Crystal Structures of Alkylperoxo and Anhydride Intermediates in an Intradiol Ring-Cleaving Dioxygenase

Crystal structures of alkylperoxo and anhydride intermediates in an intradiol ring-cleaving dioxygenase

Cory J. Knoot, Vincent M. Purpero, and John D. Lipscomb1

Department of Biochemistry Molecular Biology and Biophysics and Center for Metals in Biocatalysis, University of Minnesota, Minneapolis, MN 55455

Edited by Edward I. Solomon, Stanford University, Stanford, CA, and approved December 3, 2014 (received for review October 3, 2014)

Intradiol aromatic ring-cleaving dioxygenases use an active site, color that can be used to monitor the reaction (16). In 3,4-PCD, 3+ 3+ nonheme Fe to activate O2 and catecholic substrates for reaction. the Fe is coordinated by axial ligands Y447 and H462 and 3+ The inability of Fe to directly bind O2 presents a mechanistic conun- equatorial ligands H460, Y408, and solvent-derived hydroxide drum. The reaction mechanism of protocatechuate 3,4-dioxygenase is (Fig. 1) (11). investigated here using the alternative substrate 4-fluorocatechol. This The proposed intermediates for the reaction of 3,4-PCD with 1–8 substrate is found to slow the reaction at several steps throughout PCA and O2 are shown in Fig. 1 and designated (1, 13, 15, 17, – 1 the mechanistic cycle, allowing the intermediates to be detected in 20 23). First, PCA binds to the metal of the resting enzyme in 2 solution studies. When the reaction was initiated in an enzyme crystal, a multistep process leading to a complex where both hydroxyls are ionized and the ring of PCA is axially oriented. Formation of it was found to halt at one of two intermediates depending on the pH 2 of the surrounding solution. The X-ray crystal structure of the inter- results in dissociation of Y447 from the metal, forming a 5- 3+ coordinate (5C) complex with a vacant site trans to H460. O2 reacts mediate at pH 6.5 revealed the key alkylperoxo-Fe species, and the 3+ 3+ with 2 in a concerted process 3, generating a Fe -alkylperoxo anhydride-Fe intermediate was found for a crystal reacted at pH 8.5. species 4. Spectroscopic and computational studies suggest that Intermediates of these types have not been structurally characterized catalysis requires a physical shift of a transient six-coordinate for intradiol dioxygenases, and they validate four decades of spectro- (6C) alkyperoxo 4 to a 5C species 5, opening the axial site trans scopic, kinetic, and computational studies. In contrast to our similar in 6 – crystallo 2+ to H462. Protonation of the peroxo moiety drives O Obond crystallographic studies of an Fe -containing extradiol diox- scission and migration of one oxygen atom into the substrate ring ygenase, no evidence for a superoxo or peroxo intermediate preced- 7 to form an anhydride . Attack of the iron-bound hydroxide on the BIOCHEMISTRY ing the alkylperoxo was found. This observation and the lack of 8 2+ anhydride cleaves the substrate ring to form product complex , spectroscopic evidence for an Fe intermediate that could bind O2 which, upon release, regenerates state 1. Transient kinetic studies are consistent with concerted formation of the alkylperoxo followed have detected intermediates throughout the catalytic cycle, but the by Criegee rearrangement to yield the anhydride and ultimately ring- kinetics of their interconversion have not allowed any in- opened product. Structural comparison of the alkylperoxo intermedi- termediate between the substrate and product complexes to be ates from the intra- and extradiol dioxygenases provides a rationale trapped for detailed characterization (12, 13, 24, 25). for site specificity of ring cleavage. The use of alternative substrates has been a successful means to detect and trap intermediates in enzyme reaction cycles in dioxygenase | oxygen activation | X-ray crystallography | which the native reaction proceeds too quickly or with unfavor- reaction intermediate | Fe(III) able interconversion-rate constants (26). A powerful extension of this approach involves performing catalysis in enzyme crystals to ost ring-cleaving dioxygenases belong to the nonheme trap reactive species for characterization using X-ray crystal- mononuclear iron-containing enzyme family, many mem- lography, which has been successfully applied to a few systems M (27–31). This approach was used by our laboratory to identify bers of which play critical roles in the aerobic biodegradation of natural and synthetic aromatic compounds (1–5). Ring cleavage by these enzymes involves fission of the O–O bond of dioxygen Significance and incorporation of both atoms into the product (6). Two major classes of catechol ring-cleaving dioxygenases have been de- Vast quantities of aromatic compounds enter the environment scribed. These differ in the mode of cleavage and the oxidation due to the natural breakdown of lignin as well as industrial and + state of the iron cofactor: the Fe2 -using extradiol dioxygenases agricultural pollution. Intradiol aromatic ring-cleaving dioxy- (EDOs) cleave adjacent to the vicinal hydroxyl functions of genases play a pivotal role in the biodegradation of these catechols (meta cleavage) while intradiol dioxygenases (IDOs) aromatics. Despite exhaustive study, the mechanism of intra- + use a Fe3 cofactor and cleave between the hydroxyl functions diol dioxygenases has remained elusive because the reaction + (ortho cleavage). The Fe3 in the active site of IDOs will not bind cycle intermediates in which O2 is activated and inserted into the aromatic are too fleeting to be trapped and characterized. O and no direct spectroscopic evidence for redox cycling has 2 Here the intradiol dioxygenase reaction is carried out in been obtained (1). This means that IDOs must activate O using 2 a crystal, allowing the two reaction cycle intermediates that a unique mechanism, which differs from that of the much more 2+ – most clearly define the mechanism to be trapped and their common Fe containing oxygenases that readily form an Fe O2 structures solved. complex after the substrate binds in the active site (7–9).

The archetypal IDO is protocatechuate 3,4-dioxygenase (3,4- Author contributions: C.J.K., V.M.P., and J.D.L. designed research; C.J.K. and V.M.P. per- PCD), which catalyzes cleavage of its native substrate (3,4- formed research; C.J.K. and J.D.L. analyzed data; and C.J.K. and J.D.L. wrote the paper. β dihydroxybenzoate, PCA) to yield -carboxy-cis,cis-muconate, The authors declare no conflict of interest. thereby funneling aromatic substrates into the β-ketoadipate This article is a PNAS Direct Submission. pathway (10). Much of our mechanistic knowledge of IDOs comes from kinetic, spectroscopic, and crystallographic studies Data Deposition: The atomic coordinates and structure factors have been deposited in the – Protein Data Bank, www.pdb.org [PDB ID codes: 4WHO (resting state pH 8.5), 4WHP of 3,4-PCD, catechol 1,2-dioxygenase (11 17), and bio-inspired (resting state pH 6.5), 4WHS (E-4FC pH 8.5), 4WHQ (E-4FC-O2 pH 6.5), and 4WHR 3+ model compounds (18, 19). The 3,4-PCD coordinates the Fe in (E-4FC-O2 pH 8.5)]. a 2-His, 2-Tyr ligand set that is conserved in IDOs (11). The 1To whom correspondence should be addressed. Email: [email protected]. tyrosine ligands contribute several ligand-to-metal-charge trans- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. fer bands (LMCTs) in the visible, which give IDOs a burgundy 1073/pnas.1419118112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1419118112 PNAS Early Edition | 1of6 Downloaded by guest on September 24, 2021 flow. The time courses at 660 nm over a range of 4FC concentrations could each be fit by a single exponential expression (Fig. S1D), consistent with a one-step reaction. However, the plot of reciprocal relaxation time (RRT) versus 4FC concentration is hyperbolic (Fig. S1E). This indicates that the binding process actually occurs in at least two steps where a fast, reversible initial complex is followed by a slower ligand rearrangement or con- formational change that gives rise to the chromophoric change. The Kd of the initial complex (k-1/k1) is given by the apparent Kd (7.5 ± 0.4 mM) for the hyperbolic fit, and the forward (k2 = 34 ± −1 −1 0.9 s ) and reverse (k-2 ∼0.4 s ) rate constants for the second step are determined by analysis of the maximum and intercept values of the hyperbola (SI Materials and Methods) (34). The overall Kd for substrate binding is thus predicted to be ∼88 μM, which compares favorably to the value of 75 μM determined by a direct optical titration (Fig. S1F). The weak initial binding without a chromophoric change may indicate that 4FC first binds in the hydrophobic active site away from the iron. These experiments show

Fig. 1. Proposed intermediates in the reaction of 3,4-PCD with substrate

and O2. R represents either COOH (PCA) or F (4FC). 4FC is bound in the equatorial plane rather than the axial plane as shown in 2. It may rotate to

the axial plane to react. The πop-sym highest occupied molecular orbital of substrate is proposed to be the source of the reducing equivalents (15).

and characterize four distinct oxygenated intermediates in the catalytic cycle of the EDO homoprotocatechuate 2,3-dioxygenase (2,3-HPCD) (29, 30). Here, we have extended this in crystallo approach to the IDO 3,4-PCD as it turns over the alternative substrate 4-fluorocatechol (4FC), allowing determination of the structures of two intermediates following O2 addition. These intermediates provide a direct view of the key steps in the oxygen activation and reaction cycle of IDOs. Insight is also gained into one of the longstanding questions in oxygenase research: the basis for cleavage site specificity in the ring-cleaving dioxygenase family. Results

Characterization of the Reaction of 3,4-PCD, 4FC, and O2 in Solution. The kcat for turnover of 4FC is reduced 100-fold relative to that of PCA at 22 °C, pH 7.5, and the Km values for the organic substrate and O2 are increased eightfold and fivefold, respectively (Table S1). The reaction of 3,4-PCD with 4FC and O yielded 2 − a UV-absorbing product with a λ of 256 nm, e = 17 ± 0.8 mM 1 − max cm 1 as expected for the ortho cleavage product 3-fluoromuconate (Fig. S1A) (32). Accordingly, acidification of the reaction product to pH 2 converted it to the characteristic lactonization product of 3-fluoromuconate (Fig. S1B) (33). A comparison of the optical features of 3,4-PCD and anaerobic E-4FC is shown in Fig. 2A. IDO–catechol complexes exhibit + additional catecholate-to-Fe3 LMCT bands. Although binding Fig. 2. Transient kinetics of the reaction of E-4FC with O2.(A) Optical spectra of 4FC elicited changes in the LMCT bands (most notably at of resting 3,4-PCD (black), the anaerobic E-4FC complex (red), and the extracted > spectra of Int1 (blue) and Int2 (orange) shown at 100 μM concentration. (B) wavelengths 500 nm), the absorbance of the complex decreased Representative single wavelength data (black) for the reaction of 100 μME-4FC relative to that of E-PCA (Fig. S1C). These results suggest that with 950 μMO2 monitored at 550 nm. The blue and orange dotted traces 4FC directly coordinates the metal, but this complex has differ- represent 2- and 3-exponential fits to the data, respectively. Residuals for the ent electronic properties than E-PCA. The structural basis for fits are shown in their respective colors. (C) Plots of the dependence of the sum this change is described below. The anaerobic binding reaction of (blue) and product (orange) of the two largest RRTs for the reaction of 75 μM

3,4-PCD and 4FC at pH 7.5 at 22 °C was studied using stopped E-4FConthedissolvedO2 concentration. Error bars represent one SD.

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1419118112 Knoot et al. Downloaded by guest on September 24, 2021 that none of the steps in the 4FC binding process correspond to the amounts to a roughly 100° rotation of the aromatic plane relative −1 rate-limiting step of the reaction (kcat = 0.71 s at 22 °C). to the plane of PCA in the E-PCA complex (Fig. 3C). Two To study the reaction with O2, anaerobic E-4FC was rapidly conformations of 4FC at roughly equal occupancy were ob- mixed with oxygenated buffer at 4 °C on the stopped flow. When served, evidenced by the residual density apparent for two 100 μM E-4FC (Materials and Methods) was reacted with 950 μM positions of the fluorine. The only difference in the binding O2 at 4 °C, the time course of this pseudo first-order reaction as mode is in the position of fluorine; the iron bond distances and monitored at 550 nm could be fit well by a sum of three expo- position of the ring are essentially identical. The Fe–O4FC bond − nential functions with RRTs of 0.92, 0.50, and 0.16 s 1 (Fig. 2B). lengths are found to be 2.1 Å for both O14FC and O24FC in all of Although the magnitudes of these RRTs are similar, they are the protomers. There is no asymmetry in the Fe–hydroxyl bonds resolvable because the phases have opposite amplitudes at 550 as observed for PCA (20). In each protomer, Y447 is dissociated nm. Three phases imply that there are at least three steps in this from the metal and interacts with the phenol group of Y16; its portion of the reaction cycle. The slowest phase has an RRT very orientation is superimposable with that in the E-PCA complex. −1 similar to kcat from steady-state experiments (0.18 s at 4 °C). In The details of the ligand and Y447 refinement strategies and all previous studies, we and others have found that product re- model occupancies are given in the SI Materials and Methods lease is the rate-limiting step for IDOs, and the preceding ring- (Fig. S2). opening step is irreversible (12, 24, 25, 35). This irreversible step allows the slowest RRT to be associated with the rate constant In Crystallo Reaction of E-4FC with O2 at pH 6.5 and 7.5. Crystals of for a specific step in the reaction, presumably product release. 3,4-PCD were transferred to drops supplemented with 100 mM This assignment is supported by the observation that the spec- 4FC and reacted under aerobic atmosphere at 22 °C for 15–25 trum of the resting enzyme is observed at the completion of the min before freezing. Following the incubation, 4FC ring-cleaved reaction. Only the fastest two phases show a dependence on O2 product was detectable in the drop. Electron-density maps cal- concentration, as expected if the slow product release is pre- culated from data sets at pH 7.5 suggested an averaging of ceded by the irreversible ring-opening step (Fig. S1G). Assuming multiple species bound to the metal in each protomer. No dis- that the fastest two phases arise from two sequential steps in tinct species could be modeled into these active sites. In contrast, which at least the first is reversible, the RRTs will be described when the experiment was repeated at pH 6.5, the resulting by the two roots of a quadratic equation. Consequently, the plots electron-density maps had strong residual positive density for of the sum and products of the RRTs vs. O2 will be linear and all a metal-bound species clearly distinct from the 4FC present in of the rate constants for the steps can be extracted (Fig. 2C and the E-4FC complex. Interpretable electron density was only

SI Materials and Methods). The second fastest step is effectively observed in active site A; solvent was modeled into active sites B BIOCHEMISTRY irreversible as expected. The forward rate constant for the fastest and C to account for the missing density. The residual density in − − step (8.7 × 102 ± 0.8 M 1 s 1) is the bimolecular rate constant protomer A suggested the presence of an axially oriented ring for O2 binding. This value is 600-fold slower than that for O2 with connected density that “wrapped” around the metal in the binding to E-PCA (12). The derived minimum kinetic model for the reaction is:

= : 2 −1 −1 = : −1 = : −1 k3 8 7x10 M !s k4 0 58 s! k5 0 16 s A B E + 4FC + O2 Int1 Int2 ! E + P: −1 I491 k−3 = 0:11s

Using global analysis of the single-turnover reaction monitored at 13 wavelengths, we extracted the pure component optical spectra of the intermediates formed during the reaction R457 with O2 as shown in Fig. 2A. The intermediates have optical spectra with characteristics similar to those reported for transient 2.8 oxygenated intermediates in related enzymes (12, 24). The pres- 2.8 ence of relatively intense optical features for both of the observed 2.1 intermediates shows that the metal remains in the +3 oxidation state, because reduction of the metal would bleach the optical H460 spectrum almost entirely. The rate constants for the inter- 2.1 2.8 conversions preclude trapping of the intermediates at high yield in Wat. solution (Fig. S1H). However, the slow rate of the reaction be- Y408 H462 tween E-4FC and O2 and of each subsequent step proved to be a great advantage in trapping O2-bound intermediates in crystallo. C 4FC Y408 Crystal Structure of 3,4-PCD with 4FC Bound. Here, 3,4-PCD was crystallized in a new space group I222 with three αβ-protomers in PCA

the asymmetric unit, differentiated as protomers A, B, and C in o the model hereafter. No significant differences were noted in the ~100 resting active site structure of I222 crystals of 3,4-PCD and those H460 of the monoclinic C2 crystals previously studied, but the I222 crystals generally diffracted to higher resolution (1.3–1.6 Å limit). Data collection and refinement parameters for each of the Fig. 3. Structure of the anaerobic complex of 3,4-PCD with 4FC (E-4FC) at pH 8.5. (A) Electron-density maps of the active site of the E-4FC complex. The structures solved for this study can be found in Table S2. The j j–j j σ iron occupancy in all structures was refined at 75% based on blue mesh is the 2 Fo Fc map contoured at 1.0 after refinement with 4FC. The green mesh is the jFoj–jFcj ligand-omit map contoured at +3.8 σ.(B) ICP-MS analysis of the metal content of the enzyme batches Bonding environment of 4FC. Only one orientation of 4FC is shown for used for the crystallizations (SI Materials and Methods). clarity. Possible interactions are shown as red dashed lines and the distances Electron density maps from 3,4-PCD crystals soaked with 4FC are given in Ångstroms. (C) Comparison of E-4FC and E-PCA (PDB ID: 3PCA). at pH 8.5 under anaerobic conditions revealed direct binding of The ring of 4FC is rotated 100° relative to that of PCA. Atoms are shown as 4FC to the metal in all three protomers (Fig. 3 A and B). In spheres. Carbon atoms are gray (protein residues), yellow (4FC) or orange contrast to the axial binding mode observed in the PCA complex, (PCA). Oxygen atoms and solvent atoms are colored red, nitrogen atoms 4FC bound in the equatorial plane, trans to H460 and Y408. This blue, and the iron atom brown.

Knoot et al. PNAS Early Edition | 3of6 Downloaded by guest on September 24, 2021 A B Y408 Wat. 2.9 3.0

1.8 H462 2.5

1.9

H460 2.7 2.5

3.2 Q477 ~90° R457 C D Y408 Wat. 2.9 OH 2.1 2.3 H462 2.0 2.6 2.5 H460 2.8

~90° Q477 R457

Fig. 4. Structures of the intermediate species observed upon reacting 3,4-PCD crystals with 4FC and O2 at pH 6.5 and 8.5. (A) Two views of the electron density observed at pH 6.5 and the 4FC-alkylperoxo model (yellow). (B) Bonding environment of the 4FC-alkylperoxo intermediate. (C) Two views of the electron density observed at pH 8.5 and the 4FC-anhydride model (yellow) and hydroxide (red sphere). (D) Bonding environment of the 4FC-anhydride intermediate. In A and C, the blue mesh is the 2jFoj–jFcj map contoured at 1.0 σ after refinement with the ligands and the green mesh is the jFoj–jFcj ligand-omit map contoured at +3.5 σ.InB and D the possible interactions are shown as dashed red lines and the distances are given in Ångstroms. Atom color code is as in Fig. 3.

equatorial plane to the coordination site trans to H460 (Fig. 4A). No intermediates could be modeled into sites B and C. The bond This site is where the peroxo has been proposed to form during distance between the metal-bound hydroxide and the bridging the reaction of O2, and we accounted for this density by mod- O-atom of anhydride is 2.3 Å. The anhydride remains metal- eling an alkylperoxo species in active site A (Fig. 4 A and B, Fig. bound and interacts through the ketonized O24FC (2.0 Å) and •- S3). A 4FC-semiquinone-Fe-O2 was also tested, but the refined the ring remains in the axial orientation. The position of the models strongly supported the existence of a covalent bond be- hydroxide is roughly the same as that of the proximal oxygen of tween C24FC and O2O2 (Fig. S4 A and B). the alkylperoxo moiety, but the Fe bond distance has increased The peroxo bridges between the metal and the substrate to 2.1 Å. Critically, Y447 is found exclusively in the metal-bound (Fe–O1O2 = 1.8 Å, Fe–O2O2 = 2.5 Å, Fe–O–Oangle= 100°). orientation (Fig. S2C). Table S3 lists bond distances and angles. The refined peroxo O–O bond distance was 1.5 Å. The carbon of The possibility of reduction of the complexes and inter- 3 4FC attacked by O2 is sp -hybridized based on the observed bond mediates described here by the synchrotron beam was examined angles and the ring is puckered, consistent with oxidation of 4FC. by recording the optical spectrum of each crystal after data O24FC remains coordinated to the metal (1.9 Å), but it also collection and by comparing the structure of the alkylperoxo interacts with the guanidinium group of R457 (2.5 Å) (Fig. 4B). intermediate solved using synchrotron versus a low intensity in- In contrast, the O14FC ketone is dissociated from the metal house X-ray source (Fig. S5). Each approach showed that the + (3.8 Å) and interacts with the Q477 side chain. Y447 is found in Fe3 state is maintained after data collection. both metal bound and dissociated orientations at roughly equal occupancy (Fig. S2B). As noted, the ring of 4FC is axially ori- Discussion ented, indicating that 4FC rotates during (or before) the reaction The current results show that 3,4-PCD catalyzes ring cleavage of with O2. The Fe–ligand bond distances are considerably shorter 4FC to yield the expected intradiol product, but also that the than in the 2,3-HPCD alkylperoxo (29), consistent with the reaction occurs at a greatly decreased rate relative to that with expected higher metal-oxidation state. the native substrate, PCA. Importantly, the rate decrease is seen throughout the reaction cycle, allowing intermediates in the key In Crystallo Reaction of E-4FC with O2 at pH 8.5. The dramatic effect steps following O2 binding to be detected in solution and in of pH on stabilization of the alkylperoxo intermediate prompted crystallo. The visualization of the alkylperoxo and anhydride us to repeat the experiment at pH 8.5. The ligand-omit maps of intermediates from the IDO reaction cycle provides direct in- the crystals reacted at pH 8.5 clearly indicated splitting of the sight into the reaction mechanism of this enzyme class at a level peroxo moiety and formation of two separate metal-bound spe- that has not previously been possible. Moreover, it allows the cies. In the proposed mechanism, heterolytic O–O bond scission results of four decades of kinetic, mutational, spectroscopic, and of the peroxo leads to formation of an anhydride intermediate computational studies to be considered in a single framework. via a Criegee-type rearrangement where one O atom of O2 These aspects of the study are discussed here. migrates into the ring of substrate, and the second O atom of O2 remains metal bound as hydroxide. Based on the clear evidence Relationship to Spectroscopic and Computational Studies of the IDO of O–O bond scission, a 4FC-derived anhydride and iron- Mechanism. The structural model of the alkylperoxo intermediate hydroxide were modeled into the density (Fig. 4 C and D). A reported here is geometrically nearly identical to the 5C alkyl- lactone intermediate (present in the extradiol pathway) was also peroxo species proposed from computational studies that were tested, but the anhydride provided a better fit (Fig. S4 C and D). based on spectroscopic and structural studies of the E-PCA

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1419118112 Knoot et al. Downloaded by guest on September 24, 2021 complex (15, 22). This convergence suggests that the mechanistic 3,4-PCD insight into the ring-cleavage mechanism for 4FC is relevant to Intradiol acyl migration that for PCA. Several important aspects of the predicted mechanism are confirmed by the structure. First, the initial at- + O2 tack on the substrate by O2 is computed to be a concerted pro- 3 rearrangement cess (Fig. 1, ) that directly forms the bridging peroxo moiety. ketone This is the species we observe (Fig. 4A). In our in crystallo studies dissociated 2+ of 2,3-HPCD, a substrate radical-Fe -superoxo intermediate 2,3-HPCD alkenyl migration precedes the alkylperoxo species (29). Similar discrete precursor Extradiol + complexes, such as a substrate radical-Fe2 or substrate radical- + Fe3 -superoxo species, have been proposed in previous studies of the IDOs and models (22, 36, 37). Although not definitive, we + O2 no rearrangement ketone remains find no evidence for these species in the IDO crystal. More metal-bound significantly, the failure to observe any fully bleached inter- + mediates in the solution studies makes any Fe2 species with Fig. 5. Comparison of the geometries of the alkylperoxo intermediates in a significant lifetime unlikely. These observations are consistent 3,4-PCD and 2,3-HPCD. The coordination mode of the intermediate may dic- with the proposed concerted attack mechanism 3 (15). Second, tate acyl versus alkenyl migration and thereby cleavage specificity. The 2,3- HPCD and 3,4-PCD-PCA models are PDB structures 2IGA and 3PCA, respectively. although the initial concerted O2 attack results in a 6C iron species 4, a rapid rearrangement to a 5C iron species 5 is com- putationally predicted to be required for stability and reestab- lishment of the S = 5/2 spin state of the iron (15, 22). The 3,4-PCD and alkenyl migration to form the lactone would be fa- observed alkylperoxo species is in the predicted shifted position vored by the alternative alignment in 2,3-HPCD alkylperoxo. (Fig. 4A). Third, a subsequent shift in the position of the bridging peroxo moiety is predicted to increase the reactivity of the spe- Equatorial Binding of 4FC and its Effects on the Reaction with O2. cies and establish a geometry 6 that facilitates protonation of the Axial binding of catecholic substrates is a characteristic of all proximal peroxo oxygen (15, 22). This proton is proposed to classes of IDOs (20, 23, 35, 42). The alternative equatorial 4FC drive O–O bond cleavage and insertion. It may derive from binding mode provides insights into the driving force behind the a metal-bound solvent or deprotonation of Y447 as it returns to substrate binding-mediated dissociation of Y447. Two possible 6 origins for this rare endogenous ligand dissociation are direct the iron as seen in . The shifts in the positions of the substrate BIOCHEMISTRY and peroxo would open the exchangeable axial iron binding site displacement by a substrate hydroxyl function and/or a requirement for Y447 or solvent. In the structures reported here, the axial to retain a constant charge at the metal center (13, 14, 23). In the binding site is open in the alkylperoxo intermediate (Fig. 4A), current case, Y447 dissociates when 4FC is bound in the equatorial but Y447 is bound following O–O bond cleavage in the anhy- plane, and thus no direct competition for the iron coordination site dride intermediate (Fig. 4C). These observations strongly sup- is possible (14). On the other hand, coordination of 4FC in the port both the computed mechanism and the role of Y447 as the equatorial plane as a dianion adds two negative charges but only source of the critical proton. displaces one negative hydroxide. The charge would be rebalanced The stopped flow studies described here reveal two inter- by dissociating Y447 to its alternative site, supporting charge maintenance as the basis for ligand reorganization. mediates following addition of O2 to the E-4FC complex. Al- though it is possible that these two intermediates are the species It is unknown whether 4FC can react with O2 while bound in found in crystallo, we cannot make this correlation at present the equatorial plane. Rotation to the axial plane clearly occurs at (Fig. S5 and discussion in Supporting Information). some point during the generation of the alkylperoxo intermediate. Intriguingly, although 4FC is rotated ∼100° relative to PCA in the Intradiol Versus Extradiol Dioxygenase Ring Cleaving Specificity. The substrate complex, the two substrates experience analogous metal IDO and EDO families have each been studied extensively with environments, i.e., 4FC has a trans His ligand (H460) and cis phenolate ligand (Y408) like PCA (trans H462 and cis Y408). the goal of understanding their individual O2 activation mechanisms and the basis for their ring cleaving specificity. It is now Thus, one might expect the rates of O2 attack to be comparable. possible to directly compare representative structures of the Two steric factors might prevent rapid reaction with O2 while 4FC alkylperoxo intermediates from these families. EDOs and IDOs in bound in the equatorial plane. First, the equatorial binding mechanistically converge at an alkylperoxo intermediate, and mode of 4FC does not create a hydrophobic pocket for O2 to bind both families have been proposed to use a concerted Criegee- near the iron as we have noted for the axially bound E-PCA like mechanism for oxygen insertion into substrate (1, 38). Al- complex (Fig. S6) (20, 23). Second, the “slotlike” active site is though the current study does not address the detailed mecha- constructed to promote axial binding. As a result, equatorial nism of oxygen insertion, it does provide insight into the key substrate binding creates tight van der Waals contacts with walls of question: What determines the position of oxygen insertion when the active site (Fig. S6). This would prevent the shift in substrate ostensibly the same alkylperoxo intermediate is formed in each orientation required during the formation of the 5C alkylperoxo enzyme class (21, 36, 39–41)? One possibility supported by intermediate. No such restrictions are present for axial substrate computational studies is that the alignment of the O–O bond binding. Given these restrictions, it seems likely that 4FC is ac- relative to the scissile bonds of the substrate determines whether tually bound as an equilibrium mixture of axial and equatorial acyl (intradiol) or alkenyl (extradiol) migration occurs during the orientations with only a small fraction bound in the axial orien- rearrangement. Migration proceeds preferentially into the co- tation, making it undetectable in the electron density. If only the planar bond (22, 37). Comparison of the geometries of the EDO axial fraction is reactive toward O2, then the 600-fold reduction in and IDO alkylperoxo intermediates shows that they differ sub- the bimolecular reaction rate may be, in part, representative of the stantially in the O–O and C–C bond alignments primarily be- necessity of the substrate to reorient before it can react with O2. cause the substrate bonding to the iron is not the same. In the EDO, the peroxo and both substrate ring oxygen substituents Conclusions coordinate to the iron, while only one hydroxyl coordinates the Together with comparable studies of the EDO reaction inter- iron in the IDO (Fig. 5). In 3,4-PCD this coordination mode mediates, the structures of the alkylperoxo and anhydride inter- comes about as a result of the rearrangement described above mediates of the IDO reaction cycle make available a broad view of and is stabilized by interaction with second sphere conserved the factors that promote O2 activation and specificity in ring- residues Q477 and R457. As shown in Fig. 5, the coplanar O–Oand cleaving dioxygenases. The formation of alkylperoxo intermediates substrate C–C bond alignment favors formation of the anhydride in is widely proposed, especially in the large 2-His-1-carboxylate

Knoot et al. PNAS Early Edition | 5of6 Downloaded by guest on September 24, 2021 + family of nonheme Fe2 oxygenases. Although the mechanism of dependence of the RRTs. Under the latter condition, only the first 6 s of formation differs in the IDO case, the resulting complex is the reaction were mathematically fit to avoid complications by multiple- similar, and thus its structural parameters and the basis for its turnover. RRTs were determined while monitoring in single-wavelength reactivity are significant. Finally, this study adds to a small but mode at 550 nm for the aerobic reactions and separately at 325, 425, and growing list that show enzyme intermediates can be trapped in 660 nm for the anaerobic binding. See the Supporting Information for high yields when the reactions are carried out in a crystal. This fitting methods. approach may provide an experimental inroad into a fundamental question in enzymology concerning the mechanism of efficient Crystal Soaking. The 3,4-PCD crystallization and data collection is described in transit through multistep reaction cycles. the SI Materials and Methods. To perform the in crystallo reactions, single crystals (0.3–0.5 mm) were removed from the crystallization drop and trans- Materials and Methods ferred to drops containing 100 mM 4FC in freshly prepared mother liquor at Transient Kinetics. Transient kinetic data were collected using an Applied pH 6.5, 7.5, or 8.5. Crystals were soaked for 30 min with 4FC in mother liquor Photophysics SX.18MV stopped-flow instrument in the single wavelength containing 25% (vol/vol) glycerol as cryoprotectant and directly frozen in liq- or spectra kinetic mode. For anaerobic binding experiments, Ar-sparged uid nitrogen. For anaerobic soaks, crystals were grown anaerobically before + 100 μM3,4-PCD(Fe3 -loaded active sites) was rapidly mixed with anaer- soaking for 10 min with Ar-sparged 50 mM 4FC in mother liquor supple- obic buffer (0.3 M Hepes pH 7.5) containing various concentrations of Ar- mented with cryoprotectant.

sparged 4FC substrate. For reactions with O2, the enzyme-4FC complex was preformed in an anaerobic bag, transferred to the stopped flow and ACKNOWLEDGMENTS. Diffraction data were collected at Argonne National mixed with oxygenated Hepes buffer at 4 °C, pH 7.5. Because of the re- Laboratory, Structural Biology Center at the Advanced Photon Source. versibility of the 4FC binding, a stoichiometric complex of enzyme and 4FC Argonne is operated by UChicago Argonne, LLC, for the U.S. Department of Energy, Office of Biological and Environmental Research under contract DE- could not be formed with equal concentrations of 3,4-PCD and 4FC. The AC02-06CH11357. We thank Prof. Carrie Wilmot and Dr. Ed Hoeffner at the reaction was studied under two alternative conditions (i): single turnover University of Minnesota Kahlert Structural Biology Laboratory for many μ μ was done with excess enzyme and limiting 4FC (900 M 3,4-PCD, 250 M discussions and data collection. This work was supported by National ∼ μ 4FC, 200 M E-4FC) reacted with 1.9 mM O2 or (ii) multiple turnover Institutes of Health Grants GM24689 (to J.D.L.) and graduate traineeship conditions (150 μM 3,4-PCD, 4 mM 4FC, ∼150 μM E-4FC) to study the O2 GM08700 (to C.J.K.).

1. Lipscomb JD, Orville AM (1992) Mechanistic aspects of dihydroxybenzoate dioxyge- 22. Borowski T, Siegbahn PEM (2006) Mechanism for catechol ring cleavage by non-heme nases. Met Ions Biol Syst 28:243–298. iron intradiol dioxygenases: A hybrid DFT study. J Am Chem Soc 128(39):12941–12953. 2. Vaillancourt FH, Bolin JT, Eltis LD (2006) The ins and outs of ring-cleaving dioxyge- 23. Vetting MW, D’Argenio DA, Ornston LN, Ohlendorf DH (2000) Structure of Acineto- nases. Crit Rev Biochem Mol Biol 41(4):241–267. bacter strain ADP1 protocatechuate 3, 4-dioxygenase at 2.2 A resolution: Implications 3. Olaniran AO, Igbinosa EO (2011) Chlorophenols and other related derivatives of en- for the mechanism of an intradiol dioxygenase. Biochemistry 39(27):7943–7955. vironmental concern: Properties, distribution and microbial degradation processes. 24. Walsh TA, Ballou DP, Mayer R, Que L, Jr (1983) Rapid reaction studies on the oxy- Chemosphere 83(10):1297–1306. genation reactions of catechol dioxygenase. J Biol Chem 258(23):14422–14427. 4. Díaz E, Jiménez JI, Nogales J (2013) Aerobic degradation of aromatic compounds. Curr 25. Nakata H, Yamauchi T, Fujisawa H (1978) Studies on the reaction intermediate of Opin Biotechnol 24(3):431–442. protocatechuate 3,4-dioxygenase. Formation of enzyme-product complex. Biochim 5. Wells T, Jr, Ragauskas AJ (2012) Biotechnological opportunities with the β-ketoadi- Biophys Acta 527(1):171–181. pate pathway. Trends Biotechnol 30(12):627–637. 26. Price JC, Barr EW, Hoffart LM, Krebs C, Bollinger JM, Jr (2005) Kinetic dissection of the 6. Hayaishi O, Katagiri H, Rothberg S (1955) Mechanism of the pyrocatechase reaction. catalytic mechanism of taurine:alpha-ketoglutarate dioxygenase (TauD) from Es- – J Am Chem Soc 77(20):5450–5451. cherichia coli. Biochemistry 44(22):8138 8147. 7. Mbughuni MM, et al. (2010) Trapping and spectroscopic characterization of an FeIII- 27. Wilmot CM, Hajdu J, McPherson MJ, Knowles PF, Phillips SEV (1999) Visualization of – superoxo intermediate from a nonheme mononuclear iron-containing enzyme. Proc dioxygen bound to copper during enzyme catalysis. Science 286(5445):1724 1728. Natl Acad Sci USA 107(39):16788–16793. 28. Schlichting I, et al. (2000) The catalytic pathway of cytochrome p450cam at atomic – 8. Kovaleva EG, Lipscomb JD (2008) Versatility of biological non-heme Fe(II) centers in resolution. Science 287(5458):1615 1622. 2+ oxygen activation reactions. Nat Chem Biol 4(3):186–193. 29. Kovaleva EG, Lipscomb JD (2007) Crystal structures of Fe dioxygenase superoxo, – 9. Lipscomb JD (2008) Mechanism of extradiol aromatic ring-cleaving dioxygenases. Curr alkylperoxo, and bound product intermediates. Science 316(5823):453 457. Opin Struct Biol 18(6):644–649. 30. Kovaleva EG, Lipscomb JD (2008) Intermediate in the O-O bond cleavage reaction of – 10. Ornston LN, Stanier RY (1964) Mechanism of β-ketoadipate formation by bacteria. an extradiol dioxygenase. Biochemistry 47(43):11168 11170. 31. Jeoung J-H, Bommer M, Lin T-Y, Dobbek H (2013) Visualizing the substrate-, Nature 204(4965):1279–1283. superoxo-, alkylperoxo-, and product-bound states at the nonheme Fe(II) site of 11. Ohlendorf DH, Lipscomb JD, Weber PC (1988) Structure and assembly of proto- homogentisate dioxygenase. Proc Natl Acad Sci USA 110(31):12625–12630. catechuate 3,4-dioxygenase. Nature 336(6197):403–405. 32. Dorn E, Knackmuss HJ (1978) Chemical structure and biodegradability of halogenated 12. Bull C, Ballou DP, Otsuka S (1981) The reaction of oxygen with protocatechuate 3,4- aromatic compounds. Substituent effects on 1,2-dioxygenation of catechol. Biochem J dioxygenase from Pseudomonas putida. Characterization of a new oxygenated in- 174(1):85–94. termediate. J Biol Chem 256(24):12681–12686. + 33. Pieken WA, Kozarich JW (1990) Lactonization of cis,cis-3-halomuconates: Influence of 13. Frazee RW, et al. (1998) The axial tyrosinate Fe3 ligand in protocatechuate 3,4- pH and halo substituent on the regiochemistry. J Org Chem 55(10):3029–3035. dioxygenase influences substrate binding and product release: Evidence for new 34. Groce SL, Miller-Rodeberg MA, Lipscomb JD (2004) Single-turnover kinetics of ho- reaction cycle intermediates. Biochemistry 37(8):2131–2144. moprotocatechuate 2,3-dioxygenase. Biochemistry 43(48):15141–15153. 14. Valley MP, et al. (2005) Roles of the equatorial tyrosyl iron ligand of protocatechuate 35. Elgren TE, et al. (1997) Crystal structure and resonance Raman studies of proto- 3,4-dioxygenase in catalysis. Biochemistry 44(33):11024–11039. catechuate 3,4-dioxygenase complexed with 3,4-dihydroxyphenylacetate. Biochemistry 15. Pau MYM, Lipscomb JD, Solomon EI (2007) Substrate activation for O reactions by 2 36(38):11504–11513. oxidized metal centers in biology. Proc Natl Acad Sci USA 104(47):18355–18362. 36. Bugg TDH, Lin G (2001) Solving the riddle of the intradiol and extradiol catechol di- 16. Davis MI, et al. (2002) Spectroscopic and electronic structure studies of proto- oxygenases: How do enzymes control hydroperoxide rearrangements? Chem Commun catechuate 3,4-dioxygenase: Nature of tyrosinate-Fe(III) bonds and their contribution (Camb) (11):941–952. – to reactivity. J Am Chem Soc 124(4):602 614. 37. Georgiev V, Noack H, Borowski T, Blomberg MR, Siegbahn PE (2010) DFT study on the 17. Horsman GP, et al. (2005) Spectroscopic studies of the anaerobic enzyme-substrate catalytic reactivity of a functional model complex for intradiol-cleaving dioxygenases. – complex of catechol 1,2-dioxygenase. J Am Chem Soc 127(48):16882 16891. J Phys Chem B 114(17):5878–5885. 18. Fujii H, Funahashi Y (2002) A trigonal-bipyramidal ferric aqua complex with a steri- 38. Bugg TDH, Ramaswamy S (2008) Non-heme iron-dependent dioxygenases: Unravelling cally hindered salen ligand as a model for the active site of protocatechuate 3,4- catalytic mechanisms for complex enzymatic oxidations. Curr Opin Chem Biol 12(2): – dioxygenase. Angew Chem Int Ed Engl 41(19):3638 3641, 3517. 134–140. 19. Costas M, Mehn MP, Jensen MP, Que L, Jr (2004) Dioxygen activation at mononuclear 39. Siegbahn PEM, Haeffner F (2004) Mechanism for catechol ring-cleavage by non-heme nonheme iron active sites: Enzymes, models, and intermediates. Chem Rev 104(2): iron extradiol dioxygenases. J Am Chem Soc 126(29):8919–8932. 939–986. 40. Jo D-H, Que L, Jr (2000) Tuning the regiospecificity of cleavage in FeIII catecholate 20. Orville AM, Lipscomb JD, Ohlendorf DH (1997) Crystal structures of substrate and complexes: Tridentate facial versus meridional ligands. Angew Chem Int Ed 39(23): substrate analog complexes of protocatechuate 3,4-dioxygenase: Endogenous 4284–4287. + Fe3 ligand displacement in response to substrate binding. Biochemistry 36(33): 41. Váradi T, et al. (2013) Iron(III) complexes with meridional ligands as functional models 10052–10066. of intradiol-cleaving catechol dioxygenases. Inorg Chem 52(3):1559–1569. 21. Xin M, Bugg TDH (2008) Evidence from mechanistic probes for distinct hydroperoxide 42. Matera I, et al. (2010) Catechol 1,2-dioxygenase from the Gram-positive Rhodococcus rearrangement mechanisms in the intradiol and extradiol catechol dioxygenases. opacus 1CP: Quantitative structure/activity relationship and the crystal structures of J Am Chem Soc 130(31):10422–10430. native enzyme and catechols adducts. J Struct Biol 170(3):548–564.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1419118112 Knoot et al. Downloaded by guest on September 24, 2021