transition state stabilization afforded by GAP. The participation of GAP does not change the nature of the transition state, which remains loose in a GAP-facilitated reaction. However, GAP changes the charge distribution among the three leaving group oxygens at the transition state, suggesting that GAP increases the electrostatic potential at the β–γ bridge oxygen and the pro-S β oxygen positions. In particular, the KIE data provide strong support to the hypothesis that the arginine finger of Ras GAPs stabilizes the charge at the β–γ bridge oxygen at the transition state.