Direct electrochemistry of redox proteins

The goal of the project was to obtain more detailed insight in interactions between redox proteins and solid electrodes and the mechanisms of electron transfer. In addition to this, the influence of the protein environment on the redox properties of the active site and the possible influence of the electrode/promoter system on these properties have been considered. Because redox enzymes do not often give an unambiguous and reversible electrochemical response (if at all), electron transferring proteins have been studied. The FMN containing flavodoxins have been used as model systems for flavin enzymes such as glucose oxidase. A series of high potential iron-sulfur proteins (HiPIPs) can be regarded as models for proteins containing [4Fe-4S] clusters. The HiPlPs are also of interest because of the high oxidation state of the cluster; the sequences are known, and the three-dimensional structures of some HiPIPs are known.In Chapter 2 the electrochemical behaviour of flavodoxin from Desulfovibrio vulgaris (Hildenborough) has been characterized by staircase cyclic voltammetry (SCV) and differential pulse voltammetry (DPV). Fully oxidized flavodoxin at the bare glassy carbon electrode gave one redox couple at a potential of -218 mV (NHE) at pH=7.0 with an SCV peak current proportional to the scan rate. This response is caused by FMN, dissociated from the protein and adsorbed onto the electrode. The midpoint potential and the p K of 6.5 are equal to the values measured with free FMN in solution. When the cationic promoter neomycin was added one additional and diffusion controlled response was observed. This positively charged aminoglycoside is believed to form a flexible bridge between the negative charges on the surface of both the protein and the electrode. The midpoint potential of the observed redox couple is -413 mV (NHE) at pH 7.0 with a redox-linked pK for the reduced form of 4.8. The temperature dependence is -1.86 mV/K, yielding ΔS°=-179 J.mol -1.K -1and ΔH°=-12.4 kJ/mol. This response is believed to be the semiquinone/hydroquinone transition. Although the starting material was 100% quinone, no response was observed around the midpoint potential of the quinone to semiquinone reduction of -113 mV (NHE) at pH 7.0, determined in an EPR-monitored titration with dithionite. Digital simulation shows that the peak currents of the second reduction couple approach a maximum value after a few cycles if comproportionation of fully reduced and fully oxidized flavodoxin occurs in solution and a small amount of semiquinone is either present initially or is generated by mediation of electrode-bound FMN. In the latter case the observed increase of the peak height can be fitted with a Butler-Volmer type heterogeneous electron transfer rate between adsorbed FMN and flavodoxin of 6.3-10 -6m/s. This anomalous behaviour might have implications for the interpretation of electrochemistry on flavin enzymes like glucose oxidase. The observed peak is not per se the expected protein response or the expected (first) reduction step. The development of a catalytic current when substrate is added is no prove for direct interaction between the protein and the electrode, but can also be accomplished by electron transfer mediated by a small amount of free flavin.In Chapter 3 the detailed electrochemistry and complete EPR-monitored titrations of flavodoxin II of Azotobacter vinelandii (ATCC 478) are reported. Since wild-type flavodoxin dimerizes via disulphide bond formation between cysteine 69 residues, Cys69 has been replaced by an alanine as well as a serine residue. Redox properties of the C69A and C69S flavodoxin mutants were compared to those of wild-type flavodoxin. In the presence of the promoter neomycin, C69A and C69S flavodoxin showed a reversible response of the semiquinone/hydroquinone couple at the glassy carbon electrode, similar to the observations with D. vulgaris flavodoxin. However, addition of dithiotreitol proved to be necessary for the stabilization of the wild-type flavodoxin response. Dithiotreitol probably prevents dimerization of the protein by formation of cystine bridges. EPR- monitored redox titrations of wild-type and C69A flavodoxin at high and low pH confirm the cyclic voltammetry data. The pH dependence of the semiquinone/hydroquinone redox potentials cannot be described with a simple one-p Kred model. Instead, the presence of at least two redox-linked protonation sites is suggested: p Kred,1 = 5.39 ± 0.08, p Kox= 7 .29 ± 0.14 and p Kred,2 = 7.84 ± 0.14 with Em7 = -459 ± 4 mV and a constant potential at high pH of -485 ± 4 mV. The dependence of the semiquinone/hydroquinone potential on temperature is -0.52 ± 0.06 mV/K, yielding Δ H ° = 28.6 ± 1.5 kJ/mol. and ΔS° = -50 ± 6.2 J.mol -1.K -1. No significant differences in redox properties of wild- type, C69A and C69S flavodoxin were observed. The electrochemical data suggest that replacement of Cys69 in the vicinity of FMN by either an alanine or a serine residue does not have a measurable influence on the structure of the protein.In Chapter 4 a theoretical solution of the concentration distribution in long optical path length thin layer spectroelectrochernical cells is given by a convergent infinite summation of terms. At short times a large number of terms is required to obtain a good approximation. Alternatively, on a timescale at which the boundary of the diffusion layer has not reached the cell wall opposite to the electrode the concentration profile of the thin layer cell is equal to the profile of a semi-infinite bulk electrochemical cell. This profile is described by an error function, for which no analytical solution is available. A new three-parameter exponential approximation for this error function is presented with an accuracy better than 0.05% for all positive values of x . When the diffusion layer boundary reaches the cell wall the semi-infinite bulk model is no longer valid but then the slope of the profile has become small enough to be approximated by only the first five terms of the summation. When the composition of the bulk solution is measured by a light beam passing at grazing incidence over the electrode surface, the absorbance can be calculated from the concentration distribution by integration of the transmittance perpendicular to the light beam.In Chapter 5 the validity of the mechanism proposed in Chapter 2 for the flavodoxin response has been verified by measuring the absorbance of the semiquinone form of D. vulgaris flavodoxin during cyclic voltammetry. A long optical path length thin layer electrochemical cell (LOPTLC) was used with a layer width of 0.2 mm. Despite the non-ideal behaviour of this cell, the resulting "cyclic voltabsorptomograms" clearly show the proposed formation of semiquinone by FMN-mediated electron transfer, and comproportionation of flavodoxin in solution occurs. Simulated voltabsorptomograms qualitatively confirm this, although the observed reoxidation of flavodoxin semiquinone at low scan rates is not predicted by the Butler-Volmer model of Chapter 2.In Chapter 6 the High Potential Iron-Sulfur Proteins (HiPIPs) from Ectothiorhodospira vacuolata (iso-1 and iso-2), Chromatium vinosum , Rhodocyclus gelatinosus , Rhodocyclus tenuis (strain 2761), Rhodopila globiformis and the large (multimer) HiPIP (iso-2) from Rhodospirillum salinarum have been investigated by direct electrochemistry. Using a glassy carbon electrode with a negatively charged surface, direct, unpromoted electrochemistry was possible with the positively charged HiPlPs. With the negatively charged HiPIPs the positively charged and flexible bridging promoter poly-L-lysine was required. The stability of the response could be improved by morpholin in combination with the negatively charged proteins and by monomeric amino acids or 4,4'-dipyridyl with the positively charged HiPlPs. These 'stabilizers' apparently prevent the blocking of the electrode by denatured protein during electrochemistry. The redox potential of 500 mV found for the large HiPIP from R. salinarum is the highest HiPIP potential reported. The presence of histidines in the sequence does not per se predict a pH-dependent redox potential. Only C. vinosum and R. gelatinosus HiPIPs show a weak but significant pH dependence with a difference of 35 mV between the low and the high pH form and maximum slopes of about -20 mV/unit. Either the coupling of electron and proton transfer is indirect ('allosteric') or p Kox is only 0.6 units lower than p Kred . In the latter case an apparent dielectric constant of 48 can be calculated. The dependence of the midpoint potential on ionic strength cannot be explained by the Debye-Mickel theory alone because the linearity exceeds the limiting concentration, the slopes are much smaller than predicted by this theory (0 to -28 mV/vM) and no positive slopes are observed. Combination of the sequences, the optical spectra, the overall charges and the redox thermodynamics suggests the existence of two major groups of HiPlPs. One group consists of Chromatium -like HiPIPs with redox potentials between 300 and 350 mV, modulated only by the solvation of the cluster but not by the overall charge of the protein. The second group is formed by Ectothiorhodospira -like HiPIPs with potentials between 50 and 500 mV, largely dependent on the overall charge of the peptide and also modulated by cluster solvation. From the slope of 25 mV per unit charge an apparent dielectric constant of 84 is calculated.In Chapter 7 the reversible 2 x 1 e-reduction of the cubane cluster from oxidized to reduced to super-reduced: [4Fe-4S] 3+= [4Fe_4S] 2+= [4Fe-4S] l+has been studied in the HiPlPs of Chapter 6. Super-reduction to the 1+ state was not observed in any of these seven HiPlPs tested during cyclic voltammetry (down to -0.95 Volt). However, equilibration at low potential (pH 7.5) of Rhodopila globiformis HiPIP yielded a transient peak around -0.47 V due to the oxidation of super-reduced HiPIP adsorbed at the electrode. The peak area depends on the equilibration potential according to a one-electron Nernst curve with a half-wave potential at - 0.91 V. Reduction of R.globiformis HiPIP with titanium(III)citrate at pH 9.5 is very slow (pseudo first-order half-life of 23 min. with hundred-fold excess Ti(III)) but is reversible, and the EPR spectrum with g -values of 2.04 and 1.92 is similar to that of reduced [4Fe-4S] l+ferredoxins. Chemical or electrochemical reoxidation of the super-reduced form resulted in an EPR spectrum with g ¦¦ = 2.12 and g- = 2.03, i.e. identical to that of oxidized HiPIP. From the equilibrium concentration of super-reduced HiPIP at low concentration of Ti(III) a reduction potential of -0.64 V can be estimated. Super-reduction of the large HiPIP (iso-2) from Rhodospirillum salinarum is also possible with Ti(III) ( gz = 2.05) but the superreduced state is unstable. No super-reduction with Ti(III) was observed for the other HiPIPs. The difference between the electrochemically observed reduction potential and oxidation potential is explained by a fast and reversible conformational change upon super-reduction. The rate of super-reduction with Ti(III) is limited by the small amount (0.1 %) of the HiPIP in the 2+ state with the super-reduced conformation.It can be concluded that the interaction of redox enzymes with the glassy carbon electrode is determined primarily by the charge of the protein and of the electrode surface. With positively charged proteins no promoter is required to obtain direct electron transfer at the negatively charged electrode surface. A positively charged promoter must be added to obtain a response with negatively charged proteins. A flexible promoter that can adjust its shape to fit both the protein surface and the electrode surface gives the best results. However, the response usually deteriorates in time. This is probably caused by denaturing of the protein on the electrode surface. Van der Waals forces and hydrophobic interactions probably play an important role in this process. The electrode thereby becomes gradually less accessible for electron transfer. A good promoter therefore not only forms a flexible bridge that compensates the electrostatic repulsion, but must also protect the protein from the hydrophobic patches on the electrode. Neomycin apparently has this double- function as promoter for flavodoxins. However, the HiPIP studies show that although poly-L-lysine promotes the response of the negatively charged proteins, the response is not stable. A separate "stabilizer" can be added to improve the voltammetry of both positively and negatively charged HiPlPs. The wild-type Azotobacter vinelandii flavodoxin 11 is a special case where dithiotreitol stabilizes the response by preventing the formation of dimers in which the FMN is no longer accessible to the electrode.The redox potential of a protein is not always influenced by the charges of the peptide. This is true for both the permanent charges and for the pH-dependent charges. The HiPIP studies show that the distance between the charge and the redox-center, the dielectrics of the peptide in between, and the exposure of the charged groups to water are important factors. The dependence of the redox potentials on pH, measured with electrochemistry, do not always agree with the results of bulk-titrations. This can just be an indication that the electrochemical measurements have a much better accuracy, but some influence of the electrode surface and/or the promoter on the protein structure cannot be excluded. It is therefore important to compare the electrochemical data with the results of independent spectroscopic redox-titrations. Simultaneous spectroscopic measurements of the solution during the voltammetric experiments can give useful additional information for the deconvolution of coupled homogeneous reactions.