Conformationally-Dynamic Radical Transfer within Ribonucleotide Reductase

Abstract

Ribonucleotide reductases (RNR) catalyze the reduction of nucleotides to deoxynucleotides through a mechanism involving an essential cysteine based thiyl radical. In the E. coli class 1a RNR the thiyl radical (C₄₃₉•) is a transient species generated by radical transfer (RT) from a stable diferric-tyrosyl radical cofactor located >35 Å away across the α₂:β₂ subunit interface. RT is facilitated by sequential proton-coupled electron transfer (PCET) steps along a pathway of redox active amino acids (Y₁₂₂β ↔ [W₄₈β?] ↔ Y₃₅₆β ↔ Y₇₃₁α ↔ Y₇₃₀α ↔ C₄₃₉α). The mutant R₄₁₁A(α) disrupts the H-bonding environment and conformation of Y₇₃₁, ostensibly breaking the RT pathway in α₂. However, the R₄₁₁A protein retains significant enzymatic activity, suggesting Y₇₃₁ is conformationally dynamic on the timescale of turnover. Installation of the radical trap 3-amino tyrosine (NH₂Y) by amber codon suppression at positions Y₇₃₁ or Y₇₃₀ and investigation of the NH₂Y• trapped state in the active α₂:β₂ complex by HYSCORE spectroscopy validate that the perturbed conformation of Y₇₃₁ in R₄₁₁A-α₂ is dynamic, reforming the H-bond between Y₇₃₁ and Y₇₃₀ to allow RT to propagate to Y₇₃₀. Kinetic studies facilitated by photochemical radical generation reveal that Y₇₃₁ changes conformation on the ns-μs timescale, significantly faster than the enzymatic k_cat. Furthermore, the kinetics of RT across the subunit interface were directly assessed for the first time, demonstrating conformationally dependent RT rates that increase from 0.6 to 1.6 × 10⁴ s⁻¹ when comparing wild type to R₄₁₁A-α₂, respectively. These results illustrate the role of conformational flexibility in modulating RT kinetics by targeting the PCET pathway of radical transport.

Graphical Abstract

INTRODUCTION

Many enzymes catalyze chemical reactions that involve the movement of charge. Electron transfer (ET) between donor-acceptor pairs within the protein milieu has been well described by a tunneling model, for which ET can occur >10 Å at rates that are competitive or exceed the requisite k_cat.^1,2 While this framework describes most ET reactions well with noted exceptions,^3,4 it does not address the coupling of an electron to a proton, as is obligatory for radical transfer (RT). In proton-coupled electron transfer (PCET),⁵ the wavefunctions of both the electronic and nuclear coordinates mix and accordingly, the very different length scales between proton and electron transfer result in a much greater sensitivity of a PCET pathway to distance as compared to an ET pathway. Proteins are soft materials and exhibit significant degrees of freedom, from global changes in quaternary structure and phonon-like collective motion⁶ to local amino acid sidechain flexibility. This conformational sampling alters charge transfer distances and whereas modest perturbations on ET rates are imposed relative to those predicted from the static structural models,¹ the more sensitive PT rates may be dramatically affected and consequently the thermodynamics and kinetics of PCET may be significantly influenced.^7–9 Defining the conformational landscape of redox active groups involved in PCET is therefore essential to understanding function, especially in those enzymes that derive their activity from radicals.

Endogenous amino acid residues within proteins and enzymes are the sites of RT. Due to the correlation between local structure (e.g., H-bonding) and PT, the reactivity of redox active amino acids is expected to be tuned by conformational motions owing to changes in PCET distance and potentially the identity of donor/acceptor sites for PT. Few studies, however, have directly addressed the role of either global or local conformational dynamics on the PCET chemistry of redox active amino acids, particularly with regard to the PT component.

Of the large and growing number of enzymes known to be dependent on redox active amino acids,^10–13 ribonucleotide reductase (RNR) is an enzyme that is conformationally dynamic and employs redox active amino acid residues to perform RT and redox catalysis over an exceptionally long distance. RNR catalyzes the conversion of nucleotides (CDP, GDP, UDP, and ADP) to deoxynucleotides, the bottleneck in de novo DNA synthesis in all living organisms.^10,14 Nucleotide reduction is regulated at a number of levels including allosteric regulation through effector binding sites which governs both specificity and overall activity. Nucleotide reduction is initiated by H-atom abstraction from a thiyl radical located on a conserved cysteine residue.¹⁵ The class 1a RNR of Escherichia coli (E. coli) is proposed to be a heterodimer of homodimers with an α₂:β₂ structure based on shape complementarity of the composite α₂ ¹⁶ and β₂ ¹⁷ X-ray structures, small angle X-ray scattering (SAXS) and electron microscopy (EM) measurements¹⁸ as well as pulsed electron-electron double resonance (PELDOR) spectroscopic studies.¹⁹ The active cysteine thiyl radical in α₂ (C₄₃₉• in E. coli numbering scheme) is generated transiently under turnover conditions by a stable diferric-tyrosyl radical cofactor (Y₁₂₂•) located in the β₂ subunit >35 Å away.^18,20 Several conserved aromatic amino acid residues are proposed to be involved in RT from Y₁₂₂β to C₄₃₉α, as summarized in Figure 1a (Y₁₂₂β ↔ [W₄₈β]? ↔ Y₃₅₆β ↔ Y₇₃₁α ↔ Y₇₃₀α ↔ C₄₃₉α).^20,21 This radical hopping pathway employs sequential PCET events,^7,21–23 and PT and ET are necessarily regulated to maintain RT fidelity, and therefore enzymatic activity.

Figure 1.

Radical transfer (RT) pathway, PCET mechanism and thermodynamic landscape of the E. coli RNR. A | Redox active amino acids involved in RT and presently understood PCET mechanism at each site. Of note are the disordered nature of Y₃₅₆β, and the R₄₁₁ sidechain (green) that controls the Y₇₃₁ conformation. Whereas the tyrosine and cysteine sites of the pathway have been shown to be intimately involved in RT, the participation of W₄₈β remains an open question. B | Proposed thermodynamic landscape as determined by equilibrium measurements. Tyrosine analogues 2,3,6-F₃Y (red) and NH₂Y (blue) are shown with their relative estimated reduction potentials (pH = 8.2 vs. Y).

The mechanism of PCET that enables RT has been the subject of extensive study.^20,22,24,25 Based on the expected distance between the stable Y₁₂₂• and Y₃₅₆ in the β subunit,²⁶ as well as the involvement of a water molecule bound to the di-iron cofactor as a proton donor during transient reduction of Y₁₂₂•,²⁷ an orthogonal PCET mechanism in β has been proposed,²¹ where PT and ET occur between distinct donor/acceptor pairs.^28,29 Conversely, direct hydrogen bonding interactions of Y₇₃₁, Y₇₃₀ and C₄₃₉, revealed by high field electron nuclear double resonance (ENDOR) spectroscopy^30,31 has led to the contention that RT within α₂ occurs via collinear PCET where protons and electrons are shuttled between adjacent pathway residues.²¹ At present the mechanism of RT across the α₂β₂ subunit interface remains unknown. Recent spectroscopic studies indicate no direct H-bond is formed between Y₃₅₆β and Y₇₃₁α suggesting orthogonal PCET with alternative proton acceptors.³² Thus, during RT, Y₇₃₁ may act as a “gating” residue that facilitates a mechanistic switch of PCET, but the origin of this switch and its potential implications remain to be established.

The mechanistic role of Y₇₃₁ is further complicated by additional studies that suggest the Y₇₃₁ conformation may be dynamic. In the X-ray structure of the 3-amino tyrosine (NH₂Y) substituted Y₇₃₀NH₂Y-α₂, significant electron density is observed in an alternative Y₇₃₁ conformation that differs from the π-π stacked “Y-Y dyad” conformation.³³ A similar observation was reported in the structure of the α₂ subunit of the class 1b enzyme from Salmonella typhimurium,³⁴ although all structures to date have been solved in the absence of β₂. In the E. coli class 1a RNR, the α/β interface is relatively weakly associated and thought to be solvated to some degree^32,35 suggesting that some heterogeneity may be possible. In addition, R₄₁₁ in α₂ has been suggested to directly H-bond to Y₇₃₁ or to order a water molecule(s) that is H-bonded to Y₇₃₁,^31,36 thus stabilizing the π-π stacked conformation of the Y–Y dyad and perturbing the collective dyad reduction potential.³⁷ When A is substituted in place of R₄₁₁, Y₇₃₁• does not form an H-bond with Y₇₃₀ and moves closer to the β subunit across the α:β interface, thereby breaking the cofacial stacking of the Y–Y dyad.³⁶ To reconcile this observation with the current model for RT by PCET, Y₇₃₁• must then “flip” back to the π-π stacked conformation, reversibly and in a kinetically competent fashion, to accept a proton and electron from Y₇₃₀ to continue RT.^33,36 While the observation of steady state enzymatic activity would suggest as much, direct experimental evidence for such conformational dynamics affecting PCET is lacking. More generally, experimentally addressing both PCET and conformational dynamics is a challenge of significant interest in biophysical studies as static structural models may not represent the dominant PCET active conformations involved in charge transfer.^38,39

Probing RT dynamics in RNR are difficult to isolate because (1) the rate determining conformational change required for RT initiation in β₄₀ and timescale of (PC)ET are disparate,^41,42 and (2) RT intermediates are only fleetingly populated owing to an uphill thermodynamic landscape (>200 mV, Figure 1b).^28,42

We now address the role of Y₇₃₁ as a critical linchpin in the dynamic conformational gating of the PCET pathway and RT in RNR by employing two complementary methodologies. First, we install the non-native tyrosine analogue 3-amino tyrosine (NH₂Y) by amber codon suppression technology at either Y₇₃₁ or Y₇₃₀ within the R₄₁₁A background. This residue acts as a radical trap (E vs. Y < −200 mV,⁴³ Figure 1b), allowing us to monitor H-bonding interactions at the phenoxy radical site directly by hyperfine sublevel correlation (HYSCORE) spectroscopy, thus revealing the conformational changes occurring upon RT. Due to the significant thermodynamic sink imposed by the 3-NH₂Y, these mutants are inactive or nearly inactive, but remain useful for examination of pathway radicals. Second, we employ the photoβ₂ methodology^44,45 in which radicals can be initiated on the RT pathway by photochemical oxidation of pathway Ys. In this methodology, radical generation is gated by a short laser pulse with sufficient time resolution to examine RT kinetics, unmasking RT from slower conformational gating processes. The photoβ₂ is furnished by covalently ligating a bromomethylpyridine rhenium tricarbonyl phenanthroline ([Re]) photo-oxidant to β₂ via a S₃₅₅C mutation that is directly adjacent to the RT pathway residue Y₃₅₆. By following [Re] emission and Y₃₅₆• decay kinetics via transient optical spectroscopy, we establish a lower bound for the rate of conformational dynamics and directly assess the rate constants for radical injection from Y₃₅₆β• to Y₇₃₁α. Our results demonstrate a direct link between conformational dynamics and PCET kinetics in RNR, resulting from the structural sensitivity of PCET to its strong dependence on distance and local protein environment. Dynamics such as those reported herein may not be apparent from equilibrium measurements, but are likely critical in many other enzymes that utilize amino acids that participate in PCET.

Results

NH₂Y labelled α₂ proteins were produced by amber codon suppression methods.⁴⁶ Photoβ₂s were produced as described previously^41,45 and installation of 2,3,6-F₃Y at position 356 was accomplished by amber codon suppression as described previously.³⁵ Detailed methods are provided in the Materials and Methods section of the SI. With labelled proteins in hand, the following experiments were performed where the R₄₁₁A-α₂ proteins are compared to the otherwise “native” α₂ proteins without the R₄₁₁A mutation. For the sake of clarity, we refer to the double mutant protein constructs containing the R₄₁₁A mutation with a prime (i.e. Y₇₃₁F vs. Y₇₃₁F′ (Y₇₃₁F with R₄₁₁A mutation); Y₇₃₀NH₂Y vs. Y₇₃₀NH₂Y′ (containing the R₄₁₁A mutation).

Radical Transfer Quantitation

X-band electron paramagnetic resonance (EPR) spectroscopy was performed to quantitate RT from Y₁₂₂• to the Y_731/730NH₂Y• radical trap in the presence and absence of the R₄₁₁A mutation. Mixing of WT-β₂ (1.2 Y•/β₂) with Y₇₃₀NH₂Y-α₂, CDP (substrate) and ATP (allosteric effector) followed by freeze quenching in liquid N₂ by hand after 10 seconds resulted in 48% conversion of Y₁₂₂• signal to Y₇₃₀NH₂Y•, as determined by subtraction of the Y₇₃₀NH₂Y• spectrum from the composite spectrum (52% Y₁₂₂•) and double integration of each component (Figure S1), similar to previous reports.^46,47 These values represent nearly stoichiometric RT (excluding 5–10% total spin loss determined from small scale EPR quantitation similarly as previously described⁴⁷) as the active α₂:β₂ complex exhibits “half-site reactivity” (Figure S1 schematic), where only one of the two Y₁₂₂• of each β₂ is translocated onto the RT pathway at a time.^19,48,49 The X-band EPR experiment was repeated for Y₇₃₀NH₂Y′-α₂, Y₇₃₁NH₂Y-α₂ and Y₇₃₁NH₂Y′-α₂; significant RT to the NH₂Y trap site (>25% spin transfer) and minimal spin loss (Figure S2) was observed for all experiments.

Hydrogen Bond Environment of Radical Trap

HYSCORE spectroscopy (Figures 2 and S3–S6) of Y_731/730NH₂Y• trap states was employed to examine the perturbation of the H-bond structure of the radical by the R₄₁₁A mutation. ENDOR spectra for Y₇₃₁NH₂Y•, Y₇₃₀NH₂Y• and Y₇₃₁NH₂Y•′ have been previously recorded^31,36 and the corresponding HYSCORE spectra in Figures 2A, 2B and 2C, respectively, are consonant with these ENDOR results. The ENDOR data, in conjunction with DFT modelling, revealed the number of H-bonds at each respective NH₂Y site and the associated hyperfine coupling (HFC) constants consistent with the collinear PCET pathway of α₂. The HYSCORE method provides a direct experimental measure of the A_x HFC by linear extrapolation without requiring DFT modelling. Additionally, the HYSCORE data provide an enhanced resolution of the electron-nuclear interactions of NH₂Y• with its H-bonds by spreading the spectrum out into 2D. The HYSCORE spectrum shown in Figure 2D reveals the NH₂Y environment of Y₇₃₀NH₂Y′, which has not been measured by ENDOR; we include the HYSCORE results for Figures 2A–2C for completeness and for comparison to 2D.

Figure 2.

X-band ¹H HYSCORE spectra of radical trapped Y_73xNH₂Y-α₂s generated by mixing with WT-β₂, CDP and ATP. A | HYSCORE spectrum of Y₇₃₁NH₂Y•-α₂. B | HYSCORE spectrum of Y₇₃₁NH₂Y•′-α₂. Expanded view of dashed region in panel A. Red arrow indicates H-bond assigned to Y730. Dashed region highlighted for comparison. C | HYSCORE spectrum of Y₇₃₀NH₂Y•-α₂. Schematic representation of NH₂Y₇₃₁• based on HYSCORE data. Unresolved H-bonds due to either water or R₄₁₁ are depicted in the scheme (brown, ref. 19). D | HYSCORE spectrum of Y₇₃₀NH₂Y′•-α₂. Dashed regions in panel (i) are expanded in panel ii. Arrows in panel ii indicate H-bonds depicted in the schemes of panel iii. Dashed circle in panels Aii and Bii indicate a lost H-bond between Y₇₃₁NH₂Y• and the phenolic hydroxide of Y₇₃₀ from Y₇₃₁NH₂Y• to Y₇₃₁NH₂Y′•.

Figures 2Ai and 2Bi (see panel ii for expanded view of dashed region of panel i) show the ¹H HYSCORE measurements on Y₇₃₁NH₂Y• and Y₇₃₁NH₂Y′•. The disappearance of the cross-ridge feature (red dashed elliptical region) in Figure 2Bii indicates the loss of an H-bond with Y₇₃₀ in the R₄₁₁A variant. The HFC constants determined from an analysis of the HYSCORE spectra in Figure 2Ai via linear regression in squared-frequency units (Figures S3–S6) give A_X = 10.4 MHz and A_Y/Z = −8.3 MHz, which are in excellent agreement with the previously reported ²H ENDOR parameters for the H-bond from the Y₇₃₀ hydroxyl (10.0 MHz and −8.5 MHz, respectively, after correcting for the gyromagnetic ratio of ¹H/²H = 6.51).^{31 2}H ENDOR also demonstrated the loss of this H-bond in the Y₇₃₁NH₂Y′• trapped state, reproduced in Figure 2B (see Fig. 2Aiii and 2Biii for schematic representation). The similarity of the results between HYSCORE and ENDOR experiments reinforce the HYSCORE method as a complementary tool for examining the H-bonding environment of NH₂Y radical trapped species.

The HYSCORE spectrum of Y₇₃₀NH₂Y• shown in Figure 2C reveals four distinct H-bonding interactions (Table S1): the 3-amino protons analogous to the Y₇₃₁NH₂Y (black arrow), the hydroxyl group of Y₇₃₁ (red arrow, HFC: A_X = 5.4 MHz and A_Y/Z = −7.8 MHz), the thiol of C₄₃₉ (green arrow, HFC: A_X = 1.4 MHz and A_Y/Z = −7.0 MHz), and a weaker hydrogen bond previously assigned to water (blue arrow, HFC: A_X = 3.5 MHz and A_Y/Z = −2.3 MHz). These HFCs again agree well with the values reported by ²H ENDOR (Y₇₃₁–OH: A_X = 4.6 MHz and A_Y/Z = −7.8 MHz, C₄₃₉–SH: A_X = 3.9 MHz and A_Y/Z = −6.7 MHz, and H₂O: A_X = 2.7 MHz and A_Y/Z = −1.4 MHz). In contrast to the previously published ENDOR data, all four H-bonds are clearly resolved without the need for modelling. The fact that the experimentally determined HFCs match well with the reported ENDOR results further validates those prior studies and the DFT model. Interestingly, the spectrum of Y₇₃₀NH₂Y′• in Figure 2C is essentially identical to that of Figure 2D, with nearly identical HFC parameters (Table S1). These data suggest that radical propagation to Y₇₃₀ is associated with conformational changes in Y₇₃₁ in R₄₁₁A-α₂, which re-establish the H-bonding interaction within the Y–Y dyad.

Radical Transfer Measurements

To examine the effect of the R₄₁₁A mutation on Y₁₂₂• reduction and NH₂Y• formation, we performed stopped flow UV-vis absorption, following mixing of Y₇₃₀NH₂Y(′) with WT-β₂. Figure 3 shows the UV-vis traces of the stopped flow experiments where Y₇₃₀NH₂Y and Y₇₃₀NH₂Y′ labelled α₂ are mixed with WT-β₂ in the presence of CDP and ATP. The optical signature of NH₂Y• (λ_max = 320 nm) is distinct from native (Y₁₂₂•) or adventitiously generated Y radical (λ_max = 410 nm).^36,46 In the stopped flow experiments, mixing of Y₇₃₀NH₂Y-α₂ or Y₇₃₀NH₂Y′-α₂ resulted in biphasic transfer of the radical species from Y₁₂₂(β) to Y₇₃₀NH₂Y(α) with rate constants analogous to previous measurements.^33,36 Fits of the data for Y₇₃₀NH₂Y• formation in Y₇₃₀NH₂Y-α₂ furnished apparent rate constants of 7.4(5) and 0.63(6) s⁻¹, representing 66% and 34% of the amplitude change respectively. The fast phase is similar to the overall steady state activity of 3–8 s⁻¹ for WT-α₂ and to prior measurements for this rate constant.^33,50 The apparent rate constants for Y₇₃₀NH₂Y′• formation were quite similar to the single mutant at 6.2(7) and 0.30(2) s⁻¹. In contrast to the single mutant Y₇₃₀NH₂Y, the relative amplitudes of the two phases were 46% and 54%, respectively, for Y₇₃₀NH₂Y′. The origin of biphasic RT kinetics is at present unclear, and therefore the difference in the relative amplitude of the two kinetic phases is not immediately obvious. Both the rate and relative amplitude of the Y₇₃₀NH₂Y• formation in the presence or absence of the R₄₁₁A mutation are recapitulated in the Y₁₂₂• decay kinetics of proteins (Figure 3, data summarized in the Tables S2).

Figure 3.

Stopped flow UV-vis kinetics of: formation of NH₂Y• (λ_det = 320 nm) in Y₇₃₀NH₂Y ( ) and Y₇₃₀NH₂Y′-α₂ ( ) mixed WT-β₂; and decay of Y₁₂₂• (λ_det = 410 nm) in Y₇₃₀NH₂Y ( ) and Y₇₃₀NH₂Y′-α₂ ( ) mixed WT-β₂. Formation and decay kinetics are described by a biexponential fit ( ). Inset shows expanded view of early time fits (shaded gray region). A 10 μM solution (6 mM ATP in assay buffer) of the α₂ variants were mixed with a 10 μM solution (2 mM CDP in assay buffer) of WT-β₂. Data is the average of at least three experimental runs.

Radical Transfer Measurements Probed by Photoβ₂s

Photoβ₂ allows for the “instantaneous” generation of radicals on the PCET pathway of RNR.^44,45 Photoinitiated generation of Y• allows the RT kinetics to be obtained for the different assumed conformations of Y₇₃₁. In this photoβ₂, the [Re] chromophore is covalently ligated to a modified β₂ containing a S → C mutation at position 355 via a methylpyridinyl linker (SI Materials and Methods), directly adjacent to the RT pathway residue Y₃₅₆. Here, absorption of a photon by the [Re] unit excites a metal-to-ligand charge transfer transition. The excited state can directly act as an oxidant (excited state reduction potential of 1.97 V) of the adjacent Y₃₅₆. Electron transfer from Y₃₅₆ to the [Re*] can be monitored by excited state emission of the [Re*] (either by emission intensity or by excited state lifetime) with the judicious choice of a reference. To this end, we employ a Y₃₅₆F substitution, which is redox inert, and offers a kinetic reference to interpret the subsequent emission quenching kinetics. By this methodology we have demonstrated that the [Re*] of the photoβ₂ in complex with α₂ is selective for Y₃₅₆ oxidation relative to Y₇₃₁ in WT α₂, as evidenced by the indistinguishable [Re*] decay kinetics. Given our HYSCORE data and the previously published PELDOR data, Y₇₃₁ may become susceptible to oxidation by the [Re*] excited state. Thus, the photoβ₂ system may be employed for oxidation of the adjacent Y₃₅₆(β) or direct Y₇₃₁(α) oxidation with utilization of Y₃₅₆F-photoβ₂, both of which are constituents of the native PCET pathway.

Figure 4 compares the fluorescence spectrum of Y₃₅₆Fphotoβ₂ in the presence of WT (blue dashed line) and Y₇₃₁F-α₂ (blue solid line). The slight decay in the fluorescence intensity from Y₇₃₁F to WT-α₂ suggests Y₇₃₁ oxidation is inefficient. Conversely, in the presence of R₄₁₁A-α₂ (green dashed line), Y₃₅₆F-photoβ₂ exhibits a dramatic decrease in emission intensity relative to Y₇₃₁F′-α₂:photoβ₂ complex (green solid line), indicating significant Y₇₃₁ oxidation relative to the WT system.

Figure 4.

Emission spectra (λ_exc = 355 nm) of Y₃₅₆F-photoβ₂ in the presence of WT (blue) or R₄₁₁A-α₂ (green) with an intact (dotted lines) or blocked (Y₇₃₁F or Y₇₃₁F′, solid lines) RT pathway.

Emission lifetime measurements (Figure S7 and Table 1) recapitulate the results of steady state emission measurements. Measurement of photoβ₂ in the presence of Y₇₃₁F-α₂ provides the reference point for emission decay (defined by rate constant k₀) of the [Re] excited state ([Re*]), in the absence of on-pathway RT. [Re*] emission lifetimes in the presence of WT and Y₇₃₁F-α₂ are statistically identical, suggesting no radical injection directly into α₂. For the R₄₁₁A mutants, Y₃₅₆Fphotoβ₂ in the presence of Y₇₃₁F′ again provides the k₀ reference for RT involving proteins with the R₄₁₁A mutation. The rate constant for RT (k_PCET) is obtained from the following expression,

Table 1.

[Re*] emission decay rates and PCET kinetics for Y₃₅₆Fphotoβ₂ in complex with various α₂ mutants.

α2	τobs/ns	κPCET/105 s−1
Y731F	727 (10)	τ0
WT	714 (13)	0.3 (3)
Y731F′	1,100 (7)	τ0
Y730F′	648 (20)	6.3 (5)
C439S′	629 (6)	6.8 (2)
R411A	594 (16)	7.7 (5)

$$k_{\text{PCET}}=\frac{1}{{\tau }_0}-\frac{1}{{\tau }_{\text{obs}}}=k_{\text{obs}}-k_0$$ (1)

where k_obs is the decay rate constant of [Re*] in which a portion of the RT pathway is intact. Rapid RT injection (k_PCET = 7.7 × 10⁵ s⁻¹) from [Re*] to Y₇₃₁ is observed in the R₄₁₁A variant. Furthermore, when pathway blocks Y₇₃₀F′ and C₄₃₉S′ were compared to Y₇₃₁F′ and R₄₁₁A, a systematic increase in k_PCET from Y₇₃₀F′ to R₄₁₁A (Figures S8 and S9 and Table 1) is observed. The increase in the quenching rate constant as the pathway is extended suggests that the radical generated on Y₇₃₁ is in rapid equilibrium along the RT pathway on the timescale of the [Re*] lifetime. That the radical is delocalized along a pathway under rapid conformational dynamics is supported by the observation of clear dCDP production over the background when the Y₃₅₆F-photoβ₂ was illuminated in the presence of CDP, ATP and R₄₁₁A-α₂ (α₂ with only R₄₁₁A single point mutation) but not Y₇₃₁F′ (Figure S10).

The F-block at Y₃₅₆ shows that [Re*] bypasses the natural PCET pathway by direct oxidation of Y₇₃₁ in α₂. To investigate the role of R₄₁₁ in modulating RT across the subunit interface along an intact pathway, emission lifetime measurements were repeated by utilizing 2,3,6-F₃Y₃₅₆-photoβ₂ (Figure 1B). 2,3,6-F₃Y₃₅₆-photoβ₂ was used as a proxy for Y₃₅₆-photoβ₂ because it offers the distinct advantages of: (1) being more efficiently oxidized photochemically since it remains deprotonated under the experimental conditions (pK_a = 7);³⁵ (2) being a stronger oxidant (by 180 mV with respect to Y at pH 8.2,⁵¹) resulting in higher efficiency radical injection into α₂; and, (3) displaying a significantly shifted phenoxy radical absorption peak, which can be probed selectively within the native tyrosine background (vida infra) while minimally perturbing the protein structure.^51–53 [Re]* lifetime measurements were performed for 2,3,6-F₃Y₃₅₆-photoβ₂ in the presence of Y₇₃₁F, WT, Y₇₃₁F′, Y₇₃₀F′, C₄₃₉S′ and R₄₁₁A-α₂ and the results for which are displayed in Figure S11 together with the lifetimes summarized in Table S3. Clear emission quenching is observed when comparing Y₇₃₁F or R₄₁₁A:Y₇₃₁F to WT or R₄₁₁A, respectively, but the results indicate that the delocalization occurs from the 2,3,6-F₃Y• across the α₂:β₂ protein interface. Taken together with the Y₃₅₆F-photoβ₂ emission lifetime measurements, these data demonstrate that radical equilibration occurs among the Y₃₅₆β-Y₇₃₁α-Y₇₃₀α RT pathway against the R₄₁₁A background on a timescale which competes with [Re*] radiative decay. Abstracting direct information about the kinetics of RT event across the α₂:β₂ subunit interface is not possible with this data alone due to the kinetic complexity of the coupled processes, namely radical equilibration, quenching and substrate activation.

Transient Absorption Kinetics of RT

To address the role of the Y₇₃₁ conformational landscape on RT across the α₂:β₂ subunit interface, transient absorption (TA) experiments were performed to examine directly the photo-generated 2,3,6-F₃Y₃₅₆• and its subsequent decay kinetics. As noted above, 2,3,6-F₃Y• tyrosyl radical spectrum is red shifted ~10 nm from that observed from Y• (Figure S12) allowing the kinetics of 2,3,6-F₃Y₃₅₆ to be monitored independently from on and/or off pathway Y• species. Because charge recombination with the oxidized [Re^II] photosensitizer precludes build-up of an observable transient signal, we employed oxidative quenching (flash-quenching) by addition of Ru^III(NH₃)₆ to extend the lifetime of the [Re^II] oxidant and prevent back ET.⁴⁴ Transient absorption (TA) spectra recorded 1 μs after excitation of 2,3,6-F₃Y• alone and in the presence of WT, R₄₁₁A-α₂ and Y₇₃₁F′-α₂ are shown in Figure S13. In all cases, a 2,3,6-F₃Y• spectrum is observed, corresponding to 4% quantum yield. Comparison of 2,3,6-F₃Y• spectra from free 2,3,6-F₃Y₃₅₆ photoβ₂ and in complex with the various α₂ reveals a shift in the absorption spectrum of the radical, consistent with the emission spectrum of [Re] and anticipated electrostatic perturbations, where solvent exclusion by complexation with α₂ as well as elimination of charge in the R₄₁₁A mutants result in blue shifts of the 2,3,6-F₃Y• resonance (Figure S13). Clear interpretation of the TA spectra to discern the kinetics for RT across the α₂:β₂ interface are precluded by the complexity of the RT process, which stems from the nearly isopotential and reversible RT pathway within α₂ (see SI). The use of NH₂Y as a radical sink that greatly simplifies the kinetic description of the system, affording direct quantitation of k_PCET across the subunit interface (vide infra).

The TA kinetics of the 2,3,6-F₃Y• were measured in the presence of Y₇₃₁F-α₂ and the lifetime decay of 14.7(7) μs yielded k₀ = 6.8(3) × 10⁴ s⁻¹ (Table 2 and Figure S14), which provides a reference for the decay (defined by rate constant k₀) of the radical in the absence of on-pathway RT. In the presence of Y₇₃₀NH₂Y-α₂, the 2,3,6-F₃Y• decays with a lifetime of 13.5(2) μs yielding k_obs = 7.4(1) × 10⁴ s⁻¹. Substitution of these two rate constants into eq (1) yields the rate constant for interfacial RT of k_PCET = 6(3) × 10³ s⁻¹, which is the first measured rate constant for Y–Y radical exchange within a protein environment and across a protein-protein interface. The 2,3,6-F₃Y₃₅₆• decay was also measured for Y₇₃₁F′ and Y₇₃₀NH₂Y′ (Figure 5), yielding lifetimes of 17.9(6) and 13.9(4) μs corresponding to rate constants of 5.6(2) × 10⁴ s⁻¹ and 7.2(2) × 10⁴ s⁻¹, respectively. Again, solving for k_PCET according to eq (1) yields k_PCET = 1.6(3) × 10⁴ s⁻¹, which is modestly faster than the WT system and consistent with the R₄₁₁A mutation facilitating RT across the interface.

Table 2.

Decay kinetics of 2,3,6-F₃Y₃₅₆• measurements from transient absorption.

α2	τobs (μs )	κPCET (104 s−1)
Y731F	14.7(7)	k0
Y730NH2Y	13.5(2)	0.6 (3)
Y731F′	17.9(6)	k0
Y730NH2Y′	13.9(4)	1.6 (3)

Figure 5.

TA kinetic traces for 2,3,6-F₃Y₃₅₆photoβ₂ in complex with Y₇₃₁F′ (red) and Y₇₃₀NH₂Y′ (blue) with 1 mM CDP, 3 mM ATP and 10 mM Ru(NH₃)₆Cl₂ between 1–90 μs after excitation. Mono-exponential fits to the data are displayed in black.

Discussion

The role of conformational dynamics in PT and ET is largely unaddressed, though often essential to describe accurately protein or enzyme function.^9,40,54–56 Conformation is particularly pivotal to PCET as PT is acutely sensitive to small changes in donor-acceptor distance, which can be modulated by conformational dynamics. Indeed, the RT pathway of RNR is finely tuned to control PCET. The single point mutation R₄₁₁A in α₂ has been demonstrated to affect the conformation of a critical RT pathway residue, Y₇₃₁(α), breaking the π-π stacking of the Y₇₃₁–Y₇₃₀ dyad, which is required for efficient collinear PCET.³⁶ R₄₁₁A-α₂ maintains nucleotide reductase activity with a specific activity of 470(20) nmol min⁻¹ mg⁻¹ (1.3 s⁻¹), 17% that of WT enzyme.³⁶ The retained enzymatic activity suggests that the conformation of Y₇₃₁ is dynamic during RT, providing an impetus to address directly the conformational flexibility of Y₇₃₁ during RT and how conformational flexibility affects PCET kinetics across the α₂:β₂ protein interface.

The H-bonding environment of trapped NH₂Y• radicals has been shown by high field electron spin resonance spectroscopy to report on the structural consequences of conformational perturbations.^30,31,36 ENDOR data, in concert with PELDOR measurements have probed the radical environment at positions Y₇₃₀ and Y₇₃₁ in the native enzyme,^19,30,31 and the perturbed conformation of Y₇₃₁ in the R₄₁₁A mutant (Figure 1a).³⁶ The HYSCORE experiments reported here are complementary to ENDOR. In HYSCORE, hydrogen bonds appear as cross-ridges in a 2D correlation pattern resulting from electron-nuclear spin dipolar interactions. Tyrosyl radicals in both photosystem II and bovine liver catalase have been probed successfully by HYSCORE demonstrating that this technique is well suited to examine NH₂Y• trap states in RNR.⁵⁷

The HYSCORE spectrum of Y₇₃₁NH₂Y• shown in Figure 2A reproduces the ENDOR observation of an H-bond from Y₇₃₀ with similar HFCs. The absence of the cross-ridge pattern in the Y₇₃₁NH₂Y′• spectrum shown in Figure 2B (indicated by red ellipse) is consistent with a loss of H-bonding between Y₇₃₁• and Y₇₃₀, as has also been observed by ENDOR.³⁶ The PELDOR measurements of the Y₇₃₁NH₂Y•(α)-Y₁₂₂•(β) distance establish that the loss of this H-bonding is due to a conformational change induced by the R₄₁₁A mutation in which Y₇₃₁ “flips-out” to the α₂:β₂ interface, as shown schematically in Figure 2Biii.³⁶ The flipped-out conformation of Y₇₃₁ is manifested in the steady-state emission spectra of Figure 4 and lifetimes listed in Table 1. In the flipped-out conformation imposed by the absence of the guanidine moiety of R₄₁₁, Y₇₃₁ is more susceptible to photochemical oxidation by proximity to the [Re] unit of the photoβ₂. Accordingly [Re*] emission is quenched to a greater extent in R₄₁₁A.

Whereas the ENDOR spectrum of Y₇₃₀NH₂Y• has been reported, and is reproduced by the HYSCORE results in Figure 2C, the hydrogen-bonding environment of Y₇₃₀NH₂Y′• has not been measured. The striking result for Y₇₃₀NH₂Y• and Y₇₃₀NH₂Y′• (Figure 2D) reported here is that their HYSCORE spectra are essentially identical. These results suggest that π-π stacking of the radical in the Y–Y dyad is reformed when the radical advances to position 730 from 731 and that the H-bond between OH phenol groups flips from Y₇₃₁←Y₇₃₀ (Figure 2Aiii) to Y₇₃₁→Y₇₃₀ (Figure 2Ciii) and is maintained when the R₄₁₁A mutation is present (Figure 2Diii). This interconversion of conformations for the forward propagation of the radical is consistent with the π-π stacking enabling collinear PCET and is dynamic in nature as R₄₁₁A exhibits turnover for nucleotide reduction.

Stopped flow kinetics of Y₁₂₂• reduction and NH₂Y• formation upon mixing WT-β₂ and Y₇₃₀NH₂Y or Y₇₃₀NH₂Y′-α₂ (Figure 3) show no difference in the rate constants of RT to Y₇₃₀NH₂Y. When juxtaposed against the distinct conformational environments of Y₇₃₁ observed by HYSCORE, the lack of significant temporal differences in RT from Y₁₂₂•(β) to Y₇₃₀NH₂Y(α) or Y₇₃₀NH₂Y′ (α) in stopped-flow experiments (Figure 3, Table S2) suggests that the Y₇₃₁ conformational dynamics are masked on the turnover (millisecond) time scale and therefore the more rapid kinetic method offered by photoβ₂ was employed to examine the role of conformational motion directly.

Pathway radicals were generated using photoβ₂ implemented in two ways; injecting a radical directly into α₂ (with a Y₃₅₆F-photoβ₂ described above), or via radical generation at Y₃₅₆ of the RT pathway in β₂ (with 2,3,6-F₃Y₃₅₆-photoβ₂). Whereas, photochemical generation Y₃₅₆•(β) by the photoβ₂ has been examined extensively and occurs on the microsecond timescale,⁵⁸ the direct generation of Y₇₃₁•(α) to bypass Y₃₅₆ has been achieved by the introduction of [Re] at position 356.⁵⁹

Figure S8–S10 reveals direct radical injection within the Y₃₅₆F-photoβ₂:R₄₁₁A-α₂ complex and that it can support enzyme turnover. Consistent with the steady-state quenching experiments, transient emission decay kinetics measurements (Figure S7 and Table 1) show that the direct generation of a pathway radical from Y₃₅₆F-photoβ₂ only occurs significantly when Y₇₃₁ assumes a flipped-out conformation. Interestingly, the k_PCET rates in Table 1 follow a Dutton scaling relationship¹ where [Re*]-Y₇₃₁ distances can be calculated based on the known ΔG and λ of the reaction in both flipped up and flipped out conformations. The distances calculated from this analysis provide an estimate of the conformational displacement of Y₇₃₁ following flipping of >2.2 Å (SI), a scalar distance relative to the [Re] unit, which is in accordance with distances obtained from PELDOR measurements and from the two distinct Y₇₃₁ conformations observed in the Y₇₃₀NH₂Y α₂ structure.³⁶ This result suggests that the mechanism of PCET for Y₇₃₁ at the interface is ET/PT. We also observe that k_PCET increases in the series Y₇₃₀F′ < C₄₃₉S′ < R₄₁₁A (Table 1, Figure S9) indicating that the hole delocalizes over the RT pathway. If it were delocalized, we would expect nearest-neighbour quenching with no dependence on where the pathway block is located. Hole delocalization among RT pathway residues beyond the flipped out Y₇₃₁ on the timescale of the [Re*] lifetime necessitates electronic communication, which further supports rapid dynamics between Y₇₃₁ in a flipped-out and π-π stacked conformation. Rapid tyrosine sidechain motions (ns-μs) have been observed by NMR and supported by computation.^60–62 Thus, although the R₄₁₁A mutation shifts the equilibrium conformation of Y₇₃₁ towards the “flipped-out” conformation, rapid thermal conformational sampling allows RT to be propagated through α₂, stabilized by the H-bonding interaction with Y₇₃₀•, to the active site of nucleotide reduction.

Because Y₃₅₆ is spectrally indistinct from Y₇₃₁ and Y₇₃₀, radical injection through position Y₃₅₆ is facilitated with the use of fluorotyrosines.⁵² The spectroscopically distinct 2,3,6-F₃Y₃₅₆-photoβ₂ as a probe and Y₇₃₀NH₂Y as a radical sink in α₂ simplifies the kinetics of RT because a steady-state approximation may be imposed at Y₇₃₁ according to the following RT equilibrium,

$$2,3,6-\underset{k_{\mathit{0}}\downarrow }{{\mathrm{F}}_3{\mathrm{Y}}_{356}^{•}}\underset{k_{-\mathit{1}}}{\overset{k_{\mathit{1}}}{\rightleftarrows }}{\mathrm{Y}}_{731}\underset{k_{-\mathit{2}}}{\overset{k_{\mathit{2}}}{\rightleftarrows }}{\text{NH}}_2{\mathrm{Y}}_{730}\underset{k_{-\mathit{3}}}{\overset{k_{\mathit{3}}}{\rightleftarrows }}{\mathrm{C}}_{439}\underset{k_{-\mathit{4}}}{\overset{k_{\mathit{4}}}{\rightleftarrows }}\text{NDP}$$

In this equilibrium, the disappearance of 2,3,6-F₃Y₃₅₆• is given by,

$$\frac{\mathrm{d}[{\mathrm{F}}_3{\mathrm{Y}}_{356}•]}{\mathrm{d}t}=-(k_1+k_0)[{\mathrm{F}}_3{\mathrm{Y}}_{356}•]+k_{-1}[{\mathrm{Y}}_{731}•]$$ (2)

Due to the very slow rate, k^-2 (endergonic, ΔE > 300 mV, see Figure 1b), and k₃ (endergonic, ΔE > 300 mV), relative to the expectedly large k₂, the steady-state approximation for d[Y₇₃₁•]/dt (see SI) reduces Eq (2) to

$$\frac{\mathrm{d}[{\mathrm{F}}_3{\mathrm{Y}}_{356}•]}{\mathrm{d}t}=-(k_1+k_0)[{\mathrm{F}}_3{\mathrm{Y}}_{356}•]$$ (3)

which is effectively a reformulation of Eq (1), used to determine k_PCET. Thus, the rate of F₃Y₃₅₆• decay can be described accurately by considering only k₁, and k₀, whereupon knowing k₀ (determined by measuring d[2,3,6-F₃Y₃₅₆•]/dt in the presence of Y₇₃₁F or Y₇₃₁F′-α₂) one can solve for k₁ directly by measuring d[2,3,6-F₃Y₃₅₆•]/dt in the presence of Y₇₃₀NH₂Y or Y₇₃₀NH₂Y′-α₂, a measure of interfacial PCET. Y₃₅₆ oxidation occurs through an orthogonal PCET mechanism likely involving PT to water at the interface. An analogous mechanism is assumed for Y₇₃₁ oxidation, and the change in conformation of Y₇₃₁ may result in different PCET behaviour related to donor-acceptor distance in orthogonal PT and ET components. Our results demonstrate that k_PCET from 2,3,6-F₃Y₃₅₆β to Y₇₃₁α is enhanced in the presence of the “flipped out” conformation engendered by the R₄₁₁A mutation. Therefore, the conformational flipping in R₄₁₁A α₂ again appears rapid, enhancing RT between Y₃₅₆β and Y₇₃₁α. We speculate that the modest rate enhancement afforded by the flipped out conformation is likely due to the increased proximity of the donor acceptor pair relative to WT, analogous to the rate increase of Y₇₃₁ oxidation observed in the [Re*] quenching experiments. While the R₄₁₁A mutation enhances the interfacial RT rate, this mutation also destabilizes the α₂:β₂ protein complex. Since RT is not rate limiting, evolution appears to have favoured increasing binding affinity relative to enhancing RT. That the rate is not significantly slower in the flipped-out conformation also implies that Y₇₃₁• rapidly changes conformation to funnel the radical towards Y₇₃₀NH₂Y and conformational interconversion is not rate limiting.

Conclusions

Probing conformational dynamics and their effects on charge transfer remains a challenge in the field of biological charge transfer. Conformational control of charge transfer is particularly critical to biological PCET owing to the sensitivity of proton transfer to distance. Photoβ₂ constructs, together with advanced spectroscopic biophysical techniques such as HYSCORE, offer a powerful tool to correlate PCET with conformational dynamics in RNR, which is an exemplar of enzymatic activity derived from PCET. Herein, we demonstrate the first observation of radical transport across the α₂β₂ subunit interface in RNR, and that local conformational dynamics of a tyrosine (Y₇₃₁ α) sidechain during RT in a mutant construct of the class 1a RNR of E. coli significantly affects the PCET kinetics of this interprotein radical transfer. Dynamics analogous to those demonstrated in this report may affect other biological charge transfer reactions where protein based amino acids function in PT and/or ET by modulating donor-acceptor distances or identities.

ABBREVIATIONS

ATP

adenosine triphosphate

CDP

deoxycytosine diphosphate

dCDP

deoxycytosine diphosphate

E. coli

Escherichia coli

EM

electron microscopy

ENDOR

electron-nuclear double resonance

ET

electron transfer

EPR

electron paramagnetic resonance

F_nY

n-fluoronated tyrosine

HFC

hyperfine constant

HYSCORE

hyperfine sublevel correlation

NH₂Y

3-aminotyrosine

PCET

proton-coupled electron transfer

PELDOR

pulsed electron double resonance

PT

proton transfer

[Re]

rhenium 1,10-phenanthroline tricarbonyl, 3-bromoethylpyridine

RNR

ribonucleotide reductase

RT

radical transfer

SAXS

small angle X-ray scattering

TA

transient absorption

WT

wild type