Evaluation of a concerted vs. sequential oxygen activation mechanism in α-ketoglutarate–dependent nonheme ferrous enzymes

Significance

Many iron-dependent enzymes utilize molecular oxygen to functionalize inert C–H bonds in biology. Nonheme ferrous enzymes perform these reactions by reducing dioxygen, often in the presence of a sacrificial electron donor, to form a reactive intermediate that converts substrate to product. In α-ketoglutarate (αKG)–dependent enzymes, there has been discussion on the necessity of substrate prebinding to the αKG-bound ferrous center for productive chemistry. This study demonstrates that, when substrate is not prebound to the enzyme, the species generated in the reaction with dioxygen is not competent to perform hydrogen atom abstraction from the substrate. In contrast, when both cosubstrates are prebound to the enzyme, efficient coupling between dioxygen reduction and substrate oxidation is observed, leading to productive chemistry.

Abstract

Determining the requirements for efficient oxygen (O₂) activation is key to understanding how enzymes maintain efficacy and mitigate unproductive, often detrimental reactivity. For the α-ketoglutarate (αKG)–dependent nonheme iron enzymes, both a concerted mechanism (both cofactor and substrate binding prior to reaction with O₂) and a sequential mechanism (cofactor binding and reaction with O₂ precede substrate binding) have been proposed. Deacetoxycephalosporin C synthase (DAOCS) is an αKG-dependent nonheme iron enzyme for which both of these mechanisms have been invoked to generate an intermediate that catalyzes oxidative ring expansion of penicillin substrates in cephalosporin biosynthesis. Spectroscopy shows that, in contrast to other αKG-dependent enzymes (which are six coordinate when only αKG is bound to the Fe^II), αKG binding to Fe^II-DAOCS results in ∼45% five-coordinate sites that selectively react with O₂ relative to the remaining six-coordinate sites. However, this reaction produces an Fe^III species that does not catalyze productive ring expansion. Alternatively, simultaneous αKG and substrate binding to Fe^II-DAOCS produces five-coordinate sites that rapidly react with O₂ to form an Fe^IV=O intermediate that then reacts with substrate to produce cephalosporin product. These results demonstrate that the concerted mechanism is operative in DAOCS and by extension, other nonheme iron enzymes.

The oxygen (O₂)-activating mononuclear nonheme iron enzymes are important in antibiotic and neurotransmitter biosynthesis, hypoxia regulation, bioremediation, and DNA repair (1–7). They use a high-spin Fe^II site to activate O₂ to form iron–oxygen intermediates that transform a variety of organic substrates (8). These enzymes share a common 2-His/1-(Asp/Glu) facial triad motif to coordinate Fe^II, leaving three coordination positions for possible binding of O₂, cofactor (i.e., a source of additional electrons), and substrate (9). Using variable-temperature, variable-field magnetic circular dichroism (MCD) spectroscopy, we have shown that the nonheme iron enzymes utilize a general mechanistic strategy where the Fe^II site remains coordinatively saturated (i.e., no available site for O₂ activation) until both the cofactor and substrate are bound within the active site pocket. This ensures that the electron equivalents used to generate the iron–oxygen intermediate result in productive substrate oxidation (i.e., coupled turnover) (10).

A large subclass of the nonheme iron enzymes uses an α-ketoglutarate (αKG) cofactor as a source of two electrons for O₂ activation and has been shown to follow the general mechanistic strategy (11–13). The Fe^II resting sites are six coordinate, and the αKG cofactor coordinates to the Fe^II in a bidentate manner to form a site that remains six coordinate (12, 13). For this class of enzymes, substrates generally do not directly coordinate the Fe^II but rather, bind in the active site pocket, preserving a six-coordinate Fe^II site in the absence of αKG (14–16). However, simultaneous binding of αKG and substrate results in a coordinatively unsaturated five-coordinate Fe^II site that can now activate O₂ (11, 17). The O₂ reaction initiates oxidative decarboxylation of αKG (to form succinate and CO₂) to generate a reactive Fe^IV=O intermediate capable of substrate oxidation (18). This high-spin (S = 2) Fe^IV=O intermediate has been trapped in several αKG-dependent enzymes (19–22) and catalyzes a variety of chemical transformations initiated by a common hydrogen atom (H-atom) abstraction step from the substrate to form an Fe^III-OH and a substrate radical. Subsequently, the hydroxide (or a bound halide in the halogenases) can either rebound (resulting in hydroxylation/halogenation) or initiate a second H-atom abstraction (11) [or electron transfer (23, 24)], which leads to desaturation or ring closure/expansion (11, 18).

However, a sequential mechanism has alternatively been proposed in the literature to describe the O₂ reactivity of a number of nonheme iron enzymes (25–27) and was first proposed in deacetoxycephalosporin C synthase (DAOCS), an αKG-dependent enzyme important in the biosynthesis of cephalosporin antibiotics (2, 28, 29). DAOCS has broad substrate specificity, showing expandase activity on penicillins with different side chains, including penicillin-G (30). Unlike the other enzymes in the αKG subclass, crystallography on DAOCS showed penicillin-G substrate coordinating directly to the Fe^II and did not observe simultaneous αKG and penicillin-G binding (25). These observations resulted in a proposed sequential mechanism (Scheme 1A) for DAOCS where 1) αKG first binds and then, the site reacts with O₂ to generate an intermediate that is stored in a ferrous “planar peroxy” form until 2) substrate binding results in the generation of an Fe^IV=O intermediate for the two-electron oxidative ring expansion. However, a recent mass spectrometry study provided evidence of simultaneous αKG and penicillin-G binding that supports the general mechanistic strategy (Scheme 1B) (31). Both mechanisms predict an Fe^IV=O intermediate, which initiates the ring expansion via H-atom abstraction from the β-methyl group of penicillin to form an Fe^III-OH and a methylene radical. This primary methylene radical is proposed to convert to a tertiary radical following ring expansion (Scheme 1C), which is thought to occur through a putative thiyl radical species. Subsequent reaction between the Fe^III-OH and tertiary radical gives the ring-expanded, desaturated product and Fe^II-OH₂ (32–34).

Scheme 1.

(A) Sequential mechanism proposed for DAOCS. (B) Concerted reaction of the Fe^II/αKG/penicillin (pen) site with O₂. (C) Mechanism of DAOCS-catalyzed ring expansion, initiated by Fe^IV=O catalyzed H-atom abstraction (HAA).

In order to determine whether some nonheme iron enzymes can function via a sequential mechanism, this study uses MCD spectroscopy to define the coordination environment of the Fe^II site in resting DAOCS and in its interactions with αKG cofactor and penicillin-G substrate. These studies have 1) identified a five-coordinate Fe^II component when only αKG is bound that selectively (relative to the six-coordinate Fe^II) reacts with O₂. 2) This reactivity enabled us to evaluate and eliminate the sequential mechanism. 3) Alternatively, when both αKG and penicillin-G are bound to DAOCS, we directly observe the Fe^III species formed after H-atom abstraction along with a substrate radical that decay together to form the cephalosporin product. This study confirms that ring expansion in DAOCS and by extension, productive O₂ activation by nonheme iron enzymes in general proceed through a concerted mechanism (Scheme 1B).

Results

Interaction of αKG and Penicillin-G with Fe^II/DAOCS: Availability of a Five-Coordinate Fe^II/αKG Site.

In order to evaluate the sequential mechanism, the coordination of the Fe^II site in DAOCS and its interactions with αKG and penicillin-G substrate were first defined by MCD spectroscopy. Low-temperature MCD spectroscopy in the near-infrared region (Fig. 1 with the corresponding circular dichroism spectra in SI Appendix, Fig. S1) probes the Fe^II ligand-field transition energies and splitting that enable determination of coordination number and geometry (8). Fe^II-DAOCS has two transitions at ∼10,000 cm⁻¹ split by 1,800 cm⁻¹ (Fig. 1, blue), which are diagnostic of a six-coordinate Fe^II site. Addition of penicillin-G resulted in only small spectral perturbations (Fig. 1, green), including a small (∼100-cm⁻¹) red shift of the ligand-field transitions, a change in their intensity ratio, and a change in the ground-state splitting (from variable-temperature, variable-field MCD of these transitions) (SI Appendix, Fig. S2). These small changes indicate that penicillin-G does not directly coordinate to the Fe^II and that binding in the active site pocket only modestly perturbs the six-coordinate Fe^II center. These small coordination changes are consistent with the substrate binding mode observed in other αKG-dependent enzymes (14–16), in contrast to the crystallographic results that showed penicillin-G coordinating to the Fe^II (25).

Fig. 1.

The 7-T, 5-K MCD spectra for Fe^II-DAOCS (blue), (Fe^II/penicillin-G [penG])-DAOCS (green), (Fe^II/αKG)-DAOCS (red), and (Fe^II/αKG/penG)-DAOCS (purple) with Gaussian resolutions and transition energies indicated. In the (Fe^II/αKG)-DAOCS (red) and (Fe^II/αKG/penG)-DAOCS spectra, the presence of three bands are indicated (numbered 1–3).

Addition of αKG to Fe^II-DAOCS dramatically changes the MCD spectrum (Fig. 1, red), showing two positive transitions (8,400 and 10,400 cm⁻¹) and a negative transition at higher energy (11,500 cm⁻¹). Furthermore, MCD spectroscopy in the visible energy region revealed an Fe^II-to-αKG charge transfer transition (SI Appendix, Fig. S3), confirming bidentate αKG binding to the Fe^II as observed in other αKG-dependent enzymes (12, 15, 16, 35–37). Since a single Fe^II site can only have two ligand-field transitions in the near-infrared spectral region (8), the presence of three bands reflects a mixture of two distinct αKG-bound Fe^II centers. Previously studied αKG-bound Fe^II active sites show only a single six-coordinate site, in contrast to the Fe^II/αKG site in DAOCS (11–13, 36). Additionally, most high-spin Fe^II sites studied show all MCD transitions with positive intensity (SI Appendix, Deviation from the Sum Rule in Fe^IIMCD Spectroscopy); negatively signed transitions have only been observed in five-coordinate model complexes and the five-coordinate site of an extradiol dioxygenase (38–40).

Addition of penicillin-G to the αKG-bound Fe^II site perturbs the MCD spectrum (Fig. 1, purple), where positive bands 1 and 2 decreased in intensity while the negative intensity of band 3 increased. The presence of the Fe^II-to-αKG charge transfer transition (SI Appendix, Fig. S3) confirms that αKG is still coordinated bidentate to the Fe^II center. Small changes in the ground-state splitting from variable-temperature, variable-field MCD confirm that penicillin-G only slightly perturbs the Fe^II/αKG site (SI Appendix, Fig. S2), consistent with penicillin-G binding in the active site pocket and not directly to the Fe^II. These results demonstrate that simultaneous binding of both αKG and penicillin-G occurs in DAOCS in contrast to the crystallographic results but consistent with the mass spectrometry (25). Furthermore, the intensity changes in the Fe^II ligand-field MCD transitions indicate that substrate binding has shifted the ratio of the two species observed in the only αKG-bound mixture, decreasing the component associated with the two positive transitions and increasing the component associated with the negative transition.

In other αKG-dependent enzymes, binding of both αKG and substrate has been found to open a coordination position on the Fe^II for O₂ activation (13, 35, 36). In DAOCS, the MCD data above show that substrate binding to the αKG-bound Fe^II site increases the component of the mixture associated with the negative transition. Mössbauer spectroscopy of (Fe^II/αKG)-DAOCS also shows two ferrous components (δ₁ = 1.22 mm/s, ΔE_Q,1 = 2.17 mm/s; δ₂ = 1.24 mm/s, ΔE_Q,2 = 2.47 mm/s in Fig. 2A). On substrate binding to the αKG bound Fe^II site, the species with a larger ΔE_Q increases from 46 to 59% of total Fe^II concentration (SI Appendix, Fig. S4 and Table S1). Furthermore, X-ray absorption spectroscopy showed that, compared with resting Fe^II-DAOCS (which is six coordinate), simultaneous substrate and αKG binding produced a significant five-coordinate contribution (SI Appendix, Figs. S5–S8 and Tables S2–S7). Thus, the negative transition in MCD and the Mössbauer doublet with the larger ΔE_Q (δ = 1.24 mm/s, ΔE_Q = 2.47 mm/s) present in (Fe^II/αKG)-DAOCS are associated with a five-coordinate site that is also present to a greater (59%) extent in the (Fe^II/αKG/penicillin-G)-DAOCS sample. In contrast to other αKG-bound Fe^II active sites that are six-coordinate, the presence of this five-coordinate, therefore coordinatively unsaturated, (Fe^II/αKG)-DAOCS site allows for the evaluation of the sequential O₂ activation mechanism (Scheme 1A) proposed (25) for DAOCS and other αKG-dependent enzymes (25–27).

Fig. 2.

The 6-K Mössbauer spectra of Fe^II-DAOCS/αKG (A) reacted with one equivalent O₂ and freeze quenched at 6 s (B) and 18 s (C). Note that the red asterisks indicate the hyperfine feature associated with a small (<15%) noncatalytic Fe^III species, which results from the rapid freeze quench. SI Appendix, Reaction of Five- and Six-Coordinate (Fe^II/αKG)-DAOCS with O₂ and Fig. S9 have details.

Reaction of Five- and Six-Coordinate (DAOCS)Fe^II/αKG with O₂.

From the above MCD results, when only αKG is bound to Fe^II-DAOCS, both five- and six-coordinate sites are present. Since the two electron equivalents of the αKG cofactor are available for O₂ reduction, this enabled us to evaluate the relative reactivity of the five- and six-coordinate Fe^II sites and the nature of the species generated in this reaction. Starting from Mössbauer spectrum of the anaerobic Fe^II/αKG sample (Fig. 2A), reaction with O₂ led to the formation of a new Mössbauer doublet (Fig. 2B, orange) within 6 s with δ = 0.43 mm/s and ΔE_Q = 0.50 mm/s, consistent with a high-spin Fe^III species (SI Appendix, Reaction of Five- and Six-Coordinate (Fe^II/αKG)-DAOCS with O₂ and Figs. S9 and S10 have a description of the additional [less than 15%] hyperfine high-spin Fe^III signals associated with nonreducible sites). Associated with the formation of the Fe^III species in the Mössbauer spectrum, stopped-flow absorption spectra show the growth of two features at 335 and 380 nm (Fig. 3A) concomitant with loss of intensity at 520 nm. From the time course of the Mössbauer data in Fig. 2, the five-coordinate component converts to the Fe^III site more rapidly than the six-coordinate component. After 6 s, the six-coordinate component has only decreased by 6% of total iron concentration, while the five-coordinate component has decreased by 11%, which together result in 17% Fe^III sites. At longer times (>100 s) (SI Appendix, Figs. S9–S11), both the Fe^II doublets have converted to the Fe^III signal.

Fig. 3.

(A) Absorption spectra collected at 5 °C after equal volume mixing of a solution of 0.2 mM DAOCS, 0.2 mM Fe^II, and 0.2 mM αKG in 50 mM HEPES buffer, pH 7.5, with O₂-saturated buffer (∼2 mM). The spectra were generated by subtracting the spectrum of the DAOCS/αKG mixed with O₂-free buffer. The arrows indicate the appearance of the 335- and 380-nm features with the concurrent disappearance of the 520-nm feature. (Inset) Kinetic trace shows the growth of the absorbance at 380 nm. (B) Reaction of a solution of 0.2 mM DAOCS, 0.2 mM Fe^II, and 0.2 mM αKG in 50 mM HEPES buffer, pH 7.5, with O₂-saturated buffer (∼2 mM) monitored EPR spectroscopy with samples frozen at 45 s (red), 100 s (blue), and 6 min (black). The black arrow shows the decay of the g ∼ 2 species.

The Mössbauer Fe^II speciation and the kinetics observed from absorption spectroscopy (Fig. 3 A, Inset) cannot be simultaneously fit with two parallel five-coordinate and six-coordinate Fe^II dioxygen reactions. From the Mössbauer data, since the six-coordinate component decays at a slower rate than the five-coordinate component, it converts to the five-coordinate site, which then reacts with O₂. The formation of the 380-nm absorption (Fig. 3 A, Inset) can be fit with a second-order rate constant of 0.09 mM⁻¹s⁻¹ (SI Appendix, Fig. S12) (which corresponds to a pseudo first-order rate of 0.09 s⁻¹) and reflects the reaction of the five-coordinate component with O₂ along with an equilibrium six- to five-coordinate Fe^II conversion necessary for the O₂ reaction (SI Appendix, Fig. S12). The above results show that the five-coordinate Fe^II selectively reacts with O₂ and that the slower rate for the six-coordinate component (k_f = 0.05 s⁻¹) reflects loss of H₂O (stabilized in the active site through hydrogen bonds) to enable the O₂ reaction.

Characterization of the Fe^III Species and Its Reactivity: Evaluation of the Sequential Mechanism.

The appearance of the high-spin Fe^III species from Mössbauer spectroscopy (Fig. 2 B and C) correlates with the growth of the 335/380-nm species from absorption spectroscopy (Fig. 3A). In parallel, electron paramagnetic resonance (EPR) spectra reveal the growth of a g′ ∼ 4.3 high-spin Fe^III feature and a g ∼ 2.0 S = 1/2 radical feature (Fig. 3B). While the ferric g′ ∼ 4.3 species grows and levels off (paralleling the 380-nm absorption and the Fe^III Mössbauer doublet), the S = 1/2 radical decays with a rate of 0.02 s⁻¹. The molar absorptivities of the 335- and 380-nm bands (ε₃₃₅ = 3,400 M⁻¹cm⁻¹ and ε₃₈₀ = 2,000 M⁻¹cm⁻¹) are consistent with ligand-to-metal charge transfer transitions in high-spin Fe^III complexes and from the data in SI Appendix, Fig. S13, reflect a succinate-bound Fe^III site. Furthermore, the loss of intensity at 520 nm in Fig. 3A indicates the loss of the Fe^II-to-αKG charge transfer in the formation of the Fe^III species (note the isosbestic point) and requires that the αKG cofactor has decarboxylated to a succinate. This decarboxylation provides two electrons, and based on the behavior of other αKG-dependent Fe^II enzymes (41–44), the Fe^II would provide two more electrons to form an Fe^IV=O intermediate. The observation of the formation of an Fe^III and an organic radical suggests that this ferryl intermediate rapidly decays by H-atom abstraction (or electron and proton transfer) to form an Fe^III-OH, which would exhibit the 380- and 335-nm transitions (SI Appendix, Figs. S13–S15). It should be noted that further addition of αKG to this Fe^III-succinate species reduces it to Fe^II, enabling multiple turnovers with O₂ (SI Appendix, Figs. S16–S18), which explains the observed decay of the 380-nm species in a similar experiment done in excess αKG where this species was incorrectly assigned as an Fe^IV=O intermediate (SI Appendix, Fig. S19) (31).

From studies on the αKG- (45) and pterin-dependent hydroxylases (46), the O₂ binding step is fast and reversible relative to the succeeding chemical transformation steps. Due to the lack of a characteristic Fe^IV=O absorption (Fig. 3A) and Mössbauer features (Fig. 2), the ferryl-oxo species does not measurably accumulate (<5%) prior to formation of the Fe^III-succinate/hydroxo species. Fitting the absorption data in Fig. 3 A, Inset to the kinetic model in Scheme 2 showed that Fe^IV=O decay (k₃ > 10 s⁻¹) has to be at least two orders of magnitude faster than its formation (k₂ = 0.09 s⁻¹), reflecting its lack of accumulation.

Scheme 2.

Kinetic model and corresponding rate parameters used to simulate the stopped-flow data of the reaction of (Fe^II/αKG)-DAOCS with O₂.

When the Fe^III-succinate/hydroxo species [produced in the O₂ reaction of (Fe^II/αKG)-DAOCS] is reacted with penicillin-G substrate, it formed an Fe^II species and a species with an EPR signal with g ∼ 2.08 (SI Appendix, Figs. S20–S22), consistent with a sulfur-based radical on the penicillin-G substrate (corroborated by electronic structure calculations) (SI Appendix, Fig. S22) and with no cephalosporin product formation. Thus, the Fe^III produced in the reaction of five-coordinate (DAOCS)Fe^II/αKG with O₂ is not competent in product formation, and the Fe^IV=O that would form prior to the Fe^III species decays at a much faster rate than its formation (Scheme 2), making it inaccessible in a sequential mechanism. These results eliminate a sequential mechanism for DAOCS, consistent with other αKG nonheme iron enzymes (10).

Reactivity of the Fe^II/αKG/Penicillin-G Site with O₂: Coupled Turnover via the Concerted Mechanism.

Since the sequential mechanism does not produce the cephalosporin product, the reaction of Fe^II-DAOCS with both αKG and penicillin-G bound with O₂ was evaluated. Within the first 25 s of the reaction (Fig. 4A), an absorption feature grows at 310 nm (red arrow in Fig. 4A and SI Appendix, Fig. S23) followed by the growth of 335- and 380-nm features (black arrows in Fig. 4A), apparent in the time traces (Fig. 4B). During the growth of the 310-nm species, the intensity at 520 nm is unchanged (dark blue traces in Fig. 4 A, Inset) but decreases with an isosbestic point at 450 nm as the 335- and 380-nm features subsequently grow in (lighter traces in Fig. 4A). Although the 310-nm species does not accumulate enough for Mössbauer characterization, it can be assigned as an Fe^IV=O based on its absorption maximum (310 to 320 nm for Fe^IV=O in the literature) and the observed lag in decay of spectral features in the 500-nm region (due to growth of Fe^IV=O ligand-to-metal charge transfer transitions that overlap the decay of the Fe^II-to-αKG charge transfer) (19). The 335/380-nm intermediate is the Fe^III-succinate/hydroxo species based on its assignment in the previous section. At later times (under substoichiometric O₂ concentrations, 100 to 3,000 s) (Fig. 4C), there is a decrease in intensity in the 335- to 380-nm region, concomitant with the appearance of an intense 303-nm absorbance feature. This broad, intense 303-nm feature corresponds to the cephalosporin product, with the final product spectrum shown in SI Appendix, Fig. S24. Thus, in the presence of bound substrate, the Fe^III-succinate/hydroxo species, formed from the reaction of Fe^IV=O with substrate, productively forms the cephalosporin product.

Fig. 4.

(A) Absorption spectra collected at 5 °C after equal volume mixing of a solution of 0.26 mM DAOCS, 0.26 mM Fe^II, 8 mM penicillin-G (penG), and 0.26 mM αKG in 50 mM HEPES buffer, pH 7.5, with O₂-saturated buffer (∼2 mM), with the spectrum of DAOCS/αKG/penG mixed with O₂-free buffer subtracted. The red arrow indicates initial growth of the 310-nm feature (expanded in SI Appendix, Fig. S18) and the black arrows demonstrate the subsequent disappearance of the 520-nm feature concomitant with the growth of the 335- and 380-nm features. (Inset) Expanded view of the 500-nm region. (B) Kinetic traces for the experiment in A showing absorbances at 310 nm (purple), 335 nm (red), and 380 nm (yellow). (C) Same experiment as in A and B but mixed with 0.2 mM O₂ in buffer, with the spectrum of DAOCS/αKG/penG mixed with O₂-free buffer subtracted. The arrows demonstrate the loss of the 380-nm feature along with the growth of the 303-nm feature. (D and E) EPR samples frozen at different times along the experiment described in C, with the g′ ∼ 4.3 region in D and the g ∼ 2 region in E; signals were measured at 77 K. (F) Intensity of the g′ ∼ 4.3 (Upper; red circles) and g ∼ 2 (Upper; purple crosses) EPR plotted with the absorption time traces of the reaction; monitoring was at 303 nm (Lower; blue) and 380 nm (Lower; red). The g′ = 4.3 (red circles) EPR signal and the 380-nm time trace (red) do not completely decay due to a minor contribution of the uncoupled (DAOCS)-Fe^II/αKG + O₂ reaction described due to poor penG binding [K_M(penG) = 2 mM and (penG) = 8 mM limited by its solubility in the reaction conditions] (54).

EPR samples freeze quenched at different times in the reaction of (Fe^II/αKG/penicillin-G)-DAOCS with O₂ show (at 77 K) a g′ ∼ 4.3 signal that correlates with the Fe^III absorption feature (Fig. 4D) and a radical signal at g ∼ 2 (Fig. 4E). From Fig. 4 D, g′ ∼ 4.3 and E, g ∼ 2, these spectra decay together, which correlates with the formation of the absorption of the cephalosporin product (Fig. 4F). When EPR spectra were collected at 5 K, there is an additional resonance at g′ ∼ 7.5 that forms and decays with the g′ ∼ 4.3 and 2.0 signals and is characterized in SI Appendix, Fig. S25. In mechanistic proposals for DAOCS, the primary methyl radical rearranges to become a thiyl radical and then, a tertiary radical following ring expansion (Scheme 1C). The g value of the radical species in Fig. 4E (∼2.0) is consistent with a carbon- rather than thiyl-centered radical (47, 48) and is also distinct from the radical formed in the reaction of (Fe^II/αKG)-DAOCS with O₂ as well as the radical of the one-electron oxidized penicillin-G substrate (SI Appendix, Fig. S26).

The kinetics of the reaction of Fe^II/αKG/penicillin-G with O₂ (Fig. 4) were fit to the model in Scheme 3. In comparing the reaction in the absence and presence of substrate, the addition of substrate increases the rate of Fe^IV=O formation by an order of magnitude (k₂ = 0.9 vs. 0.09 s⁻¹ without substrate) (Scheme 2). This rate difference is an important finding and indicates that substrate binding to the αKG-bound site activates the site for a more efficient reaction with O₂. Interestingly, the rate of Fe^IV=O decay in the substrate-bound reaction is at least an order of magnitude slower (k₃ = 0.4 s⁻¹ in Scheme 3) than the Fe^IV=O decay rate observed in the substrate-free reaction (k₃ > 10 s⁻¹) (Scheme 2). In the presence of substrate, the Fe^IV=O performs an H-atom abstraction from the methyl group on the penicillin ring. In the absence of substrate, the electron and proton sources for reduction of the Fe^IV=O to Fe^III-OH are not known, but the barrier for this step must be lower than for the difficult C–H bond activation for the substrate (bond dissociation enthalpy [BDE] of the β-methyl group = 99 kcal/mol) (SI Appendix, Table S8) and can no longer be accessible when substrate is bound. Furthermore, the Fe^III species, formed after Fe^IV=O decay in the presence of substrate, is capable of abstracting a second H-atom from the substrate (that it is now a radical with a BDE = 36 kcal/mol) (SI Appendix, Table S8) to form the ring-expanded product. Thus, while the Fe^IIIOH/succinate cannot perform H-atom abstraction from the substrate in a sequential mechanism, it can on the substrate radical in the concerted mechanism, confirming the latter mechanism for αKG-dependent Fe^II enzyme catalysis.

Scheme 3.

Kinetic model and corresponding rate parameters used to simulate the stopped-flow data of the reaction of (Fe^II/αKG/penicillin-G)-DAOCS with O₂. Since there are 41% six-coordinate sites from Mössbauer, the observed (SI Appendix, Fig. S9) six- to five-coordinate conversion rate from Scheme 2 was used to generate the five-coordinate Fe^II site that binds O₂.

Discussion

This study provides important insight into the mechanism of αKG-dependent nonheme iron enzyme catalysis using DAOCS, for which past results have been controversial. Variable-temperature, variable-field MCD spectroscopy showed that both cosubstrates can simultaneously bind to the Fe^II active site and that penicillin-G substrate does not directly bind to the metal center (14–16, 25) and behaves similarly to other enzymes in this family. However, unlike other members of this family, the αKG only-bound site in DAOCS does have a five-coordinate component as demonstrated by variable-temperature, variable-field MCD, Mössbauer, and X-ray absorption spectroscopies. The five-coordinate Fe^II site selectively reacts with O₂ from the Mössbauer data (Fig. 2), while the six-coordinate component reacts slower, likely reflecting the rate of loss of a water ligand to form the reactive five-coordinate Fe^II site.

The presence of a five-coordinate component when only αKG is bound enabled evaluation of a sequential mechanism. As summarized in Scheme 4, reaction of the five-coordinate Fe^II/αKG site with O₂ forms a stable HO-Fe^III-succinate species and not an Fe^II-peroxy-succinate that had been predicted through crystallography and electronic structure calculations (25). Reaction of this Fe^III species with penicillin-G substrate results in unproductive one-electron oxidation of penicillin-G. The Fe^IV=O, which would produce this Fe^III species by reaction with a protein residue in the absence of substrate, is formed slowly and dissipates rapidly, making it kinetically inaccessible. Thus, the sequential mechanism (Scheme 1A) is not operative for productive chemistry in αKG-dependent Fe enzymes.

Scheme 4.

Uncoupled reactivity of DAOCS. Note that penG refers to the penicillin-G substrate.

As shown in Scheme 5, the concerted mechanism was evaluated by reacting the Fe^II/αKG/penicillin-G site with O₂. An Fe^IV=O species first forms but rapidly decays to an Fe^III-succinate species along with the substrate radical. Note that this species is an Fe^III-OH and not an Fe^III-oxo species [previously proposed in αKG-dependent enzyme catalysis (49)] based on its small ΔE_Q splitting from Mössbauer spectroscopy and rhombic EPR signal as observed in Fe^III-OH (but not Fe^III-oxo) model complex chemistry (50, 51). Proceeding from the Fe^III-OH intermediate and substrate radical, unlike the behavior observed in the uncoupled reaction (i.e., no substrate) (Scheme 4), both the ferric species and the organic radical decay together and productively to form the cephalosporin product. Thus, the binding of the penicillin-G substrate to the Fe^II/αKG site before reaction with O₂ is required for productive turnover, demonstrating that the concerted mechanism is operative in DAOCS. In the concerted pathway, from EPR spectroscopy one dominant g ∼ 2 radical species is observed. From mechanistic proposals for DAOCS, the primary methyl group in penicillin substrates undergoes H-atom abstraction and then, rearranges to a tertiary radical by C–S bond scission through a putative thiyl radical from stereochemical studies (34). Since the interconversion of these species is not observed, this rearrangement must be rapid: at least an order of magnitude faster than the H-atom abstraction by the Fe^IV=O. Furthermore, since there is no evidence for primary hydroxylation, the rapid rearrangement of the radical would provide a mechanism to suppress hydroxylation reactivity in the αKG-dependent hydroxylases, similar to a proposal in the literature (52).

Scheme 5.

General mechanistic strategy in DAOCS. Note that penG refers to the penicillin-G substrate.

In the concerted mechanism, the O₂ reaction to form an Fe^IV=O is an order of magnitude faster than in the absence of substrate. This rate difference has two main contributions: 1) electronic, where the substrate- and cofactor-bound Fe^II site has a more activated dπ orbital, and 2) steric, where clash between the substrate and the sixth water ligand leads to five-coordinate conversion, both of which prime the active site for efficient O₂ activation (13, 53). In the concerted reaction, the formation of cephalosporin (i.e., the second H-atom abstraction by the Fe^III-OH from the substrate radical) is rate limiting, which is likely due to the poor orientation of the rearranged penicillin-G substrate radical in the DAOCS active site. Finally, the substrate radical decays more than an order of magnitude slower than the protein residue radical (in the absence of substrate), indicating that, in the presence of substrate, autooxidation is inaccessible, likely due to substrate positioning preventing electron/proton transfer from the protein residue to the Fe^IV=O intermediate.

This study has shown that the general mechanistic strategy, where both cosubstrates are bound to the active site to generate a five-coordinate Fe^II for O₂ activation, is an effective gating mechanism that is required by nonheme iron enzymes for productive catalysis.

Materials and Methods

Details about protein expression and purification, sample preparation, spectroscopic methods, X-ray absorption experiments and data analysis, stopped-flow absorption experiments, and kinetic simulations are provided in SI Appendix, Materials and Methods.

Data Availability.

All data are included in the manuscript or SI Appendix.