Redox Properties of Tyrosine and Related Molecules

Abstract

Redox reactions of tyrosine play key roles in many biological processes, including water oxidation and DNA synthesis. We first review the redox properties of tyrosine (and other phenols) in small molecules and related polypeptides, then report work on (H20)/(Y48)-modified Pseudomonas aeruginosa azurin. The crystal structure of this protein (1.18 Å resolution) shows that H20 is strongly hydrogen bonded to Y48 (2.7–2.8 Å tyrosine-O to histidine-N distance). A firm conclusion is that proper tuning of the tyrosine potential by a proton-accepting base is critical for biological redox functions.

1. Introduction

Nature employs many different redox cofactors that promote long-range electron transfer [1]. Few cofactors are available, however, that can carry out redox chemistry at potentials near or above 1 V versus NHE. Charge transport at these high potentials is required for key biological transformations such as water oxidation and C-H bond activation. Heme iron, copper, and iron sulfur clusters are common redox “way stations” but they are typically limited to low potentials (less than about 400 mV versus NHE) [2]. Several amino acids have been implicated in high-potential redox catalysis, but the main players are tyrosine (YOH) and tryptophan (W).

Interest in enzymes that use tyrosyl radicals (YO^•) in redox catalysis has spurred investigations of a wealth of small molecule and protein-based models. Natural enzymes that use the YOH/YO^• redox couple include photosystem II [3], ribonucleotide reductases [4] cytochrome c oxidase and related oxygen reductases [5], galactose oxidase [6], prostaglandin synthase [7], and a fatty acid oxygenase [8]. Furthermore, YOH oxidation products are implicated in disease states related to oxidative stress [9]. While we know that many enzymes use YOH in catalysis or long-range electron transfer (ET), we know relatively little about its actual properties within a given enzymatic system. This review focuses on model investigations of tyrosine oxidations in chemistry and biology, including a preliminary report of our work on designing proteins containing redox active units that mimic those found in Nature.

2. Materials and methods

2.1 Protein isolation and purification

The plasmid encoding for H20/Y48 azurin was generated using the Stratagene Quikchange protocol. Protein was expressed as previously described [10]. It should be noted that H20/Y48 azurin is sensitive to pH, precipitating when the pH is below 7.5. Thus, protein was isolated as previously described, except solutions were kept at pH 8.5. Protein was exchanged into 10 mM diethanolamine (DEA)•HCl, pH 9.5 and purified on Q-Sepharose using a NaCl gradient. Protein purity was confirmed by SDS-PAGE and electrospray ionization mass spectrometry.

2.2 Crystal growth, manipulation and x-ray data collection

Cu^II azurin (~20 mg/mL) in 10 mM DEA•HCl was mixed with an equivalent volume of well solution containing 25–30% of poly(ethylene glycol) (PEG) 4000, 100 mM lithium nitrate, 10 mM copper sulfate and 100 mM Tris•HCl, pH 8. Crystals suitable for x-ray diffraction were grown using sitting drop vapor diffusion [10]. The drops were allowed to equilibrate against 0.25 mL of well solution over 5 days, by which time plate-like crystals of H20Y48 azurin were obtained. Diffraction data were collected at the Stanford Synchrotron Radiation Laboratory (SSRL) beamline 12-2. The structure was solved by molecular replacement and then refined to the resolution limit from scaling/merging statistics. The coordinates of the structure (1.18 Å resolution) have been deposited in the Protein Data Bank (PDB ID 3U25) (Table[1]).

3. Discussion

The redox chemistry of YOH is inherently “proton-coupled,” meaning that changes in electron inventory are accompanied by changes in proton content. These H⁺/e⁻ transfers are collectively called “proton-coupled electron transfer (PCET).” PCET has come to be known by many different names [11] and full treatment of this nomenclature is beyond the scope of this review. The typical starting point for understanding ET is semiclassical theory (eqn 1) [12] which describes the dependence of intramolecular rate constants on three parameters: the reaction driving force (−ΔG°); the reorganization energy (λ); and the electronic coupling between reactant and products at the transition state nuclear configuration (H_AB). Importantly, we are still seeking understanding of key parameters in eqn 1, −ΔG° and λ, most especially for reactions within and between polypeptides.

$$k_{\text{ET}}=\sqrt{\frac{4{\pi }^3}{h^2\lambda k_{\text{B}}T}}{H_{\text{AB}}}^{2}exp\left(\frac{-{(\mathrm{\Delta }G°+\lambda )}^2}{4\lambda k_{\text{B}}T}\right)$$ (1)

The zero-order Born-Oppenheimer approximation that decouples electron and nuclear motions in semiclassical ET theory (eqn. 1) is absent (along at least one coordinate) in PCET reactions. Marcus noted that eq. 1 results when the potential energy along the reaction coordinate is parabolic, as with vibrational or pseudo-vibrational motion. For perspectives on the applicability of semiclasical theory to atom transfers see references [13] and [14]. When bonds are made or broken along the reaction coordinate, simple Born-Oppenheimer separability is lost, the potential energy cannot be described by intersecting parabolic surfaces, and eqn. 1 need not apply. Theoretical models for PCET reactions have been proposed in which rate constants depend on an intrinsic-barrier parameter that resembles the reorganization parameter (λ) of eqn. 1 [15]. Experimentalists frequently employ models of this type in attempts to understand the driving force dependence of PCET reactions. Since the ad hoc parameters extracted in these studies are not the same as ET λ values (eqn. 1), we suggest that a different symbol (γ) should be used to denote the intrinsic barriers associated with PCET reactions, including biologically important cases involving tyrosine and related phenols. We anticipate that continued experimental and theoretical investigations of PCET processes will produce a conceptual framework that does not violate any of the well-established principles of semiclassical ET theory.

We begin our discussion with an introduction to the solution chemistry of phenols and tyrosine. We then consider the mechanisms that have been proposed for reactions involving small molecule models. Next, we turn to investigations of YOH oxidation/reduction in polypeptides. Finally, we discuss work on designed model systems that have shed light on tyrosine redox chemistry.

3.1 Solution thermochemistry

The Pourbaix diagrams reported by Harriman [16] provide a starting point for discussions of the solution redox chemistry of tyrosine and related phenols. They give tyrosine reduction potentials across all biologically relevant pH values that are in good agreement with those from pulse radiolysis experiments [17]. Electrochemical [18] and pulse radiolysis [19] measurements of the tyrosyl/tyrosinate or phenoxyl/phenolate couples complement studies of the H⁺/e⁻ neutral/radical. An estimate of the pKa of the tyrosyl radical cation in aqueous solution [20] completes the thermodynamic square scheme shown in Figure 1. Such schemes are useful for predicting the course of reactions involving reagents that can donate or accept H⁺ and e⁻ [21]. The free energy for X−H → X^• + H^• bond homolysis in solution (bond dissociation free energy or BDFE) can be calculated using eqn. 2 as described elsewhere [21,22].

Figure 1.

Thermodynamic cycle for tyrosine in aqueous media. The constant C_G accounts for the free energy to make H^•_(solv) from ½H₂(g) such the BDFE corresponds to the chemical reaction XH → X^• + H^•.

$$\text{BDFE}(\text{kcal}{\text{mol}}^{-1})=1.37{\text{p}K}_{\text{a}}+23.0E°+C_{\text{G}}$$ (2)

Aqueous thermochemistry is a good starting point for understanding redox reactions of tyrosine, but it provides only a qualitative picture of such reactions in polypeptides. Electrochemical and pulse radiolysis studies of di- and tri-amino acid units indicate that the redox chemistry of tyrosine can be modulated when in a polypeptide [23], but these are still far from native systems. The dielectric environment of any redox cofactor has a great impact on its reduction potential and therefore on −ΔG°, a key component of the rate of ET (eqn. 2). As an illustration of this point, we take oxidation of 2,4,6-tri-tertbutylphenol (^tBu₃PhOH) in water and acetonitrile as a prototypical example (Figure 2). Note that we choose this as an example because complete thermodynamic cycles are available in ethanol/water and in MeCN. In ethanol/water E(^tBu₃PhOH^•+/0) = 1.28 V vs. NHE as derived from simulations of irreversible voltammograms [24], and 1.2 V vs. Cp₂Fe^+/0 in MeCN [25] (1.8 V versus NHE using one common conversion, [26]). The pK_a of this phenol is 13 in ethanol/water, but 28 in MeCN. In contrast, the homolytic bond strengths only change by 3 kcal mol⁻¹ from water to MeCN. While we are unable to make quantitative predictions of redox potentials of tyrosine within proteins, the qualitative conclusion that it is more difficult to generate charged species in a low dielectric medium still holds. On the other hand, bond homolysis does not generate charged species. The thermodynamic data underscore a key component of the reactivity of tyrosine that we will return to below: oxidation and reduction of tyrosine occur with loss and gain of H⁺ + e⁻ in all but extreme cases.

Figure 2.

Thermodynamic cycles for tri-tert-butylphenol in (A) ethanol/water and (B) acetonitrile. The constant C_G for (A) is as described in Figure 1. In MeCN (B), the constant C_G relates CH₃CNH⁺ + Cp₂Fe⁰ to 1/2H₂ and then to H^•_(solv) so that the BDFE corresponds to the chemical reaction XH → X^• + H^•. A derivation and descriptions of the assumptions implicit in this value is given in reference [22].

3.2 Small molecule models for tyrosine oxidation

Small molecule models have been reported that probe the proton-coupled oxidation of tyrosine. Hammarström and co-workers have investigated photosystems where tyrosine is covalently linked to a Ru^II-tris(diimine) photosensitizer (Figure 3) [27,28] and Nocera has reported analogous photosystems where Re^I is the photosensitizer [29]. In both of these systems electron transfer is intramolecular (YOH → Ru^III) and proton transfer is intermolecular (H⁺ to water/buffer). Interestingly, relatively high reorganization parameters (γ, see above) have been estimated for these reactions (γ > 1.2 eV for ET to Ru and H⁺ loss to solvent). Interestingly, the rate constants for oxidation of tyrosine in these models show an unusual pH dependence [30] that has been the source of controversy [31–33]. The origin of this pH dependence has not been resolved.

Figure 3.

Ru^II photosystem for studying tyrosine oxidation. Flash-generated Ru^III is the oxidant. The driving force can be varied by changing the ancillary bipyridine ligands.

Solution electrochemical investigations of phenol oxidation where the proton is transferred to an exogenous base (e.g., water, buffer or nitrogen heterocycles) and the electron is transferred to an outer-sphere oxidant or electrode also have been reported. In many ways these reactions are an outgrowth of decades of work on free radical reactions of phenols and related compounds [34] where H⁺ + e⁻ are transferred from phenol (PhOH) to a single acceptor (e.g. PhOH + R^• → PhO^• + RH). In key experiments, Linschitz observed that addition of organic bases to solutions of phenol in acetonitrile facilitates phenol oxidation by excited state oxidants [35]. Likewise, in aqueous solution addition of pyridine [36] facilitates electrochemical oxidation of phenol or ascorbate/quinones. Hammarström [30] and Stanbury [31] have reported phenol oxidation studies where flash-generated Ru(bpy)₃³⁺ or IrCl₆²⁻, respectively, acts as the oxidant and water or buffer is the proton acceptor. Costentin and Savéant have extensively studied electrochemical phenol oxidation where water or buffer acts as proton acceptor [24,37,38]. These workers find that phenol oxidation at an electrode has γ closer to 0.5 eV as determined by varying the driving force for outer-sphere oxidation with proton loss to unbuffered water. In related experiments, Meyer and Thorp have used indium tin oxide electrodes to study electrocatalytic phenol oxidations by M(bpy)₃³⁺ (M = Ru, Os) complexes [39].

Mayer and co-workers have reported outer-sphere oxidation of a series of hydrogen bonded phenol systems in acetonitrile solution (Figure 4) [40–42]. In these systems proton transfer is intramolecular and electron transfer is intermolecular. While these systems may not be good biological mimics, they allow for systematic variation of the oxidant and the proton-accepting base in the H⁺/e⁻ oxidation of phenols. In another recent example [43], the proton transfer distance was varied, allowing for experimental tests of theoretical predictions of proton-coupled electron transfer [15]. Mayer’s work suggests that redox reactions of H-bonded phenol models are associated with large γ values (over 2 eV in some cases for ET self-exchange coupled to intramolecular proton transfer). These γ values were obtained by measuring rate constants as a function of driving force and fitting the data using semiclassical ET theory (eqn. 1) [29]. More importantly, these studies provide a very clear example of how a base can affect phenol reduction potentials in a low dielectric environment (the potentials are more than 0.5 V lower than those of the analogous ^tBu₃PhOH^•+/0 couple). Costentin and Savéant also have studied these (and related) phenols by electrochemistry [44], including a case in which phenol oxidation was proposed to occur via an intermolecular proton relay [45].

Figure 4.

Small molecule models for phenol oxidation where outer-sphere ET is coupled to intramolecular proton transfer.

3.2.1 Mechanisms of tyrosine/phenol oxidations

Three limiting ways to oxidize phenols to phenoxyl radicals are: (1) stepwise oxidation then deprotonation; (2) stepwise deprotonation, then oxidation; and (3) concerted oxidation/deprotonation. As noted above, tyrosine undergoes 1e⁻ oxidation or deprotonation only under forcing conditions because of its high pK_a and high reduction potential in a low dielectric environment, as a protein active site. In general, the above examples implicate concerted transfer of H⁺ + e⁻ (= H^•) as the primary mechanism of phenol oxidation. The clearest mechanistic illustrations come from Costentin and Savéant who used electrochemical simulations [37] and from Mayer’s thermodynamic arguments [40]. Take, for example, the degenerate self-exchange reaction between ^tBu₃PhOH and ^tBu₃PhO^• (Figure 5), where the free energy change for initial ET or proton transfer (PT) is the same, 43 kcal mol⁻¹; and the self-exchange barrier is 15.7 kcal mol⁻¹ [46]. Thus, the self-exchange reaction is not likely to proceed via initial ET or PT because even the ground state free energy change for those steps is well above the observed barrier. Indeed, it would appear from thermodynamic and kinetics analyses that tyrosine/phenol oxidation occurs via loss of H⁺ + e⁻ in a single step.

Figure 5.

Pathways for H⁺/e⁻ oxidation of ^tBu₃PhOH. ΔG° is the standard free energy change estimated using data from Figure 2 (MeCN). ΔG^‡ is the barrier calculated from the Eyring equation (k = (k_BT/h)exp(−ΔG^‡/RT)) for the self-exchange rate constant for H• exchange.

3.3 Peptide model systems

Pulse radiolysis and electrochemical experiments on YOH in di- and tri-peptides were mentioned above [23]. Several studies of ET between YOH and W^•+ radicals generated from reaction with N₃^•− using pulse radiolysis have been reported, some involving YOH-W dipeptides [47], while others employed more complex polypeptides with W and YOH separated by one or more amino acids [48]. It has been proposed that systems where W and YOH are separated by a polyproline spacer (Figure 6, top) [49,50] have distance decay constants (β) that are smaller than those documented for Ru-modified proteins [51]. It is our view, however, that the dynamical properties of these polypeptides greatly complicate definitive determinations of distant donor-acceptor couplings [52,53].

Figure 6.

Polyproline bridged models for studying tyrosine redox chemistry. (top) W^•+ radicals generated using pulse radiolysis. (bottom) dialkoxyphenylalanine radicals generated using laser-flash techniques. X groups that have been studied include trimethoxyphenylalanine, tryptophan, methionine, alanine and cysteine.

More recently, Giese and co-workers have extended work on polyproline-based systems to include models that probe reactions in which YOH is oxidized to YO^• via intramolecular ET (Figure 6, bottom) [54,55]. Multistep tunneling (hopping) involving a redox-active intermediate is the likely mechanism of these reactions [51]. Notably, workers in the Giese lab have employed these model polypeptides to investigate the effects of varying the intermediate amino acid on the hopping rate [56]. While these systems are not specifically geared toward elucidating the redox properties of YOH, they have shed light on the factors that influence hopping rates in biological model systems.

The increasing number of high resolution x-ray structures of enzymes that use YO•/YOH in redox catalysis has spurred the development of protein models that better mimic natural systems. Tommos and co-workers have characterized de novo designed 3- and 4-helix bundles containing W or YOH by a variety of techniques [57], including NMR. The solution structure for the 3-helix bundle with an embedded W is shown in Figure 7. [58]. In similar motifs, Tommos has investigated the electrochemical responses of phenol analogues that have been introduced by coupling mercaptophenols to a cysteine residue [59], and she and co-workers are currently studying the redox properties of 3-helix bundles where an embedded tyrosine in each protein is positioned near a proton-accepting base [60].

Figure 7.

NMR structure of an α₃W 3-helix bundle: PDB ID 1LQ7. Tryptophan is highlighted.

Barry and co-workers have reported tyrosine-containing peptides that fold in a β-hairpin motif [61]. Although the behavior of tyrosine in this β-hairpin is similar to that in water, it is apparent from electrochemical studies that protonation/deprotonation of nearby amino acids can have a substantial influence on its redox properties [62].

We turn now to recent investigations in our laboratory on multistep electron tunneling (hopping) in modified natural proteins. We have shown that long-range Cu^I oxidation in Pseudomonas aeruginosa Re^I(H124)(W122)-azurin is greatly accelerated by hole hopping through tryptophan [63], but we have not been able to observe YOH-assisted hopping in related protein models. Photogenerated YO• radicals in modified azurins have been observed by EPR [64], however, hopping to generate Cu^II is not implicated [65,66]. High potential Re^II oxidants were used to generate those YO^• radicals and kinetic parameters could not be extracted. What we have learned from this work is that fine tuning of the reduction potentials of candidate redox intermediates is absolutely critical for hopping function, and for tyrosine such tuning may be achieved by installing a basic amino acid that can hydrogen bond to the YOH proton in a modified protein. Our current investigations are focused on modified azurins where tyrosine is hydrogen bonded to His, Asp or Glu.

We start with a robust protein in which all YOH and W residues have been replaced with phenylalanine, which we call “all-Phe” azurin. Using Rosetta software [67], we surveyed candidates where YOH and His could be introduced near each other (Sheffler, W., Tinberg, C.E. Warren, J.J. Gray, H.B. and Baker, D.A. work in progress). Among these all-Phe azurin candidates, one that stood out as particularly attractive was the His20/Tyr48 protein, which we expressed and purified by standard methods. The visible absorption spectrum of the modified protein is identical with that of native azurin (the signature S-Cys to Cu^II charge transfer band maximum is at 630 nm).

Importantly, the predicted YOH···His hydrogen bonding interaction is present in the His20/Tyr48 protein (Figure 8). The O···N distance, which is 2.7–2.8 Å based on two molecules in the asymmetric unit, is similar to those in small molecule models where the phenolic oxygen is strongly hydrogen bonded to a nitrogen base [40]. The His-NH is also involved in a weak hydrogen bonding interaction (3.0–3.1 Å) with the backbone carbonyl of the adjacent Thr19. The protein backbone at position 20 is in a different conformation than in other azurin structures, presumably to accommodate the non-conservative Ile→His mutation and the YOH···His hydrogen bond. The protein also shows decreased stability at pH values below 7.5, likely due to protonation of His20, which is located in the hydrophobic core. Our work with the His20/Tyr48 model indicates that engineering redox active YOH into native protein scaffolds can provide better models of natural systems, but challenges such as protein destabilization are vital considerations.

Figure 8.

H20/Y48 azurin structure with a close-up view of the H20···Y48 hydrogen bond.

4. Concluding remarks

We have reviewed redox reactions of tyrosine and related phenols ranging from small molecule models in organic solvents to models from our lab that were built within native protein scaffolds. All these investigations share a common goal, namely elucidation of the factors that govern the electron transfer chemistry of tyrosine in biological redox machines.

Work on both small molecule and macromolecular systems indicates that a base near the phenolic proton tunes the YO^•/YOH reduction potential and also keeps the proton near the YO• radical. Examination of x-ray structures shows that enzymes that use YOH as a redox way station typically place a proton-accepting group near the phenolic proton thereby ensuring that both H⁺ and e⁻ are available to reactions. While a small molecule model can be built with and one of a wide variety of bases (amines, pyridines, imidazoles, carboxylic acids, etc.), natural systems normally employ one of three amino acids (His, Asp, Glu). In proteins where YO^• participates in catalysis (e.g. cytochrome c oxidase), the redox active YOH placed near a reactive substrate likely serves as a (net) donor/acceptor of H^•.

An important conclusion is that concerted H⁺/e^– loss is a common feature of tyrosine oxidation. Electrochemical experiments have shown that sequential deprotonation, then ET, can be operative in solution when sufficiently strong bases (pK_a ≥ 10) are present. Even stronger bases could be needed to deprotonate protein-embedded tyrosines. To the best of our knowledge, the radical cation, YOH^•+, has not been observed in studies of YOH oxidations. Conversely, reduction of YO^• is more likely to proceed via initial ET or via a concerted mechanism to avoid production of YOH^•+. It is interesting to note that in order to maintain high potentials in multistep ET pathways, oxidation of YOH and reduction of YO^• probably both proceed via a concerted mechanism; oxidation is concerted to avoid high energy intermediates and reduction is concerted because the free energy change for H⁺/e⁻ addition to YO^• is higher than E°(YO^•/−).

A great deal of the work on small molecule systems has focused on the reorganization parameter (γ) for phenol/tyrosine oxidation, but there is little consensus. Different investigators have estimated values of γ ranging from 0.5 to above 2 eV. From a biological perspective, there is no apparent advantage in utilizing cofactors that require extensive nuclear reorganization for function. As illustrated by Roth and Klinman for oxygen reduction by glucose oxidase, the protein matrix can control reorganization parameters [68].

Since the kinetics and thermodynamics of H⁺/e⁻ redox reactions of tyrosine depend so strongly on the nature of its local environment, there is great interest in building models where YOH is positioned in protein environments similar to those of natural systems. Progress in this area has been slow, mainly because control of YOH orientation in relation to nearby groups (including proton acceptors) is not an easy task. Moreover, the redox chemistry of YOH within small protein scaffolds often is complicated by the presence of other ionizable groups, highlighting the difficulty associated with keeping track of protons and electrons in biological macromolecules. Although the experiments are challenging, we believe that the study of suitably designed photosystems by time-resolved spectroscopic methods will provide key insights into the redox chemistry of tyrosine embedded in polypeptides. In such systems, positioning of the photosensitizer relative to tyrosine will be critical for ET function [63,69].

Table 1.

Crystallographic statistics.

	H20/Y48 azurin
Space group	P 21 21 21
A,B,C	48.026, 56.330, 82.910
α,β,γ	90, 90, 90
Observed reflectionsa	262814 (32071)
Unique reflectionsa	70523 (10078)
Completenessa	94.9% (84.9%)
Robsa,b	4.3% (27.1%)
I/σIa,c	15.63 (3.72)
Resolution range	1.18–20.0
High resolution bin	1.182–1.213
Rfree (e.s.u.)	21.7% (0.047 Å)
Rworking set(e.s.u.)	19.5 % (0.048 Å)
Mean B	12.878
RMS deviation: bond lengths	0.029
RMS deviation: bond angles	2.336
Refined amino acids (waters)	254 (277)

^aTotal (outershell).

^b(SUM(ABS(I(h,i)−I(h))))/(SUM(I(h,i))).

^cmean of intensity/σI of unique reflections (after merging symmetry-related observations).

List of abbreviations

ET

electron transfer

PT

proton transfer

YOH

tyrosine

YO^•

tyrosyl radical

W

tryptophan

PCET

proton-coupled electron transfer

BDFE

bond dissociation free energy

DEA

diethanolamine

Tris

tris(hydroxymethyl)aminomethane