Switching the Mechanism of NADH Photooxidation by Supramolecular Interactions.

A series of three Ru(II) polypyridine complexes was investigated for the selective photocatalytic oxidation of NAD(P)H to NAD(P)+ in water. A combination of (time-resolved) spectroscopic studies and photocatalysis experiments revealed that ligand design can be used to control the mechanism of the photooxidation: For prototypical Ru(II) complexes a 1 O2 pathway was found. Rudppz ([(tbbpy)2 Ru(dppz)]Cl2 , tbbpy=4,4'-di-tert-butyl-2,2'-bipyridine, dppz=dipyrido[3,2-a:2',3'-c]phenazine), instead, initiated the cofactor oxidation by electron transfer from NAD(P)H enabled by supramolecular binding between substrate and catalyst. Expulsion of the photoproduct NAD(P)+ from the supramolecular binding site in Rudppz allowed very efficient turnover. Therefore, Rudppz permits repetitive selective assembly and oxidative conversion of reduced naturally occurring nicotinamides by recognizing the redox state of the cofactor under formation of H2 O2 as additional product. This photocatalytic process can fuel discontinuous photobiocatalysis.

Due to their impressive catalytic activity and high stereoselectivity under ambient aqueous conditions, enzymatic oxidation reactions are recently gaining significant relevance even on an industrial level. Typically, O2 or NAD+ (oxidized nicotinamide adenine dinucleotide) are used as oxidants.[ , ] Due to the high costs of the important cofactor NAD+, a variety of enzyme‐based[ , ] as well as chemical[ , ] and electrochemical[ , ] regeneration methods have already been developed.

In order to also exploit the energy of visible light for NAD+ regeneration, processes based on either organic dyes,[ , ] photosensitive polymers or inorganic dyes such as tin porphyrines, Ru(II),[ , ] Os(II), and Ir(III)[ , ] complexes have been developed. Additionally, NADH photooxidation has recently also been successfully investigated as a mode of action in photodynamic therapy (PDT).[ , , , ] In all these light‐driven processes, NADH oxidation is either induced via formation of 1O2[ , , ] by the excited chromophore or by electron transfer (eT) onto the excited[ , , , ] or photooxidized[ , , ] dye.

A specific subclass of Ru(II) complexes, those bearing large planar ligands, is known to interact with nucleobases inside the DNA.[ , , , ] As NAD(P)H contains an adenine subunit, we thus wondered whether such preorganization might lead to improved light‐driven cofactor oxidation stimulated by supramolecular interactions. To test this hypothesis, a simple series of three prototype chromophores, namely [Ru(tbbpy)3]Cl2 (Rutbbpy), [(tbbpy)2Ru(phen)]Cl2 (Ruphen) and [(tbbpy)2Ru(dppz)]Cl2 (Rudppz) was investigated with respect to their photooxidative properties (Scheme 1, tbbpy=4,4'‐di‐tert‐butyl‐2,2'‐bipyridine, phen=1,10‐phenanthroline, dppz=dipyrido[3,2‐a:2′,3′‐c] phenazine).

Scheme 1

Molecular structure and abbreviations of the complexes (left) and reaction scheme of the investigated Ru(II) complex catalyzed NAD(P)H photooxidation (right; Rib=β‐D‐ribofuranose, ADP(P)=(phosphorylated) adenosine diphosphate).

The water soluble complexes Rutbbpy, Ruphen and Rudppz were synthesized according to established protocols (see Supporting Information and Figures S1–S21). Their photooxidative activity towards NADH under ambient aqueous conditions was evaluated using LED‐irradiation (λmax=465 nm) in presence of 500 equivalents NADH (see Supporting Information for experimental details). All three complexes proved to be active NADH photooxidation catalysts as indicated by the continuously decreasing 340 nm absorbance and 460 nm centered emission band (Figures 1A and B) both being characteristic for the reduced nicotinamide moiety. A nearly quantitative conversion of NADH to NAD+ under these conditions was obtained, if the samples were irradiated for 2 h (>99 % for Rutbbpy and Rudppz, >95 % for Ruphen, Figure S22). No spectral shifts indicative of products different to NAD+ (non‐emissive if λexc=340 nm, Figure S23) were observed. 1H NMR spectroscopy also proved the selective formation of NAD(P)+ from NAD(P)H (Figures S24–S25). Despite its negligible 1O2 formation in H2O containing solutions monitored by NIR emission spectroscopy (Figure S26), Rudppz exhibited the fastest NADH photooxidation in H2O (initial TOF>1000 h−1) followed by Ruphen and Rutbbpy (Figure 1C).

Figure 1

A) UV‐vis and B) corresponding emission spectroscopic changes during the NADH photooxidation with Rudppz in H2O. C) TON‐plot for the three photocatalysts in H2O (1 mM NADH, 2 μM photocatalyst). D) Effect of solvent deuteration on the NADH photooxidation rate. E) Effect of ROS quenchers on the NADH photooxidation rate using Rutbbpy or F) Rudppz as catalyst.

To evaluate the mechanism of the oxidation reactions, a series of control experiments was performed. NADH oxidation was neither observed in the dark nor in the absence of oxygen or Ru(II) complex (Figure S27). No UV‐vis or 1H NMR spectroscopic changes were found, if NAD+ was irradiated in presence of a photocatalyst indicating that NAD+ represented the end‐point of the reaction (Figures S27–S28). Furthermore, by adding 15 mM of the known OH radical scavenger sodium benzoate to the solution, the rate of NADH photooxidation choosing Rutbbpy as catalyst was not altered (Figure 1E). Thus, OH radicals as reactive oxygen species (ROS) were excluded to decisively contribute to the photooxidation.

However, both the addition of 15 mM NaN3 or 15 mM histidine strongly reduced the NADH photooxidation rate using Rutbbpy and Ruphen (Figure 1E, Figure S29). Based on the additives' property to act as 1O2 quenchers,[ , , ] a reaction mechanism proceeding via 1O2 formation from the 3MLCT excited state of Rutbbpy or Ruphen initiating the oxidation of NADH was proposed. The 1O2 pathway for Rutbbpy and Ruphen was further supported by the fact that significant luminescence quenching for both complexes was only observed in presence of 250 μM oxygen but not 250 μM NADH (Figure S30) as well as the finding that changing the solvent from H2O to D2O led to an accelerated photooxidation (see Figure 1D) which we ascribe to the prolonged lifetime of 1O2 in D2O. Thus, using D2O and low concentrations of Ruphen, TONs up to 3600 after 3.5 h were realized (Figure S31).

Rudppz revealed a different behavior: The addition of neither NaN3 nor histidine (Figure 1F) nor deuteration of the solvent (Figure 1D) impacted the NADH photooxidation rate considerably. Hence, we suggest an eT mechanism to be active. This is supported by monitoring the 1O2‐driven photooxidation of histidine, proceeding for Rudppz drastically slower than for Rutbbpy and Ruphen (Figures S32–S33).

To take a closer look into the eT mechanism, transient absorption (TA) spectroscopy was performed (Figure 2). When Rudppz and NADH are mixed, the femtosecond (fs‐)TA showed the formation of an excited state absorption centered at 490 nm (purple line, Figure 2A, Figures S34C and D) in both, absence and presence of oxygen. This band is likely associated to the oxygen‐sensitive, phenazine‐hydrogenated Rudppz, which is assumed to form immediately upon excitation of Rudppz. No Rudppz feature at 490 nm was observed in fs‐TA of Ruddpz without NADH (Figure S34A–B). For Rudppz mixed with NADH in the presence of oxygen, predominant signatures from NADH radicals (NADH.+/NAD.) could only be observed in the nanosecond (ns‐) TA and not in the fs‐TA spectra (Figure 2). This was due to the more dominant spectral signature of RudppzH in the fs range than the spectral signatures from the NADH radicals. Photoexcited RudppzH decays within a μs thereby showcasing the spectral signatures corresponding to NADH radicals, which typically have a much longer lifetime, only after this time delay (Figure S35).

Figure 2

A) fs‐TA and ns‐TA spectra of the Rudppz/NADH system in the presence of oxygen at 470 nm excitation along with reference spectra for NADH.+ (red and green traces).[ , , ] B) Ground‐state UV‐vis absorption spectra of the system during the course of the ns‐TA experiment.

ns‐TA reveal accumulation of Rudppz (blue line, Figure 2B). During the experiment performed under inert conditions no features of NAD radicals could be recorded anymore (Figure S35). However, as upon exposure to air Rudppz was reformed (Figure 3B) and subsequently NAD radicals were detected in the ns‐TA spectra again (black line, Figure 2A and Figure S35), Rudppz followed a similar eT based NADH photooxidation mechanism as reported for flavin dyes.[ , ] To enter a new cycle, the photochemically reduced and inactivated chromophore Rudppz has to be reoxidized by oxygen. This ability of oxygen has already been shown.

Figure 3

1H NMR spectroscopic shifts of 4 mM Rudppz (bottom spectrum in every NMR‐panel, D2O:d6‐DMSO=555 : 45 (v:v)) upon addition of equimolar amounts of NADH (A, B) or NAD+ (D, E; upper 1H NMR spectra). C) and F) schematically indicate by colored dots the interactions between Rudppz and the reduced and oxidized nicotinamide, respectively (blue: attractive, orange: repulsive).

To further verify the mechanistic differences between Rutbbpy and Ruphen vs. Rudppz, NADH photooxidation under exclusion of oxygen but in presence of H2O2 serving as oxidant for reduced chromophore species was performed. As NAD+ formation was only observed for Rudppz (Figures S36), Rutbbpy and Ruphen are strictly oxygen‐dependent catalysts whereas Rudppz also operates in presence of a different oxidant. Stability of NADH even in presence of large excess H2O2 was verified (Figure S37).

The very surprising ability of Rudppz to outcompete the significantly longer‐lived Rutbbpy and Ruphen in a photocatalytic redox reaction in H2O was further investigated by 1H NMR spectroscopy. Addition of NADH but not NAD+ led to significant mutual shifts of signals assigned to the terminal benzene ring of the dppz ligand and the adenine protons (Figure 3 and Figures S38–S41). The adenine‐mediated interaction is furthermore supported by similar 1H NMR spectroscopic shifts upon addition of structurally analogous ADP (Figures S42–S44). This further confirms, that NADH is preorganized at the dppz ligand of Rudppz via the nucleobase moiety, thus enabling efficient photooxidation of the cofactor due to close spatial proximity. No such interaction with NADH was observed for Ruphen (Figure S45–S46). Titration experiments resulted in binding constants of 30.7±7.4 M−1 for NADH and 23.5±4.3 M−1 for ADP to Rudppz (Figures S42–S44 and S47–S48). Based on these rather low numbers, a dynamic exchange on the dppz sphere was verified by adding a 10‐fold excess of ADP to a typical NADH photooxidation process. For Rudppz this resulted in an activity drop of only 12 % (no change for Ruphen, Figure S49).

Together with the observed splitting of the aliphatic tbbpy resonances being larger for NADH than for ADP (Figures S44 and S48), a molecular picture of the above‐mentioned oxidation‐state sensitive interaction was delineated (Figures 3C and F): Upon photooxidation, the charge of the nicotinamide moiety is increased from neutral to onefold positive. Consequently the coulombic repulsion from the nearby Ru center, possibly in concert with other factors such as lowered lipophilicity, dominates over attractive π‐π interactions.

Following the light induced O2 consumption of the Rudppz/NADH system in operando in combination with associated UV‐vis and emission spectroscopic data, a 1 : 1 stoichiometry between consumed oxygen and converted NADH was found (Figure 4, a→b and Figure S50). Subsequent addition of catalase after all oxygen had been consumed (Figure 4A, step c), also allowed to determine a 1 : 1 ratio between the products H2O2 and NAD+, providing evidence that two high value products during the photocatalytic process were formed that could be used for further (biochemical) reactions.

Figure 4

A) Course of O2‐concentration during a typical NADH photooxidation experiment using a PBS buffered aqueous solution containing 500 μM NADH and 5 μM Rudppz under ambient conditions. Arrow a indicates the start of blue light irradiation of the solution which ends after ca. 20 min indicated by arrow b. Addition of catalase is marked by arrow c. B) UV‐vis (solid lines) and corresponding emission spectra (dotted lines) of the reaction mixture before irradiation (black curves, a) and after complete consumption of oxygen (blue curves, b).

Considering all results, a ligand‐dependent switch of the NADH‐photooxidation mechanism in the herein investigated series of photocatalysts is postulated (Scheme 2):[ , , , ] Rutbbpy and Ruphen follow the 1O2 pathway. However, Rudppz allows for supramolecular preorganization of NADH on the dppz ligand and thus enables an eT route. Under air, the catalytically inactive intermediate Rudppz gets finally reoxidized by oxygen to close the catalytic cycle.

Scheme 2

Proposed photocatalytic mechanism for A) Rutbbpy and Ruphen as well as B) Rudppz.

To utilize the efficient eT pathway biochemically, we coupled the Rudppz mediated NAD(P)+ regeneration to an enzymatic model reaction. NADP‐dependent malic enzyme (NADP‐dependent oxaloacetate‐decarboxylating malate dehydrogenase, EC 1.1.1.40, ME) was selected, converting L‐malate into pyruvate in two steps. However, enzyme stability revealed to be the limiting factor of the envisaged photobiocatalytic process. In view of the reactivity of 1O2 towards various amino acids,[ , ] irradiation (λ=465 nm) of ME in presence of Rutbbpy, Ruphen and Rudppz led to the formation of inactive orange precipitates inhibiting further consumption of NADPH by Rudppz as well as NADP+ by ME. 1O2 formation by Rudppz most likely took place via binding into hydrophobic pockets of ME as evident from an increased and air sensitive luminescence of the complex in presence of the enzyme (Figure S51).

Therefore, a discontinuous protocol cycling between the different oxidation states of NADP was realized. First, photooxidation of NADPH was performed followed by ME addition in the dark. Both UV‐vis absorption and emission spectroscopy clearly showed a successful cycling between NADPH and NADP+ proving the bioactivity of the generated oxidized nicotinamide (Figure 5). However, there are two possible explanations for the steadily decreasing efficiency of the photobiocatalytic process. Firstly, the formation of Rudppz containing precipitates during irradiation resulted in a lower amount of active complex in solution. Secondly, a potential co‐presence of some non‐precipitated enzyme continuously regenerated NADPH.

Figure 5

Discontinuous photobiocatalytic process followed by A) UV‐vis absorption and B) emission spectroscopy. Arrows indicate the start of irradiation (blue), addition of ME after switching off the light source (black) and addition of fresh Rudppz after removal of the orange precipitates by centrifugation (red).

In conclusion, it was found that well‐balanced supramolecular interactions can be utilized to overcome possible reactivity limitations of chromophores bearing short‐lived excited states. By introducing a suitable cofactor binding site nearby a light‐absorbing moiety, the preorganization induced mechanistic switch of NADH photooxidation from an 1O2 to an eT pathway allowed Rudppz to outperform Ru(II) prototype complexes, namely Rutbbpy and Ruphen. In future, the detailed insights into the catalytic mechanism gained in this study will allow further ligand‐based optimization of visible light‐driven NAD+ regeneration, finding application in photobiocatalysis and PDT. However, in view of the 1O2 sensitivity of different amino acids, technically optimized strategies such as permanent spatial separation of enzymes and 1O2 sensitizing chromophores will have to be pursued to guarantee continuous photobiocatalysis.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.