Interaction vs Preorganization in Enzyme Catalysis. A Dispute That

Reviews Cite This: ACS Chem. Biol. 2019, 14, 1386−1392

pubs.acs.org/acschemicalbiology

Interaction vs Preorganization in Enzyme Catalysis. A Dispute That Calls for Resolution Fredric M. Menger*,† and Faruk Nome‡ †

Department of Chemistry, Emory University, Atlanta, Georgia 30322, United States Departamento de Química, Universidade Federal de Santa Catarina, Florianópolis, SC 88040-900 Brazil

Downloaded via UNIV OF SOUTHERN INDIANA on July 21, 2019 at 18:02:40 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

ABSTRACT: This essay focuses on the debate between Warshel et al. (proponents of preorganization) and Menger and Nome (proponents of spatiotemporal effects) over the source of fast enzyme catalysis. The Warshel model proposes that the main function of enzymes is to push the solvent coordinate toward the transition state. Other physical-organic factors (e.g., desolvation, entropic effects, ground state destabilization, etc.) do not, ostensibly, contribute substantially to the rate. Indeed, physical organic chemistry in its entirety was claimed to be “irrelevant to an enzyme’s active site”. Preorganization had been applied by Warshel to his “flagship” enzyme, ketosteroid isomerase, but we discuss troubling issues with their ensuing analysis. For example, the concepts of “general acid” and “general base”, known to play a role in this enzyme’s mechanism, are ignored in the text. In contrast, the spatiotemporal theory postulates that enzyme-like rates (i.e., accelerations >108) occur when two functionalities are held rigidly at contact distances less than ca. 3 Å. Numerous diverse organic systems are shown to bear this out experimentally. Many of these are intramolecular systems where distances between functionalities are known. Among them are fast intramolecular systems where strain is actually generated during the reaction, thereby excluding steric compression as a source of the observed enzyme-like rates. Finally, the account ends with structural data from four active sites of enzymes, obtained by others, all showing contact distances between substrate analogues and enzyme. To our knowledge, contact distances less than the diameter of water are found universally among enzymes, and it is to this fact that we attribute their extremely fast rates given the assumption that enzymes, whatever their particular mechanism, obey elementary chemical principles.

I

enzyme and substrate (which is what was believed by most people), but rather to a large free-energy penalty for the reorganization of the solvent in the reference reaction without the enzyme.”3 Water is ostensibly preorganized at the enzyme’s active site to preclude this energy cost. In this manner, Warshel has managed to discard much of modern bio-organic chemistry. His philosophy is stated emphatically in another Nobel lecture quote in which he states that electrostatic preorganization “helped me to explore (and frequently to eliminate) popular suggestions of factors that presumably lead to enzyme catalysis, such as entropic effects, ground state destabilization by desolvation, dynamical effects, orbital steering and more.”3 An honor-role of Warshel-jettisoned catalytic entities should also include propinquity, induced fit, distortion, charge-transfer, covalent catalysis, low-barrier hydrogen-bonds, and tunneling. Specific mention of our spatiotemporal theory2 per se seems, somehow, to have escaped the discussion. The Warshel model asserts, correctly, that solvent dipoles reorient themselves as reactants in solution move toward the transition state. This comes at the cost of free energy. Less activation energy is required at the active site of enzymes

n recent months a pronounced split has emerged among those interested in understanding the mechanism of enzyme action. On the one hand, the Warshel school maintains that enzyme catalysis results from water molecules preorganized at the active site so as to stabilize the transition state. Opposition to this view comes from ourselves and others who believe that enzyme catalysis stems from spatiotemporal effects based on geometric factors. Enzyme catalysis is such a fundamental component of living systems that, if at all possible, the dispute should not be allowed to linger. It is toward this end that the current paper has been written. A major portion of the text (following a brief discussion of the Warshel proposal) consists of experiments from our laboratories, new and old, and those of others, all supporting spatiotemporal notions. Two key 2017 papers summarize elements of the opposing positions: Jindal and Warshel, entitled “Misunderstanding the preorganization concept can lead to confusions about the origin of enzyme catalysis.”1 And a paper by Nome, Menger, and co-workers entitled “Transforming a Stable Amide into a Highly Reactive One: Capturing the Essence of Enzyme Catalysis.”2 Below we amplify our views of the two distinct enzyme mechanisms.

■

PREORGANIZATION

Received: November 27, 2018 Accepted: May 31, 2019 Published: May 31, 2019

In his Nobel lecture, Warshel claims (and we quote): “Enzyme catalysis is not due to the interaction [italics ours] between © 2019 American Chemical Society

1386

DOI: 10.1021/acschembio.8b01029 ACS Chem. Biol. 2019, 14, 1386−1392

Reviews

ACS Chemical Biology

active site.”5 (b) “Thus, it is useful to note that in most cases the main effect of the enzyme is to push the solvent coordinate toward the TS.”6 (c) “The catalytic power of enzymes is not related directly to the binding power of proteins.” “Correlated motion of an enzyme does not necessarily contribute to catalysis.” Catalysis... does not include van der Waals strain effects, charge-transfer covalent interactions, orientational entropy, and dynamical effects.” “The key to the ability to figure out the secret of enzyme catalysis... is close to impossible when one is using experimental approaches.”5 (d) “The ground states of enzyme−substrate complexes have significant electrostatic strain.”7 (e) “The interaction energy at the TS is not the origin of the catalytic effect.”1 (f) “Electrostatic preorganization effects provide the major contributing factor in enzyme catalysis and... other factors cannot give large contributions.”4 Note that we would never apply the terms “correct” or “incorrect” to Warshel’s electrostatic preorganization theory or, for that matter, to our own spatiotemporal notions. Both descriptors are inappropriate because models are, almost by definition, only partial representations of reality. Opposing models might even, to a certain extent, overlap. With this concession to the limitations of modeling (both computational and experimental), it is now time to cast a level gaze at chemical reactivity from a spatiotemporal point of view. This will then be followed by consideration of enzymes themselves.

where dipoles are, it is claimed, already favorably oriented (a phenomenon termed “preorganization”). Thus, an enzymecatalyzed reaction, paying a lower organization penalty, proceeds at a faster rate. Just how effective and exclusive the mechanism, primarily based as it is on permanent dipoles, are issues about to be explored. Consider one of Warshel’s flagship enzymes, ketosteroid isomerase, where computations were supported by thermodynamic data (e.g., pKa’s and binding constants).4 The enzyme is said to “offer perhaps the best system for illustrating the preorganization effect.”4 In any event, the first step in the enzyme reaction involves an aspartate carboxylate catalyzing the enolization of a ketone, with the barrier diminishing from 21.9 kcal/mol in water to 10.3 kcal/mol within the enzyme. The origin of this catalytic effect was attributed, in part, to the “electrostatic stabilization” of the enolate by hydrogen-bonds to Tyr57 and Asp103. Bio-organic chemists call the effect a “general-base/general-base catalysis” although the term per se is not found in the Warshel article. Warshel’s shunning classical physical-organic concepts would be more acceptable if all these concepts had each been specifically incorporated into his electrostatics model, yet even that would have sequestered valuable insights. Water molecules are a principal focus of the Warshel construct, protein group reorganization at the active site being given scant attention in the text. Water was represented by point-dipoles (an expedient necessary for computational speed and simplicity, but obviously contributing further to the uncertainty). Typical solvent reorganization energies are only a few hundred wavenumbers, while their reorganization reaction times fall into the femtosecond to picosecond range. Such lowenergy solvation dynamics seem an unlikely source of an orders-of-magnitude slower enzyme kinetics. Figuratively speaking, butterflies cannot push a racehorse. Finally, we should point out that dielectric constants for active sites are unknown, which is particularly problematic for a theory based exclusively upon electrostatics. There is perhaps an even more perplexing problem with electrostatic preorganization being biology’s chief catalytic mechanism. Models are to be used, not believed. For a model to be used, it must be useful, and to be useful it must have predictive value. But electrostatic preorganization is a theoretical construct creating structures that are difficult to test via palpable organic systems. How, after all, would one design and synthesize a particular 3-dimensional array of immobilized water dipoles? But, as will be seen, spatiotemporal theory, being a child of organic chemistry, is different in that it deals with functional-group assemblies of known geometric disposition and proven predictive power. An optimally positioned carboxyl group, for example, is viewed by us for its general acid potential and not merely as a “spectatordipole.” In any scientific dispute, it is necessary to guard against misconstruing the other’s position. This is why simple declarative sentences written by Warshel are quoted verbatim (as in our third paragraph). Direct quotes minimize our inadvertent use of polemics and so-called “misunderstandings.”1 Here, then, are some additional quotes that further illuminate Warshel’s thinking behind the preorganization concept. The quotes are necessarily out of context and yet informative and absolutely unambiguous. (a) “The problem has, however, been that physical organic chemistry experiments in solution might have been rather irrelevant to an enzyme

■

SPATIOTEMPORAL HYPOTHESIS Five past reviews of Menger describe the fundamentals of spatiotemporal theory. In order of appearance, they are entitled: “On the Source of Intramolecular and Enzymatic Reactivity”;8 “Nucleophilicity and Distance”;9 “Organic Reactivity and Geometric Disposition”;10 “Enzyme Reactivity from an Organic Perspective”;11 and “An Alternative View of Enzyme Catalysis.”12 Over the years, numerous additional experimental papers have appeared from the Nome group that strongly support the ideas in the reviews.13−16 The fact is that spatiotemporal theory is not a complex idea (one of its virtues). It simply states that enzyme-like rates (108−1012 or greater accelerations) can occur when two reactants are held rigidly at favorable geometries and at close “contact distances” (defined as less than 3 Å, the diameter of water). An array of effects, including electrostatic stabilization and environmental factors, are all amplified as a direct result of imposed contacts. Pedagogical experience tells us that one of the best ways to drive home a concept is to provide concrete examples and, with this in mind, several organic systems having remarkable enzyme-like rates are now discussed. Aspartic proteinase enzymes, which include HIV-1 proteinase, function with two aspartate carboxyls at the active site. The enzyme has been modeled with a system possessing an unactivated amide positioned between two carboxyls (see 1 below).2 Now simple amides (such as benzamide with a halflife of 3 h, 85 °C, 5.9% H2SO4) are stable under physiological conditions.17 But 1 was found to hydrolyze so rapidly relative to a benzamide reference that, owing to the presence of the carboxyl functionalities, the compound could not be isolated at 35 °C and pH = 4. Our kinetic results are consistent with spatiotemporal concepts in which two carboxylic moieties, held in a proper disposition at distances from the amide carbonyl less than the diameter of water, lead to a rate rivaling aspartic proteinase. Indeed, Jindal and Warshel themselves calculated a 17.4 kcal/mol reduction in activation energy of 1 resulting from presence of two well-positioned carboxyls.1 It is believed 1387

DOI: 10.1021/acschembio.8b01029 ACS Chem. Biol. 2019, 14, 1386−1392

Reviews

ACS Chemical Biology

distances. Kirby found, for example, that juxtapositioning an amino group and a carbon−carbon double bond leads to rapid addition with a half-life of 3.3 s at 25 °C (eq 1).18 Although the

that the impressive catalytic effect derives from one carboxyl acting as a nucleophile toward the amide carbonyl, while concurrently the other carboxyl transfers a proton to the amide nitrogen. Note that the design of the enzyme model was motivated exclusively by geometric factors independent of any solvent-reorganization considerations. Our results do not prove that aspartic proteinase operates similarly to our model, but the results do substantiate the notion that enforced interactions at close distances provide a reasonable and sufficient accelerative route available in principle to all enzymes.

exact rate acceleration cannot be determined (the corresponding intermolecular reaction, normally used as a reference, is too slow to measure), the rate enhancement attributable to spatiotemporal factors certainly falls within the enzyme range (>108). Even a relatively small degree of rotational mobility in the system, as in eq 2, can allow the functionalities to escape the required spatiotemporal contact, whereupon ring closure occurs but only at a very slow rate (2 h at 118 °C).18

We used dynamic NMR methods to measure NH/CH exchange in 3 (R = H).19 The reaction is believed to occur via proton-transfer to form an R3C−/H3NR+ ion-pair, after which a different NH proton returns to the carbon. It was found that the intramolecular base-catalyzed proton exchange is too fast to measure by NMR at −80 °C. In contrast, the reference intermolecular counterpart (where the amine and carbon acid are each maintained at 30 mM) was too slow to measure by NMR at 100 °C, a full 180 °C higher temperature. Clearly, spatiotemporal effects are profound.

An impressive catalytic effect was also observed in the hydrolysis of a phosphate diester, bis(2-(1-methyl-1Himidazolyl)phenyl) phosphate 2, a compound with two imidazoles (as has also the RNase-A enzyme) and a phosphate ester group in the same molecule.

The enzyme models presented thus far evoke a serious concern. It is possible that a compressive force between the reactive functionalities has been unwittingly introduced into the molecules during their synthesis. This could give rise to a so-called “steric acceleration” in which release of strain serves to drive the reaction. Since little hard evidence exists that enzyme catalysis itself exploits strain effects, our observed rate accelerations, although remarkably large, might in fact be irrelevant to enzymes. To address this problem, we have examined models where reactions enhance strain rather than relieve it. An example comes from the literature (eq 3).20 Thus,

The kinetics of the system are consistent with one of the imidazole groups acting as a general-base or nucleophile and the other as a general-acid. General-base catalysis and proton delivery from the general-acid catalyst to the phosphate oxygen are likely concerted. The distance in 2 between the proton accepting and donating nitrogen atoms in the reactive conformation is about 7 Å, which closely resembles the equivalent distance in RNase-A (∼6.5 Å). Distances between the catalytic imidazole groups in the enzyme and the targeted phosphate group of the substrate are fixed in a supramolecular enzyme/substrate complex. Although there is more flexibility in 2 than at the active site, 2 can transiently achieve intramolecular distances equivalent to those in the supramolecular enzyme−substrate complex. Other groups have also reported spectacular rate effects when a molecular framework holds two groups at contact

the observed 2 × 105 acceleration of the SN2 reaction in the azanorbornane (relative to the simple acyclic counterpart) 1388

DOI: 10.1021/acschembio.8b01029 ACS Chem. Biol. 2019, 14, 1386−1392

Reviews

ACS Chemical Biology

backdrop. The pillar[5]arene given below (eq 5), a catalyst that binds and splits phosphate esters, is a case in point.

must be ascribed solely to the enforced residency at bonding distances, i.e., a spatiotemporality free from steric acceleration. This follows from the fact that ring-strain is generated (not released) during the reaction. Correcting for strain-production would only increase the observed rate enhancement value of 2 × 105. A similar argument can be made for the transannular SN2 reaction in eq 4.21 Ab initio calculations reveal that the

rate of internal displacement is at the vibrational level despite an accompanying increase in ring-strain energy of ca. 40 kcal/ mol. The inescapable conclusion is that spatiotemporal rate increases cannot be dismissed as mere expressions of classical strain effects. A useful theory must not only rationalize results; it must predict them. If the spatiotemporal hypothesis has any merit, it should be possible to apply it to the a priori design of new systems that react at enzyme-like rates. We present now two conceptually different examples. We had the occasion to examine amide 4 via molecular mechanics calculations.22 The calculations told us that the molecule possesses two low-energy conformations, both of which have a carboxyl oxygen within van der Waals contact distance of the amide carbonyl carbon (2.76 and 2.80 Å, respectively). Thus, a carboxyl is poised for synchronous nucleophilic attack above the plane of the carbonyl accompanied by proton transfer. Since on this basis we predicted via spatiotemporal theory that amide cleavage should be enzyme-like, 4 was synthesized for kinetic examination. It was found that, indeed, the amide hydrolyzes at neutrality with an acceleration (expressed here as a so-called “effective molarity” = kintra/kinter) of at least 1012. An upper limit of a mere 3 kcal/mol (or 102 in rate) might conceivably be attributed to possible steric acceleration. Any steric interactions greater than this would have caused axial-to-equatorial ring-flipping, thereby placing the axial ring substituents in unreactive equatorial positions. The main point from the preceding paragraph, however, is that spatiotemporality has predictive power lacking in the more opportunistic waterpreorganization theory. Thus, geometic considerations led to predicted enzyme-like rates, a prediction subsequently verified by experiment.

It must be stated, first of all, that the conversion from intramolecularity to intermolecularity introduces a new problem: How can one control the orientations of reactant and catalytic functionalities within the complex in order to ensure contact-distances and, therefore, enzyme-like rates? The pillar[5]arene is seen to possess five imidazole groups positioned at 72° intervals around both the upper and lower edges of the macrocycle. Consequently, docking of the phosphate diester substrate into the macrocycle’s cavity would seem to favor the probability of a reaction between the ester and at least one of the imidazoles. Key physicalorganic results are given in the next paragraph: (a) NMR data showed that the phosphate substrate is encapsulated by the pillar[5]arene with inclusion of only one aromatic ring in the cavity, the other being positioned outside the cavity. (b) pKa values of the five imidazole rings are 3.41, 4.19, 4.40, 4.65, and 5.91 (compared to 7.0 for free imidazole). (c) Expulsion of the dinitrophenolate, after it is split off from the ester via nucleophilic imidazole catalysis, occurs only from the phosphate ester group docked outside the cavity. (d) Kassoc = 94 M−1 at pH = 7.4, while Kassoc = 7200 M−1 at pH = 4.3. (e) Phosphate diester hydrolysis is catalyzed by a factor of 104 in the presence of the pillar[5]arene. Although the observed 104 catalysis (relative to the reference reaction without catalyst at the same pH) is substantial, it is orders of magnitude less than that associated with enzymes. One likely reason for this behavior relates to molecular motion. If the phosphate is rotationally “loose” within the cavity, or if there is “wobble” along the imidazole-bearing side-chains then, according to spatiotemporal notions, the maximum possible rate acceleration (>108) will be negatively impacted. It is here where the “temporal” portion of the spatiotemporal model manifests itself. Clearly, to achieve maximum rates, frameworks must be immobilized such that close contact-distances are imposed. Enzymes, of course, have had a geological time-span to evolve the necessary geometries.

■

The reader has probably noticed that all the systems discussed thus far entail intramolecular catalysis; i.e., the reactant and catalyst are fixed to the same molecule. There is ample motivation for this: Intramolecularity allows distances and angles, critical to our ideas, to be defined precisely. Yet simple enzyme reactions are bimolecular (although the reaction of a substrate bound at an active site can be regarded as simulating intramolecularity). In any case, we were prompted recently to place reactive and catalytic functionalities on separate molecules. This led to the design of supramolecular catalysts as “artificial enzymes”23,24 in which spatiotemptoral concepts once again provide the philosophical

DYNAMICS Organic models and enzyme systems have an important difference that must be mentioned. Organic models (at least the good ones with enzyme-like rates) are more-or-less rigid small molelcules and are characterized by functionalities held permanently at contact distances. Enzymes, on the other hand, are flexible molecules that engage in rapid conformational dynamics indispensable to catalytic function.25−29 Thus, contact distances in enzymes are achieved when the protein samples conformation-space in a search for configurations 1389

DOI: 10.1021/acschembio.8b01029 ACS Chem. Biol. 2019, 14, 1386−1392

Reviews

ACS Chemical Biology

leading to a neutral enol. The enol then attacks the nearby oxaloacetate carbonyl in a fast step (6). Everything is perfectly aligned. Distances are too close to allow even a single water molecule to intervene between the reactive groups. Spatiotemporal effects (i.e., contact distances) dominate the catalysis. Were it not for the close contact distances, the acetyl-CoA would have been unable to enolize. This is because in the absence of enforced contact a weak acid (His-274) and a weak base (Asp-375) would be incapable of effectively enolizing a weak carbon-acid such as acetyl-CoA. In intermolecular systems, for example, general-acid/base catalysis is notoriously weak (usually less than 2 orders of magnitude). Thus, it is the short contact-distances that provides the >108 accelerations common to both model and enzyme systems. In our view, mechanistic chemistry and biochemistry overlap, a conclusion that would no doubt please proponents of Bishop Occam. (2) AMP-dependent protein kinase A catalyzes the transfer of the γ-phosphate of Mg2ATP to a serine hydroxyl of a protein substrate. Time-lapse X-ray crystallography revealed a series of complicated motions at the active site during which the Michaelis complex, primed for the phosphoryl transfer, is created.31 Within this complex, the nucleophilic hydroxyl points, 3.3 Å away, toward the reactive ATP phosphate in what has been called a “near-attack-conformation” or NAC.27 (Note: NAC is a nonquantitative concept developed subsequent to spatiotemptoral ideas but embodies much the same philosophy under a different name). Stated in another way, a critical distance between the reactants, with no intervening water molecule, has been achieved, thereby enabling a fast reaction. A complex, believed to approximate the actual Michaelis complex, is depicted below. The S21B hydroxyl is positioned at a bonding distance from the terminal ATP phosphate lying to its immediate left. Note also that the nucleophilicity of the hydroxyl is not enhanced by a general base; it is presumably sufficiently active without it. Indeed, whatever the particular mode of enzyme catalysis, contact distances at the active site are universal among enzymes. We know of no exceptions. Enzyme catalysis in the kinase-A must reflect spatiotemporal acceleration of hydroxyl attack on a Mg2+-activated phosphate group, an effect perfectly in tune with the enzyme models discussed earlier in the paper. Again, chemistry and biochemistry coincide beautifully.

favorable to the breaking and forming of bonds. Typically, conformational changes along the reaction coordinate bring functionalities closer together, ultimately to contact distances, at time-scales ranging from femtoseconds to seconds. Our enzyme models are relevant because they reveal the structure of likely productive conformations. Although organic models represent possible geometries at only a single point along the enzyme’s pathway, molecular properties at this site are crucial to enzymatic reactivity. Note that enzyme dynamics (although largely ignored in Warshel theory) is not a phenomenon disconnected from spatiotemporal effects. Rather, enzyme dynamics is Nature’s special way of searching and achieving the necessary contact distances found in organic models. This conclusion presupposes that enzymes, whatever their particular mechanism, obey the principles of reactivity revealed by bioorganic systems.

■

ENZYMES Thus far, our discussion has involved enzyme models as opposed to enzymes themselves. These models manifested remarkable enzyme-like rates owing to short contact-distances. It seems only reasonable to believe that enzymes exploit similar chemistry, i.e., that enzymes and models are subject to the same chemical principles. But since “reasonable” does not equate with “proof,” a discussion of enzymes themselves is required here. Expectations for enzymes under the tenets of the spatiotemporal hypothesis are clear: Enzyme/substrate complexes must adopt geometries such that relevant functionalities, cofactors, and metals can be maintained at contact distances (e.g.,