Photoinduced Electron Transfer Elicits a Change in the Static Dielectric Constant of a de Novo Designed Protein

Abstract

We provide a direct measure of the change in effective dielectric constant within a protein matrix after a photoinduced electron transfer (ET) reaction. A linked donor-bridge-acceptor molecule, PZn-Ph-NDI, consisting of a (porphinato)Zn donor (PZn), a phenyl bridge (Ph), and a naphthalene diimide acceptor (NDI), is shown to be a ‘meter’ to indicate protein dielectric environment. We calibrated PZn-Ph-NDI ET dynamics as a function of solvent dielectric, and computationally de novo designed a protein SCPZnI3 to bind PZn-Ph-NDI in its interior. Mapping the protein ET dynamics onto the calibrated ET catalogue shows that SCPZnI3 undergoes a switch in the effective dielectric constant following photoinduced ET, from ε_S ~8 to ~3.

Graphical abstract

Anisotropic protein interiors are often described by an effective, uniform static dielectric constant (ε_s) of 2–15.¹ The magnitude of ε_s influences critically biological electron-transfer (ET) reactions.² Through its effect on the pK_a of buried amino-acid residues, ε_s also impacts proton-coupled ET and proton-transfer processes.^1a,3 Theoretical and experimental studies of ET in proteins typically assume ε_s does not change as a result of the ET event.⁴ In this report, we provide direct evidence for a change in the protein ε_s following ET that occurs within a de novo designed protein matrix.

To interrogate dynamical changes in the magnitude of protein ε_s upon photoinduced charge separation, we utilize a donor-bridge-acceptor (DBA) molecule,⁵ with ET dynamics displaying marked sensitivity to: (i) solvent ε_s; (ii) axial metal coordination; and (iii) deuteration of water (kinetic isotope effect). We calibrated these effects for the DBA molecule (5,15-diphenyl-10-(4'-[N-yl-(N'-(11-hydroxy-3,6,9-trioxoundecyl)-naphthalene-1,4,5,8-tetracarboxylate diimide)]phenyl-porphinato)zinc(II) (PZn-Ph-NDI, Figure 1) in a variety of organic solvents, and computationally designed a tetra-α-helical protein SCPZnI3 to bind PZn-Ph-NDI in its interior.

Figure 1.

Pump-probe transient dynamics of PZn-Ph-NDI in 99:1 1,4-dioxane:N-methylimidazole. (A) Transient dynamical data at various pump-probe time delays, which are labeled on the right. Each spectrum is offset by 15 mOD for clarity. (B) Species-associated difference spectra (sans laser scatter at 560±5 nm) with associated ET time constants. Experimental conditions: λ_ex = 560 nm, P_ex = 850 nJ / pulse, T = 21 °C, magic angle polarization.

PZn-Ph-NDI exhibits signature photoinduced ET dynamics following preparation of the PZn-localized S₁ excited state, ¹(PZn)*-Ph-NDI (Figure 1A). In the low ε_s solvent system 99:1 1,4-dioxane:N-methylimidazole (NMI), ¹(PZn)*-Ph-NDI evolves to the charge-separated state ^+•PZn-Ph-NDI^•− with a charge separation (CS) time of τ_CS = 12 ps. Charge recombination (CR) to the ground state is comparatively slow: τ_CR = 102 ps. The ET dynamics are tracked by the rise and subsequent decay of sharp spectral features centered at 480 and 610 nm, characteristic of the NDI^•− radical anion.⁶ Species-associated difference spectra (Figure 1B), derived from a global fitting of the multi-wavelength transient absorption data to a sequential kinetic mechanism, also show clear signatures of the NDI^•− anion (red spectrum, prominent absorbance maxima at 480, 610, and 780 nm).

Figure 2 displays the CS and CR time constants of PZn-Ph-NDI measured in organic solvents with 1% NMI. CR dynamics (red) depend strongly on ε_s, with τ_CR decreasing monotonically from 102 ps to 4 ps. CS dynamics (blue) show a τ_CS minimum near ε_s = 9. Indeed, the observed trends for τ_CS and τ_CR are in agreement with those expected from the Marcus activated factor of ET rate theory^2a,7 (Figure 2, inset), using parameters appropriate for PZn-Ph-NDI (Supporting Information). While the ET dynamics of PZn-Ph-NDI displayed in Figure 2 approach the low ps time regime in some solvents, all time constants are well above solvent relaxation times, such that ET is not limited by dynamic solvent effects.⁸

Figure 2.

ET time constants τ_CS and τ_CR of PZn-Ph-NDI as a function of solvent static dielectric constant (ε_s). All solvents include 1% N-methylimidazole (NMI). ε_s is a volume-fraction weighted sum of component ε_s values. Shaded line widths indicate confidence intervals of the fitted time constants (bounded at 16% and 84% percentiles of the bootstrapped lifetime distributions). For the SCPZnI3 holo-monomer, τ_CR is denoted with a green X. Green vertical lines mark the range of dielectric constants consistent with the measured value of τ_CS in the SCPZnI3 holo-monomer. Purple arrows display an exemplary reduction of ET time constant magnitudes that occur upon NMI coordination, shown explicitly for anisole solvent. Inset: expected dielectric dependence of τ_CS (blue) and τ_CR (red) according to ET rate theory (see Supporting Information). DMM = dimethoxymethane, THF = tetrahydrofuran, 2-MeTHF = 2-methylTHF, DCM = dichloromethane, PhCN = benzonitrile, DMF = N,N-dimethylformamide, DMSO = dimethylsulfoxide.

NMI coordination to the PZn moiety accelerates CS and CR relative to dynamics evinced in neat solvents (Figure 2, purple, and Figure S3), resulting from the more negative oxidation potential of PZn when bound to the electron donating imidazole ligand.⁹ ET kinetics measured in solvents with 1% NMI provide a reference for comparison to kinetics measured within the protein environment of SCPZnI3, where PZn is coordinated by histidine. Accounting for axial coordination effects allows us to isolate the ε_s contribution to the ET dynamics within the SCPZnI3 protein.

The τ_CS of PZn-Ph-NDI exhibits a kinetic isotope effect (KIE) in 9:1 1,4-dioxane:H(D)₂O mixtures (Figure 3). The KIE = τ_CS(D₂O) / τ_CS(H₂O) decreases from 1.4 to 1.2 upon NMI coordination to PZn. Interestingly, although τ_CR decreases with increasing water content, consistent with the dielectric dependence in Figure 2, τ_CR does not display a KIE (Table S1, Figures S4–S7). These observations suggest that CS is more tightly coupled than CR to water H-bond interactions with the imide carbonyls.¹⁰ The τ_CS KIE is an experimental probe of PZn-Ph-NDI water solvation within ground-state SCPZnI3.

Figure 3.

Distributions of CS time constant τ_CS displayed by PZn-Ph-NDI within various solvation environments. Distributions are derived from bootstrapping the residuals of the global fit of the pump-probe data (see Supporting Information). Measurements in 1,4-dioxane contain 10% H₂O or D₂O. Measurements with SCPZnI3 were carried out in 50 mM NaPi / 150 mM NaCl buffer with 100% H₂O or D₂O. Confidence intervals in gray display 16% and 84% percentiles of the bootstrapped lifetime distributions. The optimum time constant is shown as the middle, vertical line. Black lines overlaying the bootstrap histograms are Gaussian fits to the distributions.

To explore ET dynamics in a protein environment, the tetra-α-helical protein SCPZnI3 was computationally de novo designed¹¹ to bind PZn-Ph-NDI in an interior consisting of hydrophobic amino acids (Figure 4 and Supporting Information). Electronic absorption and circular dichroism spectra of the complex in aqueous buffer confirm 1:1 binding of PZn-Ph-NDI to the interior of a helical protein and verify histidine coordination of the PZn moiety (Figure 4, B and C, Figures S21–S23). Thus the electronic environment of PZn-Ph-NDI is consistent with that of a protein interior.

Figure 4.

(A) Model of SCPZnI3 showing interior hydrophobic residues (green) surrounding PZn-Ph-NDI (orange). (B, C) Circular dichroism spectra of SCPZnI3 holoprotein. (B) Molar ellipticity per residue is consistent with a helical structure. (C) Molar ellipticity (Cotton effect) in visible region indicates achiral cofactor resides in a structured, chiral environment (see Figure S21).

The ET dynamics of PZn-Ph-NDI within the protein interior was monitored following electronic excitation of the PZn-Ph-NDI / SCPZnI3 complex in D₂O buffer and analyzed via decay-associated difference spectra derived from global fitting of the multi-wavelength transient data (Figure 5). The CR dynamics of PZn-Ph-NDI in SCPZnI3 are bi-exponential: we assign the dominant CR time τ_CR = 22 ps to the SCPZnI3 holo-dimer, and the CR time τ_CR = 69 ps to the SCPZnI3 holo-monomer, as their respective amplitudes track with the approximate 2:1 dimer:monomer ratio determined from analytical gel filtration (Supporting Information). CS occurs in monomer and dimer with the same time constant τ_CS = 7.5 ps. The overlapping distributions of τ_CS for SCPZnI3 / H(D)₂O buffer (Figure 3) do not exhibit a KIE, indicating the lack of a significant role played by water in the CS event. On the basis of this observation, we conclude that PZn-Ph-NDI is sequestered from solvent within a well-packed protein interior in both the dimer and monomer.

Figure 5.

Pump-probe transient dynamics of PZn-Ph-NDI / SCPZnI3 complex in D₂O buffer. (A) Transient dynamical data at pump-probe time delays labeled on the right. Each spectrum is offset by 10 mOD for clarity. (B) Decay-associated difference spectra (sans laser scatter at 560±5 nm) with associated τ_CR and τ_CS. D and M superscripts refer to dimer and monomer, respectively. Experimental conditions: λ_ex = 560 nm, P_ex = 920 nJ / pulse, T = 21°C, magic angle polarization.

By comparison with Figure 2, each ET time constant exhibited by PZn-Ph-NDI can be mapped to an effective ε_s. In the SCPZnI3 holo-monomer, CS and CR time constants (τ_CS = 7.5 ps, τ_CR = 69 ps) cannot be uniquely mapped to a single effective ε_s that jointly recovers both time constants. Two distinct ε_s values are needed (Figure 2, green). Accounting for the relative errors in the measured time constants (Supporting Information), as well as the non-monotonic nature of the CS dependence on solvent dielectric, the range of the effective ε_s that describes CS (τ_CS = 7.5 ps) is 8.3 to 13.4. We take ε_S^(CS) = 8.3 as a lower limit. The effective ε_s describing CR is well defined (ε_S^(CR) = 3.0 ± 0.1). Indeed, CS resembles that in 99:1 THF:NMI (ε_s ≈ 8), whereas CR resembles that in 99:1dimethoxy-methane:NMI (ε_s ≈ 3). Importantly, these assignments do not depend on any underlying model: they result from mapping dynamical data acquired for the SCPZnI3 holoprotein onto an extensive body of analogous data obtained for PZn-Ph-NDI in widely varying dielectric environments. While a protein interior is not a uniform dielectric material, these ε_S values reflect the effective dielectric probed by PZn-Ph-NDI buried in the SCPZnI3 interior, both pre- and post-charge separation.

How can the protein achieve such changes in dielectric on these fast (< 100 ps) ET time scales? These results suggest a rigidification of the protein sidechains upon formation of the ^+•PZn-Ph-NDI^•− CS state. Interestingly, the reduction in ε_S after the ET event suggests little solvational contribution to the CS state from polar water, despite the protein’s small size, and contrary to commonly accepted views.^1b The dielectric constant is a key parameter affecting CS and CR rates in biological proteins. A ε_S switching mechanism may warrant a reexamination of the energetic and electronic parameters derived from measurements of biological ET rates. Dielectric switching may provide a means to independently manipulate forward and back ET rates in natural catalytic proteins and artificial photosynthetic devices. Indeed, the ET time scales of PZn-Ph-NDI are similar to that of the primary ET in photosynthesis;¹² a decrease in protein ε_S upon the primary CS event would further slow wasteful inverted-regime CR in the photosynthetic reaction center.