Role of an active site guanine in hairpin ribozyme catalysis probed by exogenous nucleobase rescue

The hairpin ribozyme is a small catalytic RNA with reversible phosphodiester cleavage activity. Biochemical and structural studies exclude a requirement for divalent metal cation cofactors and implicate one active site nucleobase in particular, G8, in the catalytic mechanism. Our previous work demonstrated that the cleavage activity that is lost when G8 is replaced by an abasic residue is restored when certain nucleobases are provided in solution. The specificity and pH dependence of exogenous nucleobase rescue were consistent with several models of the rescue mechanism, including general acid base catalysis, electrostatic stabilization of negative charge in the transition state or a requirement for protonation to facilitate exogenous nucleobase binding. Detailed analyses of exogenous nucleobase rescue for both cleavage and ligation reactions now allow us to refine models of the rescue mechanism. Activity increased with increasing pH for both unmodified ribozyme reactions and unrescued reactions of abasic variants lacking G8. This similarity in pH dependence argues against a role for G8 as a general base catalyst, because G8 deprotonation could not be responsible for the pH-dependent transition in the abasic variant. Exogenous nucleobase rescue of both cleavage and ligation activity increased with decreasing pH, arguing against a role for rescuing nucleobases in general acid catalysis, because a nucleobase that contributes general acid catalysis in the cleavage pathway should provide general base catalysis in ligation. Analysis of the concentration dependence of cytosine rescue at high and low pH demonstrated that protonation promotes catalysis within the nucleobase-bound ribozyme complex but does not stabilize nucleobase binding in the ground state. These results support an electrostatic stabilization mechanism in which exogenous nucleobase binding counters negative charge that develops in the transition state.     

pH Dependence of catalysis by Pseudomonas aeruginosa isochorismate-pyruvate lyase: implications for transition state stabilization and the role of lysine 42

An isochorismate-pyruvate lyase with adventitious chorismate mutase activity from Pseudomonas aerugionsa (PchB) achieves catalysis of both pericyclic reactions in part by the stabilization of reactive conformations and in part by electrostatic transition-state stabilization. When the active site loop Lys42 is mutated to histidine, the enzyme develops a pH dependence corresponding to a loss of catalytic power upon deprotonation of the histidine. Structural data indicate that the change is not due to changes in active site architecture, but due to the difference in charge at this key site. With loss of the positive charge on the K42H side chain at high pH, the enzyme retains lyase activity at ～100-fold lowered catalytic efficiency but loses detectable mutase activity. We propose that both substrate organization and electrostatic transition state stabilization contribute to catalysis. However, the dominant reaction path for catalysis is dependent on reaction conditions, which influence the electrostatic properties of the enzyme active site amino acid side chains.     

The hairpin ribozyme

The hairpin ribozyme is a naturally occurring RNA that catalyzes sequence-specific cleavage and ligation of RNA. It has been the subject of extensive biochemical and structural studies, perhaps the most detailed for any catalytic RNA to date. Comparison of the structures of its constituent domains free and fully assembled demonstrates that the RNA undergoes extensive structural rearrangement. This rearrangement results in a distortion of the substrate RNA that primes it for cleavage. This ribozyme is known to achieve catalysis employing exclusively RNA functional groups. Metal ions or other catalytic cofactors are not used. Current experimental evidence points to a combination of at least four mechanistic strategies by this RNA: (1) precise substrate orientation, (2) preferential transition state binding, (3) electrostatic catalysis, and (4) general acid base catalysis.     

Mechanistic considerations for general acid-base catalysis by RNA: revisiting the mechanism of the hairpin ribozyme

Several small ribozymes carry out self-cleavage at a specific phosphodiester bond to yield 2',3'-cyclic phosphate and 5'-hydroxyl termini. Prior mechanistic and structural studies on the HDV ribozymes led to the proposal that the pK(a) of C75 is shifted toward neutrality, making it an effective general acid. Recent mechanistic studies on the hairpin ribozyme have led to models in which protonation of G8 is required for phosphodiester cleavage, either for general acid catalysis or for electrostatic stabilization. Inspection of recent crystal structures of the hairpin ribozyme, including a complex with a vanadate transition state mimic, suggests an alternative model involving general acid-base catalysis with G8 serving as the general base and A38 as the general acid. This model is consistent with the literature on the hairpin ribozyme, including pH-rate profiles of wild-type and mutant ribozymes and solvent isotope effects. General mechanistic considerations for RNA catalysis suggest that the penalty for having general acids and bases with pK(a)s removed from neutrality is not as severe as expected. These considerations suggest that general acid-base catalysis may be a common mechanistic strategy of RNA enzymes.     

Role of Asp-9 and Glu-36 in the active site of the pneumococcal CPL1 lysozyme: an evolutionary perspective of lysozyme mechanism.

The role of carboxylic amino acids Asp-9 and Glu-36 in the activity of CPL1 lysozyme was investigated by site-directed mutagenesis. The enzymatic activity of the single mutants D9E, D9N, D9H, D9K, D9A, E36D, E36Q, E36K, and E36A and of the double mutant D9A-E36A was analyzed using a highly sensitive radioactive assay. All mutants but D6K showed detectable activities. Interestingly, the mutants E36D and E36Q retained 67% and 37% activity, respectively. Amino acid replacements at position 9 turned out to be more critical for activity than at position 36. In analogy to the mechanism described for hen egg-white lysozyme, where the proton donor play a central role, we propose that, in the CPL1 lysozyme, Asp-9 might act as the proton donor for activation of the substrate, and Glu-36 could help in the stabilization of the intermediate oxocarbocation. The residual activity of lysozyme mutants lacking one or two of the acidic amino acids may be explained by the participation of a water molecule as proton donor and/or to electrostatic contributions in the active center stabilizing the transition state of the reaction. Our results are in agreement with the hypothesis that enzymes have been optimized during evolution from an ancestral protein able to bind more tightly the transition state of the substrate than the substrate itself, by the acquisition of amino acids serving a function in catalysis.     

Transition state analysis and requirement of Asp-262 general acid/base catalyst for full activation of dual-specificity phosphatase MKP3 by extracellular regulated kinase

Dual-specificity phosphatase MKP3 down-regulates mitogenic signaling through dephosphorylation of extracellular regulated kinase (ERK). Unlike a simple substrate-enzyme interaction, the noncatalytic, amino-terminal domain of MKP3 can bind efficiently to ERK, leading to activation of the phosphatase catalytic domain by as much as 100-fold toward exogenous substrates. It has been suggested that ERK activates MKP3 through the stabilization of the active phosphatase conformation, enabling general acid catalysis. Here, we investigated whether Asp-262 of MKP3 is the bona fide general acid and evaluated its contribution to the catalytic steps activated by ERK. Using site-directed mutagenesis, pH rate and Br?nsted analyses, kinetic isotope effects, and steady-state and rapid reaction kinetics, Asp-262 was identified as the authentic general acid catalyst, donating a proton to the leaving group oxygen during P-O bond cleavage. Kinetic isotope effects [(18)(V/K)(bridge), (18)(V/K)(nonbridge), and (15)(V/K)] were evaluated for the effect of ERK and of the D262N mutation on the transition state of the phosphoryl transfer reaction. The patterns of the three isotope effects for the reaction with native MKP3 in the presence of ERK are indicative of a reaction where the leaving group is protonated in the transition state, whereas in the D262N mutant, the leaving group departs as the anion. Even without general acid catalysis, the D262N mutant reaction is activated by ERK through increased phosphate affinity ( approximately 8-fold) and the partial stabilization of the transition state for phospho-enzyme intermediate formation ( approximately 4-fold). Based on these analyses, we estimate that dephosphorylation of phosphorylated ERK by the D262N mutant is >1000-fold lower than by native, activated MKP3. Also, the kinetic results suggest that Asp-262 functions as a general base during thiol-phosphate intermediate hydrolysis.     

Crystallographic determination of the mode of binding of oligosaccharides to T4 bacteriophage lysozyme: Implications for the mechanism of catalysis

Phage lysozyme has catalytic activity similar to that of hen egg white lysozyme, but the amino acid sequences of the two enzymes are completely different.The binding to phage lysozyme of several saccharides including N-acetylglucosamine (GlcNAc), N-acetylmuramic acid (MurNAc) and (GlcNAc)₃ have been determined crystallographically and shown to occupy the pronounced active site cleft. GlcNAc binds at a single location analogous to the C site of hen egg white lysozyme. MurNAc binds at the same site. (GlcNAc)₃ clearly occupies sites B and C, but the binding in site A is ill-defined.Model building suggests that, with the enzyme in the conformation seen in the crystal structure, a saccharide in the normal chair configuration cannot be placed in site D without incurring unacceptable steric interference between sugar and protein. However, as with hen egg white lysozyme, the bad contacts can be avoided by assuming the saccharide to be in the sofa conformation. Also Asp20 in T4 lysozyme is located 3 Å from carbon C₍₁₎ of saccharide D, and is in a position to stabilize the developing positive charge on a carbonium ion intermediate. Prior genetic evidence had indicated that Asp20 is critically important for catalysis. This suggests that in phage lysozyme catalysis is promoted by a combination of steric and electronic effects, acting in concert, The enzyme shape favors the binding in site D of a saccharide with the geometry of the transition state, while Asp20 stabilizes the positive charge on the oxocarbonium ion of this intermediate. Tn phage lysozyme, the identity of the proton donor is uncertain. In contrast to hen egg white lysozyme, where Glu35 is 3 Å from the glycosidic DOE bond, and is in a non-polar environment, phage lysozyme has an ion pair, Glull … Arg145, 5 Å away from the glycosidic oxygen. Possibly Glull undergoes a conformational adjustment in the presence of bound substrate, and acts as the proton donor. Alternatively, the proton might come from a bound water molecule.     

Catalytic roles for proton transfer and protonation in ribozymes

Utilization of proton transfer in catalysis, which is well known in the mechanisms of protein enzymes, has been described only relatively recently for RNA enzymes. In this article, we present a current understanding of proton transfer by nucleic acids. Rate enhancement and specificity conferred by general acid-base catalysis are discussed. We also present possibilities for electrostatic catalysis from general acids and bases as well as cationic base pairs. The microenvironments of a large RNA provide the possibility of histidine-like pK(a)s for proton transfer, as well as lysine- and arginine-like pK(a)s for electrostatic catalysis. Discussion on proton transfer focuses on the hepatitis delta virus (HDV) and hairpin ribozymes, with select examples drawn from the protein literature. Discussion on electrostatic catalysis also draws on these two ribozymes, and a postulate for electrostatic catalysis by a cationic base pair in the mechanism of peptidyl transfer in the ribosome is presented. We also provide a perspective on possibilities for phosphoryl transfer mechanisms involving phosphorane intermediates and unusual tautomeric forms of the bases. Lastly, a distinction is made between ground state and "transition state" pK(a)s. We favor a model in which changes in pH lead to changes in the distribution of reactive and nonreactive ionizations of the ribozyme molecules in the ground state, and therefore suggest that "pK(a) changes in the transition state" do not provide an acceptable explanation for observed pH-rate profiles.     

Powering DNA repair through substrate electrostatic interactions

The reaction catalyzed by the DNA repair enzyme uracil DNA glycosylase (UDG) proceeds through an unprecedented stepwise mechanism involving a positively charged oxacarbenium ion sugar and uracil anion leaving group. Here we use a novel approach to evaluate the catalytic contribution of electrostatic interactions between four essential phosphodiester groups of the DNA substrate and the cationic transition state. Our strategy was to substitute each of these phosphate groups with an uncharged (R)- or (S)-methylphosphonate linkage (MeP). We then compared the damaging effects of these methylphosphonate substitutions on catalysis with their damaging effects on binding of a cationic 1-azadeoxyribose (1-aza-dR(+)) oxacarbenium ion analogue to the UDG-uracil anion binary complex. A plot of log k(cat)/K(m) for the series of MeP-substituted substrates against log K(D) for binding of the 1-aza-dR(+) inhibitors gives a linear correlation of unit slope, confirming that the electronic features of the transition state resemble that of the 1-aza-dR(+), and that the anionic backbone of DNA is used in transition state stabilization. We estimate that all of the combined phosphodiester interactions with the substrate contribute 6-8 kcal/mol toward lowering the activation barrier, a stabilization that is significant compared to the 16 kcal/mol catalytic power of UDG. However, unlike groups of the enzyme that selectively stabilize the charged transition state by an estimated 7 kcal/mol, these phosphodiester groups also interact strongly in the ground state. To our knowledge, these results provide the first experimental evidence for electrostatic stabilization of a charged enzymatic transition state and intermediate using the anionic backbone of DNA.     

Proton relay mechanism of general acid catalysis by DNA topoisomerase IB.

Type IB topoisomerases cleave and rejoin DNA through a DNA-(3'-phosphotyrosyl)-enzyme intermediate. A constellation of conserved amino acids (Arg-130, Lys-167, Arg-223, and His-265 in vaccinia topoisomerase) catalyzes the attack of the tyrosine nucleophile (Tyr-274) at the scissile phosphodiester. Previous studies implicated Arg-223 and His-265 in transition state stabilization and Lys-167 in proton donation to the 5'-O of the leaving DNA strand. Here we find that Arg-130 also plays a major role in leaving group expulsion. The rate of DNA cleavage by vaccinia topoisomerase mutant R130K, which was slower than wild-type topoisomerase by a factor of 10(-4.3), was stimulated 2600-fold by a 5'-bridging phosphorothiolate at the cleavage site. The catalytic defect of the R130A mutant was also rescued by the 5'-S modification (190-fold stimulation), albeit to a lesser degree than R130K. We surmise that Arg-130 plays dual roles in transition state stabilization and general acid catalysis. Whereas the R130A mutation abolishes both functions, R130K permits the transition state stabilization function (via contact of lysine with the scissile phosphate) but not the proton transfer function. Our results show that the process of general acid catalysis is complex and suggest that Lys-167 and Arg-130 comprise a proton relay from the topoisomerase to the 5'-O of the leaving DNA strand.     

A thermodynamic approach to the problem of stabilization of globular protein structure: a calorimetric study

The thermal properties of five globular proteins with known spatial structure, ribonuclease, lysozyme, chymotrypsin, cytochrome c and myoglobin, are investigated by scanning microcalorimetry. It is shown that (a) heat-denaturation of these proteins can be described to a first approximation by the two-state transition model; the deviation from this model does not exceed 5% and seems to be due to the existence of highly unstable intermediates; (b) there is an interdependence between the enthalpy and entropy of transition, and the structural parameters of the protein globule (the hydrogen bond content and saturation by contacts between non-polar groups). The Gibbs energy of stabilization of the native structure is determined as a function of temperature.     

Histidine-40 of ribonuclease T1 acts as base catalyst when the true catalytic base, glutamic acid-58, is replaced by alanine

Mechanisms for the ribonuclease T1 (RNase T1; EC 3.1.27.3) catalyzed transesterification reaction generally include the proposal that Glu58 and His92 provide general base and general acid assistance, respectively [Heinemann, U., & Saenger, W. (1982) Nature (London) 299, 27-31]. This view was recently challenged by the observation that mutants substituted at position 58 retain high residual activity; a revised mechanism was proposed in which His40, and not Glu58, is engaged in catalysis as general base [Nishikawa, S., Morioka, H., Kim, H., Fuchimura, K., Tanaka, T., Uesugi, S., Hakoshima, T., Tomita, K., Ohtsuka, E., & Ikehara, M. (1987) Biochemistry 26, 8620-8624]. To clarify the functional roles of His40, Glu58, and His92, we analyzed the consequences of several amino acid substitutions (His40Ala, His40Lys, His40Asp, Glu58Ala, Glu58Gln, and His92Gln) on the kinetics of GpC transesterification. The dominant effect of all mutations is on Kcat, implicating His40, Glu58, and His92 in catalysis rather than in substrate binding. Plots of log (Kcat/Km) vs pH for wild-type, His40Lys, and Glu58Ala RNase T1, together with the NMR-determined pKa values of the histidines of these enzymes, strongly support the view that Glu58-His92 acts as the base-acid couple. The curves also show that His40 is required in its protonated form for optimal activity of wild-type enzyme. We propose that the charged His40 participates in electrostatic stabilization of the transition state; the magnitude of the catalytic defect (a factor of 2000) from the His40 to Ala replacement suggests that electrostatic catalysis contributes considerably to the overall rate acceleration. For Glu58Ala RNase T1, the pH dependence of the catalytic parameters suggests an altered mechanism in which His40 and His92 act as base and acid catalyst, respectively. The ability of His40 to adopt the function of general base must account for the significant activity remaining in Glu58-mutated enzymes.     

The active site of hydroxynitrile lyase from Prunus amygdalus: Modeling studies provide new insights into the mechanism of cyanogenesis

The FAD-dependent hydroxynitrile lyase from almond (Prunus amygdalus, PaHNL) catalyzes the cleavage of R-mandelonitrile into benzaldehyde and hydrocyanic acid. Catalysis of the reverse reaction-the enantiospecific formation of alpha-hydroxynitriles--is now widely utilized in organic syntheses as one of the few industrially relevant examples of enzyme-mediated C-C bond formation. Starting from the recently determined X-ray crystal structure, systematic docking calculations with the natural substrate were used to locate the active site of the enzyme and to identify amino acid residues involved in substrate binding and catalysis. Analysis of the modeled substrate complexes supports an enzymatic mechanism that includes the flavin cofactor as a mere "spectator" of the reaction and relies on general acid/base catalysis by the conserved His-497. Stabilization of the negative charge of the cyanide ion is accomplished by a pronounced positive electrostatic potential at the binding site. PaHNL activity requires the FAD cofactor to be bound in its oxidized form, and calculations of the pKa of enzyme-bound HCN showed that the observed inactivation upon cofactor reduction is largely caused by the reversal of the electrostatic potential within the active site. The suggested mechanism closely resembles the one proposed for the FAD-independent, and structurally unrelated HNL from Hevea brasiliensis. Although the actual amino acid residues involved in the catalytic cycle are completely different in the two enzymes, a common motif for the mechanism of cyanogenesis (general acid/base catalysis plus electrostatic stabilization of the cyanide ion) becomes evident.     

Effects on general acid catalysis from mutations of the invariant tryptophan and arginine residues in the protein tyrosine phosphatase from Yersinia

General acid catalysis in protein tyrosine phosphatases (PTPases) is accomplished by a conserved Asp residue, which is brought into position for catalysis by movement of a flexible loop that occurs upon binding of substrate. With the PTPase from Yersinia, we have examined the effect on general acid catalysis caused by mutations to two conserved residues that are integral to this conformation change. Residue Trp354 is at a hinge of the loop, and Arg409 forms hydrogen bonding and ionic interactions with the phosphoryl group of substrates. Trp354 was mutated to Phe and to Ala, and residue Arg409 was mutated to Lys and to Ala. The four mutant enzymes were studied using steady state kinetics and heavy-atom isotope effects with the substrate p-nitrophenyl phosphate. The data indicate that mutation of the hinge residue Trp354 to Ala completely disables general acid catalysis. In the Phe mutant, general acid catalysis is partially effective, but the proton is only partially transferred in the transition state, in contrast to the native enzyme where proton transfer to the leaving group is virtually complete. Mutation of Arg409 to Lys has a minimal effect on the K(m), while this parameter is increased 30-fold in the Ala mutant. The k(cat) values for R409K and for R409A are about 4 orders of magnitude lower than that for the native enzyme. General acid catalysis is rendered inoperative by the Lys mutation, but partial proton transfer during catalysis still occurs in the Ala mutant. Structural explanations for the differential effects of these mutations on movement of the flexible loop that enables general acid catalysis are presented.     

Evaluation of weak interactions of proteins and organic cations with DNA duplex structures

Molecular interactions and reactions in living cells occur with high background concentrations of organic compounds including proteins. Uncharged water-soluble polymers are commonly used cosolutes in studies on molecular crowding, and most studies argue about the effects of intracellular crowding based on results obtained using polymer cosolutes. Further investigations using protein crowders and organic cations are important in understanding the effects of cellular environments on nucleic acids with negatively charged surfaces. We assessed the effects of using model globular proteins, serum proteins, histone proteins, structurally flexible polypeptides, di- and polyamines, and uncharged polymers. Thermal stability analysis of DNA oligonucleotide structures revealed that unlike conventional polymer cosolutes, basic globular proteins (lysozyme and cytochrome c) at high concentrations stabilized long internal and bulge loop structures but not fully matched duplexes. The selective stabilization of long loop structures suggests preferential binding to unpaired nucleotides in loops through weak electrostatic interactions. Furthermore, the ability of the proteins to stabilize the loop structures was enhanced under macromolecular crowding conditions. Remarkably, the effects of basic proteins on the stability of fully matched duplexes were dissimilar to those of basic amino-acid-rich polypeptides and polyamines. This study provides new insights into the interaction of nucleic acid structures with organic cations.     

Examining the case for the effect of barrier compression on tunneling, vibrationally enhanced catalysis, catalytic entropy and related issues

The idea that tunneling is enhanced by the compression of the donor-acceptor distance has attracted significant interest. In particular, recent studies argued that this proposal is consistent with pressure effects on enzymatic reactions, and that the observed pressure effects support the idea of vibrationally enhanced catalysis. However, a careful analysis of the current works reveals serious inconsistencies in the evidence presented to support these hypotheses. Apparently, tunneling decreases upon compression, and external pressure does not lead to the applicable compression of the free energy surface. Additionally, pressure experiments do not provide actual evidence for vibrationally enhanced catalysis. Finally, the temperature dependence of the entropy change in hydride transfer reactions is shown to reflect simple electrostatic effects.     

Electrostatic contributions to binding of transition state analogues can be very different from the corresponding contributions to catalysis: phenolates binding to the oxyanion hole of ketosteroid isomerase

The relationship between binding of transition state analogues (TSAs) and catalysis is an open problem. A recent study of the binding of phenolate TSAs to ketosteroid isomerase (KSI) found a small change in the binding energy with a change in charge delocalization of the TSAs. This has been taken as proof that electrostatic effects do not contribute in a major way to catalysis. Here we reanalyze the relationship between the binding of the TSAs and the chemical catalysis by KSI as well as the binding of the transition state (TS), by computer simulation approaches. Since the simulations reproduce the relevant experimental results, they can be used to quantify the different contributions to the observed effects. It is found that the binding of the TSA and the chemical catalysis represent different thermodynamic cycles with very different electrostatic contributions. While the binding of the TSA involves a small electrostatic contribution, the chemical catalysis involves a charge transfer process and a major electrostatic contribution due to the preorganization of the active site. Furthermore, it is found that the electrostatic preorganization contributions to the binding of the enolate intermediate of KSI and the TS are much larger than the corresponding effect for the binding of the TSAs. This reflects the dependence of the preorganization on the orientation of the nonpolar form of the TSAs relative to the oxyanion hole. It seems to us that this work provides an excellent example of the need for computational studies in analyzing key experimental findings about enzyme catalysis.     

Role of His101 in the Protein Folding/Unfolding of a Goose-Type Lysozyme from Ostrich (<Emphasis Type="Italic">Struthio camelus</Emphasis>) Egg White

To understand the role of His101 in protein structure stabilization of goose-type (G-type) lysozyme, we conducted thermal unfolding/refolding experiments using native G-type lysozyme from ostrich egg white (nOEL), the recombinant G-type lysozyme (rOEL), and the mutant lysozyme, in which His101 is mutated to alanine (H101A-OEL). Thermal stability on lytic activity and in-gel refolding experiments provided similar profiles for all three OELs. Circular dichroism (CD) spectroscopy was used to determine the secondary structure of three OELs as a function of temperature. Unfolding/refolding experiments (30–90 °C) monitored by CD spectroscopy revealed an unfolding transition at 65–67 °C and a complete refolding at almost the same temperature. Notably, a slightly lower thermal stability was observed for H101A-OEL, corresponding to the calculated difference in transition free energy of thermal unfolding (??G _m) between rOEL and H101A-OEL of ?0.63 kcal/mol. To assess the effects of H101A mutation on the electrostatic behavior, we examined the pH-activity profile of the three OELs. nOEL and rOEL exhibit bimodal relationship between pH and lytic activity showing optima at pH 3.0 and 7.0, while optima for H101A-OEL activity were pH 4.0 and 6.0. Electrostatic environment surrounding His101 was affected by the H101A mutation resulting in the slightly lower thermal stability.     

Proton transfer at carbon

The viability of living systems requires that C--H bonds of biological molecules be stable in water, but that there also be a mechanism for shortening the timescale for their heterolytic cleavage through enzymatic catalysis of a variety of catabolic and metabolic reactions. An understanding of the mechanism of enzymatic catalysis of proton transfer at carbon requires the integration of results of studies to determine the structure of the enzyme-substrate complex with model studies on the mechanism for the non-enzymatic reaction in water, and the effect of the local protein environment on the stability of the transition state for this reaction. A common theme is the importance of electrostatic interactions in providing stabilization of bound carbanion intermediates of enzyme-catalyzed proton-transfer reactions.