Mechanistic Investigation of Oxygen-Activating Enzymes Using Rapid-Kinetics, Structural, and Spectroscopic Methodologies

open_access
Open Access
Author:
Wenger, Eliott
Graduate Program:
Chemistry
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
June 14, 2022
Committee Members:
Howard Salis, Outside Unit & Field Member
Amie Boal, Major Field Member
Carsten Krebs, Co-Chair of Committee
Alexey Silakov, Major Field Member
Joseph Bollinger, Co-Chair & Dissertation Advisor
Philip Bevilacqua, Program Head/Chair
Keywords:
Enzyme
Mechanism
Oxygen
Kinetics
Crystallography
Metalloenzyme
Stopped-flow
EPR-spectroscopy
Mossbauer-spectroscopy
Mass Spectrometry
Abstract:
Enzymes are proteins that catalyze chemical reactions. Enzymes that are responsible for facilitating some of the most challenging chemical transformations, such as nitrogen fixation, water oxidation, and C–H activation, contain metal cofactors that are critical for proper function. One family of metalloenzymes that has been implicated in a variety of biological roles in all three domains of life and are well-represented in the biosynthetic pathways of many molecules of medicinal and industrial interest are the Fe(II)/2-oxoglutarate dependent (Fe/2OG) oxygenases. These enzymes utilize a metal cofactor (one single non-heme atom of ferrous iron) and the common metabolite 2OG to harness the oxidizing power of molecular oxygen to cleave unactivated C–H bonds on a diverse scope of substrates. Dioxygen is activated at the iron center to form an Fe(IV)-oxo intermediate known as the ferryl. The ferryl generally decays by targeting one well-positioned C–H bond of a given substrate for cleavage through a hydrogen atom transfer (HAT) step, generating a ferric hydroxo and a substrate radical. In the majority of known Fe/2OG enzymes, and all cases occurring in humans, the substrate radical attacks the hydroxo ligand to complete a hydroxylation reaction. This low-barrier step has come to be known as oxygen rebound. Other organisms have leveraged Fe/2OG enzymes for a diverse set of chemical transformations. In Chapter 1, we review the foundational experiments that first probed the mechanism of Fe/2OG-catalyzed hydroxylation, and then explore the many known examples of other reactions catalyzed by Fe/2OG enzymes, including (but not limited to) halogenation, desaturation, oxacyclization, epimerization, endoperoxidation, and C–C bond formation. In almost all known cases, non-hydroxylating Fe/2OG enzymes still rely on the oxidizing power of a ferryl intermediate to generate a substrate radical. The varied reactivities are thought to flow from varied fates of the substrate radical, revealed by the many cases where a minor hydroxylation product is detected, establishing that rebound occurs in competition with an alternative outcome. How a given system achieves a given non-hydroxylation outcome with selectivity while avoiding rebound has thus become the central question driving research in this field. Previous research has suggested that the precise spatial disposition of both the ferryl intermediate relative to the target C–H bond and the resulting substrate radical relative to the iron-coordinated hydroxo ligand are key in determining outcome. In particular, previous work has demonstrated that inefficient HAT steps are often required to avoid rebound, and that increasing HAT efficiency can lead to increased hydroxylation. However, the nature of the rebound step, whether it is inextricably linked to the HAT efficiency or whether the two steps are separately tunable, whether enzymes that achieve non-hydroxylation outcomes actually suppress rebound, or favor the competing step, or both in combination, remained to be explored. In this body of work, we use a variety of rapid-kinetic, structural, and spectroscopic techniques to explore the relationship between oxygen rebound and competing steps that respectively lead to halogenation, oxacyclization, and desaturation, shedding light on these deep mechanistic questions. In Appendix A, we directly investigated the kinetics of oxygen rebound by characterizing three Fe/2OG enzymes that catalyze two competing hydroxylation reactions. In each of these cases, a ferryl intermediate can decay by accepting a hydrogen atom from two different substrate carbons, forming two distinct substrate radicals. In all three cases, one carbon was exclusively hydroxylated, while targeting the other carbon could lead to hydroxylation or, in certain contexts, an alternative outcome. We demonstrated that for all three systems, 18O from O2 is incorporated to different extents in the two hydroxylation products, and always to a greater extent at the carbon that is exclusively hydroxylated. Exchange with solvent only in the ferryl state is insufficient to explain this differential incorporation; additional exchange in the ferric state must be invoked. The exciting implication of this study was that the rebound step might actually be relatively inefficient in some cases, and that Fe/2OG enzymes might enforce such an inefficiency in order to achieve alternative reaction outcomes with selectivity. In Chapters 2 and 3, we explore suppression of oxygen rebound in the context of oxidative ring closure in two apparently similar systems, H6H and LolO. H6H and LolO both hydroxylate their substrates in one reaction (at C2 for LolO, and at C6 for H6H), and then in a second reaction couple the previously-installed hydroxyl to another carbon (C7 in both systems) to form an oxacycle, though LolO forms a five-membered ring while H6H installs an epoxide. Although neither oxacyclization reaction had previously received detailed mechanistic interrogation, it had been proposed that the reactions might proceed by attack of a substrate radical upon an iron-coordinated hydroxyl group. In Chapter 2, we interrogate the epoxidation reaction catalyzed by H6H. Surprisingly, our best efforts to provide evidence for coordination of the C6 hydroxyl group to the iron center suggested instead that no rearrangement of the substrate or cofactor is required. We concluded that the main factor in achieving selective epoxidation in H6H is the engineering of an extremely inefficient C7 rebound step by a particular substrate binding mode. Characterization of a variant that disrupted the substrate binding mode also disrupted the programmed suppression of rebound, leading to greater and more efficient dihydroxylation. We conclude the study by providing evidence that deprotonation of the hydroxyl group occurs after HAT and in competition with rebound, consistent with the hydroxyl group remaining uncoordinated during the reaction. In the case of LolO, which we characterize in Chapter 3, we provide several results that comport instead with the previously-proposed pathway relying on coordination of the hydroxyl group to an existing ferryl intermediate. The C7–H bond is cleaved to initiate cyclization with a rate constant 16 times faster than the rate constant measured for aberrant hydroxylation of C7 in the first reaction, suggesting that the substrate-cofactor disposition in the active site has been significantly perturbed by the addition of a hydroxyl group to C2. While only one ferryl is detected in the LolO-catalyzed hydroxylation, two different Fe(IV) species are detected during the cyclization, perhaps representing coordination of the C2 hydroxyl group to an existing ferryl. Consistent with a coordinated hydroxyl group, which would likely be deprotonated, solvent deuteration does not increase the proportion of dihydroxylation. We hope to structurally characterize this reaction in the near future. In Appendix B, we explore the LolO-catalyzed cyclization reaction in even greater detail using fluorinated analogues of the LolO substrate. The cyclized product is still formed by LolO even with both C6 hydrons substituted with fluorine. Such a substitution would likely prevent formation of a carbocation on neighboring C7, providing additional evidence for the radical coupling-based mechanism we propose in Chapter 3. In Chapter 4, we interrogate the halogenation reaction catalyzed by the Fe/2OG enzyme WelO5. Previous study with WelO5 suggested that no rearrangement of the substrate is required for selective halogenation; instead rebound is avoided through the formation of an offline ferryl. Vanadyl (V(IV)≡O) is an EPR-active molecule containing a tetravalent metal cation previously used as a structural mimic of the fleeting ferryl intermediate. We provide two distinct lines of evidence that WelO5 binds vanadyl in an offline manner, suggesting that WelO5 forms an offline ferryl during its native reaction. In addition to a crystal structure of WelO5 in complex with vanadyl, we pioneer an orientation selective double electron-electron resonance (OS-DEER) EPR technique that allowed us to measure the angle between the V-O vector of vanadyl (bound in the active site) and spin labels installed at various positions on the exterior of the protein. These experiments provide direct structural evidence for the formation of an offline ferryl as a strategy for avoiding rebound, long presumed to be a key adaptation of the Fe/2OG halogenases. In Chapter 5, we characterize a non-native desaturation catalyzed by the natively trifunctional (hydroxylation, oxacyclization, and desaturation) enzyme CAS. Previous work has demonstrated that selective desaturation is enabled by the presence of an alpha-heteroatom that allows for rapid removal of the second hydron as a proton, and that rebound is competitive in cases where a second HAT step is required for desaturation. The CAS-catalyzed desaturation lacks a heteroatom, and accordingly proceeds in competition with hydroxylation of either carbon. Measuring the kinetics of the reaction and the ratio of the three products formed in the presence of site-specific deuterium labels allowed us to resolve intrinsic rate constants for HAT from each carbon and suggested that most of the desaturation is achieved by initial oxidation of C3, followed by HAT from C4. Structural characterization of CAS and comparison to other Fe/2OG desaturase structures suggest a strict angle dependency on rebound efficiency, similar to our findings in Chapter 2. We furthermore demonstrate that an anti-periplanar substrate binding mode is enforced in all known desaturase structures, implying that such a binding mode is an important feature of this mode of catalysis. In Chapter 6, we move beyond Fe/2OG enzymes to study the mechanism of an oxygen-activating pyridoxal phosphate (PLP) dependent enzyme, RohP. A mechanism was previously proposed for the formation of two products (P1 and P2) by RohP, formed by one or two sequential oxidations of L-arginine, respectively, with each oxidation consuming one molecule of O2. We provide evidence for several predictions made by this prior study, including loss of one of the C3 hydrons in the first product P1, an overall ratio of products dependent on the O2 concentration, and absolute retention of both of the C5 hydrons in both products. However, our in-depth analysis of the enzyme in studies probing substrate binding and reaction with O2 under single- and multi-turnover conditions revealed additional unanticipated mechanistic features. These features include the release of some P1 retaining both of the C3 hydrons, establishing an additional branchpoint not proposed by the previous authors, the accumulation of a state prior to the second oxygen-reactive state (likely due to a single active site histidine residue performing two key deprotonation steps sequentially), and a cross-reaction with peroxide that proceeds by a similar mechanism and forms identical products.