The Mechanism of Inhibition of Pyruvate Formate Lyase by Methacrylate

Pyruvate Formate Lyase (PFL) catalyzes acetyl transfer from pyruvate to coenzyme a by a mechanism involving multiple amino acid radicals. A post-translationally installed glycyl radical (G734· in Escherichia coli) is essential for enzyme activity and two cysteines (C418 and C419) are proposed to form thiyl radicals during turnover, yet their unique roles in catalysis have not been directly demonstrated with both structural and electronic resolution. Methacrylate is an isostructural analog of pyruvate and an informative irreversible inhibitor of pfl. Here we demonstrate the mechanism of inhibition of pfl by methacrylate. Treatment of activated pfl with methacrylate results in the conversion of the G734· to a new radical species, concomitant with enzyme inhibition, centered at g = 2.0033. Spectral simulations, reactions with methacrylate isotopologues, and Density Functional Theory (DFT) calculations support our assignment of the radical to a C2 tertiary methacryl radical. The reaction is specific for C418, as evidenced by mass spectrometry. The methacryl radical decays over time, reforming G734·, and the decay exhibits a H/D solvent isotope effect of 3.4, consistent with H-atom transfer from an ionizable donor, presumably the C419 sulfhydryl group. Acrylate also inhibits PFL irreversibly, and alkylates C418, but we did not observe an acryl secondary radical in H2O or in D2O within 10 s, consistent with our DFT calculations and the expected reactivity of a secondary versus tertiary carbon-centered radical. Together, the results support unique roles of the two active site cysteines of PFL and a C419 S–H bond dissociation energy between that of a secondary and tertiary C–H bond.


■ INTRODUCTION
−11 It is this fleeting C• that catalyzes substrate transformations during turnover.Radical transfer from G• to a pair of cysteines in the active site of PFL is unique among known GREs, and other thiyl radical enzymes such as ribonucleotide reductases, which only contain one substrate activating cysteine. 9,12,13s a free amino acid, G• exhibits a reduction potential 110 mV lower than the corresponding C•. 14−19 Absent radical reduction potential changes due to the protein environment, 20,21 this difference in reactivity presents an intrinsic challenge to the study of the role of thiyl radicals in GRE catalysis.Consistent with this apparent difference in radical stability, no thiyl radical has been directly observed in the catalytic cycle of PFL, or any other GRE, to our knowledge.The mechanism of PFL and the role of the two cysteines are currently inferred from X-ray crystallographic structures 22−24 and the effect of mutagenesis, 25,26 isotopic substitution, 27,28 and mechanism-based inhibitors 29−34 on enzyme activity and G• characteristics, focused on the PFL from Escherichia coli.A plausible mechanistic proposal is depicted in Scheme 1, which has gained some level of consensus.In this mechanism, radical transfer occurs via a pathway composed of G 734 • ⇌ C 419 ⇌ C 418 (E. coli numbering) by sequential proton-coupled electron transfer events.The C 418 • then attacks the C2 carbonyl of pyruvate, followed by radical rearrangement and homolysis of the C1−C2 bond of pyruvate to generate an acetylated C 418 and a CO 2 • − radical anion, that abstracts a H atom from C 419 .Equilibration of C 419 • back to G 734 • completes the ping phase of the ping-pong mechanism. 27In the pong phase, H atom abstraction from the CoA sulfhydryl by C 419 •, acetyl exchange, and radical equilibration completes the catalytic cycle.While this mechanistic proposal is consistent with a majority of structural and biochemical observations, the lack of direct observations of radical chemistry with structural specificity leaves ambiguity regarding the unique mechanism and the role of the two cysteines.
Methacrylate is a substrate analog of pyruvate and acts as an irreversible mechanism-based inhibitor of PFL. 34It is particularly illuminating as it provides a compelling case for the unique role of the two thiyl radicals during the catalytic cycle.Incubation of active PFL (aPFL) containing the essential G 734 • with 1 mM methacrylate results in complete and irreversible inhibition over 30 min with no apparent effect on the G 734 •. 34 The product of inhibition is a thioether bond between C 418 and C3 of methacrylate, which is reduced stereoselectively to a S-(2-carboxy-(2S)-propyl) adduct. 34This biochemical data and the X-ray structure support the mechanism depicted in Scheme 1 and distinct roles for C 418 and C 419 .Density functional theory (DFT) calculations also support a preference for C 418 • reaction with methacrylate over C 419 • and a slow reduction of the tertiary C2 methacryl radical by the C 419 sulfhydryl, relative to inactivation. 35Unfortunately, these predictions are not supported by direct experimental evidence, limiting the mechanistic insight on the unique role of C 418 and C 419 .
Here, we report the characterization of the methacryl radical generated upon inhibition of E. coli aPFL.Higher concentrations of the inhibitor and faster quenching of the reaction were critical to evidencing the intermediate by X-band electron paramagnetic resonance (EPR) spectroscopy, revealing an inhibitor radical consistent with a tertiary C2 radical of methacrylate.Proteolytic digestion of inhibited PFL and peptide analysis by liquid chromatography-tandem mass spectrometry (LC-MS/MS) resolves a C 418 -methacrylate adduct.Density functional theory (DFT) calculations based on this model and prior studies 35 reproduce the spectroscopic properties of the C2 radical, and support the stereoselective nature of radical reduction. 34Kinetic analysis of the C2 radical decay demonstrates that the reduction of the inhibitor radical is indeed rate determining, as predicted theoretically. 35Using a combination of the methacrylate analog, acrylate, solvent kinetic isotope effects (KIE), and DFT calculations we estimate the BDE of the C 419 thiol S−H bond, placing the first energetic and mechanistic constraint on the H atom transfer chemistry of this residue.The data are consistent with a C 418 • as the radical that adds to C3, and C 419 as the H atom donor to the C2 radical.
Construction of Plasmids.In order to produce and purify E. coli PFL, we cloned the pflB gene into a modified pCm1 plasmid (Addgene #174361) using Gibson assembly by PCR. 38,39The pCm1 plasmid was modified to encode a 6 × polyhistidine tag (His-tag), followed by an SSG spacer and a TEV protease site N-terminal to the codified protein, and the previous C-terminal tag from pCm1 was removed.We term this new backbone plasmid pCm8.The pf lB gene was also cloned into pCm8 using Gibson assembly by PCR.A DNA fragment containing pf lB gene was amplified by PCR using the primers: Forward: 5′−tcagggcatgtccgagcttaatgaaaagttagcc−3′ Reverse: 5′−agcctaggttacatagattgagtgaaggtacgagtaataacgt−3′ The primers used to amplify the vector backbone were: Forward: 5′−tcaatctatgtaacctaggctgctaaacaaagccc−3′ Reverse: 5′−tcggacatgccctgaaaatacaggt−3′ DNA fragments were assembled using NEBuilder HiFi DNA Assembly Master Mix following the manufacturer instructions.Assembly reaction products were transformed into E. coli DH5α cells and streaked into chloramphenicol LB-agar plates.Successful transformants were identified using colony PCR with the Sapphir-eAmp fast PCR-hot-start master mix, as directed by the manufacturer, and the fidelity of the cloning was confirmed by Sanger sequencing through UC Berkeley DNA Sequencing Facility using the T7 promoter/terminator and pf lB specific primers: Forward: 5′−ggtccggctaccaacgctc−3′ Reverse: 5′−ccaggtcaacagccaggtc−3′ To produce PFL mutants C 418 S and C 419 S we performed sitedirected mutagenesis (SDM).For C 418 S we used the following primers: Forward: 5′−acgatgactacgctattgctagctgcgtaagcc−3′ Reverse: 5′−ggcttacgcagctagcaatagcgtagtcatcgt−3′ For C 419 S we used the following primers: Forward: 5′−ctacgctattgcttgcagcgtaagcccgatgat−3′ Reverse: 5′−atcatcgggcttacgctgcaagcaatagcgtag−3′ We used Phusion polymerase to perform SDM by PCR per the manufacturer's instructions and treated the PCR products with DpnI at 37 °C overnight and then transformed them into E. coli DH5α electrocompetent cells and streaked the cells on chloramphenicol LBagar plates.We confirmed the mutagenesis fidelity by Sanger sequencing as described above.
Protein Expression and Purification.The expression and purification of the TEV protease was performed as previously described. 36e expressed wild-type (wt), C 418 S, and C 419 S PFL using the same protocol.BL21(DE3) cells transformed with the corresponding plasmid were inoculated into a 15 mL preculture of LB broth supplemented with 50 μg/mL of chloramphenicol.The preculture was grown overnight at 37 °C in a 50 mL centrifuge tube shaking at 200 r.p.m.The following day, 15 mL of the overnight preculture were inoculated into 1.5 L of terrific broth, supplemented with 50 μg/mL of chloramphenicol, and grown at 37 °C, shaking at 200 r.p.m.Protein expression was induced by adding 0.25 mM IPTG when the OD 600 reached 1.0, and then the culture was maintained at 25 °C overnight.Cells were harvested by centrifugation at 8,000 × g for 10 min, and the collected cell paste, approximately 10−15 g wet cell paste per liter, was flash frozen in liquid N 2 and stored at −80 °C until purification.
For PLF protein purification we resuspended cell paste in resuspension buffer consisting of 50 mM potassium phosphate (KPi), 150 mM NaCl, 1 mM DTT, and 5% glycerol, adjusted to pH 8.0, and lysed the cells by French press in an Emulsiflex C3 homogenizer at 14,000 psi.Every following step was performed at 4 °C.The cell extract was clarified by centrifugation at 30,000 × g for 30 min, and the cell debris was discarded.To remove nucleic acids, 1.5% w/v streptomycin sulfate was added dropwise while stirring and incubated for 10 min before being centrifuged at 30,000 × g for 30 min.The pellet was discarded, and the supernatant was filtered through a 0.65 μm filter.The clarified lysate was applied to a 20 mL Ni-NTA column equilibrated with wash buffer composed of 30 mM imidazole, 50 mM KPi, 150 mM NaCl, 1 mM DTT, and 5% glycerol at pH 8.0.The resin was washed with 20 column volumes of wash buffer and eluted with elution buffer in which the imidazole concentration was raised to 400 mM.Fractions containing protein were pooled and concentrated using Amicon Ultra-15 50 kDa molecular weight cutoff (MWCO) centrifugal filter units and then buffer exchanged to resuspension buffer using a HiTrap desalting 5 mL column.To remove the N-terminal His-tag, PFL was incubated with TEV protease in a 1:100 ratio of protease to protein overnight and then the digestion product was injected back into the Ni-NTA column to remove the cleaved N-terminal tag, His-tagged protease, and undigested PFL.The Ni-NTA flow-through was collected and purified, and the untagged protein was concentrated and desalted as described above.The purified protein was stored in resuspension buffer supplemented with 20% w/v glycerol at −80 °C until used.
We modified a previously reported method to express and purify PFL-AE. 40BL21(DE3)pLysS cells transformed with pCAL-n-EK-pflA were inoculated into a 15 mL preculture of LB broth supplemented with 100 μg/mL of carbenicillin and 35 μg/mL of chloramphenicol.The preculture was grown overnight at 37 °C shaking at 200 r.p.m. 3 mL of cells from the overnight preculture were inoculated into 300 mL of LB broth supplemented with 100 mM KPi buffer pH 7.0, 100 μg/mL of carbenicillin and 35 μg/mL of chloramphenicol and grew at 37 °C with shaking at 200 r.p.m.PFL-AE expression was induced by adding 0.25 mM IPTG, 0.2 mM L-cysteine, and 0.2 mM (NH 4 )-Fe II (SO 4 ) 2 at OD 600 of 0.8.The culture was maintained at 30 °C for 5 h, and then L-cysteine and (NH 4 )Fe II (SO 4 ) 2 were added to a final concentration of 0.4 mM.The culture was moved to a 4 °C refrigerator and sparged with Argon overnight (15−16 h).
For PFL-AE purification, the culture flasks were moved into a coy glovebox <20 ppm of O 2 and transferred to centrifuge bottles with an O-ring seal.Cell paste was harvested by centrifugation at 8,000 × g for 5 min.All following steps were performed in the glovebox.Cells were resuspended into anaerobic lysis buffer consisting of 50 mM Tris, 50 mM KCl, 10 mM MgCl 2 , 1% w/v Triton X-100, and 5% w/v glycerol at pH 7.5 in a proportion of 1 mL of buffer per gram of wet cell paste.The resuspended cell pellet slurry was supplemented with 1 mM phenylmethylsulfonyl fluoride, 0.2 mg/mL lysozyme, 1 mM DTT and 10 μg/mL of RNase and DNase, and homogenized on ice, with gentle stirring, for 1 h.The cell slurry was then lysed by ultrasonication with a Qsonica q125 sonicator for 10 min, in cycles of 20 s on−30 s off, on ice.The sample was clarified by centrifugation at 15,000 × g for 15 min and filtered using a 0.65 μm filter.The protein extract was then injected into a 70 mL Sephacryl S-100 HR size exclusion column, equilibrated with buffer consisting of 50 mM Tris, 100 mM KCl, and 1 mM DTT at pH 7.5, and resolved at a flow rate of 0.5 mL/min.We collected the dark brown fractions and concentrated them using Amicon Ultra-0.5 10 kDa MWCO centrifugal filter units.The concentrated protein was reinjected in the same column under the same conditions to further improve protein purity.Protein-complexed iron was quantified using the ferrozine assay, 41 using iron ICP-MS standards (TraceCERT) as an iron standard.Protein concentrations were quantified using the Bradford assay, using bovine serum albumin as a standard. 42FL Activation.We activated PFL in an anaerobic VAC Atmospheres glovebox (<2 ppm of O 2 ) by mixing 50 μM PFL, 5 μM PFL-AE, 2 mM SAM, 20 mM oxamate, and 100 μM 5deazariboflavin in activation buffer containing 100 mM Tris, 100 mM KCl, 10 mM DTT, and 20% w/v glycerol at pH 7.6 in a 4 mm O.D. EPR tube.We exposed the mixture to a 1 W 405 nm LED light (Thor Laboratories) for 1 h in a thermal bath at 30 °C.The extent of activation was estimated by activity assays and EPR quantitation of the G• (see below).
PFL Activity Assays.We measured PFL activity spectrophotometrically using a multienzyme assay that couples the PFL-dependent formation of acetyl-CoA to the production of NADH by the oxidation of malate to oxaloacetate and condensation to citrate by malate dehydrogenase and citrate synthase, respectively, as previously reported. 40Reactions were initiated by addition of aPFL, and the rate of NADH production was calculated using an extinction coefficient of 6.2 mM −1 cm −1 . 43We mixed 5 μL of aPFL at an approximate concentration of 2 μM G•, with 800 μL of the activity assay mix containing 10 mM DTT, 1 mM NAD + , 10 mM malate, 2 U/mL citrate synthase, 30 U/mL malate dehydrogenase, and 0.05 mg/mL bovine serum albumin in 100 mM Tris buffer, adjusted to pH 8.1.Reaction initial velocities were determined by detecting NADH production by the absorbance peak at 340 nm, in a VAC Atmospheres glovebox using a custom fiber-coupled Ocean Optics QEPro spectrophotometer and DH-2000-BAL light source.
In order to determine aPFL kinetic parameters, we assayed activity at different concentrations of pyruvate from 0 to 10 mM (120 μM CoA), and 0 to 120 μM for CoA (10 mM pyruvate).In order to calculate specific activities (U/mg, 1 U = 1 μmol/min) we normalized the activity to the measured G• concentration determined by EPR (see below) assuming 1 mol of G• per mole dimer of active PFL, and fitted the calculated initial velocities to a Michaelis−Menten model described by eq 1, using GraphPad Prism version 8.0.0.
where V 0 represents the initial velocity, V max the maximum velocity, [S] the substrate concentration, and K m the Michaelis constant.
For enzyme inactivation kinetics we incubated samples of aPFL with 200 mM methacrylate or acrylate between 0 and 10 min.We took sample aliquots during inactivation and immediately determined the remaining PFL activity as described above.Inhibition time points were performed in duplicate and the average is reported.
X-Band EPR Spectroscopy.All EPR samples were prepared in a VAC Atmosphere glovebox with <2 ppm of O 2 .To analyze radical intermediates during aPFL inactivation we mixed 180 μL of 50 μM aPFL, in activation buffer, with 20 μL of 2 M methacrylate or acrylate in H 2 O pH adjusted to 7.0.Samples were incubated at room temperature from 10 s to 7 min.The reactions were flash frozen in EPR tubes in liquid nitrogen-cooled isopentane (<−130 °C).EPR spectra of the samples were collected using a Bruker EMXplus EPR spectrometer at 100 K with a frequency of 9.38−9.44GHz, power of 20 μW, modulation amplitude of 2 G, modulation frequency of 100 kHz, time constant of 0.01 ms, scan time of 20 s, and conversion time of 16 ms.All reported spectra are the average of 30 scans.Spin quantitation was computed from the double integral of the first harmonic signal and referenced to a 4-hydroxy-TEMPO concentration standard.All EPR spectral simulations were performed using EasySpin 6.0.0 software, and confidence intervals and standard deviations of the fitting parameters are reported. 44or preparing solvent H/D isotope effect samples, a D 2 O-based buffer was prepared by lyophilizing 100 mM KCl and 100 mM Tris at pH 7.2 and rehydrating the buffer in the same volume of D 2 O.The D 2 O-exchanged solution was measured to be pH* of 7.2, corresponding to a final pD of 7.6 according to eq 2. 45 pD pH 0.4 To ensure water content <5% in the D 2 O samples, aPFL protein samples were subject to three cycles of 5-fold concentration, followed by 5-fold dilution in D 2 O buffer using 50 kDa MWCO centrifugal filters.
Kinetic constants for the C2• formation and decay in H 2 O and in D 2 O were determined from fitting of the relative G• and C2• contributions to the composite EPR spectra as a function of time in COPASI 4.40.The reaction was fit to an irreversible G• → C2• (k 1 ) and then C2• → G•* (k 2 ) model where G•* corresponds to the inhibited PFL G•.The error associated with the fit is reported as the error for the extracted rate constants, and the solvent kinetic isotope effect is reported as Peptide Liquid Chromatography-Tandem Mass Spectrometry.We assessed the formation of covalent adducts of methacrylate and acrylate with PFL using tryptic digestion of the protein followed by liquid chromatography tandem mass spectrometry (LC-MS/MS).To digest PFL 100 μg of protein was denatured in 20 μL of 8 M urea in 100 mM ammonium bicarbonate, and 5 mM DTT was added and incubated for 30 min at 37 °C to reduce cysteines.Unreacted cysteines were alkylated with 15 mM iodoacetamide for 30 min in the dark, and the reaction was quenched by adding 20 mM DTT and incubating for 10 min.PFL samples were digested with 0.8 μg Lys-C for 2 h.Next, the digested samples were diluted to <2 M urea using 100 mM ammonium bicarbonate and 0.16 μg of Trypsin were added, and digested at 37 °C overnight.Reactions were stopped by adding formic acid to a final concentration of 1% v/v.
The digested samples (2−5 μg at 0.4−1 μg/μL) were injected into a Waters Acquity H-class Ultra High-Pressure Liquid Chromatography system coupled with a Waters Xevo G2-XS quadrupole time-offlight (qToF) mass spectrometer.The UPLC stationary phase was a Waters BEH C18 column of 50 mm × 2.1 mm × 1.7 μm and the mobile polar phase a solution of H 2 O with 0.1% v/v formic acid (A) and the apolar mobile phase acetonitrile 0.1% v/v formic acid (B).A solution of 150 pg/μL Leu-enkephalin and 100 fmol/μL Glufibrinopeptide B prepared in 25:75 acetonitrile:H 2 O with 0.1% v/v formic acid was used as a lock mass, using the average of 3 scans injected every 30 s. Peptides were resolved in a linear gradient from 20 to 35% B over 30 min with a flow rate of 0.2 mL/min at 60 °C.Eluent from the LC was injected directly into the qToF.The mass spectrometer settings were as follow: source capillary voltage 2.8 kV, sampling cone voltage 20 V, source offset 20 V, source temperature 120 °C, desolvation temperature 350 °C, cone gas flow was 50 L/h and desolvation gas flow was 800 L/h.The spectrometer was set in MS/MS detection mode in positive polarity in the resolution regime, detecting from 50 to 2000 Da, with a scan time of 3 s.Peptides corresponding with the mass of the expected tryptic peptide with the methacrylate, acrylate, or carbamidomethyl modification were fragmented using an energy ramp from 20 to 35 V and a cone voltage of 20 V. The LC-MS/MS data were analyzed using Masslynx.
Computational Analysis.We obtained coordinate files for the methacryl radical from a previous DFT geometry optimized structure using a small model containing methacrylate and the C 418 −C 419 dipeptide for EPR spectral parameter predictions. 35Rotamers were constructed in ChimeraX by iterative 10°rotation of the methacrylate moiety about the C 418 -bonded C3−C2 bond, and angles are reported relative to the initial pro-(R) H−C3−C2-C1 dihedral angle of 38°.Single point energies and EPR parameters at each angle were computed for the anionic methacryl radical in Orca 5.0.1 46 using DFT 47,48 with the B3LYP 49 unrestrained hybrid functional and a 6-311++G(d,p) basis set and a conductor-like polarization continuum model 50 dielectric and index of refraction of chloroform to model the protein environment.Spin contamination was evaluated by the computed expectation value for S 2 , which did not exceed 0.02 above the ideal value of 0.75 for a pure doublet spin system.
Reaction trajectories for methacrylate and acrylate were evaluated in two steps by geometry optimizing reactant and product states and then performing a transition state search.For reactions with methacrylate, the C4 methyl group in the methacrylate structures was replaced by a H and geometry optimized as above.The first step involved the radical Michael addition of C 418 −S• to C3, and we used a small model previously geometry optimized composed of a methyl thiyl radical and methacrylate/acrylate (step 1). 35For the second step, methacryl/acryl radical C2• reduction by C 419 −SH (step 2), a second methyl sulfide was included, again starting from a previously geometry optimized state. 35All transition states exhibited a single imaginary vibrational frequency, and energies are reported as the sum of Gibbs free energy correction (G − E el ) and electronic energy (E el ).To normalize the energy of the overall reaction, the transition state (TS) and product (P) of both step 1 and step 2 were evaluated relative to their respective reactant (R1 or R2), and then the energies of R2, TS2, and P2 were scaled to P1 (P1 = R2).
The kinetic isotope effect of methacryl radical reduction in step 2 was computed using transition state theory by recomputing the Gibbs free energy of activation (ΔG ‡ ) for the deuterated isotopologue reactant and transition states, assuming the same geometry for the deuterated and protonated reactant and transition states, and calculating k H /k D using eq 3; where k H/D are the rate of the corresponding chemical step with either H or D, R is the ideal gas constant, and T is the temperature in Kelvin.
A similar treatment was used to compute the ratio of the rate for inhibitor radical reduction by C 419 for acrylate (k acryl ) versus methacrylate (k methacryl ).

Pyruvate Formate Lyase Inactivation by Methacrylate.
To investigate the mechanism of PFL inhibition by methacrylate we recombinantly expressed PFL, including wildtype (wt), C 418 S, and C 419 S variants, as well as PFL-AE in E. coli.The PFL proteins were purified by Ni-affinity chromatography and the N-terminal 6 × His tag was subsequently removed, resulting in an overall protein purity of >95% in all cases, based on sodium dodecyl sulfate polyacrylamide gel electrophoresis (Supporting Information Figure S1).The expression yields were 10−20 mg of purified PFL protein per gram of wet cell paste.During the recombinant expression of PFL-AE we applied a continuous sparge of argon to the cell culture after induction to protect the [Fe 4 S 4 ] cluster from oxidation.Following anaerobic purification, PFL-AE was estimated to be 60% pure with an overall yield of 10 mg of protein per gram of wet cell paste (Supporting Information Figure S2).Reconstitution of the [Fe 4 S 4 ] cluster resulted in PFL-AE iron content of 2.0 ± 0.2 Fe/mol of PFL-AE, likely an underestimate based on the protein impurities.
Treatment of PFL with PFL-AE, in the presence of a photochemical reduction system, and illumination by a 405 nm LED resulted in the accumulation of an asymmetric doublet EPR signal characteristic of the E. coli aPFL G 734 • product (Supporting Information Figure S3A). 5Under our optimal conditions, PFL-AE was capable of charging 0.5−1.0G• per PFL dimer of the wt PFL substrate, characteristic of the "halfof-sites" reconstitution and reactivity of the E. coli PFL. 28The C 418 S and C 419 S PFL variants exhibited similar G• reconstitution efficiency, based on G• EPR signal quantitation (Supporting Information Figure S3B and S3C).We simulated the G• spectrum with an isotropic g iso of 2.0037, consistent with prior studies, 51 and hyperfine coupling (HFC) to a single 1 H with an isotropic hyperfine coupling constant (A αCH ) of 40.5 MHz.Coupling to additional nuclei, suggested by prior experimental and theoretical studies, 28,52 did not impactfully improve the simulation to the experimental data, and were not included in experimental simulations.Transfer of aPFL from H 2 O-based to D 2 O-based buffer results in a narrowing of the doublet signal to an apparent singlet, due to the H/D exchange of the G• and the smaller gyromagnetic ratio of deuterium relative to hydrogen, which is facilitated by the ionizable C 419 sulfhydryl group (Supporting Information Figure S3D). 26The narrowing of the G• spectrum was thus used to confirm H/D exchange of aPFL in subsequent experiments.A table summarizing these, and all subsequent EPR spectral fitting parameters and associated fitting standard deviations are provided in the Supporting Information Table S1.
To evaluate the activity of our recombinantly expressed and activated PFL we employed a multienzyme-coupled spectrophotometric assay. 27When normalized to reconstituted active sites, estimated by EPR quantitation of G•, we observed a k cat of 225 ± 9 s −1 and K m of 0.78 ± 0.09 mM for pyruvate and 12 ± 2 μM for CoA at 20 °C, consistent with prior reports of E. coli PFL kinetic constants (Supporting Information Figure S4). 27The C 418 S and C 419 S variants exhibited activities below the detection limit of the coupled optical assay, which we estimate to be <0.01 s −1 .Methacrylate is a mechanism-based suicide inhibitor of aPFL with a reported K I of 0.4 mM and k inact of 0.14 min −1 . 34Indeed, we observe time-dependent inhibition of 50 μM aPFL upon exposure to active sitesaturating (200 mM) concentrations of methacrylate, but with an apparent k inact of 1 min −1 (Figure 1).
The Methacryl Radical Intermediate.Based on our ability to temporally resolve the inhibition process, we sought to trap a radical intermediate in the inhibition of aPFL.Freezequenching the inhibition reaction at 10 s revealed a convoluted multiline EPR spectrum (Figure 2A) composed of G• and a new organic radical species coupled to various spin-active nuclei.We simulated the spectrum using the previously extracted G• parameters, which were held fixed, and allowed the simulation to model a second radical of unknown g, as well as the number and strength of potential HFCs to this second radical.Satisfactory simulations included a second radical species comprising 30.9 ± 0.6% of the total EPR intensity, with g iso of 2.0033, an isotropic HFC to three identical 1 H nuclei of 57.9 MHz, and a fourth 1 H of 69.9 MHz (Figure 2B and Table S1).Inclusion of an additional 1 H did not statistically improve the simulation fit, as judged by the residuals and the HFC confidence intervals of the additional nuclei.The spectrum is consistent with a C2• methacryl radical, which has been reported previously, 53,54 but where the radical has minimal electron density overlap with the second methylene 1 H of C3.No change in the G• signal was observed at any time for C 418 S or C 419 S (Supporting Information Figure S5).
To determine the origin of the hyperfine interactions to the radical intermediate, we repeated the inhibition freeze quenching with aPFL exchanged into D 2 O buffer (Figure 2C).At 10 s the convoluted spectrum exhibits no change to the HFCs of the radical intermediate.The reaction of aPFL with perdeuterated methacrylate in H 2 O, on the other hand, dramatically narrowed the radical intermediate (Figure 2D).These results further support a C2• methacryl radical intermediate assignment.
While our EPR data support a methacryl C2• intermediate, they do not inform on the site of reactivity with aPFL.We turned to peptide analysis of tryptic digestions of PFL by LC-MS/MS.The parental ion of the peptide containing C 418 and C 419 eluted at 12.9 min retention time, and the MS/MS fragmentation resolved both y 9 and y 10 ions corresponding to the C 419 (m/z = 990.5)and C 418 −C 419 (m/z = 1150.6)containing peptides respectively, allowing for the unambiguous determination of post-translational modifications at each residue (Supporting Information Figure S6).Following complete inhibition by methacrylate over 10 min, a new peptide is resolved at 14.8 min retention time with y 9 unchanged, but y 10 observed at an m/z of 1179.6, consistent with a methacrylate adduct to C 418 .No adduct is formed with unactivated PFL, confirming the radical nature of the inhibition mechanism, nor in G•-reconstituted C 418 S or C 419 S, suggesting both residues participate in the inactivation process.We also performed the peptide analysis on samples acid-quenched at 10 s, 30 s, and 1 min to terminate radical chemistry at the time of observation of the EPR signal, yielding the same y 10 ion with the methacrylate adduct with variable intensity (Supporting Information Figure S7).
The methacrylate product observed by LC-MS/MS and EPR is consistent with a tertiary C2• methacryl radical, yet the lack of a fifth hyperfine interaction suggests a specific poise of the radical in the protein active site.To investigate the conformational dependence of the HFC to a putative C2• we performed DFT calculations with an unrestricted hybrid B3LYP functional to simulate the EPR parameters of the radical.As a starting structure, we used a geometry optimized C2• bound to C 418 through C3 as a C 418 −C 419 dipeptide. 35The dihedral angle between the pro-(R) methylene H−C3−C2−C1 was systematically rotated about the C3−C2 bond in 10°i ncrements from −180°to +180°relative the initial angle of 38°, and the EPR g iso and HFCs were calculated (Supporting Information Figure S8A and Table S2).Two dihedral angle zones with no van der Waals clashes were consistent with the experimental HFCs, −35°and +145°, which show minimal singly occupied molecular orbital (SOMO) overlap between with the pro-(R) methylene 1 H (Supporting Information Figure S8B and S8C).The −35°rotamer yields g iso of 2.0037 with A CH3 of 56 MHz and A CH2 of 65 and 3 MHz, whereas the +145°rotamer yields g iso of 2.0029 with A CH3 of 61 MHz and A CH2 of 68 and 2.6 MHz, both in agreement with the experimental spectrum (Supporting Information Table S2).The van der Waals clashes in the other two rotamers may be alleviated by structural dynamics or relaxation, but were not investigated further.
Methacryl Radical Reactivity.The methacryl radical is unstable and decays with concomitant reformation of the G• over the course of minutes at 20 °C.We investigated the time course of C2• formation and decay during the methacrylate inactivation reaction by freeze-quenching reactions of aPFL from 10 s to 7 min (Figure 3) using EPR spectral simulations to quantify G• and C2• over time (Supporting Information Table S3).Using the EPR spectral parameters obtained from Figure 1 we fit the data to a global model to extract the rate constant for inhibition and C2• quenching (Supporting Information Figure S9).The extracted rates were 2.0 ± 0.2 min −1 for C2• formation and inhibition, and C2• decay at 1.7 ± 0.1 min −1 .No other radical species were observed, and the reaction proceeded with <30% total radical loss.S1.Insets show proposed structures consistent with experimental simulations.
To examine the mechanism of C2• reduction, we repeated the EPR analysis of the methacrylate inhibition process of aPFL in D 2 O buffer (Figure 4).We observed no solvent kinetic isotope effect (KIE) on the formation of the C2• radical (2.0 ± 0.3 min −1 ), but a solvent KIE was observed for the C2• decay and G• reformation (0.50 ± 0.04 min −1 ) of 3.4 ± 0.4, supporting a mechanism of radical reduction by H atom transfer from an ionizable amino acid.
The relatively slow reactivity of the methacryl C2• radical stands in stark juxtaposition to the native reactivity of aPFL, where no radical intermediates have been identified.This could be due to steric constraints of the covalently attached C2•, as opposed to the dissociated CO 2 • − in the native ping phase, or reactivity differences due to BDEs.−57 We performed kinetic studies to assess acrylate as a PFL suicide inhibitor by detecting PFL remaining activity at different inhibition times.Incubating aPFL with 200 mM acrylate results in activity loss with a higher apparent k inact value of >15 min −1 relative to that of methacrylate (Figure 5A).The reaction is essentially completed in less than 20 s, which we are unable to accurately characterize due to the nature of the coupled assay and liquid handling.Acrylate adds specifically, and in a radical dependent manner, to C 418 based on peptide mapping by tryptic digest LC-MS/MS (Supporting Information Figure S10).However, we were not able to  To gain insight into the relative energetics of the radical chemistry between acrylate and methacrylate, we again turned to DFT.We performed geometry optimizations and transition state searches in a truncated model from a prior study for the reaction with both acrylate and methacrylate and a thiyl radical at C 418 . 35The calculated overall energy landscape is shown in Figure 6

■ DISCUSSION
In the mechanistic study of thiyl radical enzymes, direct observation with structural resolution of radical intermediates associated with the catalytic mechanism is rarely achieved, with the notable exception of ribonucleotide reductase. 58For PFL, this has led to some debate about the role of the two active site cysteines during catalysis.Early studies on the mechanism of PFL supported a "thiohemiketal" mechanism of catalysis in which the ping phase was radical-based and the pong phase involved nucleophilic displacement. 25Peptide mapping of aPFL reacted with 14 C2-labeled pyruvate revealed radioactivity associated with C 419 , which was proposed as the site of radical reactivity and acetylation. 30The role of C 418 was suggested by the demonstration that reconstituted C 419 S PFL was charged with an acetyl unit at C 418 from acetyl-CoA. 25 The ping and pong phases appeared to be connected by an acyl shift, which was evidenced by reactions of the (C 419 -)acetylated aPFL with hypophosphite, a formate analog, forming a C 418 -ligated acetylphosphonate. 30Conversely, the X-ray crystallographic structure of the unactivated PFL bound to pyruvate 23 and peptide mapping studies with 14 C2-labeled methacrylate, 34 supported C 418 as the site of radical-based substrate activation and acetylation (Scheme 1).This latter mechanism is consistent with the observation of radical equilibration  between C 419 and G 734 in the absence of catalysis in C 418 S PFL, 26 and is widely accepted. 9,10Neither the thiohemiketal mechanism nor the mechanism described in Scheme 1 are supported by direct radical observation with structural resolution, preventing discernment between the two mechanisms.
By examining the inhibition of aPFL by methacrylate at significantly higher concentrations than those used previously, we have captured a radical intermediate by EPR spectroscopy that is kinetically consistent with inhibition.Based on the radical intermediate EPR signature when generated in H 2 O, D 2 O, and with d 5 -methacrylate, we assign this radical intermediate to a C2• tertiary radical of methacrylate, in agreement with previous proposals and theoretical studies. 34,35o connect the spectroscopic detection and assignment of the radical intermediate to structure, we characterized the sitespecific reactivity of methacrylate for C 418 by LC-MS/MS, both as an end-point and during the inactivation in acid quenched samples.The acid-quenched samples obviate the potential for acyl shifts that have been reported for PFL with other substrates or inhibitors, and confirm the site-specificity of methacrylate for C 418 .Both EPR and LC-MS/MS studies of the reaction of G•-reconstituted C 419 S showed no evidence of inhibitor radical formation or a methacrylate adduct at C 418 , respectively.These observations support the role of C 419 as a radical mediator between G 734 • and C 418 .We also gained further insight into the structure through DFT modeling, which supports two rotamers that are consistent with the simulated EPR spectral properties.These two configurations correspond to the orientation of C2• that would generate either 2-(R) or 2-(S) stereoisomer products, following reduction by H atom transfer.An inspection of the two DFT structures and the X-ray structure of the pyruvate-bound PFL (Supporting Information Figure S12), and the enantioselectivity of the reduction process, 34 strongly favors the −35°orientation, forming the 2-(S) enantiomer, where the vinyl group of methacrylate takes the position of the carbonyl in the pyruvate substrate.
The C2• radical quenching is informative regarding the second half of the ping phase of the reaction mechanism of PFL, where C2• serves as a surrogate for the CO 2 • − in the native reaction.We measured a solvent KIE of 3 for the C2• reduction in H 2 O versus in D 2 O, which confirms that the inhibitor radical is reduced by an ionizable residue capable of H atom transfer.The EPR spectrum of aPFL in D 2 O demonstrates that the C 419 sulfhydryl H/D exchanges with the solvent, and is the nearest redox-active residue to the substrate in the X-ray structure. 23The observed KIE is in agreement with the DFT calculated KIE of 3.2, consistent with the proposed mechanism of C2• reduction by C 419 .
Our observation of a long-lived C2• and DFT analysis is consistent with previous theoretical predictions of a rate determining C2• reduction. 35The accumulation of C2• suggests that the tertiary radical is more stable than the corresponding C 419 thiyl radical, but C2• reduction is ultimately driven forward by radical equilibration back to G 734 •, precluding the direct observation of a transient C 419 thiyl radical.Using acrylate as a methacrylate analog with a destabilized secondary C2•, we show that this radical is unstable, yet still specific for C 418 .Comparing the product and reactant states of the acrylate reaction by DFT, we estimate that the C 419 thiol S−H BDE is higher than a tertiary radical and nearly thermoneutral with a carbon secondary radical.
Estimates of the bond dissociation energy of the corresponding isobutanoate C2 range from 347 to 355 kJ/mol, 59,60 and the shift from a tertiary to secondary C2 increases the reactivity by +18 kJ/mol. 60Our DFT computed ΔG value for C2 reduction of the acryl radical is −8.8 kJ/mol placing the corresponding C 419 thiol between 356 and 364 kJ/mol, which is in agreement with prior experimental or theoretical studies on the free cysteine amino acid from 353 to 365 kJ/mol. 15,61,62These estimates neglect the role of the protein environment, which may significantly alter the radical reduction potentials of the inhibitor or the active site cysteines.−66 Many thiyl radical enzymes activate secondary carbon centers with high turnover frequencies (k cat > 100 s −1 ), including the class II ribonucleotide reductase, 67,68 PFL, 27 and 1,2-eliminases (e.g., choline trimethylamine-lyase, 69 propane 1,2-diol dehydrate, 70 glycerol dehydratase, 71 4-hydroxyproline dehydratase, 70 and isethionate sulfite lyase 72,73 ).Interestingly, GREs proposed to activate primary carbon centers, including X-succinate synthases (X = benzyl, 4-isopropylbenzyl, hydroxybenzyl, naphthyl-2-methyl, and 1-methylalkyl) 74,75 and C−P lyase 76 exhibit lower turnover rates (k cat < 1 s −1 ).Our results may provide a thermodynamic explanation for the kinetic privilege of the thiyl radical enzymes activating secondary, rather than primary carbon centers, although a primary carbon radical was not investigated in this study.This reactivity may also inform mechanistic investigation, where new tertiary centers may stabilize substrate analog radicals sufficiently long enough to be observed and characterized.

Scheme 1 .
Scheme 1. Proposed Reaction Mechanism of PFL with Substrate Pyruvate (top) and Inhibition by Methacrylate (bottom)

Figure 1 .
Figure 1.Inhibition of aPFL by methacrylate.aPFL was mixed with 200 mM methacrylate and incubated from 10 to 420 s at 25 °C and aliquots were taken at each time point and assayed for activity by the enzymatic coupled assay.The data were fit to a single exponential decay.Error bars represent the span of two technical replicate experiments.

Figure 2 .
Figure 2. Normalized X-band EPR spectra (black) and simulations (red) of radicals associated with PFL inactivation by methacrylate quenched at 10 s. (A) aPFL reacted with methacrylate in H 2 O. (B) Weighted simulation components assigned to G• (orange) and a methacrylate C2• (blue).(C) aPFL reacted with methacrylate in D 2 O. (D) aPFL reacted with d 5 -methacrylate in H 2 O. Experimental spectra were simulated by EasySpin and the simulation spectral parameters are reported in the Supporting Information TableS1.Insets show proposed structures consistent with experimental simulations.

Figure 3 .
Figure 3. X-band EPR spectra and relative radical content during the inhibition of aPFL by methacrylate.(A) EPR spectra (black) and associated simulations (red) during the aPFL inhibition time course from 0 to 420 s.Experimental EPR spectra were simulated in EasySpin as a convoluted spectrum of G 734 • and C2• of fixed spectral properties determined previously and varying their relative weights.(B) Spectral simulation weights for C2• (blue circles) and G• (orange squares) at each time point.Error bars represent the standard deviation of the fitted covarying weights of the fixed G• and C2• contributions.

Figure 4 .
Figure 4. X-band EPR spectra and relative radical content during the inhibition of aPFL by methacrylate.(A) EPR spectra (black) and associated simulations (red) during the aPFL inhibition time course from 0 to 800 s.Experimental EPR spectra were simulated in EasySpin as a convoluted spectrum of (D 2 O) G 734 • and C2• of fixed spectral properties determined previously and varying their relative weights.(B) Spectral simulation weights for C2• (blue circles) and G• (orange squares) at each time point.Error bars represent the standard deviation of the fitted covarying weights of the G 734 • and C2• contributions.
. The geometry optimized step 1 (thiyl radical addition of C 418 −S• to C3) and step 2 (C2• reduction by C 419 −SH) reactants, transition states, and products are shown in the Supporting Information Figure S11.The thiyl radical addition of C 418 −S• to C3 of either acrylate or methacrylate exhibited low barriers, with ΔG ‡ of 9.4 and 4.6 kJ/mol, respectively, but were both driven to the C2• product with an overall exergonic ΔG of −16.7 kJ/mol and −22.1 kJ/mol, respectively.A significant difference was observed between the inhibitors in step 2, where the predicted ΔG ‡ were +33.1 kJ/mol for acrylate and +39.6 kJ/mol for methacrylate.The difference in activation barriers between the two substrates predicts a difference in reactivity of the C2• of acrylate relative to methacrylate (k acryl /k methacryl ) of 14.For methacrylate, a KIE due to H/D exchange of the C 419 −SH(D) is calculated to be 3.2.The formation of the product of step 2, namely a C 419 −S•, is endergonic for methacrylate by +7.5 kJ/mol, whereas this reaction is exergonic by −8.8 kJ/mol for acrylate, suggesting the C 419 thiyl radical reactivity is between that the secondary C2• of acrylate and methacrylate.

Figure 5 .
Figure 5. Inhibition of aPFL by acrylate.(A) Remaining activity after inhibition of 5 μM aPFL with 200 mM methacrylate (blue diamonds) or acrylate (orange circles) from 0 to 600 s and fit to a single exponential function (black lines).(B) Normalized X-band EPR spectra reactions of aPFL with acrylate in H 2 O (top) or D 2 O (bottom).Reactions were incubated at room temperature and freeze quenched after 10 s.EPR conditions were: microwave frequency 9.30 GHz, modulation amplitude 2G, at 100 K, 30 scans were averaged for each sample.