Journal of Inorganic Biochemistry 157 (2016) 8–14
Contents lists available at ScienceDirect
Journal of Inorganic Biochemistry journal homepage: www.elsevier.com/locate/jinorgbio
Catalytic reduction of dioxygen with modified Thermus thermophilus cytochrome c552 Jonathan Husband 1, Michael S. Aaron 1, Rajneesh K. Bains, Andrew R. Lewis, Jeffrey J. Warren ⁎ Department of Chemistry, Simon Fraser University, 8888 University Drive, Burnaby, BC V5A 1S6, Canada
a r t i c l e
i n f o
Article history: Received 2 November 2015 Received in revised form 23 December 2015 Accepted 18 January 2016 Available online 20 January 2016 Keywords: Heme proteins Oxygen reduction Electrochemistry Proton relay
a b s t r a c t Efficient catalysis of the oxygen reduction reaction (ORR) is of central importance to function in fuel cells. Metalloproteins, such as laccase (Cu) or cytochrome c oxidase (Cu/Fe–heme) carry out the 4H+/4e− reduction quite efficiently, but using large, complex protein frameworks. Smaller heme proteins also can carry out ORR, but less efficiently. To gain greater insight into features that promote efficient ORR, we expressed, characterized, and investigated the electrochemical behavior of six new mutants of cytochrome c552 from Thermus thermophilus: V49S/M69A, V49T/M69A, L29D/V49S/M69A, P27A/P28A/L29D/V49S/M69A, and P27A/P28A/ L29D/V49T/M69A. Mutation to V49 causes only minor shifts to FeIII/II reduction potentials (E°′), but introduction of Ser provides a hydrogen bond donor that slightly enhances oxygen reduction activity. Mutation of L29 to D induces small shifts in heme optical spectra, but not to E°′ (within experimental error). Replacement of P27 and P28 with A in both positions induces a −50 mV shift in E°′, again with small changes to the optical spectra. Both the optical spectra and reduction potentials have signatures consistent with peroxidase enzymes. The V49S and V49T mutations have the largest impact of ORR catalysis, suggesting that increased electron density at the Fe site does not improve O2 reduction chemistry. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Heme proteins are among Nature's most diverse tools, catalyzing a vide variety of redox transformations, from oxygenation reactions in cytochromes P450 or nitric oxide synthase [1], to the reduction of dioxygen to water in cytochrome c oxidase [2] (the oxygen reduction reaction, ORR). The latter reaction is especially important in the search for an ORR catalyst that functions with minimal loss in free energy (overpotential) in fuel cell cathodes. The expense and rarity of currently employed Pt/C fuel cell catalysts spurred interest in using porphyrin complexes and heme proteins as O2 reduction catalysts [3]. The ORR is inherently a proton-coupled electron transfer (PCET) reaction [4–6], as it requires the addition of 4H+ and 4e− to O2, and scission of the O2 double bond, to produce 2H2O. Nature efficiently carries out this reaction in cytochrome c oxidase [7] and laccase enzymes [8], which incorporate multiple redox cofactors and proton-shuttling channels (proton relays). Proton transfer is critical for function, as demonstrated by the marked improvements in electrochemical ORR by ironand cobalt-porphyrin catalysts [9–11]. In protein catalysts derived from the myoglobin active site, introduction of a redox-active Cu, a cross linked tyrosine and histidine, and proton relays results in ORR catalysis; [12,13] these modified-myoglobins are active site models for ⁎ Corresponding author. E-mail address: [email protected] (J.J. Warren). 1 These authors contributed equally to this work.
http://dx.doi.org/10.1016/j.jinorgbio.2016.01.023 0162-0134/© 2016 Elsevier Inc. All rights reserved.
cytochrome c oxidase. Other myoglobins, that incorporate multiple attached to different surface histidine redox units (three Ru(NH3)2+ 5 residues), also are ORR catalysts with ascorbate or durohydroquinone as sacrificial reductants [14,15]. Engineered protein catalysts are typically not very robust (low turnover number) in part because of cofactor loss/degradation or low thermal stability. In contrast to the above heme proteins, c-type cytochromes feature a covalently anchored cofactor, which can eliminate cofactor (heme) loss as a mechanism of catalyst degradation. Moreover, these proteins exhibit facile redox chemistry and tolerate many mutations. These features have motivated their use as biocatalysts [16]. In this work, we investigate a series of new variants of thermostable cytochrome c552 (cyt c552) from Thermus thermophilus that contain proton relays and can electrochemically reduce O2. c-Type cytochromes are typically thought of as biological electron carriers [17]. Native proteins are covalently tethered to the heme vinyl groups via cysteine in a conserved CXXCH motif (where X is any amino acid). X-ray structures of yeast cyt c [18] and T. thermophilus cyt c552 [19] are shown in Fig. 1. The histidine residue at the end of the CXXCH sequence binds at the proximal position (H18 in cyt c and H15 in cyt c552, Fig. 1), and a methionine residue binds at the distal position (M80 in cyt c and M69 in cyt c552), coordinatively saturating the heme– iron. This largely precludes adventitious reactivity with substrates, such as O2. Replacement of M80 with A in yeast cyt c results in a 5-coordinate heme site that reversibly binds O2 [20]. The formal FeIII/II reduction potential (E°′) shifts to lower values and the 5-coordinate protein can
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
9
Fig. 1. Comparison of yeast cyt c active sites (left, CN−-bound structure of M80A, PDB ID 1FHB) and Thermus thermophilus cyt c552 (right, PDB ID 1DT1). The sites of distal (V49) and proximal (P26, P28, and P29) are shown for cyt c552. The distal A80 is shown for cyt c, but the analogous M69 is omitted for clarity in the structure of cyt c 552.
electrochemically reduce O2 both in solution [21] and when covalently attached to thiol-modified Au electrodes [22]. At pH 7, current onset is observed at ~0 V versus NHE. Removal of the analogous methionine residue (M69) in the more thermostable T. thermophilus cyt c552 also results in a properly folded protein [23]. Changes to the UV–visible properties in the mutant protein are in accord with those observed for M80A yeast cyt c. The M69A mutant also was demonstrated to act as a peroxidase when V49 was replaced with aspartate [24,25]. Peroxidase activity was improved by replacing an oxidizable Y with F, but peroxide-induced heme degradation was still observed over the course of several seconds [26]. Here, we take advantage of the thermal stability of M69A cyt c552 to investigate the solution chemistry of a cytochrome-based O2 reduction catalyst. We introduced serine at position 49 (V49S) to act as both a H-bonding partner and promote protonation of activated oxygen species (e.g., FeIII–O2), likely from H+ outside of the protein (mutations summarized in Fig. 2). The V49T mutant provides isostructural, polar modification. We also explored modifications to the proximal pocket that are intended to install a peroxidase-like Asp–His interaction [27] to promote binding and activation of O2. These mutations were L29D, to introduce Asp, and P27A/P28A to allow greater backbone flexibility near the proximal histidine. 2. Materials and methods All buffer salts were from J. T. Baker. Luria Broth and yeast extract was from Research Products International Corporation. K13CN was from Cambridge Isotope Laboratory. Chromatography media were from G.E. Healthcare. Water (18 MΩ cm− 1) was from a Barnstead EASYpure purification system. All gases were obtained from Praxair Canada. DNA polymerase (Q5) and DpnI were obtained from New
England Biolabs. PCR primers were from, and Eurofins Operon carried out DNA sequencing. Plasmids bearing the wild-type cytochrome c552 gene and the pec86 heme cassette were gifts from H. B. Gray (California Institute of Technology). Site-directed mutants were obtained using the Quickchange protocol (Agilent Technology). Expression and purification of all proteins were carried out according to the literature [28]. All electrochemical experiments were carried out using a Gamry Interface 1000 or a CH Instruments 600B potentiostat. Cyclic voltammetry and differential pulse voltammetry were used to collect reversible FeIII/II potentials, and cyclic voltammetry was used for experiments under air and O2. Solutions under N2, air, and O2 were sparged for 5 min before data collection. Formal potentials were referenced to an AgCl/Ag electrode calibrated versus potassium ferricyanide, and are reported with respect to the normal hydrogen electrode (NHE). Platinum wire was used as a counter electrode. Basal plane graphite electrodes were prepared according to a literature procedure [29,30]. Pyrolytic graphite was obtained from http://www.graphitestore.com. Loctite Hysol 9460 epoxy was obtained from McMaster-Carr. Electrodes were polished briefly using wet 6000 grit SiC paper before each run, washed with copious water and dried briefly using a heat gun. Optical spectra were collected on a Cary 100-Bio UV–visible spectrometer or a Photon Control SPM-002-EH CCD with a SPL-1DH deuterium/tungsten light source. Raman spectra were collected using a Renishaw inVia Raman microscope equipped with a 514 nm laser. Samples were deposited on glass as thin films to enhance signal to noise ratio; thin film spectra match solution measurements for all proteins. Infrared spectra were collected using a Perkin-Elmer Spectrum 2 Attenuated Total Reflectance (ATR) spectrometer, using 12 mM protein solutions + 20 mM K13CN. Data were smoothed using a moving average algorithm in MATLAB and the raw data are provided in the Supporting Information. Mass spectra (electrospray ionization positive ion) were
Fig. 2. Models of all cyt c552 variants. (A) Model of M69A cyt c552 variants with V49T mutations. (B) Model of M69A cyt c552 variants with V49S mutations, showing putative O2 binding. The distances were estimated by modeling the heme from oxy-cytochrome c peroxidase (PDB ID 1DCC) in the cyt c552 (PDB ID 1C52) heme site. (C) Model of L29D/M69A cyt c552 variants with V49T or V49S mutations.
10
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
collected on an Agilent 6210 TOF ESI-MS instrument. Circular dichroism spectra were collected using a Jasco J-810 spectropolarimeter and are reported as mean residue ellipticity. Single pulse 13C spectra without 1 H decoupling were collected on a Bruker 400 MHz AVANCE III NMR spectrometer operating at 100.6 MHz for 13C using a Bruker 5 mm BBFO probe. No 2H locking was employed and the spectral sweep for locking was turned off. Each sample 300 μL was added to a 5 mm outer diameter D2O-matched Shigemi tube and the magnetic field homogeneity was optimized manually by observing the 1H NMR signal of water and the sample shimming was checked using the 13C signal of the excess, free 13CN− ions. An 8 μs 90-degree radio frequency excitation pulse was used with a spectral width of 1000 ppm, 16,384 complex data points acquired (acquisition time 0.041 s), and a recycle delay of 0.025 s [31–33]. The 13C transmitter offset was set to − 3950 or −3600 ppm and the probe was carefully matched and tuned for each sample. Following 16 dummy scans, 400,000–800,000 scans were accumulated (24–48 h total acquisition time). 500 Hz exponential apodization was applied before Fourier transformation and spectral baselines were manually corrected using a multipoint spline function. The chemical shift scale was referenced using an external sample of K13CN in H2O (set to +168.6 ppm) run immediately following acquisition without any adjustment of the magnetic field, but with the probe re-tuned and matched using a transmitter offset of 100 ppm. 3. Results and discussion We expressed and purified the known [24] M69A variant of T. thermophilus cyt c552 and 6 new variants: V49S/M69A, V49T/M69A, L29D/V49S/M69A, P27A/P28A/L29D/V49S/M69A, and P27A/P28A/ L29D/V49T/M69A. The proteins were characterized using mass spectrometry, circular dichroism (CD) spectroscopy, and UV–visible spectrophotometry. Recombinant cyt c522 expressed in Escherichia coli is cleaved between Ala2 and Asp3; [19] our observed masses for all protein variants are consistent with this observation. Calculated and observed masses (including spectra) are given in the Supporting Information. The CD spectra of the ferric M69A, V49S/M69A, and V49T/M69A proteins are identical to each other, and to that of the wild-type protein [28,19]. Introduction of Asp at position 29 causes minor changes in the CD spectra between 225 and 250 nm; we conclude that the protein structures in all mutants are largely intact, with minor perturbations for the L29D-containing proteins. Optical spectra of the M69A, V49S/M69A, and V49T/M69A ferric proteins (data summarized in Table 1, Fig. 3) are consistent with a highspin, 5-coordinate heme, as for M80A cyt c [20]. Notably, the optical spectra of these proteins are not sensitive to pH between 3 and 8. Introduction of the L29D causes a small (3 nm) red shift in the Soret band, and shifts in the Q-bands that are reminiscent of optical signatures observed in heme peroxidases [34,35]. The optical spectra of these proteins are sensitive to pH, which could be due to protonation/ deprotonation of an iron-ligated H2O, as suggested for the analogous
Table 1 UV–vis features of cyt c552 proteins.
M69A FeIII (FeII–CN)
Soret
Visible
398 418
493 522
572 566
L29D/V49S/M69A and L29D/V49T/M69A 401 529 FeIII (pH 3) FeIII (pH 8) 406 531 417 523 (FeII–CN)
551
M69A, V49S/M69AAAT, and V49T/M69AAAS 405 528 FeIII (pH 3) 406 528 FeIII (pH 8) 417 522 (FeII–CN)
629 650 553
623
621
650
M80A cyt c [20–22]. Finally, introduction of Ala at positions 27 and 28, yields proteins with optical spectra similar to the L29D/V49T(S)/M69A proteins, but with a more pronounced band near 650 nm exhibits less pH dependence. The UV–vis features in this last set of proteins are similar to heme peroxidase proteins [36]. The ferric proteins are readily reduced using sodium dithionite and readily bind cyanide (Table 1). We also investigated the electronic structure of the hemes using resonance Raman spectroscopy (see Supporting Information). The spectra are different from WT cyt c552 and peroxidase enzymes [19,37] and exhibit only small changes between mutants. For the P27A/P28A/L29D/V49T/ M69A and P27A/P28A/L29D/V49S/M69A mutants there is a 1–3 cm−1 red shift in Raman bands, roughly consistent with observations for peroxidases [37] where the heme site is more electron rich. We also probed the electronic structure of the Met69Ala variant using 13C NMR and IR spectroscopies. The 13C chemical shift of cyanide in FeIII–CN hemes correlates with the electron density at the iron center [31–33]. In a work with structurally related (but smaller) Hydrogenbacter thermophilus (Ht) cytochrome c552, heme reduction potentials correlate with 13CN chemical shits and were used as a probe of electron density at heme–Fe [33]. We observed δ(13CN) = −3900 ppm (Fig. 4) for V49X/M69A (X = L, T, S) cyt c552, which is similar to Ht cyt c552, δ(13CN) = − 3948 ppm [33], and lower than for horse heart cyt c [δ(13CN) = − 3761 ppm] or yeast cyt c peroxidase [δ(13CN) = −3543 ppm] [31]. Under identical conditions, we did not observe a 13CN signal for the other mutants. Broad singlets were observed in some samples, which may suggest ligand exchange at the active site, and indicates that our mutations perturbed binding of anionic ligands to FeIII. This result also is in accord with the pH dependent UV– vis spectra. Selected 13CN-ligated proteins also were probed using IR spectroscopy. Shifts in C–N stretches have been used as probes of heme active sites, although the molar absorptivity can be very weak [38]. Characteristic IR spectra of FeIII M69A, V49T/M69A, and P27A/P28A/L29D/V49T/M69A are shown in Fig. 5. In all cases, we observe a more prominent band at 2038 cm−1, characteristic of free 13CN, and a weaker band centered at 2079 ± 1 cm−1 that we attribute to Fe-ligated 13CN. The corresponding 13 CN stretches in hemoglobin, myoglobin and horseradish peroxidase are 2078, 2079 and 2085 cm−1, respectively [38]. In dimeric cyt c, where M80 is misligated such that CN− can access the distal pocket, the 13CN stretch is 2078 cm− 1 [39]. These data suggest that the FeIII site in our mutants is somewhat more electron rich than wild-type cyt c, but it does not appear that we entirely reproduce the electronics of peroxidase sites. This could be due to differences in how the distal pocket interacts with ligands (e.g., in peroxidase enzymes, see Ref. [40]), which we are still investigating. The redox properties of each protein were investigated under inert atmosphere and under O2 (see below). Wild type cyt c552 has E°′ = 0.232 (by titration) and 0.200 (at modified gold electrodes) [19]. Note that all reduction potentials are reported versus NHE. Representative voltammograms are shown in Fig. 6 (all data at 10 mM NaPi + 100 mM NaCl pH 7, unless otherwise noted). Our measurement at pyrolytic graphite electrodes is 0.220 ± 0.01 V, in good agreement with the reported values. Removal of the coordinating M69 results in a significant shift in reduction potential to − 0.065 V. Replacement of hydrophobic V49 with T or S results in small anodic shifts in reduction potential (20–60 mV). Introduction of Asp (L29D) in the proximal pocket does not affect the FeIII/II reduction potentials, likely because it cannot rotate to interact with H15 or the heme due to the presence of two proline residues (P27, P28). Replacement of Pro27 and Pro 28 with Ala induces a cathodic shift by about 60 mV to a potential window similar to many heme peroxidases (around − 0.150 to − 0.250 V) [27]. (See Table 2.) Our measured value of E°′(FeIII/II) for M69A cyt c552 is higher than recently reported (E°′ = 0.130 V [23]) using very similar experimental conditions and pH. The main difference is buffer concentration (10 mM versus 50 mM [23]), so the origin of this discrepancy is not
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
11
Fig. 3. (A) Optical spectra of ferric M69A, V49S/M69A, and V49T/M69A cyt. c552 mutants. (B) Optical spectra of ferric L29D/V49S/M69A, and L29D/V49T/M69A cyt. c552 mutants at pH 3 (- - -) and pH 8 (—). (C) Optical spectra of ferric P27A/P28A/L29D/V49S/M69A, and P27A/P28A/L29D/V49T/M69A cyt. c552 mutants at pH 3 (- - -) and pH 8 (—).
readily apparent. For comparison, the reported FeIII/II reduction potentials for M80A cyt c range from E°′ = 0.185 V (pH 7, solution titration) [41] to ca. − 0.200 V (pH 7, immobilized on Au electrodes) [21,22]. The reduction potentials are pH-dependent (pH 3, E°′ = − 0.056 V [22] and pH 5, E°′ = −0.092 V [21]), and it was suggested that deprotonation of a heme–iron bound water gives rise to this behavior. Our measured reduction potential is more consistent with that of the FeIII– OH2 species of Met80Ala cyt c at pH 3 (E°′ = −0.056 V) [22]. The UV–
Fig. 4. 13C NMR spectrum of 5 mM Met69Ala cyt c552 + 20 mM K13CN in 25 mM sodium phosphate buffer, pH 7.
vis spectra for our cyt c552 mutants at different pH do not suggest changes in protonation state for M69A, but the optical spectra of other mutants do depend on pH (e.g., Fig. 3) (Table 2).
Fig. 5. Infrared spectra of 13CN-ligated cyt c552.
12
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
Fig. 6. (Left) Cyclic voltammograms for V49S/M69A (•••) and V49T/M69A (—) cyt c552 and (Right) Differential pulse voltammograms for M69A(- - -), V49S/M69A (•••), and V49T/M69A (—) cyt c552. Data collected in 10 mM NaPi buffer, 100 mM NaCl, pH 7 at room temperature using a freshly polished basal plane graphite electrode (Pt wire counter).
The presence of even trace amounts of O2 results in irreversible cyclic voltammograms for V49S/M69A, V49T/M69A, L29D/V49S/M69A, P27A/P28A/L29D/V49S/M69A, and P27A/P28A/L29D/V49T/M69A; M69A is somewhat less reactive toward O2, but will still react in the FeII state under air, analogous to the behavior of Met80Ala cyt c [20, 22]. Voltammograms collected in deoxygenated and air-saturated protein solutions are shown in Fig. 7. The magnitude of the current response is roughly proportional to the concentration of O2 with CVs in O2-saturated buffer ([O2] ~ 1 mM) showing about 3–4 times more current than CVs in air-saturated buffer ([O2] ~ 0.3 mM), but at high [O2] the current response is convoluted with background response from the electrode (see Supporting Information). Under air, our observed current responses (~ 5–7 μA, ~ 80 μA cm− 2) are slightly less than immobilized cyt c [22] (~ 15 μA), albeit at slower scan rates (10 mV s−1 versus 20 mV s−1, respectively). Our cyt c552 variants show current production for ≥ 60 min under controlled potential electrolysis conditions (15 μM protein, −0.1 V versus NHE applied potential, under air, constant stirring). For the L29D/ V49S/M29A protein, about 50% of the current response is lost after 60 min. The color of the protein solutions does not bleach appreciably, suggesting that the heme chromophore is intact. We can estimate a turnover rate of ~ 2 × 10−4 s− 1 (see Supporting Information), far below that for native heme–copper oxidases and related models (≥50 s−1) [12,13,42], or Ru(NH3)5-modified myoglobin (0.6 s−1) [14]. As observed for other mutants of M69A cyt c552 [24,25], our proteins react rapidly and irreversibly with peroxides. Heme degradation by reaction of peroxides with cyt c is known [43], resulting in several oxidized products [44]. Addition of 10 equivalents (100 μM) H2O2 or 3chloroperbenzoic acid (mCPBA) results in disappearance of the optical signature of the heme chromophore, with no apparent buildup of
intermediates by UV–vis spectroscopy (see Supporting Information). Additional peroxide causes more heme degradation. Based on this observation, H2O2 is not likely to be a major product of our experiments under air and O2, and if peroxide is produced, the ready availability of reducing equivalents can convert it to water. Finally, we tested the ability of the proteins to bind and activate nitrite, in analogy to related work on cyt c [22]. Addition of up to 2 mM NaNO2 to 6 μM cyt c552 variants had no effect on the heme absorption spectrum, suggesting that NO− 2 does not bind very strongly to the heme–iron. CV experiments also showed no electrochemical response, other than that of the FeIII/II couple from the protein (Supporting Information), suggesting that the FeII state also has a low affinity for NO− 2 or that NO− 2 reduction chemistry is very slow. The modest ORR activity and lack of activity with NO− 2 set our cyt c552 variants apart from immobilized cyt c [21,22] or modified myoglobins (Mbs) [12–15]. This could be because we are catalyzing a multielectron process, requiring proteins to remain near the electrode for complete catalytic cycles. For immobilized proteins, strong coupling between the heme and the electrode facilitates multi-electron reactions, while for Mbs, multiple redox cofactors (e.g., RuII(NH3)5-His [14]) are built directly into the scaffold. We also note that the electron transfer characteristics for cyt c and cyt c522 are similar; a direct comparison is available for proteins on alkylthiol self-assembled monolayers: [45]. The differences in reactivity likely are not due to intrinsic barriers. Differences in protein conformation associated with immobilization also could lead to changes in redox reactivity. Protein interaction with our carbon electrodes is probably very different from previous work where proteins were directly attached to protein surfaces [21,22]. Such potential differences in conformation can direct comparison a challenge. Alternatively, catalysis may be slower in the less
Fig. 7. (Left) Cyclic voltammograms for V49S/M69A cyt c552 in N2 (- - -) and air (––) saturated buffers. (Center) Cyclic voltammograms for L29D/V49S/M69A cyt c552 in N2 (- - -) and air (––) saturated buffers. (Right) Cyclic voltammograms for P27A/P28A/L29D/V49S/M69A cyt c552 in N2 (- - -) and air (––) saturated buffers. Data collected in 10 mM NaPi buffer, 100 mM NaCl, pH 7 at room temperature using a freshly polished basal plane graphite electrode (Pt wire counter).
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14 Table 2 Reduction potentials (E°′) for cyt c552 proteins. Protein
E°′ (pH 7)a
Wild type M69A V49S/M69A V49T/M69A L29D/V49S/M69A L29D/V49T/M69A P27A/P28A/L29D/V49S/M69A P27A/P28A/L29D/V49T/M69A
0.220 −0.065 −0.120 −0.080 −0.120 −0.080 −0.170 −0.120
a Collected in Ar-sparged 10 mM sodium phosphate buffer with 100 mM added NaCl, pH 7.
conformationally flexible cyt c552 (e.g., cyt c unfolds with 2 M guanidine·HCl [46] and cyt c552 unfolds with 4 M [28]). Cyt c552 contains additional helices proximal to the heme, and β-sheets near the heme propionates, additionally shielding the distal site compared to cyt c (Fig. 1). These differences in flexibility could be one origin of slow catalysis, and have a larger effect on more difficult reactions (e.g., NO− 2 reductions), where even cyt c exhibits modest reactivity [22]. The heme pocket is large enough to accommodate nitrite based on observed reactivity with H2O2 and mCPBA. We note that these peroxide shunt-like reactions are more favorable than NO− 2 reduction [1,24,25], which may enhance reactivity. Finally, we note that cyt c also contains a Tyr in the distal site. In addition to potential roles as a proton relay, Tyr also could reduce intermediates in ORR chemistry (e.g., H+/e− transfer to FeIII(O2•) to give FeIIIOOH, or reduction of Compound I or Compound II-like intermediates). 4. Conclusions and broader implications Modified cytochrome c552 can electrochemically reduce O2 at carbon electrodes. Introduction of hydrogen bond donors in the distal heme pocket induced small negative shifts in reduction potential, but slightly enhanced reactivity with O2 compared to the removal of the axial Met alone (M69A). Introduction of biologically-inspired of H-bond donating residues in the active site only marginally improved catalytic currents and produced small negative shifts in E°′. Mutations to increase peroxidase-like character in the proximal heme site did not improve catalysis, but did shift potentials to lower values to approximately the same range as peroxidase enzymes. The high reactivity, and degradation, of the enzymes in the presence of peroxide, suggests that O2 is mostly reduced to H2O during electrocatalysis. Catalysis is slower compared to related yeast cyt c, which we propose is due to differences in electrode interactions or due to a less flexible active site in thermally robust cyt c552. We are currently exploring the peroxidase activity of these proteins, as well as the effect of the introduction of additional redox active or hydrogen bonding amino acids near the heme–iron. Other oxygen-binding and oxygen-activating heme proteins (e.g., myoglobin and cytochrome c peroxidase) have different hydrogen bond donors in the active site. Histidine, arginine, and asparagine are functionally important in peroxidases [47, 48], and it is known that the distal pocket accommodates different ligands. Likewise, in myoglobin, the location of the distal noncoordinating His substantially affects reactivity with peroxide [49]. The mutations that we made to cyt c552 were intended to introduce isostructural (conservative) mutations to the distal pocket and peroxidase-like mutations to the proximal pocket. While introduction of H-bond donating Ser or Thr slightly improved catalytic currents for O2 reduction, these enhancements were smaller than expected and a more promising direction may be to introduce His, which can act as both a Brønsted acid and base, and is a better analog to known O2− binding and O2-activating hemes. Finally, in addition to proton-relays, high overpotentials and multi-electron chemistries are still key challenges for artificial O2 reduction. As noted above, natural and artificial proteins
13
with multiple redox cofactors, or direct electrode interface, are the best catalysts. To address these challenges, we are currently investigating both alterative proton relay groups (e.g., His) and cofactor substitution [50]. Acknowledgments The Simon Fraser University President's Research Startup Grant and the National Sciences and Engineering Research Council (RGPIN05559 to J. J. W.) supported this work. John Thompson assisted with the collection of Raman spectra and Kara Bren provided helpful advice on 13C NMR acquisition. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jinorgbio.2016.01.023. References [1] M. Sono, M.P. Roach, E.D. Coulter, J.H. Dawson, Chem. Rev. 96 (1996) 2841–2887. [2] V.R.I. Kaila, M.I. Verkhovsky, M. Wikström, Chem. Rev. 110 (2010) 7062–7081. [3] F. Jaouen, E. Proietti, M. Lefevre, R. Chenitz, J.-P. Dodelet, G. Wu, H.T. Chung, C.M. Johnston, P. Zelenay, Energy Environ. Sci. 4 (2011) 114–130. [4] C. Costentin, M. Robert, J.-M. Savéant, C. Tard, Acc. Chem. Res. 47 (2013) 271–280. [5] D.R. Weinberg, C.J. Gagliardi, J.F. Hull, C.F. Murphy, C.A. Kent, B.C. Westlake, A. Paul, D.H. Ess, D.G. McCafferty, T.J. Meyer, Chem. Rev. 112 (2012) 4016–4093. [6] J.J. Warren, T.A. Tronic, J.M. Mayer, Chem. Rev. 110 (2010) 6961–7001. [7] V. Sharma, G. Enkavi, I. Vattulainen, T. Róg, M. Wikström, Proc. Natl. Acad. Sci. U. S. A. 112 (2015) 2040–2045. [8] I. Bento, C. Silva, Z. Chen, L. Martins, P. Lindley, C. Soares, BMC Struct. Biol. 10 (2010) 28–42. [9] C.T. Carver, B.D. Matson, J.M. Mayer, J. Am. Chem. Soc. 134 (2012) 5444–5447. [10] R. McGuire Jr., D.K. Dogutan, T.S. Teets, J. Suntivich, Y. Shao-Horn, D.G. Nocera, Chem. Sci. 1 (2010) 411–414. [11] K. Mittra, S. Chatterjee, S. Samanta, A. Dey, Inorg. Chem. 52 (2013) 14317–14325. [12] K.D. Miner, A. Mukherjee, Y.-G. Gao, E.L. Null, I.D. Petrik, X. Zhao, N. Yeung, H. Robinson, Y. Lu, Angew. Chem. Int. Ed. 51 (2012) 5589–5592. [13] Y. Yu, C. Cui, X. Liu, I.D. Petrik, J. Wang, Y. Lu, J. Am. Chem. Soc. 137 (2015) 11570–11573. [14] R. Margalit, I. Pecht, H.B. Gray, J. Am. Chem. Soc. 105 (1983) 301–302. [15] C.-M. Che, H.-J. Chiang, R. Margalit, H.B. Gray, Catal. Lett. 1 (1988) 51–54. [16] R. Vazquez-Duhalt, J. Mol. Catal. B Enzym. 7 (1999) 241–249. [17] R.A. Scott, G.A. Mauk, Cytochrome c: A Multidisciplinary Approach, University Science Books, Sausalito, CA, 1995. [18] L. Banci, I. Bertini, K.L. Bren, H.B. Gray, P. Sompornpisut, P. Turano, Biochemistry 34 (1995) 11385–11398. [19] J.A. Fee, Y. Chen, T.R. Todaro, K.L. Bren, K.M. Patel, M.G. Hill, E. Gomez-Moran, T.M. Loehr, J. Ai, L. Thöny-Meyer, P.A. Williams, E. Sturam, V. Sridhar, D.E. Mcree, Protein Sci. 9 (2000) 2074–2084. [20] Y. Lu, D.R. Casimiro, K.L. Bren, J.H. Richards, H.B. Gray, Proc. Natl. Acad. Sci. U. S. A. 90 (1993) 11456–11459. [21] S. Casalini, G. Battistuzzi, M. Borsari, C.A. Bortolotti, A. Ranieri, M. Sola, J. Phys. Chem. B 112 (2008) 1555–1563. [22] S. Casalini, G. Battistuzzi, M. Borsari, A. Ranieri, M. Sola, J. Am. Chem. Soc. 130 (2008) 15099–15104. [23] R.K. Behera, H. Nakajima, J. Rajbongshi, Y. Watanabe, S. Mazumdar, Biochemistry 52 (2013) 1373–1384. [24] Y. Ichikawa, H. Nakajima, Y. Watanabe, ChemBioChem 7 (2006) 1582–1589. [25] H. Nakajima, Y. Ichikawa, Y. Satake, N. Takatani, S.K. Manna, J. Rajbongshi, S. Mazumdar, Y. Watanabe, ChemBioChem 9 (2008) 2954–2957. [26] H. Nakajima, K. Ramanathan, N. Kawaba, Y. Watanabe, Dalton Trans. 39 (2010) 3105–3114. [27] G. Battistuzzi, M. Bellei, C.A. Bortolotti, M. Sola, Arch. Biochem. Biophys. 500 (2010) 21–36. [28] S. Yamada, F. Bouley, N. D., G.E. Keller, W.C. Ford, H.B. Gray, J.R. Winkler, Proc. Natl. Acad. Sci. U. S. A. 110 (2013) 1606–1610. [29] J.D. Blakemore, N.D. Schley, D. Balcells, J.F. Hull, G.W. Olack, C.D. Incarvito, O. Eisenstein, G.W. Brudvig, R.H. Crabtree, J. Am. Chem. Soc. 132 (2010) 16017–16029. [30] J.D. Blakemore, A. Gupta, J.J. Warren, B.S. Brunschwig, H.B. Gray, J. Am. Chem. Soc. 135 (2013) 18288–18291. [31] H. Fujii, J. Am. Chem. Soc. 124 (2002) 5936–5937. [32] H. Fujii, T. Yoshida, Inorg. Chem. 45 (2006) 6816–6827. [33] S. Bowman, E. J., K. Bren, L, Inorg. Chem. 49 (2010) 7890–7897. [34] L.B. Vitello, M. Huang, J.E. Erman, Biochemistry 29 (1990) 4283–4288. [35] C.W. Conroy, P. Tyma, P.H. Daum, J.E. Erman, Biochim. Biophys. Acta 537 (1978) 62–69. [36] T. Yonetani, H. Anni, J. Biol. Chem. 262 (1987) 9547–9554. [37] G. Smulevich, J.M. Mauro, L.A. Fishel, A.M. English, J. Kraut, T.G. Spiro, Biochemistry 27 (1988) 5477–5485.
14
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
[38] S. Yoshikawa, D.H. O'Keeffe, W.S. Caughey, J. Biol. Chem. 260 (1985) 3518–3528. [39] A.D. Nugraheni, S. Nagao, S. Yanagisawa, T. Ogura, S. Hirota, J. Biol. Inorg. Chem. 18 (2013) 383–390. [40] S.L. Edwards, T.L. Poulos, J. Biol. Chem. 265 (1990) 2588–2595. [41] C.J. Wallace, I. Clark-Lewis, J. Biol. Chem. 267 (1992) 3852–3861. [42] H.J. Lee, R.B. Gennis, P. Ädelroth, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 17661–17666. [43] J.A. Villegas, A.G. Mauk, R. Vazquez-Duhalt, Chem. Biol. 7 (2000) 237–244. [44] J.N. Rodriguez-Lopez, J. Hernández-Ruiz, F. Garcia-Cánovas, R.N.F. Thorneley, M. Acosta, M.B. Arnao, J. Biol. Chem. 272 (1997) 5469–5476. [45] S. Lecomte, P. Hildebrandt, T. Soulimane, J. Phys. Chem. B 103 (1999) 10053–10064.
[46] E.V. Pletneva, H.B. Gray, J.R. Winkler, J. Mol. Biol. 345 (2005) 855–867. [47] S. Nagano, M. Tanaka, K. Ishimori, Y. Watanabe, I. Morishima, Biochemistry 35 (1996) 14251–14258. [48] L.B. Vitello, J.E. Erman, M.A. Miller, J. Wang, J. Kraut, Biochemistry 32 (1993) 9807–9818. [49] T. Matsui, S.-i. Ozaki, E. Liong, G.N. Phillips, Y. Watanabe, J. Biol. Chem. 274 (1999) 2838–2844. [50] S.M. Ibrahim, H. Nakajima, T. Ohta, K. Ramanathan, N. Takatani, Y. Naruta, Y. Watanabe, Biochemistry 50 (2011) 9826–9835.
Contents lists available at ScienceDirect
Journal of Inorganic Biochemistry journal homepage: www.elsevier.com/locate/jinorgbio
Catalytic reduction of dioxygen with modified Thermus thermophilus cytochrome c552 Jonathan Husband 1, Michael S. Aaron 1, Rajneesh K. Bains, Andrew R. Lewis, Jeffrey J. Warren ⁎ Department of Chemistry, Simon Fraser University, 8888 University Drive, Burnaby, BC V5A 1S6, Canada
a r t i c l e
i n f o
Article history: Received 2 November 2015 Received in revised form 23 December 2015 Accepted 18 January 2016 Available online 20 January 2016 Keywords: Heme proteins Oxygen reduction Electrochemistry Proton relay
a b s t r a c t Efficient catalysis of the oxygen reduction reaction (ORR) is of central importance to function in fuel cells. Metalloproteins, such as laccase (Cu) or cytochrome c oxidase (Cu/Fe–heme) carry out the 4H+/4e− reduction quite efficiently, but using large, complex protein frameworks. Smaller heme proteins also can carry out ORR, but less efficiently. To gain greater insight into features that promote efficient ORR, we expressed, characterized, and investigated the electrochemical behavior of six new mutants of cytochrome c552 from Thermus thermophilus: V49S/M69A, V49T/M69A, L29D/V49S/M69A, P27A/P28A/L29D/V49S/M69A, and P27A/P28A/ L29D/V49T/M69A. Mutation to V49 causes only minor shifts to FeIII/II reduction potentials (E°′), but introduction of Ser provides a hydrogen bond donor that slightly enhances oxygen reduction activity. Mutation of L29 to D induces small shifts in heme optical spectra, but not to E°′ (within experimental error). Replacement of P27 and P28 with A in both positions induces a −50 mV shift in E°′, again with small changes to the optical spectra. Both the optical spectra and reduction potentials have signatures consistent with peroxidase enzymes. The V49S and V49T mutations have the largest impact of ORR catalysis, suggesting that increased electron density at the Fe site does not improve O2 reduction chemistry. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Heme proteins are among Nature's most diverse tools, catalyzing a vide variety of redox transformations, from oxygenation reactions in cytochromes P450 or nitric oxide synthase [1], to the reduction of dioxygen to water in cytochrome c oxidase [2] (the oxygen reduction reaction, ORR). The latter reaction is especially important in the search for an ORR catalyst that functions with minimal loss in free energy (overpotential) in fuel cell cathodes. The expense and rarity of currently employed Pt/C fuel cell catalysts spurred interest in using porphyrin complexes and heme proteins as O2 reduction catalysts [3]. The ORR is inherently a proton-coupled electron transfer (PCET) reaction [4–6], as it requires the addition of 4H+ and 4e− to O2, and scission of the O2 double bond, to produce 2H2O. Nature efficiently carries out this reaction in cytochrome c oxidase [7] and laccase enzymes [8], which incorporate multiple redox cofactors and proton-shuttling channels (proton relays). Proton transfer is critical for function, as demonstrated by the marked improvements in electrochemical ORR by ironand cobalt-porphyrin catalysts [9–11]. In protein catalysts derived from the myoglobin active site, introduction of a redox-active Cu, a cross linked tyrosine and histidine, and proton relays results in ORR catalysis; [12,13] these modified-myoglobins are active site models for ⁎ Corresponding author. E-mail address: [email protected] (J.J. Warren). 1 These authors contributed equally to this work.
http://dx.doi.org/10.1016/j.jinorgbio.2016.01.023 0162-0134/© 2016 Elsevier Inc. All rights reserved.
cytochrome c oxidase. Other myoglobins, that incorporate multiple attached to different surface histidine redox units (three Ru(NH3)2+ 5 residues), also are ORR catalysts with ascorbate or durohydroquinone as sacrificial reductants [14,15]. Engineered protein catalysts are typically not very robust (low turnover number) in part because of cofactor loss/degradation or low thermal stability. In contrast to the above heme proteins, c-type cytochromes feature a covalently anchored cofactor, which can eliminate cofactor (heme) loss as a mechanism of catalyst degradation. Moreover, these proteins exhibit facile redox chemistry and tolerate many mutations. These features have motivated their use as biocatalysts [16]. In this work, we investigate a series of new variants of thermostable cytochrome c552 (cyt c552) from Thermus thermophilus that contain proton relays and can electrochemically reduce O2. c-Type cytochromes are typically thought of as biological electron carriers [17]. Native proteins are covalently tethered to the heme vinyl groups via cysteine in a conserved CXXCH motif (where X is any amino acid). X-ray structures of yeast cyt c [18] and T. thermophilus cyt c552 [19] are shown in Fig. 1. The histidine residue at the end of the CXXCH sequence binds at the proximal position (H18 in cyt c and H15 in cyt c552, Fig. 1), and a methionine residue binds at the distal position (M80 in cyt c and M69 in cyt c552), coordinatively saturating the heme– iron. This largely precludes adventitious reactivity with substrates, such as O2. Replacement of M80 with A in yeast cyt c results in a 5-coordinate heme site that reversibly binds O2 [20]. The formal FeIII/II reduction potential (E°′) shifts to lower values and the 5-coordinate protein can
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
9
Fig. 1. Comparison of yeast cyt c active sites (left, CN−-bound structure of M80A, PDB ID 1FHB) and Thermus thermophilus cyt c552 (right, PDB ID 1DT1). The sites of distal (V49) and proximal (P26, P28, and P29) are shown for cyt c552. The distal A80 is shown for cyt c, but the analogous M69 is omitted for clarity in the structure of cyt c 552.
electrochemically reduce O2 both in solution [21] and when covalently attached to thiol-modified Au electrodes [22]. At pH 7, current onset is observed at ~0 V versus NHE. Removal of the analogous methionine residue (M69) in the more thermostable T. thermophilus cyt c552 also results in a properly folded protein [23]. Changes to the UV–visible properties in the mutant protein are in accord with those observed for M80A yeast cyt c. The M69A mutant also was demonstrated to act as a peroxidase when V49 was replaced with aspartate [24,25]. Peroxidase activity was improved by replacing an oxidizable Y with F, but peroxide-induced heme degradation was still observed over the course of several seconds [26]. Here, we take advantage of the thermal stability of M69A cyt c552 to investigate the solution chemistry of a cytochrome-based O2 reduction catalyst. We introduced serine at position 49 (V49S) to act as both a H-bonding partner and promote protonation of activated oxygen species (e.g., FeIII–O2), likely from H+ outside of the protein (mutations summarized in Fig. 2). The V49T mutant provides isostructural, polar modification. We also explored modifications to the proximal pocket that are intended to install a peroxidase-like Asp–His interaction [27] to promote binding and activation of O2. These mutations were L29D, to introduce Asp, and P27A/P28A to allow greater backbone flexibility near the proximal histidine. 2. Materials and methods All buffer salts were from J. T. Baker. Luria Broth and yeast extract was from Research Products International Corporation. K13CN was from Cambridge Isotope Laboratory. Chromatography media were from G.E. Healthcare. Water (18 MΩ cm− 1) was from a Barnstead EASYpure purification system. All gases were obtained from Praxair Canada. DNA polymerase (Q5) and DpnI were obtained from New
England Biolabs. PCR primers were from, and Eurofins Operon carried out DNA sequencing. Plasmids bearing the wild-type cytochrome c552 gene and the pec86 heme cassette were gifts from H. B. Gray (California Institute of Technology). Site-directed mutants were obtained using the Quickchange protocol (Agilent Technology). Expression and purification of all proteins were carried out according to the literature [28]. All electrochemical experiments were carried out using a Gamry Interface 1000 or a CH Instruments 600B potentiostat. Cyclic voltammetry and differential pulse voltammetry were used to collect reversible FeIII/II potentials, and cyclic voltammetry was used for experiments under air and O2. Solutions under N2, air, and O2 were sparged for 5 min before data collection. Formal potentials were referenced to an AgCl/Ag electrode calibrated versus potassium ferricyanide, and are reported with respect to the normal hydrogen electrode (NHE). Platinum wire was used as a counter electrode. Basal plane graphite electrodes were prepared according to a literature procedure [29,30]. Pyrolytic graphite was obtained from http://www.graphitestore.com. Loctite Hysol 9460 epoxy was obtained from McMaster-Carr. Electrodes were polished briefly using wet 6000 grit SiC paper before each run, washed with copious water and dried briefly using a heat gun. Optical spectra were collected on a Cary 100-Bio UV–visible spectrometer or a Photon Control SPM-002-EH CCD with a SPL-1DH deuterium/tungsten light source. Raman spectra were collected using a Renishaw inVia Raman microscope equipped with a 514 nm laser. Samples were deposited on glass as thin films to enhance signal to noise ratio; thin film spectra match solution measurements for all proteins. Infrared spectra were collected using a Perkin-Elmer Spectrum 2 Attenuated Total Reflectance (ATR) spectrometer, using 12 mM protein solutions + 20 mM K13CN. Data were smoothed using a moving average algorithm in MATLAB and the raw data are provided in the Supporting Information. Mass spectra (electrospray ionization positive ion) were
Fig. 2. Models of all cyt c552 variants. (A) Model of M69A cyt c552 variants with V49T mutations. (B) Model of M69A cyt c552 variants with V49S mutations, showing putative O2 binding. The distances were estimated by modeling the heme from oxy-cytochrome c peroxidase (PDB ID 1DCC) in the cyt c552 (PDB ID 1C52) heme site. (C) Model of L29D/M69A cyt c552 variants with V49T or V49S mutations.
10
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
collected on an Agilent 6210 TOF ESI-MS instrument. Circular dichroism spectra were collected using a Jasco J-810 spectropolarimeter and are reported as mean residue ellipticity. Single pulse 13C spectra without 1 H decoupling were collected on a Bruker 400 MHz AVANCE III NMR spectrometer operating at 100.6 MHz for 13C using a Bruker 5 mm BBFO probe. No 2H locking was employed and the spectral sweep for locking was turned off. Each sample 300 μL was added to a 5 mm outer diameter D2O-matched Shigemi tube and the magnetic field homogeneity was optimized manually by observing the 1H NMR signal of water and the sample shimming was checked using the 13C signal of the excess, free 13CN− ions. An 8 μs 90-degree radio frequency excitation pulse was used with a spectral width of 1000 ppm, 16,384 complex data points acquired (acquisition time 0.041 s), and a recycle delay of 0.025 s [31–33]. The 13C transmitter offset was set to − 3950 or −3600 ppm and the probe was carefully matched and tuned for each sample. Following 16 dummy scans, 400,000–800,000 scans were accumulated (24–48 h total acquisition time). 500 Hz exponential apodization was applied before Fourier transformation and spectral baselines were manually corrected using a multipoint spline function. The chemical shift scale was referenced using an external sample of K13CN in H2O (set to +168.6 ppm) run immediately following acquisition without any adjustment of the magnetic field, but with the probe re-tuned and matched using a transmitter offset of 100 ppm. 3. Results and discussion We expressed and purified the known [24] M69A variant of T. thermophilus cyt c552 and 6 new variants: V49S/M69A, V49T/M69A, L29D/V49S/M69A, P27A/P28A/L29D/V49S/M69A, and P27A/P28A/ L29D/V49T/M69A. The proteins were characterized using mass spectrometry, circular dichroism (CD) spectroscopy, and UV–visible spectrophotometry. Recombinant cyt c522 expressed in Escherichia coli is cleaved between Ala2 and Asp3; [19] our observed masses for all protein variants are consistent with this observation. Calculated and observed masses (including spectra) are given in the Supporting Information. The CD spectra of the ferric M69A, V49S/M69A, and V49T/M69A proteins are identical to each other, and to that of the wild-type protein [28,19]. Introduction of Asp at position 29 causes minor changes in the CD spectra between 225 and 250 nm; we conclude that the protein structures in all mutants are largely intact, with minor perturbations for the L29D-containing proteins. Optical spectra of the M69A, V49S/M69A, and V49T/M69A ferric proteins (data summarized in Table 1, Fig. 3) are consistent with a highspin, 5-coordinate heme, as for M80A cyt c [20]. Notably, the optical spectra of these proteins are not sensitive to pH between 3 and 8. Introduction of the L29D causes a small (3 nm) red shift in the Soret band, and shifts in the Q-bands that are reminiscent of optical signatures observed in heme peroxidases [34,35]. The optical spectra of these proteins are sensitive to pH, which could be due to protonation/ deprotonation of an iron-ligated H2O, as suggested for the analogous
Table 1 UV–vis features of cyt c552 proteins.
M69A FeIII (FeII–CN)
Soret
Visible
398 418
493 522
572 566
L29D/V49S/M69A and L29D/V49T/M69A 401 529 FeIII (pH 3) FeIII (pH 8) 406 531 417 523 (FeII–CN)
551
M69A, V49S/M69AAAT, and V49T/M69AAAS 405 528 FeIII (pH 3) 406 528 FeIII (pH 8) 417 522 (FeII–CN)
629 650 553
623
621
650
M80A cyt c [20–22]. Finally, introduction of Ala at positions 27 and 28, yields proteins with optical spectra similar to the L29D/V49T(S)/M69A proteins, but with a more pronounced band near 650 nm exhibits less pH dependence. The UV–vis features in this last set of proteins are similar to heme peroxidase proteins [36]. The ferric proteins are readily reduced using sodium dithionite and readily bind cyanide (Table 1). We also investigated the electronic structure of the hemes using resonance Raman spectroscopy (see Supporting Information). The spectra are different from WT cyt c552 and peroxidase enzymes [19,37] and exhibit only small changes between mutants. For the P27A/P28A/L29D/V49T/ M69A and P27A/P28A/L29D/V49S/M69A mutants there is a 1–3 cm−1 red shift in Raman bands, roughly consistent with observations for peroxidases [37] where the heme site is more electron rich. We also probed the electronic structure of the Met69Ala variant using 13C NMR and IR spectroscopies. The 13C chemical shift of cyanide in FeIII–CN hemes correlates with the electron density at the iron center [31–33]. In a work with structurally related (but smaller) Hydrogenbacter thermophilus (Ht) cytochrome c552, heme reduction potentials correlate with 13CN chemical shits and were used as a probe of electron density at heme–Fe [33]. We observed δ(13CN) = −3900 ppm (Fig. 4) for V49X/M69A (X = L, T, S) cyt c552, which is similar to Ht cyt c552, δ(13CN) = − 3948 ppm [33], and lower than for horse heart cyt c [δ(13CN) = − 3761 ppm] or yeast cyt c peroxidase [δ(13CN) = −3543 ppm] [31]. Under identical conditions, we did not observe a 13CN signal for the other mutants. Broad singlets were observed in some samples, which may suggest ligand exchange at the active site, and indicates that our mutations perturbed binding of anionic ligands to FeIII. This result also is in accord with the pH dependent UV– vis spectra. Selected 13CN-ligated proteins also were probed using IR spectroscopy. Shifts in C–N stretches have been used as probes of heme active sites, although the molar absorptivity can be very weak [38]. Characteristic IR spectra of FeIII M69A, V49T/M69A, and P27A/P28A/L29D/V49T/M69A are shown in Fig. 5. In all cases, we observe a more prominent band at 2038 cm−1, characteristic of free 13CN, and a weaker band centered at 2079 ± 1 cm−1 that we attribute to Fe-ligated 13CN. The corresponding 13 CN stretches in hemoglobin, myoglobin and horseradish peroxidase are 2078, 2079 and 2085 cm−1, respectively [38]. In dimeric cyt c, where M80 is misligated such that CN− can access the distal pocket, the 13CN stretch is 2078 cm− 1 [39]. These data suggest that the FeIII site in our mutants is somewhat more electron rich than wild-type cyt c, but it does not appear that we entirely reproduce the electronics of peroxidase sites. This could be due to differences in how the distal pocket interacts with ligands (e.g., in peroxidase enzymes, see Ref. [40]), which we are still investigating. The redox properties of each protein were investigated under inert atmosphere and under O2 (see below). Wild type cyt c552 has E°′ = 0.232 (by titration) and 0.200 (at modified gold electrodes) [19]. Note that all reduction potentials are reported versus NHE. Representative voltammograms are shown in Fig. 6 (all data at 10 mM NaPi + 100 mM NaCl pH 7, unless otherwise noted). Our measurement at pyrolytic graphite electrodes is 0.220 ± 0.01 V, in good agreement with the reported values. Removal of the coordinating M69 results in a significant shift in reduction potential to − 0.065 V. Replacement of hydrophobic V49 with T or S results in small anodic shifts in reduction potential (20–60 mV). Introduction of Asp (L29D) in the proximal pocket does not affect the FeIII/II reduction potentials, likely because it cannot rotate to interact with H15 or the heme due to the presence of two proline residues (P27, P28). Replacement of Pro27 and Pro 28 with Ala induces a cathodic shift by about 60 mV to a potential window similar to many heme peroxidases (around − 0.150 to − 0.250 V) [27]. (See Table 2.) Our measured value of E°′(FeIII/II) for M69A cyt c552 is higher than recently reported (E°′ = 0.130 V [23]) using very similar experimental conditions and pH. The main difference is buffer concentration (10 mM versus 50 mM [23]), so the origin of this discrepancy is not
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
11
Fig. 3. (A) Optical spectra of ferric M69A, V49S/M69A, and V49T/M69A cyt. c552 mutants. (B) Optical spectra of ferric L29D/V49S/M69A, and L29D/V49T/M69A cyt. c552 mutants at pH 3 (- - -) and pH 8 (—). (C) Optical spectra of ferric P27A/P28A/L29D/V49S/M69A, and P27A/P28A/L29D/V49T/M69A cyt. c552 mutants at pH 3 (- - -) and pH 8 (—).
readily apparent. For comparison, the reported FeIII/II reduction potentials for M80A cyt c range from E°′ = 0.185 V (pH 7, solution titration) [41] to ca. − 0.200 V (pH 7, immobilized on Au electrodes) [21,22]. The reduction potentials are pH-dependent (pH 3, E°′ = − 0.056 V [22] and pH 5, E°′ = −0.092 V [21]), and it was suggested that deprotonation of a heme–iron bound water gives rise to this behavior. Our measured reduction potential is more consistent with that of the FeIII– OH2 species of Met80Ala cyt c at pH 3 (E°′ = −0.056 V) [22]. The UV–
Fig. 4. 13C NMR spectrum of 5 mM Met69Ala cyt c552 + 20 mM K13CN in 25 mM sodium phosphate buffer, pH 7.
vis spectra for our cyt c552 mutants at different pH do not suggest changes in protonation state for M69A, but the optical spectra of other mutants do depend on pH (e.g., Fig. 3) (Table 2).
Fig. 5. Infrared spectra of 13CN-ligated cyt c552.
12
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
Fig. 6. (Left) Cyclic voltammograms for V49S/M69A (•••) and V49T/M69A (—) cyt c552 and (Right) Differential pulse voltammograms for M69A(- - -), V49S/M69A (•••), and V49T/M69A (—) cyt c552. Data collected in 10 mM NaPi buffer, 100 mM NaCl, pH 7 at room temperature using a freshly polished basal plane graphite electrode (Pt wire counter).
The presence of even trace amounts of O2 results in irreversible cyclic voltammograms for V49S/M69A, V49T/M69A, L29D/V49S/M69A, P27A/P28A/L29D/V49S/M69A, and P27A/P28A/L29D/V49T/M69A; M69A is somewhat less reactive toward O2, but will still react in the FeII state under air, analogous to the behavior of Met80Ala cyt c [20, 22]. Voltammograms collected in deoxygenated and air-saturated protein solutions are shown in Fig. 7. The magnitude of the current response is roughly proportional to the concentration of O2 with CVs in O2-saturated buffer ([O2] ~ 1 mM) showing about 3–4 times more current than CVs in air-saturated buffer ([O2] ~ 0.3 mM), but at high [O2] the current response is convoluted with background response from the electrode (see Supporting Information). Under air, our observed current responses (~ 5–7 μA, ~ 80 μA cm− 2) are slightly less than immobilized cyt c [22] (~ 15 μA), albeit at slower scan rates (10 mV s−1 versus 20 mV s−1, respectively). Our cyt c552 variants show current production for ≥ 60 min under controlled potential electrolysis conditions (15 μM protein, −0.1 V versus NHE applied potential, under air, constant stirring). For the L29D/ V49S/M29A protein, about 50% of the current response is lost after 60 min. The color of the protein solutions does not bleach appreciably, suggesting that the heme chromophore is intact. We can estimate a turnover rate of ~ 2 × 10−4 s− 1 (see Supporting Information), far below that for native heme–copper oxidases and related models (≥50 s−1) [12,13,42], or Ru(NH3)5-modified myoglobin (0.6 s−1) [14]. As observed for other mutants of M69A cyt c552 [24,25], our proteins react rapidly and irreversibly with peroxides. Heme degradation by reaction of peroxides with cyt c is known [43], resulting in several oxidized products [44]. Addition of 10 equivalents (100 μM) H2O2 or 3chloroperbenzoic acid (mCPBA) results in disappearance of the optical signature of the heme chromophore, with no apparent buildup of
intermediates by UV–vis spectroscopy (see Supporting Information). Additional peroxide causes more heme degradation. Based on this observation, H2O2 is not likely to be a major product of our experiments under air and O2, and if peroxide is produced, the ready availability of reducing equivalents can convert it to water. Finally, we tested the ability of the proteins to bind and activate nitrite, in analogy to related work on cyt c [22]. Addition of up to 2 mM NaNO2 to 6 μM cyt c552 variants had no effect on the heme absorption spectrum, suggesting that NO− 2 does not bind very strongly to the heme–iron. CV experiments also showed no electrochemical response, other than that of the FeIII/II couple from the protein (Supporting Information), suggesting that the FeII state also has a low affinity for NO− 2 or that NO− 2 reduction chemistry is very slow. The modest ORR activity and lack of activity with NO− 2 set our cyt c552 variants apart from immobilized cyt c [21,22] or modified myoglobins (Mbs) [12–15]. This could be because we are catalyzing a multielectron process, requiring proteins to remain near the electrode for complete catalytic cycles. For immobilized proteins, strong coupling between the heme and the electrode facilitates multi-electron reactions, while for Mbs, multiple redox cofactors (e.g., RuII(NH3)5-His [14]) are built directly into the scaffold. We also note that the electron transfer characteristics for cyt c and cyt c522 are similar; a direct comparison is available for proteins on alkylthiol self-assembled monolayers: [45]. The differences in reactivity likely are not due to intrinsic barriers. Differences in protein conformation associated with immobilization also could lead to changes in redox reactivity. Protein interaction with our carbon electrodes is probably very different from previous work where proteins were directly attached to protein surfaces [21,22]. Such potential differences in conformation can direct comparison a challenge. Alternatively, catalysis may be slower in the less
Fig. 7. (Left) Cyclic voltammograms for V49S/M69A cyt c552 in N2 (- - -) and air (––) saturated buffers. (Center) Cyclic voltammograms for L29D/V49S/M69A cyt c552 in N2 (- - -) and air (––) saturated buffers. (Right) Cyclic voltammograms for P27A/P28A/L29D/V49S/M69A cyt c552 in N2 (- - -) and air (––) saturated buffers. Data collected in 10 mM NaPi buffer, 100 mM NaCl, pH 7 at room temperature using a freshly polished basal plane graphite electrode (Pt wire counter).
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14 Table 2 Reduction potentials (E°′) for cyt c552 proteins. Protein
E°′ (pH 7)a
Wild type M69A V49S/M69A V49T/M69A L29D/V49S/M69A L29D/V49T/M69A P27A/P28A/L29D/V49S/M69A P27A/P28A/L29D/V49T/M69A
0.220 −0.065 −0.120 −0.080 −0.120 −0.080 −0.170 −0.120
a Collected in Ar-sparged 10 mM sodium phosphate buffer with 100 mM added NaCl, pH 7.
conformationally flexible cyt c552 (e.g., cyt c unfolds with 2 M guanidine·HCl [46] and cyt c552 unfolds with 4 M [28]). Cyt c552 contains additional helices proximal to the heme, and β-sheets near the heme propionates, additionally shielding the distal site compared to cyt c (Fig. 1). These differences in flexibility could be one origin of slow catalysis, and have a larger effect on more difficult reactions (e.g., NO− 2 reductions), where even cyt c exhibits modest reactivity [22]. The heme pocket is large enough to accommodate nitrite based on observed reactivity with H2O2 and mCPBA. We note that these peroxide shunt-like reactions are more favorable than NO− 2 reduction [1,24,25], which may enhance reactivity. Finally, we note that cyt c also contains a Tyr in the distal site. In addition to potential roles as a proton relay, Tyr also could reduce intermediates in ORR chemistry (e.g., H+/e− transfer to FeIII(O2•) to give FeIIIOOH, or reduction of Compound I or Compound II-like intermediates). 4. Conclusions and broader implications Modified cytochrome c552 can electrochemically reduce O2 at carbon electrodes. Introduction of hydrogen bond donors in the distal heme pocket induced small negative shifts in reduction potential, but slightly enhanced reactivity with O2 compared to the removal of the axial Met alone (M69A). Introduction of biologically-inspired of H-bond donating residues in the active site only marginally improved catalytic currents and produced small negative shifts in E°′. Mutations to increase peroxidase-like character in the proximal heme site did not improve catalysis, but did shift potentials to lower values to approximately the same range as peroxidase enzymes. The high reactivity, and degradation, of the enzymes in the presence of peroxide, suggests that O2 is mostly reduced to H2O during electrocatalysis. Catalysis is slower compared to related yeast cyt c, which we propose is due to differences in electrode interactions or due to a less flexible active site in thermally robust cyt c552. We are currently exploring the peroxidase activity of these proteins, as well as the effect of the introduction of additional redox active or hydrogen bonding amino acids near the heme–iron. Other oxygen-binding and oxygen-activating heme proteins (e.g., myoglobin and cytochrome c peroxidase) have different hydrogen bond donors in the active site. Histidine, arginine, and asparagine are functionally important in peroxidases [47, 48], and it is known that the distal pocket accommodates different ligands. Likewise, in myoglobin, the location of the distal noncoordinating His substantially affects reactivity with peroxide [49]. The mutations that we made to cyt c552 were intended to introduce isostructural (conservative) mutations to the distal pocket and peroxidase-like mutations to the proximal pocket. While introduction of H-bond donating Ser or Thr slightly improved catalytic currents for O2 reduction, these enhancements were smaller than expected and a more promising direction may be to introduce His, which can act as both a Brønsted acid and base, and is a better analog to known O2− binding and O2-activating hemes. Finally, in addition to proton-relays, high overpotentials and multi-electron chemistries are still key challenges for artificial O2 reduction. As noted above, natural and artificial proteins
13
with multiple redox cofactors, or direct electrode interface, are the best catalysts. To address these challenges, we are currently investigating both alterative proton relay groups (e.g., His) and cofactor substitution [50]. Acknowledgments The Simon Fraser University President's Research Startup Grant and the National Sciences and Engineering Research Council (RGPIN05559 to J. J. W.) supported this work. John Thompson assisted with the collection of Raman spectra and Kara Bren provided helpful advice on 13C NMR acquisition. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jinorgbio.2016.01.023. References [1] M. Sono, M.P. Roach, E.D. Coulter, J.H. Dawson, Chem. Rev. 96 (1996) 2841–2887. [2] V.R.I. Kaila, M.I. Verkhovsky, M. Wikström, Chem. Rev. 110 (2010) 7062–7081. [3] F. Jaouen, E. Proietti, M. Lefevre, R. Chenitz, J.-P. Dodelet, G. Wu, H.T. Chung, C.M. Johnston, P. Zelenay, Energy Environ. Sci. 4 (2011) 114–130. [4] C. Costentin, M. Robert, J.-M. Savéant, C. Tard, Acc. Chem. Res. 47 (2013) 271–280. [5] D.R. Weinberg, C.J. Gagliardi, J.F. Hull, C.F. Murphy, C.A. Kent, B.C. Westlake, A. Paul, D.H. Ess, D.G. McCafferty, T.J. Meyer, Chem. Rev. 112 (2012) 4016–4093. [6] J.J. Warren, T.A. Tronic, J.M. Mayer, Chem. Rev. 110 (2010) 6961–7001. [7] V. Sharma, G. Enkavi, I. Vattulainen, T. Róg, M. Wikström, Proc. Natl. Acad. Sci. U. S. A. 112 (2015) 2040–2045. [8] I. Bento, C. Silva, Z. Chen, L. Martins, P. Lindley, C. Soares, BMC Struct. Biol. 10 (2010) 28–42. [9] C.T. Carver, B.D. Matson, J.M. Mayer, J. Am. Chem. Soc. 134 (2012) 5444–5447. [10] R. McGuire Jr., D.K. Dogutan, T.S. Teets, J. Suntivich, Y. Shao-Horn, D.G. Nocera, Chem. Sci. 1 (2010) 411–414. [11] K. Mittra, S. Chatterjee, S. Samanta, A. Dey, Inorg. Chem. 52 (2013) 14317–14325. [12] K.D. Miner, A. Mukherjee, Y.-G. Gao, E.L. Null, I.D. Petrik, X. Zhao, N. Yeung, H. Robinson, Y. Lu, Angew. Chem. Int. Ed. 51 (2012) 5589–5592. [13] Y. Yu, C. Cui, X. Liu, I.D. Petrik, J. Wang, Y. Lu, J. Am. Chem. Soc. 137 (2015) 11570–11573. [14] R. Margalit, I. Pecht, H.B. Gray, J. Am. Chem. Soc. 105 (1983) 301–302. [15] C.-M. Che, H.-J. Chiang, R. Margalit, H.B. Gray, Catal. Lett. 1 (1988) 51–54. [16] R. Vazquez-Duhalt, J. Mol. Catal. B Enzym. 7 (1999) 241–249. [17] R.A. Scott, G.A. Mauk, Cytochrome c: A Multidisciplinary Approach, University Science Books, Sausalito, CA, 1995. [18] L. Banci, I. Bertini, K.L. Bren, H.B. Gray, P. Sompornpisut, P. Turano, Biochemistry 34 (1995) 11385–11398. [19] J.A. Fee, Y. Chen, T.R. Todaro, K.L. Bren, K.M. Patel, M.G. Hill, E. Gomez-Moran, T.M. Loehr, J. Ai, L. Thöny-Meyer, P.A. Williams, E. Sturam, V. Sridhar, D.E. Mcree, Protein Sci. 9 (2000) 2074–2084. [20] Y. Lu, D.R. Casimiro, K.L. Bren, J.H. Richards, H.B. Gray, Proc. Natl. Acad. Sci. U. S. A. 90 (1993) 11456–11459. [21] S. Casalini, G. Battistuzzi, M. Borsari, C.A. Bortolotti, A. Ranieri, M. Sola, J. Phys. Chem. B 112 (2008) 1555–1563. [22] S. Casalini, G. Battistuzzi, M. Borsari, A. Ranieri, M. Sola, J. Am. Chem. Soc. 130 (2008) 15099–15104. [23] R.K. Behera, H. Nakajima, J. Rajbongshi, Y. Watanabe, S. Mazumdar, Biochemistry 52 (2013) 1373–1384. [24] Y. Ichikawa, H. Nakajima, Y. Watanabe, ChemBioChem 7 (2006) 1582–1589. [25] H. Nakajima, Y. Ichikawa, Y. Satake, N. Takatani, S.K. Manna, J. Rajbongshi, S. Mazumdar, Y. Watanabe, ChemBioChem 9 (2008) 2954–2957. [26] H. Nakajima, K. Ramanathan, N. Kawaba, Y. Watanabe, Dalton Trans. 39 (2010) 3105–3114. [27] G. Battistuzzi, M. Bellei, C.A. Bortolotti, M. Sola, Arch. Biochem. Biophys. 500 (2010) 21–36. [28] S. Yamada, F. Bouley, N. D., G.E. Keller, W.C. Ford, H.B. Gray, J.R. Winkler, Proc. Natl. Acad. Sci. U. S. A. 110 (2013) 1606–1610. [29] J.D. Blakemore, N.D. Schley, D. Balcells, J.F. Hull, G.W. Olack, C.D. Incarvito, O. Eisenstein, G.W. Brudvig, R.H. Crabtree, J. Am. Chem. Soc. 132 (2010) 16017–16029. [30] J.D. Blakemore, A. Gupta, J.J. Warren, B.S. Brunschwig, H.B. Gray, J. Am. Chem. Soc. 135 (2013) 18288–18291. [31] H. Fujii, J. Am. Chem. Soc. 124 (2002) 5936–5937. [32] H. Fujii, T. Yoshida, Inorg. Chem. 45 (2006) 6816–6827. [33] S. Bowman, E. J., K. Bren, L, Inorg. Chem. 49 (2010) 7890–7897. [34] L.B. Vitello, M. Huang, J.E. Erman, Biochemistry 29 (1990) 4283–4288. [35] C.W. Conroy, P. Tyma, P.H. Daum, J.E. Erman, Biochim. Biophys. Acta 537 (1978) 62–69. [36] T. Yonetani, H. Anni, J. Biol. Chem. 262 (1987) 9547–9554. [37] G. Smulevich, J.M. Mauro, L.A. Fishel, A.M. English, J. Kraut, T.G. Spiro, Biochemistry 27 (1988) 5477–5485.
14
J. Husband et al. / Journal of Inorganic Biochemistry 157 (2016) 8–14
[38] S. Yoshikawa, D.H. O'Keeffe, W.S. Caughey, J. Biol. Chem. 260 (1985) 3518–3528. [39] A.D. Nugraheni, S. Nagao, S. Yanagisawa, T. Ogura, S. Hirota, J. Biol. Inorg. Chem. 18 (2013) 383–390. [40] S.L. Edwards, T.L. Poulos, J. Biol. Chem. 265 (1990) 2588–2595. [41] C.J. Wallace, I. Clark-Lewis, J. Biol. Chem. 267 (1992) 3852–3861. [42] H.J. Lee, R.B. Gennis, P. Ädelroth, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 17661–17666. [43] J.A. Villegas, A.G. Mauk, R. Vazquez-Duhalt, Chem. Biol. 7 (2000) 237–244. [44] J.N. Rodriguez-Lopez, J. Hernández-Ruiz, F. Garcia-Cánovas, R.N.F. Thorneley, M. Acosta, M.B. Arnao, J. Biol. Chem. 272 (1997) 5469–5476. [45] S. Lecomte, P. Hildebrandt, T. Soulimane, J. Phys. Chem. B 103 (1999) 10053–10064.
[46] E.V. Pletneva, H.B. Gray, J.R. Winkler, J. Mol. Biol. 345 (2005) 855–867. [47] S. Nagano, M. Tanaka, K. Ishimori, Y. Watanabe, I. Morishima, Biochemistry 35 (1996) 14251–14258. [48] L.B. Vitello, J.E. Erman, M.A. Miller, J. Wang, J. Kraut, Biochemistry 32 (1993) 9807–9818. [49] T. Matsui, S.-i. Ozaki, E. Liong, G.N. Phillips, Y. Watanabe, J. Biol. Chem. 274 (1999) 2838–2844. [50] S.M. Ibrahim, H. Nakajima, T. Ohta, K. Ramanathan, N. Takatani, Y. Naruta, Y. Watanabe, Biochemistry 50 (2011) 9826–9835.
