The Influence of pH on Long-Range Electron Transfer and Proton-Coupled Electron Transfer in Ruthenium-Modified Azurin

Abstract

Long-range electron transfer (ET) is an essential component of all biological systems. Reactions of metalloproteins are important in this context. Recent work on protein “charge ladders” has revealed how the redox state of embedded metal ions can influence the ionization of amino acid residues at protein surface sites. Inspired by these observations, we carried out a variable pH investigation of intramolecular ET reactions in a ruthenium-modified protein system built on azurin from Pseudomonas aeruginosa. We also generate a Pourbaix diagram that describes the variable pH redox behavior of a Ru model complex, Ru(2,2′-bipyridyl)₂(imidazole)₂(PF₆)₂. The intramolecular ET rate constants for the oxidation of azurin-Cu⁺ by flash-quench-generated Ru³⁺ oxidants do not follow the predictions of the semi-classical ET rate expression with fixed values of reorganization energy (λ) and electronic coupling (H_DA). Based on the pH dependence of the Ru^3+/2+ redox couple, we propose a model where pure ET is operative at acidic pH values (≤ 7) and the mechanism changes to proton-coupled electron transfer at pH ≥ 7.5. The implications of this mechanistic proposal are discussed in the context of biological redox reactions and with respect to other examples of intramolecular ET reactions in the literature.

1. Introduction

The workings of biological redox systems involving electron flow through and between proteins presents an intricate puzzle to scientists [1,2,3,4,5,6]. Biological electron transfer (ET) processes have a large number of variables that can influence the function and the malfunction of reactions that are crucial for life, as discussed in the above (and many other) references. One key variable is protonation, or proton transfer (PT), which is another of the most fundamental chemical reactions. Such reactions where ET and PT occur together (simultaneously or sequentially) have been termed “proton-coupled electron transfer” (PCET) [7,8,9,10,11]. In the end, an ongoing challenge to understanding biological redox chemistry is to understand how protons influence the rates, energetics, and selectivity of redox reactions.

Recent research has highlighted the importance of the local environment of cofactors in modulating the properties of redox cofactors and their ET reaction. One earlier example is the hydrogen-bonded tyrosine_Z in photosystem II, which facilitates reversible PCET in the net water oxidation reaction [12,13,14]. In work on Pseudomonas aeruginosa azurin, the proximity of charged amino acid residues near a redox-active tyrosine also influences its thermodynamic properties [15,16]. Such observations also have been made in peptide-based models for the study of redox-active amino acid residues [17,18,19,20,21]. With respect to the embedded azurin-Cu ion, second sphere (i.e., non-coordinated) amino acid residues also can exert influence on the redox properties of the Cu site [22]. In general, the above examples can be classified as short-range effects, but comparatively less is known about the long-range influence of protein structures and amino acid composition on embedded sites. In this work, we set out to address some of these questions, in particular with respect to pH and ionization of surface sites.

The identity and composition of protein surface sites are known to influence the properties of embedded cofactors. For example, a series of azurin variants show how hydrophobic amino acid residues can systematically affect Cu^2+/1+ reduction potentials [23]. While it may not be surprising that amino acid variations influence embedded cofactors (even at a distance), there are numerous examples of pH-dependent reduction potentials, especially in metalloproteins. A few examples include azurin [24], ferredoxins [25], Rieske proteins [26], b-type cytochromes [27,28], and c-type cytochromes [29,30,31]. That list is far from comprehensive, and one can reasonably assume that nearly all redox proteins will have redox properties that depend on pH. In azurin, for example, it has been proposed that the ionization state of (mainly) surface amino acid residues influences the properties of the cofactor [24,32].

Recent work from Shaw and co-workers demonstrated that a change in redox state of an embedded cofactor influences the ionization state of surface amino acid residues [33,34,35]. Such behaviors are called “charge regulation”, wherein the unit change in charge of the embedded cofactor is compensated by changes in the pK_a of other amino acid residues [36,37,38,39]. In this way, most redox proteins are better described as PCET proteins. Stated more explicitly, redox cycling of metal cofactors can involve proton transfer reactions at protein surfaces, including sites that are distant from the metal ion. In addition to thermodynamics considerations, PCET reactions can have different associated reorganization energy (cf. examples from theoretical treatments of PCET in ref. [40]). In short, the coupling of ET reactions and proton transfer reactions can have a wide range of effects on the rate(s) of biological redox chemistry.

Azurin is one of the most studied models of biological ET reactions. As noted above, its pH-dependent redox chemistry has been described [24], its charge regulation theoretically treated [33], as well as experimentally probed [32]. Residues His35 and His83 were identified as sites that change their pK_a upon a redox change in the Cu ion. That work is in good agreement with computation work from Ullmann and co-corkers, who explored the protonation states of the two histidine sites [32]. While the discrete order of redox and protonation reactions was not determined, the two sites are tightly coupled. Herein, we use the known Ru-His83-modified azurin variant (Figure 1) to explore the pH dependence of intramolecular ET. We employ the well-known [41] Ru(bpy)₂(imidazole)^3+/2+ system to initiate intramolecular ET reactions (bpy = 2,2′-bipyridyl).

2. Results

The pH-dependent properties of azurin and of the model complex Ru(bpy)₂(im)₂²⁺ were first investigated. The azurin Cu^2+/1+ redox couple has a known pH dependence, but in a different buffer system [24]. Although the buffer system is likely to be only a minor contribution, we also collected data using an APB (acetate–phosphate–borate) buffer to ensure that a different buffer did not influence the potentials. While we did not undertake studies at the same level of detail as that of the past report, the measured potentials at pH values between 4 and 8 were in accord with that of a previous work [24]. The potentials for the other pH values indicated below are calculated using the model set out in reference [24]. Additional details are set out in the Supporting Information.

Next, the pH-dependent electrochemistry of the Ru(bpy)₂(im)₂^3+/2+ couple was investigated using cyclic voltammetry (Figure 2). Cobalt(II)tris(2,2′-bipyridyl)(PF₆)₂ was used as a pH independent internal standard (E = 0.30 V versus SHE [42]). The pH-dependent photophysical properties of the Ru²⁺ complexes have been described at pH values between 10 and 14 [43], but not the pH dependence at lower pH values. To the best of our knowledge, the pH-dependent reduction potentials are not reported, though the Ru^3+/2+ potential at pH 3 is known [44]. That value (1.00 V versus SHE) is in agreement with our measured value between pH 4 and 7 (E = 1.014 ± 0.005 V). At pH values higher than 7.5, the Ru^3+/2+ decreases in potential by 55 mV per decade, which is close to the value expected for a 1H⁺/1e⁻ process (59 mV per pH). From these data, the pK_a of the Ru³⁺ complex is 7.6 ± 0.1, and the reported pK_a for the corresponding Ru²⁺ complex is 11.9 ± 0.2 [43].

The modification and purification of azurin at His83 with a [Ru(bpy)₂(im)]²⁺ label was carried out according to the literature using Ru(bpy)₂CO₃ for initial labeling and incubation in imidazole-containing buffer to fill the remaining coordination site [41]. The efficiency is about 80%, following purification. Successful modification was evinced by characteristic UV-Vis absorption peaks at 628, 490, and 350 nm, which are characteristic of the azurin Cu site and for the Ru complex. The spectrum of the Ru-labeled protein is shown in the Supporting Information (Figure S2). Intramolecular ET rates were determined using a home-built transient absorption spectrometer [45,46]. To induce excitation, the samples were exposed using a wavelength of 480 nm. This laser light was generated via an optical parametric oscillator (OPO) laser pumped with frequency-tripled (355 nm) light from a neodymium-doped yttrium aluminum garnet (Nd-YAG) laser. This experimental arrangement involved the utilization of a 10 mM concentration of [Ru(NH₃)₆]³⁺ as a quencher for reversible redox reactions.

Samples with varying pH values were probed using time-resolved transient absorbance spectroscopy to investigate the influence of pH on intramolecular ET kinetics. A typical flash-quench transient experimental set-up was used (see Supporting Information, Figure S3) with Ru(NH₃)₆Cl₃ as a reversible ET quencher [47,48]. The kinetic traces were analyzed as described previously [15,16]. A representative kinetics trace in shown in Figure 3 and all kinetics traces are shown in the Supporting Information (Figure S4). The obtained ET rate constants are set out in

Table 1. Summary of experimental driving forces and ET rate constants and the corresponding calculated ET rate constants.

pH	−∆G° 1	kET (×10−6) 2	kET,calc (×10−6) 3
4	−0.671	1.2 ± 0.1	2.18
5	−0.671	1.4 ± 0.1	2.18
6	−0.673	1.3 ± 0.1	2.18
7	−0.698	1.3 ± 0.1	2.20
8	−0.693	1.3 ± 0.1	2.20
9	−0.655	0.3 ± 0.1	2.14
9.5	−0.634	0.4 ± 0.1	2.07

¹ Errors in −∆G° values are ±0.01 V. ² Calculated values of k_ET use Equation (1) and fixed values of λ (0.7 eV), the Ru-Cu distance (17 Å), H_AB at close contact is 186 cm⁻¹, and the distance decay constant (β = 1.1 Å⁻¹). ³ Errors in k_ET,calc are approximately ±2 × 10⁴ s⁻¹ based on errors in −∆G°.

. The driving forces (−∆G°) for the reactions at different pH values were computed based on the electrochemical data from above, and these values also are shown in Table 1.

3. Discussion

The starting point for rationalizing long-range intramolecular ET reactions is the semiclassical rate expression (Equation (1)). Here, ∆G° is the reaction driving force (given by the difference in formal reduction potentials), λ is the reorganization energy, H_AB is the electronic coupling matrix element, and the other physical constants have their usual meanings. The value of λ for azurin is 0.7 to 0.8 eV [47,49,50,51], and here we use 0.7 eV, but it should be noted that our conclusions also hold if 0.8 eV is used. For H_AB, the value at close contact (3 Å) is 186 cm⁻¹, and the value at large distances (r) is determined using a square barrier model: H_AB = (186 cm⁻¹)·exp (0.5 βr). The value of the distance decay constant β is an empirical value for proteins (1.1 Å⁻¹) [48,52].

$$k_{\mathrm{E}\mathrm{T}}=\sqrt{{\displaystyle \frac{4{\pi }^3}{h^2\lambda RT}}}{H_{AB}}^{2}exp\left(-{\displaystyle \frac{{(\Delta \mathrm{G}°+\mathsf{\lambda })}^2}{4\lambda RT}}\right)$$

(1)

Using the above values as constants, and the experimentally determined driving forces (−∆G°), it is straightforward to compute rate constants for ET (k_ET,calc) in RuHis83 azurin at each pH value (Table 1). We also provide corresponding values with a different value of λ in Table S2 (see Supporting Information). The agreement between experiment and theory is good at pH values between 4 and 8, but less so at pH 9 and 9.5. The calculated rate constants show that, in principle, ET should be fastest at pH 7, but this is not possible to confirm within experimental uncertainty. At pH values of 9 and above, the rate constants are predicted to be smaller in magnitude, but the discrepancy between experiment and theory is larger than that for the other pH values. Unfortunately, the stability of RuHis83-azurin at pH values about 9.5 is decreased, especially during irradiation in time-resolved spectroscopy experiments.

The above ET analysis (i.e., values in Table 1) does not account for all aspects of the pH dependence of the azurin protein and the Ru label; only the driving force (−∆G°) is include in the calculations of k_ET. The azurin Cu^2+/+ potential is pH-dependent between about pH 5.5 and 8, and the Ru potential is pH-dependent between about pH 7.5 and pH 11.9 (see above). The pH dependence of azurin is not described by a simple 1H⁺/1e⁻ Nernstian model, indicating that multiple sites are involved with the pH dependence [24]. This is corroborated by the work on charge regulation in azurin and in other proteins [33,34]. Overall, the change in the pK_a of surface residues was shown to correlate with the total reorganization energy (likely outer-sphere) in several metalloproteins. With respect to the RuHis83-azurin studied here, it seems that any of the changes in driving force and reorganization energy are small in order to produce proportionally small changes in observed and calculated ET rate constants.

The lack of pH dependence on Cu⁺ oxidation lower than pH 8 for Ru-His83-azurin is worth brief comments. Ullmann and co-workers have explored, in detail, the underpinnings of the pH-dependent reduction of Cu⁺ in azurin, using experimental data as a benchmark [32]. In particular, there exists a high probability of His83 and His35 being protonated upon reduction at pH values between 6 and 8. In Ru-His83-azurin, the binding of the Ru-label at His83 blocks that site from protonation, but His35 remains accessible. In Ullman’s work, they demonstrate that protonation of His35 upon Cu⁺ reduction is coupled to a peptide flip. In this context, one could expect more pH dependence; however, we lack more detailed mechanistic information. The lack of pH dependence in Ru-His83-azurin might suggest that proton transfer or peptide flip occurs after Cu²⁺ oxdidation, but more work is clearly needed to make firm conclusions.

As noted above, the discrepancy between experiment and theory is more pronounced at pH 9 and pH 9.5. At these pH values, the reduction potentials for azurin are pH-independent, so it must be the redox behavior of the Ru label that contributes to the observed change in k_ET. There are two limiting possibilities. Both possibilities involve the oxidation of Cu⁺ by a deprotonated Ru³⁺, that is, an imidazolate- or histidinate-ligated Ru³⁺ (Figure 4). The pK_a values of oxidized complexes are always lower (more acidic) than the corresponding reduced complexes [53,54]. Thus, it can be expected that, following the flash quench reaction, the Ru³⁺ label will rapidly deprotonate. The first possibility is that Cu⁺ is oxidized by Ru³⁺-imidazolate to form Cu²⁺ and a Ru²⁺ imidazolate which is subsequently re-protonated. The value for E° (Ru³⁺-imidazolate/Ru²⁺-imidazolate) can be estimated as 0.47 V versus SHE using our data and literature pK_a values [43]. The calculated value of k_ET is between 5.2 × 10⁵ and 1.0 × 10⁶ s⁻¹ for values of λ between 0.7 and 0.8 eV. The calculated rate constants are in good agreement with experimental work.

The second option is that the reduction of Ru³⁺-imidazolate occurs via concerted addition of H⁺ and e⁻ in one step. This has been called, equivalently, “concerted proton-electron transfer” or “electron proton transfer” [54,55]. Such reactions are known to have larger associated values of λ [56,57]. Thus, it is straightforward to use Equation (1) to calculate pH-dependent λ values with ET rate constants and driving forces fixed at our experimental values. These values are set out in

Table 2. Calculated reorganization energy (λ).

pH	λ 1
4	0.88
5	0.85
6	0.87
7	0.89
8	0.85
9	1.10
9.5	1.03

¹ Values are computed using Equation (1) with rate constants and driving forces fixed to the experimental values in Table 1.

. The increase in λ is reasonable based on other literature examples [56,57], especially one involving ET and PCET of a ruthenium diimine complex [57].

To the best of our knowledge, there is only one other example of the pH dependence of intramolecular ET in azurin [58]. In broad terms, those authors observe the same general trend, with larger ET rate constants at low pH values. However, the magnitude of the change in k_ET is larger (over a factor of 10). We emphasize that those authors use a different system in which the kinetics of Cu²⁺ reduction by disulfide radicals are monitored. The authors discuss the pH dependence of the Cu^2+/+ couple in azurin, but any pH dependence of disulfide redox reactions is not considered explicitly. As is the case for our Ru-modified system, the rate constants cannot be satisfactorily modeled using Equation (1) with fixed values of H_AB and/or λ. Ultimately, the authors propose small changes to the degree of electronic coupling (H_AB).

Taking the observations from reference [58] in account, as well as our results from Ru-His83 azurin, we propose that the pathways shown Figure 4 are the likely origin of the changes in ET rate constant as a function of pH in both modified and unmodified azurins. A change in mechanism to PCET with larger associated values of λ, or significant changes in reduction potentials due to deprotonation, give rise to similar decreases in k_ET as pH is increased. With respect to λ, and as is expected from the Marcus additivity postulate, if the reorganization energy of one redox site changes, the value of λ for the entire system changes and thus the observed rate constant.

The results from this study underscore the importance of PCET in biological redox systems. The concept of charge regulation [33,34] provides a means to rationalize pH-dependent reduction potentials and outer-sphere reorganization energy. By spreading the changes to surface pK_a across several weakly coupled sites, biomolecules can have a less pronounced (non-Nernstian) pH dependence. This also maintains the naturally low reorganization energy of biological metal sites. In the RU-modified azurin studied here, the pH dependence of the Ru site has a strong pH dependence and therefore a larger impact of ET rate constants. At near-neutral pH values, any change in mechanism is compensated by an increase in driving force (Table 1), but at high pH Cu⁺, oxidation likely occurs via a different mechanism, which is reflected in the observed rate constants. Ultimately, it seems that biomolecules have evolved coupled pathways that control both electron and proton transfers, even in “simple” electron carriers like azurin.

4. Materials and Methods

All chemicals were from Sigma-Aldrich and used as received, unless otherwise noted. Ru(2,2′-bipyridyl)₂Cl₂ was from Strem Chemicals. The complexes Co(2,2′-bipyridyl)₃(PF₆)₂ and Ru(2,2′-bipyridyl)₂CO₃ were prepared according to the literature [59,60]. Ru(NH₃)₆Cl₃ was recrystallized before use [61]. All buffer salts were from Fisher Scientific and used as received. BL21-DE3 E. coli were from Thermo-Fisher Scientific. All media for purification of protein samples was from GE Healthcare (Cytiva). Protein samples were concentrated using an Amicon ultrafiltration cell under N₂ positive pressure with Sartorius Hydrosart 10K ultrafiltration membranes. All experiments utilized purified water purified to a resistivity of 18 MΩ cm⁻¹ using a Barnstead EASYpure system. The APB buffer used in all variable pH experiments comprised 20 mM each of sodium acetate, dibasic potassium phosphate, and sodium borate, as well as 100 mM potassium chloride. The pH of each solution was adjusted using 1 M HCl or 1 M KOH, as appropriate.

Optical spectra were collected using a Cary 100 Bio UV-Visible spectrophotometer. Electrochemical experiments were carried out using a standard three-electrode setup with an edge plane graphite working electrode, a glassy carbon counter electrode, and a silver/silver chloride reference electrode. Cobalt(III) tris(2,2′-pipyridyl) was used as an internal standard. Cyclic voltammetry data were collected using a CH Instruments CHI6171B Electrochemical Analyzer.

The plasmid for expression of wild-type Pseudomonas aeruginosa azurin was a gift from Harry B. Gray and John H. Richards (California Institute of Technology). Azurin was expressed and purified according to the literature. modification with Ru(bpy)₂ at His83 was carried out as previously described. Final protein purification used a Pharmacia fast protein liquid chromatography system. Azurin concentrations were assessed using UV-vis spectroscopy and the known molar absorptivity of the azurin charge transfer band ε (628 nm) = 5600 M⁻¹ cm⁻¹.

The model complex Ru(bpy)₂(im)₂(PF₆) was synthesized following a literature procedure (bpy = 2,2′-bipyridiyl). To facilitate dissolution of the PF₆⁻ salts for electrochemistry in water, concentrated DMSO stocks were prepared and diluted to a final concentration of 0.1 mM in APB buffers at each pH value. Azurin samples also were assayed at 0.1 mM concentrations. Cyclic voltammograms were collected using scan rates of 100 mV s⁻¹. The working electrode was polished between each scan with 0.3 µm alumina micropolish powder and a microfiber lab (Buheler). For measurements of azurin, potassium ferricyanide (K₃Fe(CN)₆) was employed as an external standard and measurements of Ru(bpy)₂(im)₂(PF₆)₂ used Co(bpy)₃(PF₆) as an internal standard.

Transient absorption experiments were conducted using a home-built spectrometer that has been described elsewhere. The excitation of the samples was achieved using a 480 nm laser beam with an energy of 6 ± 1 mJ per pulse. Protein samples were reduced with ascorbic acid, desalted into the appropriate pH buffer and deoxygenated with a 15–20 pump-backfill cycles with N₂ gas. Initial luminescence decay traces were collected prior to adding Ru(NH₃)₆Cl₃ to a final concentration of 12–18 mM; the concentration did not affect the observed ET rate constants. Samples were again deoxygenated following the addition of quencher and time-resolved luminescence, and absorbance data were collected. The resulting kinetics traces were analyzed using the MATLAB (version R2023b) Curve Fitting toolbox.