Two Tryptophans Are Better Than One in Accelerating Electron Flow through a Protein

Abstract

We have constructed and structurally characterized a Pseudomonas aeruginosa azurin mutant Re126WWCu^I, where two adjacent tryptophan residues (W124 and W122, indole separation 3.6–4.1 Å) are inserted between the Cu^I center and a Re photosensitizer coordinated to the imidazole of H126 (Re^I(H126)(CO)₃(4,7-dimethyl-1,10-phenanthroline)⁺). Cu^I oxidation by the photoexcited Re label (*Re) 22.9 Å away proceeds with a ∼70 ns time constant, similar to that of a single-tryptophan mutant (∼40 ns) with a 19.4 Å Re–Cu distance. Time-resolved spectroscopy (luminescence, visible and IR absorption) revealed two rapid reversible electron transfer steps, W124 → *Re (400–475 ps, K₁ ≅ 3.5–4) and W122 → W124^•+ (7–9 ns, K₂ ≅ 0.55–0.75), followed by a rate-determining (70–90 ns) Cu^I oxidation by W122^•+ ca. 11 Å away. The photocycle is completed by 120 μs recombination. No photochemical Cu^I oxidation was observed in Re126FWCu^I, whereas in Re126WFCu^I, the photocycle is restricted to the ReH126W124 unit and Cu^I remains isolated. QM/MM/MD simulations of Re126WWCu^I indicate that indole solvation changes through the hopping process and W124 → *Re electron transfer is accompanied by water fluctuations that tighten W124 solvation. Our finding that multistep tunneling (hopping) confers a ∼9000-fold advantage over single-step tunneling in the double-tryptophan protein supports the proposal that hole-hopping through tryptophan/tyrosine chains protects enzymes from oxidative damage.

Short abstract

Insertion of two tryptophan residues between a rhenium photooxidant and the copper(I) center in Pseudomonas aeruginosa azurin accelerates electron flow by almost 4 orders of magnitude.

Introduction

Single-step tunneling (ET) in proteins can move electrons between donor and acceptor sites separated by about 25 Å on a millisecond time scale.¹⁻⁴ Inserting redox-active groups between the terminal donor and acceptor accelerates electron transport (ET_hop) by splitting the reaction pathway into shorter tunneling steps,^1,2,4−9 achieving much higher charge migration rates and extending the charge separation range. Many natural redox systems employ multistep tunneling (hopping), transferring an electron sequentially along a series of redox proteins or cofactors. A case in point is the Ralstonia eutropha O₂-tolerant [NiFe]-hydrogenase, where electrons travel from the active site to the protein surface through a series of Fe–S clusters involving tunneling steps of 10.7, 9.7, and 8.7 Å;¹⁰ even more striking is the respiratory complex I, where an electron is transported over 90 Å through a redox chain consisting of a flavin mononucleotide and a series of Fe–S clusters.^11,12 Electron hopping also takes place in photosynthesis—both within reaction centers and when moving the photoseparated holes and electrons along the chloroplast membrane.⁶

Tryptophan and/or tyrosine residues are of special importance as hole hopping intermediates. In the prototypal radical enzyme ribonucleotide reductase,¹³⁻¹⁵ substrate reaction is triggered by 35-Å hole transfer across a chain of Tyr residues to the nucleotide binding site;^5,16−20 and in photolyases and cryptochromes, a photogenerated hole moves over ∼15 Å in ∼30 ps from a photoexcited flavin through a chain of three precisely positioned tryptophans.^21,22 We recently proposed that hole transfer through Trp/Tyr chains protects oxidases and oxygenases by moving potentially destructive oxidizing equivalents (holes) to protein surfaces where they can be disarmed by cellular reductants.^1,23−27 While our hypothesis is well supported by bioinformatics analysis of the structures of redox enzymes, it calls for deeper mechanistic investigations of the hopping mechanism.

The blue copper protein Pseudomonas aeruginosa azurin is an excellent platform to investigate ET mechanisms, owing to the presence of a reversible Cu^II/I redox couple in a robust structure that allows for multiple mutations and covalent attachment of a Ru^II or Re^I photooxidant to a surface histidine (H) at a defined position.^7,8,28−33 Although azurin does not contain chains of aromatic amino acids,²⁷ tryptophan^7,34,35 and nitrotyrosine⁸ residues can be introduced into the redox pathways by site mutations with retention of the protein structure. In particular, replacing a lysine (K122) by tryptophan (W) results in dramatic (>100×) acceleration of Cu^I oxidation by a photoexcited Re metallolabel in Re124W122Cu^I azurin (Re = Re^I(CO)₃(dmp)(H124)⁺; CT excited state (*Re) = Re^II(CO)₃(dmp^•–)(H124)⁺; dmp = 4,7-dimethyl-1,10-phenanthroline).^7,9 The reaction involves W122 oxidation as the first ET step in the mechanism (Scheme 1A). Electron (hole) hopping also can occur across a hydrophobic protein–protein interface, as was observed in {Re126T124W122Cu^I}₂, where Re excitation in one subunit leads to oxidation of the tryptophan and Cu^I in the neighboring subunit.³⁴ Expanding the azurin electron hopping system to include mutants with two closely spaced aromatic amino acid residues provides a well-characterized minimal model to investigate multiple hopping (Scheme 1B). We report here on a series of structurally characterized azurin mutants labeled with a Re photooxidant at H126 and containing either tryptophan or phenylalanine (F) at positions 124 and 122. (Mutants are abbreviated Re126WWCu, Re126WFCu, and Re126FWCu, where the first and second letters specify the 124 and 122 residues, respectively. All other naturally occurring Trp and Tyr residues were replaced by Phe.) ET_hop reactions were studied in Cu^I azurins, whereas the corresponding Cu^II and Zn^II forms were used to evaluate redox reactivity of the Re···W···W moiety in isolation. Systematic spectroscopic and kinetics investigations of photoinduced ET_hop in these mutants have shed light on factors that control multiple hole hopping along tryptophan chains.

Scheme 1. (A) Photoinduced ET Cycle of Re124W122Cu^I (ref (7)) and (B) Analogous Photoinduced ET Cycle of Re126WWCu^I.

(A) *³CT and ³CT denote hot and relaxed excited states of the Re label, respectively. The photocycle starts with optical excitation to the ¹CT state *Re^II(CO)₃(dmp^•–)H124W122AzCu^I followed by several relaxation steps, establishing an equilibrium between ³CT and the charge-separated (CS) state Re^I(CO)₃(dmp^•–)H124(W122^•+)Cu^I. The oxidized tryptophan intermediate W122^•+ then undergoes ∼30 ns reduction by Cu^I over a ∼11 Å distance, forming the redox product (RP) Re^I(CO)₃(dmp^•–)H124W122Cu^II. The cycle is completed by ∼3 μs dmp^•– → Cu^II back electron transfer across 19.4 Å (refs (7 and 9)). (B) The time constants were determined in this work. The Re–Cu charge separation takes place over 23 Å via hopping through two Trp residues. The hot *³CT state and its relaxation were omitted for clarity. ³CT is a mixed Re → dmp MLCT/dmp-intraligand state.

Results and Discussion

Structures

X-ray crystal structures of Re126WWCu^II (PDB ID: 6MJS), Re126WFCu^II (6MJT), and Re126FWCu^II (6MJR) were determined (Table S1) to resolutions of 1.85, 1.9, and 2.0 Å, respectively, and the regions of the redox cofactors are shown in Figure 1. The shortest ET-relevant distances are reported in Table 1. The Re–W124–W122 hopping sequence is characterized by multiple short (3.5–4.0 Å) contacts between mutually T-oriented aromatic groups, and the dmp methyl groups are in close proximity to the W124 indole. Structures of all three mutants are superimposable, and replacement of either of the two tryptophans in Re126WWCu^II with phenylalanine switches off one of the hopping steps without altering the overall geometry or length of the Re–Cu ET_hop pathway. In particular, the Re chromophore becomes redox-isolated in Re126FWCu^I, where the W122 residue is too far from the Re site for ET to compete with ³CT decay. The W122–Cu pathway (∼11 Å) is the same in Re126WWCu^II as in single-tryptophan mutants Re126FWCu^II, Re126T124W122Cu^II,³⁴ and Re124W122Cu^II(7) (Figure S1). In Re126WFCu^I, the photocycle is largely limited to ET in the Re126W124 unit, owing to the long W124–Cu distance.

Figure 1.

Structures of ReH126-azurin mutants showing intramolecular distances between the redox cofactors. WW: Re126WWCu^II (PDB ID: 6MJS). FW: Re126FWCu^II (6MJR). WF: Re126WFCu^II – chain B (6MJT; in chain A, the W122-indole is oriented backward and the Re(CO)3(dmp) unit is tilted leftwards). Lower right: Superposition of ET-relevant regions and protein folds of Re126WWCu^II (green), Re126FWCu^II (pink), and Re126WFCu^II – chain B (light blue) demonstrates their structural similarity. Packing of Re126WWCu^II and Re126FWCu^II in the respective asymmetric unit is shown in Figure S2.

Table 1. Shortest Atom–Atom Intramolecular Distances between Redox-Active Sites^a.

distance	Re126WWCuII	Re126WFCuIIb	Re126FWCuII	Re124W122CuIIh
Re–W124	6.9	7.6
dmp–W124	3.5c	3.7d
W124–W122	3.9e
Re–W122	11.4		11.1	6.3
dmp–W122	7.8		7.1	3.4
Cu–W122	10.7	15.7f	10.7	10.8
Cu–dmp	20.6	19.9	20.2	16.0g
Cu–Re	22.9	22.7	23.3	19.4
angle (deg)
dmp–W124	67.7			20.8i
W124–W122	78.7

^aOnly aromatic C and N atoms, as well as Re and Cu, are considered. Values averaged over the molecules comprising the unit cell.

^bTwo molecules with different Re/W122-indole orientations are present. The listed distances are pertinent to the molecule with closer contacts.

^cAn additional close contact (3.9 Å) exists between the W124 indole ring and C(CH₃-dmp).

^dClosest distance between the indole ring and C(CH₃-dmp) = 3.5 Å.

^eThe distances in the four molecules comprising the asymmetric unit are in the range 3.6–4.1 Å.

^fCu–W124 distance.

^gClosest distance between Cu and C(CH₃-dmp) = 15.3 Å.

^hPDB ID: 2I7O; see Figure S1.

ⁱdmp-W122.

Re126WWCu^II and Re126FWCu^II pack in the asymmetric unit so that redox cofactors on different protein monomers interact with each other (Figure S2). Assuming that similar dimerization occurs in solution,^34,36 it is likely that intermolecular ET_hop will be observed at higher protein concentrations.

Photoinduced Electron Transport

The ET_hop kinetics of Re126WWCu^I and its variants were studied following the protocol established for Re124W122Cu^I (Scheme 1).⁷ Pulsed laser excitation of the Re label at 400 or 355 nm triggers a sequence of ET steps whose kinetics were followed by measuring the decay of *Re luminescence at 560 nm and absorption-time profiles at 500 (*Re and ReH126(CO)₃(dmp^•–)) and 632.8 nm (Cu^II formation and decay). Time-resolved IR (TRIR) spectroscopy in the range of CO stretching vibrations was used to distinguish ground, excited, and reduced forms of the Re label (negative (bleach) bands and positive features shifted to higher and lower frequencies upon excitation, respectively).³⁷ In all UV–visible transient spectroscopic experiments, the protein concentration was kept below 40 μM to minimize contributions from intermolecular ET_hop.³⁴ Additional insights were provided by measurements with other mutants: all reactivity is confined to the Re126WW moiety in ReH126WWZn^II and ReH126WWCu^II; ET between *Re and the proximal Trp in isolation was probed in ReH126WFCu^I, and the Re label is effectively removed from the redox pathway by phenylalanine in ReH126FWCu^I. Results are summarized in Figure 2 and Table 2, and Scheme 1B outlines the mechanism together with elementary rate constants extracted from kinetics simulations (vide infra).

Figure 2.

Transient absorption and luminescence time profiles measured on dilute (<40 μM) Re-azurin solutions. (A) Transient absorption of Re126 azurins at 632.8 nm: Re126WWCu^I, 32 μM; Re126WWZn, 17 μM; Re126FWCu^I, 30 μM; Re126FWCu^I, 24 μM. (B) Comparison of the Cu^II transient absorption signals (632.8 nm) for Re126WWCu^I (22 μM, red) and Re124W122Cu^I (27 μM, black, scaled by a factor of 22/27) measured under virtually identical excitation conditions (1–1.5 mJ/pulse). (C) Luminescence decay of Re126 azurins at 560 nm. (D) Multiexponential luminescence decay of Re124W122Cu^I in the pico-nanosecond range.

Table 2. Kinetics Fitting Parameters from Time-Resolved Luminescence and Transient Absorbance Measurements on Re126 Azurins with 355 nm Excitation.

	τ1 /ps	τ2 /ns	τ3 /ns	τ4 /μs	τ5 /μs	τ6 /ms
Re126WWCuI
luminescence τ	270 ± 20	4 ± 1	81 ± 6
(% amplitude)	(61 ± 3)	(15 ± 2)	(23 ± 1)
TA τ			68 ± 5	1.2 ± 0.1	123 ± 10	4.6 ± 0.5
amplitude (632.8 nm)			–0.011	0.005	0.020	0.002
amplitude (500 nm)			0.004	0.003	0.008	0.0008
Re126WWCuII
luminescence	290 ± 10	4 ± 2	79 ± 7
(% amplitude)	(71 ± 1)	(9 ± 1)	(20 ± 2)
Re126WWZnII
luminescence	430 ± 40	10 ± 3	100 ± 7
(% amplitude)	(45 ± 2)	(18 ± 2)	(36 ± 2)
TA (500 nm) τ			125 ± 30	4 ± 3a
Re126WFCuI
luminescence	200 ± 10	3 ± 1	71 ± 8
(% amplitude)	(73 ± 2)	(14 ± 1)	(13 ± 1)
TA (500 nm) τ				0.9 ± 0.1
Re126FWCuI
luminescence	340 ± 150	3 ± 2	120 ± 100	1.15 ± 0.2
(% amplitude)	(36 ± 6)	(14 ± 6)	(5 ± 4)	(45 ± 5)
TA (500 nm) τ				1.10 ± 0.15	20 ± 10a

^aMinor component.

*Re luminescence is strongly quenched by W124 in Re126WWCu^I, Re126WWZn^II, and Re126WFCu^I, decaying with fast multiexponential kinetics (Table 2). On the other hand, Re126FWCu^I (Figure 2C) exhibits a long luminescence decay time (1.15 μs) consistent with an unquenched ³CT excited state. (Similar values were found for redox-inactive Re126T124X122Cu^I (X = K or F, 730 ns) and Re124F122Cu^I (1.3 μs).^7,34) Pico- and nanosecond TRIR spectra (Figure 3) demonstrate that the *Re ³CT excited state decays to produce a charge-separated (CS) state with a reduced Re complex, Re^I(H126)(CO)₃(dmp^•–).^37,38 Luminescence decay kinetics obtained for all mutants containing W124 are similar (Table 2), indicating that *Re reduction by W124 is the common reaction step, regardless of the metal (Cu^I, Cu^II, Zn^II) or the 122 amino acid (W, F).

Figure 3.

Difference TRIR spectra of Re126WWCu^I measured at selected time delays after 400 nm, 50 fs excitation. Measured in ∼1.8 mM/D₂O solution, 20 mM KP_i (pD ≅ 7.1). Blue and red labels denote features due to the ³CT state (*Re) and reduced Re^I(H126)(CO)₃(dmp^•–) in the two CS states (and RP at later time delays). Negative bands correspond to depleted ground-state population. The spectral features evolve in the directions of the arrows. Time evolution of the highest CT band is largely determined by excited-state relaxation (ref (38); the simultaneous decay and rise of ³CT and CS features on late-picosecond and early nanosecond time scales confirm (ultra)fast reduction of the excited Re label. Because of the high concentration used (∼1.8 mM), the kinetics correspond to a combination of intra- and intermolecular processes.

Nanosecond transient absorption (TA) measurements revealed large absorbance increases indicative of Cu^II formation (632.8 nm) only in Re126WWCu^I and Re124W122Cu^I (Figure 2); Cu^II formation was not observed in low-concentration solutions of Re126WFCu^I, Re126FWCu^I, or Re126WWZn^II, whose much weaker transient absorption at 632.8 nm (Figure 2A) originates from the ³CT and CS states. Hence, fast Cu^I photooxidation requires the presence of both W124 and W122 in the ET_hop pathway. The ∼68 ns rise of Re126WWCu^I 632.8 nm TA parallels that observed⁷ for Re124W122Cu^I (∼40 ns), despite different Re–Cu distances (22.9 and 19.4 Å, respectively), indicating an analogous rate-determining step (the ∼11 Å W122^•+→Cu^I “hole hop”). TA kinetics (632.8 nm) measured under virtually identical excitation conditions show that the Cu^II yield for Re126WWCu^I is 1.5–2.4 times lower than for Re124W122Cu^I, where ET_hop involves a single W122 intermediate (Figure 2B and SI - Section S3).

The Re^I(H126)(CO)₃(dmp^•–) → Cu^II recombination reaction closes the photocycle; simultaneous fitting of the 632.8 and 500 nm kinetics gives a ∼120 μs time constant for this process. The back-reaction time constant accords with the estimate (∼150 μs) for single-step dmp^•– → Cu^II tunneling (SI-Section S2). The almost 2000-fold difference in the charge-separation and recombination time scales reflects the different mechanisms: multistep and single-step tunneling, respectively. The charge-separation/recombination advantage increases with ET_hop range: Re124W122Cu^I shows an ∼80-fold difference over 19.4 Å.

The overall performance of the Re126WWCu^I photocycle can be assessed by comparing Cu^II formation kinetics (monitored at 632.8 nm) with the time constant of single-step Cu^I → *Re ET (∼630 μs) estimated from the value reported³⁰ for Re83WT-azurinCu^I (770 ns, r = 16.8 Å) by correcting for the longer Cu–Re distance (22.9 Å) in Re126WWCu^I (see SI-Section S2). Given an unquenched ³CT lifetime of about 1 μs,⁷ photoinduced Cu^I oxidation should not be observable if single-step tunneling were the only operative mechanism. Instead, 355 nm, ∼8 ns laser-pulse excitation of ≤40 μM Re126WWCu^I solutions led to Cu^II (RP, Scheme 1) formation (∼68 ns time constant), followed by ∼120 μs ground-state recovery. Remarkably, hole hopping through the two intervening tryptophan residues accelerates Cu^I oxidation by a factor of 9000 compared to single-step tunneling.

The solution to the rate law for the kinetics model outlined in Scheme 1B (beginning from³CT) is a 4-exponential function. One of the empirical rate constants in the solution is equal to the elementary rate constant k₉. The remaining empirical rate constants are functions of k_1–8, given by the roots of a third-order polynomial. To obtain estimates for these elementary rate constants, we solved the rate law numerically (see SI, Section S3). Values used for k₃ and k₄ were fixed, based on measurements in Re124W122Cu^I and Re126FWCu^I, respectively.⁷ We simulated the kinetics for 4.8 × 10⁷ combinations of the remaining parameters (see SI, Section S3) and retained those combinations in which the empirical rate constants were in satisfactory agreement with the observed luminescence rate constants and relative amplitudes, the TA kinetics, and the relative Cu^II yield (Table 3).

Table 3. Results from Numerical Solutions to the Rate Law Implied by Scheme 1B.

Re126WWCuI	Inputs: k3–1 = 60 ns; k4–1 = 1.15 μs; k9–1 = 120 μs
elementary rate constants	k1–1 /ps	K1	k5–1 /ns	K2	k7–1 /ns	k8–1 /ns
	400–475	1.5–2.0	7–9	0.55–0.75	125–750	60–90
yield and empirical time constants	τ1 /ps	τ2 /ns	τ3 /ns	τ4 /μs	Φ124/Φ126
	260–280	3.5–4	70–90	120	2.0–2.2

Re126WWCuII	Inputs: k3–1 = 60 ns;k4–1 = 1.15 μs; k8 = k9 = 0
elementary rate constants	k1–1 /ps	K1	k5–1 /ns	K2	k7–1 /ns
	375–425	2.25–3.25	9–21	0.25–0.75	100–350
empirical time constants	τ1 /ps	τ2 /ns	τ3 /ns
	280–300	3.5–4.5	70–90

Re126WFCuI	Inputs: k4–1 = 1.15 μs; k5 = k6 = k8 = k9 = 0
elementary rate constants	k1–1 /ps	K1	k3–1 /ns
	234	5.7	61 ns
empirical time constants	τ1 /ps	τ2 /ns
	200	71

The kinetics simulations of Re126WWCu^I revealed that W124 and W122 are distinct hopping intermediates (if there were a single highly delocalized {W124,W122}^•+ intermediate, the relative Cu^II yield would be much higher). To account for the low Cu^II yield, the CS1 ⇌ CS2 equilibrium must be shifted to the left (K₂ = 0.55–0.75), indicating that hole-localization on W124 proximal to Re is thermodynamically preferred. The simulations suggest that the W122 → W124^•+ ET time constant (k₅^–1) can be constrained to 7–9 ns, but the Re^I(H126)(CO)₃(dmp^•–) → W122^•+ (CS2 → GS) ET time constant is less well-defined (k₅^–1 = 125–750 ns).

The experimental kinetics (Table 2) show several minor components (e.g., the 1.2 μs and 4.6 ms TA decays in Re126WWCu^I) that are not recovered by simulations. These features can be attributed to intermolecular ET_hop in azurin dimers, as observed for Re126T124W122Cu^I,³⁴ and supported by observations of concentration-dependent luminescence decay kinetics in Re126WWCu^II and photoredox activity in Re126FWCu^I at higher concentrations.

Further insight into the nature of the electronic states and intermediates of Re126WWCu^I was obtained by QM/MM molecular dynamics (MD) simulations of the solvated protein,^39,40 where the quantum part (QM) consisted of -Re(H126)(CO)₃(dmp)L125W124G123W122-, and the rest of the system was treated by MM (Figure S12). The QM calculations employed density functional theory (DFT) techniques with the PBE0 functional^41,42 and D3 dispersion correction;⁴³ see the Supporting Information for computational details. For each case, several QM/MM/MD trajectories, which differed in initial conditions, were run for up to 10 ps after equilibration.

In agreement with the proposed mechanism, TDDFT MD simulations found ³CT to be the lowest excited state of solvated Re126WWCu^I. As usual for Re carbonyl-diimines, it arises from Re → dmp metal-to-ligand charge transfer (MLCT) and dmp-intraligand excitations⁴⁴⁻⁴⁸ whose relative contributions vary in time (Figure S13). The ³CT state is closely followed in energy by several CS states. Whereas distances and angles among redox cofactors (Re, dmp, indoles) do not exhibit any major or systematic differences in the ³CT, CS1, and CS2 states (Figure S14), indole solvation was found to be very sensitive to the actual charge distribution and appears to be a dominant component of the ET reorganization. Each indole NH is strongly solvated by a single water molecule at about 2 Å with other water molecules lying farther away (Figures 4, S15). Oxidation of either one of the two tryptophans is accompanied by H-bonding and tighter solvation. In particular, W124 solvation tightens upon oxidation to W124^•+ in the CS1 state where the NH---OH₂ shortens by about 0.1 Å relative to ³CT, owing to a ca. + 0.15 e^– increase in charge on the indole N atom. Restoring uncharged W124 in CS2 relaxes its solvation and shifts water molecules toward W122. Contrasting behavior was found for W122, whose solvation is similar in the ground, ³CT, and CS1 states but tightens in CS2, where the W122^•+ NH group is strongly bound to a single water molecule (Figure 4). Comparing the two tryptophans reveals that the W122 indole is generally solvated less tightly than W124. Surprisingly, in CS2, the distances between W124 and W122^•+ indoles and their respective closest water molecules are comparable despite different charges (Figure 4). The generally weaker W122 solvation could be related to its steric shielding by the -S118A119L120- backbone 3.4–4.4 Å away (Figure 1). Such an asymmetric environment makes W124 a slightly stronger reductant than W122, favors single-indole hole localization in CS1 and CS2 over a delocalized {W124; W122}^•+ intermediate, and shifts the CS1 ⇌ CS2 equilibrium to the left (K₂ < 1), limiting the Cu^II formation yield.

Figure 4.

Left: Distribution function of individual water molecules around W124 and W122 indole NH groups. The dotted line at 1.9 Å helps to visualize the differences. Right: Snapshots showing water molecules within 2.5 Å of the indole-NH hydrogen atoms (the brown sphere represents the Re atom).

Concluding Remarks

Multistep electron tunneling dramatically increases the range over which electrons can be transported through proteins. Our prior study revealed that hole hopping through a single intervening tryptophan residue could accelerate electron transport by a factor of ∼10².⁷ Our present study demonstrates that electron hopping through two adjacent tryptophan residues in Re126WWCu^I accelerates Cu^I oxidation by a factor of ∼10⁴ relative to single-step Cu^I → *Re tunneling. The timetable for electron tunneling/transport (Figure 5) illustrates the advantage of multistep tunneling in the Re-azurin construct: hopping through Trp^•+ intermediates enables sub-microsecond electron transport across more than 20 Å.

Figure 5.

Plot of Cu^I oxidation time constant (τ_Cu) as a function of the metal–metal separation (R_MM) in constructs of Ru- (circles) and Re-derivatized (squares) azurins. Reactions of all Ru-azurins and Re83-azurin (solid square) involve single-step tunneling. The two open squares illustrate the reactions of Re124WCu^I and Re126WWCu^I.

The advantage of hopping over single-step tunneling is sensitive to the structure of the hopping system and the driving forces associated with the individual ET steps. Taking the Re124WCu^I and Re126WWCu^I constructs as models, we simulated the ET_hop kinetics for all possible values of redox-site distances (Figure 6). A key component in the hopping advantage is the single-step tunneling distance. The longer distance in the Re126WWCu^I model gives an optimum double-hop advantage ∼50 times greater than the single-hop advantage in the Re124WCu^I model. The optimum advantage of a double-hop over a single-hop (via Trp) increases as the single-step distance increases. In a comparison of single- and double hopping through Trp over a 20-Å separation between Re and Cu, optimized double-hopping provides just a 2.5-fold advantage over a single hop; at a 23-Å Re–Cu separation, optimized double-hopping is 10 times better than a single hop.

Figure 6.

Simulated single-Trp (A) and double-Trp (B) hopping advantages for constructs modeled on Re124WCu^I (d(Re–Cu) = 20 Å) and Re126WWCu^I (d(Re–Cu) = 23 Å). Optimum positioning of the intervening Trp in the Re124WCu^I construct (r₁ = 7 Å, r₂ = 13 Å, colinear) leads to a predicted 10^3.8 hopping advantage. In the Re126WWCu^I model, optimum positioning of the two Trp residues (r₁ = 6 Å, r₂ = 7 Å, r₃ = 10 Å, colinear) produces a 10^5.5 hopping advantage over single-step tunneling.

Our observations suggest that protein constructs containing multiple closely spaced Trp (or Tyr) residues could support transport of high potential holes across distances of 30 Å or more. Such facile movement of holes through polypeptides necessitates careful placement of oxidizable residues in enzymes that operate at high potentials. We have found that chains of three or more Trp and Tyr residues separated by ≤5 Å are relatively common in the structures of redox enzymes, particularly those that participate in reactions with oxygen.^23−25,27 These Trp/Tyr chains may play protective antioxidant roles by disarming highly oxidizing intermediates when reactions with intended substrates are disrupted.^25,26 Kinetics measurements with Re124WCu^I and Re126WWCu^I demonstrate that once a hole is injected into the indole ring of a Trp residue, it can rapidly migrate to a nearby indole, even when environmental disparities produce an unfavorable free-energy gradient. Current efforts are aimed at elucidating whether Trp/Tyr chains play functional as well as protective roles in high-potential enzymatic redox catalysis.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscentsci.8b00882.

• X-ray data refinement statistics and validation, figures of the Re124W122Cu^II structure and packing of the neighboring chains of Re126WWCu^II and Re126FWCu^II, estimates of single-step tunneling rates, details and summary of kinetics modeling, QM/MM dynamics simulations: structure of the modeled system, trajectories of structural parameters and of indole NH-water distances (PDF)

The authors declare no competing financial interest.

Notes

Safety Statement: No unexpected or unusually high safety hazards were encountered.