Direct interfacial Y731 oxidation in α2 by a photoβ2 subunit of E. coli class Ia ribonucleotide reductase

Proton-coupled electron transfer (PCET) is a fundamental mechanism important in a wide range of biological processes including the universal reaction catalysed by ribonucleotide reductases (RNRs) in making de novo, the building blocks required for DNA replication and repair.


Introduction
Enzymatic control of coupled proton and electron transfer 1-4 is critical in managing biological processes ranging from energy storage (photosystem II) [5][6][7][8][9] and conversion (cytochrome c oxidase) 10 to the synthesis of DNA precursors (ribonucleotide reductase). [11][12][13][14] To better understand biological PCET, we have undertaken studies of this process in class Ia RNRs, which catalyse the conversion of nucleoside diphosphates (NDPs) to deoxynucleoside diphosphates (dNDPs)-a process required for synthesis and repair of DNA in all organisms. 15,16 Catalysis by the class I RNRs proceeds by a radical mechanism requiring coupling of radical transport over 35Å involving PCET across the two subunits to substrate turnover. The long distance, reversibility, and rate-limiting conformational gating of radical transport have made study of this process challenging. To overcome this challenge, we have developed two methodologies: photoRNRs [17][18][19][20][21] and site-specic incorporation of unnatural amino acids in place of pathway residues. [22][23][24] E. coli class Ia RNR has served as the paradigm for this long distance radical transport. It is composed of two homodimeric subunits: a 2 and b 2 . A docking model for this complex, 25 substantiated by recent biochemical and biophysical studies, 26,27 has provided the working model for the radical transport pathway shown in Fig. 1. The active site for NDP reduction resides in a 2 , where the cysteine radical (C 439 c) must be transiently generated during each turnover by the essential diferric-tyrosyl radical (Y 122 c) cofactor in b 2 . This long range oxidation requires a multi-step radical hopping mechanism that involves a specic pathway including four tyrosines (Y 122 and Y 356 in b 2 ; Y 731 and Y 730 in a 2 ) 11,28 and potentially W 48 in b 2 . 11 Recent attention has focused on the detection of the proposed transient radical intermediates and identication of the operative PCET mechanism at each site. Mössbauer studies have established that Y 122 c reduction in b 2 is triggered by binding of substrate and effector to a 2 29 and involves proton donation from the water at Fe 1 (Fig. 1). This process involves orthogonal PCET wherein the proton and electron come from different residues. High-eld electron paramagnetic resonance (Hf EPR) and deuterium electron nuclear double resonance (ENDOR) have provided atomic level resolution of local hydrogen bond interactions, specically the co-linearity of the PCET within a 2 . Additionally, signicant shis in g x values together with the assignment of hyperne coupling features from the ENDOR spectra of various amino-substituted RNR mutants propose an important role for electrostatics at the a 2 :b 2 interface. 30 However, the disordered C-terminal tail of b 2 where Y 356 resides has made interrogation of the chemistry at the subunit interface challenging (Fig. 1).
Rate limiting conformational gating in RNR obscures radical transport across the subunit interface, prompting us to develop photoRNRs to trigger radical initiation with light to avoid this gating and to potentially enable the observation of Yc at the interface. Radical injection kinetics were initially made possible using a 19mer peptide photoRNR, which corresponded to the identical 19 residues of the C-terminal tail of b 2 along with a modication that appended a photooxidant (rhenium phenanthroline [Re]) adjacent to Y 356 or uorinated derivatives. This peptide photoRNR enables nucleotide reduction in the presence of a 2 and light and allows for observation of radical injection into a 2 . 21 Radical injection was only realized in the presence of an intact Y 731 -Y 730 dyad within a 2 , providing important support for co-linear PCET within this subunit.
More recently, photoRNRs have been generated in which the peptide with the photooxidant is replaced by the full-length b 2 containing a site-specically incorporated [Re] photooxidant at residue 355 using a S 355 C-b 2 mutant. 31  Additionally, by leveraging the greater acidity of 2,3,5-F 3 Y to enable deprotonation at neutral pH, this residue furnishes an ionizable reporter that varies with experimental pH. In turn, siteselective removal of a single proton at position Y 731 (a) provides the rst protein:protein scaffold of RNR that permits the investigation of the effect of a modied proton microenvironment on radical transport on transient time scales (sub ms) at the interface. In the absence of Y 356 , radical injection is only achieved when position Y 731 is deprotonated. In addition to con-rming the complexity of RNR in maintaining a well-organized PCET pathway, this work introduces and highlights the importance of a well-dened proton exit channel out of a 2 involving the key pathway residues, Y 356 and Y 731 , at the subunit interface.

Experimental
Modied RNR subunits were constructed, expressed, puried, modied, and characterized as previously reported or with Fig. 1 Current model of radical transport pathway in class Ia RNR leading to nucleotide reduction as determined by the docking model 16 and diagonal distance measurements acquired by PELDOR spectroscopy. 34 Key redox active amino acids and known distance measurements involved in PCET pathway are shown. Residue W 48 is grayed indicating the absence of experimental evidence supporting its participation in the PCET radical mechanism. Residue Y 356 is shown at the interface for illustrative purposes. Salmon arrows indicate electron transfer (ET) and green arrows represent proton transfer (PT). In a 2 , co-linear PCET is denoted by the dual-colored arrow, and the proposed bi-directional PCET in b 2 is indicated by the orthogonality of the ET and PT pathways. minor modication. 19,23,24,31,35 Protein concentrations were measured by absorbance at 280 nm using: Purity of protein constructs was assessed by SDS-PAGE (Fig. S1 †). All measurements were conducted in assay buffer at pH 7.6 (50 mM HEPES, 15 MgSO 4 , 1 mM EDTA; unless otherwise specied). Measurement of the dissociation constant (K D ) between [Re 356 ]-b 2 and wt-a 2 was performed by a spectrophotometric competitive inhibition assay as previously reported. 17,21 Measurement of the pK a of the phenolic proton of 2,3,5-F 3 Y 731 within the assembled 2,3,5-F 3 Y 731 -a 2 :[Re 356 ]-b 2 complex was performed by uorometric titration as previously reported. 21 The details of methods that deviate from published procedures are provided in the ESI. † Similarly, photoinitiated nucleotide reduction activity assays were performed according to published methods. [17][18][19]32 Error bars represent 2s resulting from photolysis on $three independent samples.
Time-resolved spectroscopic measurements were performed using a home-built nanosecond laser system previously described. 21,[31][32][33] Each sample was prepared prior to photolysis and measurements were performed in triplicate. The calculation of the uncertainty in experimental measurements to 95% condence limits (2s) is described in the ESI (eqn S1-S5 †). To probe radical initiation across the a 2 :b 2 interface, specic variants of each subunit were required. To directly target the intersubunit radical transport step of Y 731 oxidation and subsequent radical injection into a 2 , we chose to circumvent Y 356 oxidation entirely. In contrast to previous systems where photooxidants were placed adjacent to Y 356 at position 355, [Re] replaces Y 356 in this study. The new construct maintains the mutations C 268 S and C 305 S, and preserves catalytic activity. Additionally, the mutation, Y 356 C, thus enables alkylation with [Re]-Br to yield [Re 356 ]-b 2 . To examine the effect of a proton at position Y 731 , this residue was replaced with 2,3,5-F 3 Y, solution pK a ¼ 6.4, compared with Y (pK a ¼ $10). The photoRNR construct is illustrated in Fig. 2. These two subunit modications allow for direct oxidization of Y 731 by a photob 2 .

Results and discussion
Construction, expression, isolation, and labelling to generate [Re 356 ]-b 2 were performed as previously reported with minor modications. 31 Unlabelled and reconstituted Y 356 C-b 2 (holo) is inactive (0.16(5) U) towards nucleotide reduction (wt-b 2 activity ¼ 6000-8000 U), as expected given the absence of Y 356 . We note that labelling does not preclude binding (K D , ¼ 0.43 (11) mM), as measured in the competitive inhibition assay shown in Fig. S2. † This value is in agreement with previously reported values for active photoRNR b 2 mutants, 32 and not signicantly altered from wt-RNR. 12,36 [  (Fig. S3 †).
Construction of 2,3,5-F 3 Y 731 -a 2 was achieved using in vivo nonsense suppression methodology previously developed to install uorotyrosine reporters in class Ia RNR. 24 The specic activity of 2,3,5-F 3 Y 731 -a 2 under catalytic conditions (375(5) U) is diminished relative to wt-a 2 (1800-2500 U), while under singleturnover activity (2.75(4) equiv. dCDP/a 2 ) is comparable to wt-a 2 ($3 dCDP equiv./a 2 ). This decrease in catalytic activity, though comparable to that of wild-type (wt), is consistent with previous reports of this mutant. 37 The deprotonation of 2,3,5-F 3 Y 731 is expected to perturb the rate of radical generation at position Y 731 . The kinetic penalty associated with proton transfer is alleviated by removal of the proton when experimental pH > pK a , thus requiring measurement of the precise pK a of 2,3,5-F 3 Y 731 in the assembled construct. As previously reported, [19][20][21]32,33,38 the increase in the rate of photooxidation for tyrosinate relative to tyrosine, 39 makes [Re] emission a reporter of the protonation state of nearby tyrosine residues. Fluorometric titration of the [Re 356 ]b 2 :2,3,5-F 3 Y 731 -a 2 complex reveals the pK a of 2,3,5-F 3 Y 731 to be 6.7(1), Fig. 3. Accordingly, $90% of 2,3,5-F 3 Y 731 residues are deprotonated under experimental conditions at the optimal operating pH for RNR (pH 7.6). Given the thermodynamically unfavourable acidity of the tyrosyl radical cation (pK a ¼ À2) a PCET process managing both the electron and proton transfers is mandated. 40,41 Photoinitiated substrate turnover To establish that the photoRNR construct is competent to generate dCDP, the [Re 356 ]-b 2 :2,3,5-F 3 Y 731 -a 2 complex in the presence of [ 3 H]-CDP and ATP was photolyzed for 10 min (l > 313 nm) and dCDP was measured by scintillation counting. The results of this single turnover experiment are shown in Fig. 4. Perturbation of the enzyme by the introduction of [Re] results in a reduced level of turnover that is 5-10% relative to wt-RNR under the same pH conditions. Notwithstanding, the presence of photogenerated products establish the relevance of [Re 356 ]-b 2 :2,3,5-F 3 Y 731 -a 2 complex to the natural enzyme. Attenuated enzymatic activity is also detected at pH > 7.6 which is consistent with the observed pH rate proles for the wt enzyme and uorotyrosine derivatized RNR constructs. 42

Radical injection kinetics
The photogeneration of product for the [Re 356 ]-b 2 :2,3,5-F 3 Y 731a 2 complex prompted us to undertake radical injection kinetics studies. Using the [Re]* emission lifetime as a reporter for radical injection, ns TA laser spectroscopy on the [Re 356 ]-b 2 :a 2 complexes in the presence of CDP and ATP was conducted. The emission decay lifetimes for each construct were measured at pH ¼ 7.6, where maximum turnover was observed, and are summarized in Table 1; representative traces are included in Accordingly, this quenching rate constant, k q , is equivalent to the radical generation rate, and from eqn (1) it is calculated to Samples were illuminated with l exc ¼ 315 nm, scanned from 420-700 nm at 0.5 nm intervals, integrated for 1 s at each data point, and averaged from three scans. The collected emission plots were integrated for fluorescence intensity and plotted against pH. Data were fit to an internal sigmoidal logistic function in OriginPro 8.5. The inflection point of the monoprotic pH-titration curve (x 0 ) is the pK a :  a Lifetime of emission decay measured on 10 mM [Re 356 ]-b 2 , 25 mM a 2 (as indicated), 1 mM CDP, 3 mM ATP in assay buffer (pH 7.6), l exc ¼ 355 nm, l obs ¼ 600 nm. Errors shown in parentheses represent 2s resulting from measurement on $3 independent samples. b Emission quenching rate constant, k q , determined from eqn (1). Error in quenching rate constants calculated as shown in ESI. 2) Â 10 4 s À1 ). This acceleration of radical injection into a 2 is in accordance with the differing protonation states of tyrosine in the two constructs. Under the experimental conditions of pH ¼ 7.6, 2,3,5-F 3 Y 731 is deprotonated and hence quenching occurs by ET rather than PCET, which results in faster tyrosine oxidation, despite being $50-100 mV more difficult to oxidize than native tyrosine at this pH (as determined from solution peak potentials (E p ) of N-acetyl and C-amide protected uorotyrosines measured by differential pulse voltammetry). 22 For clarity, obtaining precise single reside midpoint potentials in protein constructs is extremely challenging, and thus thermodynamic considerations must be guided from these measured E p values, 11 despite recent ndings that suggest signicant deviations of the formal reduction potential and solution E p values. 43 To investigate how the proton at position Y 731 affects interfacial PCET and subsequent interfacial radical injection kinetics, emission quenching of the [Re]* within the [Re 356 ]b 2 :2,3,5-F 3 Y 731 -a 2 complex was monitored over the activity accessible pH region of RNR. A plot of the [Re]* decay lifetimes for [Re 356 ]-b 2 alone and in the three [Re 356 ]-b 2 :a 2 variant complexes monitored as a function of pH are shown in Fig. S5; † representative emission decay traces are provided in Fig. S4. † The quenching rate constants, k q , as a function of pH may be determined from eqn (1); these rate constants are plotted in Fig. 5. Quenching by Y 731 (wt)-a 2 (red circles, ) and 2,3,5-F 3 Y 731 -a 2 (light blue squares, ) are referenced to the control, Y 731 F-a 2 . Within our error limits, little to no dependence of k q is observed for the [Re 356 ]-b 2 :Y 731 (wt)-a 2 complex ( ), as native tyrosine is protonated throughout the pH window. Conversely, a large pH dependence is observed for k q when [Re 356 ]-b 2 :2,3,5-F 3 Y 731 -a 2 is compared to Y 731 F ( ). Guided by our measurement of the pK a of 2,3,5-F 3 Y 731 -a 2 , shown in Fig. 3, we ascribe the observed differences in quenching to the relative ratio of deprotonated:protonated forms of 2,3,5-F 3 Y 731 as pH is varied. The large k q s at high pH is expected owing to deprotonation of F 3 Y 731 whereas quenching at low pH approaches that of Y 731 where both of the tyrosines are protonated. Scheme 1 summarizes the decay pathways for [Re]* in the different variants. The presence of tyrosine introduces an excited-state decay pathway via radical generation. For the case of Y 731 (wt)-a 2 , oxidation occurs by PCET whereas for 2,3,5-F 3 Y 731 -a 2 occurs by ET. These differing mechanisms of tyrosine oxidation for the two variants in the absence of Y 356 introduce the possibility that Y 356 may be involved with facilitating proton removal at the interface, though future experiments are needed to establish its specic role. While a co-linear Y 356 -Y 731 p-stacked mode where Y 356 acts directly as the proton acceptor for Y 731 is unlikely in light of recent Hf EPR/ENDOR data on amino-substituted tyrosine derivatives at various pathway positions, the strongly perturbed g x values of NH 2 Y 356 c indicate that Y 356 may communicate with Y 731 through a network of water molecules. 30 Additional evidence supporting the involvement of Y 356 in modulating Y 731 oxidation was also observed in [Re 355 ]-b 2 construct. 32 The photooxidation kinetics of Y 731 , which are summarized in Table S1, † indicate that Y 731 radical generation is enhanced by the presence of Y 356 as [Re 355 ]-b 2 oxidation of Y 731 is 2.1(1.2) times faster for Y 356 than for F 356 . The presence of Y 356 may facilitate proton removal from a 2 via the interface, thus assisting in PCET.
In this directed study, whereby Y 356 is absent by virtue of its replacement with [Re 356 ], efficient injection of a radical into a 2 is realized only when a proton is removed from the pathway by the introduction of 2,3,5-F 3 Y 731 . While this result does not implicate Y 356 directly as the proton acceptor for Y 731 , it supports the contention that Y 356 is in communication with Y 731 at the a 2 :b 2 subunit interface and that Y 356 enables the PCET required for efficient radical transport. Further investigations of this contention are underway along with studies to assess the role of possible contributions from other residues, or perhaps metal ions that may also be involved in managing protons at the interface.

Conclusions
Replacement of Y 356 by a [Re] photooxidant and installation of 2,3,5-F 3 Y at position 731 in a 2 furnishes a photoRNR that specically targets intersubunit radical transport. This construct supports photoinitiated substrate turnover, conrming its delity to the natural system. Time-resolved emission studies reveal that 2,3,5-F 3 Y 731 is oxidized at a rate 3 times faster than native Y 731 even though the non-natural amino acid is thermodynamically more difficult to oxidize at pH 7.6 (DE p $ 50-100 mV). These results emphasize the enzymatic imperative for coupling the proton and electron to allow for efficient radical transport. In conjunction with the parallel studies of [Re 355 ]-b 2 , these results suggest the importance of a well-coordinated proton exit channel involving Y 356 and Y 731 as key interfacial residues for radical transport across the a 2 :b 2 interface.