Tracking light-induced electron transfer toward O2 in a hybrid photoredox-laccase system

Summary Photobiocatalysis uses light to perform specific chemical transformations in a selective and efficient way. The intention is to couple a photoredox cycle with an enzyme performing multielectronic catalytic activities. Laccase, a robust multicopper oxidase, can be envisioned to use dioxygen as a clean electron sink when coupled to an oxidation photocatalyst. Here, we provide a detailed study of the coupling of a [Ru(bpy)3]2+ photosensitizer to laccase. We demonstrate that efficient laccase reduction requires an electron relay like methyl viologen. In the presence of dioxygen, electrons transiently stored in superoxide ions are scavenged by laccase to form water instead of H2O2. The net result is the photo accumulation of highly oxidizing [Ru(bpy)3]3+. This study provides ground for the use of laccase in tandem with a light-driven oxidative process and O2 as one-electron transfer relay and as four-electron substrate to be a sustainable final electron acceptor in a photocatalytic process.


INTRODUCTION
Electron transfer (ET) reactions are of fundamental importance in biological processes such as photosynthesis and respiration (Winkler, 2005). Understanding how electrons are transported in these complex systems may help chemists discover new tools to perform sustainable multielectronic chemical transformations. Currently, chemists and biologists are gathering their efforts to use light energy to drive enzymes to realize both chemical oxidation and reduction reactions of high importance for our societies (Lee et al., 2018;Litman et al., 2018;Guo et al., 2018;Schmermund et al., 2019). These targets are gaining much attention from both a fundamental and applied point of view due to the fact that to date we do not have catalytic systems that can match the activities and specificities of the biological catalysts. Hence, discoveries in this field may help develop sustainable chemical processes and also help in the design of more simple and robust photocatalytic systems (Pellegrin and Odobel, 2011).
Laccases are robust enzymes used in many industrial and biotechnological processes (Mate and Alcalde, 2017;Couto and Herrera, 2006;Agrawal et al., 2018;Riva, 2006). In nature, laccases couple the oxidation of organic substrates such as phenol derivatives, to the reduction of O 2 to water (Madhavi and Lele, 2009;Tron, 2013). The catalytic mechanism of laccases has been extensively studied (Solomon et al., 2008;Jones and Solomon, 2015;Heppner et al., 2014). At a type 1 (T1) Cu ion site the one-electron oxidation of substrate molecules occurs, and electrons are sequentially transferred inside the protein to a trinuclear copper catalytic center (TNC) where O 2 is eventually reduced in a four-electron, four-proton reaction (Jones and Solomon, 2015;Bertrand et al., 2002). The TNC consists of a pair of coupled type 3 (T3) Cu ions and a mononuclear type 2 (T2) Cu ion (Solomon et al., 1996;Hakulinen and Rouvinen, 2015). Although laccases are active on a wide range of natural substrates including phenol, polyphenols, hydroxyindoles, and benzenethiols among others, their oxidative power is locked by the one-electron oxidation process taking place at the T1 site whose potential ranges from 420 to 790 mV versus NHE, thus limiting the scope for the oxidation of organic substrates. Inspired by this biological tandem of oxidative and reductive processes performed by MCOs we have been interested in demonstrating that we can employ O 2 as the final oxidant by coupling a photoredox cycle with a laccase. Light-induced ET from the excited state of a photosensitizer
Time-resolved emission measurements with increasing LAC3 concentrations indicate a dynamic quenching process of the ruthenium excited state by laccase with a diffusion-limited bimolecular rate constant of 6.1 10 9 $M À1 s À1 ( Figure S2). This emission quenching could be due to energy and/or electron transfer from the sensitizer to the laccase. If all the quenching were due to ET a quantum yield of 12% of laccase reduction (compared with the concentration of excited state formed by the laser excitation) would be expected for a concentration of laccase of 37 mM. However, the amplitude of the measured absorption changes at long times, i.e., after decay of the excited state (Figure 1, inset) indicates a yield for the laser-flash-induced formation of the charge-separated state (CSS) of [Ru(bpy) 3 ] 3+ and reduced LAC3 (Cu I ) of only 0.7% (see Figure 1 for details). This yield is 17 times lower than what is expected for the products of the quenching process, suggesting that less than 10% of the quenching reactions lead to an effective reduction of laccase despite the high driving force for the ET from the ruthenium to the copper center T1 of -DG = 1.52 eV estimated from the redox potentials (E(Ru III /Ru II *) = À0.84 V; E(Cu II /Cu I ) = 0.68 V) (Kalyanasundaram, 1982a;Balland et al., 2008;Bock et al., 1979). These findings suggest that the major part of quenching is related to a Fö rster-type energy transfer process that can be expected to be rather efficient due to the strong spectral overlap between the [Ru(bpy) 3 ] 2+ * emission spectrum and the absorption band of the Cu II of the T1 site ( Figure S1) (Winkler, 2013). The Fö rster radius (R 0 = 2.6 nm) calculated for this interaction implies that upon encounter, through-space resonance energy transfer will largely outcompete ET, the rate of which is seven orders of magnitude slower at that distance (see supplemental information, transparent methods Section and Figure S5). ll OPEN ACCESS 2 iScience 24, 102378, April 23, 2021 iScience Article Interestingly, a careful comparison of the kinetics of the emission transient at 610 nm, expected to represent the decay of the [Ru(bpy) 3 ] 2+ * excited state, and the absorption transient at 610 nm reveals significant differences in the kinetics (Figure 1, compare blue and dashed pink lines). It appears clearly that the transient absorption trace recorded at 610 nm contains, in addition to the positive absorption attributable to the [Ru(bpy) 3 ] 2+ * excited state that has a small absorption at this wavelength (Yoshimura et al., 1993), a transient bleaching that can be related to a loss of Cu 2+ absorption of the T1 site. The transient bleaching at 610 nm (green dashed line in Figure 1) is well described by two exponential functions for its formation and decay with rate constants of 10 7 s À1 and 2 3 10 6 s À1 , respectively. Considering a concentration of laccase of 37 mM, the observed forward rate is too fast to be compatible with a diffusion-limited reaction between the sensitizer and the laccase. This fact leads us to postulate the formation of a relatively strong association complex between the [Ru(bpy) 3 ] 2+ chromophore and LAC3 (K a = 4.75 mM À1 ; determined by kinetic simulations described below), present in equilibrium with the separated species before and directly after the excitation flash. Such association complexes between ruthenium polypyridine and other transition metal complexes with blue copper proteins have been reported before (Brunschwig et al., 1985;Goldberg and Pecht, 1976). This interaction is usually attributed to the presence of negatively charged regions on the protein surface interacting with the positively charged metal complexes. The crystal structure of a laccase from Trametes versicolor revealed a dominance of negative charges of the protein and the substrate binding site as a small negatively charged cavity near copper T1 (Piontek et al., 2002). However, interactions other than electrostatic may occur (Kurzeev et al., 2009). For LAC3, molecular docking calculations predict a proximal binding position for [Ru(bpy) 3 ] 2+ at a distance as short as 11.2 Å from the Cu T1 (Robert et al., 2017). We note that the association constant K a = 4.75 mM À1 implies that under the conditions of the experiment shown in Figure 1, 12.5% of the [Ru(bpy) 3 ] 2+ complexes (1.9 mM of 15 mM employed) are in association with a LAC3 protein.   (Kalyanasundaram, 1982b); εT1 610 = 5600 M À1 cm À1 (Solomon et al., 2004). Small contributions to absorption changes at 450 nm due to T1 reduction and at 610 nm due to Ru 3+ were neglected. The dashed pink line is the scaled kinetics of emission at 610 nm, indicating that the 610 nm absorption transient contains contribution from other species than the excited state. Green dashed line: difference between the dashed pink line and the 610 nm emission kinetics representing transient bleaching of T1 absorption. Bottom: residuals for best fit according to the reaction scheme described in the text (Scheme 1) and in the supplemental information. Best fit traces are superimposed on the red and blue experimental traces as black lines both in the main window and in the inset.

OPEN ACCESS
iScience 24, 102378, April 23, 2021 3 iScience Article The question that arises then is whether the observed bleaching of T1 absorption in the association complex should be attributed to an electron or energy transfer process from the [Ru(bpy) 3 ] 2+ excited state to the copper T1. We exclude formation of the doublet excited state of the T1 copper ion, as this state was shown to decay to the ground state in less than 1 ps, which would certainly not allow us to observe a bleaching of T1 absorption persisting for hundreds of ns (Delfino et al., 2015). Formation of another, more long-lived excited state of the T1 Cu ion via Fö rster resonance energy transfer or via simultaneous exchange of an electron between T1 and the excited [Ru(bpy) 3 ] 2+ complex (Dexter mechanism) could explain the transient bleaching of T1 absorption. However, both energy transfer processes should lead to a concomitant recovery of the [Ru(bpy) 3 ] 2+ ground state for which no indication is detectable in the transient absorption trace recorded at 450 nm. For this reason and to account for the formation of stable Ru 3+ and reduced T1 (Figure 1) we attribute the observed transient bleaching of T1 to a photoinduced electron transfer reaction according to (Equation A): where the rate constant k cs describes charge separation from the excited state of the chromophore yielding the oxidized state of the chromophore (Ru 3+ ) and reduced laccase (T1 -), and k br describes charge recombination within the complex. It should be noted that the transiently formed species Ru 3+ is characterized by an extinction coefficient close to the one of the *Ru 2+ excited state (see legend of Figure 1) that leads to only very weak modifications of the absorption at 450 nm. The dataset of Figure 1 was analyzed by a global kinetic simulation based on the complete reaction scheme (Scheme 1, see also transparent methods section in the supplemental information), and an excellent fit to the data was obtained (solid black traces and residuals in Figure 1) with all rate constants well defined. From the k on and k off values describing formation of the association complex at two different concentrations of LAC3 an association constant of K a = 4.75 mM À1 was determined together with values for the rate constants k cs and k br , of 8.5 3 10 6 s À1 and 1.8 3 10 6 s À1 , respectively. Considering the large difference in driving forces for the charge separation and charge recombination reactions (-DG cs = 1.52 eV versus -DG br = 0.58 eV), the rate constants are surprisingly close. To obtain more information on the driving force dependance of these ET rates the same process was investigated using [Ru ( 2+ and LAC3 were determined as k cs = 6.6 3 10 6 s À1 and k br = 1.2 3 10 6 s À1 , close to the values found in the case of [Ru(bpy) 3 ] 2+ .
With the known driving forces for the two reactions in eq (A) for the two sensitizers, the rates could be analyzed in the framework of Marcus theory permitting an estimation of the reorganization energy and distances for the ET reactions ( Figure S4; see also transparent methods Section in the supplemental information). A value of 1.05 G 0.07 eV for the reorganization energy of ET for both charge separation and recombination can be deduced, which places the two ET reactions with high driving force in the inverted region of the Marcus curve ( Figure S4). The value for the reorganization energy is consistent with values reported in literature for oxidation or reduction of T1 copper sites (0.6-0.8 eV) and the low internal reorganization energy of the Ru 2+ /Ru 3+ (or Ru 2+ */Ru 3+ ) transition (Di Bilio et al., 1997;Farver et al., 2004;Wijma et al., 2007;Mines et al., 1996). Also, the distances extracted from the Marcus analysis, r = 12.5-13.5 Å , compare reasonably well with the Ru II -to-T1 Cu II distance of 11.2 Å found by non-covalent docking simulations (Robert et al., 2017). Interestingly, when the rates for charge separation and recombination were analyzed separately (red and blue fit curves in Figure S4) the donor-acceptor distance for charge recombination was found to be 1 Å larger than the distance for charge separation. Clearly additional data points on the Marcus plot would be desirable to draw definite conclusions, but it is tempting to attribute this difference to the location of the electron transferred upon charge separation on the peripheral bipyridine ligand in the *Ru 2+ excited state in contrast to the location of the positive charge receiving the electron during back ET on the central metal ion Ru 3+ . The fact that all rates determined by our global simulation can be consistently described by Marcus theory with meaningful parameters for distances and reorganization energy further supports the proposed mechanism of electron transfer described by eq. (A).
The flash-induced energy and electron transfer reactions occurring upon excitation of [Ru(bpy) 3 ] 2+ in a solution containing LAC3 are summarized in Scheme 1 and values for the case of [Ru(bpz) 3 ] 2+ are shown in Scheme S1 in the supplemental information. It can be concluded from this investigation that diffusioncontrolled electron transfer from [Ru(bpy) 3 ] 2+ * to LAC3 is short circuited by Fö rster energy transfer because the latter operates over longer distances and that only excitation of a sensitizer present in a Ru:LAC3 association complex at the time of photon absorption leads to electron transfer to the enzyme. The rightside mechanism in Scheme 1 further predicts that the yield of effective laccase reduction is given by the competition between the dissociation of the [Ru(bpy) 3 ] 3+ :LAC3complex described by the rate k off and charge recombination within the complex described by the rate k br . The association constant for the charge separated state can be determined from the value of k off resulting from the global fit (k off = 1.2 3 10 5 s À1 ), yielding K a cs = 8.5 mM À1 . The slightly increased affinity of the sensitizer for LAC3 in the charge separated state (K a cs = 8.5 mM À1 versus K a = 4.75 mM À1 ) can be attributed to an increased electrostatic interaction between oxidized Ru 3+ and reduced laccase. The ratio of k off /(k off + k br ) then gives a yield of 6% for the effective formation of Ru 3+ and LAC3from the charge separated state. With only 12.5% of the ruthenium sensitizers engaged in such a complex, the overall quantum yield of persistent, not only transient, laccase reduction (and Ru 3+ formation) drops to 0.75%. Put together, these different phenomena explain well the observed low yield of photoreduction of laccase from the excited state of the ruthenium sensitizer.
Although the simple reaction sequence proposed to occur in the association complexes (right panel of Scheme 1) describes well the spectroscopic data, it is surprising that no energy transfer reaction from [Ru(bpy) 3 ] 2+ * to LAC3 had to be considered in the analysis. In other words, it appears that in the complex the rate of energy transfer is slower than the rate of ET. This situation is reminiscent of a recent observation in a flexible BODIPY-C 60 dyad where photoinduced ET occurs in the folded conformer (R = 8.8 Å ), whereas energy transfer is observed in extended conformers (R = 17-20 Å ) (Tran et al., 2020). Similarly, in studies on single-walled carbon nanotubes tethered with porphyrins, excited-state energy transfer was found to be absent in the sample with a short tether (Li et al., 2004). Our results therefore seem not to be a unique case where at close distance the rate of energy transfer appears much slower than expected from the classical Fö rster rate-distance dependence. Although the different distance dependences between Fö rster energy transfer and ET (R À6 versus exponential) is expected to lead to a crossing of the k(r) curves with the rate of electron transfer becoming faster than the rate of energy transfer for short distances, in our case this ll OPEN ACCESS iScience 24, 102378, April 23, 2021 5 iScience Article crossing occurs at too short distance to provide a simple explanation for the dominance of ET over energy transfer ( Figure S5). It might be argued that Fö rster energy transfer theory is an approximation based on point dipole-dipole interaction restricted to weak coupling, valid at distances greater than R 0 (dashed green line in Figure S5), and extrapolation to significantly shorter distances might have to be taken with care F} orster, 1959). However, there exist a huge body of experimental data where the theoretical predictions hold even for quit short distances (Stryer and Haugland, 1967). We therefore favor another explication related to the spin multiplicity in our system with a *Ru 2+ triplet state and a doublet ground state of the copper T1. Although examples of triplet-singlet and triplet-doublet energy transfer reactions have been described (Naqvi, 1981;Cravcenco et al., 2019), the selection rules for angular momentum conservation might result in a strong decrease of the energy transfer rates. A particularly striking example is the study by McCusker and coworkers on a model system where energy transfer was occurring or inhibited as a function of the spin state of the acceptor (Guo et al., 2011). In our system, even two orders of magnitude decrease in energy transfer rate due to the doublet spin state of the T1 Cu(II) could have little effect at long distances (zR 0 ) where the rate of electron transfer is slow but could be crucial at short distances where the rate of electron transfer is fast enough to compete ( Figure S5). It would be interesting to investigate on the predominance of energy versus electron transfer in covalently linked ruthenium-laccase constructs where the distance can be varied in a controlled manner.
Considering the results described above it appears clearly that an efficient reduction of laccase cannot be obtained directly from the excited state of the sensitizer. In the following we assessed the possibility to overcome these limitations by using methyl viologen (MV 2+ ) as an electron relay between [Ru(bpy) 3 ] 2+ * and LAC3. Systems combining [Ru(bpy) 3 ] 2+ and MV 2+ are probably the most extensively studied photoredox systems (Bock et al., 1974;Mandal and Hoffman, 1984;. Acting as an oxidative quencher to the photosensitizer's excited state, this well-known mediator is converted into a long-lived reduced species that could deliver electrons extracted from the chromophore to the enzyme even when the latter is present at low concentration. Such an approach has helped to convey electrons to redox enzymes such as hydrogenases under anaerobic conditions (Noji et al., 2014;Honda et al., 2016;Okura et al., 1979).

Laccase photoreduction via MV ,+
To act as mediator, the chemical species to be used must meet several requirements. Besides a suitable redox potential, it should have good solubility in aqueous solution, the oxidized form (necessarily present in mM concentration for good dynamic quenching of the photosensitizer's excited state) should have little absorption in the visible to avoid absorption of excitation light, and, most importantly, the reduced molecule must be able to act as a reductant for the enzyme. In contrast to many other small redox molecules investigated as mediator, MV 2+ was found fit for this duty. It offers the additional advantage that the characteristic absorption features of its reduced state at 395 nm and 605 nm permit an easy tracking of its reduction and reoxidation. Under anaerobic conditions, the emission of [Ru(bpy) 3 ] 2+ * was quenched by the addition of 10 mM MV 2+ with a shortening of the lifetime of the excited state from 600 ns to 110 ns. The classic oxidative quenching of the [Ru(bpy) 3 ] 2+ * by MV 2+ involves a diffusion-limited encounter followed by an intermolecular electron transfer process (Kalyanasundaram, 1982c;. The charge separated state (CSS) [Ru(bpy) 3 ] 3+ -MV ,+ formed by dissociation of the encounter complex in competition with the back ET is described by a cage escape yield of about 20% . The addition of LAC3 (17 mM) to the [Ru(bpy) 3 ] 2+ /MV 2+ mixture did not result in a significant change of the [Ru(bpy) 3 ] 2+ * excited state lifetime. This experimental observation suggests that the dominating deactivation pathway of the excited triplet state [Ru(bpy) 3 ] 2+ * in the presence of both MV 2+ and LAC3 is an ET process to MV 2+ , which is faster than energy transfer to the laccase because of the much higher concentration of MV 2+ compared with LAC3. Transient absorption spectra were recorded at different time delays after the laser flash ( Figure S6). At 1 ms after the laser pulse, a bleaching around 450 nm and the typical positive absorption features of the MV ,+ radical (395 nm and 605 nm) were observed indicating the formation of about 1 mM of the [Ru(bpy) 3 ] 3+ -MV ,+ CSS by the oxidative quenching mechanism discussed above. Spectral evolution up to 500 ms shows that the signature of reduced methyl viologen disappears completely followed by a bleaching around 610 nm, which denotes the consumption of the MV ,+ radical and the loss of oxidized T1 in LAC3. These findings can be assigned to the charge shift from MV ,+ to the T1 locus of the laccase while the bleaching due to the [Ru(bpy) 3 ] 3+ species partially persists ( Figure S6, red trace).
The absorption transients at specific wavelengths ( Figure 2)  iScience Article the generation of 7.8 mM [Ru(bpy) 3 ] 2+ * triplet excited state as deduced from the maximum bleaching observed at 450 nm immediately after the laser flash (see legend of Figure 1 for extinction coefficients used). As described before, the excited [Ru(bpy) 3 ] 2+ * transfers an electron to the MV 2+ leading to the formation of 1.52 mM [Ru(bpy) 3 ] 3+ detected by the absorption bleach at 450 nm after decay of the excited state, and the same amount of MV ,+ radical, observed by the absorption increase at 610 nm at early times after the flash (εMV +, 605 = 13700 M À1 cm À1 ) (Watanabe and Honda, 1982). From these numbers a quantum yield of ca. 19.5% for formation of the MV ,+ radical is estimated. This yield agrees with reported values for the cage escape yield for reduced MV ,+ from the encounter complex (Kalyanasundaram, 1982c;. In the absence of LAC3 the MV ,+ radical disappears in a second-order reaction (k = 3.1 3 10 9 M À1 s À1 ) due to charge recombination with [Ru(bpy) 3 ] 3+ yielding mirror-like absorption transients at 450 and 610 nm (Figure 2 blue traces). In the presence of 16 mM LAC3 (red traces) the decay of MV ,+ at 610 nm is strongly accelerated, and the absorption transient turns into a bleaching indicating loss of oxidized T1 of the laccase. With higher concentration of LAC3 the decay of MV ,+ is further accelerated (Figure 2, green traces). The recovery of [Ru(bpy) 3 ] 2+ (bleaching at 450 nm) on the other hand slows down in the presence of laccase, and the lifetime of both species, reduced T1 and oxidized chromophore, extends into the ms time range. Using the extinction coefficients at 610 nm εT1 610 z 5600 M À1 cm À1 and εRu 3+ 610 z 2000 M À1 cm À1 (Kalyanasundaram, 1982b), a concentration of 0.7 and 0.9 mM of reduced T1 was estimated to be formed after the laser flash for the two laccase concentrations, which represents a yield of 46 and 59% compared with the quantity of MV ,+ produced. This clearly illustrates that MV ,+ can reduce the laccase (DG = À1.13 eV) with diffusion-limited kinetics (k 2 = 3.8 3 10 8 M À1 s À1 ). Importantly, the quantum yield of formation of [Ru(bpy) 3 ] 3+ and [Cu I ] depends only on the competition between the forward ET from MV ,+ to LAC3 and the charge recombination of the [Ru(bpy) 3 ] 3+ -MV ,+ CSS. This competition is in favor of laccase reduction even for low concentrations of LAC3 because the flash-induced concentration of MV ,+ (and [Ru(bpy) 3 ] 3+ , around 1 mM) is much smaller than the LAC3 concentration. It is important to recall that in this system, optimization of laccase reduction is equivalent to optimization of formation of the oxidized form of the ruthenium chromophore. The energetics and kinetics of the reactions occurring in the [Ru(bpy) 3 ] 2+ /MV 2+ /LAC3 system are summarized in Scheme S2 in the supplemental information.

Laccase photoreduction under continuous illumination in absence of O 2
Time-resolved laser flash photolysis studies such as those presented above are an important tool to investigate the sequence of light-induced reaction steps and determine the relevant rate constants. On the other hand, for the bimolecular rate constants for charge recombination between MV ,+ and Ru 3+ (3.1 3 10 9 M À1 s À1 ), electron transfer from MV ,+ to T1 (0.38 3 10 9 M À1 s ÀÀ1 ), and oxidation of reduced T1 by Ru 3+ (charge recombination, 0.46 3 10 9 M À1 s À1 ). Performing the experiment at pH 6 gave comparable results. First, we studied LAC3 photoreduction in the absence of O 2 . In this context it is important to recall that under anaerobic conditions the fully oxidized form of the enzyme corresponds to the so-called ''resting oxidized'' (RO) form, which is structurally and functionally different from the fully oxidized native intermediate (NI) produced during catalytic turnover in the presence of O 2 . RO is proposed to evolve from NI by protonation keeping all four Cu centers oxidized but ambiguity concerning the details of reduction of the Cu centers in RO exists (Solomon et al., 2008;Jones and Solomon, 2015;Heppner et al., 2013Heppner et al., , 2014. A solution of sensitizer and LAC3 containing 10 mM EDTA and different concentrations of the MV 2+ electron relay was irradiated with a white LED (filtered, 450<l < 700 nm) in the absence of O 2 , and the evolution of absorption at 610 nm was followed as a function of illumination time ( Figures 3A and S7). Before illumination, the absorption at 610 nm is exclusively attributable to the oxidized (Cu II ) T1 site. Upon illumination this absorption decreases with a concave shape to reach zero, indicating complete reduction of the Cu II T1 site. With increasing concentration of the electron mediator, MV 2+ , the time interval needed for T1 reduction is strongly decreased ( Figure 3A). The plot of the reduction rate of T1 (defined as the reciprocal of the time needed for complete reduction of T1) as a function of the concentration of MV 2+ reveals a linear dependence ( Figure S8) with an acceleration by a factor of >10 for a MV 2+ concentration of 300 mM compared with the system without MV 2+ .The linear dependence can be traced back to the linear dependence on MV 2+ concentration of the quantum yield of MV 2+ reduction prevailing at these low concentrations ( Figure S8). Rescaling of the absorption transients to match the time needed for disappearance of T1 absorption reveals very similar shapes of the kinetics in the absence and presence of MV 2+ (for all concentrations studied, Figure S9), strongly suggesting that there is no fundamental difference in the mechanism of reduction of the T1 site whether reduction occurs directly from the excited state of the chromophore or via MV ,+ . After the absorption at 610 nm has decayed close to zero, indicating accumulation of reduced T1, a re-increase of absorption at this wavelength is observed ( Figure 3A) that can be clearly assigned to the accumulation of reduced methyl viologen radical MV ,+ because its typical absorption band at 395 nm grows concomitantly ( Figure S7). Interestingly, especially at low concentrations of MV 2+  iScience Article we noted a lag time between complete T1 reduction and the start of accumulation of MV ,+ (Figures 3A and S7 right). As observed for the reduction of T1, this time is shorter for higher concentrations of MV 2+ . It seems reasonable to attribute this delay to the time needed for the total reduction of laccase (four reducing equivalents). The fact that no MV ,+ radical is detectable before complete laccase reduction implies that the rate of generation of the MV ,+ radical species by the excited sensitizer is slower than its consumption by the laccase enzyme. To put numbers to this, at the highest MV 2+ concentration a rate of formation of MV ,+ of 2 mM/s is estimated. This means that with the employed concentration of LAC3 (30 mM) the maximum rate of photoreduction of LAC3 is < 0.067 s À1 .
To investigate the observed kinetics, we performed a global simulation of the absorption changes recorded with different concentrations of MV 2+ (Figure 3). The presence of four different reduction states of the laccase that can interact with two potential electron donors, [Ru(bpy) 3 ] 2+ * or MV ,+ , makes the system rather inimical to analyze and requires some simplifying assumptions. First, it seems sound to assume that electron input into the LAC3 from either [Ru(bpy) 3 ] 2+ * or MV ,+ occurs via the T1 site (as supported by the laser flash data, at least for the first reduction). Furthermore, we neglect the details of reactions occurring within association complexes between [Ru(bpy) 3 ] 2+ and LAC3 and modeled laccase reduction from the sensitizer's excited state by a low-yield process. With these simplifications and because many rate constants are reasonably well known from kinetic flash photolysis measurements and are independent of MV 2+ concentration, a reasonable fit to the experimental data could be obtained, which reproduces the major features of the photoreduction assay with a common set of parameters (solid lines in Figure 3A). With this model in hand, we can now investigate the critical parameters with the aim to extract information on the mechanism governing the photoreduction of laccase and to determinate the rate constants for the reactions describing internal electron transfer (IET) from T1 to the TNC center. In the following we will present only the major results; details of the analysis are given in the transparent methods section in the supplemental information.
We use the nomenclature of Sekretaryova et al. (Sekretaryova et al., 2019) to describe the redox states of laccase from fully oxidized (T1 ox TNC 0 ) to the fully reduced form (T1 red TNC 3 ; with TNC x , x = number of electrons on the TNC). The initial absorption decrease at 610 nm describes the loss of absorption from oxidized T1 and corresponds to the consumption of three electrons, leading to reduced T1 and doubly reduced TNC (T1 red TNC 2 , dashed blue trace in Figure 3B). The first two IET steps from T1 to TNC, leading to the formation of T1 red TNC 2 , are fast compared with the photoproduction of MV ,+ , which is the rate limiting step. Accordingly, the analysis was found to be not sensitive to the rates of forward electron transfer from T1 to the TNC, provided they are faster than about 0.1 s À1 . We assumed for the fits rate constants of 1 s À1 as found in a pulse radiolysis study (Farver et al., 2011). The lag time following reduction of T1 corresponds to the uptake by the laccase of the fourth electron leading to full reduction of the enzyme (T1 red TNC 3 , solid blue line in Figure 3B). As can be seen in the profiles of the intermediates, this last reduction step corresponds to the transition T1 red TNC 2 / T1 red TNC 3 (compare dashed and solid blue lines between 230 and 420 s) meaning that T1 stays almost fully reduced during the process although T1 is the entrance side for the electron. In fact, the kinetic parameters determined from the global fit clearly show that the third reduction of the TNC from T1 red is significantly endergonic (k + /kz 1/4500; DG z +200 meV), in contrast to the first two IET steps that are close to equilibrium (Scheme 3). Because during catalytic turnover all internal electron transfer steps leading to full reduction of the oxidized native intermediate (NI) are fast, even in high-potential laccases as LAC3 used in our study , (Sekretaryova et al., 2019;Heppner et al., 2014;Solomon et al., 1996), we attribute this sluggish third-electron reduction of the TNC to the fact that under the anaerobic conditions of this experiment the oxidized enzyme is not in its native intermediate form but in the resting oxidized (RO) state well described in the literature (Tollin et al., 1993). The characteristics of internal electron transfer deduced from our experiment are in agreement with results from pulse radiolysis and direct electrochemical measurements reporting a potential of the T2 Cu site in the RO state of around 400 mV versus NHE, i.e. about 300 mV lower than the potentials of T1 and T3 (Farver et al., 2011;Shleev et al., 2005). Although there is ''insufficient driving force'' for ET from reduced T1 to T2, our results show that electron transfer occurs because the overall energetics is still favorable due to the low potential of the MV ,+ ''substrate.'' The possibility of uphill intramolecular electron transfer under certain conditions had also been proposed based on electrochemical and quantum chemical studies (Shleev et al., 2012 iScience Article appears that monitoring the accumulation of the reduced MV ,+ radical is an indirect but valuable tool to obtain information about IET events within the laccase enzyme, complementing measurements of absorption changes related to the oxidation state of T1, which are the only observables in the absence of MV 2+ . Interestingly, the consistent analysis performed on this set of experiments, i.e., laccase photoreduction in the absence and the presence of MV 2+ , reveals that in the absence of MV 2+ , where laccase reduction occurs via the excited state of the chromophore, the enzyme is only reduced by three electrons ( Figure S10). The reason for this is the unfavorable pre-equilibrium resulting from the endergonic nature of the third reduction of the TNC, which makes the IET T1 red TNC 2 / T1 ox TNC 3 too slow to be compatible with the short lifetime of the [Ru(bpy) 3 ] 2+ * excited state. In contrast, the MV ,+ species is long-lived and can progressively reduce the state T1 ox TNC 3 present at low concentration in the above-mentioned equilibrium with T1 red TNC 2 to form T1 red TNC 3 . Of note, in the presence of dioxygen where all internal electron transfer steps are nearly isoenergetic, the complete, four-electron reduction of laccase via [Ru(bpy) 3 ] 2+ * is possible.
Finally, to obtain a satisfactory global fit of the absorption transients describing accumulation of MV ,+ after full, four-electron reduction of the enzyme ( Figure 3A), we had to include in the simulation a feature that accounts for a lack of the quantity of MV ,+ accumulating (reaction A15 in the supplemental information, transparent methods Section). The missing quantity, most clearly visible in the traces for low concentration of MV 2+ (see Figure S9), corresponds roughly to the quantity of enzyme present (30 mM). We therefore tentatively attribute this feature to the formation of adducts of MV 2+ , with the enzyme being most efficient when LAC3 becomes fully reduced. The ''bound'' MV 2+ might still be reducible by [Ru(bpy) 3 ] 2+ * but the dissociation rate for MV ,+ or Ru 3+ from the Ru 3+ :MV ,+ :LAC3 complex might be too low to avoid charge recombination between MV ,+ and Ru 3+ .

Dioxygen consumption
With the finding that the presence of an electron mediator helps transferring electrons faster and more efficiently to the LAC3, we then performed a set of dioxygen consumption experiments to characterize the catalytic activity of the laccase when reduced by a sensitizer/electron mediator/electron donor photoredox system. Under these aerobic conditions, O 2 competes with the laccase for the electron from the MV ,+ species as it is well known that in aqueous solution MV ,+ reacts rapidly with O 2 to form O 2 ,-(or O 2 H , at our operating pH) (Farrington et al., 1973). This competition between laccase and O 2 is clearly evident from transient absorption measurements revealing a markedly reduced efficiency of flash-induced T1 reduction with increasing amounts of O 2 ( Figure S11). Surprisingly though, when light-induced dioxygen consumption was measured with a Clark electrode, the effect of laccase was much stronger than expected from this simple competition scheme. Figure 4 shows the rate of O 2 consumption as a function of LAC3 concentration in the absence and presence (250 mM) of MV 2+ . In the absence of MV 2+ , the rate increases linearly with the concentration of LAC3, a behavior easily explained by an increased reduction of laccase from the excited state of the chromophore due to increased formation of association complexes (see above). When MV 2+ is present, but no laccase, a fast dioxygen consumption rate is observed, which corresponds to the two-electron reduction of O 2 to H 2 O 2 via dismutation of O 2 ,-(eqs. B and D, Figure 5

O (C)+(F) or 4x(B)+(E)+(F)
When laccase accepts electrons, a four-electron/four-proton total reduction of O 2 is performed by the enzyme (eqs. C and F) in addition to the two-electron/two-proton partial direct reduction of O 2 to H 2 O 2 (eqs. B and D). A substantial addition of an enzyme consuming half as much O 2 for the same concentration of photogenerated MV ,+ could thus show up in the dioxygen consumption curve as a decrease in the dioxygen consumption rate compared with MV 2+ alone. Therefore, the pattern in Figure 4 can be explained by a competitive scheme where O 2 is being progressively substituted by LAC3 as electron recipient from MV ,+ because the two paths distinctively produce end products, either H 2 O 2 or H 2 O (Scheme 2). When enough laccase is present to compete with O 2 for electrons from MV ,+ the rate increases and reaches a saturation value given by the light intensity and efficiency of the sacrificial electron donor. However, the effect of laccase on the rate of O 2 consumption at laccase concentrations much lower than the concentration of dissolved O 2 (250 mM) is surprisingly pronounced. This suggests that laccase reduction is more efficient than expected from a simple competition between laccase and O 2 for electrons from MV ,+ as observed in Figure S11. We therefore propose that laccase can be reduced not only by MV ,+ but also and quite significantly by the O 2 ,formed via electron transfer from MV ,+ to O 2 (eq. 4xB, E). Analysis of the O 2 consumption kinetics in the absence and presence of laccase ( Figures 5 and S12) supports the interpretation that it is indeed O 2 ,that is the dominant reductant for the laccase under these conditions. Laccase reduction by O 2 ,had also been reported based on pulse-radiolysis studies (Guissani et al., 1982;(Pecht et al., 1977)). The two processes (eqs. (C)+(F) or 4x(B)+(E)+(F)) should then significantly prevent an accumulation of H 2 O 2 . As a matter of fact, H 2 O 2 concentration dropped dramatically as laccase was introduced in dioxygen consumption experiments ( Figure S14). The reaction sequence for photoreduction of laccase under aerobic conditions is summarized in Scheme 2.
However, if O 2 ,can replace MV ,+ for reduction of LAC3 as suggested by the O 2 consumption experiments, then the question arises why is the bleaching of T1 absorption, as detected in the laser flash experiment ( Figure S11), strongly reduced in the presence of O 2 ? To answer this question, we performed a global kinetic analysis of the oxygen consumption profile in the absence and presence of laccase (Figure 5). It is interesting to note that a first analysis of the O 2 consumption kinetics (see Figure S12 and transparent methods Section in the supplemental information) supports a slow reaction of superoxide with LAC3 with a rate not exceeding 5 3 10 6 M À1 s À1 (yielding a time constant of >1 ms for a concentration of O 2 of 250 mM), about two orders of magnitude slower than the rate of laccase reduction by MV +, iScience Article (see legend of Figure 2 and Scheme S2 in the supplemental information). If this applies also for the flashinduced process, T1 reduction would be too slow to be detected on the timescale of the transient absorption measurement (1 ms) but could be still fast enough to be operational under the continuous light conditions of the dioxygen consumption assay. Such a slow reduction of T1 by O 2 ,has been described in pulse radiolysis studies where laccase was shown to be an efficient radical scavenger (Guissani et al., 1982;Goldberg and Pecht, 1978). In these studies, O 2 ,-, as other reducing radicals, was found to interact with the laccase not through a direct reduction of the T1 center but instead by fast formation of a transient adduct with the protein followed by slow intra-or intermolecular electron transfer from the adduct site to the T1. A refined kinetic simulation shows that such a scenario is fully compatible with the kinetics of O 2 consumption shown in Figure 5, provided that the adduct formation is reversible (see transparent methods Section in the supplemental information). This scenario can account for the two orders of magnitude difference in the apparent bimolecular rate constants for laccase reduction via MV ,+ or O 2 ,in a more convincing way because such a difference in bimolecular interaction rates would be difficult to rationalize. Unfortunately, neither the exact locus of the O 2 ,adduct nor a definite mechanism for the intramolecular electron transfer have been determined in these pulse radiolysis investigations.
Although not favored by the authors of these earlier studies, it is tempting to hypothesize on the possibility of O 2 ,interacting with the enzyme at the level of the TNC considering the fact that this is the natural binding site for O 2 in laccases. The data presented here do not allow to draw conclusions on this topic, but they might give indications for experiments dedicated to address these questions. Importantly though, our results clearly evidence the specificity of the enzyme acting as a sink for photogenerated electrons that it can efficiently recover either from an electron mediator or from a superoxide radical and use them for the reduction of O 2 to H 2 O. To visualize the action of laccase Figure S13 shows the evolution of the concentration of key species during the photocatalytic cycle. , and not MV ,+ , is the predominant reductant for the laccase under these conditions. See Figures S12 and S13 in the supplemental information for further discussion.

12
iScience 24, 102378, April 23, 2021 iScience Article Conclusion Using synthetic chromophores to initiate light-driven electron transfer processes for photo-biocatalysis can provide unique solutions to perform green chemical transformations. However, an innate limitation in elaborating such systems resides in the presence of unproductive energy transfer processes that can outcompete ET processes (Winkler, 2013). In this study, we clearly show that detrimental energy transfer can be short circuited by the addition of a reversible electron acceptor, which transfers electrons from [Ru(bpy) 3 ] 2+ * to a laccase with high quantum yields. Even when the electron relay is reducing enough to interact with O 2 to form superoxide radicals, the laccase enzyme can efficiently accept electrons from the latter species to catalyze O 2 reduction and thereby short circuit the formation of H 2 O 2 . In this way O 2 functions as both, an electron shuttle to and oxidizing substrate for the laccase, leading to an efficient photocatalytic cycle. With the presented thorough description of the charge transfer and dioxygen reduction steps involving this MCO, we show that the dioxygen from air, usually carefully eliminated from reaction media in synthetic systems mimicking photosynthesis, can become a key element in photo-catalysis as a safe, renewable, and inexpensive alternative to sacrificial electron acceptors. In parallel to the consumption of O 2 in such systems, a powerful oxidant, [Ru(bpy) 3 ] 3+ (1300 mV versus NHE), is photo produced, ready to drive the oxidation of more tenacious organic substrates (Scheme 3). Among future challenges ahead of us lie the use of the oxidative power of the oxidized chromophore for further chemical transformations in lieu of the sacrificial electron donor used in this study. Work along these lines is underway in our labs. In a broader context, this study also highlights the potential of light-controlled electron delivery to enzymes as an interesting alternative to stopped-flow techniques to reveal functional details of internal charge transfer and charge accumulation at the catalytic sites. The interested reader is referred to Figure S15 in the transparent methods section of the supplemental information for some illustration.

Limitations of the study
This work provides insight on coupling of laccase with a photoredox system to create highly oxidizing species using dioxygen as terminal electron acceptor. As the next step we are working on the application of this system for oxidative catalysis by replacing the sacrificial electron donor EDTA used in this study by a catalytic system for oxygen atom transfer reactions. Further work is required to analyze the effect of pH on both the efficiency of laccase reduction and the oxidation catalytic cycle.

Resource availability Lead contact
Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Winfried Leibl (winfried.leibl@cea.fr).

Materials availability
This study did not generate new unique reagents.

Data and code availability
The published article includes all datasets/code generated or analyzed during this study.  Fig. 1, Fig. 2, and Fig. 3. and different concentrations of LAC3 in B&R buffer, pH = 6 after excitation at 455 nm. Right: Stern-Volmer plot derived from monoexponential fits to the data. The decrease in amplitude of the emission in presence of laccase is mainly attributable to an inner filter effect caused by the absorption of the oxidized laccase overlapping the spectral region of [Ru(bpy)3] 2+ emission (see Fig. S1). For the highest LAC3 concentration part of decrease in initial emission amplitude is due to a dilution effect. Related to Fig. 1 and Fig. S1. Figure S3. Transient absorption kinetics of a solution of [Ru(bpz)3] 2+ (15 µM) and LAC3 (36 µM) in B&R buffer, pH = 5.7 after laser flash excitation at 455 nm. Inset: zoom of the data at long times. Bottom: residuals for best fit according to the reaction scheme described in Scheme S1 below. The dashed pink line is the scaled kinetics of emission at 610 nm indicating that the 610 nm transient contains contribution from other species than the excited state. Related to Fig. 1 and Scheme S1.  Fig. 1, Fig. S3, Scheme 1, and Scheme S1.
Scheme S1. Interactions between the [Ru(bpz)3] 2+ photosensitizer and the LAC3 enzyme identified by laser flash photolysis experiments. Reactions leading to formation of reduced laccase (LAC3 -) and oxidized photosensitizer (Ru 3+ ) are indicated by red arrows. In the bimolecular pathway (left) Förster resonance energy transfer quenching of the [Ru(bpz)3] 2+ * excited state largely outcompetes with electron transfer to the T1 Cu II center in LAC3. In the association complex (right) charge recombination is about 10 times faster than dissociation leading to a low yield for formation of the target state Ru 3+ LAC3 -(red). Parameter values were deduced from analysis of the data of Fig. S3 (pH 5.7). Related to Fig. 1, Fig. S3, and Scheme 1. Figure S5. Rates for Förster energy transfer (blue, Eq. 1) and electron transfer (red, Eq. 4) as a function of donor-acceptor distance r. For the electron transfer rate, values for charge separation (-G = 1.52 eV and = 1.12 eV) from Fig. S4 were used and the experimental rates for ET between the two chromophores and LAC3 are indicated by dots. The black dashed line represents the maximal electron transfer rate (-G = ). The dashed blue line shows energy transfer rates 100 times slower than the Förster rates to account for a hypothetical effect of selection rules related to the triplet/doublet spin multiplicity (see text). The value of the Förster radius R0 = 26.4 Å is represented by the dashed green line. See Methods Section for equations and details.
At the employed concentrations of LAC3 the mean distance between the photosensitizers and LAC3 are larger than 30 Å. If the position of LAC3 is taken at r = 0 Å a photosensitizer approaching (from the left side of the graph) will have to cross a k(r) curve providing a rate faster than the diffusion limit for an interaction to occur. This occurs at about 23 Å in the case of "slow" Förster energy transfer and at much larger distance for normal Förster energy transfer. At these large distances Marcus ET is many orders of magnitude slower and therefore cannot occur. Related to Fig. 1, Scheme 1, Fig. S3, and Scheme S1. The total quantum efficiency, limited by the escape yield of MV 2+ for formation of the charge-separated state, could be improved using a bichromophoric sensitizer . Related to Fig. 1, Fig. 2, Fig. S2, and Fig. S3. The initial absorption spectrum (red) shows contributions from [Ru(bpy)3] 2+ ( =450 nm) and the Cu II T1 site of laccase in its oxidized form ( =500-700 nm). After 1.6 min of illumination the 610 nm absorption band has disappeared leaving [Ru(bpy)3] 2+ absorption ( =450 nm, blue). After a lag phase without spectral changes the typical spectrum of MV +• radical (395 nm, 605 nm) starts growing (black). Right: Characteristics of the time dependence of absorption changes at 610 nm with a phase of Cu II T1 reduction (1) followed by a lag phase (2) and accumulation of MV +• radical (3). Related to Figure 3. Figure S8. Effect of MV 2+ concentration on the rate of photoreduction of T1 in a [Ru(bpy)3] 2+ /EDTA/MV 2+ system. Rates k were taken as the reciprocal of the time of illumination needed for the total bleach of the absorption at 610 nm. k0 is the rate in absence of MV 2+ . Blue symbols indicate the simulated quantum yield of MV 2+ reduction from the chromophore excited state which occurs in competition with the intrinsic decay of the excited state, considering the escape yield for MV •+ of 20%. The good correlation shows that laccase reduction is controlled by the yield of photoproduction of MV •+ . Related to Figure 3A.    In the absence of laccase, the constant slope is determined by the dismutation of superoxide which is faster than its production (the steady state concentration of O2is very low, see Fig. S13). In the presence of laccase, the initial O2 consumption rate is zero because oxidation of O2 -(regenerating O2, no net consumption) is faster than dismutation. In other words, laccase prevents accumulation of sufficient concentration of O2for dismutation to become kinetically competitive. Once LAC3 has recovered four electrons, the enzyme starts to consume O2 by its normal catalytic activity. The action of O2as a reversible electron relay between MV •+ and LAC3 is not contributing to O2 consumption. Related to Fig. 5. Figure S13. Visualization of the evolution of the concentration of key species during photocatalytic oxygen consumption measurements in the absence (dashed lines) and in the presence of 16 µM LAC3. Dox is the concentration of oxidized electron donor. In presence of LAC3 production of H2O2 is strongly reduced (yellow traces) and production of H2O is strongly accelerated.
In the absence of LAC3 O2 •quickly reaches a low stead-state concentration (0.6 µM) given by the ratio of the constant rate of photoproduction via Ru* and MV •+ and the effective rate of dismutation which increases quadratically with [O2 •-]. When LAC3 is present O2 •level rises slowly because O2 •is initially efficiently oxidized by LAC3 until a steady-state level is established which is controlled by an ET step within the laccase which is rate limiting for O2 reduction catalysis by laccase. Related to Fig. 5, Fig. S12, and Scheme 2.  Simulation of a typical anaerobic photoreduction experiment to visualize the sensitivity of the absorption changes at 610 nm on rate constants for internal electron transfer from T1 to the TNC. Presence of excess MV 2+ (2 mM) is assumed which permits to neglect direct, inefficient interaction of the chromophore with laccase (reactions C1-C6 above). Light intensity was chosen to produce 3 µM/s of MV •+ reductant. A) Variation of the rate of the first IET from T1 to TNC; B) Variation of the rate of the second IET from T1 to TNC; C) Variation of the rate for reverse IET from the fully reduced TNC to T1. The latter corresponds to a change from an exergonic to an endergonic reduction step for the storage of the third electron on the TNC. All forward rate constants besides the ones varied are assumed not limiting (10 s -1 ). All backward rate constants besides the one varied in panel C are set to 0. Related to Fig. 3.

Materials
Tris(2,2'-bipyridyl)dichlororuthenium(II) hexahydrate (99.5%), methyl viologen dichloride hydrate (98%) and benzyl viologen dichloride (97%) were purchased from Sigma-Aldrich. The laccase LAC3 from Trametes sp. C30 was produced as a recombinant enzyme in Aspergillus niger as described. (Klonowska et al., 2005, Mekmouche et al., 2014 Laser flash photolysis For transient absorption and emission measurements an Edinburgh Instruments LP920 flash photolysis spectrometer system was used with a Continuum Surelite OPO laser (5 ns pulse duration, typical energy 10 mJ) for excitation at 460 nm, a pulsed 450 W Xenon lamp for the probe light, and either a PMT (Hamamatsu) or ICCD camera (Andor) as detectors. Presented data are typically averages over 10-20 measurements. Samples were prepared in sealed quartz cuvettes containing Britton-Robinson (B&R) buffer at the indicated pH and were purged for 20 min with argon prior to each experiment.

Light-induced dioxygen consumption studies
Dioxygen consumption was measured by polarography using a model 781 oxygen meter (Strathkelvin Instruments) or a Hansatech Oxygraph system with a micro-Clark electrode fitted to a temperaturecontrolled glass chamber. Irradiation of the sample was performed through the glass chamber with a Dolan JENNER LED (250 mW.cm -2 ) equipped with filters (450 nm <<700 nm). For the data in Fig. S11 irradiation was done with a Flexilux 600 Longlife 150 W 21 V halogen lamp through an ultraviolet cut-off (λ≥375 nm) and a 23% neutral density filter.
H2O2 quantification H2O2 production after illumination as a function of laccase concentration was determined by HRP assay. 40 µL of the reaction solution, filtered through a 3 kDa filter was added to a solution of 240 µL ABTS (2 mg/mL) in phosphate buffer pH 8.0 and 10 µL HRP (Sigma, 0.103 mg/mL) in 100 mM phosphate buffer pH 8 was added. After 5 minutes of incubation the absorbance at 420 nm was recorded. Calibration was done with reference measurements in the absence of light.
Kinetic spectroscopic studies UV-vis absorption spectra were recorded on a Cary 50 spectrophotometer equipped with a homemade irradiation setup including a Dolan JENNER LED (250 mW.cm -2 ) equipped with filters (450 nm <<700 nm). For each MV 2+ concentration a quartz cuvette was filled with the reaction mixture and sealed in a glovebox prior recording the corresponding spectrum.

A) Simulation of flash-induced absorption changes in absence of MV 2+
The simulation is based on the reaction scheme shown in Scheme 1: The following rate constants were fixed: kA1= k-A6 = 1·10 9 M -1 s -1 ; kA2= kA7 = 1·10 9 s -1 ; kA8 = 1.7·10 6 s -1 . Extinction coefficients are given in the legend of Fig. 1. In the simulation excitation was induced after establishment of the association equilibria (reactions A1 and A6).

C)
Simulation of anaerobic photoreduction of laccase (Fig. 3) A simplified reaction scheme was used including the following reaction sequence: Initial concentrations of chromophore-laccase association complexes (RuL) and free chromophore (Rufree) were chosen according to the parameters determined before (Scheme 1). Other initial concentrations are those used in the experiment. Reaction 10 takes into account the escape yield . For computational reasons, effective electron transfer in association complexes between Ru and LAC3 were simulated by apparent rates for electron transfer (reactions 3-6) in competition with the other deactivation pathway for decay of the Ru excited state (reaction 2) and relative excitation rate (reaction 1). Potential charge recombination between Ru III free and T1red was neglected. All reactions except intramolecular ET in the laccase (17-19) and binding of MV 2+ to fully reduced laccase (reaction 15) is considered irreversible. For some reactions, rate constants are known from independent laser flash kinetic measurements.
The essential result of the fit to the data is that, under the anaerobic conditions of this experiment where the fully oxidized form of the laccase enzyme corresponds to the resting oxidized form (Solomon, 2016), the third reduction of the TNC (reaction 19) is energetically uphill by about 200 meV. Values for rate constants for internal electron transfer in the laccase derived from the simulation: k-C17=0.02 s -1 , k-C18=0.85 s -1 , k-C19 ≈4500 s -1 .
The traces in Fig. S10 were produced by plotting the flux integrals for reactions C7 and C20.
Simulations of a typical anaerobic photoreduction experiment to visualize the sensitivity of the absorption changes at 610 nm on rate constants for internal electron transfer from T1 to the TNC (reactions C17-C19) are shown in Fig. S15.
Other rate constants as before.

Calculation of energy and electron transfer rates
Förster-type resonance energy transfer: To investigate the competition between energy transfer quenching and photo-induced electro transfer quenching of the ruthenium excited state it is useful to simulate the distance-dependence of both processes. The rate of oscillating dipole energy transfer between a donor and acceptor molecule can be deduced from equation (1) proposed by Förster (Főrster, 1959, Lakowicz, 2006 where τD the excited state lifetime of the donor in the absence of acceptor, R0 is the Förster distance at which energy transfer is 50% efficient, and r is the distance between donor and acceptor. The coupling between the dipolar transitions of donor and acceptor is determined by the Förster radius given by 0 6 = 9000 * ln(10) * 2 * * 128 * 5 * 4 * (2) where κ 2 is the dipole orientation factor, ΦD the quantum yield of the donor emission, n the refractive index, NA Avogadro's number, and J the spectral overlap integral in units of mol -1 dm 3 cm 3 , which contains the normalized emission spectrum of the donor, FD, and the absorption spectrum of the acceptor, εA as a function of wavenumber, ν. J was evaluated from the data in Fig. S1 to yield J = 7.4·10 -14 mol -1 dm 3 cm 3 . Using as values for the other parameters in eq (2) κ 2 = 0.476 (Steinberg, 1971), ΦD = 0.042 (Kalyanasundaram, 1982), and n = 1.4 Lakowicz (2006) R0 is determined to R0 = 26.4 Å. Together with the value τD = 588 ns from Fig. S2 this allows to draw the kEnT(r) plot shown in Fig. S5.