Edited by Harry B. Gray, California Institute of Technology, Pasadena, CA, and approved October 13, 2015 (received for review August 25, 2015)
The mononuclear Mo enzymes are ubiquitous throughout life, and the notion that their activity arises from Mo(VI/V/IV) redox cycling is a central dogma of bioinorganic chemistry. We prove that YedY, a structurally simple mononuclear Mo enzyme, operates via a strikingly different mechanism: the catalytically active state is generated from addition of three electrons and three protons to the Mo(V) form of the enzyme, suggesting for the first time (to our knowledge) that organic-ligand–based electron transfer reactions at the pyranopterin play a role in catalysis. We showcase Fourier-transformed alternating-current voltammetry as a technique with powerful utility in metalloenzyme studies, allowing the simultaneous measurement of redox catalysis and the underlying electron transfer reactions.
A long-standing contradiction in the field of mononuclear Mo enzyme research is that small-molecule chemistry on active-site mimic compounds predicts ligand participation in the electron transfer reactions, but biochemical measurements only suggest metal-centered catalytic electron transfer. With the simultaneous measurement of substrate turnover and reversible electron transfer that is provided by Fourier-transformed alternating-current voltammetry, we show that Escherichia coli YedY is a mononuclear Mo enzyme that reconciles this conflict. In YedY, addition of three protons and three electrons to the well-characterized “as-isolated” Mo(V) oxidation state is needed to initiate the catalytic reduction of either dimethyl sulfoxide or trimethylamine N-oxide. Based on comparison with earlier studies and our UV-vis redox titration data, we assign the reversible one-proton and one-electron reduction process centered around +174 mV vs. standard hydrogen electrode at pH 7 to a Mo(V)-to-Mo(IV) conversion but ascribe the two-proton and two-electron transition occurring at negative potential to the organic pyranopterin ligand system. We predict that a dihydro-to-tetrahydro transition is needed to generate the catalytically active state of the enzyme. This is a previously unidentified mechanism, suggested by the structural simplicity of YedY, a protein in which Mo is the only metal site.
Most living species require a Mo enzyme (1), and apart from nitrogenase, all of these Mo-containing proteins are part of the large family of “mononuclear Mo” enzymes. The general ability of mononuclear Mo enzymes to catalyze two-electron oxygen atom transfer reactions has been attributed to the Mo(IV)/Mo(V)/Mo(VI) oxidation state cycling of the active site, and this mechanism is a common part of undergraduate syllabuses (1, 2). Escherichia coli YedY is a mononuclear Mo enzyme (3), and based on sequence homology, the majority of sequenced Gram-negative bacterial genomes encode a YedY-like protein (3⇓–5). Uniquely for a mononuclear Mo enzyme, it has not been possible to form the YedY Mo(VI) state in experiments using ferricyanide as an oxidizing agent, and an unusually positive reduction potential for the Mo(V/IV) transition [+132 mV vs. standard hydrogen electrode (SHE) at pH 7] was determined from EPR experiments (6). Although the physiological substrate of YedY is unknown, a possible role in the reduction of reactive nitrogen species is suggested by experiments on the pathogen Campylobacter jejuni, where deletion of the Cj0379 YedY homolog generated a mutant that is deficient in chicken colonization and has a nitrosative stress phenotype (4). YedY catalysis can be assayed by measuring the two-electron reduction of either dimethyl sulfoxide (DMSO) or trimethylamine N-oxide (TMAO) (3), but the inaccessibility of the YedY Mo(VI) means the enzyme mechanism does not proceed via the common two-electron Mo-redox cycle. In small-molecule analogs of mononuclear Mo enzymes, the pterin ligands are described as “noninnocent,” meaning that the redox processes could be ligand or metal based (7). This study explores the possibility that ligand-based redox chemistry plays a role in YedY catalysis.
YedY has been structurally characterized via both X-ray crystallography and X-ray absorption spectroscopy (XAS) (3, 8, 9). In most mononuclear Mo enzymes, heme groups and iron sulfur clusters are found within the same protein as the Mo center, but the only metal site in YedY is Mo, making this enzyme a helpfully simple system for studying redox chemistry (Fig. 1) (1, 3). Within the active site, the X-ray structure was interpreted to show Mo(V) in a square pyramidal environment (3), identical to other members of the “sulfite oxidase” family of mononuclear Mo enzymes. In contrast, XAS has suggested a pseudooctahedral Mo center (8, 9) with an additional O (from Glu104) or N (from Asn45) axial ligand coordinating trans to the apical oxo group (Fig. 1) (8). The equatorial ligation is provided by one oxygen-containing ligand and three sulfur donor atoms, one provided by cysteine (Cys102) and two from the pyranopterin cofactor, which binds to the Mo in a bidentate fashion via the enedithiolate side chain (3).
Structure of Escherichia coli YedY. (A and B) The protein structure, PDB ID code 1XDQ (3). (C) The active site in the as-isolated Mo(V) state containing the dihydro form of pyranopterin.
A 2012 computational study provided evidence that two different oxidation states can be accessed by protein-bound pyranopterin ligands (10). Conformational analysis and electronic structure calculations were used to assign redox states to the pyranopterin ligands in all known mononuclear Mo enzyme structures (10). It was concluded that, although enzymes from the sulfite oxidase family (such as YedY) contain pyranopterin ligands in the “dihydro” form, the xanthine dehydrogenase family of enzymes contain the two-proton, two-electron more reduced “tetrahydro” form of the pyranopterin (10).
Traditionally, redox potential measurements of enzymes have required substrate-free conditions to either permit a solution equilibrium to be established (spectroscopic redox titrations) or to prevent catalytic signals from masking the noncatalytic response (film electrochemistry). Fourier-transformed alternating-current voltammetry (FTacV) is a technique that offers the ability to measure catalytic chemical redox reactions and reversible electron transfer processes in a single experiment (11, 12). In the FTacV measurement, a large-amplitude sine wave of frequency f is superimposed on a linear voltage–time sweep (11, 13⇓–15) and the resulting current–time response is measured and then Fourier transformed (FT) into the frequency domain to give a power spectrum of harmonic contributions at frequencies f, 2f, 3f, etc. Band selection of the individual harmonics followed by inverse FT resolves the data back into the time domain. The higher harmonic components only arise from fast, reversible redox reactions, devoid of catalysis and baseline contributions, but the aperiodic (dc) component (f = 0) gives the same catalytic information as a traditional dc cyclic voltammetry (dcV) experiment, and can therefore show catalytic turnover (13, 14, 16).
In this study, we both discover previously unidentified mononuclear Mo enzyme redox chemistry as well as demonstrate the significant advantages of using FTacV to probe the mechanism of a redox active enzyme.
We prove that YedY can reversibly form three different oxidation states, i.e., the enzyme can undergo two different redox transitions. This is shown in Fig. 2A, which contains dcV YedY electrochemical data measured under conditions of pH 7 and 25 °C. The enzyme has been adsorbed onto the surface of the electrode, and the signals that are observed are typical for noncatalytic redox-enzyme “film” electrochemistry (SI Appendix, Fig. S1) (17). In Fig. 2A, both enzyme-redox transitions are visible as “peak” Faradaic signals at around +170 and −250 mV (control experiments confirm that these signals are not present with a YedY-free electrode). The YedY signals were stable over at least 20 continuous 100 mV⋅s−1 cyclic voltammograms, but only one scan is shown for clarity.
Redox transitions of YedY measured by dcV at a scan rate of 100 mV⋅s−1. (A) A cyclic voltammetry measurement of YedY in 50 mM Mes buffer, pH 7, is shown by the black solid line. The baseline is shown by the red dashed line; the baseline-subtracted signal (scaled by a multiplication factor of 20) is depicted by the gray solid lines. (B) pH dependence of peak midpoint potentials derived from voltammetry experiments conducted in 50 mM buffer solution of either acetate (pH 4 and 5), Mes (pH 6 and 7), or Tris (pH 8 and 9). Error bars shown within data point circles reflect the SE calculated from at least three repeat experiments. Other conditions were as follows: stationary electrode, and temperature, 25 °C.
A significant limitation of the protein film dcV technique is that the Faradaic noncatalytic enzyme signals are very small in relation to the “background” signal from the non-Faradaic (double-layer charging) electrode process. To permit analysis of the YedY-only redox chemistry, we have computed the non-Faradaic response using a polynomial function and then subtracted this from the experimental data to give pure Faradaic data. The baseline-subtracted signals thus obtained are scaled by a 20-fold multiplication factor in Fig. 2A. From Fig. 2A, the integrated area of the baseline-subtracted negative potential process, centered around −250 mV, is ∼1.8 times larger than the integrated area of the baseline-subtracted positive potential process centered around +170 mV. We therefore conclude that almost twice as many electrons are passed in the negative potential redox transition relative to the positive potential redox transition.
As shown in SI Appendix, Fig. S2B, the redox transition measured for YedY at positive potential is well modeled by a Nernstian one-electron process (equivalent to a peak width at half-height, δ, of 90 mV) with a pH 7 midpoint potential Em, 7 = +174 ± 4 mV. We attribute this process to the Mo(V/IV) redox transition. Our assignment of the positive potential redox process as a metal-based transition is supported by UV-vis solution spectroelectrochemistry measurements made from 750 to 320 nm under an atmosphere of Ar and shown in SI Appendix, Fig. S3. As-purified YedY, known to be in the Mo(V) state, exhibits two peaks in the optical spectrum (6, 18): a broad absorbance centered at 503 nm and another at ∼360 nm. Upon lowering the potential from +0.21 to −0.09 V, i.e., passing through the positive potential redox transition, both spectral signals are bleached, indicating a metal-based reduction. These spectral changes can be reversed by raising the potential back to +0.21 V. No spectral changes accompany the negative potential redox reaction, i.e., no UV-vis changes are measured when the solution potential is stepped between −0.44 and −0.09 V.
Using film electrochemistry, there is no evidence of any further redox transitions at more positive potentials, even when the potential range is extended to the solvent/electrode limit (SI Appendix, Fig. S4), so in agreement with other techniques, we also cannot observe a Mo(V/VI) redox transition.
Analysis of the negative potential baseline-subtracted dcV wave shape shown in SI Appendix, Fig. S2C, suggests a cooperative, nonsimultaneous two-electron charge transfer process, i.e., one electron is transferred and then a second electron follows onto the same center (19⇓⇓–22). We measure Em, 7 = −248 ± 1 mV, but using dcV it is very difficult to derive more precise mechanistic information regarding the separate one-electron processes that combine to give the “envelope” signal.
Experiments at different pH values reveal that the Em values for the Mo(V/IV)-assigned reaction and the negative potential redox processes change by −53 and −55 mV per pH unit, respectively (Fig. 2B), close to the −59 mV per pH unit expected for a one proton per electron process at 25 °C (23). “Trumpet plots” of the reductive and oxidative peak potentials vs. scan rate show greater peak separation at lower scan rates for the Mo(V/IV) signals compared with the negative potential transition, suggesting that the Mo-based redox process has a slower electron transfer rate (SI Appendix, Fig. S5) (24). Combining all of the dcV information, we suggest a 1H+ + 1e− process for the Mo(V/IV) transition and a faster 2H+ + 2e− process for the negative potential transition, over the pH range measured.
Analogous to the dcV experiment shown in Fig. 2, FTacV was used to interrogate YedY redox chemistry over a wide potential range and this is shown in SI Appendix, Fig. S6. There are two significant differences between the dc and ac results; first, relative to dcV, the signal-to-background response of YedY at around −250 mV is much larger in the higher harmonic components of the FTacV measurements; second, whereas in dcV the peak area for the negative potential YedY signal is approximately double the peak area of the Mo(V/IV) signal, in FTacV the Mo(V/IV)-assigned signal is not visible above the noise. The fact that very little Mo(V/IV) signal is observed means that the 9-Hz frequency applied in the FTacV outpaces Mo-based electron transfer processes, and we can therefore state that YedY’s negative potential electron transfer processes is much faster than the Mo(V/IV) redox transition. Based on analysis of the FTacV signals, we define the Mo(V/IV) process as a quasireversible electron transfer reaction with an apparent heterogeneous charge transfer rate
To learn more about the negative potential process, Fig. 3A shows 9-Hz FTacV measurements focused on the YedY reversible redox transition centered at around −250 mV. At least 12 harmonic components are detected, and, in stark contrast to dcV, no baseline subtraction is required before analysis of the higher harmonic signals because YedY-free controls confirm there is negligible baseline contribution from a bare electrode. The potentials of the central maxima of the odd harmonics and the central minima of the even harmonics provide a direct measure of the midpoint potential, and these values agree with the baseline-subtracted dcV Em data. Similar to changing the scan rate in dcV, changing the frequency in FTacV provides a qualitative means of assessing the electron transfer rate. As shown in Fig. 3B, when the same negative potential range is interrogated using a range of frequencies, well-defined ac harmonics are observed up to f = 219 Hz, indicating that the low-potential YedY redox reaction involves extremely fast electron transfer.
FTacV measurements of the YedY two-electron redox transition. (A) Gray solid lines show, in descending order, the 7th, 8th, 9th, and 10th ac harmonic signals measured for YedY using FTacV with a frequency of 9 Hz. The response from a bare (YedY-free) graphite electrode under the same conditions is shown by light red dotted line. (B) Gray solid lines compare the eighth ac harmonic from YedY FTacV experiments at different frequencies, as denoted in the graph. (A and B) Black dashed lines depict simulated data for a 1e− + 1e− mechanism with parameters (E01)app = −239 mV, (E02)app = −261 mV, (k01)app = (k02)app = 2·104 s−1, Ru = 50 Ω, polynomial capacitance, Γ = 1.15 pmol⋅cm−2 (9 Hz), Γapp = 0.86 pmol⋅cm−2 (39 Hz), and Γapp = 0.66 pmol⋅cm−2 (219 Hz). Other experimental conditions were as follows: scan rate, 15.83 mV⋅s−1; amplitude, 150 mV; buffer solution of 50 mM Mes and 2 M NaCl, pH 7, 25 °C.
The ability of FTacV to separately resolve catalytic and electron transfer steps in a single experiment is shown in Fig. 4. Solution assays have shown that YedY catalyses the reduction of N- or S-oxides with concomitant oxidation of reduced benzyl viologen (3). The enzyme has the largest specificity constant, kcat/KM, for the substrates DMSO and TMAO, where kcat refers to the catalytic turnover rate and KM denotes the Michaelis constant (3). When YedY is adsorbed onto a graphite electrode and then placed in a solution of DMSO, the negative catalytic reduction current, revealed by the aperiodic dc component of the data (Fig. 4B), steadily increases as the electrode potential is lowered below approximately −0.3 V at pH 7 (“control” enzyme-free electrode experiments in the presence of DMSO show no reductive current; SI Appendix). In contrast, the aperiodic component resembles a “blank” electrode in the presence of YedY but absence of substrate (Fig. 4B). The reduction potential for DMSO is +160 mV at pH 7, and work by Heffron et al. (25) shows that the enzyme E. coli DMSO reductase is capable of reducing DMSO at more positive voltages than YedY. In Fig. 4B, the “onset potential,” i.e., the electrochemical voltage required to initiate YedY-catalyzed DMSO reduction, is therefore an enzyme-specific property and not a substrate-related behavior.
FTacV of YedY in the presence and absence of DMSO substrate. (A) Sixth harmonic component of a 219-Hz FTacV experiment on YedY in the absence (black solid line) and presence (gray dashed line) of 200 mM DMSO. The response of a blank or bare (YedY-free) graphite electrode in the absence of DMSO, measured using the same FTacV parameters, is shown by the gray solid line. (B) The aperiodic dc component of the same FTacV experiment on YedY in the absence (black solid line) and presence (gray solid line) of DMSO. Other conditions were as follows: scan rate, 15.83 mV⋅s−1; amplitude, 150 mV; buffer solution of 50 mM Mes and 2 M NaCl, pH 7, 25 °C; and stationary electrode.
The sixth harmonic signal for YedY, shown in Fig. 4A, is unchanged in the presence or absence of 200 mM DMSO, showing that substrate has not affected the electron transfer properties of the negative potential redox process: it remains very clearly distinguishable, revealing the redox potential without any need for background subtraction. Comparison between the high harmonic data in the presence and absence of substrate therefore suggests that the negative potential two-electron, two-proton reduction process generates the catalytically active state of YedY, because catalysis does not commence until the potential is sufficiently negative for this reaction to have occurred. Experiments at different pH values further support the hypothesis that the most reduced state of the enzyme reacts with substrate because the onset potential of YedY-catalyzed DMSO reduction changes between pH 5 and 8 in exactly the same way as the potential of the two-electron noncatalytic redox signal (SI Appendix, Fig. S7). Data extracted from experiments at pH 7 and different DMSO concentrations also corroborate published solution assay measurements with KM = 35 ± 5 mM at pH 7, 25 °C, and kcat = 4.2 ± 0.9 s−1 at −359 mV, pH 7, 25 °C (SI Appendix, Fig. S8) (3, 8, 26). Electrocatalytic experiments have also been conducted using TMAO as a substrate; these are shown in SI Appendix, Fig. S9. With TMAO as a substrate, we again observe that the onset potential of YedY catalysis is approximately −0.3 V at pH 7, far more negative than the equilibrium redox potential for the substrate [TMAO has reduction potential +130 mV at pH 7 (27)].
Simulation of the FTacV data makes it possible to harness the technique’s ability to provide a quantitative measure of electron transfer rates and deconvolution of separate redox potentials in a single experiment, an insight that we cannot access with dcV (22). Fig. 3A shows simulations of the FTacV data using the 1e− + 1e− mechanism described in SI Appendix. The same parameters were used to simulate all of the harmonic signals. To minimize the parameter space used in simulations, the value for uncompensated resistance, Ru, was derived from a separate impedance spectroscopy measurement, and Γapp, the apparent coverage of enzyme on the electrode, was estimated from dcV measurements. Both electron transfer steps exhibit fast kinetics as reflected by
When simulating the FTacV data obtained from experiments at different frequencies (Fig. 3B), all of the same parameters were used except for the enzyme-electrode coverage value, Γapp, which was lowered with increasing frequency from 1.15 pmol⋅cm−2 in the first 9-Hz measurement, to 0.615 pmol⋅cm−2 for f = 519 Hz (Fig. 3B and SI Appendix, Fig. S12). This trend did not reflect true enzyme desorption because a final measurement at 9 Hz yielded data that was best simulated using Γapp of 0.9 pmol⋅cm−2. As described in a recent theoretical study (28), kinetic dispersion, meaning that different enzyme orientations on the electrode surface have different electron transfer rates (k0app), is believed to be the major reason that Γapp decreases as the frequency increases. Dispersion is a common observation in protein film electrochemistry measurements (29).
We present dcV and FTacV data that prove that E. coli YedY forms three stable oxidation states. Relative to the well-characterized Mo(V) form of the enzyme, formation of the catalytically active state requires addition of three electrons and three protons and we summarize our proposed mechanism in Fig. 5. We assign the YedY redox transition that has Em, 7 = +174 ± 4 mV to the Mo(V/IV) process and ascribe the redox transition with
Proposed redox state cycling of YedY.
Using spectroelectrochemistry, it has been demonstrated that application of −0.09 V is sufficient to bleach the absorbance peaks observed in UV-vis spectra of the as-isolated, Mo(V), enzyme. The disappearance of the absorbance centered at 503 nm is consistent with our assignment of the positive potential redox process being Mo-based; dithiolene-S ligand to metal charge transfer processes give rise to this spectral feature so it should be a reporter signal for changes to the metal redox state (18). Both dcV trumpet plot data and low-intensity FTacV harmonic currents indicate that the Mo(V/IV) redox reaction has a slow electron transfer rate, from 3 to 6 s−1, which suggests structural reorganization. This correlates with the XAS mechanism that six-coordinate Mo(V) is reduced to a five-coordinate Mo(IV) species (9).
Varying the pH from 4 to 9 causes the electrochemically determined Em(Mo(V/IV)) value to decrease by 53 mV per pH unit, which indicates a one-electron, one-proton transition in agreement with the proposed XAS mechanism: Mo(V)−OH + 1H+ + 1e− → Mo(IV)−OH2 (9). There is a discrepancy between the midpoint potentials we measure using electrochemistry and those reported from an EPR redox titration, with respective Em, 7 values of +174 and +132 mV (6). The EPR data are also pH independent over a range of pH 6–8 (6). Whereas the dcV electrochemical data could be accurately simulated as a one-electron Nernstian process, the EPR Nernst plots were fit as physically impossible 1.3 and 1.63 electron processes for the oxidative and reductive titrations, respectively. As noted in the EPR study (6), the complex spectroscopic data are difficult to interpret and we suggest that the disparity in midpoint potential values may reflect this challenge. We also note that the electrochemical and EPR redox potential measurements are made on very different timescales as protein film electrochemistry affords the advantage of “wiring” the enzyme to the electrode, permitting rapid potential control, whereas achieving solution redox potential equilibration for EPR requires many minutes.
Our experiments at highly oxidizing potentials confirm the unusual stability of the YedY Mo(V) state with respect to oxidation, setting a limiting value of Em, 7(Mo(V/VI)) > +600 mV. The Mo(VI) oxidation state is therefore defined as physiologically irrelevant and thus plays no direct role in a catalytic reaction mechanism.
FTacV permitted simultaneous measurement of the putative pyranopterin redox transition and catalysis. The onset potential for enzymatic reduction of either DMSO or TMAO is more negative than the redox potential of either substrate (27) and instead correlates with the pyranopterin-assigned two-electron, two-proton reversible redox transition across the pH range 5–8. We assign this process to the pyranopterin cofactor because the structural simplicity of YedY is such that there are no other putative redox active centers apart from the Mo (3). In sulfite oxidase fold enzymes such as YedY, which is crystallized in the Mo(V) oxidation state, the geometry of the pyranopterin is consistent with a 10,10a-dihydro form (10); in Fig. 5, we show the three-electron, three-proton reduced catalytically active Mo(IV) form of the enzyme with a tetrahydro pyranopterin ligand. We have chosen to display the two-electron reduced pyranopterin ligand in a ring-closed tetrahydro state because this is consistent with the structure found most frequently at the active site of Mo-containing enzymes (10). However, it should be noted that an alternative, ring-opened confirmation exists at the same oxidation state level (7). For simple pterins, reversible two-electron and two-proton transitions between tetrahydro and dihydro forms are well known, as is further oxidation of the dihydro state, so pyranopterin redox reactions would be expected on the basis of chemical analogs (7, 30). It has been proposed that a nitrate reductase undergoes reversible enzymatic inactivation under oxidizing conditions because the pyranopterin converts from the tetrahydro to the dihydro state; however, all of the substrate reactions were still thought to be solely metal based (31). Our experiments therefore provide the first evidence (to our knowledge) of catalytically relevant pyranopterin redox chemistry (Fig. 5).
It is not possible to conclude whether the two electrons for substrate reduction are supplied directly by the pyranopterin or whether reduction of the pyranopterin ligand activates the Mo in such a way as to promote changes to the metal redox state catalysis. In the reduced tetrahydro state, the dithiolene chelate of a pyranopterin has increased electron-donating ability to Mo, which will decrease the Mo reduction potentials, because relative to the oxidized dihydro form π-delocalization is lost between the dithiolene chelate and the pterin ring (10). This could make the Mo(VI) state indirectly accessible. There is no conclusive structural information about how the substrate coordinates to YedY; XAS experiments on the Mo(V) state at pH 8 showed a possible long-range coordination of TMAO to Mo, but DMSO coordination was undetectable (8). To probe substrate binding, future experiments would need to be conducted under reducing conditions.
Comparison of Fig. 2 and Fig. 3 demonstrates how the complete absence of background current in the higher harmonic FTacV YedY signals results in much better defined noncatalytic redox peaks, overcoming the need for the significant baseline subtraction that is required in analysis of dcV data. Simulation of the higher harmonic YedY FTacV responses has also provided detailed information on the thermodynamics (apparent E0), kinetics (apparent k0 at E0app), and mechanism of the pyranopterin-assigned electron transfer. Simulation of the FTacV data suggests that the pyranopterin redox reaction proceeds via a sequential 1e− + 1e− process and is extremely fast [(k01)app, (k02)app ≥ 2.0⋅104 s−1]. Previous measurements of biological electron transfer rates using FTacV have been predominantly on metal centers and much slower electron transfer rates have been calculated (13, 16, 22, 32⇓–34), supporting our conclusion that the delocalized organic ligand is the site for oxidation state changes.
Escherichia coli YedY was prepared as described previously (3, 6) and stored in a buffer solution of 20 mM 3-(N-morpholino)propanesulfonic acid, pH 7, which was also used for the UV-vis spectroelectrochemical experiments. For film electrochemistry experiments, a protein concentration of ∼10 mg⋅mL−1 was used, and for spectroelectrochemistry, the protein concentration was 5.3 mg⋅mL−1.
All film electrochemistry experiment solutions were prepared using deionized water from a Pur1te Select water purification system (7.4 MΩ⋅cm). The buffer salts were as follows: pH 4–5, 50 mM acetate; pH 6–7, 50 mM 2-(N-morpholino)ethanesulfonic acid (Mes); or pH 8–9, 50 mM Tris(hydroxymethyl)aminomethane (Tris). The pH was adjusted by addition of NaOH or HCl for Mes and Tris buffers, and acetic acid for acetate buffer. Additional supporting electrolyte of NaCl was used where stated. DMSO (Fisher) or TMAO (Sigma) were used as enzyme substrates. All solids were of at least 99% (wt/wt) purity.
Direct-current voltammograms were measured with an Ivium CompactStat potentiostat, and FTacV measurements were performed using custom-made instrumentation described elsewhere (11, 14). All electrochemical experiments were performed under a N2 atmosphere in a glove box (manufactured by University of York Chemistry Mechanical and Electronic Workshops). A conventional three-electrode setup was used with the pyrolytic graphite edge working electrode (0.03-cm2 geometric area, made in-house), Pt wire counter electrode (Advent Research Materials), and saturated calomel reference electrode (Scientific Laboratory Supplies), all located in an all-glass, water-jacketed electrochemical cell (manufactured by University of York Chemistry Glassblower). The connection to the working electrode was made via an OrigaTrod electrode rotator, and the electrode was rotated during dcV catalysis experiments to ensure that substrate and product mass transport did not limit the enzyme activity. The electrode was not rotated for FTacV catalysis experiments. All experiments were performed at 25 °C, and all potentials have been converted to the SHE scale using the correction +241 mV vs Hg|Hg2Cl2|KCl(sat.) at 25 °C (23). A clean working electrode surface was obtained by abrading the pyrolytic graphite edge surface with P1200 sandpaper and then rinsing with deionized water. Enzyme was adsorbed onto the electrode surface by pipetting on 4 µL of YedY solution and allowing this to dry for ∼10 min. All values quoted are the average of at least three experiments, and the error bars are the standard errors (SE) calculated from all repeat data.
Simulations of FTacV data were based on a Butler−Volmer formalism for heterogeneous electron transfer kinetics (22) and the mechanism described in SI Appendix, and were performed using the Monash Electrochemistry Simulator (MECSim) digital simulation software package (35). The charge transfer coefficient was assumed to be α = 0.5 in all simulations, and all other simulation parameters were optimized to give a close fit between theoretical and experimental data using a heuristic approach.
We gratefully acknowledge Shannon Murphy’s technical assistance in protein purification and thank Prof. Paul Walton and Prof. Anne-Kathrin Duhme-Klair (both University of York) for useful scientific discussions. A.P. and H.A. thank the University of York for funding and Biotechnology and Biological Sciences Research Council (BB/F017316/1). H.A., A.N.S., A.M.B., and A.P. also acknowledge the Royal Society for funding this work via the International Exchange Scheme. Research in the J.H.W. lab was funded by the Canadian Institutes of Health Research Grant MOP15292.
Author contributions: H.A. and A.P. designed research; H.A. performed research; M.K., R.A.R., J.H.W., and A.M.B. contributed new reagents/analytic tools; H.A. and A.N.S. analyzed data; and H.A., A.N.S., R.A.R., J.H.W., A.M.B., and A.P. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The graphical data reported in this paper have been deposited with the University of York Library (www.york.ac.uk/library/info-for/researchers/datasets/).
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1516869112/-/DCSupplemental.
